Test blob indexing

indexer/blob.go

@@ -10,8 +10,8 @@ import (
 )
 var (
-	skipTooLargeBlob = fmt.Errorf("Blob should be skipped: Too large")
-	skipBinaryBlob   = fmt.Errorf("Blob should be skipped: binary")
+	SkipTooLargeBlob = fmt.Errorf("Blob should be skipped: Too large")
+	SkipBinaryBlob   = fmt.Errorf("Blob should be skipped: binary")
 )
 const (
@@ -21,9 +21,9 @@ const (
 func isSkipBlobErr(err error) bool {
 	switch err {
-	case skipTooLargeBlob:
+	case SkipTooLargeBlob:
 		return true
-	case skipBinaryBlob:
+	case SkipBinaryBlob:
 		return true
 	}
@@ -60,7 +60,7 @@ func GenerateBlobID(parentID, filename string) string {
 func BuildBlob(file *git.File, parentID, commitSHA string) (*Blob, error) {
 	if file.Size > maxBlobSize {
-		return nil, skipTooLargeBlob
+		return nil, SkipTooLargeBlob
 	}
 	reader, err := file.Blob()
@@ -78,7 +78,7 @@ func BuildBlob(file *git.File, parentID, commitSHA string) (*Blob, error) {
 	}
 	if DetectBinary(b) {
-		return nil, skipBinaryBlob
+		return nil, SkipBinaryBlob
 	}
 	content := tryEncodeBytes(b)

indexer/blob_test.go

+package indexer_test
+
+import (
+	"encoding/json"
+	"io"
+	"io/ioutil"
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+
+	"gitlab.com/gitlab-org/es-git-go/git"
+	"gitlab.com/gitlab-org/es-git-go/indexer"
+)
+
+const (
+	sha = "9876543210987654321098765432109876543210"
+	oid = "0123456789012345678901234567890123456789"
+)
+
+func readerFunc(data string, err error) func() (io.ReadCloser, error) {
+	return func() (io.ReadCloser, error) {
+		return ioutil.NopCloser(strings.NewReader(data)), err
+	}
+}
+
+func validFile(path, content string) *git.File {
+	return &git.File{
+		Path: path,
+		Blob: readerFunc(content, nil),
+		Size: int64(len(content)),
+		Oid:  oid,
+	}
+}
+
+func validBlob(file *git.File, content, language string) *indexer.Blob {
+	return &indexer.Blob{
+		Type:      "blob",
+		ID:        indexer.GenerateBlobID(parentID, file.Path),
+		OID:       oid,
+		RepoID:    parentID,
+		CommitSHA: sha,
+		Content:   content,
+		Path:      file.Path,
+		Filename:  file.Path,
+		Language:  language,
+	}
+}
+
+func TestBuildBlob(t *testing.T) {
+	file := validFile("foo/bar", "foo")
+	expected := validBlob(file, "foo", "Text")
+
+	actual, err := indexer.BuildBlob(file, expected.RepoID, expected.CommitSHA)
+	assert.NoError(t, err)
+
+	assert.Equal(t, expected, actual)
+
+	expectedJSON := `{
+		"commit_sha" : "` + expected.CommitSHA + `",
+		"content"    : "` + expected.Content + `",
+		"file_name"  : "` + expected.Filename + `",
+		"language"   : "` + expected.Language + `",
+		"oid"        : "` + expected.OID + `",
+		"path"       : "` + expected.Path + `",
+		"rid"        : "` + expected.RepoID + `",
+		"type"       : "blob"
+	}`
+
+	actualJSON, err := json.Marshal(actual)
+	assert.NoError(t, err)
+	assert.JSONEq(t, expectedJSON, string(actualJSON))
+}
+
+func TestBuildBlobSkipsLargeBlobs(t *testing.T) {
+	file := validFile("foo/bar", "foo")
+	file.Size = 1024*1024 + 1
+
+	blob, err := indexer.BuildBlob(file, parentID, sha)
+	assert.Error(t, err, indexer.SkipTooLargeBlob)
+	assert.Nil(t, blob)
+}
+
+func TestBuildBlobSkipsBinaryBlobs(t *testing.T) {
+	file := validFile("foo/bar", "foo\x00")
+
+	blob, err := indexer.BuildBlob(file, parentID, sha)
+	assert.Equal(t, err, indexer.SkipBinaryBlob)
+	assert.Nil(t, blob)
+}
+
+func TestBuildBlobDetectsLanguageByFilename(t *testing.T) {
+	file := validFile("Makefile.am", "foo")
+	blob, err := indexer.BuildBlob(file, parentID, sha)
+
+	assert.NoError(t, err)
+	assert.Equal(t, "Makefile", blob.Language)
+}
+
+func TestBuildBlobDetectsLanguageByExtension(t *testing.T) {
+	file := validFile("foo.rb", "foo")
+	blob, err := indexer.BuildBlob(file, parentID, sha)
+
+	assert.NoError(t, err)
+	assert.Equal(t, "Ruby", blob.Language)
+}
+
+func TestGenerateBlobID(t *testing.T) {
+	assert.Equal(t, "projectID_path", indexer.GenerateBlobID("projectID", "path"))
+}