feat(too/cmd/migrate): support parse BUILD.bazel for java migrate (#4335)

Supports parsing java specifc fields from BUILD.bazel and populate to
librarian.yaml in migration. `include_samples` and `rest_numeric_enums`
has majority and missing default of `true`, they are included in reverse
in librarian.yaml, so only a few needs to be explicitly set.

For #4306 
For #4130

Author: zhumin8

doc/config-schema.md

@@ -141,6 +141,15 @@ This document describes the schema for the librarian.yaml.
 | `module_path_version` | string | Is the version of the Go module path. |
 | `nested_module` | string | Is the name of a nested module directory. |
 
+## JavaAPI Configuration
+
+| Field | Type | Description |
+| :--- | :--- | :--- |
+| `additional_protos` | list of string | Is a list of additional proto files to include in generation. |
+| `no_samples` | bool | Determines whether to generate samples for the API. |
+| `path` | string | Is the source path. |
+| `no_rest_numeric_enums` | bool | Determines whether to use numeric enums in REST requests for the API. |
+
 ## JavaModule Configuration
 
 | Field | Type | Description |
@@ -160,6 +169,7 @@ This document describes the schema for the librarian.yaml.
 | `library_type_override` | string | Allows the "library_type" field in .repo-metadata.json to be overridden. |
 | `min_java_version` | int | Is the minimum Java version required. |
 | `name_pretty_override` | string | Allows the "name_pretty" field in .repo-metadata.json to be overridden. |
+| `java_apis` | list of [JavaAPI](#javaapi-configuration) (optional) | Is a list of Java-specific API configurations. |
 | `product_documentation_override` | string | Allows the "product_documentation" field in .repo-metadata.json to be overridden. |
 | `recommended_package` | string | Is the recommended package name. |
 | `billing_not_required` | bool | Indicates whether the API does NOT require billing. This is typically false. |

internal/config/language.go

@@ -488,6 +488,9 @@ type JavaModule struct {
 	// to be overridden.
 	NamePrettyOverride string `yaml:"name_pretty_override,omitempty"`
 
+	// JavaAPIs is a list of Java-specific API configurations.
+	JavaAPIs []*JavaAPI `yaml:"java_apis,omitempty"`
+
 	// ProductDocumentationOverride allows the "product_documentation" field in
 	// .repo-metadata.json to be overridden.
 	ProductDocumentationOverride string `yaml:"product_documentation_override,omitempty"`
@@ -505,3 +508,18 @@ type JavaModule struct {
 	// RpcDocumentation is the URL for the RPC documentation.
 	RpcDocumentation string `yaml:"rpc_documentation,omitempty"`
 }
+
+// JavaAPI represents configuration for a single API within a Java module.
+type JavaAPI struct {
+	// AdditionalProtos is a list of additional proto files to include in generation.
+	AdditionalProtos []string `yaml:"additional_protos,omitempty"`
+
+	// NoSamples determines whether to generate samples for the API.
+	NoSamples bool `yaml:"no_samples,omitempty"`
+
+	// Path is the source path.
+	Path string `yaml:"path,omitempty"`
+
+	// NoRestNumericEnums determines whether to use numeric enums in REST requests for the API.
+	NoRestNumericEnums bool `yaml:"no_rest_numeric_enums,omitempty"`
+}

tool/cmd/migrate/java.go

@@ -29,6 +29,60 @@ const (
 	generationConfigFileName = "generation_config.yaml"
 )
 
