Replace test cases with all 6823 queries from ClickHouse stateless tests

parser/parser_test.go

@@ -18,12 +18,30 @@ type testMetadata struct {
 	Source string `json:"source,omitempty"`
 }
 
+// astJSON represents the structure of ast.json from ClickHouse EXPLAIN AST
+type astJSON struct {
+	Meta []struct {
+		Name string `json:"name"`
+		Type string `json:"type"`
+	} `json:"meta"`
+	Data []struct {
+		Explain string `json:"explain"`
+	} `json:"data"`
+	Rows       int `json:"rows"`
+	Statistics struct {
+		Elapsed  float64 `json:"elapsed"`
+		RowsRead int     `json:"rows_read"`
+		BytesRead int    `json:"bytes_read"`
+	} `json:"statistics"`
+	Error bool `json:"error,omitempty"`
+}
+
 // TestParser tests the parser using test cases from the testdata directory.
 // Each subdirectory in testdata represents a test case with:
 // - query.sql: The SQL query to parse
+// - ast.json: Expected AST from ClickHouse EXPLAIN AST
 // - metadata.json (optional): Metadata including:
 //   - todo: true if the test is not yet expected to pass
-//   - source: URL to the source file in ClickHouse repository
 func TestParser(t *testing.T) {
 	testdataDir := "testdata"
 
@@ -37,21 +55,24 @@ func TestParser(t *testing.T) {
 			continue
 		}
 
-		testName := entry.Name()
-		testDir := filepath.Join(testdataDir, testName)
+		testDir := filepath.Join(testdataDir, entry.Name())
+
+		t.Run(entry.Name(), func(t *testing.T) {
+			t.Parallel()
 
-		t.Run(testName, func(t *testing.T) {
 			// Create context with 1 second timeout
 			ctx, cancel := context.WithTimeout(context.Background(), 1*time.Second)
 			defer cancel()
 
-			// Read the query
+			// Read the query (only first line, as ast.json was generated from first statement)
 			queryPath := filepath.Join(testDir, "query.sql")
 			queryBytes, err := os.ReadFile(queryPath)
 			if err != nil {
 				t.Fatalf("Failed to read query.sql: %v", err)
 			}
-			query := strings.TrimSpace(string(queryBytes))
+			// Get first line only (ast.json contains AST for first statement)
+			lines := strings.SplitN(string(queryBytes), "\n", 2)
+			query := strings.TrimSpace(lines[0])
 
 			// Read optional metadata
 			var metadata testMetadata
@@ -62,9 +83,19 @@ func TestParser(t *testing.T) {
 				}
 			}
 
-			// Log source if available
-			if metadata.Source != "" {
-				t.Logf("Source: %s", metadata.Source)
+			// Read expected AST from ClickHouse
+			var expectedAST astJSON
+			astPath := filepath.Join(testDir, "ast.json")
+			if astBytes, err := os.ReadFile(astPath); err == nil {
+				if err := json.Unmarshal(astBytes, &expectedAST); err != nil {
+					t.Fatalf("Failed to parse ast.json: %v", err)
+				}
+			}
+
+			// Skip tests where ClickHouse also couldn't parse the query
+			if expectedAST.Error {
+				t.Skipf("ClickHouse also failed to parse this query")
+				return
 			}
 
 			// Parse the query
@@ -94,6 +125,8 @@ func TestParser(t *testing.T) {
 				}
 				t.Fatalf("JSON marshal error: %v\nQuery: %s", jsonErr, query)
 			}
+
+			// TODO: Compare parsed AST against expectedAST.Data
 		})
 	}
 }

parser/testdata/00001_count_hits/ast.json

@@ -0,0 +1 @@
+{"error": true}

parser/testdata/00001_count_hits/query.sql

@@ -0,0 +1,2 @@
+-- Tags: stateful
+SELECT count() FROM test.hits

parser/testdata/00001_select_1/ast.json

@@ -0,0 +1,37 @@
+{
+	"meta":
+	[
+		{
+			"name": "explain",
+			"type": "String"
+		}
+	],
+
+	"data":
+	[
+		{
+			"explain": "SelectWithUnionQuery (children 1)"
+		},
+		{
+			"explain": " ExpressionList (children 1)"
+		},
+		{
+			"explain": "  SelectQuery (children 1)"
+		},
+		{
+			"explain": "   ExpressionList (children 1)"
+		},
+		{
+			"explain": "    Literal UInt64_1"
+		}
+	],
+
+	"rows": 5,
+
+	"statistics":
+	{
+		"elapsed": 0.001936758,
+		"rows_read": 5,
+		"bytes_read": 177
+	}
+}

parser/testdata/00002_count_visits/ast.json

@@ -0,0 +1 @@
+{"error": true}

parser/testdata/00002_count_visits/query.sql

@@ -0,0 +1,2 @@
+-- Tags: stateful
+SELECT sum(Sign) FROM test.visits

parser/testdata/00002_system_numbers/ast.json

@@ -0,0 +1,25 @@
+{
+	"meta":
+	[
+		{
+			"name": "explain",
+			"type": "String"
+		}
+	],
+
+	"data":
+	[
+		{
+			"explain": "Set"
+		}
+	],
+
+	"rows": 1,
+
+	"statistics":
+	{
+		"elapsed": 0.001602544,
+		"rows_read": 1,
+		"bytes_read": 11
+	}
+}

