Switch Unicode Escaping to a VSCode-like system (#19990)

This PR rewrites the invisible unicode detection algorithm to more
closely match that of the Monaco editor on the system. It provides a
technique for detecting ambiguous characters and relaxes the detection
of combining marks.

Control characters are in addition detected as invisible in this
implementation whereas they are not on monaco but this is related to
font issues.

Close #19913

Signed-off-by: Andrew Thornton <art27@cantab.net>

modules/charset/ambiguous.go

@@ -0,0 +1,54 @@
+// This file is generated by modules/charset/ambiguous/generate.go DO NOT EDIT
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package charset
+
+import (
+	"sort"
+	"strings"
+	"unicode"
+
+	"code.gitea.io/gitea/modules/translation"
+)
+
+// AmbiguousTablesForLocale provides the table of ambiguous characters for this locale.
+func AmbiguousTablesForLocale(locale translation.Locale) []*AmbiguousTable {
+	key := locale.Language()
+	var table *AmbiguousTable
+	var ok bool
+	for len(key) > 0 {
+		if table, ok = AmbiguousCharacters[key]; ok {
+			break
+		}
+		idx := strings.LastIndexAny(key, "-_")
+		if idx < 0 {
+			key = ""
+		} else {
+			key = key[:idx]
+		}
+	}
+	if table == nil {
+		table = AmbiguousCharacters["_default"]
+	}
+
+	return []*AmbiguousTable{
+		table,
+		AmbiguousCharacters["_common"],
+	}
+}
+
+func isAmbiguous(r rune, confusableTo *rune, tables ...*AmbiguousTable) bool {
+	for _, table := range tables {
+		if !unicode.Is(table.RangeTable, r) {
+			continue
+		}
+		i := sort.Search(len(table.Confusable), func(i int) bool {
+			return table.Confusable[i] >= r
+		})
+		(*confusableTo) = table.With[i]
+		return true
+	}
+	return false
+}

modules/charset/ambiguous/generate.go

@@ -0,0 +1,178 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package main
+
+import (
+	"bytes"
+	"flag"
+	"fmt"
+	"go/format"
+	"os"
+	"sort"
+	"text/template"
+	"unicode"
+
+	"code.gitea.io/gitea/modules/json"
+
+	"golang.org/x/text/unicode/rangetable"
+)
+
+// ambiguous.json provides a one to one mapping of ambiguous characters to other characters
+// See https://github.com/hediet/vscode-unicode-data/blob/main/out/ambiguous.json
+
+type AmbiguousTable struct {
+	Confusable []rune
+	With       []rune
+	Locale     string
+	RangeTable *unicode.RangeTable
+}
+
+type RunePair struct {
+	Confusable rune
+	With       rune
+}
+
+var verbose bool
+
+func main() {
+	flag.Usage = func() {
+		fmt.Fprintf(os.Stderr, `%s: Generate AmbiguousCharacter
+
+Usage: %[1]s [-v] [-o output.go] ambiguous.json
+`, os.Args[0])
+		flag.PrintDefaults()
+	}
+
+	output := ""
+	flag.BoolVar(&verbose, "v", false, "verbose output")
+	flag.StringVar(&output, "o", "ambiguous_gen.go", "file to output to")
+	flag.Parse()
+	input := flag.Arg(0)
+	if input == "" {
+		input = "ambiguous.json"
+	}
+
+	bs, err := os.ReadFile(input)
+	if err != nil {
+		fatalf("Unable to read: %s Err: %v", input, err)
+	}
+
+	var unwrapped string
+	if err := json.Unmarshal(bs, &unwrapped); err != nil {
+		fatalf("Unable to unwrap content in: %s Err: %v", input, err)
+	}
+
+	fromJSON := map[string][]uint32{}
+	if err := json.Unmarshal([]byte(unwrapped), &fromJSON); err != nil {
+		fatalf("Unable to unmarshal content in: %s Err: %v", input, err)
+	}
+
+	tables := make([]*AmbiguousTable, 0, len(fromJSON))
+	for locale, chars := range fromJSON {
+		table := &AmbiguousTable{Locale: locale}
+		table.Confusable = make([]rune, 0, len(chars)/2)
+		table.With = make([]rune, 0, len(chars)/2)
+		pairs := make([]RunePair, len(chars)/2)
+		for i := 0; i < len(chars); i += 2 {
+			pairs[i/2].Confusable, pairs[i/2].With = rune(chars[i]), rune(chars[i+1])
+		}
+		sort.Slice(pairs, func(i, j int) bool {
+			return pairs[i].Confusable < pairs[j].Confusable
+		})
+		for _, pair := range pairs {
+			table.Confusable = append(table.Confusable, pair.Confusable)
+			table.With = append(table.With, pair.With)
+		}
+		table.RangeTable = rangetable.New(table.Confusable...)
+		tables = append(tables, table)
+	}
+	sort.Slice(tables, func(i, j int) bool {
+		return tables[i].Locale < tables[j].Locale
+	})
+	data := map[string]interface{}{
+		"Tables": tables,
+	}
+
+	if err := runTemplate(generatorTemplate, output, &data); err != nil {
+		fatalf("Unable to run template: %v", err)
+	}
+}
+
+func runTemplate(t *template.Template, filename string, data interface{}) error {
+	buf := bytes.NewBuffer(nil)
+	if err := t.Execute(buf, data); err != nil {
+		return fmt.Errorf("unable to execute template: %w", err)
+	}
+	bs, err := format.Source(buf.Bytes())
+	if err != nil {
+		verbosef("Bad source:\n%s", buf.String())
+		return fmt.Errorf("unable to format source: %w", err)
+	}
+	file, err := os.Create(filename)
+	if err != nil {
+		return fmt.Errorf("failed to create file %s because %w", filename, err)
+	}
+	defer file.Close()
+	_, err = file.Write(bs)
+	if err != nil {
+		return fmt.Errorf("unable to write generated source: %w", err)
+	}
+	return nil
+}
+
+var generatorTemplate = template.Must(template.New("ambiguousTemplate").Parse(`// This file is generated by modules/charset/ambiguous/generate.go DO NOT EDIT
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package charset
+
+import "unicode"
+
+// This file is generated from https://github.com/hediet/vscode-unicode-data/blob/main/out/ambiguous.json
+
+// AmbiguousTable matches a confusable rune with its partner for the Locale
+type AmbiguousTable struct {
+	Confusable []rune
+	With       []rune
+	Locale     string
+	RangeTable *unicode.RangeTable
+}
+
+// AmbiguousCharacters provides a map by locale name to the confusable characters in that locale
+var AmbiguousCharacters = map[string]*AmbiguousTable{
+	{{range .Tables}}{{printf "%q:" .Locale}} {
+			Confusable: []rune{ {{range .Confusable}}{{.}},{{end}} },
+			With: []rune{ {{range .With}}{{.}},{{end}} },
+			Locale: {{printf "%q" .Locale}},
+			RangeTable: &unicode.RangeTable{
+				R16: []unicode.Range16{
+			{{range .RangeTable.R16 }}		{Lo:{{.Lo}}, Hi:{{.Hi}}, Stride: {{.Stride}}},
+			{{end}}	},
+				R32: []unicode.Range32{
+			{{range .RangeTable.R32}}		{Lo:{{.Lo}}, Hi:{{.Hi}}, Stride: {{.Stride}}},
+			{{end}}	},
+				LatinOffset: {{.RangeTable.LatinOffset}},
+			},
+		},
+	{{end}}
+}
+
+`))
+
+func logf(format string, args ...interface{}) {
+	fmt.Fprintf(os.Stderr, format+"\n", args...)
+}
+
+func verbosef(format string, args ...interface{}) {
+	if verbose {
+		logf(format, args...)
+	}
+}
+
+func fatalf(format string, args ...interface{}) {
+	logf("fatal: "+format+"\n", args...)
+	os.Exit(1)
+}

modules/charset/ambiguous_gen_test.go

@@ -0,0 +1,32 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package charset
+
+import (
+	"sort"
+	"testing"
+	"unicode"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestAmbiguousCharacters(t *testing.T) {
+	for locale, ambiguous := range AmbiguousCharacters {
+		assert.Equal(t, locale, ambiguous.Locale)
+		assert.Equal(t, len(ambiguous.Confusable), len(ambiguous.With))
+		assert.True(t, sort.SliceIsSorted(ambiguous.Confusable, func(i, j int) bool {
+			return ambiguous.Confusable[i] < ambiguous.Confusable[j]
+		}))
+
+		for _, confusable := range ambiguous.Confusable {
+			assert.True(t, unicode.Is(ambiguous.RangeTable, confusable))
+			i := sort.Search(len(ambiguous.Confusable), func(j int) bool {
+				return ambiguous.Confusable[j] >= confusable
+			})
+			found := i < len(ambiguous.Confusable) && ambiguous.Confusable[i] == confusable
+			assert.True(t, found, "%c is not in %d", confusable, i)
+		}
+	}
+}

modules/charset/breakwriter.go

@@ -0,0 +1,44 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package charset
+
+import (
+	"bytes"
+	"io"
+)
+
+// BreakWriter wraps an io.Writer to always write '\n' as '<br>'
+type BreakWriter struct {
+	io.Writer
+}
+
+// Write writes the provided byte slice transparently replacing '\n' with '<br>'
+func (b *BreakWriter) Write(bs []byte) (n int, err error) {
+	pos := 0
+	for pos < len(bs) {
+		idx := bytes.IndexByte(bs[pos:], '\n')
+		if idx < 0 {
+			wn, err := b.Writer.Write(bs[pos:])
+			return n + wn, err
+		}
+
+		if idx > 0 {
+			wn, err := b.Writer.Write(bs[pos : pos+idx])
+			n += wn
+			if err != nil {
+				return n, err
+			}
+		}
+
+		if _, err = b.Writer.Write([]byte("<br>")); err != nil {
+			return n, err
+		}
+		pos += idx + 1
+
+		n++
+	}
+
+	return n, err
+}

modules/charset/breakwriter_test.go

@@ -0,0 +1,69 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package charset
+
+import (
+	"strings"
+	"testing"
+)
+
+func TestBreakWriter_Write(t *testing.T) {
+	tests := []struct {
+		name    string
+		kase    string
+		expect  string
+		wantErr bool
+	}{
+		{
+			name:   "noline",
+			kase:   "abcdefghijklmnopqrstuvwxyz",
+			expect: "abcdefghijklmnopqrstuvwxyz",
+		},
+		{
+			name:   "endline",
+			kase:   "abcdefghijklmnopqrstuvwxyz\n",
+			expect: "abcdefghijklmnopqrstuvwxyz<br>",
+		},
+		{
+			name:   "startline",
+			kase:   "\nabcdefghijklmnopqrstuvwxyz",
+			expect: "<br>abcdefghijklmnopqrstuvwxyz",
+		},
+		{
+			name:   "onlyline",
+			kase:   "\n\n\n",
+			expect: "<br><br><br>",
+		},
+		{
+			name:   "empty",
+			kase:   "",
+			expect: "",
+		},
+		{
+			name:   "midline",
+			kase:   "\nabc\ndefghijkl\nmnopqrstuvwxy\nz",
+			expect: "<br>abc<br>defghijkl<br>mnopqrstuvwxy<br>z",
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			buf := &strings.Builder{}
+			b := &BreakWriter{
+				Writer: buf,
+			}
+			n, err := b.Write([]byte(tt.kase))
+			if (err != nil) != tt.wantErr {
+				t.Errorf("BreakWriter.Write() error = %v, wantErr %v", err, tt.wantErr)
+				return
+			}
+			if n != len(tt.kase) {
+				t.Errorf("BreakWriter.Write() = %v, want %v", n, len(tt.kase))
+			}
+			if buf.String() != tt.expect {
+				t.Errorf("BreakWriter.Write() wrote %q, want %v", buf.String(), tt.expect)
+			}
+		})
+	}
+}

modules/charset/escape.go

@@ -1,236 +1,58 @@
-// Copyright 2021 The Gitea Authors. All rights reserved.
+// Copyright 2022 The Gitea Authors. All rights reserved.
 // Use of this source code is governed by a MIT-style
 // license that can be found in the LICENSE file.
 
