From 5f226d0cd4049babcb0242bdbd614d870a6269de Mon Sep 17 00:00:00 2001
From: Ozbolt Menegatti <ozbolt.menegatti@gmail.com>
Date: Sun, 2 Jun 2019 12:53:16 +0200
Subject: [PATCH] fixing matching of agreements with msd

---
 wani.py | 50 ++++++++++++++++++++++++++------------------------
 1 file changed, 26 insertions(+), 24 deletions(-)

diff --git a/wani.py b/wani.py
index 036e05d..4043669 100644
--- a/wani.py
+++ b/wani.py
@@ -186,7 +186,7 @@ class LemmaCR(ComponentRepresentation):
 
 class LexisCR(ComponentRepresentation):
     def _render(self):
-        return self.data
+        return self.data['lexis']
     
 class WordFormAllCR(ComponentRepresentation):
     def _render(self):
@@ -224,10 +224,13 @@ class WordFormAnyCR(ComponentRepresentation):
 class WordFormMsdCR(WordFormAnyCR):
     def __init__(self, *args):
         super().__init__(*args)
-        self.backup_word = None
+        self.lemma = None
+        self.msd = None
 
     def check_msd(self, word):
-        selectors = self.data
+        if 'msd' not in self.data:
+            return True
+        selectors = self.data['msd']
         for key, value in selectors.items():
             t = word.msd[0]
             v = TAGSET[t].index(key.lower())
@@ -241,39 +244,37 @@ class WordFormMsdCR(WordFormAnyCR):
     pass
     
     def add_word(self, word):
-        if self.backup_word is None:
-            msd = self.word_renderer.get_lemma_msd(word.lemma, word.msd)
-            WordLemma = namedtuple('WordLemmaOnly', 'msd most_frequent_text lemma text')
-            self.backup_word = WordLemma(msd=msd, most_frequent_text=lambda *x: None, lemma=None, text=None)
+        if self.lemma is None:
+            self.lemma = word.lemma
+            self.msd = word.msd
 
         if self.check_msd(word):
             super().add_word(word)
     
     def _render(self):
-        self.words.append(self.backup_word)
+        msd = self.word_renderer.get_lemma_msd(self.lemma, self.msd)
+        WordLemma = namedtuple('WordLemmaOnly', 'msd most_frequent_text lemma text')
+        backup_word = WordLemma(msd=msd, most_frequent_text=lambda *x: None, lemma=None, text=None)
+        self.words.append(backup_word)
         return super()._render()
 
-class WordFormAgreementCR(ComponentRepresentation):
+class WordFormAgreementCR(WordFormMsdCR):
     def __init__(self, data, word_renderer):
         super().__init__(data, word_renderer)
-        self.agree_with, self.data = self.data
         self.rendition_candidate = None
     
     def get_agreement(self):
-        return self.agree_with
+        return self.data['other']
     
     def match(self, word_msd):
-        word_category = self.words[0].msd[0]
-        word_lemma = self.words[0].lemma
-        agreements = self.data
 
         existing = [(w.msd, w.text) for w in self.words]
 
-        for candidate_msd, candidate_text in self.word_renderer.available_words(word_lemma, existing):
-            if word_category != candidate_msd[0]:
+        for candidate_msd, candidate_text in self.word_renderer.available_words(self.lemma, existing):
+            if self.msd[0] != candidate_msd[0]:
                 continue
 
-            if WordFormAgreementCR.check_agreement(word_msd, candidate_msd, agreements):
+            if WordFormAgreementCR.check_agreement(word_msd, candidate_msd, self.data['agreement']):
                 self.rendition_candidate = candidate_text
                 return True
 
@@ -320,11 +321,9 @@ class WordFormAgreementCR(ComponentRepresentation):
 
 class ComponentRendition:
     def __init__(self):
-        self.more = None
+        self.more = {}
         self.representation_factory = ComponentRepresentation
     
-    def _set_more(self, m):
-        self.more = m
     
     def add_feature(self, feature):
         if 'rendition' in feature:
@@ -335,21 +334,24 @@ class ComponentRendition:
                 self.representation_factory = WordFormAnyCR
             elif feature['rendition'] == "lexis":
                 self.representation_factory = LexisCR
-                self.more = feature['string']
+                self.more['lexis'] = feature['string']
             else:
                 raise NotImplementedError("Representation rendition: {}".format(feature))
 
         elif 'selection' in feature:
             if feature['selection'] == "msd":
-                self.representation_factory = WordFormMsdCR
-                self.more = {k: v for k, v in feature.items() if k != 'selection'}
+                # could already be agreement
+                if self.representation_factory != WordFormAgreementCR:
+                    self.representation_factory = WordFormMsdCR
+                self.more['msd'] = {k: v for k, v in feature.items() if k != 'selection'}
             elif feature['selection'] == "all":
                 self.representation_factory = WordFormAllCR
             elif feature['selection'] == 'agreement':
                 assert(feature['head'][:4] == 'cid_')
                 assert(feature['msd'] is not None)
                 self.representation_factory = WordFormAgreementCR
-                self.more = (feature['head'][4:], feature['msd'].split('+'))
+                self.more['agreement'] = feature['msd'].split('+')
+                self.more['other'] = feature['head'][4:]
             else:
                 raise NotImplementedError("Representation selection: {}".format(feature))