Changeset 385 for trunk/genshi/input.py

Timestamp:

Oct 22, 2006, 4:57:40 PM (17 years ago)

Author:

cmlenz

Message:

• The HTMLParser class and the HTML function now accept an encoding parameter to properly deal with bytestring input (defaults to UTF-8).
• The TemplateLoader class can now also be initialized from a string for the search path, for cases where the search path contains only a single directory.

File:

• trunk/genshi/input.py (modified) (4 diffs)

trunk/genshi/input.py

@@ -77 +77 @@
 
     def __init__(self, source, filename=None):
-        """Initialize the parser for the given XML text.
+        """Initialize the parser for the given XML input.
         
         @param source: the XML text as a file-like object
@@ -251 +251 @@
                               'param'])
 
-    def __init__(self, source, filename=None):
+    def __init__(self, source, filename=None, encoding='utf-8'):
+        """Initialize the parser for the given HTML input.
+        
+        @param source: the HTML text as a file-like object
+        @param filename: the name of the file, if known
+        @param filename: encoding of the file; ignored if the input is unicode
+        """
         html.HTMLParser.__init__(self)
         self.source = source
         self.filename = filename
+        self.encoding = encoding
         self._queue = []
         self._open_tags = []
@@ -322 +329 @@
 
     def handle_data(self, text):
+        if not isinstance(text, unicode):
+            text = text.decode(self.encoding, 'replace')
         self._enqueue(TEXT, text)
 
@@ -344 +353 @@
 
 
-def HTML(text):
-    return Stream(list(HTMLParser(StringIO(text))))
+def HTML(text, encoding='utf-8'):
+    return Stream(list(HTMLParser(StringIO(text), encoding=encoding)))
 
 def _coalesce(stream):