escape_lt_in_attrs serialization option

rubys · rubys · commit ddd8755afeb8 · 2007-06-21T01:03:47.000Z
--HG--
extra : convert_revision : svn%3Aacbfec75-9323-0410-a652-858a13e371e0/trunk%40788
diff --git a/parse.py b/parse.py
@@ -97,7 +97,7 @@ def printOutput(parser, document, opts):
         for opt in ['inject_meta_charset', 'strip_whitespace', 'sanitize',
                     'omit_optional_tags', 'quote_attr_values', 'quote_char',
                     'use_best_quote_char', 'minimize_boolean_attributes',
-                    'use_trailing_solidus']:
+                    'use_trailing_solidus', 'escape_lt_in_attrs']:
             kwargs[opt] = getattr(opts,opt)
         if not kwargs['quote_char']: del kwargs['quote_char']
         tokens = treewalkers.getTreeWalker(opts.treebuilder)(document)
@@ -178,6 +178,10 @@ def getOptParser():
                       default=False, dest="use_trailing_solidus",
                       help="use trailing solidus")
 
+    parser.add_option("", "--escape_lt_in_attrs", action="store_true",
+                      default=False, dest="escape_lt_in_attrs",
+                      help="escape less than signs in attribute values")
+
     parser.add_option("", "--sanitize", action="store_true", default=False,
                       dest="sanitize", help="sanitize")
 
diff --git a/src/serializer.py b/src/serializer.py
@@ -64,6 +64,7 @@ class HTMLSerializer(object):
 
     use_trailing_solidus = False
     space_before_trailing_solidus = True
+    escape_lt_in_attrs = False
 
     omit_optional_tags = True
 
@@ -77,7 +78,7 @@ def __init__(self, **kwargs):
         for attr in ("quote_attr_values", "quote_char", "use_best_quote_char",
           "minimize_boolean_attributes", "use_trailing_solidus",
           "space_before_trailing_solidus", "omit_optional_tags",
-          "strip_whitespace", "inject_meta_charset"):
+          "strip_whitespace", "inject_meta_charset", "escape_lt_in_attrs"):
             setattr(self, attr, kwargs.get(attr, getattr(self, attr)))
         self.errors = []
         self.strict = False
@@ -142,6 +143,7 @@ def serialize(self, treewalker, encoding=None):
                             quote_attr = reduce(lambda x,y: x or (y in v),
                                 spaceCharacters + "<>\"'", False)
                         v = v.replace("&", "&amp;")
+                        if self.escape_lt_in_attrs: v = v.replace("<", "&lt;")
                         if encoding:
                             v = encode(v, encoding)
                         if quote_attr:
