Fixes go-gitea#16558 CSV delimiter determiner

modules/csv/csv.go

@@ -9,9 +9,11 @@ import (
 	stdcsv "encoding/csv"
 	"errors"
 	"io"
+	"path/filepath"
 	"regexp"
 	"strings"
 
+	"code.gitea.io/gitea/modules/markup"
 	"code.gitea.io/gitea/modules/translation"
 	"code.gitea.io/gitea/modules/util"
 )
@@ -26,8 +28,8 @@ func CreateReader(input io.Reader, delimiter rune) *stdcsv.Reader {
 	return rd
 }
 
-// CreateReaderAndGuessDelimiter tries to guess the field delimiter from the content and creates a csv.Reader.
-func CreateReaderAndGuessDelimiter(rd io.Reader) (*stdcsv.Reader, error) {
+// CreateReaderAndDetermineDelimiter tries to guess the field delimiter from the content and creates a csv.Reader.
+func CreateReaderAndDetermineDelimiter(ctx *markup.RenderContext, rd io.Reader) (*stdcsv.Reader, error) {
 	var data = make([]byte, 1e4)
 	size, err := rd.Read(data)
 	if err != nil {
@@ -37,7 +39,7 @@ func CreateReaderAndGuessDelimiter(rd io.Reader) (*stdcsv.Reader, error) {
 		return nil, err
 	}
 
-	delimiter := guessDelimiter(data[:size])
+	delimiter := determineDelimiter(ctx, data[:size])
 
 	var newInput io.Reader
 	if size < 1e4 {
@@ -49,6 +51,27 @@ func CreateReaderAndGuessDelimiter(rd io.Reader) (*stdcsv.Reader, error) {
 	return CreateReader(newInput, delimiter), nil
 }
 
+// determineDelimiter takes a RenderContext and if it isn't nil and the Filename has an extension that specifies the delimiter,
+// it is used as the delimiter. Otherwise we call guessDelimiter with the data passed
+func determineDelimiter(ctx *markup.RenderContext, data []byte) rune {
+	extension := ".csv"
+	if ctx != nil {
+		extension = strings.ToLower(filepath.Ext(ctx.Filename))
+	}
+
+	var delimiter rune
+	switch extension {
+	case ".tsv":
+		delimiter = '\t'
+	case ".psv":
+		delimiter = '|'
+	default:
+		delimiter = guessDelimiter(data)
+	}
+
+	return delimiter
+}
+
 // guessDelimiter scores the input CSV data against delimiters, and returns the best match.
 // Reads at most 10k bytes & 10 lines.
 func guessDelimiter(data []byte) rune {

modules/csv/csv_test.go

@@ -17,10 +17,10 @@ func TestCreateReader(t *testing.T) {
 	assert.Equal(t, ',', rd.Comma)
 }
 
-func TestCreateReaderAndGuessDelimiter(t *testing.T) {
+func TestCreateReaderAndDetermineDelimiter(t *testing.T) {
 	input := "a;b;c\n1;2;3\n4;5;6"
 
-	rd, err := CreateReaderAndGuessDelimiter(strings.NewReader(input))
+	rd, err := CreateReaderAndDetermineDelimiter(nil, strings.NewReader(input))
 	assert.NoError(t, err)
 	assert.Equal(t, ';', rd.Comma)
 }
@@ -35,6 +35,7 @@ func TestGuessDelimiter(t *testing.T) {
 		"1,2,3;4,5,6;7,8,9\na;b;c":  ';',
 		"\"1,2,3,4\";\"a\nb\"\nc;d": ';',
 		"<br/>":                     ',',
+		"name\temail\tnote\nJohn Doe\tjohn@doe.com\tThis,note,had,a,lot,of,commas,to,test,delimters": '\t',
 	}
 
 	for k, v := range kases {

routers/web/repo/compare.go

@@ -22,6 +22,7 @@ import (
 	csv_module "code.gitea.io/gitea/modules/csv"
 	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/markup"
 	"code.gitea.io/gitea/modules/setting"
 	"code.gitea.io/gitea/modules/upload"
 	"code.gitea.io/gitea/modules/util"
@@ -105,7 +106,7 @@ func setCsvCompareContext(ctx *context.Context) {
 
 		errTooLarge := errors.New(ctx.Locale.Tr("repo.error.csv.too_large"))
 
-		csvReaderFromCommit := func(c *git.Commit) (*csv.Reader, error) {
+		csvReaderFromCommit := func(ctx *markup.RenderContext, c *git.Commit) (*csv.Reader, error) {
 			blob, err := c.GetBlobByPath(diffFile.Name)
 			if err != nil {
 				return nil, err
@@ -121,14 +122,14 @@ func setCsvCompareContext(ctx *context.Context) {
 			}
 			defer reader.Close()
 
-			return csv_module.CreateReaderAndGuessDelimiter(charset.ToUTF8WithFallbackReader(reader))
+			return csv_module.CreateReaderAndGuessDelimiter(ctx, charset.ToUTF8WithFallbackReader(reader))
 		}
 
-		baseReader, err := csvReaderFromCommit(baseCommit)
+		baseReader, err := csvReaderFromCommit(&markup.RenderContext{Filename: diffFile.OldName}, baseCommit)
 		if err == errTooLarge {
 			return CsvDiffResult{nil, err.Error()}
 		}
-		headReader, err := csvReaderFromCommit(headCommit)
+		headReader, err := csvReaderFromCommit(&markup.RenderContext{Filename: diffFile.Name}, headCommit)
 		if err == errTooLarge {
 			return CsvDiffResult{nil, err.Error()}
 		}

services/gitdiff/csv_test.go

@@ -194,16 +194,16 @@ c,d,e`,
 
 		var baseReader *csv.Reader
 		if len(c.base) > 0 {
-			baseReader, err = csv_module.CreateReaderAndGuessDelimiter(strings.NewReader(c.base))
+			baseReader, err = csv_module.CreateReaderAndDetermineDelimiter(nil, strings.NewReader(c.base))
 			if err != nil {
-				t.Errorf("CreateReaderAndGuessDelimiter failed: %s", err)
+				t.Errorf("CreateReaderAndDetermineDelimiter failed: %s", err)
 			}
 		}
 		var headReader *csv.Reader
 		if len(c.head) > 0 {
-			headReader, err = csv_module.CreateReaderAndGuessDelimiter(strings.NewReader(c.head))
+			headReader, err = csv_module.CreateReaderAndDetermineDelimiter(nil, strings.NewReader(c.head))
 			if err != nil {
-				t.Errorf("CreateReaderAndGuessDelimiter failed: %s", err)
+				t.Errorf("CreateReaderAndDetermineDelimiter failed: %s", err)
 			}
 		}