plint

commit da866fcbd6d798f5ba5f73bfb224b8f95dde1857
parent 3e6eb41e7c0b2a85b5fd4230cc3801e5d92278b9
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Sat, 21 Sep 2013 11:08:16 +0200

fix problem with apostrophes followed by spaces

Diffstat:
common.py
 | 
5
+++--
verse.py
 | 
15
++++++++++++++-

2 files changed, 17 insertions(+), 3 deletions(-)
diff --git a/common.py b/common.py
@@ -6,6 +6,7 @@ import re
 
 vowels = 'aeiouyϾ'
 consonants = "bcçdfghjklmnpqrstvwxzñ'"
+apostrophes = "'’"
 legal = vowels + consonants + ' -'
 
 # a variant of x-sampa such that all french phonemes are one-character
@@ -44,8 +45,8 @@ def norm_spaces(text):
 
 def rm_punct(text, rm_all=False, rm_apostrophe=False):
   """Remove punctuation from text"""
-  text = re.sub("’", "'", text) # no weird apostrophes
-  text = re.sub("' ", "'", text) # space after apostrophes
+  text = re.sub("[" + apostrophes + "]", "'", text) # no weird apostrophes
+  text = re.sub("' *", "'", text) # space after apostrophes
   if rm_apostrophe:
     text = re.sub("'", "", text)
   text = re.sub("'*$", "", text) # apostrophes at end of line
diff --git a/verse.py b/verse.py
@@ -1,7 +1,7 @@
 #!/usr/bin/python3
 
 import common
-from common import consonants, normalize, is_consonants, is_vowels, sure_end_fem, strip_accents_one
+from common import apostrophes, consonants, normalize, is_consonants, is_vowels, sure_end_fem, strip_accents_one
 import re
 import vowels
 import haspirater
@@ -66,6 +66,19 @@ class Verse:
     self.chunks = [[{'original': y, 'text': normalize(y, rm_apostrophe=True)}
       for y in x] for x in pre_chunks]
 
+    # collapse apostrophes
+    self.chunks2 = []
+    acc = []
+    for w in self.chunks:
+      if re.search("[" + apostrophes + "]\s*$", w[-1]['original']):
+        acc += w
+      else:
+        self.chunks2.append(acc + w)
+        acc = []
+    if len(acc) > 0:
+      self.chunks2.append(acc)
+    self.chunks = self.chunks2
+
     # check forbidden characters
     for w in self.chunks:
       for y in w: