spaCy/spacy/lang/zh/__init__.py

339 lines
12 KiB
Python
Raw Normal View History

2020-09-29 19:10:22 +00:00
from typing import Optional, List, Dict, Any, Callable, Iterable
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
from enum import Enum
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
import tempfile
import srsly
import warnings
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
from pathlib import Path
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
from ...errors import Warnings, Errors
2017-05-08 20:29:04 +00:00
from ...language import Language
from ...scorer import Scorer
2017-05-08 20:29:04 +00:00
from ...tokens import Doc
2020-09-29 19:10:22 +00:00
from ...training import validate_examples, Example
2020-09-30 08:20:14 +00:00
from ...util import DummyTokenizer, registry, load_config_from_str
from .lex_attrs import LEX_ATTRS
from .stop_words import STOP_WORDS
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
from ... import util
2020-09-29 19:09:10 +00:00
# fmt: off
_PKUSEG_INSTALL_MSG = "install spacy-pkuseg with `pip install spacy-pkuseg==0.0.26`"
2020-09-29 19:09:10 +00:00
# fmt: on
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
DEFAULT_CONFIG = """
[nlp]
2019-08-18 13:09:16 +00:00
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
[nlp.tokenizer]
@tokenizers = "spacy.zh.ChineseTokenizer"
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
segmenter = "char"
2020-09-29 19:10:22 +00:00
[initialize]
[initialize.tokenizer]
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
pkuseg_model = null
pkuseg_user_dict = "default"
"""
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
class Segmenter(str, Enum):
char = "char"
jieba = "jieba"
pkuseg = "pkuseg"
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
@classmethod
def values(cls):
return list(cls.__members__.keys())
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
@registry.tokenizers("spacy.zh.ChineseTokenizer")
2020-10-03 15:20:18 +00:00
def create_chinese_tokenizer(segmenter: Segmenter = Segmenter.char):
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
def chinese_tokenizer_factory(nlp):
2020-09-29 19:10:22 +00:00
return ChineseTokenizer(nlp, segmenter=segmenter)
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
return chinese_tokenizer_factory
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
class ChineseTokenizer(DummyTokenizer):
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
def __init__(
self, nlp: Language, segmenter: Segmenter = Segmenter.char,
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
):
self.vocab = nlp.vocab
if isinstance(segmenter, Segmenter):
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
segmenter = segmenter.value
self.segmenter = segmenter
self.pkuseg_seg = None
self.jieba_seg = None
if segmenter not in Segmenter.values():
warn_msg = Warnings.W103.format(
lang="Chinese",
segmenter=segmenter,
supported=", ".join(Segmenter.values()),
default="'char' (character segmentation)",
)
warnings.warn(warn_msg)
self.segmenter = Segmenter.char
2020-09-30 09:46:45 +00:00
if segmenter == Segmenter.jieba:
self.jieba_seg = try_jieba_import()
def initialize(
self,
get_examples: Optional[Callable[[], Iterable[Example]]] = None,
2020-09-30 09:46:45 +00:00
*,
nlp: Optional[Language] = None,
2020-09-30 09:46:45 +00:00
pkuseg_model: Optional[str] = None,
pkuseg_user_dict: Optional[str] = "default",
2020-09-30 09:46:45 +00:00
):
if self.segmenter == Segmenter.pkuseg:
if pkuseg_user_dict is None:
pkuseg_user_dict = pkuseg_model
2020-09-30 09:46:45 +00:00
self.pkuseg_seg = try_pkuseg_import(
pkuseg_model=pkuseg_model, pkuseg_user_dict=pkuseg_user_dict,
)
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
def __call__(self, text: str) -> Doc:
if self.segmenter == Segmenter.jieba:
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
words = list([x for x in self.jieba_seg.cut(text, cut_all=False) if x])
(words, spaces) = util.get_words_and_spaces(words, text)
return Doc(self.vocab, words=words, spaces=spaces)
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
elif self.segmenter == Segmenter.pkuseg:
if self.pkuseg_seg is None:
raise ValueError(Errors.E1000)
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
words = self.pkuseg_seg.cut(text)
(words, spaces) = util.get_words_and_spaces(words, text)
return Doc(self.vocab, words=words, spaces=spaces)
# warn if segmenter setting is not the only remaining option "char"
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
if self.segmenter != Segmenter.char:
warn_msg = Warnings.W103.format(
lang="Chinese",
segmenter=self.segmenter,
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
supported=", ".join(Segmenter.values()),
default="'char' (character segmentation)",
)
warnings.warn(warn_msg)
# split into individual characters
words = list(text)
(words, spaces) = util.get_words_and_spaces(words, text)
return Doc(self.vocab, words=words, spaces=spaces)
