Rewrote the GenshiAdapter to transform tokens (provided by the SimpleTree treewalker) into Genshi's Stream events (which will be transformed back into tokens by the Genshi treewalker)

tbroyer · tbroyer · commit 8c9eb12a506d · 2007-06-05T22:26:04.000Z
--HG--
extra : convert_revision : svn%3Aacbfec75-9323-0410-a652-858a13e371e0/trunk%40667
diff --git a/src/treewalkers/genshistream.py b/src/treewalkers/genshistream.py
@@ -25,11 +25,12 @@ def __iter__(self):
                 if previous[0] == END:
                     depth -= 1
             previous = event
-        if ignore_until is None or ignore_until <= depth:
-            for token in self.tokens(previous, None):
-                yield token
-        elif ignore_until is not None:
-            raise ValueError("Illformed DOM event stream: void element without END_ELEMENT")
+        if previous is not None:
+            if ignore_until is None or ignore_until <= depth:
+                for token in self.tokens(previous, None):
+                    yield token
+            elif ignore_until is not None:
+                raise ValueError("Illformed DOM event stream: void element without END_ELEMENT")
 
     def tokens(self, event, next):
         kind, data, pos = event
diff --git a/tests/test_treewalkers.py b/tests/test_treewalkers.py
@@ -110,50 +110,52 @@ def PullDOMAdapter(node):
 except ImportError:
     pass
 
-if "ElementTree" in locals():
-    try:
-        from genshi.core import Attrs, QName
-        from genshi.core import START, END, DOCTYPE, TEXT, COMMENT
-
-        def GenshiAdapter(tree):
-            if not(hasattr(tree, "tag")):
-                tree = tree.getroot()
-    
-            if tree.tag in ("<DOCUMENT_ROOT>", "<DOCUMENT_FRAGMENT>"):
-                if tree.text:
-                    yield TEXT, tree.text, (None, -1, -1)
-                for child in tree.getchildren():
-                    for item in GenshiAdapter(child):
-                        yield item
-
-            elif tree.tag == "<!DOCTYPE>":
-                yield DOCTYPE, QName(tree.text), (None, -1, -1)
-
-            elif type(tree.tag) == type(ElementTree.Comment):
-                yield COMMENT, tree.text, (None, -1, -1)
+try:
+    from genshi.core import QName, Attrs
+    from genshi.core import START, END, TEXT, COMMENT, DOCTYPE
+
+    def GenshiAdapter(tree):
+        text = None
+        for token in treewalkers.getTreeWalker("simpletree")(tree):
+            type = token["type"]
+            if type in ("Characters", "SpaceCharacters"):
+                if text is None:
+                    text = token["data"]
+                else:
+                    text += token["data"]
+            elif text is not None:
+                yield TEXT, text, (None, -1, -1)
+                text = None
+
+            if type in ("StartTag", "EmptyTag"):
+                yield (START,
+                       (QName(token["name"]),
+                        Attrs([(QName(attr),value) for attr,value in token["data"]])),
+                       (None, -1, -1))
+                if type == "EmptyTag":
+                    type = "EndTag"
+
+            if type == "EndTag":
+                yield END, QName(token["name"]), (None, -1, -1)
+
+            elif type == "Comment":
+                yield COMMENT, token["data"], (None, -1, -1)
+
+            elif type == "Doctype":
+                yield DOCTYPE, token["name"], (None, -1, -1)
 
             else:
-                tag_name = QName(tree.tag.lstrip('{'))
-                attrs = Attrs([(QName(attr.lstrip('{')), value)
-                               for attr, value in tree.items()])
-    
-                yield START, (tag_name, attrs), (None, -1, -1)
-                if tree.text:
-                    yield TEXT, tree.text, (None, -1, -1)
-                for child in tree.getchildren():
-                    for item in GenshiAdapter(child):
-                        yield item
-                yield END, tag_name, (None, -1, -1)
-
-            if tree.tail:
-                yield TEXT, tree.tail, (None, -1, -1)
-
-        treeTypes["genshi"] = \
-            {"builder": treebuilders.getTreeBuilder("etree", ElementTree),
-             "adapter": GenshiAdapter,
-             "walker":  treewalkers.getTreeWalker("genshi")}
-    except ImportError:
-        pass
+                pass # FIXME: What to do?
+
+        if text is not None:
+            yield TEXT, text, (None, -1, -1)
+
+    treeTypes["genshi"] = \
+        {"builder": treebuilders.getTreeBuilder("simpletree"),
+         "adapter": GenshiAdapter,
+         "walker":  treewalkers.getTreeWalker("genshi")}
+except ImportError:
+    pass
 
 def concatenateCharacterTokens(tokens):
     charactersToken = None
