added regex to replace "lj user" tags with normal links.

links are set to class='lj-user' too allow for potential css manipulation in
the future

Signed-off-by: Greg Hewgill <greg@hewgill.com>

diff --git a/convertdump.py b/convertdump.py
index f098a58f2a2e0b29fd0fbaca86ceb3a5a7e75509..937403b43bf1b137fa8ecca42e795298613efd2f 100755 (executable)
--- a/convertdump.py
+++ b/convertdump.py
@@ -30,6 +30,7 @@ import os
 import codecs
 import sys
 import getopt
+import re
 
 from time import strptime, strftime
 
@@ -85,7 +86,9 @@ def addEntryForId(outDoc, element, username, id, includeSecure):
     # Create an event node (special case because for some reason there are two
     # 'event' elements in the pydump output, which is probably LJ's fault)
     event = inDoc.getElementsByTagName("event")[0]
-    appendTextNode(outDoc, entry, "event", getNodeText(event, "event"))
+    eventText = getNodeText(event, "event")
+
+    appendTextNode(outDoc, entry, "event", replaceLJTags(eventText))
 
     security = getNodeText(inDoc, "security")
 
@@ -152,8 +155,8 @@ def addCommentsForId(outDoc, entry, username, id):
             getNodeText(comment, "subject"))
 
         # Create an event element
-        appendTextNode(outDoc, outComment, "event", 
-            getNodeText(comment, "body"))
+        bodyText = getNodeText(comment, "body")
+        appendTextNode(outDoc, outComment, "event", replaceLJTags(bodyText))
 
         # Create the author element
         author = outDoc.createElement("author")
@@ -172,6 +175,13 @@ def addCommentsForId(outDoc, entry, username, id):
         if(parentId != ""): 
             appendTextNode(outDoc, outComment, "parent_itemid", parentId)
 
+def replaceLJTags(entry):
+    # regex to replace <lj user="jeebus" /> tags
+    fixedUserTags = re.sub("<lj user=\"(.*?)\" ?/?>", "<a href=\"http://\\1.livejournal.com/\" class=\"lj-user\">\\1</a>", entry)
+
+    return fixedUserTags
+
+
 def usage():
     print( "Usage: convertdump.py [arguments]" )
     print( """