...

[iramuteq] / corpus.py

diff --git a/corpus.py b/corpus.py
index 0635f2a..e683630 100644 (file)
--- a/corpus.py
+++ b/corpus.py
@@ -38,8 +38,7 @@ def copycorpus(corpus) :
 
 class Corpus :
     """Corpus class
-    list of uci
-
+    list of text
     """
     def __init__(self, parent, parametres = {}, read = False) :
         self.parent = parent
@@ -510,6 +509,17 @@ class Corpus :
                 table_uce[uces[uce]][i] = 1
         table_uce.insert(0, list_act)
         return table_uce      
+    
+    def make_pondtable_with_classe(self, uces, list_act) :
+        table_uce = [[0 for val in list_act] for line in range(0,len(uces))]
+        uces = dict([[uce, i] for i, uce in enumerate(uces)])
+        for i, lem in enumerate(list_act) :
+            uceseff = self.getlemuceseff(lem)
+            lemuces = list(set(uceseff.keys()).intersection(uces))
+            for uce in lemuces :
+                table_uce[uces[uce]][i] = uceseff[uce]
+        table_uce.insert(0, list_act)
+        return table_uce 
 
     def parse_active(self, gramact, gramsup = None) :
         log.info('parse actives')
@@ -538,6 +548,8 @@ class Corpus :
         allactives = [[self.lems[lem].freq, lem] for lem in self.lems if self.lems[lem].act == key and self.lems[lem].freq >= 3]
         self.activenb = len(allactives)
         allactives = sorted(allactives, reverse = True)
+        if self.activenb == 0 :
+            return [], 0
         if len(allactives) <= nbmax :
             log.info('nb = %i - eff min = %i ' % (len(allactives), allactives[-1][0]))
             return [val[1] for val in allactives], allactives[-1][0]
@@ -834,8 +846,7 @@ class MakeUciStat :
         ucinb = corpus.getucinb()
         ucisize = corpus.getucisize()
         ucimean = float(sum(ucisize))/float(ucinb)
-        detoile = corpus.make_etoiles_dict()
-        
+        detoile = corpus.make_etoiles_dict() 
 
 class Uci :
     def __init__(self, iduci, line, paraset = None) :