Fix XmlTagUrlParser and make Python 3 compatible

URLs within a sitemap file were not being captured.

linkcheck/parser/sitemap.py

@@ -29,7 +29,10 @@ class XmlTagUrlParser(object):
         self.tag = tag
         self.parser = ParserCreate()
         self.parser.buffer_text = True
-        self.parser.returns_unicode = True
+        try:
+            self.parser.returns_unicode = True
+        except AttributeError:
+            pass  # Python 3
         self.parser.StartElementHandler = self.start_element
         self.parser.EndElementHandler = self.end_element
         self.parser.CharacterDataHandler = self.char_data
@@ -37,7 +40,7 @@ class XmlTagUrlParser(object):
     def parse(self, url_data):
         """Parse XML URL data."""
         self.url_data = url_data
-        self.loc = False
+        self.in_tag = False
         self.url = u""
         data = url_data.get_raw_content()
         isfinal = True
@@ -45,6 +48,7 @@ class XmlTagUrlParser(object):
             self.parser.Parse(data, isfinal)
         except ExpatError as expaterr:
             self.url_data.add_warning(expaterr.message,tag=WARN_XML_PARSE_ERROR)
+
     def start_element(self, name, attrs):
         """Set tag status for start element."""
         self.in_tag = (name == self.tag)
@@ -65,7 +69,7 @@ class XmlTagUrlParser(object):
 
     def char_data(self, data):
         """If inside the wanted tag, append data to URL."""
-        if self.loc:
+        if self.in_tag:
             self.url += data
 
 
@@ -77,4 +81,3 @@ def parse_sitemap(url_data):
 def parse_sitemapindex(url_data):
     """Parse XML sitemap index data."""
     XmlTagUrlParser(u"loc").parse(url_data)
-

tests/checker/data/sitemap.xml

@@ -6,4 +6,10 @@
       <changefreq>monthly</changefreq>
       <priority>0.8</priority>
    </url>
+   <url>
+      <loc>http://www.example.com/?ascii=nø</loc>
+      <lastmod>2005-01-01</lastmod>
+      <changefreq>monthly</changefreq>
+      <priority>0.8</priority>
+   </url>
 </urlset> 

tests/checker/data/sitemap.xml.result

@@ -2,3 +2,11 @@ url http://localhost:%(port)d/%(datadir)s/sitemap.xml
 cache key http://localhost:%(port)d/%(datadir)s/sitemap.xml
 real url http://localhost:%(port)d/%(datadir)s/sitemap.xml
 valid
+url http://www.example.com/
+cache key http://www.example.com/
+real url http://www.example.com/
+valid
+url http://www.example.com/?ascii=nø
+cache key http://www.example.com/?ascii=n%%C3%%B8
+real url http://www.example.com/?ascii=n%%C3%%B8
+valid