minor changes

bitext_align.py

@@ -25,18 +25,23 @@ def master_align(text0, text1, lang0, lang1):
     """ Takes two equivalent texts (original and trnslation) and returns 
         aligned texts. """
     df0 = frame_from_text(text0, source=lang0, target=lang1)
+    print('A')
     df1 = frame_from_text(text1, source=lang1, target=lang0, is1=True)
+    print('B')
     # returns dfs with ['sent', 'trans', 'rellen', 'relpos']
     anchors = anchors_from_frames(df0, df1, score_funct, score_threshold, window=2)
+    print('C')
     alignments = intermediate_align(df0, df1, anchors, lookahead=4)
+    print('D')
     textdict0, textdict1 = textdicts_from_alignments(df0, df1, alignments)
+    print('E')
     return textdict0, textdict1
 
 
-def frame_from_text(text, source='ru', target='en', is0=True): # 
+def frame_from_text(text, source='ru', target='en', is1=False): # 
     """  """ # 
     cols = [c+str(int(is1)) for c in ['sent','trans','rellen','relpos']]
-    frame = pd.DataFrame(colmns=cols)
+    frame = pd.DataFrame(columns=cols)
     frame[cols[0]] = sent_tokenize(text)
     frame[cols[1]] = frame[cols[0]].apply(lambda x: translate_client.translate(x, source_language=source, target_language=target, model='nmt')['translatedText'])
     frame[cols[2]] = frame[cols[0]].apply(lambda x: len(x))
@@ -55,7 +60,7 @@ def anchors_from_frames(frame0, frame1, window): #
     pairdf['rellen_ratio'] = (pairdf.rellen0/pairdf.rellen1).apply(gr1)
     pairdf['minlev'] = pairdf[['lev0', 'lev1']].min(axis=1)
     pairdf['maxlev'] = pairdf[['lev0', 'lev1']].min(axis=1)
-    pairdf['isanchor'] = pairdf.minlev<045 & pairdf.maxlev<0.6 & pairdf.rellen_ratio<1.3
+    pairdf['isanchor'] = pairdf.minlev<0.45 & pairdf.maxlev<0.6 & pairdf.rellen_ratio<1.3
     return pairdf[pairdf.isanchor][['index0','index1']]
 
 
@@ -113,11 +118,11 @@ def generate_pairdf(frame0, frame1, window):
     pairdf = pd.DataFrame(columns=['index0', 'index1'])
     ranges0 = frame0.relpos0
     ranges1 = frame1.relpos1
-    overlap = [i,j for (i,(a,b)),(j,(c,d)) in cp(enumerate(ranges0), enumerate(ranges1)) if get_overlap(a,b,c,d)>0]
+    overlap = [(i,j) for (i,(a,b)),(j,(c,d)) in cp(enumerate(ranges0), enumerate(ranges1)) if get_overlap(a,b,c,d)>0]
     allpairs = []
     for i,j in overlap:
-    	for k in range(-window:window+1):
-    		for l in range(-window:window+1):
+        for k in range(-window, window+1):
+            for l in range(-window, window+1):
                 allpairs.append()
     allpairs = sorted(list(set(allpairs)))
     pairdf[pairdf.columns] = pd.DataFrame(allpairs).values