Finished parsing the 'tree' xml files.

The '.tree' files define the structure of the help contents.

3 files changed, 66 insertions, 3 deletions

diff --git a/ooo-help-parser.py b/ooo-help-parser.py
index 165b511..0525156 100755
--- a/ooo-help-parser.py
+++ b/ooo-help-parser.py
@@ -3,7 +3,7 @@
 import sys, os, os.path, optparse
 sys.path.append(sys.path[0]+"/source")
 
-import globals, expatimpl, docbook, node
+import globals, expatimpl, docbook, node, tree
 
 def processTreeFiles (tree_dir):
     if not os.path.isdir(tree_dir):
@@ -13,6 +13,7 @@ def processTreeFiles (tree_dir):
     if tree_dir[-1] == '/':
         tree_dir = tree_dir[:-1]
 
+    # Parse the tree files to build dom structures.
     rootNodes = {}
     for entry in os.listdir(tree_dir):
         entry = tree_dir + '/' + entry
@@ -31,7 +32,11 @@ def processTreeFiles (tree_dir):
         p = expatimpl.TreeParser(strm)
         p.parse()
         rootNodes[basename] = p.root
-        node.prettyPrint(sys.stdout, p.root)
+
+    # Build document tree.
+    builder = tree.TreeBuilder(rootNodes)
+    builder.build()
+
 
 def main ():
     parser = optparse.OptionParser()
diff --git a/source/node.py b/source/node.py
index 0296993..dc37dcb 100644
--- a/source/node.py
+++ b/source/node.py
@@ -57,10 +57,12 @@ class Content(NodeBase):
         self.content = content
 
 class Element(NodeBase):
-    def __init__ (self, name, attrs={}):
+    def __init__ (self, name, attrs=None):
         NodeBase.__init__(self, NodeType.Element)
         self.name = name
         self.attrs = attrs
+        if self.attrs == None:
+            self.attrs = {}
 
     def getContent (self):
         text = ''
@@ -76,6 +78,13 @@ class Element(NodeBase):
                 text += child.getContent()
         return text
 
+    def getAttrByName (self, name):
+        if not self.attrs.has_key(name):
+            return None
+        return self.attrs[name]
+
+    def setAttr (self, name, val):
+        self.attrs[name] = val
 
 encodeTable = {
     '>': 'gt',
diff --git a/source/tree.py b/source/tree.py
new file mode 100644
index 0000000..b859e6e
--- /dev/null
+++ b/source/tree.py
@@ -0,0 +1,49 @@
+
+import sys
+import globals, node
+
+class TreeBuilder:
+    def __init__ (self, treeroots):
+        self.treeroots = treeroots
+        self.root = node.Root()
+
+    def build (self):
+
+        for key in self.treeroots.keys():
+            treeroot = self.treeroots[key]
+            treeview = treeroot.firstChild()
+            # tree_view node has one or more help_section elements.
+            for helpsection in treeview.getChildNodes():
+                if helpsection.nodeType != node.NodeType.Element or helpsection.name != 'help_section':
+                    continue
+
+                # help_section element has a 'title' attribute.
+                title = helpsection.getAttrByName('title')
+                elem = self.root.appendElement('section')
+                elem.setAttr('title', title)
+                for child in helpsection.getChildNodes():
+                    self.__walkNode(child, elem)
+
+        node.prettyPrint(sys.stdout, self.root)
+
+    def __walkNode (self, src, dest):
+        if src.nodeType != node.NodeType.Element:
+            return
+
+        # 'topic' is a leaf node while 'node' is a non-leaf node.
+        if src.name == 'topic':
+            # the 'id' attribute contains the corresponding xhp file path, while 
+            # the content of this element is the title of this section.
+            xhppath = src.getAttrByName('id')
+            title = src.getContent()
+            elem = dest.appendElement('section')
+            elem.setAttr('title', title)
+            elem.setAttr('path', xhppath)
+        elif src.name == 'node':
+            # A 'node' element is just a branch point; it doesn't have any content.
+            title = src.getAttrByName('title')
+            elem = dest.appendElement('section')
+            elem.setAttr('title', title)
+            for child in src.getChildNodes():
+                self.__walkNode(child, elem)
+