+//go:generate go run invisible/generate.go -v -o ./invisible_gen.go
+
+//go:generate go run ambiguous/generate.go -v -o ./ambiguous_gen.go ambiguous/ambiguous.json
+
 package charset
 
 import (
-	"bytes"
-	"fmt"
 	"io"
 	"strings"
-	"unicode"
-	"unicode/utf8"
 
-	"golang.org/x/text/unicode/bidi"
+	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/translation"
 )
 
-// EscapeStatus represents the findings of the unicode escaper
-type EscapeStatus struct {
-	Escaped      bool
-	HasError     bool
-	HasBadRunes  bool
-	HasControls  bool
-	HasSpaces    bool
-	HasMarks     bool
-	HasBIDI      bool
-	BadBIDI      bool
-	HasRTLScript bool
-	HasLTRScript bool
-}
-
-// Or combines two EscapeStatus structs into one representing the conjunction of the two
-func (status EscapeStatus) Or(other EscapeStatus) EscapeStatus {
-	st := status
-	st.Escaped = st.Escaped || other.Escaped
-	st.HasError = st.HasError || other.HasError
-	st.HasBadRunes = st.HasBadRunes || other.HasBadRunes
-	st.HasControls = st.HasControls || other.HasControls
-	st.HasSpaces = st.HasSpaces || other.HasSpaces
-	st.HasMarks = st.HasMarks || other.HasMarks
-	st.HasBIDI = st.HasBIDI || other.HasBIDI
-	st.BadBIDI = st.BadBIDI || other.BadBIDI
-	st.HasRTLScript = st.HasRTLScript || other.HasRTLScript
-	st.HasLTRScript = st.HasLTRScript || other.HasLTRScript
-	return st
-}
+// RuneNBSP is the codepoint for NBSP
+const RuneNBSP = 0xa0
 
-// EscapeControlString escapes the unicode control sequences in a provided string and returns the findings as an EscapeStatus and the escaped string
-func EscapeControlString(text string) (EscapeStatus, string) {
+// EscapeControlHTML escapes the unicode control sequences in a provided html document
+func EscapeControlHTML(text string, locale translation.Locale, allowed ...rune) (escaped *EscapeStatus, output string) {
 	sb := &strings.Builder{}
-	escaped, _ := EscapeControlReader(strings.NewReader(text), sb)
-	return escaped, sb.String()
-}
+	outputStream := &HTMLStreamerWriter{Writer: sb}
+	streamer := NewEscapeStreamer(locale, outputStream, allowed...).(*escapeStreamer)
 
-// EscapeControlBytes escapes the unicode control sequences  a provided []byte and returns the findings as an EscapeStatus and the escaped []byte
-func EscapeControlBytes(text []byte) (EscapeStatus, []byte) {
-	buf := &bytes.Buffer{}
-	escaped, _ := EscapeControlReader(bytes.NewReader(text), buf)
-	return escaped, buf.Bytes()
+	if err := StreamHTML(strings.NewReader(text), streamer); err != nil {
+		streamer.escaped.HasError = true
+		log.Error("Error whilst escaping: %v", err)
+	}
+	return streamer.escaped, sb.String()
 }
 
-// EscapeControlReader escapes the unicode control sequences  a provided Reader writing the escaped output to the output and returns the findings as an EscapeStatus and an error
-func EscapeControlReader(text io.Reader, output io.Writer) (escaped EscapeStatus, err error) {
-	buf := make([]byte, 4096)
-	readStart := 0
-	runeCount := 0
-	var n int
-	var writePos int
-
-	lineHasBIDI := false
-	lineHasRTLScript := false
-	lineHasLTRScript := false
-
-readingloop:
-	for err == nil {
-		n, err = text.Read(buf[readStart:])
-		bs := buf[:n+readStart]
-		n = len(bs)
-		i := 0
+// EscapeControlReaders escapes the unicode control sequences in a provider reader and writer in a locale and returns the findings as an EscapeStatus and the escaped []byte
+func EscapeControlReader(reader io.Reader, writer io.Writer, locale translation.Locale, allowed ...rune) (escaped *EscapeStatus, err error) {
+	outputStream := &HTMLStreamerWriter{Writer: writer}
+	streamer := NewEscapeStreamer(locale, outputStream, allowed...).(*escapeStreamer)
 
-		for i < len(bs) {
-			r, size := utf8.DecodeRune(bs[i:])
-			runeCount++
-
-			// Now handle the codepoints
-			switch {
-			case r == utf8.RuneError:
-				if writePos < i {
-					if _, err = output.Write(bs[writePos:i]); err != nil {
-						escaped.HasError = true
-						return
-					}
-					writePos = i
-				}
-				// runes can be at most 4 bytes - so...
-				if len(bs)-i <= 3 {
-					// if not request more data
-					copy(buf, bs[i:])
-					readStart = n - i
-					writePos = 0
-					continue readingloop
-				}
-				// this is a real broken rune
-				escaped.HasBadRunes = true
-				escaped.Escaped = true
-				if err = writeBroken(output, bs[i:i+size]); err != nil {
-					escaped.HasError = true
-					return
-				}
-				writePos += size
-			case r == '\n':
-				if lineHasBIDI && !lineHasRTLScript && lineHasLTRScript {
-					escaped.BadBIDI = true
-				}
-				lineHasBIDI = false
-				lineHasRTLScript = false
-				lineHasLTRScript = false
-
-			case runeCount == 1 && r == 0xFEFF: // UTF BOM
-				// the first BOM is safe
-			case r == '\r' || r == '\t' || r == ' ':
-				// These are acceptable control characters and space characters
-			case unicode.IsSpace(r):
-				escaped.HasSpaces = true
-				escaped.Escaped = true
-				if writePos < i {
-					if _, err = output.Write(bs[writePos:i]); err != nil {
-						escaped.HasError = true
-						return
-					}
-				}
-				if err = writeEscaped(output, r); err != nil {
-					escaped.HasError = true
-					return
-				}
-				writePos = i + size
-			case unicode.Is(unicode.Bidi_Control, r):
-				escaped.Escaped = true
-				escaped.HasBIDI = true
-				if writePos < i {
-					if _, err = output.Write(bs[writePos:i]); err != nil {
-						escaped.HasError = true
-						return
-					}
-				}
-				lineHasBIDI = true
-				if err = writeEscaped(output, r); err != nil {
-					escaped.HasError = true
-					return
-				}
-				writePos = i + size
-			case unicode.Is(unicode.C, r):
-				escaped.Escaped = true
-				escaped.HasControls = true
-				if writePos < i {
-					if _, err = output.Write(bs[writePos:i]); err != nil {
-						escaped.HasError = true
-						return
-					}
-				}
-				if err = writeEscaped(output, r); err != nil {
-					escaped.HasError = true
-					return
-				}
-				writePos = i + size
-			case unicode.Is(unicode.M, r):
-				escaped.Escaped = true
-				escaped.HasMarks = true
-				if writePos < i {
-					if _, err = output.Write(bs[writePos:i]); err != nil {
-						escaped.HasError = true
-						return
-					}
-				}
-				if err = writeEscaped(output, r); err != nil {
-					escaped.HasError = true
-					return
-				}
-				writePos = i + size
-			default:
-				p, _ := bidi.Lookup(bs[i : i+size])
-				c := p.Class()
-				if c == bidi.R || c == bidi.AL {
-					lineHasRTLScript = true
-					escaped.HasRTLScript = true
-				} else if c == bidi.L {
-					lineHasLTRScript = true
-					escaped.HasLTRScript = true
-				}
-			}
-			i += size
-		}
-		if n > 0 {
-			// we read something...
-			// write everything unwritten
-			if writePos < i {
-				if _, err = output.Write(bs[writePos:i]); err != nil {
-					escaped.HasError = true
-					return
-				}
-			}
-
-			// reset the starting positions for the next read
-			readStart = 0
-			writePos = 0
-		}
-	}
-	if readStart > 0 {
-		// this means that there is an incomplete or broken rune at 0-readStart and we read nothing on the last go round
-		escaped.Escaped = true
-		escaped.HasBadRunes = true
-		if err = writeBroken(output, buf[:readStart]); err != nil {
-			escaped.HasError = true
-			return
-		}
-	}
-	if err == io.EOF {
-		if lineHasBIDI && !lineHasRTLScript && lineHasLTRScript {
-			escaped.BadBIDI = true
-		}
-		err = nil
-		return
+	if err = StreamHTML(reader, streamer); err != nil {
+		streamer.escaped.HasError = true
+		log.Error("Error whilst escaping: %v", err)
 	}
-	escaped.HasError = true
-	return escaped, err
+	return streamer.escaped, err
 }
 
-func writeBroken(output io.Writer, bs []byte) (err error) {
-	_, err = fmt.Fprintf(output, `<span class="broken-code-point">&lt;%X&gt;</span>`, bs)
-	return err
-}
+// EscapeControlString escapes the unicode control sequences in a provided string and returns the findings as an EscapeStatus and the escaped string
+func EscapeControlString(text string, locale translation.Locale, allowed ...rune) (escaped *EscapeStatus, output string) {
+	sb := &strings.Builder{}
+	outputStream := &HTMLStreamerWriter{Writer: sb}
+	streamer := NewEscapeStreamer(locale, outputStream, allowed...).(*escapeStreamer)
 
