Created a way to search all mails with no pagination for indexing

3 files changed, 29 insertions, 8 deletions

diff --git a/service/pixelated/adapter/mailbox_indexer_listener.py b/service/pixelated/adapter/mailbox_indexer_listener.py
index a7309118..d8e0f81e 100644
--- a/service/pixelated/adapter/mailbox_indexer_listener.py
+++ b/service/pixelated/adapter/mailbox_indexer_listener.py
@@ -31,7 +31,7 @@ class MailboxIndexerListener(object):
         self.querier = soledad_querier
 
     def newMessages(self, exists, recent):
-        indexed_idents = set(self.SEARCH_ENGINE.search('tag:' + self.mailbox_name.lower()))
+        indexed_idents = set(self.SEARCH_ENGINE.search('tag:' + self.mailbox_name.lower(), all_mails=True))
         soledad_idents = self.querier.idents_by_mailbox(self.mailbox_name)
 
         missing_idents = soledad_idents.difference(indexed_idents)
diff --git a/service/pixelated/adapter/search.py b/service/pixelated/adapter/search.py
index eef97c56..cfb5aaa0 100644
--- a/service/pixelated/adapter/search.py
+++ b/service/pixelated/adapter/search.py
@@ -116,19 +116,40 @@ class SearchEngine(object):
             results = searcher.search(query, **options)
         return results
 
-    def search(self, query, window=25, page=1):
+    def search(self, query, window=25, page=1, all_mails=False):
+        query = self.prepare_query(query)
+
+        if(all_mails):
+            return self._search_all_mails(query)
+        else:
+            return self._paginated_search_mails(query, window, page)
+
+
+    def _search_all_mails(self, query):
+        with self._index.searcher() as searcher:
+            results = searcher.search(query, limit=None)
+            return [mail['ident'] for mail in results]
+
+    def _paginated_search_mails(self, query, window, page):
         page = int(page) if int(page) > 1 else 1
         window = int(window)
 
-        query = query.replace('\"', '')
-        query = query.replace('-in:', 'AND NOT tag:')
-        query = query.replace('in:all', '*')
-
         with self._index.searcher() as searcher:
-            query = QueryParser('raw', self._index.schema).parse(query)
             results = searcher.search_page(query, page, pagelen=window)
             return [mail['ident'] for mail in results]
 
+
+    def prepare_query(self, query):
+        query = (
+            query
+            .replace('\"', '')
+            .replace('-in:', 'AND NOT tag:')
+            .replace('in:all', '*')
+        )
+        return QueryParser('raw', self._index.schema).parse(query)
+       
+                    
+
     def remove_from_index(self, mail_id):
         writer = self._index.writer()
         try:
diff --git a/service/test/unit/adapter/mailbox_indexer_listener_test.py b/service/test/unit/adapter/mailbox_indexer_listener_test.py
index 1f28ec7b..291739e0 100644
--- a/service/test/unit/adapter/mailbox_indexer_listener_test.py
+++ b/service/test/unit/adapter/mailbox_indexer_listener_test.py
@@ -40,7 +40,7 @@ class MailboxListenerTest(unittest.TestCase):
 
     def test_reindex_missing_idents(self):
         search_engine = mock()
-        when(search_engine).search('tag:inbox').thenReturn(['ident1', 'ident2'])
+        when(search_engine).search('tag:inbox', all_mails=True).thenReturn(['ident1', 'ident2'])
 
         MailboxIndexerListener.SEARCH_ENGINE = search_engine