Removed most of irrelevant syntactic realizations from tei-xml

common/management/commands/valunifier_tei.py

@@ -8,7 +8,7 @@ from xml.sax.saxutils import escape
 
 from unifier.models import UnifiedFrame2SlowalFrameMapping, \
                            UnifiedFrameArgumentSlowalFrameMapping
-from connections.models import ArgumentConnection
+from connections.models import ArgumentConnection, ExampleConnection
 
 from collections import defaultdict
 
@@ -325,7 +325,7 @@ def write_lexical_unit(parent, lexical_unit, mapping):
         opinion_symbol.attrib['value'] = slowal_frame.opinion.key
         
 def write_alternations(parent, unified_frame, lexical_unit, mapping, used_schemata):
-    alternations = prepare_alternations(mapping, used_schemata)
+    alternations = prepare_alternations(lexical_unit, mapping, used_schemata)
     for key in sorted(alternations.keys()):
         alternation_fs = etree.SubElement(parent, 'fs')
         alternation_fs.attrib['type'] = 'aternation'
@@ -344,7 +344,9 @@ def write_alternations(parent, unified_frame, lexical_unit, mapping, used_schema
             phrases_f = etree.SubElement(connection_fs, 'f')
             write_phrases_coll(phrases_f, schema_hooks)
 
-def prepare_alternations(mapping, used_schemata):
+def prepare_alternations(lexical_unit, mapping, used_schemata):
+    connections_info = analyse_connections(mapping)
+    
     argument_mappings = UnifiedFrameArgumentSlowalFrameMapping.objects.filter(unified_frame_mapping = mapping)
     alternations = defaultdict(lambda: defaultdict(lambda: []))
     for argument_mapping in argument_mappings:
@@ -354,14 +356,41 @@ def prepare_alternations(mapping, used_schemata):
         argument_realization = ArgumentConnection.objects.get(argument = sargument)
         by_schema_realizations = argument_realization.schema_connections.all()
         for schema_hook in by_schema_realizations:
-            subentry = schema_hook.subentry
-            schema = schema_hook.schema
-            used_schemata.add((subentry, schema))
-            alternation = schema_hook.alternation
-            alternations[(subentry.id, schema.id, alternation)][uargument].append(schema_hook)
+            if valid_connection(lexical_unit, schema_hook, connections_info):
+                subentry = schema_hook.subentry
+                schema = schema_hook.schema
+                used_schemata.add((subentry, schema))
+                alternation = schema_hook.alternation
+                alternations[(subentry.id, schema.id, alternation)][uargument].append(schema_hook)
 
     return alternations
 
+def analyse_connections(mapping):
+    sframe = mapping.slowal_frame
+    lus = sframe.lexical_units.all()
+    lu_schema_connections = defaultdict(lambda: set())
+    connected_lu_count = defaultdict(lambda: 0)
+    for lu in lus:
+        examples = ExampleConnection.objects.filter(lexical_unit=lu)
+        for example in examples:
+            scs = example.schema_connections.all()
+            if len(scs) > 0:
+                schema_hook = scs[0]
+                schema_id = (schema_hook.subentry.id, schema_hook.schema.id)
+                lu_schema_connections[lu].add(schema_id)
+                connected_lu_count[schema_id] += 1
+    return (lu_schema_connections, connected_lu_count)
+
+def valid_connection(lexical_unit, schema_hook, connections_info):
+    schema_id = (schema_hook.subentry.id, schema_hook.schema.id)
+    lu_schema_connections, connected_lu_count = connections_info
+    if schema_id in lu_schema_connections[lexical_unit]:
+        return True
+    if connected_lu_count[schema_id] == 0:
+        # @TODO: check for "się"
+        return True
+    return False
+
 def write_phrases_coll(parent, phrases_list):
     vColl = etree.SubElement(parent, 'vColl')
     vColl.attrib['org'] = 'set'