From 14a1b00f498887864aab18e42a102bf39bbcf9ab Mon Sep 17 00:00:00 2001
From: Gerben <gerben@treora.com>
Date: Wed, 6 Aug 2014 16:07:05 -0700
Subject: [PATCH 1/2] Search using a match query rather than term filter

---
 annotator/annotation.py    | 20 +++++++++++---------
 annotator/elasticsearch.py | 14 +++++++-------
 2 files changed, 18 insertions(+), 16 deletions(-)

diff --git a/annotator/annotation.py b/annotator/annotation.py
index 31e09af..33f6da7 100644
--- a/annotator/annotation.py
+++ b/annotator/annotation.py
@@ -113,18 +113,20 @@ def _build_query(cls, query=None, offset=None, limit=None,
         # attempt to expand query to include uris for other representations
         # using information we may have on hand about the Document
         if 'uri' in query:
-            term_filter = q['query']['filtered']['filter']
+            clauses = q['query']['bool']
             doc = document.Document.get_by_uri(query['uri'])
             if doc:
-                new_terms = []
-                for term in term_filter['and']:
-                    if 'uri' in term['term']:
-                        term = {'or': []}
+                for clause in clauses['must']:
+                    # Rewrite the 'uri' clause to match any of the document URIs
+                    if 'match' in clause and 'uri' in clause['match']:
+                        uri_matchers = []
                         for uri in doc.uris():
-                            term['or'].append({'term': {'uri': uri}})
-                    new_terms.append(term)
-
-                term_filter['and'] = new_terms
+                            uri_matchers.append({'match': {'uri': uri}})
+                        del clause['match']
+                        clause['bool'] = {
+                            'should': uri_matchers,
+                            'minimum_should_match': 1
+                        }
 
         if es.authorization_enabled:
             # Apply a filter to the results.
diff --git a/annotator/elasticsearch.py b/annotator/elasticsearch.py
index b284fdd..c16b35f 100644
--- a/annotator/elasticsearch.py
+++ b/annotator/elasticsearch.py
@@ -232,15 +232,15 @@ def _csv_split(s, delimiter=','):
 
 def _build_query(query, offset, limit):
     # Base query is a filtered match_all
-    q = {'match_all': {}}
-
-    if query:
-        f = {'and': []}
-        q = {'filtered': {'query': q, 'filter': f}}
+    match_clauses = [
+        # We start with a single match_all because Elasticsearch considers an
+        # empty conjunction to be false..
+        {'match_all': {}}
+    ]
 
     # Add a term query for each keyword
     for k, v in iteritems(query):
-        q['filtered']['filter']['and'].append({'term': {k: v}})
+        match_clauses.append({'match': {k: v}})
 
     return {
         'sort': [{'updated': {
@@ -255,7 +255,7 @@ def _build_query(query, offset, limit):
         }}],
         'from': max(0, offset),
         'size': min(RESULTS_MAX_SIZE, max(0, limit)),
-        'query': q
+        'query': {'bool': {'must': match_clauses}}
     }
 
 

From 63735a2f7b606722e1a7411dd9049df40f23ea53 Mon Sep 17 00:00:00 2001
From: Gerben <gerben@treora.com>
Date: Wed, 24 Sep 2014 13:05:34 +0300
Subject: [PATCH 2/2] Only add match_all to query when needed

---
 annotator/elasticsearch.py | 16 ++++++----------
 1 file changed, 6 insertions(+), 10 deletions(-)

diff --git a/annotator/elasticsearch.py b/annotator/elasticsearch.py
index c16b35f..d6b8a5c 100644
--- a/annotator/elasticsearch.py
+++ b/annotator/elasticsearch.py
@@ -231,16 +231,12 @@ def _csv_split(s, delimiter=','):
 
 
 def _build_query(query, offset, limit):
-    # Base query is a filtered match_all
-    match_clauses = [
-        # We start with a single match_all because Elasticsearch considers an
-        # empty conjunction to be false..
-        {'match_all': {}}
-    ]
-
-    # Add a term query for each keyword
-    for k, v in iteritems(query):
-        match_clauses.append({'match': {k: v}})
+    # Create a match query for each keyword
+    match_clauses = [{'match': {k: v}} for k, v in iteritems(query)]
+
+    if len(match_clauses) == 0:
+        # Elasticsearch considers an empty conjunction to be false..
+        match_clauses.append({'match_all': {}})
 
     return {
         'sort': [{'updated': {