Merge pull request #38 from keith-hall/processing-instructions

add support for processing instructions (even before the root element)

lxml_parser.py

@@ -41,9 +41,13 @@ class LocationAwareComment(etree.CommentBase):
     tag_pos = None
 
 
+class LocationAwareProcessingInstruction(etree.PIBase):
+    tag_pos = None
+
+
 # http://stackoverflow.com/questions/36246014/lxml-use-default-class-element-lookup-and-treebuilder-parser-target-at-the-sam
 class LocationAwareXMLParser:
-    RE_SPLIT_XML = re.compile('<!\[CDATA\[|\]\]>|[<>]')
+    RE_SPLIT_XML = re.compile(r'<!\[CDATA\[|\]\]>|[<>]')
 
     def __init__(self, position_offset = 0, **parser_options):
         def getLocation(index=None):
@@ -126,6 +130,8 @@ def _reset(self):
         self._most_recent = None
         self._in_tail = None
         self._all_namespaces = collections.OrderedDict()
+        self._addprevious = []
+        self._root = None
 
     def _flush(self):
         if self._text:
@@ -163,25 +169,43 @@ def element_end(self, tag, location=None):
     def text_data(self, data, location=None):
         self._text.append(data)
 
+    def pi(self, target, data, location=None):
+        self._flush()
+        self._appendNode(self.create_pi(target, data))
+        self._most_recent.tag_pos = location
+        self._in_tail = True
+
     def comment(self, text, location=None):
-        if self._most_recent is not None:
-            self._flush()
-            self._appendNode(self.create_comment(text))
-            self._most_recent.tag_pos = location
-            self._in_tail = True
+        self._flush()
+        self._appendNode(self.create_comment(text))
+        self._most_recent.tag_pos = location
+        self._in_tail = True
 
     def create_comment(self, text):
         return LocationAwareComment(text)
 
+    def create_pi(self, target, data):
+        return LocationAwareProcessingInstruction(target, data)
+
     def _appendNode(self, node):
         if self._element_stack: # if we have anything on the stack
             self._element_stack[-1].append(node) # append the node as a child to the last/top element on the stack
+        elif self._root is None and isinstance(node, etree.ElementBase):
+            self._root = node
+            for item in self._addprevious:
+                node.addprevious(item)
+        elif self._most_recent is not None and self._root is not None:
+            # after the root element
+            self._most_recent.addnext(node)
+        else:
+            # store this element to add before the root node when we encounter it
+            self._addprevious.append(node)
         self._all_elements.append(node)
         self._most_recent = node
 
     def document_end(self):
         """Return the root node and a list of all elements (and comments) found in the document, to keep their proxy alive."""
-        return (self._most_recent, self._all_namespaces, self._all_elements)
+        return (self._root, self._all_namespaces, self._all_elements)
 
 
 def lxml_etree_parse_xml_string_with_location(xml_chunks, position_offset = 0, should_stop = None):
@@ -206,7 +230,7 @@ def lxml_etree_parse_xml_string_with_location(xml_chunks, position_offset = 0, s
 def getNodeTagRange(node, position_type):
     """Given a node and position type (open or close), return the node's position."""
     pos = None
-    if isinstance(node, LocationAwareComment):
+    if isinstance(node, LocationAwareComment) or isinstance(node, LocationAwareProcessingInstruction):
         pos = node.tag_pos
     else:
         pos = getattr(node, position_type + '_tag_pos')

sublime_lxml.py

@@ -105,6 +105,8 @@ def get_nodes_from_document(nodes):
             element = node.getparent() # get the parent
             if element is None: # some nodes are not actually part of the original document we parsed, for example when using the substring function. so there is no way to find the original node, and therefore the location
                 continue
+        elif isinstance(node, etree.PIBase):
+            element = node
         elif isinstance(node, etree.CommentBase):
             element = node
         elif isinstance(node, etree.ElementBase):
@@ -154,7 +156,7 @@ def ensureTagNameEndPosIsSet(node, open_pos):
             if next_node is not None:
                 text_end_pos = getNodeTagRegion(view, next_node, 'open').begin()
             yield sublime.Region(text_begin_pos, text_end_pos)
-        elif isinstance(node, etree.CommentBase):
+        elif isinstance(node, etree.CommentBase) or isinstance(node, etree.PIBase):
             yield open_pos
         elif attr_name is None or attribute_position_type is None or attribute_position_type in ('element', 'parent'):
             # position type 'open' <|name| attr1="test"></name> "Goto name in open tag"

xpath.py

@@ -958,7 +958,7 @@ def completions_axis_specifiers():
         return [(completion + '\taxis', completion + '::') for completion in completions]
 
     def completions_node_types():
-        completions = ['text', 'node', 'comment'] # 'processing-instruction' is also a valid XPath node type, but not parsed into an ElementTree, so useless to show it in suggestions/completions
+        completions = ['text', 'node', 'comment', 'processing-instruction']
         return [(completion + '\tnode type', completion + '()') for completion in completions]
 
     def completions_functions():