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
def pkuseg_update_user_dict(self, words: List[str], reset: bool = False):
if self.segmenter == Segmenter.pkuseg:
if reset:
try:
import spacy_pkuseg
2020-05-21 12:04:57 +00:00
self.pkuseg_seg.preprocesser = spacy_pkuseg.Preprocesser(None)
except ImportError:
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
msg = (
"spacy_pkuseg not installed: unable to reset pkuseg "
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
"user dict. Please " + _PKUSEG_INSTALL_MSG
)
raise ImportError(msg) from None
for word in words:
2020-05-21 12:04:57 +00:00
self.pkuseg_seg.preprocesser.insert(word.strip(), "")
else:
warn_msg = Warnings.W104.format(target="pkuseg", current=self.segmenter)
warnings.warn(warn_msg)
def score(self, examples):
validate_examples(examples, "ChineseTokenizer.score")
return Scorer.score_tokenization(examples)
def _get_config(self) -> Dict[str, Any]:
return {
"segmenter": self.segmenter,
}
def _set_config(self, config: Dict[str, Any] = {}) -> None:
self.segmenter = config.get("segmenter", Segmenter.char)
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
def to_bytes(self, **kwargs):
pkuseg_features_b = b""
pkuseg_weights_b = b""
pkuseg_processors_data = None
if self.pkuseg_seg:
with tempfile.TemporaryDirectory() as tempdir:
self.pkuseg_seg.feature_extractor.save(tempdir)
self.pkuseg_seg.model.save(tempdir)
tempdir = Path(tempdir)
with open(tempdir / "features.msgpack", "rb") as fileh:
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
pkuseg_features_b = fileh.read()
with open(tempdir / "weights.npz", "rb") as fileh:
pkuseg_weights_b = fileh.read()
pkuseg_processors_data = (
_get_pkuseg_trie_data(self.pkuseg_seg.preprocesser.trie),
self.pkuseg_seg.postprocesser.do_process,
sorted(list(self.pkuseg_seg.postprocesser.common_words)),
sorted(list(self.pkuseg_seg.postprocesser.other_words)),
)
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
serializers = {
"cfg": lambda: srsly.json_dumps(self._get_config()),
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
"pkuseg_features": lambda: pkuseg_features_b,
"pkuseg_weights": lambda: pkuseg_weights_b,
"pkuseg_processors": lambda: srsly.msgpack_dumps(pkuseg_processors_data),
}
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
return util.to_bytes(serializers, [])
def from_bytes(self, data, **kwargs):
2020-05-21 12:24:38 +00:00
pkuseg_data = {"features_b": b"", "weights_b": b"", "processors_data": None}
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
def deserialize_pkuseg_features(b):
2020-05-21 12:24:38 +00:00
pkuseg_data["features_b"] = b
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
def deserialize_pkuseg_weights(b):
2020-05-21 12:24:38 +00:00
pkuseg_data["weights_b"] = b
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
def deserialize_pkuseg_processors(b):
2020-05-21 12:24:38 +00:00
pkuseg_data["processors_data"] = srsly.msgpack_loads(b)
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
deserializers = {
"cfg": lambda b: self._set_config(srsly.json_loads(b)),
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
"pkuseg_features": deserialize_pkuseg_features,
"pkuseg_weights": deserialize_pkuseg_weights,
"pkuseg_processors": deserialize_pkuseg_processors,
}
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
util.from_bytes(data, deserializers, [])
2020-05-21 12:24:38 +00:00
if pkuseg_data["features_b"] and pkuseg_data["weights_b"]:
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
with tempfile.TemporaryDirectory() as tempdir:
tempdir = Path(tempdir)
with open(tempdir / "features.msgpack", "wb") as fileh:
2020-05-21 12:24:38 +00:00
fileh.write(pkuseg_data["features_b"])
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
with open(tempdir / "weights.npz", "wb") as fileh:
2020-05-21 12:24:38 +00:00
fileh.write(pkuseg_data["weights_b"])
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
try:
import spacy_pkuseg
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
except ImportError:
raise ImportError(
"spacy_pkuseg not installed. To use this model, "
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
+ _PKUSEG_INSTALL_MSG
) from None
self.pkuseg_seg = spacy_pkuseg.pkuseg(str(tempdir))
2020-05-21 12:24:38 +00:00
if pkuseg_data["processors_data"]:
processors_data = pkuseg_data["processors_data"]
(user_dict, do_process, common_words, other_words) = processors_data
self.pkuseg_seg.preprocesser = spacy_pkuseg.Preprocesser(user_dict)
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
self.pkuseg_seg.postprocesser.do_process = do_process
self.pkuseg_seg.postprocesser.common_words = set(common_words)
self.pkuseg_seg.postprocesser.other_words = set(other_words)
return self
def to_disk(self, path, **kwargs):
path = util.ensure_path(path)
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
def save_pkuseg_model(path):
if self.pkuseg_seg:
if not path.exists():
path.mkdir(parents=True)
self.pkuseg_seg.model.save(path)