-func writeEscaped(output io.Writer, r rune) (err error) {
-	_, err = fmt.Fprintf(output, `<span class="escaped-code-point" data-escaped="[U+%04X]"><span class="char">%c</span></span>`, r, r)
-	return err
+	if err := streamer.Text(text); err != nil {
+		streamer.escaped.HasError = true
+		log.Error("Error whilst escaping: %v", err)
+	}
+	return streamer.escaped, sb.String()
 }

modules/charset/escape_status.go

@@ -0,0 +1,28 @@
+// Copyright 2021 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package charset
+
+// EscapeStatus represents the findings of the unicode escaper
+type EscapeStatus struct {
+	Escaped      bool
+	HasError     bool
+	HasBadRunes  bool
+	HasInvisible bool
+	HasAmbiguous bool
+}
+
+// Or combines two EscapeStatus structs into one representing the conjunction of the two
+func (status *EscapeStatus) Or(other *EscapeStatus) *EscapeStatus {
+	st := status
+	if status == nil {
+		st = &EscapeStatus{}
+	}
+	st.Escaped = st.Escaped || other.Escaped
+	st.HasError = st.HasError || other.HasError
+	st.HasBadRunes = st.HasBadRunes || other.HasBadRunes
+	st.HasAmbiguous = st.HasAmbiguous || other.HasAmbiguous
+	st.HasInvisible = st.HasInvisible || other.HasInvisible
+	return st
+}

modules/charset/escape_stream.go

@@ -0,0 +1,297 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package charset
+
+import (
+	"fmt"
+	"regexp"
+	"sort"
+	"strings"
+	"unicode"
+	"unicode/utf8"
+
+	"code.gitea.io/gitea/modules/translation"
+
+	"golang.org/x/net/html"
+)
+
+// VScode defaultWordRegexp
+var defaultWordRegexp = regexp.MustCompile(`(-?\d*\.\d\w*)|([^\` + "`" + `\~\!\@\#\$\%\^\&\*\(\)\-\=\+\[\{\]\}\\\|\;\:\'\"\,\.\<\>\/\?\s\x00-\x1f]+)`)
+
+func NewEscapeStreamer(locale translation.Locale, next HTMLStreamer, allowed ...rune) HTMLStreamer {
+	return &escapeStreamer{
+		escaped:                 &EscapeStatus{},
+		PassthroughHTMLStreamer: *NewPassthroughStreamer(next),
+		locale:                  locale,
+		ambiguousTables:         AmbiguousTablesForLocale(locale),
+		allowed:                 allowed,
+	}
+}
+
+type escapeStreamer struct {
+	PassthroughHTMLStreamer
+	escaped         *EscapeStatus
+	locale          translation.Locale
+	ambiguousTables []*AmbiguousTable
+	allowed         []rune
+}
+
+func (e *escapeStreamer) EscapeStatus() *EscapeStatus {
+	return e.escaped
+}
+
+// Text tells the next streamer there is a text
+func (e *escapeStreamer) Text(data string) error {
+	sb := &strings.Builder{}
+	pos, until, next := 0, 0, 0
+	if len(data) > len(UTF8BOM) && data[:len(UTF8BOM)] == string(UTF8BOM) {
+		_, _ = sb.WriteString(data[:len(UTF8BOM)])
+		pos = len(UTF8BOM)
+	}
+	for pos < len(data) {
+		nextIdxs := defaultWordRegexp.FindStringIndex(data[pos:])
+		if nextIdxs == nil {
+			until = len(data)
+			next = until
+		} else {
+			until, next = nextIdxs[0]+pos, nextIdxs[1]+pos
+		}
+
+		// from pos until until we know that the runes are not \r\t\n or even ' '
+		runes := make([]rune, 0, next-until)
+		positions := make([]int, 0, next-until+1)
+
+		for pos < until {
+			r, sz := utf8.DecodeRune([]byte(data)[pos:])
+			positions = positions[:0]
+			positions = append(positions, pos, pos+sz)
+			types, confusables, _ := e.runeTypes(r)
+			if err := e.handleRunes(data, []rune{r}, positions, types, confusables, sb); err != nil {
+				return err
+			}
+			pos += sz
+		}
+
+		for i := pos; i < next; {
+			r, sz := utf8.DecodeRune([]byte(data)[i:])
+			runes = append(runes, r)
+			positions = append(positions, i)
+			i += sz
+		}
+		positions = append(positions, next)
+		types, confusables, runeCounts := e.runeTypes(runes...)
+		if runeCounts.needsEscape() {
+			if err := e.handleRunes(data, runes, positions, types, confusables, sb); err != nil {
+				return err
+			}
+		} else {
+			_, _ = sb.Write([]byte(data)[pos:next])
+		}
+		pos = next
+	}
+	if sb.Len() > 0 {
+		if err := e.PassthroughHTMLStreamer.Text(sb.String()); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func (e *escapeStreamer) handleRunes(data string, runes []rune, positions []int, types []runeType, confusables []rune, sb *strings.Builder) error {
+	for i, r := range runes {
+		switch types[i] {
+		case brokenRuneType:
+			if sb.Len() > 0 {
+				if err := e.PassthroughHTMLStreamer.Text(sb.String()); err != nil {
+					return err
+				}
+				sb.Reset()
+			}
+			end := positions[i+1]
+			start := positions[i]
+			if err := e.brokenRune([]byte(data)[start:end]); err != nil {
+				return err
+			}
+		case ambiguousRuneType:
+			if sb.Len() > 0 {
+				if err := e.PassthroughHTMLStreamer.Text(sb.String()); err != nil {
+					return err
+				}
+				sb.Reset()
+			}
+			if err := e.ambiguousRune(r, confusables[0]); err != nil {
+				return err
+			}
+			confusables = confusables[1:]
+		case invisibleRuneType:
+			if sb.Len() > 0 {
+				if err := e.PassthroughHTMLStreamer.Text(sb.String()); err != nil {
+					return err
+				}
+				sb.Reset()
+			}
+			if err := e.invisibleRune(r); err != nil {
+				return err
+			}
+		default:
+			_, _ = sb.WriteRune(r)
+		}
+	}
+	return nil
+}
+
+func (e *escapeStreamer) brokenRune(bs []byte) error {
+	e.escaped.Escaped = true
+	e.escaped.HasBadRunes = true
+
+	if err := e.PassthroughHTMLStreamer.StartTag("span", html.Attribute{
+		Key: "class",
+		Val: "broken-code-point",
+	}); err != nil {
+		return err
+	}
+	if err := e.PassthroughHTMLStreamer.Text(fmt.Sprintf("<%X>", bs)); err != nil {
+		return err
+	}
+
+	return e.PassthroughHTMLStreamer.EndTag("span")
+}
+
+func (e *escapeStreamer) ambiguousRune(r, c rune) error {
+	e.escaped.Escaped = true
+	e.escaped.HasAmbiguous = true
+
+	if err := e.PassthroughHTMLStreamer.StartTag("span", html.Attribute{
+		Key: "class",
+		Val: "ambiguous-code-point tooltip",
+	}, html.Attribute{
+		Key: "data-content",
+		Val: e.locale.Tr("repo.ambiguous_character", r, c),
+	}); err != nil {
+		return err
+	}
+	if err := e.PassthroughHTMLStreamer.StartTag("span", html.Attribute{
+		Key: "class",
+		Val: "char",
+	}); err != nil {
+		return err
+	}
+	if err := e.PassthroughHTMLStreamer.Text(string(r)); err != nil {
+		return err
+	}
+	if err := e.PassthroughHTMLStreamer.EndTag("span"); err != nil {
+		return err
+	}
+
+	return e.PassthroughHTMLStreamer.EndTag("span")
+}
+
+func (e *escapeStreamer) invisibleRune(r rune) error {
+	e.escaped.Escaped = true
+	e.escaped.HasInvisible = true
+
+	if err := e.PassthroughHTMLStreamer.StartTag("span", html.Attribute{
+		Key: "class",
+		Val: "escaped-code-point",
+	}, html.Attribute{
+		Key: "data-escaped",
+		Val: fmt.Sprintf("[U+%04X]", r),
+	}); err != nil {
+		return err
+	}
+	if err := e.PassthroughHTMLStreamer.StartTag("span", html.Attribute{
+		Key: "class",
+		Val: "char",
+	}); err != nil {
+		return err
+	}
+	if err := e.PassthroughHTMLStreamer.Text(string(r)); err != nil {
+		return err
+	}
+	if err := e.PassthroughHTMLStreamer.EndTag("span"); err != nil {
+		return err
+	}
+
+	return e.PassthroughHTMLStreamer.EndTag("span")
+}
+
+type runeCountType struct {
+	numBasicRunes                int
+	numNonConfusingNonBasicRunes int
+	numAmbiguousRunes            int
+	numInvisibleRunes            int
+	numBrokenRunes               int
+}
+
+func (counts runeCountType) needsEscape() bool {
+	if counts.numBrokenRunes > 0 {
+		return true
+	}
+	if counts.numBasicRunes == 0 &&
+		counts.numNonConfusingNonBasicRunes > 0 {
+		return false
+	}
+	return counts.numAmbiguousRunes > 0 || counts.numInvisibleRunes > 0
+}
+
+type runeType int
+
+const (
+	basicASCIIRuneType runeType = iota //nolint // <- This is technically deadcode but its self-documenting so it should stay
+	brokenRuneType
+	nonBasicASCIIRuneType
+	ambiguousRuneType
+	invisibleRuneType
+)
+
+func (e *escapeStreamer) runeTypes(runes ...rune) (types []runeType, confusables []rune, runeCounts runeCountType) {
+	types = make([]runeType, len(runes))
+	for i, r := range runes {
+		var confusable rune
+		switch {
+		case r == utf8.RuneError:
+			types[i] = brokenRuneType
+			runeCounts.numBrokenRunes++
+		case r == ' ' || r == '\t' || r == '\n':
+			runeCounts.numBasicRunes++
+		case e.isAllowed(r):
+			if r > 0x7e || r < 0x20 {
+				types[i] = nonBasicASCIIRuneType
+				runeCounts.numNonConfusingNonBasicRunes++
+			} else {
+				runeCounts.numBasicRunes++
+			}
+		case unicode.Is(InvisibleRanges, r):
+			types[i] = invisibleRuneType
+			runeCounts.numInvisibleRunes++
+		case unicode.IsControl(r):
+			types[i] = invisibleRuneType
+			runeCounts.numInvisibleRunes++
+		case isAmbiguous(r, &confusable, e.ambiguousTables...):
+			confusables = append(confusables, confusable)
+			types[i] = ambiguousRuneType
+			runeCounts.numAmbiguousRunes++
+		case r > 0x7e || r < 0x20:
+			types[i] = nonBasicASCIIRuneType
+			runeCounts.numNonConfusingNonBasicRunes++
+		default:
+			runeCounts.numBasicRunes++
+		}
+	}
+	return types, confusables, runeCounts
+}
+
+func (e *escapeStreamer) isAllowed(r rune) bool {
+	if len(e.allowed) == 0 {
+		return false
+	}
+	if len(e.allowed) == 1 {
+		return e.allowed[0] == r
+	}
+
+	return sort.Search(len(e.allowed), func(i int) bool {
+		return e.allowed[i] >= r
+	}) >= 0
+}

modules/charset/escape_test.go

@@ -8,6 +8,8 @@ import (
 	"reflect"
 	"strings"
 	"testing"
+
+	"code.gitea.io/gitea/modules/translation"
 )
 
