WIP

aelaguiz · aelaguiz · commit 913aa1f2f544 · 2014-12-28T19:51:41.000-06:00
diff --git a/html5lib/html5parser.py b/html5lib/html5parser.py
@@ -168,6 +168,7 @@ def mainLoop(self):
         ParseErrorToken = tokenTypes["ParseError"]
 
         for token in self.normalizedTokens():
+            #log.debug(u"Token {}".format(token))
             new_token = token
             while new_token is not None:
                 currentNode = self.tree.openElements[-1] if self.tree.openElements else None
diff --git a/html5lib/tests/test_jinja.py b/html5lib/tests/test_jinja.py
@@ -0,0 +1,57 @@
+import html5lib
+import unittest
+import logging
+
+log = logging.getLogger(__name__)
+
+
+def dump(tree, tabs=0):
+    log.debug(u"{}Tag '{}' - {} children - Value = {} - Text = {}".format(
+        "".join(["\t" for i in range(tabs)]), tree.tag, len(tree), tree.attrib['value'] if 'value' in tree.attrib else None, tree.text))
+
+    for child in tree:
+        dump(child, tabs + 1)
+
+
+class JinjaTestCase(unittest.TestCase):
+    def setUp(self):
+        self.parser = html5lib.HTMLParser(strict=True, namespaceHTMLElements=False, tree=html5lib.treebuilders.getTreeBuilder("etree", fullTree=True))
+
+    def assertTree(self, root, spec):
+        self.assertEqual(len(root), len(spec))
+
+        for child, spec_child in zip(root, spec):
+            self.assertEqual(child.tag, spec_child['tag'])
+
+            if 'text' in spec_child:
+                self.assertEqual(child.text, spec_child['text'])
+
+            if 'value' in spec_child:
+                self.assertEqual(child.attrib['value'], spec_child['value'])
+
+            if 'children' in spec_child:
+                self.assertTree(child, spec_child['children'])
+            else:
+                self.assertEqual(len(child), 0)
+
+            if 'attrs' in spec_child:
+                for k, v in spec_child['attrs'].iteritems():
+                    self.assertIn(k, child.attrib)
+                    self.assertEqual(v, child.attrib[k])
+
+    def test_open_block(self):
+        html_string = """
+            <div class="basket-site dropdown-menu" role="menu">
+                {% set cart = active_cart() %}
+                {% for cart_product in cart.products %}
+                {% endfor %}
+
+                <div class="controls clearfix">
+            {#
+                    <button type="button" class="btn btn-link">VIEW BASKET</button>
+            #}
+                </div>
+            </div>
+        """
+        tree = self.parser.parseFragment(html_string)
+        dump(tree)
diff --git a/html5lib/tokenizer.py b/html5lib/tokenizer.py
@@ -311,6 +311,14 @@ def jinjaOpenState(self):
             #})
 
             self.state = self.jinjaStatementState
+        elif data == "#":
+            #self.tokenQueue.append({
+                #"type": tokenTypes["JinjaStatementStartTag"],
+                #"name": "{%", "data": {},
+                #"selfClosing": False
+            #})
+
+            self.state = self.jinjaCommentState
 
         #self.state = self.dataState
         return True
@@ -359,6 +367,28 @@ def jinjaVariableEndState(self):
         #self.state = self.dataState
         return True
 
+    def jinjaCommentEndState(self):
+        # We got a {
+        data = self.stream.char()
+
+        if data == "}":
+            #self.tokenQueue.append({
+                #"type": tokenTypes["JinjaVariableEndTag"],
+                #"name": "}}", "data": [],
+                #"selfClosing": False
+            #})
+            self.state = self.dataState
+        elif data is EOF:
+            self.tokenQueue.append({"type": tokenTypes["ParseError"], "data":
+                                    "expected-jinja-comment-closing-tag-but-got-eof",
+                                    "datavars": {"data": data}})
+            self.state = self.dataState
+        else:
+            self.state = self.jinjaStatementState
+
+        #self.state = self.dataState
+        return True
+
     def jinjaStatementState(self):
         data = self.stream.char()
 
@@ -375,6 +405,24 @@ def jinjaStatementState(self):
 
         return True
 
+    def jinjaCommentState(self):
+        data = self.stream.char()
+
+        log.debug(u"Jinja comment state '{}'".format(data))
+
+        if data == "#":
+            self.state = self.jinjaCommentEndState
+        elif data is EOF:
+            self.tokenQueue.append({"type": tokenTypes["ParseError"], "data":
+                                    "missing-jinja-comment-closing-brace"})
+            self.state = self.dataState
+        else:
+            chars = self.stream.charsUntil(("#", "\u0000"))
+            #self.tokenQueue.append({"type": tokenTypes["JinjaStatementTag"], "data":
+                                    #data + chars})
+
+        return True
+
     def jinjaVariableState(self):
         data = self.stream.char()
 
