Merge pull request #8793 from dolthub/macneale4/dirty-branch-column

Add the `dirty` column to the dolt_branches table

go/libraries/doltcore/sqle/dtables/branches_table.go

@@ -15,6 +15,7 @@
 package dtables
 
 import (
+	"errors"
 	"fmt"
 	"io"
 
@@ -26,6 +27,7 @@ import (
 	"github.com/dolthub/dolt/go/libraries/doltcore/schema"
 	"github.com/dolthub/dolt/go/libraries/doltcore/sqle/dsess"
 	"github.com/dolthub/dolt/go/libraries/doltcore/sqle/index"
+	"github.com/dolthub/dolt/go/store/hash"
 )
 
 const branchesDefaultRowCount = 10
@@ -90,6 +92,7 @@ func (bt *BranchesTable) Schema() sql.Schema {
 	if !bt.remote {
 		columns = append(columns, &sql.Column{Name: "remote", Type: types.Text, Source: bt.tableName, PrimaryKey: false, Nullable: true})
 		columns = append(columns, &sql.Column{Name: "branch", Type: types.Text, Source: bt.tableName, PrimaryKey: false, Nullable: true})
+		columns = append(columns, &sql.Column{Name: "dirty", Type: types.Boolean, Source: bt.tableName, PrimaryKey: false, Nullable: true})
 	}
 	return columns
 }
@@ -114,6 +117,7 @@ type BranchItr struct {
 	table    *BranchesTable
 	branches []string
 	commits  []*doltdb.Commit
+	dirty    []bool
 	idx      int
 }
 
@@ -145,26 +149,36 @@ func NewBranchItr(ctx *sql.Context, table *BranchesTable) (*BranchItr, error) {
 
 	branchNames := make([]string, len(branchRefs))
 	commits := make([]*doltdb.Commit, len(branchRefs))
+	dirtyBits := make([]bool, len(branchRefs))
 	for i, branch := range branchRefs {
 		commit, err := ddb.ResolveCommitRefAtRoot(ctx, branch, txRoot)
-
 		if err != nil {
 			return nil, err
 		}
 
+		var dirty bool
+		if !remote {
+			dirty, err = isDirty(ctx, ddb, commit, branch, txRoot)
+			if err != nil {
+				return nil, err
+			}
+		}
+
 		if branch.GetType() == ref.RemoteRefType {
 			branchNames[i] = "remotes/" + branch.GetPath()
 		} else {
 			branchNames[i] = branch.GetPath()
 		}
 
+		dirtyBits[i] = dirty
 		commits[i] = commit
 	}
 
 	return &BranchItr{
 		table:    table,
 		branches: branchNames,
 		commits:  commits,
+		dirty:    dirtyBits,
 		idx:      0,
 	}, nil
 }
@@ -182,6 +196,7 @@ func (itr *BranchItr) Next(ctx *sql.Context) (sql.Row, error) {
 
 	name := itr.branches[itr.idx]
 	cm := itr.commits[itr.idx]
+	dirty := itr.dirty[itr.idx]
 	meta, err := cm.GetCommitMeta(ctx)
 
 	if err != nil {
@@ -211,8 +226,53 @@ func (itr *BranchItr) Next(ctx *sql.Context) (sql.Row, error) {
 			remoteName = branch.Remote
 			branchName = branch.Merge.Ref.GetPath()
 		}
-		return sql.NewRow(name, h.String(), meta.Name, meta.Email, meta.Time(), meta.Description, remoteName, branchName), nil
+		return sql.NewRow(name, h.String(), meta.Name, meta.Email, meta.Time(), meta.Description, remoteName, branchName, dirty), nil
+	}
+}
+
+// isDirty returns true if the working ref points to a dirty branch.
+func isDirty(ctx *sql.Context, ddb *doltdb.DoltDB, commit *doltdb.Commit, branch ref.DoltRef, txRoot hash.Hash) (bool, error) {
+	wsRef, err := ref.WorkingSetRefForHead(branch)
+	if err != nil {
+		return false, err
+	}
+	ws, err := ddb.ResolveWorkingSetAtRoot(ctx, wsRef, txRoot)
+	if err != nil {
+		if errors.Is(err, doltdb.ErrWorkingSetNotFound) {
+			// If there is no working set for this branch, then it is never dirty. This happens on servers commonly.
+			return false, nil
+		}
+		return false, err
+	}
+
+	workingRoot := ws.WorkingRoot()
+	workingRootHash, err := workingRoot.HashOf()
+	if err != nil {
+		return false, err
+	}
+	stagedRoot := ws.StagedRoot()
+	stagedRootHash, err := stagedRoot.HashOf()
+	if err != nil {
+		return false, err
+	}
+
+	dirty := false
+	if workingRootHash != stagedRootHash {
+		dirty = true
+	} else {
+		cmRt, err := commit.GetRootValue(ctx)
+		if err != nil {
+			return false, err
+		}
+		cmRtHash, err := cmRt.HashOf()
+		if err != nil {
+			return false, err
+		}
+		if cmRtHash != workingRootHash {
+			dirty = true
+		}
 	}
+	return dirty, nil
 }
 