 type escapeControlTest struct {
@@ -25,37 +27,37 @@ var escapeControlTests = []escapeControlTest{
 		name:   "single line western",
 		text:   "single line western",
 		result: "single line western",
-		status: EscapeStatus{HasLTRScript: true},
+		status: EscapeStatus{},
 	},
 	{
 		name:   "multi line western",
 		text:   "single line western\nmulti line western\n",
 		result: "single line western\nmulti line western\n",
-		status: EscapeStatus{HasLTRScript: true},
+		status: EscapeStatus{},
 	},
 	{
 		name:   "multi line western non-breaking space",
 		text:   "single line western\nmulti line western\n",
 		result: `single line<span class="escaped-code-point" data-escaped="[U+00A0]"><span class="char"> </span></span>western` + "\n" + `multi line<span class="escaped-code-point" data-escaped="[U+00A0]"><span class="char"> </span></span>western` + "\n",
-		status: EscapeStatus{Escaped: true, HasLTRScript: true, HasSpaces: true},
+		status: EscapeStatus{Escaped: true, HasInvisible: true},
 	},
 	{
 		name:   "mixed scripts: western + japanese",
 		text:   "日属秘ぞしちゅ。Then some western.",
 		result: "日属秘ぞしちゅ。Then some western.",
-		status: EscapeStatus{HasLTRScript: true},
+		status: EscapeStatus{},
 	},
 	{
 		name:   "japanese",
 		text:   "日属秘ぞしちゅ。",
 		result: "日属秘ぞしちゅ。",
-		status: EscapeStatus{HasLTRScript: true},
+		status: EscapeStatus{},
 	},
 	{
 		name:   "hebrew",
 		text:   "עד תקופת יוון העתיקה היה העיסוק במתמטיקה תכליתי בלבד: היא שימשה כאוסף של נוסחאות לחישוב קרקע, אוכלוסין וכו'. פריצת הדרך של היוונים, פרט לתרומותיהם הגדולות לידע המתמטי, הייתה בלימוד המתמטיקה כשלעצמה, מתוקף ערכה הרוחני. יחסם של חלק מהיוונים הקדמונים למתמטיקה היה דתי - למשל, הכת שאסף סביבו פיתגורס האמינה כי המתמטיקה היא הבסיס לכל הדברים. היוונים נחשבים ליוצרי מושג ההוכחה המתמטית, וכן לראשונים שעסקו במתמטיקה לשם עצמה, כלומר כתחום מחקרי עיוני ומופשט ולא רק כעזר שימושי. עם זאת, לצדה",
-		result: "עד תקופת יוון העתיקה היה העיסוק במתמטיקה תכליתי בלבד: היא שימשה כאוסף של נוסחאות לחישוב קרקע, אוכלוסין וכו'. פריצת הדרך של היוונים, פרט לתרומותיהם הגדולות לידע המתמטי, הייתה בלימוד המתמטיקה כשלעצמה, מתוקף ערכה הרוחני. יחסם של חלק מהיוונים הקדמונים למתמטיקה היה דתי - למשל, הכת שאסף סביבו פיתגורס האמינה כי המתמטיקה היא הבסיס לכל הדברים. היוונים נחשבים ליוצרי מושג ההוכחה המתמטית, וכן לראשונים שעסקו במתמטיקה לשם עצמה, כלומר כתחום מחקרי עיוני ומופשט ולא רק כעזר שימושי. עם זאת, לצדה",
-		status: EscapeStatus{HasRTLScript: true},
+		result: `עד תקופת <span class="ambiguous-code-point tooltip" data-content="repo.ambiguous_character"><span class="char">י</span></span><span class="ambiguous-code-point tooltip" data-content="repo.ambiguous_character"><span class="char">ו</span></span><span class="ambiguous-code-point tooltip" data-content="repo.ambiguous_character"><span class="char">ו</span></span><span class="ambiguous-code-point tooltip" data-content="repo.ambiguous_character"><span class="char">ן</span></span> העתיקה היה העיסוק במתמטיקה תכליתי בלבד: היא שימשה כאוסף של נוסחאות לחישוב קרקע, אוכלוסין וכו&#39;. פריצת הדרך של היוונים, פרט לתרומותיהם הגדולות לידע המתמטי, הייתה בלימוד המתמטיקה כשלעצמה, מתוקף ערכה הרוחני. יחסם של חלק מהיוונים הקדמונים למתמטיקה היה דתי - למשל, הכת שאסף סביבו פיתגורס האמינה כי המתמטיקה היא הבסיס לכל הדברים. היוונים נחשבים ליוצרי מושג ההוכחה המתמטית, וכן לראשונים שעסקו במתמטיקה לשם עצמה, כלומר כתחום מחקרי עיוני ומופשט ולא רק כעזר שימושי. עם זאת, לצדה`,
+		status: EscapeStatus{Escaped: true, HasAmbiguous: true},
 	},
 	{
 		name: "more hebrew",
@@ -64,12 +66,12 @@ var escapeControlTests = []escapeControlTest{
 			המתמטיקאי הבולט הראשון ביוון העתיקה, ויש האומרים בתולדות האנושות, הוא תאלס (624 לפנה"ס - 546 לפנה"ס בקירוב).[1] לא יהיה זה משולל יסוד להניח שהוא האדם הראשון שהוכיח משפט מתמטי, ולא רק גילה אותו. תאלס הוכיח שישרים מקבילים חותכים מצד אחד של שוקי זווית קטעים בעלי יחסים שווים (משפט תאלס הראשון), שהזווית המונחת על קוטר במעגל היא זווית ישרה (משפט תאלס השני), שהקוטר מחלק את המעגל לשני חלקים שווים, ושזוויות הבסיס במשולש שווה-שוקיים שוות זו לזו. מיוחסות לו גם שיטות למדידת גובהן של הפירמידות בעזרת מדידת צילן ולקביעת מיקומה של ספינה הנראית מן החוף.
 
 			בשנים 582 לפנה"ס עד 496 לפנה"ס, בקירוב, חי מתמטיקאי חשוב במיוחד - פיתגורס. המקורות הראשוניים עליו מועטים, וההיסטוריונים מתקשים להפריד את העובדות משכבת המסתורין והאגדות שנקשרו בו. ידוע שסביבו התקבצה האסכולה הפיתגוראית מעין כת פסבדו-מתמטית שהאמינה ש"הכל מספר", או ליתר דיוק הכל ניתן לכימות, וייחסה למספרים משמעויות מיסטיות. ככל הנראה הפיתגוראים ידעו לבנות את הגופים האפלטוניים, הכירו את הממוצע האריתמטי, הממוצע הגאומטרי והממוצע ההרמוני והגיעו להישגים חשובים נוספים. ניתן לומר שהפיתגוראים גילו את היותו של השורש הריבועי של 2, שהוא גם האלכסון בריבוע שאורך צלעותיו 1, אי רציונלי, אך תגליתם הייתה למעשה רק שהקטעים "חסרי מידה משותפת", ומושג המספר האי רציונלי מאוחר יותר.[2] אזכור ראשון לקיומם של קטעים חסרי מידה משותפת מופיע בדיאלוג "תאיטיטוס" של אפלטון, אך רעיון זה היה מוכר עוד קודם לכן, במאה החמישית לפנה"ס להיפאסוס, בן האסכולה הפיתגוראית, ואולי לפיתגורס עצמו.[3]`,
-		result: `בתקופה מאוחרת יותר, השתמשו היוונים בשיטת סימון מתקדמת יותר, שבה הוצגו המספרים לפי 22 אותיות האלפבית היווני. לסימון המספרים בין 1 ל-9 נקבעו תשע האותיות הראשונות, בתוספת גרש ( ' ) בצד ימין של האות, למעלה; תשע האותיות הבאות ייצגו את העשרות מ-10 עד 90, והבאות את המאות. לסימון הספרות בין 1000 ל-900,000, השתמשו היוונים באותן אותיות, אך הוסיפו לאותיות את הגרש דווקא מצד שמאל של האותיות, למטה. ממיליון ומעלה, כנראה השתמשו היוונים בשני תגים במקום אחד.
+		result: `בתקופה מאוחרת יותר, השתמשו היוונים בשיטת סימון מתקדמת יותר, שבה הוצגו המספרים לפי 22 אותיות האלפבית היווני. לסימון המספרים בין 1 ל-9 נקבעו תשע האותיות הראשונות, בתוספת גרש ( &#39; ) בצד ימין של האות, למעלה; תשע האותיות הבאות ייצגו את העשרות מ-10 עד 90, והבאות את המאות. לסימון הספרות בין 1000 ל-900,000, השתמשו היוונים באותן אותיות, אך הוסיפו לאותיות את הגרש דווקא מצד שמאל של האותיות, למטה. ממיליון ומעלה, כנראה השתמשו היוונים בשני תגים במקום אחד.
 
-			המתמטיקאי הבולט הראשון ביוון העתיקה, ויש האומרים בתולדות האנושות, הוא תאלס (624 לפנה"ס - 546 לפנה"ס בקירוב).[1] לא יהיה זה משולל יסוד להניח שהוא האדם הראשון שהוכיח משפט מתמטי, ולא רק גילה אותו. תאלס הוכיח שישרים מקבילים חותכים מצד אחד של שוקי זווית קטעים בעלי יחסים שווים (משפט תאלס הראשון), שהזווית המונחת על קוטר במעגל היא זווית ישרה (משפט תאלס השני), שהקוטר מחלק את המעגל לשני חלקים שווים, ושזוויות הבסיס במשולש שווה-שוקיים שוות זו לזו. מיוחסות לו גם שיטות למדידת גובהן של הפירמידות בעזרת מדידת צילן ולקביעת מיקומה של ספינה הנראית מן החוף.
+			המתמטיקאי הבולט הראשון ביוון העתיקה, ויש האומרים בתולדות האנושות, הוא תאלס (624 לפנה&#34;<span class="ambiguous-code-point tooltip" data-content="repo.ambiguous_character"><span class="char">ס</span></span> - 546 לפנה&#34;<span class="ambiguous-code-point tooltip" data-content="repo.ambiguous_character"><span class="char">ס</span></span> בקירוב).[1] לא יהיה זה משולל יסוד להניח שהוא האדם הראשון שהוכיח משפט מתמטי, ולא רק גילה אותו. תאלס הוכיח שישרים מקבילים חותכים מצד אחד של שוקי זווית קטעים בעלי יחסים שווים (משפט תאלס הראשון), שהזווית המונחת על קוטר במעגל היא זווית ישרה (משפט תאלס השני), שהקוטר מחלק את המעגל לשני חלקים שווים, ושזוויות הבסיס במשולש שווה-שוקיים שוות זו לזו. מיוחסות לו גם שיטות למדידת גובהן של הפירמידות בעזרת מדידת צילן ולקביעת מיקומה של ספינה הנראית מן החוף.
 