+type javaGAPICInfo struct {
+	NoRestNumericEnums bool
+	NoSamples          bool
+	AdditionalProtos   []string
+}
+
+func parseJavaBazel(googleapisDir, dir string) (*javaGAPICInfo, error) {
+	file, err := parseBazel(googleapisDir, dir)
+	if err != nil {
+		return nil, err
+	}
+	if file == nil {
+		return nil, nil
+	}
+	info := &javaGAPICInfo{}
+	// 1. From java_gapic_library
+	if rules := file.Rules("java_gapic_library"); len(rules) > 0 {
+		if len(rules) > 1 {
+			log.Printf("Warning: multiple java_gapic_library in %s/BUILD.bazel, using first", dir)
+		}
+		rule := rules[0]
+		info.NoRestNumericEnums = rule.AttrLiteral("rest_numeric_enums") == "False"
+	}
+	// 2. From java_gapic_assembly_gradle_pkg
+	if rules := file.Rules("java_gapic_assembly_gradle_pkg"); len(rules) > 0 {
+		if len(rules) > 1 {
+			log.Printf("Warning: multiple java_gapic_assembly_gradle_pkg in %s/BUILD.bazel, using first", dir)
+		}
+		rule := rules[0]
+		info.NoSamples = rule.AttrLiteral("include_samples") == "False"
+	}
+	// 3. From proto_library_with_info
+	if rules := file.Rules("proto_library_with_info"); len(rules) > 0 {
+		if len(rules) > 1 {
+			log.Printf("Warning: multiple proto_library_with_info in %s/BUILD.bazel, using first", dir)
+		}
+		rule := rules[0]
+		// Search for specific common resource targets in deps
+		if deps := rule.AttrStrings("deps"); len(deps) > 0 {
+			protoMappings := map[string]string{
+				"//google/cloud:common_resources_proto":  "google/cloud/common_resources.proto",
+				"//google/cloud/location:location_proto": "google/cloud/location/locations.proto",
+				"//google/iam/v1:iam_policy_proto":       "google/iam/v1/iam_policy.proto",
+			}
+			for _, dep := range deps {
+				if protoPath, ok := protoMappings[dep]; ok {
+					info.AdditionalProtos = append(info.AdditionalProtos, protoPath)
+				}
+			}
+		}
+	}
+	return info, nil
+}
+
 // GAPICConfig represents the GAPIC configuration in generation_config.yaml.
 type GAPICConfig struct {
 	ProtoPath string `yaml:"proto_path"`
@@ -73,10 +127,17 @@ func runJavaMigration(ctx context.Context, repoPath string) error {
 	if err != nil {
 		return err
 	}
-	cfg := buildConfig(gen)
+	src, err := fetchSource(ctx)
+	if err != nil {
+		return errFetchSource
+	}
+	cfg := buildConfig(gen, src.Dir)
 	if cfg == nil {
 		return fmt.Errorf("no libraries found to migrate")
 	}
+	// The directory name in Googleapis is present for migration code to look
+	// up API details. It shouldn't be persisted.
+	cfg.Sources.Googleapis.Dir = ""
 	if err := librarian.RunTidyOnConfig(ctx, cfg); err != nil {
 		return errTidyFailed
 	}
@@ -89,18 +150,36 @@ func readGenerationConfig(path string) (*GenerationConfig, error) {
 }
 
 // buildConfig converts a GenerationConfig to a Librarian Config.
-func buildConfig(gen *GenerationConfig) *config.Config {
+func buildConfig(gen *GenerationConfig, googleapisDir string) *config.Config {
 	var libs []*config.Library
 	for _, l := range gen.Libraries {
 		name := l.LibraryName
 		if name == "" {
 			name = l.APIShortName
 		}
 		var apis []*config.API
+		var javaAPIs []*config.JavaAPI
 		for _, g := range l.GAPICs {
-			if g.ProtoPath != "" {
-				apis = append(apis, &config.API{Path: g.ProtoPath})
+			if g.ProtoPath == "" {
+				continue
+			}
+			apis = append(apis, &config.API{Path: g.ProtoPath})
+
+			info, err := parseJavaBazel(googleapisDir, g.ProtoPath)
+			if err != nil {
+				log.Printf("Warning: failed to parse BUILD.bazel for %s: %v", g.ProtoPath, err)
+				continue
+			}
+			if info == nil {
+				continue
+			}
+			javaAPI := &config.JavaAPI{
+				Path:               g.ProtoPath,
+				NoRestNumericEnums: info.NoRestNumericEnums,
+				AdditionalProtos:   info.AdditionalProtos,
+				NoSamples:          info.NoSamples,
 			}
+			javaAPIs = append(javaAPIs, javaAPI)
 		}
 		libs = append(libs, &config.Library{
 			Name:         name,
@@ -119,6 +198,7 @@ func buildConfig(gen *GenerationConfig) *config.Config {
 				ExcludedDependencies:         l.ExcludedDependencies,
 				ExcludedPoms:                 l.ExcludedPoms,
 				ExtraVersionedModules:        l.ExtraVersionedModules,
+				JavaAPIs:                     javaAPIs,
 				GroupID:                      l.GroupID,
 				IssueTrackerOverride:         l.IssueTracker,
 				LibraryTypeOverride:          l.LibraryType,
@@ -139,8 +219,7 @@ func buildConfig(gen *GenerationConfig) *config.Config {
 		Language: "java",
 		Default:  &config.Default{},
 		Sources: &config.Sources{
-			// hardcoded for local testing
-			Googleapis: &config.Source{Dir: "../../googleapis"},
+			Googleapis: &config.Source{Dir: googleapisDir},
 		},
 		Libraries: libs,
 	}

