Traitement correct des espaces insécables

songbook/plastex-songs.py

@@ -3,6 +3,8 @@
 
 import plasTeX
 
+from songbook.plastex import processUnbreakableSpace
+
 def split_linebreak(texlist):
     return_list = []
     current = []
@@ -11,7 +13,7 @@ def split_linebreak(texlist):
             return_list.append(current)
             current = []
         else:
-            current.append(token.textContent.encode('utf-8'))
+            current.append(processUnbreakableSpace(token).textContent.encode('utf-8'))
     if current:
         return_list.append(current)
     return return_list
@@ -31,7 +33,7 @@ class beginsong(plasTeX.Command):
         args = {}
         for (key, val) in self.attributes['args'].iteritems():
             if isinstance(val, plasTeX.DOM.Element):
-                args[key] = val.textContent.encode('utf-8')
+                args[key] = processUnbreakableSpace(val).textContent.encode('utf-8')
             elif isinstance(val, unicode):
                 args[key] = val.encode('utf-8')
             elif isinstance(val, str):

songbook/plastex.py

@@ -2,12 +2,25 @@
 # -*- coding: utf-8 -*-
 
 from plasTeX.TeX import TeX
+from plasTeX.Base.LaTeX import Sentences
+
 import codecs
 import copy
 import locale
 import os
 import sys
 
+def processUnbreakableSpace(node):
+    """Replace '~' and '\ ' in node by nodes that will be rendered as unbreakable space.
+
+    Return node object for convenience.
+    """
+    if type(node) == Sentences.InterWordSpace or (type(node) == Sentences.NoLineBreak and node.source == '~ '):
+        node.unicode = unichr(160)
+    for child in node.childNodes:
+        processUnbreakableSpace(child)
+
+    return node
 
 def simpleparse(text):
     """Parse a simple LaTeX string.
@@ -15,7 +28,7 @@ def simpleparse(text):
     tex = TeX()
     tex.input(text.decode('utf8'))
     doc = tex.parse()
-    return doc.textContent
+    return processUnbreakableSpace(doc.textContent)
 
 class SongParser:
     """Analyseur syntaxique de fichiers .sg"""