Support json index type (#130)

Co-authored-by: rafie <rafi@redislabs.com>

Author: AvitalFineRedis

.circleci/config.yml

@@ -91,7 +91,7 @@ jobs:
         default: latest
     docker:
       - image: circleci/python:<<parameters.python_version >>
-      - image: redislabs/redisearch:edge
+      - image: redislabs/redismod:edge
     steps:
       - build_and_test
 

pyproject.toml

@@ -26,6 +26,7 @@ python = ">=2.7,<=2.9.0 || >= 3.5.0"
 redis = "^3.5.3"
 six = "^1.16.0"
 rmtest = {git = "https://github.com/RedisLabs/rmtest"}
+rejson = "^0.5.4"
 hiredis = [
     {version = "1.1.0", python = "~2.7"},
     {version = "^2.0.0", python = "^3.6"},

redisearch/client.py

@@ -8,6 +8,7 @@
 from .query import Query
 from ._util import to_string
 from .aggregation import AggregateRequest, AggregateResult, Cursor
+from enum import Enum
 
 
 class Field(object):
@@ -87,53 +88,56 @@ def __init__(self, name, separator=',', **kwargs):
         Field.__init__(self, name, args=[Field.TAG, self.SEPARATOR, separator], **kwargs)
 
 
+class IndexType(Enum):
+    """
+    Enum of the currently supported index types.
+    """
+    HASH = 1
+    JSON = 2
+
 class IndexDefinition(object):
     """
-    IndexDefinition is used to define a index definition for automatic indexing on Hash update
+    IndexDefinition is used to define a index definition for automatic indexing on Hash or Json update.
     """
 
-    ON = 'ON'
-    HASH = 'HASH'
-    PREFIX = 'PREFIX'
-    FILTER = 'FILTER'
-    LANGUAGE_FIELD = 'LANGUAGE_FIELD'
-    LANGUAGE = 'LANGUAGE'
-    SCORE_FIELD = 'SCORE_FIELD'
-    SCORE = 'SCORE'
-    PAYLOAD_FIELD = 'PAYLOAD_FIELD'
-        
-    def __init__(self, prefix=[], filter=None, language_field=None, language=None, score_field=None, score=1.0, payload_field=None):
+    def __init__(self, prefix=[], filter=None, language_field=None, language=None, score_field=None, score=1.0, payload_field=None, index_type=None):
+        args = []
 
-        args = [self.ON, self.HASH]
+        if index_type is IndexType.HASH:
+            args.extend(['ON', 'HASH'])
+        elif index_type is IndexType.JSON:
+            args.extend(['ON', 'JSON'])
+        elif index_type is not None:
+            raise RuntimeError("index_type must be one of {}".format(list(IndexType)))
 
         if len(prefix) > 0:
-            args.append(self.PREFIX)
+            args.append('PREFIX')
             args.append(len(prefix))
             for p in prefix:
                 args.append(p)
 
         if filter is not None:
-            args.append(self.FILTER)
+            args.append('FILTER')
             args.append(filter)
 
         if language_field is not None:
-            args.append(self.LANGUAGE_FIELD)
+            args.append('LANGUAGE_FIELD')
             args.append(language_field)
 
         if language is not None:
-            args.append(self.LANGUAGE)
+            args.append('LANGUAGE')
             args.append(language)
 
         if score_field is not None:
-            args.append(self.SCORE_FIELD)
+            args.append('SCORE_FIELD')
             args.append(score_field)
 
         if score is not None:
-            args.append(self.SCORE)
+            args.append('SCORE')
             args.append(score)
 
         if payload_field is not None:
-            args.append(self.PAYLOAD_FIELD)
+            args.append('PAYLOAD_FIELD')
             args.append(payload_field)
 
         self.args = args

redisearch/result.py

@@ -3,12 +3,14 @@
 from .document import Document
 from ._util import to_string
 
+
+
 class Result(object):
     """
     Represents the result of a search query, and has an array of Document objects
     """
 
-    def __init__(self, res, hascontent, duration=0, has_payload = False, with_scores = False):
+    def __init__(self, res, hascontent, duration=0, has_payload=False, with_scores=False):
         """
         - **snippets**: An optional dictionary of the form {field: snippet_size} for snippet formatting
         """
@@ -45,9 +47,14 @@ def __init__(self, res, hascontent, duration=0, has_payload = False, with_scores
             except KeyError:
                 pass
 
+            try:
+                fields['json'] = fields['$']
+                del fields['$']
+            except KeyError:
+                pass
+
             doc = Document(id, score=score, payload=payload, **fields) if with_scores else Document(id, payload=payload, **fields)
             self.docs.append(doc)
 
     def __repr__(self):
-
         return 'Result{%d total, docs: %s}' % (self.total, self.docs)

test/test.py

@@ -1,4 +1,7 @@
+import json
 import os, sys
+
+
 sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
 
 from rmtest import ModuleTestCase
@@ -12,8 +15,10 @@
 import six
 
 from redisearch import *
+from redisearch.client import IndexType
 import redisearch.aggregation as aggregations
 import redisearch.reducers as reducers
+import rejson
 
 WILL_PLAY_TEXT = os.path.abspath(os.path.dirname(__file__)) + '/will_play_text.csv.bz2'
 
@@ -487,6 +492,7 @@ def testAutoComplete(self):
     def testNoIndex(self):
         # Creating a client with a given index name
         client = self.getCleanClient('idx')
+        client.redis.flushdb()
 
         client.create_index(
             (TextField('field'),
@@ -946,7 +952,10 @@ def testAggregations(self):
             self.assertEqual('RediSearch', res[23])
             self.assertEqual(2, len(res[25]))
 
-    def testIndexDefiniontion(self):
+    def testIndexDefinition(self):
+        """
+        Create definition and test its args
+        """
         conn = self.redis()
 
         with conn as r:
@@ -955,19 +964,24 @@ def testIndexDefiniontion(self):
                 return
             client = Client('test', port=conn.port)
 
+            self.assertRaises(RuntimeError, IndexDefinition, prefix=['hset:', 'henry'], index_type='json')
+
             definition = IndexDefinition(prefix=['hset:', 'henry'],
             filter='@f1==32', language='English', language_field='play',
-            score_field='chapter', score=0.5, payload_field='txt' )
+            score_field='chapter', score=0.5, payload_field='txt', index_type=IndexType.JSON)
 
-            self.assertEqual(['ON','HASH', 'PREFIX',2,'hset:','henry',
-            'FILTER','@f1==32','LANGUAGE_FIELD','play','LANGUAGE','English',
-            'SCORE_FIELD','chapter','SCORE',0.5,'PAYLOAD_FIELD','txt'],
+            self.assertEqual(['ON', 'JSON', 'PREFIX', 2, 'hset:', 'henry',
+            'FILTER', '@f1==32', 'LANGUAGE_FIELD', 'play', 'LANGUAGE', 'English',
+            'SCORE_FIELD', 'chapter', 'SCORE', 0.5, 'PAYLOAD_FIELD', 'txt'],
             definition.args)
 
             self.createIndex(client, num_docs=500, definition=definition)
 
-
-    def testCreateClientDefiniontion(self):
+    def testCreateClientDefinition(self):
+        """
+        Create definition with no index type provided,
+        and use hset to test the client definition (the default is HASH).
+        """
         conn = self.redis()
 
         with conn as r:
@@ -987,5 +1001,57 @@ def testCreateClientDefiniontion(self):
             info = client.info()
             self.assertEqual(495, int(info['num_docs']))
 
+    def testCreateClientDefinitionHash(self):
+        """
+        Create definition with IndexType.HASH as index type (ON HASH),
+        and use hset to test the client definition.
+        """
+        conn = self.redis()
+
+        with conn as r:
+            r.flushdb()
+            if not check_version(r, 20000):
+                return
+            client = Client('test', port=conn.port)
+
+            definition = IndexDefinition(prefix=['hset:', 'henry'], index_type=IndexType.HASH)
+            self.createIndex(client, num_docs=500, definition=definition)
+
+            info = client.info()
+            self.assertEqual(494, int(info['num_docs']))
+
+            r.hset('hset:1', 'f1', 'v1');
+
+            info = client.info()
+            self.assertEqual(495, int(info['num_docs']))
+
+    def testCreateClientDefinitionJson(self):
+        """
+        Create definition with IndexType.JSON as index type (ON JSON),
+        and use json client to test it.
+        """
+        conn = self.redis()
+
+        with conn as r:
+            r.flushdb()
+            if not check_version(r, 20200):
+                return
+
+            client = Client('json1', port=conn.port)
+
+            definition = IndexDefinition(prefix=['king:'], index_type=IndexType.JSON)
+            client.create_index((TextField('$.name'),), definition=definition)
+
+            rj = rejson.Client(host='localhost', port=conn.port, decode_responses=True)
+            rj.jsonset('king:1', rejson.Path.rootPath(), {'name': 'henry'})
+            rj.jsonset('king:2', rejson.Path.rootPath(), {'name': 'james'})
+
+            res = client.search('henry')
+            self.assertEqual(res.docs[0].id, 'king:1')
+            self.assertIsNone(res.docs[0].payload)
+            self.assertEqual(res.docs[0].json, '{"name":"henry"}')
+            self.assertEqual(res.total, 1)
+
+
 if __name__ == '__main__':
     unittest.main()