tool/cmd/migrate/java_test.go

@@ -15,6 +15,7 @@
 package main
 
 import (
+	"context"
 	"errors"
 	"os"
 	"testing"
@@ -24,6 +25,13 @@ import (
 )
 
 func TestRunJavaMigration(t *testing.T) {
+	fetchSource = func(ctx context.Context) (*config.Source, error) {
+		return &config.Source{
+			Commit: "abcd123",
+			SHA256: "sha123",
+			Dir:    "../../internal/testdata/googleapis",
+		}, nil
+	}
 	for _, test := range []struct {
 		name     string
 		repoPath string
@@ -88,7 +96,7 @@ func TestBuildConfig(t *testing.T) {
 				Language: "java",
 				Default:  &config.Default{},
 				Sources: &config.Sources{
-					Googleapis: &config.Source{Dir: "../../googleapis"},
+					Googleapis: &config.Source{Dir: "../../internal/testdata/googleapis"},
 				},
 				Libraries: []*config.Library{
 					{
@@ -118,7 +126,7 @@ func TestBuildConfig(t *testing.T) {
 				Language: "java",
 				Default:  &config.Default{},
 				Sources: &config.Sources{
-					Googleapis: &config.Source{Dir: "../../googleapis"},
+					Googleapis: &config.Source{Dir: "../../internal/testdata/googleapis"},
 				},
 				Libraries: []*config.Library{
 					{
@@ -154,7 +162,7 @@ func TestBuildConfig(t *testing.T) {
 				Language: "java",
 				Default:  &config.Default{},
 				Sources: &config.Sources{
-					Googleapis: &config.Source{Dir: "../../googleapis"},
+					Googleapis: &config.Source{Dir: "../../internal/testdata/googleapis"},
 				},
 				Libraries: []*config.Library{
 					{
@@ -215,7 +223,7 @@ func TestBuildConfig(t *testing.T) {
 				Language: "java",
 				Default:  &config.Default{},
 				Sources: &config.Sources{
-					Googleapis: &config.Source{Dir: "../../googleapis"},
+					Googleapis: &config.Source{Dir: "../../internal/testdata/googleapis"},
 				},
 				Libraries: []*config.Library{
 					{
@@ -254,7 +262,48 @@ func TestBuildConfig(t *testing.T) {
 		},
 	} {
 		t.Run(test.name, func(t *testing.T) {
-			got := buildConfig(test.gen)
+			got := buildConfig(test.gen, "../../internal/testdata/googleapis")
+			if diff := cmp.Diff(test.want, got); diff != "" {
+				t.Errorf("mismatch (-want +got):\n%s", diff)
+			}
+		})
+	}
+}
+
+func TestParseJavaBazel(t *testing.T) {
+	for _, test := range []struct {
+		name          string
+		googleapisDir string
+		buildPath     string
+		want          *javaGAPICInfo
+	}{
+		{
+			name:          "success",
+			googleapisDir: "testdata/parse-bazel/success",
+			buildPath:     "google/cloud/bigquery/analyticshub/v1",
+			want: &javaGAPICInfo{
+				NoRestNumericEnums: true,
+				NoSamples:          false,
+				AdditionalProtos: []string{
+					"google/cloud/common_resources.proto",
+				},
+			},
+		},
+		{
+			name:          "no GAPIC rules",
+			googleapisDir: "testdata/parse-bazel/no-gapic-rule",
+			want: &javaGAPICInfo{
+				AdditionalProtos: []string{
+					"google/cloud/common_resources.proto",
+				},
+			},
+		},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			got, err := parseJavaBazel(test.googleapisDir, test.buildPath)
+			if err != nil {
+				t.Fatal(err)
+			}
 			if diff := cmp.Diff(test.want, got); diff != "" {
 				t.Errorf("mismatch (-want +got):\n%s", diff)
 			}

