* Refactor symbols, so that frequency rank can be derived from the orth id of a word.

bin/init_model.py

@@ -168,6 +168,11 @@ def setup_vocab(get_lex_attr, tag_map, src_dir, dst_dir):
             probs[word] = oov_prob
 
     lexicon = []
+    for word, prob in reversed(sorted(list(probs.items()), key=lambda item: item[1])):
+        # First encode the strings into the StringStore. This way, we can map
+        # the orth IDs to frequency ranks
+        orth = vocab.strings[word]
+    # Now actually load the vocab
     for word, prob in reversed(sorted(list(probs.items()), key=lambda item: item[1])):
         lexeme = vocab[word]
         lexeme.prob = prob

setup.py

@@ -165,7 +165,8 @@ MOD_NAMES = ['spacy.parts_of_speech', 'spacy.strings',
              'spacy.tokens.doc', 'spacy.tokens.spans', 'spacy.tokens.token',
              'spacy.serialize.packer', 'spacy.serialize.huffman', 'spacy.serialize.bits',
              'spacy.cfile', 'spacy.matcher',
-             'spacy.syntax.ner']
+             'spacy.syntax.ner',
+             'spacy.symbols']
 
 
 if __name__ == '__main__':

spacy/attrs.pxd

@@ -1,5 +1,6 @@
 # Reserve 64 values for flag features
 cpdef enum attr_id_t:
+    NULL_ATTR
     IS_ALPHA
     IS_ASCII
     IS_DIGIT
@@ -14,8 +15,7 @@ cpdef enum attr_id_t:
     IS_STOP
     IS_OOV
     
-    FLAG13 = 13
+    FLAG14 = 14
-    FLAG14
     FLAG15
     FLAG16
     FLAG17

spacy/attrs.pyx

@@ -0,0 +1,90 @@
+ATTR_IDS = {
+    "NULL_ATTR": NULL_ATTR,
+    "IS_ALPHA": IS_ALPHA,
+    "IS_ASCII": IS_ASCII,
+    "IS_DIGIT": IS_DIGIT,
+    "IS_LOWER": IS_LOWER,
+    "IS_PUNCT": IS_PUNCT,
+    "IS_SPACE": IS_SPACE,
+    "IS_TITLE": IS_TITLE,
+    "IS_UPPER": IS_UPPER,
+    "LIKE_URL": LIKE_URL,
+    "LIKE_NUM": LIKE_NUM,
+    "LIKE_EMAIL": LIKE_EMAIL,
+    "IS_STOP": IS_STOP,
+    "IS_OOV": IS_OOV,
+
+    "FLAG14": FLAG14,
+    "FLAG15": FLAG15,
+    "FLAG16": FLAG16,
+    "FLAG17": FLAG17,
+    "FLAG18": FLAG18,
+    "FLAG19": FLAG19,
+    "FLAG20": FLAG20,
+    "FLAG21": FLAG21,
+    "FLAG22": FLAG22,
+    "FLAG23": FLAG23,
+    "FLAG24": FLAG24,
+    "FLAG25": FLAG25,
+    "FLAG26": FLAG26,
+    "FLAG27": FLAG27,
+    "FLAG28": FLAG28,
+    "FLAG29": FLAG29,
+    "FLAG30": FLAG30,
+    "FLAG31": FLAG31,
+    "FLAG32": FLAG32,
+    "FLAG33": FLAG33,
+    "FLAG34": FLAG34,
+    "FLAG35": FLAG35,
+    "FLAG36": FLAG36,
+    "FLAG37": FLAG37,
+    "FLAG38": FLAG38,
+    "FLAG39": FLAG39,
+    "FLAG40": FLAG40,
+    "FLAG41": FLAG41,
+    "FLAG42": FLAG42,
+    "FLAG43": FLAG43,
+    "FLAG44": FLAG44,
+    "FLAG45": FLAG45,
+    "FLAG46": FLAG46,
+    "FLAG47": FLAG47,
+    "FLAG48": FLAG48,
+    "FLAG49": FLAG49,
+    "FLAG50": FLAG50,
+    "FLAG51": FLAG51,
+    "FLAG52": FLAG52,
+    "FLAG53": FLAG53,
+    "FLAG54": FLAG54,
+    "FLAG55": FLAG55,
+    "FLAG56": FLAG56,
+    "FLAG57": FLAG57,
+    "FLAG58": FLAG58,
+    "FLAG59": FLAG59,
+    "FLAG60": FLAG60,
+    "FLAG61": FLAG61,
+    "FLAG62": FLAG62,
+    "FLAG63": FLAG63,
+
+    "ID": ID,
+    "ORTH": ORTH,
+    "LOWER": LOWER,
+    "NORM": NORM,
+    "SHAPE": SHAPE,
+    "PREFIX": PREFIX,
+    "SUFFIX": SUFFIX,
+
+    "LENGTH": LENGTH,
+    "CLUSTER": CLUSTER,
+    "LEMMA": LEMMA,
+    "POS": POS,
+    "TAG": TAG,
+    "DEP": DEP,
+    "ENT_IOB": ENT_IOB,
+    "ENT_TYPE": ENT_TYPE,
+    "HEAD": HEAD,
+    "SPACY": SPACY,
+    "PROB": PROB,
+}
+
+# ATTR IDs, in order of the symbol
+ATTR_NAMES = [key for key, value in sorted(ATTR_IDS.items(), key=lambda item: item[1])]

