Updating for new representations, for now only parsing

wani.py

@@ -129,7 +129,8 @@ class RestrictionType(Enum):
 class Rendition(Enum):
     Lemma = 0
     WordForm = 1
-    Unknown = 2
+    Lexis = 2
+    Unknown = 3
 
 class Order(Enum):
     FromTo = 0
@@ -164,8 +165,9 @@ class Order(Enum):
             raise NotImplementedError("Should not be here: Order match")
 
 class ComponentRendition:
-    def __init__(self, rendition=Rendition.Unknown):
-        self.word_form = {}
+    def __init__(self, rendition=Rendition.Unknown, lexis=None):
+        assert(type(rendition) is Rendition)
+        self.lexis = lexis
         self.rendition = rendition
 
     def render(self, word):
@@ -173,6 +175,8 @@ class ComponentRendition:
             return word.lemma
         elif self.rendition == Rendition.WordForm:
             return word.text
+        elif self.rendition == Rendition.Lexis:
+            return self.lexis
         elif self.rendition == Rendition.Unknown:
             return None
         else:
@@ -182,27 +186,30 @@ class ComponentRendition:
         return str(self.rendition)
 
 
-# dont know...
 class StructureSelection(Enum):
     All = 0
-    Frequency = 1
+    Msd = 1
 
 class ComponentRepresentation:
+    @staticmethod
     def new(s):
         if 'rendition' in s:
             if s['rendition'] == "lemma":
                 return ComponentRendition(Rendition.Lemma)
             elif s['rendition'] == "word_form":
                 return ComponentRendition(Rendition.WordForm)
+            elif s['rendition'] == "lexis":
+                assert(s['string'] is not None)
+                return ComponentRendition(Rendition.Lexis, s['string'])
             else:
                 raise NotImplementedError("Rendition: {}".format(s))
         elif 'selection' in s:
-            if s['selection'] == "frequency":
-                return StructureSelection.Frequency
+            if s['selection'] == "msd":
+                return StructureSelection.Msd
             elif s['selection'] == "all":
                 return StructureSelection.All
             else:
-                return {s['selection']: s['value']}
+                raise NotImplementedError("Selection: {}".format(s))
         else:
             return None
 
@@ -376,7 +383,7 @@ class Component:
 
     def set_representation(self, representation):
         cr = None
-        if representation is not None:
+        if len(representation) > 0:
             self.representation = []
 
             for feature in representation:
@@ -414,9 +421,6 @@ class Component:
                 others, r2 = next_component.find_next(deps, comps, restrs, reprs)
                 to_ret.extend(others)
 
-                if StructureSelection.Frequency in (r1, r2):
-                    representation = StructureSelection.Frequency
-
         return to_ret, representation
 
     def name_str(self):
@@ -566,7 +570,7 @@ class SyntacticStructure:
         for comp in definitions:
             n = comp.get('cid')
             restrs[n] = None
-            forms[n] = None
+            forms[n] = []
 
             for el in comp:
                 if el.tag.startswith("restriction"):
@@ -575,29 +579,29 @@ class SyntacticStructure:
                 elif el.tag.startswith("representation"):
                     st.add_representation(n, el, forms)
                 else:
-                    raise NotImplementedError("definition??")
+                    raise NotImplementedError("Unknown definition: {} in structure {}".format(el.tag, st.id))
 
         fake_root_component = Component({'cid': '#', 'type': 'other'})
         st.components, st.selection = fake_root_component.find_next(deps, comps, restrs, forms)
         return st
 
-    def add_representation(self, n, el, forms):
-        if el.tag == "representation":
-            els = [el]
-        elif el.tag == "representation_and":
-            els = list(el)
-        else:
-            raise NotImplementedError("Unknown representation tag: {}".format(el.tag))
+    def add_representation(self, n, rep_el, forms):
+        if rep_el.tag == "representation_and":
+            rep_el = rep_el[0]
+            logging.warning("Only using first reprentation in representation_and in structure {}".format(self.id))
 
-        for el in els:
-            if el.get('basic') == 'form':
-                assert(forms[n] is None)
-                forms[n] = el
-            elif el.get('basic') == "agreement":
+        assert(rep_el.tag == "representation")
+        for el in rep_el:
+            assert(el.tag == "feature")
+            if 'rendition' in el.attrib:
+                forms[n].append(el)
+            elif 'selection' in el.attrib and el.attrib["selection"] != "agreement":
+                forms[n].append(el)
+            elif 'selection' in el.attrib:
                 self.add_agreement(n, el)
             else:
-                logging.warning("Strange representation (basic={}) in structure {}. Skipping"
-                        .format(el.get('basic'), self.id))
+                logging.warning("Strange representation feature in structure {}. Skipping"
+                        .format(self.id))
                 continue
 
     def add_agreement(self, n, el):
@@ -605,12 +609,13 @@ class SyntacticStructure:
 
         n1 = n
         n2 = el.get('head')[4:]
-        agreement_str = next(iter(el)).get('agreement')
+        agreement_str = el.get('msd')
+        assert(agreement_str is not None)
 
         self.agreements.append({
             'n1': n1,
             'n2': n2,
-            'match': agreement_str.split('|')})
+            'match': agreement_str.split('+')})
 
     def __str__(self):
         comp_str = "\n".join(str(comp) for comp in self.components)