tool/cmd/migrate/legacylibrarian.go

@@ -17,7 +17,6 @@ package main
 import (
 	"cmp"
 	"context"
-	"errors"
 	"fmt"
 	"os"
 	"path/filepath"
@@ -26,7 +25,6 @@ import (
 	"sort"
 	"strings"
 
-	"github.com/bazelbuild/buildtools/build"
 	"github.com/googleapis/librarian/internal/config"
 	"github.com/googleapis/librarian/internal/fetch"
 	"github.com/googleapis/librarian/internal/legacylibrarian/legacyconfig"
@@ -283,7 +281,7 @@ func buildGoLibraries(input *MigrationInput) ([]*config.Library, error) {
 		}
 		// Read Go GAPIC configurations from BUILD.bazel.
 		for _, api := range library.APIs {
-			info, err := parseBazel(input.googleapisDir, api.Path)
+			info, err := parseGoBazel(input.googleapisDir, api.Path)
 			if err != nil {
 				return nil, err
 			}
@@ -379,28 +377,22 @@ func readRepoConfig(path string) (*RepoConfig, error) {
 	return yaml.Read[RepoConfig](configFile)
 }
 
-// parseBazel parses the BUILD.bazel file in the given directory to extract information from
+// parseGoBazel parses the BUILD.bazel file in the given directory to extract information from
 // the go_gapic_library rule.
-func parseBazel(googleapisDir, dir string) (*goGAPICInfo, error) {
-	path := filepath.Join(googleapisDir, dir, "BUILD.bazel")
-	if _, err := os.Stat(path); errors.Is(err, os.ErrNotExist) {
-		// Skip Not Exist error for testing purpose.
-		return nil, nil
-	}
-	data, err := os.ReadFile(path)
+func parseGoBazel(googleapisDir, dir string) (*goGAPICInfo, error) {
+	file, err := parseBazel(googleapisDir, dir)
 	if err != nil {
 		return nil, err
 	}
-	file, err := build.ParseBuild(path, data)
-	if err != nil {
-		return nil, err
+	if file == nil {
+		return nil, nil
 	}
 	rules := file.Rules("go_gapic_library")
 	if len(rules) == 0 {
 		return &goGAPICInfo{DisableGAPIC: true}, nil
 	}
 	if len(rules) > 1 {
-		return nil, fmt.Errorf("file %s contains multiple go_gapic_library rules", path)
+		return nil, fmt.Errorf("%s/BUILD.bazel contains multiple go_gapic_library rules", dir)
 	}
 	rule := rules[0]
 	importPath, clientPkg := parseImportPathFromBuild(rule.AttrString("importpath"))

tool/cmd/migrate/legacylibrarian_test.go

@@ -900,7 +900,7 @@ func TestParseBazel(t *testing.T) {
 		},
 	} {
 		t.Run(test.name, func(t *testing.T) {
-			got, err := parseBazel(test.googleapisDir, test.buildPath)
+			got, err := parseGoBazel(test.googleapisDir, test.buildPath)
 			if err != nil {
 				t.Fatal(err)
 			}
@@ -924,7 +924,7 @@ func TestParseBazel_Error(t *testing.T) {
 		},
 	} {
 		t.Run(test.name, func(t *testing.T) {
-			_, err := parseBazel("", test.dir)
+			_, err := parseGoBazel("", test.dir)
 			if err == nil {
 				t.Fatalf("parseBazel(%q): expected error", test.dir)
 			}