 // Close closes the iterator.

go/libraries/doltcore/sqle/sqlselect_test.go

@@ -784,6 +784,7 @@ func BasicSelectTests() []SelectTest {
 					"Initialize data repository",
 					"",
 					"",
+					true, // Test setup has a dirty workspace.
 				},
 			},
 			ExpectedSqlSchema: sql.Schema{
@@ -795,6 +796,7 @@ func BasicSelectTests() []SelectTest {
 				&sql.Column{Name: "latest_commit_message", Type: gmstypes.Text},
 				&sql.Column{Name: "remote", Type: gmstypes.Text},
 				&sql.Column{Name: "branch", Type: gmstypes.Text},
+				&sql.Column{Name: "dirty", Type: gmstypes.Boolean},
 			},
 		},
 	}

go/libraries/doltcore/sqle/testutil.go

@@ -452,7 +452,7 @@ func CreateEmptyTestTable(dEnv *env.DoltEnv, tableName string, sch schema.Schema
 	return dEnv.UpdateWorkingRoot(ctx, newRoot)
 }
 
-// CreateTestDatabase creates a test database with the test data set in it.
+// CreateTestDatabase creates a test database with the test data set in it. Has a dirty workspace as well.
 func CreateTestDatabase() (*env.DoltEnv, error) {
 	ctx := context.Background()
 	dEnv, err := CreateEmptyTestDatabase()

integration-tests/bats/sql-server.bats

@@ -1996,4 +1996,44 @@ EOF
     run dolt --data-dir datadir1 sql-server --data-dir datadir2
     [ $status -eq 1 ]
     [[ "$output" =~ "cannot specify both global --data-dir argument and --data-dir in sql-server config" ]] || false
-}
+}
+
+# This is really a test of the dolt_Branches system table, but due to needing a server with multiple dirty branches
+# it was easier to test it with a sql-server.
+@test "sql-server: dirty branches listed properly in dolt_branches table" {
+    skiponwindows "Missing dependencies"
+
+    cd repo1
+    dolt checkout main
+    dolt branch br1 # Will be a clean commit, ahead of main.
+    dolt branch br2 # will be a dirty branch, on main.
+    dolt branch br3 # will be a dirty branch, on br1
+    start_sql_server repo1
+
+    dolt --use-db "repo1" --branch br1 sql -q "CREATE TABLE tbl (i int primary key)"
+    dolt --use-db "repo1" --branch br1 sql -q "CALL DOLT_COMMIT('-Am', 'commit it')"
+
+    dolt --use-db "repo1" --branch br2 sql -q "CREATE TABLE tbl (j int primary key)"
+
+    # Fast forward br3 to br1, then make it dirty.
+    dolt --use-db "repo1" --branch br3 sql -q "CALL DOLT_MERGE('br1')"
+    dolt --use-db "repo1" --branch br3 sql -q "CREATE TABLE othertbl (k int primary key)"
+
+    stop_sql_server 1 && sleep 0.5
+
+    run dolt sql -q "SELECT name,dirty FROM dolt_branches"
+    [ "$status" -eq 0 ]
+    [[ "$output" =~ "br1  | false" ]] || false
+    [[ "$output" =~ "br2  | true " ]] || false
+    [[ "$output" =~ "br3  | true" ]] || false
+    [[ "$output" =~ "main | false" ]] || false
+
+    # Verify that the dolt_branches table show the same output, regardless of the checked out branch.
+    dolt checkout br1
+    run dolt sql -q "SELECT name,dirty FROM dolt_branches"
+    [ "$status" -eq 0 ]
+    [[ "$output" =~ "br1  | false" ]] || false
+    [[ "$output" =~ "br2  | true " ]] || false
+    [[ "$output" =~ "br3  | true" ]] || false
+    [[ "$output" =~ "main | false" ]] || false
+}

integration-tests/mysql-client-tests/node/workbenchTests/branches.js

@@ -58,6 +58,7 @@ export const branchTests = [
         latest_commit_message: "Initialize data repository",
         remote: "",
         branch: "",
+        dirty: 0,
       },
       {
         name: "mybranch",
@@ -68,6 +69,7 @@ export const branchTests = [
         latest_commit_message: "Create table test",
         remote: "",
         branch: "",
+        dirty: 0,
       },
     ],
     matcher: branchesMatcher,