index query limit: ensure 'limit' is correctly applied to XmlQueryInd…

…exSource, fixes ukwa/ukwa-pywb#49 (#523)

pywb/warcserver/index/indexsource.py

@@ -243,6 +243,10 @@ def load_index(self, params):
             raise BadRequestException('matchType={0} is not supported'.format(matchType=matchType))
 
         try:
+            limit = params.get('limit')
+            if limit:
+                query = 'limit: {0} '.format(limit) + query
+
             # OpenSearch API requires double-escaping
             # TODO: add option to not double escape if needed
             query_url = self.query_api_url + '?q=' + quote_plus(query + quote_plus(url))

pywb/warcserver/index/test/test_xmlquery_indexsource.py

@@ -9,9 +9,14 @@
 import pytest
 
 
+query_url = None
+
+
 # ============================================================================
 def mock_get(self, url):
     string = ''
+    global query_url
+    query_url = url
     if quote_plus(XmlQueryIndexSource.EXACT_QUERY) in url:
         if quote_plus(quote_plus('http://example.com/some/path')) in url:
             string = URL_RESPONSE_2
@@ -65,12 +70,14 @@ def do_query(self, params):
 
     @patch('pywb.warcserver.index.indexsource.requests.sessions.Session.get', mock_get)
     def test_exact_query(self):
-        res, errs = self.do_query({'url': 'http://example.com/'})
+        res, errs = self.do_query({'url': 'http://example.com/', 'limit': 100})
+
         expected = """\
 com,example)/ 20180112200243 example.warc.gz
 com,example)/ 20180216200300 example.warc.gz"""
         assert(key_ts_res(res) == expected)
         assert(errs == {})
+        assert query_url == 'http://localhost:8080/path?q=limit%3A+100+type%3Aurlquery+url%3Ahttp%253A%252F%252Fexample.com%252F'
 
 
     @patch('pywb.warcserver.index.indexsource.requests.sessions.Session.get', mock_get)
@@ -82,6 +89,8 @@ def test_exact_query_2(self):
         assert(key_ts_res(res) == expected)
         assert(errs == {})
 
+        assert query_url == 'http://localhost:8080/path?q=type%3Aurlquery+url%3Ahttp%253A%252F%252Fexample.com%252Fsome%252Fpath'
+
 
     @patch('pywb.warcserver.index.indexsource.requests.sessions.Session.get', mock_get)
     def test_prefix_query(self):