sqlc-dev · kyleconroy · Dec 26, 2025 · Dec 26, 2025 · Dec 26, 2025 · Dec 26, 2025
diff --git a/CLAUDE.md b/CLAUDE.md
@@ -10,6 +10,14 @@ go run ./cmd/next-test
 
 This tool finds all tests with `todo: true` in their metadata and returns the one with the shortest `query.sql` file.
 
+To find the next format roundtrip test to work on, run:
+
+```bash
+go run ./cmd/next-test -format
+```
+
+This finds tests with `todo_format: true` in their metadata.
+
 ## Workflow
 
 1. Run `go run ./cmd/next-test` to find the next test to implement
@@ -42,6 +50,16 @@ go test ./parser/... -check-skipped -v 2>&1 | grep "PASSES NOW"
 
 Tests that output `PASSES NOW` can have their `todo` flag removed from `metadata.json`. This helps identify when parser improvements fix multiple tests at once.
 
+## Checking for Newly Passing Format Tests
+
+After implementing format changes, run:
+
+```bash
+go test ./parser/... -check-format -v 2>&1 | grep "FORMAT PASSES NOW"
+```
+
+Tests that output `FORMAT PASSES NOW` can have their `todo_format` flag removed from `metadata.json`.
+
 ## Test Structure
 
 Each test in `parser/testdata/` contains:
@@ -52,7 +70,8 @@ Each test in `parser/testdata/` contains:
 
 ### Metadata Options
 
-- `todo: true` - Test is pending implementation
+- `todo: true` - Test is pending parser/explain implementation
+- `todo_format: true` - Format roundtrip test is pending implementation
 - `skip: true` - Skip test entirely (e.g., causes infinite loop)
 - `explain: false` - Skip test (e.g., ClickHouse couldn't parse it)
 - `parse_error: true` - Query is intentionally invalid SQL

diff --git a/cmd/next-test/main.go b/cmd/next-test/main.go
@@ -2,14 +2,18 @@ package main
 
 import (
 	"encoding/json"
+	"flag"
 	"fmt"
 	"os"
 	"path/filepath"
 	"sort"
 )
 
+var formatFlag = flag.Bool("format", false, "Find tests with todo_format instead of todo")
+
 type testMetadata struct {
 	Todo       bool  `json:"todo,omitempty"`
+	TodoFormat bool  `json:"todo_format,omitempty"`
 	Explain    *bool `json:"explain,omitempty"`
 	Skip       bool  `json:"skip,omitempty"`
 	ParseError bool  `json:"parse_error,omitempty"`
@@ -21,6 +25,8 @@ type todoTest struct {
 }
 
 func main() {
+	flag.Parse()
+
 	testdataDir := "parser/testdata"
 	entries, err := os.ReadDir(testdataDir)
 	if err != nil {
@@ -49,9 +55,15 @@ func main() {
 			continue
 		}
 
-		// Only include tests marked as todo
-		if !metadata.Todo {
-			continue
+		// Check for todo or todo_format based on flag
+		if *formatFlag {
+			if !metadata.TodoFormat {
+				continue
+			}
+		} else {
+			if !metadata.Todo {
+				continue
+			}
 		}
 
 		// Skip tests with skip or explain=false or parse_error
@@ -72,8 +84,13 @@ func main() {
 		})
 	}
 
+	todoType := "todo"
+	if *formatFlag {
+		todoType = "todo_format"
+	}
+
 	if len(todoTests) == 0 {
-		fmt.Println("No todo tests found!")
+		fmt.Printf("No %s tests found!\n", todoType)
 		return
 	}
 
@@ -86,7 +103,7 @@ func main() {
 	next := todoTests[0]
 	testDir := filepath.Join(testdataDir, next.name)
 
-	fmt.Printf("Next test: %s\n\n", next.name)
+	fmt.Printf("Next %s test: %s\n\n", todoType, next.name)
 
 	// Print query.sql contents
 	queryPath := filepath.Join(testDir, "query.sql")
@@ -99,5 +116,5 @@ func main() {
 		fmt.Printf("\nExpected EXPLAIN output:\n%s\n", string(explainBytes))
 	}
 
-	fmt.Printf("\nRemaining todo tests: %d\n", len(todoTests))
+	fmt.Printf("\nRemaining %s tests: %d\n", todoType, len(todoTests))
 }
diff --git a/internal/format/statements.go b/internal/format/statements.go
@@ -8,6 +8,9 @@ import (
 
 // formatSelectWithUnionQuery formats a SELECT with UNION query.
 func formatSelectWithUnionQuery(sb *strings.Builder, q *ast.SelectWithUnionQuery) {
+	if q == nil {
+		return
+	}
 	for i, sel := range q.Selects {
 		if i > 0 {
 			sb.WriteString(" UNION ")
@@ -23,6 +26,9 @@ func formatSelectWithUnionQuery(sb *strings.Builder, q *ast.SelectWithUnionQuery
 
