Optionally produce html as output

--HG--
extra : convert_revision : svn%3Aacbfec75-9323-0410-a652-858a13e371e0/trunk%40658

parse.py

@@ -11,11 +11,11 @@
 
 #RELEASE remove
 from src import html5parser, liberalxmlparser
-from src import treebuilders
+from src import treebuilders, serializer, treewalkers
 #END RELEASE
 #RELEASE add
 #from html5lib import html5parser, liberalxmlparser
-#from html5lib import treebuilders
+#from html5lib import treebuilders, serializer, treewalkers
 #END RELEASE
 
 def convertTreeDump(treedump):
@@ -51,16 +51,7 @@ def parse():
  sys.stderr.write("No filename provided. Use -h for help\n")
  sys.exit(1)
 
- if opts.treebuilder is not None:
- try:
- treebuilder = eval("treebuilders." + opts.treebuilder).TreeBuilder
- except ImportError, name:
- sys.stderr.write("Treebuilder %s not found\n"%name)
- raise
- except Exception, foo:
- treebuilder = treebuilders.simpletree.TreeBuilder
- else:
- treebuilder = treebuilders.simpletree.TreeBuilder
+ treebuilder = treebuilders.getTreeBuilder(opts.treebuilder)
 
  if opts.xml:
  p = liberalxmlparser.XHTMLParser(tree=treebuilder)
@@ -96,6 +87,10 @@ def printOutput(parser, document, opts):
  if not opts.no_tree:
  if opts.xml:
  sys.stdout.write(document.toxml("utf-8"))
+ elif opts.html:
+ tokens = treewalkers.getTreeWalker(opts.treebuilder)(document)
+ for text in serializer.HTMLSerializer().serialize(tokens):
+ sys.stdout.write(text.encode('utf-8'))
  elif opts.hilite:
  sys.stdout.write(document.hilite("utf-8"))
  else:
@@ -121,14 +116,17 @@ def getOptParser():
  dest="no_tree", help="Do not print output tree")
 
  parser.add_option("-b", "--treebuilder", action="store", type="string",
- dest="treebuilder")
+ dest="treebuilder", default="simpleTree")
 
  parser.add_option("-e", "--error", action="store_true", default=False,
  dest="error", help="Print a list of parse errors")
 
  parser.add_option("-x", "--xml", action="store_true", default=False,
  dest="xml", help="Output as xml")
 
+ parser.add_option("", "--html", action="store_true", default=False,
+ dest="html", help="Output as html")
+
  parser.add_option("", "--hilite", action="store_true", default=False,
  dest="hilite", help="Output as formatted highlighted code.")
 

src/serializer.py

@@ -249,6 +249,7 @@ def __init__(self, **kwargs):
  if attr in kwargs:
  setattr(self, attr, kwargs[attr])
  self.errors = []
+ self.strict = False
 
  def serialize(self, treewalker, encoding=None):
  in_cdata = False

src/treewalkers/_base.py

@@ -95,7 +95,7 @@ def getNextSibling(self, node):
  def getParentNode(self, node):
  raise NotImplementedError
 
- def walk(self):
+ def __iter__(self):
  currentNode = self.tree
  while currentNode is not None:
  details = self.getNodeDetails(currentNode)