-			בשנים 582 לפנה"ס עד 496 לפנה"ס, בקירוב, חי מתמטיקאי חשוב במיוחד - פיתגורס. המקורות הראשוניים עליו מועטים, וההיסטוריונים מתקשים להפריד את העובדות משכבת המסתורין והאגדות שנקשרו בו. ידוע שסביבו התקבצה האסכולה הפיתגוראית מעין כת פסבדו-מתמטית שהאמינה ש"הכל מספר", או ליתר דיוק הכל ניתן לכימות, וייחסה למספרים משמעויות מיסטיות. ככל הנראה הפיתגוראים ידעו לבנות את הגופים האפלטוניים, הכירו את הממוצע האריתמטי, הממוצע הגאומטרי והממוצע ההרמוני והגיעו להישגים חשובים נוספים. ניתן לומר שהפיתגוראים גילו את היותו של השורש הריבועי של 2, שהוא גם האלכסון בריבוע שאורך צלעותיו 1, אי רציונלי, אך תגליתם הייתה למעשה רק שהקטעים "חסרי מידה משותפת", ומושג המספר האי רציונלי מאוחר יותר.[2] אזכור ראשון לקיומם של קטעים חסרי מידה משותפת מופיע בדיאלוג "תאיטיטוס" של אפלטון, אך רעיון זה היה מוכר עוד קודם לכן, במאה החמישית לפנה"ס להיפאסוס, בן האסכולה הפיתגוראית, ואולי לפיתגורס עצמו.[3]`,
-		status: EscapeStatus{HasRTLScript: true},
+			בשנים 582 לפנה&#34;<span class="ambiguous-code-point tooltip" data-content="repo.ambiguous_character"><span class="char">ס</span></span> עד 496 לפנה&#34;<span class="ambiguous-code-point tooltip" data-content="repo.ambiguous_character"><span class="char">ס</span></span>, בקירוב, חי מתמטיקאי חשוב במיוחד - פיתגורס. המקורות הראשוניים עליו מועטים, וההיסטוריונים מתקשים להפריד את העובדות משכבת המסתורין והאגדות שנקשרו בו. ידוע שסביבו התקבצה האסכולה הפיתגוראית מעין כת פסבדו-מתמטית שהאמינה ש&#34;הכל מספר&#34;, או ליתר דיוק הכל ניתן לכימות, וייחסה למספרים משמעויות מיסטיות. ככל הנראה הפיתגוראים ידעו לבנות את הגופים האפלטוניים, הכירו את הממוצע האריתמטי, הממוצע הגאומטרי והממוצע ההרמוני והגיעו להישגים חשובים נוספים. ניתן לומר שהפיתגוראים גילו את היותו של השורש הריבועי של 2, שהוא גם האלכסון בריבוע שאורך צלעותיו 1, אי רציונלי, אך תגליתם הייתה למעשה רק שהקטעים &#34;חסרי מידה משותפת&#34;, ומושג המספר האי רציונלי מאוחר יותר.[2] אזכור ראשון לקיומם של קטעים חסרי מידה משותפת מופיע בדיאלוג &#34;תאיטיטוס&#34; של אפלטון, אך רעיון זה היה מוכר עוד קודם לכן, במאה החמישית לפנה&#34;<span class="ambiguous-code-point tooltip" data-content="repo.ambiguous_character"><span class="char">ס</span></span> להיפאסוס, בן האסכולה הפיתגוראית, ואולי לפיתגורס עצמו.[3]`,
+		status: EscapeStatus{Escaped: true, HasAmbiguous: true},
 	},
 	{
 		name: "Mixed RTL+LTR",
@@ -79,10 +81,7 @@ then resh (ר), and finally heh (ה) (which should appear leftmost).`,
 		result: `Many computer programs fail to display bidirectional text correctly.
 For example, the Hebrew name Sarah (שרה) is spelled: sin (ש) (which appears rightmost),
 then resh (ר), and finally heh (ה) (which should appear leftmost).`,
-		status: EscapeStatus{
-			HasRTLScript: true,
-			HasLTRScript: true,
-		},
+		status: EscapeStatus{},
 	},
 	{
 		name: "Mixed RTL+LTR+BIDI",
@@ -90,32 +89,27 @@ then resh (ר), and finally heh (ה) (which should appear leftmost).`,
 			For example, the Hebrew name Sarah ` + "\u2067" + `שרה` + "\u2066\n" +
 			`sin (ש) (which appears rightmost), then resh (ר), and finally heh (ה) (which should appear leftmost).`,
 		result: `Many computer programs fail to display bidirectional text correctly.
-			For example, the Hebrew name Sarah <span class="escaped-code-point" data-escaped="[U+2067]"><span class="char">` + "\u2067" + `</span></span>שרה<span class="escaped-code-point" data-escaped="[U+2066]"><span class="char">` + "\u2066" + `</span></span>` + "\n" +
+			For example, the Hebrew name Sarah ` + "\u2067" + `שרה` + "\u2066\n" +
 			`sin (ש) (which appears rightmost), then resh (ר), and finally heh (ה) (which should appear leftmost).`,
-		status: EscapeStatus{
-			Escaped:      true,
-			HasBIDI:      true,
-			HasRTLScript: true,
-			HasLTRScript: true,
-		},
+		status: EscapeStatus{},
 	},
 	{
 		name:   "Accented characters",
 		text:   string([]byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}),
 		result: string([]byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}),
-		status: EscapeStatus{HasLTRScript: true},
+		status: EscapeStatus{},
 	},
 	{
 		name:   "Program",
 		text:   "string([]byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba})",
 		result: "string([]byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba})",
-		status: EscapeStatus{HasLTRScript: true},
+		status: EscapeStatus{},
 	},
 	{
 		name:   "CVE testcase",
 		text:   "if access_level != \"user\u202E \u2066// Check if admin\u2069 \u2066\" {",
-		result: `if access_level != "user<span class="escaped-code-point" data-escaped="[U+202E]"><span class="char">` + "\u202e" + `</span></span> <span class="escaped-code-point" data-escaped="[U+2066]"><span class="char">` + "\u2066" + `</span></span>// Check if admin<span class="escaped-code-point" data-escaped="[U+2069]"><span class="char">` + "\u2069" + `</span></span> <span class="escaped-code-point" data-escaped="[U+2066]"><span class="char">` + "\u2066" + `</span></span>" {`,
-		status: EscapeStatus{Escaped: true, HasBIDI: true, BadBIDI: true, HasLTRScript: true},
+		result: `if access_level != &#34;user<span class="escaped-code-point" data-escaped="[U+202E]"><span class="char">` + "\u202e" + `</span></span> <span class="escaped-code-point" data-escaped="[U+2066]"><span class="char">` + "\u2066" + `</span></span>// Check if admin<span class="escaped-code-point" data-escaped="[U+2069]"><span class="char">` + "\u2069" + `</span></span> <span class="escaped-code-point" data-escaped="[U+2066]"><span class="char">` + "\u2066" + `</span></span>&#34; {`,
+		status: EscapeStatus{Escaped: true, HasInvisible: true},
 	},
 	{
 		name: "Mixed testcase with fail",
@@ -124,10 +118,10 @@ then resh (ר), and finally heh (ה) (which should appear leftmost).`,
 			`sin (ש) (which appears rightmost), then resh (ר), and finally heh (ה) (which should appear leftmost).` +
 			"\nif access_level != \"user\u202E \u2066// Check if admin\u2069 \u2066\" {\n",
 		result: `Many computer programs fail to display bidirectional text correctly.
-			For example, the Hebrew name Sarah <span class="escaped-code-point" data-escaped="[U+2067]"><span class="char">` + "\u2067" + `</span></span>שרה<span class="escaped-code-point" data-escaped="[U+2066]"><span class="char">` + "\u2066" + `</span></span>` + "\n" +
+			For example, the Hebrew name Sarah ` + "\u2067" + `שרה` + "\u2066\n" +
 			`sin (ש) (which appears rightmost), then resh (ר), and finally heh (ה) (which should appear leftmost).` +
-			"\n" + `if access_level != "user<span class="escaped-code-point" data-escaped="[U+202E]"><span class="char">` + "\u202e" + `</span></span> <span class="escaped-code-point" data-escaped="[U+2066]"><span class="char">` + "\u2066" + `</span></span>// Check if admin<span class="escaped-code-point" data-escaped="[U+2069]"><span class="char">` + "\u2069" + `</span></span> <span class="escaped-code-point" data-escaped="[U+2066]"><span class="char">` + "\u2066" + `</span></span>" {` + "\n",
-		status: EscapeStatus{Escaped: true, HasBIDI: true, BadBIDI: true, HasLTRScript: true, HasRTLScript: true},
+			"\n" + `if access_level != &#34;user<span class="escaped-code-point" data-escaped="[U+202E]"><span class="char">` + "\u202e" + `</span></span> <span class="escaped-code-point" data-escaped="[U+2066]"><span class="char">` + "\u2066" + `</span></span>// Check if admin<span class="escaped-code-point" data-escaped="[U+2069]"><span class="char">` + "\u2069" + `</span></span> <span class="escaped-code-point" data-escaped="[U+2066]"><span class="char">` + "\u2066" + `</span></span>&#34; {` + "\n",
+		status: EscapeStatus{Escaped: true, HasInvisible: true},
 	},
 	{
 		// UTF-8/16/32 all use the same codepoint for BOM
@@ -135,15 +129,16 @@ then resh (ר), and finally heh (ה) (which should appear leftmost).`,
 		name:   "UTF BOM",
 		text:   "\xef\xbb\xbftest",
 		result: "\xef\xbb\xbftest",
-		status: EscapeStatus{HasLTRScript: true},
+		status: EscapeStatus{},
 	},
 }
 
 func TestEscapeControlString(t *testing.T) {
 	for _, tt := range escapeControlTests {
 		t.Run(tt.name, func(t *testing.T) {
-			status, result := EscapeControlString(tt.text)
-			if !reflect.DeepEqual(status, tt.status) {
+			locale := translation.NewLocale("en_US")
+			status, result := EscapeControlString(tt.text, locale)
+			if !reflect.DeepEqual(*status, tt.status) {
 				t.Errorf("EscapeControlString() status = %v, wanted= %v", status, tt.status)
 			}
 			if result != tt.result {
@@ -153,20 +148,6 @@ func TestEscapeControlString(t *testing.T) {
 	}
 }
 
-func TestEscapeControlBytes(t *testing.T) {
-	for _, tt := range escapeControlTests {
-		t.Run(tt.name, func(t *testing.T) {
-			status, result := EscapeControlBytes([]byte(tt.text))
-			if !reflect.DeepEqual(status, tt.status) {
-				t.Errorf("EscapeControlBytes() status = %v, wanted= %v", status, tt.status)
-			}
-			if string(result) != tt.result {
-				t.Errorf("EscapeControlBytes()\nresult= %v,\nwanted= %v", result, tt.result)
-			}
-		})
-	}
-}
-
 func TestEscapeControlReader(t *testing.T) {
 	// lets add some control characters to the tests
 	tests := make([]escapeControlTest, 0, len(escapeControlTests)*3)
@@ -184,16 +165,7 @@ func TestEscapeControlReader(t *testing.T) {
 		test.text = addPrefix("\u001E", test.text)
 		test.result = addPrefix(`<span class="escaped-code-point" data-escaped="[U+001E]"><span class="char">`+"\u001e"+`</span></span>`, test.result)
 		test.status.Escaped = true
-		test.status.HasControls = true
-		tests = append(tests, test)
-	}
-
-	for _, test := range escapeControlTests {
-		test.name += " (+Mark)"
-		test.text = addPrefix("\u0300", test.text)
-		test.result = addPrefix(`<span class="escaped-code-point" data-escaped="[U+0300]"><span class="char">`+"\u0300"+`</span></span>`, test.result)
-		test.status.Escaped = true
-		test.status.HasMarks = true
+		test.status.HasInvisible = true
 		tests = append(tests, test)
 	}
 