 // formatSelectQuery formats a SELECT query.
 func formatSelectQuery(sb *strings.Builder, q *ast.SelectQuery) {
+	if q == nil {
+		return
+	}
 	sb.WriteString("SELECT ")
 
 	if q.Distinct {

diff --git a/parser/parser_test.go b/parser/parser_test.go
@@ -17,9 +17,14 @@ import (
 // Use with: go test ./parser -check-skipped -v
 var checkSkipped = flag.Bool("check-skipped", false, "Run skipped todo tests to see which ones now pass")
 
+// checkFormat runs skipped todo_format tests to see which ones now pass.
+// Use with: go test ./parser -check-format -v
+var checkFormat = flag.Bool("check-format", false, "Run skipped todo_format tests to see which ones now pass")
+
 // testMetadata holds optional metadata for a test case
 type testMetadata struct {
 	Todo       bool   `json:"todo,omitempty"`
+	TodoFormat bool   `json:"todo_format,omitempty"` // true if format roundtrip test is pending
 	Source     string `json:"source,omitempty"`
 	Explain    *bool  `json:"explain,omitempty"`
 	Skip       bool   `json:"skip,omitempty"`
@@ -170,6 +175,33 @@ func TestParser(t *testing.T) {
 				}
 			}
 
+			// Check Format output (roundtrip test)
+			// Skip if todo_format is true, unless -check-format flag is set
+			if !metadata.TodoFormat || *checkFormat {
+				formatted := parser.Format(stmts)
+				expected := strings.TrimSpace(query)
+				if formatted != expected {
+					if metadata.TodoFormat {
+						if *checkFormat {
+							t.Logf("FORMAT STILL FAILING:\nExpected:\n%s\n\nGot:\n%s", expected, formatted)
+						}
+					} else {
+						t.Errorf("Format output mismatch\nExpected:\n%s\n\nGot:\n%s", expected, formatted)
+					}
+				} else if metadata.TodoFormat && *checkFormat {
+					// Automatically remove the todo_format flag from metadata.json
+					metadata.TodoFormat = false
+					updatedBytes, err := json.Marshal(metadata)
+					if err != nil {
+						t.Errorf("Failed to marshal updated metadata: %v", err)
+					} else if err := os.WriteFile(metadataPath, append(updatedBytes, '\n'), 0644); err != nil {
+						t.Errorf("Failed to write updated metadata.json: %v", err)
+					} else {
+						t.Logf("FORMAT ENABLED - removed todo_format flag from: %s", entry.Name())
+					}
+				}
+			}
+
 			// If we get here with a todo test and -check-skipped is set, the test passes!
 			// Automatically remove the todo flag from metadata.json
 			if metadata.Todo && *checkSkipped {