self.pkuseg_seg.feature_extractor.save(path)
def save_pkuseg_processors(path):
if self.pkuseg_seg:
data = (
_get_pkuseg_trie_data(self.pkuseg_seg.preprocesser.trie),
self.pkuseg_seg.postprocesser.do_process,
sorted(list(self.pkuseg_seg.postprocesser.common_words)),
sorted(list(self.pkuseg_seg.postprocesser.other_words)),
)
srsly.write_msgpack(path, data)
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
serializers = {
"cfg": lambda p: srsly.write_json(p, self._get_config()),
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
"pkuseg_model": lambda p: save_pkuseg_model(p),
"pkuseg_processors": lambda p: save_pkuseg_processors(p),
}
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
return util.to_disk(path, serializers, [])
def from_disk(self, path, **kwargs):
path = util.ensure_path(path)
def load_pkuseg_model(path):
try:
import spacy_pkuseg
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
except ImportError:
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
if self.segmenter == Segmenter.pkuseg:
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
raise ImportError(
"spacy_pkuseg not installed. To use this model, "
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
+ _PKUSEG_INSTALL_MSG
) from None
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
if path.exists():
self.pkuseg_seg = spacy_pkuseg.pkuseg(path)
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
def load_pkuseg_processors(path):
try:
import spacy_pkuseg
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
except ImportError:
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
if self.segmenter == Segmenter.pkuseg:
raise ImportError(self._pkuseg_install_msg) from None
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
if self.segmenter == Segmenter.pkuseg:
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
data = srsly.read_msgpack(path)
(user_dict, do_process, common_words, other_words) = data
self.pkuseg_seg.preprocesser = spacy_pkuseg.Preprocesser(user_dict)
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
self.pkuseg_seg.postprocesser.do_process = do_process
self.pkuseg_seg.postprocesser.common_words = set(common_words)
self.pkuseg_seg.postprocesser.other_words = set(other_words)
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
serializers = {
"cfg": lambda p: self._set_config(srsly.read_json(p)),
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
"pkuseg_model": lambda p: load_pkuseg_model(p),
"pkuseg_processors": lambda p: load_pkuseg_processors(p),
}
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
util.from_disk(path, serializers, [])
2017-12-28 09:13:58 +00:00
class ChineseDefaults(Language.Defaults):
2020-09-30 08:20:14 +00:00
config = load_config_from_str(DEFAULT_CONFIG)
lex_attr_getters = LEX_ATTRS
stop_words = STOP_WORDS
writing_system = {"direction": "ltr", "has_case": False, "has_letters": False}
2017-12-28 09:13:58 +00:00
2019-03-11 16:10:50 +00:00
class Chinese(Language):
lang = "zh"
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
Defaults = ChineseDefaults
2016-05-05 09:39:12 +00:00
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
2020-09-30 09:46:45 +00:00
def try_jieba_import() -> None:
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
try:
import jieba
2020-09-30 09:46:45 +00:00
# segment a short text to have jieba initialize its cache in advance
list(jieba.cut("作为", cut_all=False))
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
return jieba
except ImportError:
2020-09-30 09:46:45 +00:00
msg = (
"Jieba not installed. To use jieba, install it with `pip "
" install jieba` or from https://github.com/fxsjy/jieba"
)
raise ImportError(msg) from None
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
2020-09-30 09:46:45 +00:00
def try_pkuseg_import(pkuseg_model: str, pkuseg_user_dict: str) -> None:
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
try:
import spacy_pkuseg
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
except ImportError:
msg = "spacy_pkuseg not installed. To use pkuseg, " + _PKUSEG_INSTALL_MSG
2020-09-30 09:46:45 +00:00
raise ImportError(msg) from None
try:
return spacy_pkuseg.pkuseg(pkuseg_model, pkuseg_user_dict)
Refactor pipeline components, config and language data (#5759) * Update with WIP * Update with WIP * Update with pipeline serialization * Update types and pipe factories * Add deep merge, tidy up and add tests * Fix pipe creation from config * Don't validate default configs on load * Update spacy/language.py Co-authored-by: Ines Montani <ines@ines.io> * Adjust factory/component meta error * Clean up factory args and remove defaults * Add test for failing empty dict defaults * Update pipeline handling and methods * provide KB as registry function instead of as object * small change in test to make functionality more clear * update example script for EL configuration * Fix typo * Simplify test * Simplify test * splitting pipes.pyx into separate files * moving default