linguist: Use the colors from go-enry

We're about to go all-in on go-enry for the language detection. Which
means we'll soon no longer can load the language colors from a json file
provided by the linguist gem. We could copy the file, but go-enry
provides colors as well. So use those instead.

This also made it possible so turn the Color() function into a normal
function, instead of a receiver function.

This has a little disadvantage: it's no longer possible to load a custom
set of colors. We believe this isn't used, so we're willing to give up
on that.

3 files changed, 9 insertions, 136 deletions

diff --git a/internal/gitaly/linguist/linguist.go b/internal/gitaly/linguist/linguist.go
index 9e4fa648c..e22b583ab 100644
--- a/internal/gitaly/linguist/linguist.go
+++ b/internal/gitaly/linguist/linguist.go
@@ -8,7 +8,6 @@ import (
 	"io"
 	"os"
 	"os/exec"
-	"path/filepath"
 
 	"github.com/go-enry/go-enry/v2"
 	"github.com/grpc-ecosystem/go-grpc-middleware/logging/logrus/ctxlogrus"
@@ -22,37 +21,19 @@ import (
 	"gitlab.com/gitlab-org/gitaly/v15/internal/metadata/featureflag"
 )
 
-// Language is used to parse Linguist's language.json file.
-type Language struct {
-	Color string `json:"color"`
-}
-
 // ByteCountPerLanguage represents a counter value (bytes) per language.
 type ByteCountPerLanguage map[string]uint64
 
 // Instance is a holder of the defined in the system language settings.
 type Instance struct {
-	cfg      config.Cfg
-	colorMap map[string]Language
+	cfg config.Cfg
 }
 
 // New loads the name->color map from the Linguist gem and returns initialized
 // instance to use back to the caller or an error.
 func New(cfg config.Cfg) (*Instance, error) {
-	jsonReader, err := openLanguagesJSON(cfg)
-	if err != nil {
-		return nil, err
-	}
-	defer jsonReader.Close()
-
-	var colorMap map[string]Language
-	if err := json.NewDecoder(jsonReader).Decode(&colorMap); err != nil {
-		return nil, err
-	}
-
 	return &Instance{
-		cfg:      cfg,
-		colorMap: colorMap,
+		cfg: cfg,
 	}, nil
 }
 
@@ -90,8 +71,8 @@ func (inst *Instance) Stats(ctx context.Context, repo *localrepo.Repo, commitID
 }
 
 // Color returns the color Linguist has assigned to language.
-func (inst *Instance) Color(language string) string {
-	if color := inst.colorMap[language].Color; color != "" {
+func Color(language string) string {
+	if color := enry.GetColor(language); color != "#cccccc" {
 		return color
 	}
 
@@ -119,46 +100,6 @@ func (inst *Instance) startGitLinguist(ctx context.Context, repoPath string, com
 	return internalCmd, nil
 }
 
-func openLanguagesJSON(cfg config.Cfg) (io.ReadCloser, error) {
-	if jsonPath := cfg.Ruby.LinguistLanguagesPath; jsonPath != "" {
-		// This is a fallback for environments where dynamic discovery of the
-		// linguist path via Bundler is not working for some reason, for example
-		// https://gitlab.com/gitlab-org/gitaly/issues/1119.
-		return os.Open(jsonPath)
-	}
-
-	linguistPathSymlink, err := os.CreateTemp("", "gitaly-linguist-path")
-	if err != nil {
-		return nil, err
-	}
-	defer func() { _ = os.Remove(linguistPathSymlink.Name()) }()
-
-	if err := linguistPathSymlink.Close(); err != nil {
-		return nil, err
-	}
-
-	// We use a symlink because we cannot trust Bundler to not print garbage
-	// on its stdout.
-	rubyScript := `FileUtils.ln_sf(Bundler.rubygems.find_name('github-linguist').first.full_gem_path, ARGV.first)`
-	cmd := exec.Command("bundle", "exec", "ruby", "-rfileutils", "-e", rubyScript, linguistPathSymlink.Name())
-	cmd.Dir = cfg.Ruby.Dir
-
-	// We have learned that in practice the command we are about to run is a
-	// canary for Ruby/Bundler configuration problems. Including stderr and
-	// stdout in the gitaly log is useful for debugging such problems.
-	cmd.Stderr = os.Stderr
-	cmd.Stdout = os.Stdout
-
-	if err := cmd.Run(); err != nil {
-		if exitError, ok := err.(*exec.ExitError); ok {
-			err = fmt.Errorf("%v; stderr: %q", exitError, exitError.Stderr)
-		}
-		return nil, err
-	}
-
-	return os.Open(filepath.Join(linguistPathSymlink.Name(), "lib", "linguist", "languages.json"))
-}
-
 func (inst *Instance) enryStats(ctx context.Context, repo *localrepo.Repo, commitID string, catfileCache catfile.Cache) (ByteCountPerLanguage, error) {
 	stats, err := newLanguageStats(repo)
 	if err != nil {
diff --git a/internal/gitaly/linguist/linguist_test.go b/internal/gitaly/linguist/linguist_test.go
index 6e9887074..a3b98859a 100644
--- a/internal/gitaly/linguist/linguist_test.go
+++ b/internal/gitaly/linguist/linguist_test.go
@@ -10,8 +10,6 @@ import (
 	"strings"
 	"testing"
 
-	"github.com/go-enry/go-enry/v2"
-	enrydata "github.com/go-enry/go-enry/v2/data"
 	"github.com/stretchr/testify/require"
 	"gitlab.com/gitlab-org/gitaly/v15/internal/git"
 	"gitlab.com/gitlab-org/gitaly/v15/internal/git/catfile"
@@ -28,56 +26,6 @@ func TestMain(m *testing.M) {
 	testhelper.Run(m)
 }
 
-// TestNew_knownLanguages tests the compatibility between the Ruby and the Go
-// implementation. This test will be removed together with the Ruby implementation.
-func TestNew_knownLanguages(t *testing.T) {
-	t.Parallel()
-
-	cfg := testcfg.Build(t, testcfg.WithRealLinguist())
-
-	linguist, err := New(cfg)
-	require.NoError(t, err)
-
-	t.Run("by name", func(t *testing.T) {
-		linguistLanguages := make([]string, 0, len(linguist.colorMap))
-		for language := range linguist.colorMap {
-			linguistLanguages = append(linguistLanguages, language)
-		}
-
-		enryLanguages := make([]string, 0, len(enrydata.IDByLanguage))
-		for language := range enrydata.IDByLanguage {
-			enryLanguages = append(enryLanguages, language)
-		}
-
-		require.ElementsMatch(t, linguistLanguages, enryLanguages)
-	})
-
-	t.Run("with their color", func(t *testing.T) {
-		exclude := map[string]struct{}{}
-
-		linguistLanguages := make(map[string]string, len(linguist.colorMap))
-		for language, color := range linguist.colorMap {
-			if color.Color == "" {
-				exclude[language] = struct{}{}
-				continue
-			}
-			linguistLanguages[language] = color.Color
-		}
-
-		enryLanguages := make(map[string]string, len(enrydata.IDByLanguage))
-		for language := range enrydata.IDByLanguage {
-			if _, excluded := exclude[language]; excluded {
-				continue
-			}
-
-			color := enry.GetColor(language)
-			enryLanguages[language] = color
-		}
-
-		require.Equal(t, linguistLanguages, enryLanguages)
-	})
-}
-
 func TestInstance_Stats(t *testing.T) {
 	testhelper.NewFeatureSets(featureflag.GoLanguageStats).
 		Run(t, testInstanceStats)
@@ -290,14 +238,9 @@ func TestInstance_Stats_unmarshalJSONError(t *testing.T) {
 	require.False(t, ok, "expected the error not be a json Syntax Error")
 }
 
-func TestInstance_Color(t *testing.T) {
+func TestColor(t *testing.T) {
 	t.Parallel()
 
-	cfg := testcfg.Build(t, testcfg.WithRealLinguist())
-
-	ling, err := New(cfg)
-	require.NoError(t, err)
-
 	for _, tc := range []struct {
 		language      string
 		expectedColor string
@@ -307,27 +250,15 @@ func TestInstance_Color(t *testing.T) {
 		{language: "HTML", expectedColor: "#e34c26"},
 		{language: "Markdown", expectedColor: "#083fa1"},
 		{language: "Javascript", expectedColor: "#75712c"},
-		{language: "SSH Config", expectedColor: "#2d519e"},
+		{language: "SSH Config", expectedColor: "#d1dbe0"},    // grouped into INI by go-enry
 		{language: "Wozzle Wuzzle", expectedColor: "#3adbcf"}, // non-existing language
 	} {
 		t.Run(tc.language, func(t *testing.T) {
-			require.Equal(t, tc.expectedColor, ling.Color(tc.language), "color value for '%v'", tc.language)
+			require.Equal(t, tc.expectedColor, Color(tc.language), "color value for '%v'", tc.language)
 		})
 	}
 }
 
-func TestNew_loadLanguagesCustomPath(t *testing.T) {
-	jsonPath, err := filepath.Abs("testdata/fake-languages.json")
-	require.NoError(t, err)
-
-	cfg := testcfg.Build(t, testcfg.WithBase(config.Cfg{Ruby: config.Ruby{LinguistLanguagesPath: jsonPath}}))
-
-	ling, err := New(cfg)
-	require.NoError(t, err)
-
-	require.Equal(t, "foo color", ling.Color("FooBar"))
-}
-
 // filenameForCache returns the filename where the cache is stored, depending on
 // the feature flag.
 func filenameForCache(ctx context.Context) string {
diff --git a/internal/gitaly/service/commit/languages.go b/internal/gitaly/service/commit/languages.go
index 4363011c0..8fef28db8 100644
--- a/internal/gitaly/service/commit/languages.go
+++ b/internal/gitaly/service/commit/languages.go
@@ -10,6 +10,7 @@ import (
 	"strings"
 
 	"gitlab.com/gitlab-org/gitaly/v15/internal/git"
+	"gitlab.com/gitlab-org/gitaly/v15/internal/gitaly/linguist"
 	"gitlab.com/gitlab-org/gitaly/v15/internal/helper"
 	"gitlab.com/gitlab-org/gitaly/v15/internal/helper/text"
 	"gitlab.com/gitlab-org/gitaly/v15/proto/go/gitalypb"
@@ -61,7 +62,7 @@ func (s *server) CommitLanguages(ctx context.Context, req *gitalypb.CommitLangua
 		l := &gitalypb.CommitLanguagesResponse_Language{
 			Name:  lang,
 			Share: float32(100*count) / float32(total),
-			Color: s.linguist.Color(lang),
+			Color: linguist.Color(lang),
 			Bytes: stats[lang],
 		}
 		resp.Languages = append(resp.Languages, l)