plint

commit 484c48a9b590a86af96c1c7d32d17ba4159c6a44
parent d7c04b0f6cbf165720881422f20cadc60eabe641
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Sun, 27 Jul 2014 14:06:25 +0200

fix crash on "quinqu-"

Diffstat:
verse.py
 | 
15
++++++++++++---

1 file changed, 12 insertions(+), 3 deletions(-)
diff --git a/verse.py b/verse.py
@@ -145,9 +145,6 @@ class Verse:
         if w[i]['text'].endswith('g') and len(w[i+1]['text']) >= 2:
           if w[i+1]['text'][1] in "eéèa":
             w[i+1]['text'] = w[i+1]['text'][1:]
-    # remove empty chunks created by simplifications
-    for i, w in enumerate(self.chunks):
-      self.chunks[i] = [x for x in w if len(x['text']) > 0]
 
     # remove leading and trailing crap
     for w in self.chunks:
@@ -157,6 +154,18 @@ class Verse:
         while len(w[p]['text']) > 0 and w[p]['text'][-1] in ' -':
           w[p]['text'] = w[p]['text'][:-1]
 
+    # collapse empty chunks created by simplifications
+    for i, w in enumerate(self.chunks):
+      new_chunks = []
+      for x in self.chunks[i]:
+        if len(x['text']) > 0:
+          new_chunks.append(x)
+        else:
+          # propagate the original text
+          # newly empty chunks cannot be the first ones
+          new_chunks[-1]['original'] += x['original']
+      self.chunks[i] = new_chunks
+
     # sigles
     for i, w in enumerate(self.chunks):
       if len(w) == 1 and is_consonants(w[0]['text']):