diff --git a/parser/testdata/00001_count_hits/metadata.json b/parser/testdata/00001_count_hits/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00001_select_1/metadata.json b/parser/testdata/00001_select_1/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00002_count_visits/metadata.json b/parser/testdata/00002_count_visits/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00002_system_numbers/metadata.json b/parser/testdata/00002_system_numbers/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00003_reinterpret_as_string/metadata.json b/parser/testdata/00003_reinterpret_as_string/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00004_shard_format_ast_and_remote_table/metadata.json b/parser/testdata/00004_shard_format_ast_and_remote_table/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00004_top_counters/metadata.json b/parser/testdata/00004_top_counters/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00005_filtering/metadata.json b/parser/testdata/00005_filtering/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00005_shard_format_ast_and_remote_table_lambda/metadata.json b/parser/testdata/00005_shard_format_ast_and_remote_table_lambda/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00006_agregates/metadata.json b/parser/testdata/00006_agregates/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00006_extremes_and_subquery_from/metadata.json b/parser/testdata/00006_extremes_and_subquery_from/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00007_uniq/metadata.json b/parser/testdata/00007_uniq/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00008_array_join/metadata.json b/parser/testdata/00008_array_join/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00008_uniq/metadata.json b/parser/testdata/00008_uniq/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00009_array_join_subquery/metadata.json b/parser/testdata/00009_array_join_subquery/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00009_uniq_distributed/metadata.json b/parser/testdata/00009_uniq_distributed/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00010_big_array_join/metadata.json b/parser/testdata/00010_big_array_join/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00010_quantiles_segfault/metadata.json b/parser/testdata/00010_quantiles_segfault/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00011_array_join_alias/metadata.json b/parser/testdata/00011_array_join_alias/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00011_sorting/metadata.json b/parser/testdata/00011_sorting/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00012_array_join_alias_2/metadata.json b/parser/testdata/00012_array_join_alias_2/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00012_sorting_distributed/metadata.json b/parser/testdata/00012_sorting_distributed/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00013_create_table_with_arrays/metadata.json b/parser/testdata/00013_create_table_with_arrays/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00013_sorting_of_nested/metadata.json b/parser/testdata/00013_sorting_of_nested/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00014_filtering_arrays/metadata.json b/parser/testdata/00014_filtering_arrays/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00014_select_from_table_with_nested/metadata.json b/parser/testdata/00014_select_from_table_with_nested/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00015_totals_and_no_aggregate_functions/metadata.json b/parser/testdata/00015_totals_and_no_aggregate_functions/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00015_totals_having_constants/metadata.json b/parser/testdata/00015_totals_having_constants/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00016_any_if_distributed_cond_always_false/metadata.json b/parser/testdata/00016_any_if_distributed_cond_always_false/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00016_totals_having_constants/metadata.json b/parser/testdata/00016_totals_having_constants/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00017_aggregation_uninitialized_memory/metadata.json b/parser/testdata/00017_aggregation_uninitialized_memory/metadata.json
@@ -1 +1 @@
-{"todo": true}
+{"todo":true,"todo_format":true}
diff --git a/parser/testdata/00017_in_subquery_with_empty_result/metadata.json b/parser/testdata/00017_in_subquery_with_empty_result/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00018_distinct_in_subquery/metadata.json b/parser/testdata/00018_distinct_in_subquery/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00019_shard_quantiles_totals_distributed/metadata.json b/parser/testdata/00019_shard_quantiles_totals_distributed/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00020_distinct_order_by_distributed/metadata.json b/parser/testdata/00020_distinct_order_by_distributed/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00020_sorting_arrays/metadata.json b/parser/testdata/00020_sorting_arrays/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00021_1_select_with_in/metadata.json b/parser/testdata/00021_1_select_with_in/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00021_2_select_with_in/metadata.json b/parser/testdata/00021_2_select_with_in/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00021_3_select_with_in/metadata.json b/parser/testdata/00021_3_select_with_in/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00021_sorting_arrays/metadata.json b/parser/testdata/00021_sorting_arrays/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00022_func_higher_order_and_constants/metadata.json b/parser/testdata/00022_func_higher_order_and_constants/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00022_merge_prewhere/metadata.json b/parser/testdata/00022_merge_prewhere/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00023_agg_select_agg_subquery/metadata.json b/parser/testdata/00023_agg_select_agg_subquery/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00023_totals_limit/metadata.json b/parser/testdata/00023_totals_limit/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00024_random_counters/metadata.json b/parser/testdata/00024_random_counters/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00024_unused_array_join_in_subquery/metadata.json b/parser/testdata/00024_unused_array_join_in_subquery/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00025_implicitly_used_subquery_column/metadata.json b/parser/testdata/00025_implicitly_used_subquery_column/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00026_shard_something_distributed/metadata.json b/parser/testdata/00026_shard_something_distributed/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00027_argMinMax/metadata.json b/parser/testdata/00027_argMinMax/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00027_distinct_and_order_by/metadata.json b/parser/testdata/00027_distinct_and_order_by/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00027_simple_argMinArray/metadata.json b/parser/testdata/00027_simple_argMinArray/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00028_shard_big_agg_aj_distributed/metadata.json b/parser/testdata/00028_shard_big_agg_aj_distributed/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00030_alter_table/metadata.json b/parser/testdata/00030_alter_table/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00030_array_enumerate_uniq/metadata.json b/parser/testdata/00030_array_enumerate_uniq/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00031_array_enumerate_uniq/metadata.json b/parser/testdata/00031_array_enumerate_uniq/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00031_parser_number/metadata.json b/parser/testdata/00031_parser_number/metadata.json
@@ -1 +1 @@
-{"todo": true}
+{"todo":true,"todo_format":true}
diff --git a/parser/testdata/00032_aggregate_key64/metadata.json b/parser/testdata/00032_aggregate_key64/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00032_fixed_string_to_string/metadata.json b/parser/testdata/00032_fixed_string_to_string/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00033_aggregate_key_string/metadata.json b/parser/testdata/00033_aggregate_key_string/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00033_fixed_string_to_string/metadata.json b/parser/testdata/00033_fixed_string_to_string/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00034_aggregate_key_fixed_string/metadata.json b/parser/testdata/00034_aggregate_key_fixed_string/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00034_fixed_string_to_number/metadata.json b/parser/testdata/00034_fixed_string_to_number/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00035_aggregate_keys128/metadata.json b/parser/testdata/00035_aggregate_keys128/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00036_aggregate_hashed/metadata.json b/parser/testdata/00036_aggregate_hashed/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00036_array_element/metadata.json b/parser/testdata/00036_array_element/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00037_totals_limit/metadata.json b/parser/testdata/00037_totals_limit/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00037_uniq_state_merge1/metadata.json b/parser/testdata/00037_uniq_state_merge1/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00038_totals_limit/metadata.json b/parser/testdata/00038_totals_limit/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00038_uniq_state_merge2/metadata.json b/parser/testdata/00038_uniq_state_merge2/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}
diff --git a/parser/testdata/00039_primary_key/metadata.json b/parser/testdata/00039_primary_key/metadata.json
@@ -1 +1 @@
-{}
+{"todo_format":true}