@@ -201,13 +173,13 @@ func TestEscapeControlReader(t *testing.T) {
 		t.Run(tt.name, func(t *testing.T) {
 			input := strings.NewReader(tt.text)
 			output := &strings.Builder{}
-			status, err := EscapeControlReader(input, output)
+			status, err := EscapeControlReader(input, output, translation.NewLocale("en_US"))
 			result := output.String()
 			if err != nil {
 				t.Errorf("EscapeControlReader(): err = %v", err)
 			}
 
-			if !reflect.DeepEqual(status, tt.status) {
+			if !reflect.DeepEqual(*status, tt.status) {
 				t.Errorf("EscapeControlReader() status = %v, wanted= %v", status, tt.status)
 			}
 			if result != tt.result {
@@ -223,5 +195,5 @@ func TestEscapeControlReader_panic(t *testing.T) {
 	for i := 0; i < 6826; i++ {
 		bs = append(bs, []byte("—")...)
 	}
-	_, _ = EscapeControlBytes(bs)
+	_, _ = EscapeControlString(string(bs), translation.NewLocale("en_US"))
 }

modules/charset/htmlstream.go

@@ -0,0 +1,201 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package charset
+
+import (
+	"fmt"
+	"io"
+
+	"golang.org/x/net/html"
+)
+
+// HTMLStreamer represents a SAX-like interface for HTML
+type HTMLStreamer interface {
+	Error(err error) error
+	Doctype(data string) error
+	Comment(data string) error
+	StartTag(data string, attrs ...html.Attribute) error
+	SelfClosingTag(data string, attrs ...html.Attribute) error
+	EndTag(data string) error
+	Text(data string) error
+}
+
+// PassthroughHTMLStreamer is a passthrough streamer
+type PassthroughHTMLStreamer struct {
+	next HTMLStreamer
+}
+
+func NewPassthroughStreamer(next HTMLStreamer) *PassthroughHTMLStreamer {
+	return &PassthroughHTMLStreamer{next: next}
+}
+
+var _ (HTMLStreamer) = &PassthroughHTMLStreamer{}
+
+// Error tells the next streamer in line that there is an error
+func (p *PassthroughHTMLStreamer) Error(err error) error {
+	return p.next.Error(err)
+}
+
+// Doctype tells the next streamer what the doctype is
+func (p *PassthroughHTMLStreamer) Doctype(data string) error {
+	return p.next.Doctype(data)
+}
+
+// Comment tells the next streamer there is a comment
+func (p *PassthroughHTMLStreamer) Comment(data string) error {
+	return p.next.Comment(data)
+}
+
+// StartTag tells the next streamer there is a starting tag
+func (p *PassthroughHTMLStreamer) StartTag(data string, attrs ...html.Attribute) error {
+	return p.next.StartTag(data, attrs...)
+}
+
+// SelfClosingTag tells the next streamer there is a self-closing tag
+func (p *PassthroughHTMLStreamer) SelfClosingTag(data string, attrs ...html.Attribute) error {
+	return p.next.SelfClosingTag(data, attrs...)
+}
+
+// EndTag tells the next streamer there is a end tag
+func (p *PassthroughHTMLStreamer) EndTag(data string) error {
+	return p.next.EndTag(data)
+}
+
+// Text tells the next streamer there is a text
+func (p *PassthroughHTMLStreamer) Text(data string) error {
+	return p.next.Text(data)
+}
+
+// HTMLStreamWriter acts as a writing sink
+type HTMLStreamerWriter struct {
+	io.Writer
+	err error
+}
+
+// Write implements io.Writer
+func (h *HTMLStreamerWriter) Write(data []byte) (int, error) {
+	if h.err != nil {
+		return 0, h.err
+	}
+	return h.Writer.Write(data)
+}
+
+// Write implements io.StringWriter
+func (h *HTMLStreamerWriter) WriteString(data string) (int, error) {
+	if h.err != nil {
+		return 0, h.err
+	}
+	return h.Writer.Write([]byte(data))
+}
+
+// Error tells the next streamer in line that there is an error
+func (h *HTMLStreamerWriter) Error(err error) error {
+	if h.err == nil {
+		h.err = err
+	}
+	return h.err
+}
+
+// Doctype tells the next streamer what the doctype is
+func (h *HTMLStreamerWriter) Doctype(data string) error {
+	_, h.err = h.WriteString("<!DOCTYPE " + data + ">")
+	return h.err
+}
+
+// Comment tells the next streamer there is a comment
+func (h *HTMLStreamerWriter) Comment(data string) error {
+	_, h.err = h.WriteString("<!--" + data + "-->")
+	return h.err
+}
+
+// StartTag tells the next streamer there is a starting tag
+func (h *HTMLStreamerWriter) StartTag(data string, attrs ...html.Attribute) error {
+	return h.startTag(data, attrs, false)
+}
+
+// SelfClosingTag tells the next streamer there is a self-closing tag
+func (h *HTMLStreamerWriter) SelfClosingTag(data string, attrs ...html.Attribute) error {
+	return h.startTag(data, attrs, true)
+}
+
+func (h *HTMLStreamerWriter) startTag(data string, attrs []html.Attribute, selfclosing bool) error {
+	if _, h.err = h.WriteString("<" + data); h.err != nil {
+		return h.err
+	}
+	for _, attr := range attrs {
+		if _, h.err = h.WriteString(" " + attr.Key + "=\"" + html.EscapeString(attr.Val) + "\""); h.err != nil {
+			return h.err
+		}
+	}
+	if selfclosing {
+		if _, h.err = h.WriteString("/>"); h.err != nil {
+			return h.err
+		}
+	} else {
+		if _, h.err = h.WriteString(">"); h.err != nil {
+			return h.err
+		}
+	}
+	return h.err
+}
+
+// EndTag tells the next streamer there is a end tag
+func (h *HTMLStreamerWriter) EndTag(data string) error {
+	_, h.err = h.WriteString("</" + data + ">")
+	return h.err
+}
+
+// Text tells the next streamer there is a text
+func (h *HTMLStreamerWriter) Text(data string) error {
+	_, h.err = h.WriteString(html.EscapeString(data))
+	return h.err
+}
+
+// StreamHTML streams an html to a provided streamer
+func StreamHTML(source io.Reader, streamer HTMLStreamer) error {
+	tokenizer := html.NewTokenizer(source)
+	for {
+		tt := tokenizer.Next()
+		switch tt {
+		case html.ErrorToken:
+			if tokenizer.Err() != io.EOF {
+				return tokenizer.Err()
+			}
+			return nil
+		case html.DoctypeToken:
+			token := tokenizer.Token()
+			if err := streamer.Doctype(token.Data); err != nil {
+				return err
+			}
+		case html.CommentToken:
+			token := tokenizer.Token()
+			if err := streamer.Comment(token.Data); err != nil {
+				return err
+			}
+		case html.StartTagToken:
+			token := tokenizer.Token()
+			if err := streamer.StartTag(token.Data, token.Attr...); err != nil {
+				return err
+			}
+		case html.SelfClosingTagToken:
+			token := tokenizer.Token()
+			if err := streamer.StartTag(token.Data, token.Attr...); err != nil {
+				return err
+			}
+		case html.EndTagToken:
+			token := tokenizer.Token()
+			if err := streamer.EndTag(token.Data); err != nil {
+				return err
+			}
+		case html.TextToken:
+			token := tokenizer.Token()
+			if err := streamer.Text(token.Data); err != nil {
+				return err
+			}
+		default:
+			return fmt.Errorf("unknown type of token: %d", tt)
+		}
+	}
+}

