Unicode handling of URLs fix

[rss2maildir.git] / rss2maildir.py

diff --git a/rss2maildir.py b/rss2maildir.py
index 353787c178dc4610e71495edea31281ece9f0425..2af32bc9105a49201e2bbba49d87b5e2cbcda1da 100755 (executable)
--- a/rss2maildir.py
+++ b/rss2maildir.py
@@ -151,7 +151,7 @@ class HTML2Text(HTMLParser):
             elif tag_name == u'a':
                 for attr in attrs:
                     if attr[0].lower() == u'href':
-                        self.urls.append(attr[1])
+                        self.urls.append(attr[1].decode('utf-8'))
                 self.curdata = self.curdata + u'`'
                 self.opentags.append(tag_name)
                 return
@@ -235,8 +235,7 @@ class HTML2Text(HTMLParser):
         if tag_thats_done in ["h1", "h2", "h3", "h4", "h5", "h6"]:
             underline = u''
             underlinechar = u'='
-            headingtext = unicode( \
-                self.curdata.encode("utf-8").strip(), "utf-8")
+            headingtext = self.curdata
             seperator = u'\n' + u' '*self.indentlevel
             headingtext = seperator.join( \
                 textwrap.wrap( \
@@ -257,7 +256,7 @@ class HTML2Text(HTMLParser):
                 underline = u' ' * self.indentlevel \
                     + underlinechar * len(headingtext)
             self.text = self.text \
-                + headingtext.encode("utf-8") + u'\n' \
+                + headingtext + u'\n' \
                 + underline
         elif tag_thats_done in [u'p', u'div']:
             paragraph = unicode( \
@@ -414,7 +413,7 @@ class HTML2Text(HTMLParser):
     def handle_data(self, data):
         if len(self.opentags) == 0:
             self.opentags.append(u'p')
-        self.curdata = self.curdata + unicode(data, "utf-8")
+        self.curdata = self.curdata + data.decode("utf-8")
 
     def handle_entityref(self, name):
         entity = name
@@ -724,11 +723,13 @@ if __name__ == "__main__":
     elif scp.has_option("general", "state_dir"):
         new_state_dir = scp.get("general", "state_dir")
         try:
-            mode = os.stat(state_dir)[stat.ST_MODE]
+            mode = os.stat(new_state_dir)[stat.ST_MODE]
             if not stat.S_ISDIR(mode):
                 sys.stderr.write( \
                     "State directory (%s) is not a directory\n" %(state_dir))
                 sys.exit(1)
+            else:
+                state_dir = new_state_dir
         except:
             # try to create it
             try: