From 2805068ca858f8d445d9ac0bdf6e87713a1d2b9e Mon Sep 17 00:00:00 2001 From: Matthew Honnibal Date: Tue, 14 Oct 2014 15:21:03 +1100 Subject: [PATCH] * Have tokens track tuples that record the start offset and pos tag as well as a lexeme pointer --- spacy/tokens.pxd | 17 ++++++-- spacy/tokens.pyx | 106 +++++++++++++++++++++++++++-------------------- 2 files changed, 73 insertions(+), 50 deletions(-) diff --git a/spacy/tokens.pxd b/spacy/tokens.pxd index 8fd58ea8c..38918e092 100644 --- a/spacy/tokens.pxd +++ b/spacy/tokens.pxd @@ -2,19 +2,28 @@ from spacy.lexeme cimport LexemeC from libcpp.vector cimport vector +cdef struct Token: + int i + int pos + LexemeC* lex + + cdef class Tokens: - cdef vector[LexemeC*] *v + cdef vector[Token] v + + cdef int extend(self, int i, LexemeC** lexemes, int n) except -1 + cdef int push_back(self, int i, LexemeC* lexeme) except -1 cpdef int id(self, size_t i) except -1 - cpdef unicode string(self, size_t i) cpdef float prob(self, size_t i) except 1 cpdef int cluster(self, size_t i) except * cpdef bint check_orth_flag(self, size_t i, size_t flag_id) except * cpdef bint check_dist_flag(self, size_t i, size_t flag_id) except * cpdef unicode string_view(self, size_t i, size_t view_id) - cpdef unicode casefix(self, size_t i) - cpdef unicode shape(self, size_t i) + cpdef unicode orig(self, size_t i) + cpdef unicode norm(self, size_t i) + cpdef unicode shape(self, size_t i) cpdef unicode unsparse(self, size_t i) cpdef unicode asciied(self, size_t i) cpdef bint is_alpha(self, size_t i) except * diff --git a/spacy/tokens.pyx b/spacy/tokens.pyx index c15ad7de1..44bca0b45 100644 --- a/spacy/tokens.pyx +++ b/spacy/tokens.pyx @@ -25,127 +25,141 @@ cdef class Tokens: """ def __cinit__(self, string_length=0): size = int(string_length / 3) if string_length >= 3 else 1 - self.v = new vector[LexemeC*]() + self.v = vector[Token]() self.v.reserve(size) def __getitem__(self, i): - return Lexeme(self.v.at(i)) + return Lexeme(self.v.at(i).lex) def __len__(self): return self.v.size() - def __dealloc__(self): - del self.v + cdef int push_back(self, int idx, LexemeC* lexeme) except -1: + self.v.push_back(Token(idx, 0, lexeme)) + return idx + lexeme.ints[LexInt_length] - def append(self, Lexeme lexeme): - self.v.push_back(lexeme._c) + cdef int extend(self, int idx, LexemeC** lexemes, int n) except -1: + cdef int i + if lexemes == NULL: + return idx + elif n == 0: + i = 0 + while lexemes[i] != NULL: + idx = self.push_back(idx, lexemes[i]) + else: + for i in range(n): + idx = self.push_back(idx, lexemes[i]) + return idx cpdef int id(self, size_t i) except -1: - return self.v.at(i).ints[LexInt_i] - - cpdef unicode string(self, size_t i): - cdef bytes utf8_string = self.v.at(i).strings[LexStr_orig] - cdef unicode string = utf8_string.decode('utf8') - return string + return self.v.at(i).lex.ints[LexInt_id] cpdef float prob(self, size_t i) except 1: - return self.v.at(i).floats[LexFloat_prob] + return self.v.at(i).lex.floats[LexFloat_prob] cpdef int cluster(self, size_t i) except *: - return self.v.at(i).ints[LexInt_cluster] + return self.v.at(i).lex.ints[LexInt_cluster] cpdef bint check_orth_flag(self, size_t i, size_t flag_id) except *: - return lexeme_check_orth_flag(self.v.at(i), flag_id) + return lexeme_check_orth_flag(self.v.at(i).lex, flag_id) cpdef bint check_dist_flag(self, size_t i, size_t flag_id) except *: - return lexeme_check_dist_flag(self.v.at(i), flag_id) + return lexeme_check_dist_flag(self.v.at(i).lex, flag_id) cpdef unicode string_view(self, size_t i, size_t view_id): - return lexeme_get_string(self.v.at(i), view_id) + return lexeme_get_string(self.v.at(i).lex, view_id) # Provide accessor methods for the features supported by the language. # Without these, clients have to use the underlying string_view and check_flag # methods, which requires them to know the IDs. - cpdef unicode casefix(self, size_t i): - return lexeme_get_string(self.v.at(i), LexStr_casefix) + + cpdef unicode orig(self, size_t i): + cdef bytes utf8_string = self.v.at(i).lex.strings[LexStr_orig] + cdef unicode string = utf8_string.decode('utf8') + return string + + cpdef unicode norm(self, size_t i): + cdef bytes utf8_string = self.v.at(i).lex.strings[LexStr_norm] + cdef unicode string = utf8_string.decode('utf8') + return string cpdef unicode shape(self, size_t i): - return lexeme_get_string(self.v.at(i), LexStr_shape) + return lexeme_get_string(self.v.at(i).lex, LexStr_shape) cpdef unicode unsparse(self, size_t i): - return lexeme_get_string(self.v.at(i), LexStr_unsparse) + return lexeme_get_string(self.v.at(i).lex, LexStr_unsparse) cpdef unicode asciied(self, size_t i): - return lexeme_get_string(self.v.at(i), LexStr_asciied) + return lexeme_get_string(self.v.at(i).lex, LexStr_asciied) cpdef bint is_alpha(self, size_t i) except *: - return lexeme_check_orth_flag(self.v.at(i), LexOrth_alpha) + return lexeme_check_orth_flag(self.v.at(i).lex, LexOrth_alpha) cpdef bint is_ascii(self, size_t i) except *: - return lexeme_check_orth_flag(self.v.at(i), LexOrth_ascii) + return lexeme_check_orth_flag(self.v.at(i).lex, LexOrth_ascii) cpdef bint is_digit(self, size_t i) except *: - return lexeme_check_orth_flag(self.v.at(i), LexOrth_digit) + return lexeme_check_orth_flag(self.v.at(i).lex, LexOrth_digit) cpdef bint is_lower(self, size_t i) except *: - return lexeme_check_orth_flag(self.v.at(i), LexOrth_lower) + return lexeme_check_orth_flag(self.v.at(i).lex, LexOrth_lower) cpdef bint is_punct(self, size_t i) except *: - return lexeme_check_orth_flag(self.v.at(i), LexOrth_punct) + return lexeme_check_orth_flag(self.v.at(i).lex, LexOrth_punct) cpdef bint is_space(self, size_t i) except *: - return lexeme_check_orth_flag(self.v.at(i), LexOrth_space) + return lexeme_check_orth_flag(self.v.at(i).lex, LexOrth_space) cpdef bint is_title(self, size_t i) except *: - return lexeme_check_orth_flag(self.v.at(i), LexOrth_title) + return lexeme_check_orth_flag(self.v.at(i).lex, LexOrth_title) cpdef bint is_upper(self, size_t i) except *: - return lexeme_check_orth_flag(self.v.at(i), LexOrth_upper) + return lexeme_check_orth_flag(self.v.at(i).lex, LexOrth_upper) cpdef bint can_adj(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_adj) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_adj) cpdef bint can_adp(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_adp) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_adp) cpdef bint can_adv(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_adv) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_adv) cpdef bint can_conj(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_conj) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_conj) cpdef bint can_det(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_det) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_det) cpdef bint can_noun(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_noun) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_noun) cpdef bint can_num(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_num) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_num) cpdef bint can_pdt(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_pdt) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_pdt) cpdef bint can_pos(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_pos) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_pos) cpdef bint can_pron(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_pron) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_pron) cpdef bint can_prt(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_prt) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_prt) cpdef bint can_punct(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_punct) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_punct) cpdef bint can_verb(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_verb) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_verb) cpdef bint oft_lower(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_lower) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_lower) cpdef bint oft_title(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_title) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_title) cpdef bint oft_upper(self, size_t i) except *: - return lexeme_check_dist_flag(self.v.at(i), LexDist_upper) + return lexeme_check_dist_flag(self.v.at(i).lex, LexDist_upper)