modules/charset/invisible/generate.go

@@ -0,0 +1,111 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package main
+
+import (
+	"bytes"
+	"flag"
+	"fmt"
+	"go/format"
+	"os"
+	"text/template"
+
+	"golang.org/x/text/unicode/rangetable"
+)
+
+// InvisibleRunes these are runes that vscode has assigned to be invisible
+// See https://github.com/hediet/vscode-unicode-data
+var InvisibleRunes = []rune{
+	9, 10, 11, 12, 13, 32, 127, 160, 173, 847, 1564, 4447, 4448, 6068, 6069, 6155, 6156, 6157, 6158, 7355, 7356, 8192, 8193, 8194, 8195, 8196, 8197, 8198, 8199, 8200, 8201, 8202, 8203, 8204, 8205, 8206, 8207, 8234, 8235, 8236, 8237, 8238, 8239, 8287, 8288, 8289, 8290, 8291, 8292, 8293, 8294, 8295, 8296, 8297, 8298, 8299, 8300, 8301, 8302, 8303, 10240, 12288, 12644, 65024, 65025, 65026, 65027, 65028, 65029, 65030, 65031, 65032, 65033, 65034, 65035, 65036, 65037, 65038, 65039, 65279, 65440, 65520, 65521, 65522, 65523, 65524, 65525, 65526, 65527, 65528, 65532, 78844, 119155, 119156, 119157, 119158, 119159, 119160, 119161, 119162, 917504, 917505, 917506, 917507, 917508, 917509, 917510, 917511, 917512, 917513, 917514, 917515, 917516, 917517, 917518, 917519, 917520, 917521, 917522, 917523, 917524, 917525, 917526, 917527, 917528, 917529, 917530, 917531, 917532, 917533, 917534, 917535, 917536, 917537, 917538, 917539, 917540, 917541, 917542, 917543, 917544, 917545, 917546, 917547, 917548, 917549, 917550, 917551, 917552, 917553, 917554, 917555, 917556, 917557, 917558, 917559, 917560, 917561, 917562, 917563, 917564, 917565, 917566, 917567, 917568, 917569, 917570, 917571, 917572, 917573, 917574, 917575, 917576, 917577, 917578, 917579, 917580, 917581, 917582, 917583, 917584, 917585, 917586, 917587, 917588, 917589, 917590, 917591, 917592, 917593, 917594, 917595, 917596, 917597, 917598, 917599, 917600, 917601, 917602, 917603, 917604, 917605, 917606, 917607, 917608, 917609, 917610, 917611, 917612, 917613, 917614, 917615, 917616, 917617, 917618, 917619, 917620, 917621, 917622, 917623, 917624, 917625, 917626, 917627, 917628, 917629, 917630, 917631, 917760, 917761, 917762, 917763, 917764, 917765, 917766, 917767, 917768, 917769, 917770, 917771, 917772, 917773, 917774, 917775, 917776, 917777, 917778, 917779, 917780, 917781, 917782, 917783, 917784, 917785, 917786, 917787, 917788, 917789, 917790, 917791, 917792, 917793, 917794, 917795, 917796, 917797, 917798, 917799, 917800, 917801, 917802, 917803, 917804, 917805, 917806, 917807, 917808, 917809, 917810, 917811, 917812, 917813, 917814, 917815, 917816, 917817, 917818, 917819, 917820, 917821, 917822, 917823, 917824, 917825, 917826, 917827, 917828, 917829, 917830, 917831, 917832, 917833, 917834, 917835, 917836, 917837, 917838, 917839, 917840, 917841, 917842, 917843, 917844, 917845, 917846, 917847, 917848, 917849, 917850, 917851, 917852, 917853, 917854, 917855, 917856, 917857, 917858, 917859, 917860, 917861, 917862, 917863, 917864, 917865, 917866, 917867, 917868, 917869, 917870, 917871, 917872, 917873, 917874, 917875, 917876, 917877, 917878, 917879, 917880, 917881, 917882, 917883, 917884, 917885, 917886, 917887, 917888, 917889, 917890, 917891, 917892, 917893, 917894, 917895, 917896, 917897, 917898, 917899, 917900, 917901, 917902, 917903, 917904, 917905, 917906, 917907, 917908, 917909, 917910, 917911, 917912, 917913, 917914, 917915, 917916, 917917, 917918, 917919, 917920, 917921, 917922, 917923, 917924, 917925, 917926, 917927, 917928, 917929, 917930, 917931, 917932, 917933, 917934, 917935, 917936, 917937, 917938, 917939, 917940, 917941, 917942, 917943, 917944, 917945, 917946, 917947, 917948, 917949, 917950, 917951, 917952, 917953, 917954, 917955, 917956, 917957, 917958, 917959, 917960, 917961, 917962, 917963, 917964, 917965, 917966, 917967, 917968, 917969, 917970, 917971, 917972, 917973, 917974, 917975, 917976, 917977, 917978, 917979, 917980, 917981, 917982, 917983, 917984, 917985, 917986, 917987, 917988, 917989, 917990, 917991, 917992, 917993, 917994, 917995, 917996, 917997, 917998, 917999,
+}
+
+var verbose bool
+
+func main() {
+	flag.Usage = func() {
+		fmt.Fprintf(os.Stderr, `%s: Generate InvisibleRunesRange
+
+Usage: %[1]s [-v] [-o output.go]
+`, os.Args[0])
+		flag.PrintDefaults()
+	}
+
+	output := ""
+	flag.BoolVar(&verbose, "v", false, "verbose output")
+	flag.StringVar(&output, "o", "invisible_gen.go", "file to output to")
+	flag.Parse()
+
+	// First we filter the runes to remove
+	// <space><tab><newline>
+	filtered := make([]rune, 0, len(InvisibleRunes))
+	for _, r := range InvisibleRunes {
+		if r == ' ' || r == '\t' || r == '\n' {
+			continue
+		}
+		filtered = append(filtered, r)
+	}
+
+	table := rangetable.New(filtered...)
+	if err := runTemplate(generatorTemplate, output, table); err != nil {
+		fatalf("Unable to run template: %v", err)
+	}
+}
+
+func runTemplate(t *template.Template, filename string, data interface{}) error {
+	buf := bytes.NewBuffer(nil)
+	if err := t.Execute(buf, data); err != nil {
+		return fmt.Errorf("unable to execute template: %w", err)
+	}
+	bs, err := format.Source(buf.Bytes())
+	if err != nil {
+		verbosef("Bad source:\n%s", buf.String())
+		return fmt.Errorf("unable to format source: %w", err)
+	}
+	file, err := os.Create(filename)
+	if err != nil {
+		return fmt.Errorf("failed to create file %s because %w", filename, err)
+	}
+	defer file.Close()
+	_, err = file.Write(bs)
+	if err != nil {
+		return fmt.Errorf("unable to write generated source: %w", err)
+	}
+	return nil
+}
+
+var generatorTemplate = template.Must(template.New("invisibleTemplate").Parse(`// This file is generated by modules/charset/invisible/generate.go DO NOT EDIT
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package charset
+
+import "unicode"
+
+var InvisibleRanges = &unicode.RangeTable{
+	R16: []unicode.Range16{
+{{range .R16 }}		{Lo:{{.Lo}}, Hi:{{.Hi}}, Stride: {{.Stride}}},
+{{end}}	},
+	R32: []unicode.Range32{
+{{range .R32}}		{Lo:{{.Lo}}, Hi:{{.Hi}}, Stride: {{.Stride}}},
+{{end}}	},
+	LatinOffset: {{.LatinOffset}},
+}
+`))
+
+func logf(format string, args ...interface{}) {
+	fmt.Fprintf(os.Stderr, format+"\n", args...)
+}
+
+func verbosef(format string, args ...interface{}) {
+	if verbose {
+		logf(format, args...)
+	}
+}
+
+func fatalf(format string, args ...interface{}) {
+	logf("fatal: "+format+"\n", args...)
+	os.Exit(1)
+}

modules/charset/invisible_gen.go

@@ -0,0 +1,37 @@
+// This file is generated by modules/charset/invisible/generate.go DO NOT EDIT
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package charset
+
+import "unicode"
+
+var InvisibleRanges = &unicode.RangeTable{
+	R16: []unicode.Range16{
+		{Lo: 11, Hi: 13, Stride: 1},
+		{Lo: 127, Hi: 160, Stride: 33},
+		{Lo: 173, Hi: 847, Stride: 674},
+		{Lo: 1564, Hi: 4447, Stride: 2883},
+		{Lo: 4448, Hi: 6068, Stride: 1620},
+		{Lo: 6069, Hi: 6155, Stride: 86},
+		{Lo: 6156, Hi: 6158, Stride: 1},
+		{Lo: 7355, Hi: 7356, Stride: 1},
+		{Lo: 8192, Hi: 8207, Stride: 1},
+		{Lo: 8234, Hi: 8239, Stride: 1},
+		{Lo: 8287, Hi: 8303, Stride: 1},
+		{Lo: 10240, Hi: 12288, Stride: 2048},
+		{Lo: 12644, Hi: 65024, Stride: 52380},
+		{Lo: 65025, Hi: 65039, Stride: 1},
+		{Lo: 65279, Hi: 65440, Stride: 161},
+		{Lo: 65520, Hi: 65528, Stride: 1},
+		{Lo: 65532, Hi: 65532, Stride: 1},
+	},
+	R32: []unicode.Range32{
+		{Lo: 78844, Hi: 119155, Stride: 40311},
+		{Lo: 119156, Hi: 119162, Stride: 1},
+		{Lo: 917504, Hi: 917631, Stride: 1},
+		{Lo: 917760, Hi: 917999, Stride: 1},
+	},
+	LatinOffset: 2,
+}

options/locale/locale_en-US.ini

@@ -1035,13 +1035,13 @@ file_view_rendered = View Rendered
 file_view_raw = View Raw
 file_permalink = Permalink
 file_too_large = The file is too large to be shown.
-bidi_bad_header = `This file contains unexpected Bidirectional Unicode characters!`
-bidi_bad_description = `This file contains unexpected Bidirectional Unicode characters that may be processed differently from what appears below. If your use case is intentional and legitimate, you can safely ignore this warning. Use the Escape button to reveal hidden characters.`
-bidi_bad_description_escaped = `This file contains unexpected Bidirectional Unicode characters. Hidden unicode characters are escaped below. Use the Unescape button to show how they render.`
-unicode_header = `This file contains hidden Unicode characters!`
-unicode_description = `This file contains hidden Unicode characters that may be processed differently from what appears below. If your use case is intentional and legitimate, you can safely ignore this warning. Use the Escape button to reveal hidden characters.`
-unicode_description_escaped = `This file contains hidden Unicode characters. Hidden unicode characters are escaped below. Use the Unescape button to show how they render.`
-line_unicode = `This line has hidden unicode characters`
+invisible_runes_header = `This file contains invisible Unicode characters!`
+invisible_runes_description = `This file contains invisible Unicode characters that may be processed differently from what appears below. If your use case is intentional and legitimate, you can safely ignore this warning. Use the Escape button to reveal hidden characters.`
+ambiguous_runes_header = `This file contains ambiguous Unicode characters!`
+ambiguous_runes_description = `This file contains ambiguous Unicode characters that may be confused with others in your current locale. If your use case is intentional and legitimate, you can safely ignore this warning. Use the Escape button to highlight these characters.`
+invisible_runes_line = `This line has invisible unicode characters`
+ambiguous_runes_line = `This line has ambiguous unicode characters`
+ambiguous_character = `%[1]c [U+%04[1]X] is confusable with %[2]c [U+%04[2]X]`
 
 escape_control_characters = Escape
 unescape_control_characters = Unescape

routers/web/repo/blame.go

@@ -40,7 +40,7 @@ type blameRow struct {
 	CommitMessage  string
 	CommitSince    gotemplate.HTML
 	Code           gotemplate.HTML
-	EscapeStatus   charset.EscapeStatus
+	EscapeStatus   *charset.EscapeStatus
 }
 
 // RefBlame render blame page
@@ -235,7 +235,7 @@ func renderBlame(ctx *context.Context, blameParts []git.BlamePart, commitNames m
 	}
 	lines := make([]string, 0)
 	rows := make([]*blameRow, 0)
-	escapeStatus := charset.EscapeStatus{}
+	escapeStatus := &charset.EscapeStatus{}
 
 	i := 0
 	commitCnt := 0
@@ -280,7 +280,7 @@ func renderBlame(ctx *context.Context, blameParts []git.BlamePart, commitNames m
 			fileName := fmt.Sprintf("%v", ctx.Data["FileName"])
 			line = highlight.Code(fileName, language, line)
 
-			br.EscapeStatus, line = charset.EscapeControlString(line)
+			br.EscapeStatus, line = charset.EscapeControlHTML(line, ctx.Locale)
 			br.Code = gotemplate.HTML(line)
 			rows = append(rows, br)
 			escapeStatus = escapeStatus.Or(br.EscapeStatus)

routers/web/repo/lfs.go

@@ -309,7 +309,7 @@ func LFSFileGet(ctx *context.Context) {
 
 		// Building code view blocks with line number on server side.
 		escapedContent := &bytes.Buffer{}
-		ctx.Data["EscapeStatus"], _ = charset.EscapeControlReader(rd, escapedContent)
+		ctx.Data["EscapeStatus"], _ = charset.EscapeControlReader(rd, escapedContent, ctx.Locale)
 
 		var output bytes.Buffer
 		lines := strings.Split(escapedContent.String(), "\n")

routers/web/repo/view.go

@@ -328,35 +328,31 @@ func renderReadmeFile(ctx *context.Context, readmeFile *namedBlob, readmeTreelin
 	if markupType := markup.Type(readmeFile.name); markupType != "" {
 		ctx.Data["IsMarkup"] = true
 		ctx.Data["MarkupType"] = markupType
-		var result strings.Builder
-		err := markup.Render(&markup.RenderContext{
+
+		ctx.Data["EscapeStatus"], ctx.Data["FileContent"], err = markupRender(ctx, &markup.RenderContext{
 			Ctx:          ctx,
 			RelativePath: path.Join(ctx.Repo.TreePath, readmeFile.name), // ctx.Repo.TreePath is the directory not the Readme so we must append the Readme filename (and path).
 			URLPrefix:    readmeTreelink,
 			Metas:        ctx.Repo.Repository.ComposeDocumentMetas(),
 			GitRepo:      ctx.Repo.GitRepo,
-		}, rd, &result)
+		}, rd)
 		if err != nil {
-			log.Error("Render failed: %v then fallback", err)
+			log.Error("Render failed for %s in %-v: %v Falling back to rendering source", readmeFile.name, ctx.Repo.Repository, err)
 			buf := &bytes.Buffer{}
-			ctx.Data["EscapeStatus"], _ = charset.EscapeControlReader(rd, buf)
+			ctx.Data["EscapeStatus"], _ = charset.EscapeControlReader(rd, buf, ctx.Locale)
 			ctx.Data["FileContent"] = strings.ReplaceAll(
 				gotemplate.HTMLEscapeString(buf.String()), "\n", `<br>`,
 			)
-		} else {
-			ctx.Data["EscapeStatus"], ctx.Data["FileContent"] = charset.EscapeControlString(result.String())
 		}
 	} else {
 		ctx.Data["IsRenderedHTML"] = true
 		buf := &bytes.Buffer{}
-		ctx.Data["EscapeStatus"], err = charset.EscapeControlReader(rd, buf)
+		ctx.Data["EscapeStatus"], err = charset.EscapeControlReader(rd, &charset.BreakWriter{Writer: buf}, ctx.Locale, charset.RuneNBSP)
 		if err != nil {
 			log.Error("Read failed: %v", err)
 		}
 
