...

[iramuteq] / corpus.py

diff --git a/corpus.py b/corpus.py
index 4ba60d1..6e667b3 100644 (file)
--- a/corpus.py
+++ b/corpus.py
@@ -177,6 +177,10 @@ class Corpus :
         query = 'SELECT uces FROM uces where id IN (%s) ORDER BY id' % formesid
         res = self.cformes.execute(query)
         return list(set(list(itertools.chain(*[[int(val) for val in row[0].split()] if not isinstance(row[0], int) else [row[0]] for row in res]))))
+    
+    def gettgentxt(self, tgen):
+        sts = self.gettgenst(tgen)
+        return list(set([self.getucefromid(val).uci for val in sts]))
 
     def getlemucis(self, lem) :
         uces = self.getlemuces(lem)
@@ -397,6 +401,26 @@ class Corpus :
                         tgenoccurrences[t][etoiles[i]] += sum([lemuceeff[uce] for uce in concern])
         return tgenoccurrences, totoccurrences
     
+    def make_tgen_profile(self, tgen, ucecl, uci = False) :
+        log.info('tgen/classes')
+        if uci :
+            tab = [[lem] + [len(set(self.gettgentxt(tgen[lem])).intersection(classe)) for classe in ucecl] for lem in tgen]
+        else :
+            tab = [[lem] + [len(set(self.gettgenst(tgen[lem])).intersection(classe)) for classe in ucecl] for lem in tgen]
+        tab = [[line[0]] + [val for val in line[1:]] for line in tab if sum(line[1:]) >= 3]
+        return tab
+        #i = 0
+        #nam = 'total'
+        #while nam + `i` in tgen :
+        #    i += 1
+        #nam = nam + `i`
+        #last = [nam] + [`len(classe)` for classe in ucecl]
+        #tab += [last]
+        #line0 = ['tgen'] + ['_'.join(['cluster', `i+1`]) for i in range(len(ucecl))]
+        #tab = [line0] + tab
+        #with open(fileout, 'w') as f :
+        #    f.write('\n'.join(['\t'.join(line) for line in tab]).encode(self.parametres['syscoding']))        
+    
     def make_efftype_from_etoiles(self, etoiles) :
         dtype = {}
         etuces = [[] for et in etoiles]
@@ -1573,6 +1597,8 @@ class Builder :
                 filein = parametres['dictionary']
             else :
                 filein = None
+            if dial.corpusname.GetValue() != '' :
+                parametres['corpus_name'] = dial.corpusname.GetValue()
             dial.Destroy()
             ReadLexique(self.parent, lang = parametres['lang'], filein = filein)
             if parametres['lang'] != 'other' and  os.path.exists(self.parent.DictPath.get(parametres['lang']+'_exp', 'french_exp')):