Dump command prototype (#263)

Author: bplunkett-stripe

README.md

@@ -138,7 +138,12 @@ echo "CREATE TABLE bar (id varchar(255), message TEXT NOT NULL);" > schema/bar.s
 Apply the schema to a fresh database. [The connection string spec can be found here](https://www.postgresql.org/docs/current/libpq-connect.html#LIBPQ-CONNSTRING).
 Setting the `PGPASSWORD` env var will override any password set in the connection string and is recommended.
 ```bash
-pg-schema-diff apply --from-dsn "postgres://postgres:postgres@localhost:5432/postgres" --to-dir schema 
+pg-schema-diff apply --from-dsn "postgres://postgres:postgres@localhost:5432/postgres" --to-dir schema
+```
+
+Alternatively, if you have an existing database, you can dump its schema to use as a starting point:
+```bash
+mkdir -p schema && pg-schema-diff dump --dsn "postgres://postgres:postgres@localhost:5432/postgres" > schema/schema.sql
 ```
 
 ## 2. Updating schema

cmd/pg-schema-diff/dump_cmd.go

@@ -0,0 +1,105 @@
+package main
+
+import (
+	"context"
+	"database/sql"
+	"fmt"
+	"strings"
+
+	"github.com/jackc/pgx/v4"
+	"github.com/spf13/cobra"
+	"github.com/stripe/pg-schema-diff/pkg/diff"
+	"github.com/stripe/pg-schema-diff/pkg/log"
+	"github.com/stripe/pg-schema-diff/pkg/tempdb"
+)
+
+func buildDumpCmd() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "dump",
+		Short: "Dump the schema of a database as SQL DDL statements (effectively pg_dump)",
+	}
+
+	connFlags := createConnectionFlags(cmd, "", "The database to dump")
+
+	var includeSchemas []string
+	var excludeSchemas []string
+	cmd.Flags().StringArrayVar(&includeSchemas, "include-schema", nil, "Include the specified schema in the dump")
+	cmd.Flags().StringArrayVar(&excludeSchemas, "exclude-schema", nil, "Exclude the specified schema from the dump")
+
+	cmd.RunE = func(cmd *cobra.Command, args []string) error {
+		connConfig, err := parseConnectionFlags(connFlags)
+		if err != nil {
+			return err
+		}
+
+		cmd.SilenceUsage = true
+
+		plan, err := generateDump(cmd.Context(), generateDumpParams{
+			connConfig:     connConfig,
+			includeSchemas: includeSchemas,
+			excludeSchemas: excludeSchemas,
+		})
+		if err != nil {
+			return err
+		}
+
+		cmdPrintln(cmd, dumpToSql(plan))
+		return nil
+	}
+
+	return cmd
+}
+
+type generateDumpParams struct {
+	connConfig     *pgx.ConnConfig
+	includeSchemas []string
+	excludeSchemas []string
+}
+
+func generateDump(ctx context.Context, params generateDumpParams) (diff.Plan, error) {
+	connPool, err := openDbWithPgxConfig(params.connConfig)
+	if err != nil {
+		return diff.Plan{}, fmt.Errorf("opening database connection: %w", err)
+	}
+	defer connPool.Close()
+	connPool.SetMaxOpenConns(defaultMaxConnections)
+
+	tempDbFactory, err := tempdb.NewOnInstanceFactory(ctx, func(ctx context.Context, dbName string) (*sql.DB, error) {
+		cfg := params.connConfig.Copy()
+		cfg.Database = dbName
+		return openDbWithPgxConfig(cfg)
+	}, tempdb.WithRootDatabase(params.connConfig.Database))
+	if err != nil {
+		return diff.Plan{}, fmt.Errorf("creating temp db factory: %w", err)
+	}
+	defer func() {
+		if err := tempDbFactory.Close(); err != nil {
+			log.SimpleLogger().Errorf("error shutting down temp db factory: %v", err)
+		}
+	}()
+
+	plan, err := diff.Generate(ctx, diff.DDLSchemaSource([]string{}), diff.DBSchemaSource(connPool),
+		diff.WithTempDbFactory(tempDbFactory),
+		diff.WithIncludeSchemas(params.includeSchemas...),
+		diff.WithExcludeSchemas(params.excludeSchemas...),
+		diff.WithDoNotValidatePlan(),
+		diff.WithNoConcurrentIndexOps(),
+	)
+	if err != nil {
+		return diff.Plan{}, fmt.Errorf("generating plan: %w", err)
+	}
+	return plan, nil
+}
+
+// dumpToSql converts the plan to clean SQL without metadata
+func dumpToSql(plan diff.Plan) string {
+	sb := strings.Builder{}
+	for i, stmt := range plan.Statements {
+		sb.WriteString(stmt.DDL)
+		sb.WriteString(";")
+		if i < len(plan.Statements)-1 {
+			sb.WriteString("\n\n")
+		}
+	}
+	return sb.String()
+}

cmd/pg-schema-diff/dump_cmd_test.go

@@ -0,0 +1,41 @@
+package main
+
+func (suite *cmdTestSuite) TestDumpCmd() {
+	type testCase struct {
+		name        string
+		args        []string
+		dynamicArgs []dArgGenerator
+
+		// outputContains is a list of substrings that are expected to be contained in the stdout output of the command.
+		outputContains []string
+		// expectErrContains is a list of substrings that are expected to be contained in the error returned by
+		// cmd.RunE. This is DISTINCT from stdErr.
+		expectErrContains []string
+	}
+
+	for _, tc := range []testCase{
+		{
+			name: "dump database with table",
+			dynamicArgs: []dArgGenerator{
+				tempDsnDArg(suite.pgEngine, "dsn", []string{
+					"CREATE TABLE foobar(id INT PRIMARY KEY, name TEXT NOT NULL)",
+				}),
+			},
+			outputContains: []string{
+				"CREATE TABLE",
+				"foobar",
+				"id",
+				"name",
+			},
+		},
+	} {
+		suite.Run(tc.name, func() {
+			suite.runCmdWithAssertions(runCmdWithAssertionsParams{
+				args:              append([]string{"dump"}, tc.args...),
+				dynamicArgs:       tc.dynamicArgs,
+				outputContains:    tc.outputContains,
+				expectErrContains: tc.expectErrContains,
+			})
+		})
+	}
+}

cmd/pg-schema-diff/main.go

@@ -13,6 +13,7 @@ func buildRootCmd() *cobra.Command {
 	}
 	rootCmd.AddCommand(buildPlanCmd())
 	rootCmd.AddCommand(buildApplyCmd())
+	rootCmd.AddCommand(buildDumpCmd())
 	rootCmd.AddCommand(buildVersionCmd())
 	return rootCmd
 }