fix(gazelle): embed default stdlib list for non-Bazel Go builds

gazelle/python/BUILD.bazel

@@ -1,5 +1,6 @@
 load("@bazel_gazelle//:def.bzl", "gazelle_binary")
 load("@bazel_skylib//rules:copy_file.bzl", "copy_file")
+load("@bazel_skylib//rules:diff_test.bzl", "diff_test")
 load("@io_bazel_rules_go//go:def.bzl", "go_library", "go_test")
 load(":gazelle_test.bzl", "gazelle_test")
 
@@ -17,15 +18,10 @@ go_library(
         "std_modules.go",
         "target.go",
     ],
-    # NOTE @aignas 2023-12-03: currently gazelle does not support embedding
-    # generated files, but 3.11.txt is generated by a build rule.
-    #
-    # You will get a benign error like when running gazelle locally:
-    # > 8 gazelle: .../rules_python/gazelle/python/std_modules.go:24:3: pattern 3.11.txt: matched no files
-    #
-    # See following for more info:
-    # https://github.com/bazelbuild/bazel-gazelle/issues/1513
-    embedsrcs = ["stdlib_list.txt"],  # keep # TODO: use user-defined version?
+    embedsrcs = [
+        "stdlib_list/default.txt",
+        "stdlib_list/selected.txt",  # keep # TODO: use user-defined version?
+    ],
     importpath = "github.com/bazel-contrib/rules_python/gazelle/python",
     visibility = ["//visibility:public"],
     deps = [
@@ -62,10 +58,16 @@ copy_file(
             "//conditions:default": "@python_stdlib_list//:stdlib_list/lists/3.14.txt",
         },
     ),
-    out = "stdlib_list.txt",
+    out = "stdlib_list/selected.txt",
     allow_symlink = True,
 )
 
+diff_test(
+    name = "stdlib_list_default_test",
+    file1 = "stdlib_list/default.txt",
+    file2 = "@python_stdlib_list//:stdlib_list/lists/3.14.txt",
+)
+
 # gazelle:exclude testdata/
 # gazelle:exclude extensions.bzl
 # Exclude test-only Starlark helper from Gazelle to avoid generating an unnecessary bzl_library

gazelle/python/std_modules.go

@@ -16,22 +16,39 @@ package python
 
 import (
 	"bufio"
-	_ "embed"
-	"strings"
+	"bytes"
+	"embed"
+	"io/fs"
 )
 
 var (
-	//go:embed stdlib_list.txt
-	stdlibList string
+	//go:embed stdlib_list/*.txt
+	stdlibFS   embed.FS
 	stdModules map[string]struct{}
 )
 
 func init() {
-	stdModules = make(map[string]struct{})
-	scanner := bufio.NewScanner(strings.NewReader(stdlibList))
+	var err error
+	stdModules, err = loadStdModules(stdlibFS)
+	if err != nil {
+		panic(err)
+	}
+}
+
+func loadStdModules(fsys fs.ReadFileFS) (map[string]struct{}, error) {
+	data, err := fsys.ReadFile("stdlib_list/selected.txt")
+	if err != nil {
+		data, err = fsys.ReadFile("stdlib_list/default.txt")
+		if err != nil {
+			return nil, err
+		}
+	}
+	modules := make(map[string]struct{})
+	scanner := bufio.NewScanner(bytes.NewReader(data))
 	for scanner.Scan() {
-		stdModules[scanner.Text()] = struct{}{}
+		modules[scanner.Text()] = struct{}{}
 	}
+	return modules, scanner.Err()
 }
 
 func isStdModule(m Module) bool {

gazelle/python/std_modules_test.go

@@ -16,6 +16,7 @@ package python
 
 import (
 	"testing"
+	"testing/fstest"
 
 	"github.com/stretchr/testify/assert"
 )
@@ -25,3 +26,27 @@ func TestIsStdModule(t *testing.T) {
 	assert.True(t, isStdModule(Module{Name: "os.path"}))
 	assert.False(t, isStdModule(Module{Name: "foo"}))
 }
+
+func TestLoadStdModules(t *testing.T) {
+	t.Run("prefers selected.txt when present", func(t *testing.T) {
+		fsys := fstest.MapFS{
+			"stdlib_list/default.txt":  &fstest.MapFile{Data: []byte("default_only\n")},
+			"stdlib_list/selected.txt": &fstest.MapFile{Data: []byte("selected_only\n")},
+		}
+		modules, err := loadStdModules(fsys)
+		if assert.NoError(t, err) {
+			assert.Contains(t, modules, "selected_only")
+			assert.NotContains(t, modules, "default_only")
+		}
+	})
+
+	t.Run("falls back to default.txt when selected.txt is absent", func(t *testing.T) {
+		fsys := fstest.MapFS{
+			"stdlib_list/default.txt": &fstest.MapFile{Data: []byte("default_only\n")},
+		}
+		modules, err := loadStdModules(fsys)
+		if assert.NoError(t, err) {
+			assert.Contains(t, modules, "default_only")
+		}
+	})
+}