drime

commit fe81b24132a9a162c58a30a73100bda85ea2e6dc
parent 11d1c425321e7d9d5f83e645b84e2d43b2c876ba
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Tue, 27 Dec 2011 02:47:07 +0100

more versatile

Diffstat:
query.py
 | 
17
++++++++++-------

1 file changed, 10 insertions(+), 7 deletions(-)
diff --git a/query.py b/query.py
@@ -135,30 +135,33 @@ def do_query(word, phon, minsyll, maxsyll, elide, gender, offset, size):
     row['freq'] = float(row['freq'])
     row['phon_rhyme'] = lcs(phon, row['phon'])
     row['word_rhyme'] = lcs(word, row['word'])
+    bases = row['orig'].split(',')
+    for i in range(len(bases)):
+      bases[i] = bases[i].split('|')
+    row['orig'] = bases
     row['key'] = (
         -row['phon_rhyme'], # phon_rhyme desc
         -row['word_rhyme'], # eye_rhyme desc
-        #TODO row['base'] != row['word'], # same as base
+        #-len([x for x in row['orig'] if x[1] == row['word']]), # same as base
         -row['freq'], # frequency desc
         row['word'] # alphabetical order
         )
+
     result.append(row)
   print ("DONE2")
 
   result2 = []
   seen = set()
   for row in sorted(result, key=operator.itemgetter('key')):
-    bases = row['orig'].split(',')
     ok = False
-    for i in range(len(bases)):
-      bases[i] = bases[i].split('|')
-      if bases[i][1] not in seen:
+    for i in range(len(row['orig'])):
+      if row['orig'][i][1] not in seen:
         ok = True
-      seen.add(bases[i][1])
+      seen.add(row['orig'][i][1])
     if ok:
       row['orig'] = ', '.join(
             [a[0] + (' ('+a[1]+')' if row['word'] != a[1] else '')
-              for a in bases])
+              for a in row['orig']])
       row['phon'] = to_xsampa(row['phon'])
       result2.append(row)
       seen.add(row['word'])