french dictionary

[iramuteq] / corpus.py

diff --git a/corpus.py b/corpus.py
index 2e85ed5..0635f2a 100644 (file)
--- a/corpus.py
+++ b/corpus.py
@@ -77,10 +77,10 @@ class Corpus :
                 gramtype = self.parent.lexique[word][1]
                 lem = self.parent.lexique[word][0]
             elif word.isdigit() :
-                gramtype = 'num'
+                gramtype = u'num'
                 lem = word
             else :
-                gramtype = 'nr'
+                gramtype = u'nr'
                 lem = word
             self.formes[word] =  Word(word, gramtype, len(self.formes), lem)
             self.idformesuces[self.formes[word].ident] = {self.ucis[-1].uces[-1].ident : 1}
@@ -275,8 +275,12 @@ class Corpus :
         if self.iduces is None :
             self.iduces = dict([[uce.ident, uce] for uci in self.ucis for uce in uci.uces])
 
-    def make_lexitable(self, mineff, etoiles) :
-        tokeep = [lem for lem in self.lems if self.lems[lem].freq >= mineff]
+    def make_lexitable(self, mineff, etoiles, gram = 0) :
+        if gram == 0 :
+            grams = {1:'', 2:''}
+        else :
+            grams = {gram :''}
+        tokeep = [lem for lem in self.lems if self.lems[lem].freq >= mineff and self.lems[lem].act in grams]
         etuces = [[] for et in etoiles]
         for uci in self.ucis :
             get = list(set(uci.etoiles).intersection(etoiles))
@@ -434,7 +438,7 @@ class Corpus :
                 f.write(guce.encode(self.parametres['syscoding']) + '\n\n')
 
     def export_classe(self, outf, classe, lem = False) :
-        sts = self.lc[classe] 
+        sts = self.lc[classe - 1] 
         res = self.getconcorde(sts)
         self.make_iduces()
         with open(outf, 'w') as f :
@@ -1081,7 +1085,7 @@ class BuildCorpus :
     def firstclean(self, txt) :
         txt = txt.replace(u'’',"'")
         txt = txt.replace(u'œ', u'oe')
-        return txt.replace('...',u' £$£ ').replace('?',' ? ').replace('.',' . ').replace('!', ' ! ').replace(',',' , ').replace(';', ' ; ').replace(':',' : ').replace(u'…', ' £$£ ')
+        return txt.replace('...',u' £$£ ').replace('?',' ? ').replace('.',' . ').replace('!', ' ! ').replace(',',' , ').replace(';', ' ; ').replace(':',' : ').replace(u'…', u' £$£ ')
 
     def make_cleans(self, txt) :
         for clean in self.cleans :