rule: fix squash rule with convert_dates (#827)

rule: fix squash rule with convert_dates

Author: ajnavarro

integration_test.go

@@ -8,6 +8,7 @@ import (
 	"path/filepath"
 	"strings"
 	"testing"
+	"time"
 
 	"github.com/src-d/gitbase/cmd/gitbase/command"
 	"github.com/src-d/gitbase/internal/rule"
@@ -354,6 +355,30 @@ func TestIntegration(t *testing.T) {
 			WHERE r.ref_name = 'HEAD' AND IS_VENDOR(cf.file_path)`,
 			[]sql.Row{{".gitignore"}, {"vendor/foo.go"}},
 		},
+		{
+			`
+			SELECT f.file_path, c.commit_author_when
+			FROM repositories r
+			NATURAL JOIN commits c
+			NATURAL JOIN commit_files cf
+			NATURAL JOIN files f
+			WHERE r.repository_id = 'worktree'
+				AND c.commit_hash = '6ecf0ef2c2dffb796033e5a02219af86ec6584e5'
+				AND f.file_path NOT REGEXP '^vendor.*'
+				AND NOT IS_BINARY(f.blob_content)
+			LIMIT 10
+			`,
+			[]sql.Row{
+				{".gitignore", time.Date(2015, time.April, 5, 21, 30, 47, 0, time.UTC)},
+				{"CHANGELOG", time.Date(2015, time.April, 5, 21, 30, 47, 0, time.UTC)},
+				{"LICENSE", time.Date(2015, time.April, 5, 21, 30, 47, 0, time.UTC)},
+				{"binary.jpg", time.Date(2015, time.April, 5, 21, 30, 47, 0, time.UTC)},
+				{"go/example.go", time.Date(2015, time.April, 5, 21, 30, 47, 0, time.UTC)},
+				{"json/long.json", time.Date(2015, time.April, 5, 21, 30, 47, 0, time.UTC)},
+				{"json/short.json", time.Date(2015, time.April, 5, 21, 30, 47, 0, time.UTC)},
+				{"php/crappy.php", time.Date(2015, time.April, 5, 21, 30, 47, 0, time.UTC)},
+			},
+		},
 	}
 
 	var pid uint64

internal/rule/squashjoins.go

@@ -111,7 +111,32 @@ func squashProjects(parent, child *plan.Project) (sql.Node, bool) {
 	projections := make([]sql.Expression, len(parent.Expressions()))
 	schema := child.Child.Schema()
 
+	// When squashing two projects, it's possible that the parent project has
+	// a reference to a column defined in the child project.
+	// For that reason, we need to gather the new columns that were defined
+	// in the child project in order to replace the reference to those in the
+	// parent project with the new column definition.
+	var newColumns = make(map[string]sql.Expression)
+	for _, e := range child.Expressions() {
+		if _, ok := e.(*expression.GetField); !ok {
+			var name string
+			if n, ok := e.(sql.Nameable); ok {
+				name = n.Name()
+			} else {
+				name = e.String()
+			}
+
+			newColumns[name] = e
+		}
+	}
+
 	for i, e := range parent.Expressions() {
+		if f, ok := e.(*expression.GetField); ok && f.Table() == "" {
+			if expr, ok := newColumns[f.Name()]; ok {
+				e = expr
+			}
+		}
+
 		fe, err := fixFieldIndexes(e, schema)
 		if err != nil {
 			return nil, false

internal/rule/squashjoins_test.go

@@ -81,9 +81,6 @@ func TestAnalyzeSquashNaturalJoins(t *testing.T) {
 	project, ok := exchange.Child.(*plan.Project)
 	require.True(ok)
 
-	project, ok = project.Child.(*plan.Project)
-	require.True(ok)
-
 	filter, ok := project.Child.(*plan.Filter)
 	require.True(ok)