Add binary files support (#503)

internal/git/mime.go

@@ -0,0 +1,89 @@
+package git
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/gabriel-vasile/mimetype"
+)
+
+type MimeType struct {
+	ContentType string
+}
+
+func (mt MimeType) IsText() bool {
+	return strings.Contains(mt.ContentType, "text/")
+}
+
+func (mt MimeType) IsCSV() bool {
+	return strings.Contains(mt.ContentType, "text/csv")
+}
+
+func (mt MimeType) IsImage() bool {
+	return strings.Contains(mt.ContentType, "image/")
+}
+
+func (mt MimeType) IsSVG() bool {
+	return strings.Contains(mt.ContentType, "image/svg+xml")
+}
+
+func (mt MimeType) IsPDF() bool {
+	return strings.Contains(mt.ContentType, "application/pdf")
+}
+
+func (mt MimeType) IsAudio() bool {
+	return strings.Contains(mt.ContentType, "audio/")
+}
+
+func (mt MimeType) IsVideo() bool {
+	return strings.Contains(mt.ContentType, "video/")
+}
+
+func (mt MimeType) CanBeHighlighted() bool {
+	return mt.IsText() && !mt.IsCSV()
+}
+
+func (mt MimeType) CanBeEmbedded() bool {
+	return mt.IsImage() || mt.IsPDF() || mt.IsAudio() || mt.IsVideo()
+}
+
+func (mt MimeType) CanBeRendered() bool {
+	return mt.IsText() || mt.IsImage() || mt.IsSVG() || mt.IsPDF() || mt.IsAudio() || mt.IsVideo()
+}
+
+func (mt MimeType) CanBeEdited() bool {
+	return mt.IsText() || mt.IsSVG()
+}
+
+func (mt MimeType) RenderType() string {
+	t := strings.Split(mt.ContentType, "/")
+	str := ""
+	if len(t) == 2 {
+		str = fmt.Sprintf("(%s)", strings.ToUpper(t[1]))
+	}
+
+	// More user friendly description
+	if mt.IsImage() || mt.IsSVG() {
+		return fmt.Sprintf("Image %s", str)
+	}
+	if mt.IsAudio() {
+		return fmt.Sprintf("Audio %s", str)
+	}
+	if mt.IsVideo() {
+		return fmt.Sprintf("Video %s", str)
+	}
+	if mt.IsPDF() {
+		return "PDF"
+	}
+	if mt.IsCSV() {
+		return "CSV"
+	}
+	if mt.IsText() {
+		return "Text"
+	}
+	return "Binary"
+}
+
+func DetectMimeType(data []byte) MimeType {
+	return MimeType{mimetype.Detect(data).String()}
+}

internal/git/output_parser.go

@@ -3,27 +3,23 @@ package git
 import (
 	"bufio"
 	"bytes"
-	"encoding/csv"
 	"fmt"
 	"io"
+	"regexp"
 	"strings"
 )
 
 type File struct {
-	Filename    string `json:"filename"`
-	Size        uint64 `json:"size"`
-	HumanSize   string `json:"human_size"`
-	OldFilename string `json:"-"`
-	Content     string `json:"content"`
-	Truncated   bool   `json:"truncated"`
-	IsCreated   bool   `json:"-"`
-	IsDeleted   bool   `json:"-"`
-}
-
-type CsvFile struct {
-	File
-	Header []string
-	Rows   [][]string
+	Filename    string   `json:"filename"`
+	Size        uint64   `json:"size"`
+	HumanSize   string   `json:"human_size"`
+	OldFilename string   `json:"-"`
+	Content     string   `json:"content"`
+	Truncated   bool     `json:"truncated"`
+	IsCreated   bool     `json:"-"`
+	IsDeleted   bool     `json:"-"`
+	IsBinary    bool     `json:"-"`
+	MimeType    MimeType `json:"-"`
 }
 
 type Commit struct {
@@ -62,6 +58,8 @@ func truncateCommandOutput(out io.Reader, maxBytes int64) (string, bool, error)
 	return string(buf), truncated, nil
 }
 
+var reLogBinaryNames = regexp.MustCompile(`Binary files (.+) and (.+) differ`)
+
 // inspired from https://github.com/go-gitea/gitea/blob/main/services/gitdiff/gitdiff.go
 func parseLog(out io.Reader, maxFiles int, maxBytes int) ([]*Commit, error) {
 	var commits []*Commit
@@ -206,6 +204,20 @@ loopLog:
 						currentFile.IsCreated = true
 					case strings.HasPrefix(line, "deleted file"):
 						currentFile.IsDeleted = true
+					case strings.HasPrefix(line, "Binary files"):
+						currentFile.IsBinary = true
+						names := reLogBinaryNames.FindStringSubmatch(line)
+						if names[1][2:] != names[2][2:] {
+							if currentFile.IsCreated {
+								currentFile.Filename = convertOctalToUTF8(names[2])[2:]
+							}
+							if currentFile.IsDeleted {
+								currentFile.Filename = convertOctalToUTF8(names[1])[2:]
+							}
+						} else {
+							currentFile.OldFilename = convertOctalToUTF8(names[1])[2:]
+							currentFile.Filename = convertOctalToUTF8(names[2])[2:]
+						}
 					case strings.HasPrefix(line, "--- "):
 						name := convertOctalToUTF8(line[4 : len(line)-1])
 						if parseRename && currentFile.IsDeleted {
@@ -344,27 +356,3 @@ func skipToNextCommit(input *bufio.Reader) (line string, err error) {
 	}
 	return line, err
 }
-
-func ParseCsv(file *File) (*CsvFile, error) {
-
-	reader := csv.NewReader(strings.NewReader(file.Content))
-	records, err := reader.ReadAll()
-	if err != nil {
-		return nil, err
-	}
-
-	header := records[0]
-	numColumns := len(header)
-
-	for i := 1; i < len(records); i++ {
-		if len(records[i]) != numColumns {
-			return nil, fmt.Errorf("CSV file has invalid row at index %d", i)
-		}
-	}
-
-	return &CsvFile{
-		File:   *file,
-		Header: header,
-		Rows:   records[1:],
-	}, nil
-}