Changeset 3478 in products for quintagroup.plonegooglesitemaps/branches/two_request/utils.py

Timestamp:

Jul 27, 2012 2:35:19 PM (12 years ago)

Author:

potar

Message:

added urlFilter; rewrote code

File:

• quintagroup.plonegooglesitemaps/branches/two_request/utils.py (copied) (copied from quintagroup.plonegooglesitemaps/branches/1.7.1/quintagroup/plonegooglesitemaps/browser/utils.py) (2 diffs)

quintagroup.plonegooglesitemaps/branches/two_request/utils.py

@@ -11 +11 @@
 _marker = []
 
-
+    
 def searchAndReplace(string, what, withs):
     """Emulate sed command s/"""
+    #TODO: Better solution would be to compile
+    #'what' and 'withs'
     res = re.sub(what, withs, string)
     return res
 OPERATORS = {'s': searchAndReplace, }
 
+def urlFilter(url, operations):
+    """ Method dedicated to filter url according with operations """
+    for operator, what, withs in operations:
+        url = OPERATORS[operator](url, what, withs.replace("\\", ""))
+    return url 
 
-def applyOperations(objects, operations):
-    """Parse Operations """
-    operations = [OPERATIONS_PARSE.match(op).groups() for op in operations]
-    result = {}
+def getUrlsObjects(objects):
+    """Method dedicated to get url for each object"""
+    results = set()
     for ob in objects:
         url = _marker
@@ -29 +35 @@
         if url in [Missing_Value, _marker]:
             url = ICanonicalLink(ob.getObject()).canonical_link
-        for operator, what, withs in operations:
-            url = OPERATORS[operator](url, what, withs.replace("\\", ""))
         #TODO: Remove or replace following condition
         #it is senseless in the case we need intelligent
         #result set. Better condition would be to place
         #freshest brain into result
-        if url in result.keys():
+        if url in results:
             continue
         #TODO: replace brain with only data necessary to
         #generate sitemap
+        results.add(url)
+        yield url, ob
+
+def applyOperations(objects, operations):
+    """Parse Operations"""
+    operations = [OPERATIONS_PARSE.match(op).groups() for op in operations]
+    result = {}
+    for url, ob in getUrlsObjects(objects):
+        url = urlFilter(url, operations)
         result[url] = ob
     return result
-
 
 def additionalURLs(context):