Refactoring lexis/morphology matchers, now "pickable".

wani.py

@@ -442,66 +442,70 @@ def determine_ppb(rgx):
     else:
         return 4
 
-def build_morphology_regex(restriction):
+class MorphologyRegex:
-    restr_dict = {}
+    def __init__(self, restriction):
-    for feature in restriction:
+        self.min_msd_length = 1
-        feature_dict = dict(feature.items())
 
-        match_type = True
+        restr_dict = {}
-        if "filter" in feature_dict:
+        for feature in restriction:
-            assert feature_dict['filter'] == "negative"
+            feature_dict = dict(feature.items())
-            match_type = False
-            del feature_dict['filter']
 
-        assert len(feature_dict) == 1
+            match_type = True
-        key, value = next(iter(feature_dict.items()))
+            if "filter" in feature_dict:
-        restr_dict[key] = (value, match_type)
+                assert feature_dict['filter'] == "negative"
+                match_type = False
+                del feature_dict['filter']
 
-    assert 'POS' in restr_dict
+            assert len(feature_dict) == 1
-    category = restr_dict['POS'][0].capitalize()
+            key, value = next(iter(feature_dict.items()))
-    cat_code = CODES[category]
+            restr_dict[key] = (value, match_type)
-    rgx = [cat_code] + CATEGORY_BASES[cat_code]
 
-    del restr_dict['POS']
+        assert 'POS' in restr_dict
-    min_msd_length = 1
+        category = restr_dict['POS'][0].capitalize()
+        cat_code = CODES[category]
+        rgx = [cat_code] + CATEGORY_BASES[cat_code]
 
-    for attribute, (value, typ) in restr_dict.items():
+        del restr_dict['POS']
-        index = TAGSET[cat_code].index(attribute.lower())
-        assert index >= 0
 
-        if '|' in value:
+        for attribute, (value, typ) in restr_dict.items():
-            match = "".join(CODES[val] for val in value.split('|'))
+            index = TAGSET[cat_code].index(attribute.lower())
-        else:
+            assert index >= 0
-            match = CODES[value]
 
-        match = "[{}{}]".format("" if typ else "^", match)
+            if '|' in value:
-        rgx[index + 1] = match
+                match = "".join(CODES[val] for val in value.split('|'))
+            else:
+                match = CODES[value]
 
-        if typ:
+            match = "[{}{}]".format("" if typ else "^", match)
-            min_msd_length = max(index + 1, min_msd_length)
+            rgx[index + 1] = match
 
-    re_objects = [re.compile(r) for r in rgx]
+            if typ:
-    def matcher(text):
+                self.min_msd_length = max(index + 1, self.min_msd_length)
-        if len(text) <= min_msd_length:
+
+        self.re_objects = [re.compile(r) for r in rgx]
+        self.rgx = rgx
+    
+    def __call__(self, text):
+        if len(text) <= self.min_msd_length:
             return False
 
-        for c, r in zip(text, re_objects):
+        for c, r in zip(text, self.re_objects):
             if not r.match(c):
                 return False
         return True
 
-    return rgx, matcher
 
+class LexisRegex:
+    def __init__(self, restriction):
+        restr_dict = {}
+        for feature in restriction:
+            restr_dict.update(feature.items())
 
-def build_lexis_regex(restriction):
+        assert "lemma" in restr_dict
-    restr_dict = {}
+        self.match_list = restr_dict['lemma'].split('|')
-    for feature in restriction:
-        restr_dict.update(feature.items())
     
-    assert "lemma" in restr_dict
+    def __call__(self, text):
-    match_list = restr_dict['lemma'].split('|')
+        return text in self.match_list
-
-    return match_list, lambda text: text in match_list
 
 
 class Restriction:
@@ -517,13 +521,12 @@ class Restriction:
         restriction_type = restriction_tag.get('type')
         if restriction_type == "morphology":
             self.type = RestrictionType.Morphology
-            present, self.matcher = build_morphology_regex(list(restriction_tag))
+            self.matcher = MorphologyRegex(list(restriction_tag))
-            self.present = " ".join(present)
+            self.ppb = determine_ppb(self.matcher.rgx)
-            self.ppb = determine_ppb(present)
 
         elif restriction_type == "lexis":
             self.type = RestrictionType.Lexis
-            self.present, self.matcher = build_lexis_regex(list(restriction_tag))
+            self.matcher = LexisRegex(list(restriction_tag))
         else:
             raise NotImplementedError()
 
@@ -802,7 +805,7 @@ def get_lemma_features(et):
 
     result = {}
     for pos in lf.iter('POS'):
-        rgx_list, _ = build_morphology_regex(pos)
+        rgx_list = MorphologyRegex(pos).rgx
         rgx_str = ""
         for position in rgx_list:
             if position == ".":