Read posts with unicode instead binary

dynastie/generators/index.py

@@ -25,6 +25,7 @@ import xml
 from xml.parsers.expat import *
 import xml.parsers.expat
 from xml.dom.minidom import parse, parseString
+import codecs
 from dynastie.generators.generator import DynastieGenerator, StrictUTF8Writer
 from django.db import models
 from dynastie.generators import markdown2
@@ -202,7 +203,10 @@ class Index(DynastieGenerator):
                 break
 
             try:
-                dom = parseString(code[start:end+11])
+                try:
+                    dom = parseString(code[start:end+11])
+                except UnicodeEncodeError:
+                    dom = parseString(code[start:end+11].encode('utf-8'))
             except xml.dom.DOMException as e:
                 self.addError('Error parsing ' + self.filename)
                 break
@@ -210,7 +214,6 @@ class Index(DynastieGenerator):
             res = self.createCode(dom, dom.firstChild)
             if res:
                 code = code.replace(code[start:end+11], res)
-
         return code
 
     def _have_I_right(self, user, post_id):
@@ -246,7 +249,8 @@ class Index(DynastieGenerator):
                     if not post: continue
                     new_content = self._loadPostContent(post)
                     if new_content:
-                        text = text.replace('[[' + str(post_id) + ']]', new_content)
+                        p = '[[' + str(post_id) + ']]'
+                        text = text.replace(p, new_content)
             if internal_posts: return text
         # HTML replace
         if not post or (post and post.content_format == Post.CONTENT_HTML):
@@ -267,7 +271,7 @@ class Index(DynastieGenerator):
                     if not post: break
                     new_content = self._loadPostContent(post)
                     if new_content:
-                        text = text.replace(text[start:end+18], new_content.encode('utf-8'))
+                        text = text.replace(text[start:end+18], new_content)
         return text
 
     def _loadPostContent(self, post):
@@ -287,7 +291,7 @@ class Index(DynastieGenerator):
                 filename = filename2
 
         if not filename in self.hash_posts_content:
-            f = open(filename, 'rb')
+            f = codecs.open(filename, 'rb', 'utf-8')
             post_content = f.read()
             f.close()
             self.parent_posts.append(post.id)
@@ -498,7 +502,7 @@ class Index(DynastieGenerator):
 
         # Remove <pre> after <div class="highlight">
         code = code[28:-13]
-        code = '<div class="highlight">' + code + '</div>'
+        code = u'<div class="highlight">' + unicode(code, 'utf-8') + u'</div>'
 
         return code
 

dynastie/sites/blog.soutade.fr/_ljdc.xml

@@ -1,5 +1,23 @@
 <?xml version="1.0" encoding="utf-8"?>
 <ljdc>
+    <entry>
+        <id>138932499327</id>
+        <address>http://thecodinglove.com/post/138932499327/seeing-my-own-code-after-years</address>
+        <title>Seeing my own code after years</title>
+        <img>http://tclhost.com/nDmdJau.gif</img>
+    </entry>
+    <entry>
+        <id>137804239758</id>
+        <address>http://thecodinglove.com/post/137804239758/when-you-discover-a-serious-bug-during-a-client</address>
+        <title>When you discover a serious bug during a client...</title>
+        <img>http://tclhost.com/rzUXIEl.gif</img>
+    </entry>
+    <entry>
+        <id>137344478111</id>
+        <address>http://thecodinglove.com/post/137344478111/arriving-late-to-an-important-meeting</address>
+        <title>Arriving late to an important meeting</title>
+        <img>http://tclhost.com/CXW1sJ4.gif</img>
+    </entry>
     <entry>
         <id>132019829293</id>
         <address>http://thecodinglove.com/post/132019829293/when-a-mad-boss-looking-for-me-enters-the-open</address>