configs to each component file * fix batch_size type * removing default values from component constructors where possible (TODO: test 4725) * skip instead of xfail * Add test for config -> nlp with multiple instances * pipeline.pipes -> pipeline.pipe * Tidy up, document, remove kwargs * small cleanup/generalization for Tok2VecListener * use DEFAULT_UPSTREAM field * revert to avoid circular imports * Fix tests * Replace deprecated arg * Make model dirs require config * fix pickling of keyword-only arguments in constructor * WIP: clean up and integrate full config * Add helper to handle function args more reliably Now also includes keyword-only args * Fix config composition and serialization * Improve config debugging and add visual diff * Remove unused defaults and fix type * Remove pipeline and factories from meta * Update spacy/default_config.cfg Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> * Update spacy/default_config.cfg * small UX edits * avoid printing stack trace for debug CLI commands * Add support for language-specific factories * specify the section of the config which holds the model to debug * WIP: add Language.from_config * Update with language data refactor WIP * Auto-format * Add backwards-compat handling for Language.factories * Update morphologizer.pyx * Fix morphologizer * Update and simplify lemmatizers * Fix Japanese tests * Port over tagger changes * Fix Chinese and tests * Update to latest Thinc * WIP: xfail first Russian lemmatizer test * Fix component-specific overrides * fix nO for output layers in debug_model * Fix default value * Fix tests and don't pass objects in config * Fix deep merging * Fix lemma lookup data registry Only load the lookups if an entry is available in the registry (and if spacy-lookups-data is installed) * Add types * Add Vocab.from_config * Fix typo * Fix tests * Make config copying more elegant * Fix pipe analysis * Fix lemmatizers and is_base_form * WIP: move language defaults to config * Fix morphology type * Fix vocab * Remove comment * Update to latest Thinc * Add morph rules to config * Tidy up * Remove set_morphology option from tagger factory * Hack use_gpu * Move [pipeline] to top-level block and make [nlp.pipeline] list Allows separating component blocks from component order – otherwise, ordering the config would mean a changed component order, which is bad. Also allows initial config to define more components and not use all of them * Fix use_gpu and resume in CLI * Auto-format * Remove resume from config * Fix formatting and error * [pipeline] -> [components] * Fix types * Fix tagger test: requires set_morphology? Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com> Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com> Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
2020-07-22 11:42:59 +00:00
except FileNotFoundError:
2020-09-30 09:46:45 +00:00
msg = "Unable to load pkuseg model from: " + pkuseg_model
raise FileNotFoundError(msg) from None
2017-05-03 09:01:42 +00:00
Add pkuseg and serialization support for Chinese (#5308) * Add pkuseg and serialization support for Chinese Add support for pkuseg alongside jieba * Specify model through `Language` meta: * split on characters (if no word segmentation packages are installed) ``` Chinese(meta={"tokenizer": {"config": {"use_jieba": False, "use_pkuseg": False}}}) ``` * jieba (remains the default tokenizer if installed) ``` Chinese() Chinese(meta={"tokenizer": {"config": {"use_jieba": True}}}) # explicit ``` * pkuseg ``` Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "use_jieba": False, "use_pkuseg": True}}}) ``` * The new tokenizer setting `require_pkuseg` is used to override `use_jieba` default, which is intended for models that provide a pkuseg model: ``` nlp_pkuseg = Chinese(meta={"tokenizer": {"config": {"pkuseg_model": "default", "require_pkuseg": True}}}) nlp = Chinese() # has `use_jieba` as `True` by default nlp.from_bytes(nlp_pkuseg.to_bytes()) # `require_pkuseg` overrides `use_jieba` when calling the tokenizer ``` Add support for serialization of tokenizer settings and pkuseg model, if loaded * Add sorting for `Language.to_bytes()` serialization of `Language.meta` so that the (emptied, but still present) tokenizer metadata is in a consistent position in the serialized data Extend tests to cover all three tokenizer configurations and serialization * Fix from_disk and tests without jieba or pkuseg * Load cfg first and only show error if `use_pkuseg` * Fix blank/default initialization in serialization tests * Explicitly initialize jieba's cache on init * Add serialization for pkuseg pre/postprocessors * Reformat pkuseg install message
2020-04-18 15:01:53 +00:00
def _get_pkuseg_trie_data(node, path=""):
data = []
for c, child_node in sorted(node.children.items()):
data.extend(_get_pkuseg_trie_data(child_node, path + c))
if node.isword:
data.append((path, node.usertag))
return data
2019-08-18 13:09:16 +00:00
__all__ = ["Chinese"]