add benchmark phrase query (#1366)

### What problem does this PR solve? * In the benchmark, es client and quickwit client support phrase query * Recover phrase query sqllogic test ### Type of change - [x] New Feature - [x] Refactoring
infiniflow · Jun 21, 2024 · 47fa26f · 47fa26f
1 parent b945c7c
commit 47fa26f
Show file tree

Hide file tree

Showing 4 changed files with 40 additions and 16 deletions.
diff --git a/docs/references/benchmark.md b/docs/references/benchmark.md
@@ -114,7 +114,7 @@ Following are commands to issue a single query so that you can compare results a
 ```base
 curl -X GET "http://localhost:9200/elasticsearch_enwiki/_search" -H 'Content-Type: application/json' -d'{"size":10,"_source":"doctitle","query": {"match": { "body": "wraysbury istorijos" }}}'
 
-curl -X GET "http://localhost:7280/api/v1/_elastic/qucikwit_enwiki/_search" -H 'Content-Type: application/json' -d'{"query": {"query_string": {"query": "wraysbury istorijos", "fields": [ "body" ] } },"sort": ["_score"]}'
+curl -X GET "http://localhost:7280/api/v1/_elastic/qucikwit_enwiki/_search" -H 'Content-Type: application/json' -d'{"query": {"query_string": {"query": "wraysbury istorijos", "fields": [ "body" ] } },"sort": ["_score"],"size":10}'
 
 psql -h 0.0.0.0 -p 5432 -c "SELECT doctitle, ROW_ID(), SCORE() FROM infinity_enwiki SEARCH MATCH TEXT ('body', 'wraysbury istorijos', 'topn=10');"
 ```

diff --git a/python/benchmark/clients/elasticsearch_client.py b/python/benchmark/clients/elasticsearch_client.py
@@ -118,7 +118,14 @@ def get_fulltext_query_content(self, query: str, is_and: bool = False) -> Any:
                 }
             }
         else:
-            ret = {"query": {"match": {"body": query}}}
+            is_phrase = False
+            query = query.lstrip()
+            if query.startswith('"'):
+                is_phrase = True
+            if is_phrase:
+                ret = {"query": {"match_phrase": {"body": query}}}
+            else:
+                ret = {"query": {"match": {"body": query}}}
         return ret
 
     def setup_clients(self, num_threads=1):

diff --git a/python/benchmark/clients/quickwit_client.py b/python/benchmark/clients/quickwit_client.py
@@ -186,10 +186,27 @@ def get_fulltext_query_content(self, query: str, is_and: bool = False) -> Any:
                 }
             }
         else:
-            ret = {
-                "query": {"query_string": {"query": query, "fields": ["body"]}},
-                "sort": ["_score"],
-            }
+            is_phrase = False
+            query = query.lstrip()
+            if query.startswith('"'):
+                is_phrase = True
+
+            if is_phrase:
+                ret = {
+                    "query": {
+                        "match_phrase": {"body": query}
+                    },
+                    "sort": ["_score"],
+                    "size": self.data["topK"]
+                }
+            else:
+                ret = {
+                    "query": {
+                        "query_string": {"query": query, "fields": ["body"]}
+                    },
+                    "sort": ["_score"],
+                    "size": self.data["topK"]
+                }
         return ret
 
     def setup_clients(self, num_threads=1):
@@ -202,7 +219,7 @@ def do_single_query(self, query_id, client_id) -> list[Any]:
         query = self.queries[query_id]
         client = self.clients[client_id]
         if self.data["mode"] == "fulltext":
-            body = self.get_fulltext_query_content(query)
+            body = self.get_fulltext_query_content(query=query)
             body["size"] = self.data["topK"]
 
             result = client.search(

diff --git a/test/sql/dql/fulltext/fulltext.slt b/test/sql/dql/fulltext/fulltext.slt
@@ -28,17 +28,17 @@ SELECT doctitle, docdate, ROW_ID(), SCORE() FROM enwiki SEARCH MATCH TEXT ('body
 Anarchism 30-APR-2012 03:25:17.000 0 21.620300
 
 # only phrase
-# query TTI rowsort
-# SELECT doctitle, docdate, ROW_ID(), SCORE() FROM enwiki SEARCH MATCH TEXT ('body^5', '"social customs"', 'topn=3;block_max=compare');
-# ----
-# Anarchism 30-APR-2012 03:25:17.000 6 20.881144
+query TTI rowsort
+SELECT doctitle, docdate, ROW_ID(), SCORE() FROM enwiki SEARCH MATCH TEXT ('body^5', '"social customs"', 'topn=3;block_max=compare');
+----
+Anarchism 30-APR-2012 03:25:17.000 6 20.881144
 
 # phrase and term
-# query TTI rowsort
-# SELECT doctitle, docdate, ROW_ID(), SCORE() FROM enwiki SEARCH MATCH TEXT ('doctitle,body^5', '"social customs" harmful', 'topn=3');
-# ----
-# Anarchism 30-APR-2012 03:25:17.000 0 21.620300
-# Anarchism 30-APR-2012 03:25:17.000 6 20.881144
+query TTI rowsort
+SELECT doctitle, docdate, ROW_ID(), SCORE() FROM enwiki SEARCH MATCH TEXT ('doctitle,body^5', '"social customs" harmful', 'topn=3');
+----
+Anarchism 30-APR-2012 03:25:17.000 0 21.620300
+Anarchism 30-APR-2012 03:25:17.000 6 20.881144
 
 # copy data from csv file
 query I