Fixed incorrect encoding of symbols & handle directory walking myself.

2 files changed, 19 insertions, 10 deletions

diff --git a/ooo-help-parser.py b/ooo-help-parser.py
index d91f08c..c96dee0 100755
--- a/ooo-help-parser.py
+++ b/ooo-help-parser.py
@@ -1,6 +1,6 @@
 #!/usr/bin/env python
 
-import sys, os.path, optparse
+import sys, os, os.path, optparse
 sys.path.append(sys.path[0]+"/source")
 
 import globals, expatimpl
@@ -14,6 +14,18 @@ def main ():
         parser.print_help()
         sys.exit(1)
 
+    filepaths = []
+    for fpath in args:
+        if os.path.isdir(fpath):
+            # get all .xhp files recursively under this directory.
+            for root, dirs, files in os.walk(fpath):
+                for file in files:
+                    if os.path.splitext(file)[1] == '.xhp':
+                        filepaths.append(root + '/' + file)
+        elif os.path.isfile(fpath):
+            if os.path.splitext(fpath)[1] == '.xhp':
+                filepaths.append(os.path.abspath(fpath))
+
     fd = sys.stdout
     if options.output != None:
         if os.path.isdir(options.output):
@@ -22,11 +34,7 @@ def main ():
         fd = open(options.output, 'w')
 
     filesParsed = 0
-    for fpath in args:
-        if not os.path.isfile(fpath):
-            globals.error(fpath + " is not a valid file.  Skipping.")
-            continue
-
+    for fpath in filepaths:
 #       globals.info("processing " + fpath)
         file = open(fpath, 'r')
         strm = file.read()
diff --git a/source/expatimpl.py b/source/expatimpl.py
index 5d10569..295d88a 100644
--- a/source/expatimpl.py
+++ b/source/expatimpl.py
@@ -121,24 +121,25 @@ class Parser:
         if node.nodeType == NodeType.Node:
             hasChildren = len(node.children) > 0
 
-            line = "<%s"%node.name
+            # We add '<' and '>' (or '/>') after the element content gets 
+            # encoded.
+            line = node.name
             if len(node.attrs) > 0:
                 keys = node.attrs.keys()
                 keys.sort()
                 for key in keys:
                     line += " " + key + '="' + node.attrs[key] + '"'
             if hasChildren and not nodesToSkip.has_key(node.name):
-                line += ">\n"
                 line = encodeString(line)
+                line = "<%s>\n"%line
                 fd.write (indent + line)
                 for child in node.children:
                     self.printNode(fd, child, level+1)
                 line = "</%s>\n"%node.name
-                line = encodeString(line)
                 fd.write (indent + line)
             else:
-                line += "/>\n"
                 line = encodeString(line)
+                line = "<%s/>\n"%line
                 fd.write (indent + line)
 
         elif node.nodeType == NodeType.Content: