Merge pull request #170 from go-enry/tiny-fixes-2

test: fix LanguagesByFilename case & improve logging

.github/workflows/pyTest.yml

@@ -21,7 +21,8 @@ jobs:
         if [ -f python/requirements.dev.txt ]; then pip install -r python/requirements.dev.txt; fi
     - name: Build and install package
       run: |
-        pip install -e python
+        pip install setuptools wheel
+        pip -v install --no-use-pep517 -e python
     - name: Test
       run: |
         pytest python/

common_test.go

@@ -4,12 +4,12 @@ import (
 	"fmt"
 	"io/ioutil"
 	"os"
-	"os/exec"
 	"path/filepath"
 	"strings"
 	"testing"
 
 	"github.com/go-enry/go-enry/v2/data"
+	"github.com/go-enry/go-enry/v2/internal/tests"
 
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
@@ -19,43 +19,8 @@ import (
 const linguistURL = "https://github.com/github/linguist.git"
 const linguistClonedEnvVar = "ENRY_TEST_REPO"
 
-// not a part of the test Suite as benchmark does not use testify
 func maybeCloneLinguist() (string, bool, error) {
-	var err error
+	return tests.MaybeCloneLinguist(linguistClonedEnvVar, linguistURL, data.LinguistCommit)
-	linguistTmpDir := os.Getenv(linguistClonedEnvVar)
-	isCleanupNeeded := false
-	isLinguistCloned := linguistTmpDir != ""
-	if !isLinguistCloned {
-		linguistTmpDir, err = ioutil.TempDir("", "linguist-")
-		if err != nil {
-			return "", false, err
-		}
-
-		isCleanupNeeded = true
-		cmd := exec.Command("git", "clone", "--depth", "100", linguistURL, linguistTmpDir)
-		if err := cmd.Run(); err != nil {
-			return linguistTmpDir, isCleanupNeeded, err
-		}
-	}
-
-	cwd, err := os.Getwd()
-	if err != nil {
-		return linguistTmpDir, isCleanupNeeded, err
-	}
-
-	if err = os.Chdir(linguistTmpDir); err != nil {
-		return linguistTmpDir, isCleanupNeeded, err
-	}
-
-	cmd := exec.Command("git", "checkout", data.LinguistCommit)
-	if err := cmd.Run(); err != nil {
-		return linguistTmpDir, isCleanupNeeded, err
-	}
-
-	if err = os.Chdir(cwd); err != nil {
-		return linguistTmpDir, isCleanupNeeded, err
-	}
-	return linguistTmpDir, isCleanupNeeded, nil
 }
 
 type enryBaseTestSuite struct {
@@ -275,7 +240,7 @@ func (s *enryTestSuite) TestGetLanguagesByFilename() {
 		{name: "TestGetLanguagesByFilename_7", filename: "_vimrc", expected: []string{"Vim Script"}},
 		{name: "TestGetLanguagesByFilename_8", filename: "pom.xml", expected: []string{"Maven POM"}},
 		{name: "TestGetLanguagesByFilename_9", filename: "", expected: nil},
-		{name: "TestGetLanguagesByFilename_10", filename: "hi.py", expected: []string{"Python"}},
+		{name: "TestGetLanguagesByFilename_10", filename: "hi.py", expected: nil}, // LanguagesByFilename is only for the well-known file names (.gitignore, etc)
 	}
 
 	for _, test := range tests {

internal/code-generator/generator/generator_test.go

@@ -5,11 +5,12 @@ import (
 	"fmt"
 	"io/ioutil"
 	"os"
-	"os/exec"
 	"path/filepath"
 	"strings"
 	"testing"
 
+	"github.com/go-enry/go-enry/v2/data"
+	"github.com/go-enry/go-enry/v2/internal/tests"
 	"github.com/go-enry/go-enry/v2/internal/tokenizer"
 
 	"github.com/stretchr/testify/assert"
@@ -18,8 +19,8 @@ import (
 )
 
 var (
-	linguistURL          = "https://github.com/github/linguist.git"
 	linguistClonedEnvVar = "ENRY_TEST_REPO"
+	linguistURL          = "https://github.com/github/linguist.git"
 	commit               = "bf853f1c663903e3ee35935189760191f1c45e1c"
 	samplesDir           = "samples"
 	languagesFile        = filepath.Join("lib", "linguist", "languages.yml")
@@ -120,38 +121,11 @@ func Test_GeneratorTestSuite(t *testing.T) {
 	suite.Run(t, new(GeneratorTestSuite))
 }
 
-func (s *GeneratorTestSuite) maybeCloneLinguist() {
-	var err error
-	s.tmpLinguistDir = os.Getenv(linguistClonedEnvVar)
-	isLinguistCloned := s.tmpLinguistDir != ""
-	if !isLinguistCloned {
-		s.tmpLinguistDir, err = ioutil.TempDir("", "linguist-")
-		require.NoError(s.T(), err)
-
-		s.T().Logf("Cloning Linguist repo to '%s' as %s was not set\n",
-			s.tmpLinguistDir, linguistClonedEnvVar)
-		cmd := exec.Command("git", "clone", "--depth", "100", linguistURL, s.tmpLinguistDir)
-		err = cmd.Run()
-		require.NoError(s.T(), err)
-		s.isCleanupNeeded = true
-	}
-
-	cwd, err := os.Getwd()
-	require.NoError(s.T(), err)
-
-	err = os.Chdir(s.tmpLinguistDir)
-	require.NoError(s.T(), err)
-
-	cmd := exec.Command("git", "checkout", commit)
-	err = cmd.Run()
-	require.NoError(s.T(), err)
-
-	err = os.Chdir(cwd)
-	require.NoError(s.T(), err)
-}
-
 func (s *GeneratorTestSuite) SetupSuite() {
-	s.maybeCloneLinguist()
+	var err error
+	s.tmpLinguistDir, s.isCleanupNeeded, err = tests.MaybeCloneLinguist(linguistClonedEnvVar, linguistURL, data.LinguistCommit)
+	require.NoError(s.T(), err)
+
 	s.testCases = []testCase{
 		{
 			name:        "Extensions()",

internal/tests/utils.go

@@ -0,0 +1,48 @@
+package tests
+
+import (
+	"fmt"
+	"io/ioutil"
+	"os"
+	"os/exec"
+)
+
+// Re-used by the packages: enry (test), enry (benchmark) and code-generator (test).
+// Does not rely on testify, panics on errors so that there always is a trace to the caller.
+func MaybeCloneLinguist(envVar, url, commit string) (string, bool, error) {
+	var err error
+	linguistTmpDir := os.Getenv(envVar)
+	isCleanupNeeded := false
+	isLinguistCloned := linguistTmpDir != ""
+	if !isLinguistCloned {
+		linguistTmpDir, err = ioutil.TempDir("", "linguist-")
+		if err != nil {
+			return "", false, err
+		}
+
+		isCleanupNeeded = true
+		cmd := exec.Command("git", "clone", "--depth", "150", url, linguistTmpDir)
+		if err := cmd.Run(); err != nil {
+			panic(fmt.Errorf("%s: %w", cmd.String(), err))
+		}
+	}
+
+	cwd, err := os.Getwd()
+	if err != nil {
+		panic(err)
+	}
+
+	if err = os.Chdir(linguistTmpDir); err != nil {
+		panic(err)
+	}
+
+	cmd := exec.Command("git", "checkout", commit)
+	if err := cmd.Run(); err != nil {
+		panic(fmt.Errorf("%s: %w", cmd.String(), err))
+	}
+
+	if err = os.Chdir(cwd); err != nil {
+		panic(fmt.Errorf("%s: %w", cmd.String(), err))
+	}
+	return linguistTmpDir, isCleanupNeeded, nil
+}