spacy/matcher.pyx

@@ -15,7 +15,7 @@ from libcpp.vector cimport vector
 from murmurhash.mrmr cimport hash64
 
 from .attrs cimport LENGTH, ENT_TYPE, ORTH, NORM, LEMMA, LOWER, SHAPE
-from .attrs cimport FLAG13, FLAG14, FLAG15, FLAG16, FLAG17, FLAG18, FLAG19, FLAG20, FLAG21, FLAG22, FLAG23, FLAG24, FLAG25
+from .attrs cimport FLAG14, FLAG15, FLAG16, FLAG17, FLAG18, FLAG19, FLAG20, FLAG21, FLAG22, FLAG23, FLAG24, FLAG25
 from .tokens.doc cimport get_token_attr
 from .tokens.doc cimport Doc
 from .vocab cimport Vocab

spacy/parts_of_speech.pxd

@@ -1,23 +1,24 @@
-# Google universal tag set
+from .symbols cimport *
+
+
 cpdef enum univ_pos_t:
-    NO_TAG
+    NO_TAG = EMPTY_VALUE
-    ADJ
+    ADJ = POS_adj
-    ADP
+    ADP = POS_adp
-    ADV
+    ADV = POS_adv
-    AUX
+    AUX = POS_aux
-    CONJ
+    CONJ = POS_conj
-    DET
+    DET = POS_det
-    INTJ
+    INTJ = POS_intj
-    NOUN
+    NOUN = POS_noun
-    NUM
+    NUM = POS_num
-    PART
+    PART = POS_part
-    PRON
+    PRON = POS_pron
-    PROPN
+    PROPN = POS_propn
-    PUNCT
+    PUNCT = POS_punct
-    SCONJ
+    SCONJ = POS_sconj
-    SYM
+    SYM = POS_sym
-    VERB
+    VERB = POS_verb
-    X
+    X = POS_x
-    EOL
+    EOL = POS_eol
-    SPACE
+    SPACE = POS_space
-    N_UNIV_TAGS

spacy/vocab.pyx

@@ -67,6 +67,21 @@ cdef class Vocab:
         self._by_hash = PreshMap()
         self._by_orth = PreshMap()
         self.strings = StringStore()
+        # Load strings in a special order, so that we have an onset number for
+        # the vocabulary. This way, when words are added in order, the orth ID
+        # is the frequency rank of the word, plus a certain offset. The structural
+        # strings are loaded first, because the vocab is open-class, and these
+        # symbols are closed class.
+        #for attr_name in sorted(ATTR_NAMES.keys()):
+        #    _ = self.strings[attr_name]
+        #for univ_pos_name in sorted(UNIV_POS_NAMES.keys()):
+        #    _ = self.strings[pos_name]
+        #for morph_name in sorted(UNIV_MORPH_NAMES.keys()):
+        #    _ = self.strings[morph_name]
+        #for entity_type_name in sorted(ENTITY_TYPES.keys()):
+        #    _ = self.strings[entity_type_name]
+        #for tag_name in sorted(TAG_MAP.keys()):
+        #    _ = self.strings[tag_name]
         self.get_lex_attr = get_lex_attr
         self.morphology = Morphology(self.strings, tag_map, lemmatizer)
         self.serializer_freqs = serializer_freqs