# HG changeset patch
# User paulb
# Date 1123097701 0
# Node ID 0ee67e24c3832b44f32b253b47718361b25ce279
# Parent  d527a2402d64dab4f8348095fddd4725783f9861
[project @ 2005-08-03 19:34:59 by paulb]
Added tentative HTML parsing support.
Introduced various libxml2 options to explicitly prevent things like
network access and output of errors/warnings.

diff -r d527a2402d64 -r 0ee67e24c383 libxml2dom/__init__.py
--- a/libxml2dom/__init__.py	Fri Jul 22 22:52:40 2005 +0000
+++ b/libxml2dom/__init__.py	Wed Aug 03 19:35:01 2005 +0000
@@ -360,21 +360,21 @@
 def createDocument(namespaceURI, localName, doctype):
     return Document(Node_createDocument(namespaceURI, localName, doctype))
 
-def parse(stream_or_string):
+def parse(stream_or_string, html=0):
     if hasattr(stream_or_string, "read"):
         stream = stream_or_string
-        return parseString(stream.read())
+        return parseString(stream.read(), html)
     else:
-        return parseFile(stream_or_string)
+        return parseFile(stream_or_string, html)
 
-def parseFile(s):
-    return Document(Node_parseFile(s))
+def parseFile(s, html=0):
+    return Document(Node_parseFile(s, html))
 
-def parseString(s):
-    return Document(Node_parseString(s))
+def parseString(s, html=0):
+    return Document(Node_parseString(s, html))
 
-def parseURI(uri):
-    return Document(Node_parseURI(uri))
+def parseURI(uri, html=0):
+    return Document(Node_parseURI(uri, html))
 
 def toString(node, encoding=None):
     return Node_toString(node.as_native_node(), encoding)
diff -r d527a2402d64 -r 0ee67e24c383 libxml2dom/macrolib/macrolib.py
--- a/libxml2dom/macrolib/macrolib.py	Fri Jul 22 22:52:40 2005 +0000
+++ b/libxml2dom/macrolib/macrolib.py	Wed Aug 03 19:35:01 2005 +0000
@@ -334,35 +334,48 @@
         Node_appendChild(d, root)
     return d
 
-def parse(stream_or_string):
+def parse(stream_or_string, html=0):
     if hasattr(stream_or_string, "read"):
         stream = stream_or_string
-        return parseString(stream.read())
+        return parseString(stream.read(), html)
     else:
-        return parseFile(stream_or_string)
+        return parseFile(stream_or_string, html)
 
-def parseFile(s):
+def parseFile(s, html=0):
     # NOTE: Switching off validation and remote DTD resolution.
-    context = libxml2mod.xmlCreateFileParserCtxt(s)
-    libxml2mod.xmlParserSetValidate(context, 0)
-    libxml2mod.xmlCtxtUseOptions(context, 0)
-    libxml2mod.xmlParseDocument(context)
-    return libxml2mod.xmlParserGetDoc(context)
+    if not html:
+        context = libxml2mod.xmlCreateFileParserCtxt(s)
+        libxml2mod.xmlParserSetValidate(context, 0)
+        libxml2mod.xmlCtxtUseOptions(context, XML_PARSE_NOERROR | XML_PARSE_NOWARNING | XML_PARSE_NONET)
+        libxml2mod.xmlParseDocument(context)
+        return libxml2mod.xmlParserGetDoc(context)
+    else:
+        return libxml2mod.htmlReadFile(s, None, HTML_PARSE_NOERROR | HTML_PARSE_NOWARNING | HTML_PARSE_NONET)
 
-def parseString(s):
+def parseString(s, html=0):
     # NOTE: Switching off validation and remote DTD resolution.
-    context = libxml2mod.xmlCreateMemoryParserCtxt(s, len(s))
-    libxml2mod.xmlParserSetValidate(context, 0)
-    libxml2mod.xmlCtxtUseOptions(context, 0)
-    libxml2mod.xmlParseDocument(context)
-    return libxml2mod.xmlParserGetDoc(context)
+    if not html:
+        context = libxml2mod.xmlCreateMemoryParserCtxt(s, len(s))
+        libxml2mod.xmlParserSetValidate(context, 0)
+        libxml2mod.xmlCtxtUseOptions(context, XML_PARSE_NOERROR | XML_PARSE_NOWARNING | XML_PARSE_NONET)
+        libxml2mod.xmlParseDocument(context)
+        return libxml2mod.xmlParserGetDoc(context)
+    else:
+        # NOTE: URL given as None.
+        html_url = None
+        return libxml2mod.htmlReadMemory(s, len(s), html_url, None,
+            HTML_PARSE_NOERROR | HTML_PARSE_NOWARNING | HTML_PARSE_NONET)
 
-def parseURI(uri):
-    context = libxml2mod.xmlCreateURLParserCtxt(url)
-    libxml2mod.xmlParserSetValidate(context, 0)
-    libxml2mod.xmlCtxtUseOptions(context, 0)
-    libxml2mod.xmlParseDocument(context)
-    return libxml2mod.xmlParserGetDoc(context)
+def parseURI(uri, html=0):
+    # NOTE: Switching off validation and remote DTD resolution.
+    if not html:
+        context = libxml2mod.xmlCreateURLParserCtxt(url)
+        libxml2mod.xmlParserSetValidate(context, 0)
+        libxml2mod.xmlCtxtUseOptions(context, XML_PARSE_NOERROR | XML_PARSE_NOWARNING | XML_PARSE_NONET)
+        libxml2mod.xmlParseDocument(context)
+        return libxml2mod.xmlParserGetDoc(context)
+    else:
+        raise NotSupportedError, "parseURI does not yet support HTML"
 
 def toString(node, encoding=None):
     return libxml2mod.serializeNode(node, encoding, 0)
@@ -373,4 +386,13 @@
 def toFile(node, f, encoding=None):
     libxml2mod.saveNodeTo(node, f, encoding, 0)
 
+# libxml2mod constants.
+
+HTML_PARSE_NOERROR = 32
+HTML_PARSE_NOWARNING = 64
+HTML_PARSE_NONET = 2048
+XML_PARSE_NOERROR = 32
+XML_PARSE_NOWARNING = 64
+XML_PARSE_NONET = 2048
+
 # vim: tabstop=4 expandtab shiftwidth=4