From 42bc3ad73bbf4434b799e341903dcea7d9cd8401 Mon Sep 17 00:00:00 2001
From: Matthew Honnibal <honnibal+gh@gmail.com>
Date: Sat, 9 Mar 2019 00:20:29 +0000
Subject: [PATCH] Fix class mapping for morphologizer

---
 spacy/morphology.pyx             | 10 +++++++---
 spacy/pipeline/morphologizer.pyx |  2 +-
 2 files changed, 8 insertions(+), 4 deletions(-)

diff --git a/spacy/morphology.pyx b/spacy/morphology.pyx
index cead0756b..eab7c9e31 100644
--- a/spacy/morphology.pyx
+++ b/spacy/morphology.pyx
@@ -1066,16 +1066,20 @@ FEATURES = [
 FEATURE_NAMES = {get_string_id(name): name for name in FEATURES}
 
 FEATURE_FIELDS = {feature: FIELDS[feature.split('_', 1)[0]] for feature in FEATURES}
+FIELD_SIZES = Counter(FEATURE_FIELDS.values())
+for field in FIELD_SIZES:
+    FIELD_SIZES[field] += 1
 for feat_id, name in FEATURE_NAMES.items():
     FEATURE_FIELDS[feat_id] = FEATURE_FIELDS[name]
-
-FIELD_SIZES = Counter(FEATURE_FIELDS.values())
+# Mapping of feature names to their position in total vector
 FEATURE_OFFSETS = {}
+# Mapping of field names to their first position in total vector.
 FIELD_OFFSETS = {}
 _seen_fields = Counter()
 for i, feature in enumerate(FEATURES):
     field = FEATURE_FIELDS[feature]
-    FEATURE_OFFSETS[feature] = _seen_fields[field]
+    # Add 1 for the NIL class, on each field
+    FEATURE_OFFSETS[feature] = _seen_fields[field] + 1
     if _seen_fields[field] == 0:
         FIELD_OFFSETS[field] = i
     _seen_fields[field] += 1 
diff --git a/spacy/pipeline/morphologizer.pyx b/spacy/pipeline/morphologizer.pyx
index 7d0ad42cf..589373f80 100644
--- a/spacy/pipeline/morphologizer.pyx
+++ b/spacy/pipeline/morphologizer.pyx
@@ -88,7 +88,7 @@ class Morphologizer(Pipe):
                     if doc_guesses[j, k] == 0:
                         doc_feat_ids[j, k] = 0
                     else:
-                        doc_feat_ids[j, k] = offset + doc_guesses[j, k]
+                        doc_feat_ids[j, k] = offset + (doc_guesses[j, k]-1)
                 # Get the set of feature names.
                 feats = {FEATURES[f] for f in doc_feat_ids[j] if f != 0}
                 # Now add the analysis, and set the hash.