make sure that content is defined before testing for search patterns

[mediawiki_dump_tools.git] / wikiq

diff --git a/wikiq b/wikiq
index 0543a3383b57ae27281882885bad08978c6de5f9..2c1ef3857e9cd5c02e09000fae3ab9125853cf64 100755 (executable)
--- a/wikiq
+++ b/wikiq
@@ -146,7 +146,7 @@ class RegexPair(object):
         if self.has_groups:
 
             # if there are matches of some sort in this revision content, fill the lists for each cap_group
-            if self.pattern.search(content) is not None:
+            if content is not None and self.pattern.search(content) is not None:
                 m = self.pattern.finditer(content)
                 matchobjects = list(m)
 
@@ -174,7 +174,7 @@ class RegexPair(object):
         # there are no capture groups, we just search for all the matches of the regex
         else:
             #given that there are matches to be made
-            if self.pattern.search(content) is not None:
+            if content is not None and self.pattern.search(content) is not None:
                 m = self.pattern.findall(content)
                 temp_dict[self.label] = ', '.join(m)
             else:
@@ -367,6 +367,8 @@ class WikiqParser():
                     rev_data['collapsed_revs'] = rev.collapsed_revs
 
                 if self.persist != PersistMethod.none:
+                    # initialize an empty dictionary before assigning things into it. this catches bugs if the first revision is deleted                    
+                    old_rev_data = {}
                     if rev.deleted.text:
                         for k in ["token_revs", "tokens_added", "tokens_removed", "tokens_window"]:
                             old_rev_data[k] = None