More on parsing xhp files.

+ support writing to a file.
+ encode non-ascii characters to avoid unicode error by python
  interpreter.

3 files changed, 66 insertions, 18 deletions

diff --git a/ooo-help-parser.py b/ooo-help-parser.py
index 2a3a255..d91f08c 100755
--- a/ooo-help-parser.py
+++ b/ooo-help-parser.py
@@ -7,20 +7,40 @@ import globals, expatimpl
 
 def main ():
     parser = optparse.OptionParser()
+    parser.set_defaults(output=None)
+    parser.add_option("-o", "--output", dest="output", help="write output to FILE", metavar="FILE")
     options, args = parser.parse_args()
     if len(args) == 0:
         parser.print_help()
         sys.exit(1)
 
+    fd = sys.stdout
+    if options.output != None:
+        if os.path.isdir(options.output):
+            globals.error("cannot create output file: " + optiont.output)
+            sys.exit(1)
+        fd = open(options.output, 'w')
+
+    filesParsed = 0
     for fpath in args:
-        if not os.path.isfile(args[0]):
-            globals.error(args[0] + " is not a valid file.  Skipping.")
-    
-        file = open(args[0], 'r')
+        if not os.path.isfile(fpath):
+            globals.error(fpath + " is not a valid file.  Skipping.")
+            continue
+
+#       globals.info("processing " + fpath)
+        file = open(fpath, 'r')
         strm = file.read()
         file.close()
         p = expatimpl.Parser(strm)
         p.parse()
+        p.printSummary(fd)
+        p.prettyPrint(fd)
+        filesParsed += 1
+
+    globals.info("%d files have been processed"%filesParsed)
+    if fd != sys.stdout:
+        fd.close()
+
 
 if __name__ == '__main__':
     main()
diff --git a/source/expatimpl.py b/source/expatimpl.py
index a147162..15e40c5 100644
--- a/source/expatimpl.py
+++ b/source/expatimpl.py
@@ -1,4 +1,4 @@
-import xml.parsers.expat
+import xml.parsers.expat, sys
 import globals
 
 class NodeType:
@@ -36,6 +36,26 @@ class Node(NodeBase):
         self.name = name
         self.attrs = attrs
 
+# ============================================================================
+
+class Meta:
+    def __init__ (self):
+        self.title = None
+        self.filename = None
+
+# ============================================================================
+
+def encodeNonAscii (sin):
+    sout = ''
+    for c in sin:
+        if ord(c) >= 128:
+            sout += "\\x%2.2x"%ord(c)
+        else:
+            sout += c
+
+    return sout
+
+# ============================================================================
 
 class Parser:
 
@@ -75,16 +95,18 @@ class Parser:
         p.EndElementHandler = self.endElement
         p.CharacterDataHandler = self.character
         p.Parse(self.strm, 1)
-        self.prettyPrint()
 
-    def prettyPrint (self):
+    def printSummary (self, fd):
+        pass
+
+    def prettyPrint (self, fd):
         if len(self.root.children) != 1:
             return
 
         node = self.root.firstChild()
-        self.printNode(node, 0)
+        self.printNode(fd, node, 0)
 
-    def printNode (self, node, level):
+    def printNode (self, fd, node, level):
         singleIndent = ' '*4
         indent = singleIndent*level
         if node.nodeType == NodeType.Node:
@@ -97,18 +119,21 @@ class Parser:
                 for key in keys:
                     line += " " + key + '="' + node.attrs[key] + '"'
             if hasChildren:
-                line += ">"
-                print (indent + line)
+                line += ">\n"
+                line = encodeNonAscii(line)
+                fd.write (indent + line)
                 for child in node.children:
-                    self.printNode(child, level+1)
-                line = "</%s>"%node.name
-                print (indent + line)
+                    self.printNode(fd, child, level+1)
+                line = "</%s>\n"%node.name
+                line = encodeNonAscii(line)
+                fd.write (indent + line)
             else:
-                line += "/>"
-                print (indent + line)
-
+                line += "/>\n"
+                line = encodeNonAscii(line)
+                fd.write (indent + line)
 
         elif node.nodeType == NodeType.Content:
             if len(node.content) > 0:
-                print (indent + node.content)
+                content = encodeNonAscii(node.content)
+                fd.write (indent + content + "\n")
 
diff --git a/source/globals.py b/source/globals.py
index f7e6388..746fcba 100644
--- a/source/globals.py
+++ b/source/globals.py
@@ -4,6 +4,9 @@ import sys
 def error (msg):
     sys.stderr.write(msg + "\n")
 
+def info (msg):
+    sys.stderr.write(msg + "\n")
+
 class Exception:
     def __init__ (self, msg):
         self.msg = msg