add benchmark phrase query

Author: Ma-cat

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');"
 ```

python/benchmark/clients/elasticsearch_client.py

@@ -108,7 +108,7 @@ def upload(self):
         self.client.indices.forcemerge(index=self.table_name, wait_for_completion=True)
 
     # https://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-match-query.html
-    def get_fulltext_query_content(self, query: str, is_and: bool = False) -> Any:
+    def get_fulltext_query_content(self, query: str, is_and: bool = False, is_phrase: bool = False) -> Any:
         ret = None
         if is_and:
             terms = query.split()
@@ -118,7 +118,10 @@ def get_fulltext_query_content(self, query: str, is_and: bool = False) -> Any:
                 }
             }
         else:
-            ret = {"query": {"match": {"body": query}}}
+            if is_phrase:
+                ret = {"query": {"match_phrase": {"body": query}}}
+            else:
+                ret = {"query": {"match": {"body": query}}}
         return ret
 
     def setup_clients(self, num_threads=1):
@@ -149,7 +152,7 @@ def do_single_query(self, query_id, client_id) -> list[Any]:
             ]
             return result
         elif self.data["mode"] == "fulltext":
-            body = self.get_fulltext_query_content(query)
+            body = self.get_fulltext_query_content(query, is_phrase=self.data["is_phrase_query"])
             result = self.client.search(
                 index=self.table_name,
                 body=body,

python/benchmark/clients/quickwit_client.py

@@ -176,7 +176,7 @@ def upload(self):
         else:
             raise TypeError("Unsupported file type")
 
-    def get_fulltext_query_content(self, query: str, is_and: bool = False) -> Any:
+    def get_fulltext_query_content(self, query: str, is_and: bool = False, is_phrase=False) -> Any:
         ret = None
         if is_and:
             terms = query.split()
@@ -186,10 +186,22 @@ def get_fulltext_query_content(self, query: str, is_and: bool = False) -> Any:
                 }
             }
         else:
-            ret = {
-                "query": {"query_string": {"query": query, "fields": ["body"]}},
-                "sort": ["_score"],
-            }
+            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 +214,7 @@ 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, is_phrase=self.data["is_phrase_query"])
             body["size"] = self.data["topK"]
 
             result = client.search(

python/benchmark/configs/elasticsearch_tantivy.json

@@ -6,6 +6,7 @@
   "data_path": "datasets/tantivy/wiki-articles.json",
   "insert_batch_size": 8192,
   "query_path": "datasets/tantivy/operations.txt",
+  "is_phrase_query": false,
   "result_path": "datasets/tantivy/elasticsearch_result.jsonl",
   "mode": "fulltext",
   "topK": 10,

python/benchmark/configs/quickwit_tantivy.json

@@ -6,6 +6,7 @@
   "data_path": "datasets/tantivy/wiki-articles.json",
   "insert_batch_size": 8192,
   "query_path": "datasets/tantivy/operations.txt",
+  "is_phrase_query": false,
   "result_path": "datasets/tantivy/quickwit_result.jsonl",
   "mode": "fulltext",
   "topK": 10,

src/unit_test/storage/invertedindex/search/query_match.cpp

@@ -278,7 +278,6 @@ void QueryMatchTest::InsertData(const String& db_name, const String& table_name)
                 auto& row = datas_[block_id];
                 for (SizeT i = 0; i < column_vectors.size(); ++i) {
                     auto &column = row[i];
-                    fmt::print("insert {}\n", column);
                     column_vectors[i].AppendByStringView(column, ',');
                 }
                 block_entry->IncreaseRowCount(1);
@@ -352,9 +351,8 @@ void QueryMatchTest::QueryMatch(const String& db_name,
             auto phrase_iterator = dynamic_cast<PhraseDocIterator *>(doc_iterator.get());
             auto res_df = phrase_iterator->GetDF();
             auto res_phrase_freq = phrase_iterator->GetPhraseFreq();
-            fmt::print("res_df: {}, res_phrase_freq: {}\n", res_df, res_phrase_freq);
-//            EXPECT_EQ(res_df, expected_doc_freq);
-//            EXPECT_EQ(res_phrase_freq, expected_matched_freq);
+            EXPECT_EQ(res_df, expected_doc_freq);
+            EXPECT_EQ(res_phrase_freq, expected_matched_freq);
         } else {
             EXPECT_EQ(doc_iterator->GetType(), DocIteratorType::kTermIterator);
             auto term_iterator = dynamic_cast<TermDocIterator *>(doc_iterator.get());

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