Include git search instead of replace on RowTransformations as well.

src/main/python/ExtractDomain.py

@@ -16,21 +16,18 @@
 from urllib.parse import urlparse
 
 def ExtractDomain(url, source = ""):
-  if url is None: 
+  if url is None:
     return None
   host = None
   try:
     host = urlparse(url).hostname
   except Exception as e:
-    # it's okay 
-    pass 
-  if (host is not None or source == ""): 
+    # it's okay
+    pass
+  if (host is not None or source == ""):
     return host
   try:
     host = urlparse(source).hostname
     return host
   except Exception as e:
     return None
-
-
-

src/main/python/RowTransformations.py

@@ -54,7 +54,7 @@ def keepUrls(row, urls):
 def keepUrlPatterns(row, urlREs):
   url = row['url']
   for pattern in urlREs:
-    if re.match(pattern, url) is not None:
+    if re.search(pattern, url) is not None:
       return True
   return False
 
@@ -67,7 +67,7 @@ def keepLanguages(row, langs):
 def keepContent(row, contentREs):
   contents = row['contentString']
   for pattern in contentREs:
-    if re.match(pattern, contents) is not None:
+    if re.search(pattern, contents) is not None:
       return True
   return False