parser/testdata/00002_system_numbers/query.sql

@@ -0,0 +1,14 @@
+SET send_logs_level = 'fatal';
+
+SELECT * FROM system.numbers LIMIT 3;
+SELECT sys_num.number FROM system.numbers AS sys_num WHERE number > 2 LIMIT 2;
+SELECT number FROM system.numbers WHERE number >= 5 LIMIT 2;
+SELECT * FROM system.numbers WHERE number == 7 LIMIT 1;
+SELECT number AS n FROM system.numbers WHERE number IN(8, 9) LIMIT 2;
+select number from system.numbers limit 0;
+select x from system.numbers limit 1; -- { serverError UNKNOWN_IDENTIFIER }
+SELECT x, number FROM system.numbers LIMIT 1; -- { serverError UNKNOWN_IDENTIFIER }
+SELECT * FROM system.number LIMIT 1; -- { serverError UNKNOWN_TABLE }
+SELECT * FROM system LIMIT 1; -- { serverError UNKNOWN_TABLE }
+SELECT * FROM numbers LIMIT 1; -- { serverError UNKNOWN_TABLE }
+SELECT sys.number FROM system.numbers AS sys_num LIMIT 1; -- { serverError UNKNOWN_IDENTIFIER }

parser/testdata/00003_reinterpret_as_string/ast.json

@@ -0,0 +1,70 @@
+{
+	"meta":
+	[
+		{
+			"name": "explain",
+			"type": "String"
+		}
+	],
+
+	"data":
+	[
+		{
+			"explain": "SelectWithUnionQuery (children 1)"
+		},
+		{
+			"explain": " ExpressionList (children 1)"
+		},
+		{
+			"explain": "  SelectQuery (children 4)"
+		},
+		{
+			"explain": "   ExpressionList (children 1)"
+		},
+		{
+			"explain": "    Identifier number"
+		},
+		{
+			"explain": "   TablesInSelectQuery (children 1)"
+		},
+		{
+			"explain": "    TablesInSelectQueryElement (children 1)"
+		},
+		{
+			"explain": "     TableExpression (children 1)"
+		},
+		{
+			"explain": "      TableIdentifier system.numbers"
+		},
+		{
+			"explain": "   Function equals (children 1)"
+		},
+		{
+			"explain": "    ExpressionList (children 2)"
+		},
+		{
+			"explain": "     Function reinterpretAsString (children 1)"
+		},
+		{
+			"explain": "      ExpressionList (children 1)"
+		},
+		{
+			"explain": "       Identifier number"
+		},
+		{
+			"explain": "     Literal 'Ё'"
+		},
+		{
+			"explain": "   Literal UInt64_1"
+		}
+	],
+
+	"rows": 16,
+
+	"statistics":
+	{
+		"elapsed": 0.001337951,
+		"rows_read": 16,
+		"bytes_read": 614
+	}
+}

parser/testdata/00003_reinterpret_as_string/query.sql

@@ -0,0 +1 @@
+SELECT number FROM system.numbers WHERE reinterpretAsString(number) = 'Ё' LIMIT 1

parser/testdata/00004_shard_format_ast_and_remote_table/ast.json

@@ -0,0 +1 @@
+{"error": true}

parser/testdata/00004_shard_format_ast_and_remote_table/query.sql

@@ -0,0 +1,3 @@
+-- Tags: shard
+
+SELECT (dummy AS x) - 1 FROM remote('127.0.0.{2,3}', system, one)

parser/testdata/00004_top_counters/ast.json

@@ -0,0 +1 @@
+{"error": true}

parser/testdata/00004_top_counters/query.sql

@@ -0,0 +1,3 @@
+-- Tags: stateful
+SELECT CounterID, count() AS c FROM test.hits GROUP BY CounterID ORDER BY c DESC LIMIT 10;
+SELECT CounterID, count() AS c FROM test.hits GROUP BY CounterID ORDER BY c DESC LIMIT 10 SETTINGS optimize_aggregation_in_order = 1

parser/testdata/00005_filtering/ast.json

@@ -0,0 +1 @@
+{"error": true}

parser/testdata/00005_filtering/query.sql

@@ -0,0 +1,3 @@
+-- Tags: stateful
+SELECT count() FROM test.hits WHERE AdvEngineID != 0
+

parser/testdata/00005_shard_format_ast_and_remote_table_lambda/ast.json

@@ -0,0 +1 @@
+{"error": true}

parser/testdata/00005_shard_format_ast_and_remote_table_lambda/query.sql

@@ -0,0 +1,3 @@
+-- Tags: shard
+
+SELECT count() FROM remote('127.0.0.{2,3}', system, one) WHERE arrayExists((x) -> x = 1, [1, 2, 3])

parser/testdata/00006_agregates/ast.json

@@ -0,0 +1 @@
+{"error": true}

parser/testdata/00006_agregates/query.sql

@@ -0,0 +1,2 @@
+-- Tags: stateful
+SELECT sum(AdvEngineID), count(), avg(ResolutionWidth) FROM test.hits

parser/testdata/00006_extremes_and_subquery_from/ast.json

@@ -0,0 +1,25 @@
+{
+	"meta":
+	[
+		{
+			"name": "explain",
+			"type": "String"
+		}
+	],
+
+	"data":
+	[
+		{
+			"explain": "Set"
+		}
+	],
+
+	"rows": 1,
+
+	"statistics":
+	{
+		"elapsed": 0.001759471,
+		"rows_read": 1,
+		"bytes_read": 11
+	}
+}