...

[iramuteq] / corpusNG.py

diff --git a/corpusNG.py b/corpusNG.py
index 6a027dc..19cfd29 100644 (file)
--- a/corpusNG.py
+++ b/corpusNG.py
@@ -512,18 +512,31 @@ class Corpus :
             et = etoile.split('_')
             if et[0] in det :
                 try :
             et = etoile.split('_')
             if et[0] in det :
                 try :

-                    if et[1] in det[et[0]] :
-                        det[et[0]][et[1]] += 1
+                    endet = '_'.join(et[1:])
+                    if endet in det[et[0]] :
+                        det[et[0]][endet] += 1

                     else :
                     else :

-                        det[et[0]][et[1]] = 1
+                        det[et[0]][endet] = 1

                 except IndexError :
                     det[et[0]] += 1
             else :
                 try :
                 except IndexError :
                     det[et[0]] += 1
             else :
                 try :

-                    det[et[0]] = {et[1] :1}
+                    endet = '_'.join(et[1:])
+                    det[et[0]] = {endet :1}

                 except IndexError :
                     det[et[0]] = 1
                 except IndexError :
                     det[et[0]] = 1

-        print det
+        return det
+
+    def make_etline(self, listet) :
+        etuces = [[] for et in listet]
+        for uci in self.ucis :
+            get = list(set(uci.etoiles).intersection(listet))
+            if len(get) > 1 :
+                return '2 variables sur la meme ligne'
+            elif get != [] :
+                etuces[listet.index(get[0])] += [uce.ident for uce in uci.uces]
+        return etuces
+            

 
     def make_and_write_profile_et(self, ucecl, fileout) :
         log.info('etoiles/classes')
 
     def make_and_write_profile_et(self, ucecl, fileout) :
         log.info('etoiles/classes')


@@ -906,8 +919,9 @@ class BuildCorpus :
             self.cleans.append(self.dolower)
         if self.corpus.parametres.get('firstclean', 1) :
             self.cleans.append(self.firstclean)
             self.cleans.append(self.dolower)
         if self.corpus.parametres.get('firstclean', 1) :
             self.cleans.append(self.firstclean)

-        self.rule = self.corpus.parametres.get('keep_caract', u"^a-zA-Z0-9àÀâÂäÄáÁéÉèÈêÊëËìÌîÎïÏòÒôÔöÖùÙûÛüÜçÇßœŒ’ñ.:,;!?*'_")
-        self.cleans.append(self.docharact)
+        if self.corpus.parametres['charact'] :
+            self.rule = self.corpus.parametres.get('keep_caract', u"^a-zA-Z0-9àÀâÂäÄáÁéÉèÈêÊëËìÌîÎïÏòÒôÔöÖùÙûÛüÜçÇßœŒ’ñ.:,;!?*'_")
+            self.cleans.append(self.docharact)

         if self.corpus.parametres.get('expressions', 1) :
             self.cleans.append(self.make_expression)
         if self.corpus.parametres.get('apos', 1) :
         if self.corpus.parametres.get('expressions', 1) :
             self.cleans.append(self.make_expression)
         if self.corpus.parametres.get('apos', 1) :