Merge pull request #78 from ziadsawalha/mongo

Add Text Search Parsing

README.md

@@ -11,7 +11,7 @@ Common Python libraries for:
 - [WSGI Middleware](#middleware)
 - [REST API Tooling](#rest)
 - [Date/Time (chronos)](#chronos)
-- [MongoDB Backedn Wrapper](#mongo)
+- [MongoDB Backend Wrapper](#mongo)
 
 ## <a name="config"></a>Config
 

simpl/db/mongodb.py

@@ -115,6 +115,55 @@ def scrub(data):
         raise ValidationError("Input '%s' not permitted: %s" % (data, exc))
 
 
+def build_text_search(strings, name_field='name'):
+    """Build mongodb query that performs text search for string(s).
+
+    This is the backend implementation of the front-end search box on a list.
+    It performs text search on a text index and a regex search on a `name`
+    field. The value(s) for this are parsed by `rest.process_params` from the
+    `q` query param.
+
+    :param list strings: strings to search on.
+    :keyword str name: field to search on in addition to search index.
+
+    For example, searching the following collection for 'john':
+
+    #  Name    Description (in text index)
+    -  ------  ---------------------------
+    1  John    Big boss
+    2  Johnny  Boss' son
+    3  Thomas  John's first employee
+    4  Henry   He quit
+
+    Search returns all records except the last one:
+    #1 - match on name
+    #2 - partial match on name
+    #3 - text index match on description
+
+    >>> import pprint
+    >>> pprint.pprint(build_text_search(['john']))
+    {'$or': [{'$text': {'$search': 'john'}},
+             {'name': {'$options': 'i', '$regex': 'john'}}]}
+
+    >>> import pprint
+    >>> pprint.pprint(build_text_search(['john', 'tom'], name_field='objName'))
+    {'$or': [{'$text': {'$search': 'john tom'}},
+             {'$or': [{'objName': {'$options': 'i', '$regex': 'john'}},
+                      {'objName': {'$options': 'i', '$regex': 'tom'}}]}]}
+    """
+    assert isinstance(strings, list)
+
+    text_search = {'$text': {'$search': ' '.join(strings)}}
+
+    searches = [{name_field: {'$regex': s, '$options': 'i'}} for s in strings]
+    if len(searches) == 1:
+        name_search = searches[0]
+    else:
+        name_search = {'$or': searches}
+
+    return {'$or': [text_search, name_search]}
+
+
 class SimplDB(object):
 
     """Database wrapper.

simpl/rest.py

@@ -289,6 +289,12 @@ def process_params(request, standard_params=STANDARD_QUERY_PARAMS,
         sort = list(itertools.chain(*(
             comma_separated_strings(str(k)) for k in sort)))
         query_fields['sort'] = sort
+    if 'q' in request.query:
+        search = request.query.getall('q')
+        search = list(itertools.chain(*(
+            comma_separated_strings(k) for k in search
+            if k)))
+        query_fields['q'] = search
     return query_fields
 
 

test-requirements.txt

@@ -7,6 +7,7 @@ mock
 mongobox
 nose
 nose-ignore-docstring
+prettyprint
 pylint
 pyyaml
 requests

tests/test_db_mongodb.py

@@ -1,3 +1,4 @@
+# coding=utf-8
 # All Rights Reserved.
 #    Licensed under the Apache License, Version 2.0 (the "License"); you may
 #    not use this file except in compliance with the License. You may obtain
@@ -33,25 +34,6 @@ def tune(self):
         pass  # bypass async tuning in tests
 
 
-def build_search_string(strings):
-    """Emulate `rest.process_params` parsing of `q` param."""
-    assert isinstance(strings, list)
-    if len(strings) == 1:
-        name = {'name': strings[0]}
-    else:
-        name = {
-            '$or': [
-                {'name': {'$regex': s, '$options': 'i'}} for s in strings
-            ]
-        }
-    return {
-        '$or': [
-            {'$text': {'$search': ' '.join(strings)}},
-            name
-        ]
-    }
-
-
 class TestMongoDB(unittest.TestCase):
 
     """Test :mod:`simpl.db.mongodb`."""
@@ -164,22 +146,22 @@ def test_text_search(self):
         self.db.prose.save("B", {"name": "Johnny Walker",
                                  "keywords": "whisky health"})
         # Single word
-        search = build_search_string(['john'])
+        search = mongodb.build_text_search(['john'])
         result = self.db.prose.list(**search)
         self.assertEqual(result[0][0]['name'], "John Adams")
 
         # Second word
-        search = build_search_string(['adams'])
+        search = mongodb.build_text_search(['adams'])
         result = self.db.prose.list(**search)
         self.assertEqual(result[0][0]['name'], "John Adams")
 
         # Multiple words in search
-        search = build_search_string(['wealth', 'health'])
+        search = mongodb.build_text_search(['wealth', 'health'])
         result = self.db.prose.list(**search)
         self.assertEqual(result[1], 2)
 
         # Keyword
-        search = build_search_string(['whisky'])
+        search = mongodb.build_text_search(['whisky'])
         result = self.db.prose.list(**search)
         self.assertEqual(result[0][0]['name'], "Johnny Walker")
 
@@ -417,6 +399,5 @@ def test_skip_if_filtered(self):
         )
         self.assertIsNone(obj)
 
-
 if __name__ == '__main__':
     unittest.main()

tests/test_rest.py

@@ -262,11 +262,18 @@ def test_sort(self):
 
     def test_standard(self):
         request = bottle.BaseRequest(environ={
-            'QUERY_STRING': 'limit=100&offset=0&q=txt&facets=status'
+            'QUERY_STRING': 'limit=100&offset=0&facets=status'
         })
         results = rest.process_params(request)
         self.assertEqual(results, {})
 
+    def test_text(self):
+        request = bottle.BaseRequest(environ={
+            'QUERY_STRING': 'q=txt'
+        })
+        results = rest.process_params(request)
+        self.assertEqual(results, {'q': ['txt']})
+
     def test_invalid(self):
         request = bottle.BaseRequest(environ={
             'QUERY_STRING': 'foo=bar'