-		ctx.Data["FileContent"] = strings.ReplaceAll(
-			gotemplate.HTMLEscapeString(buf.String()), "\n", `<br>`,
-		)
+		ctx.Data["FileContent"] = buf.String()
 	}
 }
 
@@ -498,32 +494,30 @@ func renderFile(ctx *context.Context, entry *git.TreeEntry, treeLink, rawLink st
 		if markupType != "" && !shouldRenderSource {
 			ctx.Data["IsMarkup"] = true
 			ctx.Data["MarkupType"] = markupType
-			var result strings.Builder
 			if !detected {
 				markupType = ""
 			}
 			metas := ctx.Repo.Repository.ComposeDocumentMetas()
 			metas["BranchNameSubURL"] = ctx.Repo.BranchNameSubURL()
-			err := markup.Render(&markup.RenderContext{
+			ctx.Data["EscapeStatus"], ctx.Data["FileContent"], err = markupRender(ctx, &markup.RenderContext{
 				Ctx:          ctx,
 				Type:         markupType,
 				RelativePath: ctx.Repo.TreePath,
 				URLPrefix:    path.Dir(treeLink),
 				Metas:        metas,
 				GitRepo:      ctx.Repo.GitRepo,
-			}, rd, &result)
+			}, rd)
 			if err != nil {
 				ctx.ServerError("Render", err)
 				return
 			}
 			// to prevent iframe load third-party url
 			ctx.Resp.Header().Add("Content-Security-Policy", "frame-src 'self'")
-			ctx.Data["EscapeStatus"], ctx.Data["FileContent"] = charset.EscapeControlString(result.String())
 		} else if readmeExist && !shouldRenderSource {
 			buf := &bytes.Buffer{}
 			ctx.Data["IsRenderedHTML"] = true
 
-			ctx.Data["EscapeStatus"], _ = charset.EscapeControlReader(rd, buf)
+			ctx.Data["EscapeStatus"], _ = charset.EscapeControlReader(rd, buf, ctx.Locale)
 
 			ctx.Data["FileContent"] = strings.ReplaceAll(
 				gotemplate.HTMLEscapeString(buf.String()), "\n", `<br>`,
@@ -570,12 +564,13 @@ func renderFile(ctx *context.Context, entry *git.TreeEntry, treeLink, rawLink st
 				log.Error("highlight.File failed, fallback to plain text: %v", err)
 				fileContent = highlight.PlainText(buf)
 			}
-			status, _ := charset.EscapeControlReader(bytes.NewReader(buf), io.Discard)
-			ctx.Data["EscapeStatus"] = status
-			statuses := make([]charset.EscapeStatus, len(fileContent))
+			status := &charset.EscapeStatus{}
+			statuses := make([]*charset.EscapeStatus, len(fileContent))
 			for i, line := range fileContent {
-				statuses[i], fileContent[i] = charset.EscapeControlString(line)
+				statuses[i], fileContent[i] = charset.EscapeControlHTML(line, ctx.Locale)
+				status = status.Or(statuses[i])
 			}
+			ctx.Data["EscapeStatus"] = status
 			ctx.Data["FileContent"] = fileContent
 			ctx.Data["LineEscapeStatus"] = statuses
 		}
@@ -613,20 +608,17 @@ func renderFile(ctx *context.Context, entry *git.TreeEntry, treeLink, rawLink st
 			rd := io.MultiReader(bytes.NewReader(buf), dataRc)
 			ctx.Data["IsMarkup"] = true
 			ctx.Data["MarkupType"] = markupType
-			var result strings.Builder
-			err := markup.Render(&markup.RenderContext{
+			ctx.Data["EscapeStatus"], ctx.Data["FileContent"], err = markupRender(ctx, &markup.RenderContext{
 				Ctx:          ctx,
 				RelativePath: ctx.Repo.TreePath,
 				URLPrefix:    path.Dir(treeLink),
 				Metas:        ctx.Repo.Repository.ComposeDocumentMetas(),
 				GitRepo:      ctx.Repo.GitRepo,
-			}, rd, &result)
+			}, rd)
 			if err != nil {
 				ctx.ServerError("Render", err)
 				return
 			}
-
-			ctx.Data["EscapeStatus"], ctx.Data["FileContent"] = charset.EscapeControlString(result.String())
 		}
 	}
 
@@ -645,6 +637,23 @@ func renderFile(ctx *context.Context, entry *git.TreeEntry, treeLink, rawLink st
 	}
 }
 
+func markupRender(ctx *context.Context, renderCtx *markup.RenderContext, input io.Reader) (escaped *charset.EscapeStatus, output string, err error) {
+	markupRd, markupWr := io.Pipe()
+	defer markupWr.Close()
+	done := make(chan struct{})
+	go func() {
+		sb := &strings.Builder{}
+		// We allow NBSP here this is rendered
+		escaped, _ = charset.EscapeControlReader(markupRd, sb, ctx.Locale, charset.RuneNBSP)
+		output = sb.String()
+		close(done)
+	}()
+	err = markup.Render(renderCtx, input, markupWr)
+	_ = markupWr.CloseWithError(err)
+	<-done
+	return escaped, output, err
+}
+
 func safeURL(address string) string {
 	u, err := url.Parse(address)
 	if err != nil {

routers/web/repo/wiki.go

@@ -239,9 +239,28 @@ func renderViewPage(ctx *context.Context) (*git.Repository, *git.TreeEntry) {
 		Metas:     ctx.Repo.Repository.ComposeDocumentMetas(),
 		IsWiki:    true,
 	}
-
-	var buf strings.Builder
-	if err := markdown.Render(rctx, bytes.NewReader(data), &buf); err != nil {
+	buf := &strings.Builder{}
+
+	renderFn := func(data []byte) (escaped *charset.EscapeStatus, output string, err error) {
+		markupRd, markupWr := io.Pipe()
+		defer markupWr.Close()
+		done := make(chan struct{})
+		go func() {
+			// We allow NBSP here this is rendered
+			escaped, _ = charset.EscapeControlReader(markupRd, buf, ctx.Locale, charset.RuneNBSP)
+			output = buf.String()
+			buf.Reset()
+			close(done)
+		}()
+
+		err = markdown.Render(rctx, bytes.NewReader(data), markupWr)
+		_ = markupWr.CloseWithError(err)
+		<-done
+		return escaped, output, err
+	}
+
+	ctx.Data["EscapeStatus"], ctx.Data["content"], err = renderFn(data)
+	if err != nil {
 		if wikiRepo != nil {
 			wikiRepo.Close()
 		}
@@ -249,11 +268,10 @@ func renderViewPage(ctx *context.Context) (*git.Repository, *git.TreeEntry) {
 		return nil, nil
 	}
 
-	ctx.Data["EscapeStatus"], ctx.Data["content"] = charset.EscapeControlString(buf.String())
-
 	if !isSideBar {
 		buf.Reset()
-		if err := markdown.Render(rctx, bytes.NewReader(sidebarContent), &buf); err != nil {
+		ctx.Data["sidebarEscapeStatus"], ctx.Data["sidebarContent"], err = renderFn(sidebarContent)
+		if err != nil {
 			if wikiRepo != nil {
 				wikiRepo.Close()
 			}
@@ -261,14 +279,14 @@ func renderViewPage(ctx *context.Context) (*git.Repository, *git.TreeEntry) {
 			return nil, nil
 		}
 		ctx.Data["sidebarPresent"] = sidebarContent != nil
-		ctx.Data["sidebarEscapeStatus"], ctx.Data["sidebarContent"] = charset.EscapeControlString(buf.String())
 	} else {
 		ctx.Data["sidebarPresent"] = false
 	}
 
 	if !isFooter {
 		buf.Reset()
-		if err := markdown.Render(rctx, bytes.NewReader(footerContent), &buf); err != nil {
+		ctx.Data["footerEscapeStatus"], ctx.Data["footerContent"], err = renderFn(footerContent)
+		if err != nil {
 			if wikiRepo != nil {
 				wikiRepo.Close()
 			}
@@ -276,7 +294,6 @@ func renderViewPage(ctx *context.Context) (*git.Repository, *git.TreeEntry) {
 			return nil, nil
 		}
 		ctx.Data["footerPresent"] = footerContent != nil
-		ctx.Data["footerEscapeStatus"], ctx.Data["footerContent"] = charset.EscapeControlString(buf.String())
 	} else {
 		ctx.Data["footerPresent"] = false
 	}

services/gitdiff/gitdiff.go

@@ -32,6 +32,7 @@ import (
 	"code.gitea.io/gitea/modules/lfs"
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/setting"
+	"code.gitea.io/gitea/modules/translation"
 
 	"github.com/sergi/go-diff/diffmatchpatch"
 	stdcharset "golang.org/x/net/html/charset"
@@ -169,11 +170,11 @@ func getDiffLineSectionInfo(treePath, line string, lastLeftIdx, lastRightIdx int
 }
 
 // escape a line's content or return <br> needed for copy/paste purposes
-func getLineContent(content string) DiffInline {
+func getLineContent(content string, locale translation.Locale) DiffInline {
 	if len(content) > 0 {
-		return DiffInlineWithUnicodeEscape(template.HTML(html.EscapeString(content)))
+		return DiffInlineWithUnicodeEscape(template.HTML(html.EscapeString(content)), locale)
 	}
-	return DiffInline{Content: "<br>"}
+	return DiffInline{EscapeStatus: &charset.EscapeStatus{}, Content: "<br>"}
 }
 
 // DiffSection represents a section of a DiffFile.
@@ -267,26 +268,26 @@ func init() {
 
 // DiffInline is a struct that has a content and escape status
 type DiffInline struct {
-	EscapeStatus charset.EscapeStatus
+	EscapeStatus *charset.EscapeStatus
 	Content      template.HTML
 }
 
 // DiffInlineWithUnicodeEscape makes a DiffInline with hidden unicode characters escaped
-func DiffInlineWithUnicodeEscape(s template.HTML) DiffInline {
-	status, content := charset.EscapeControlString(string(s))
+func DiffInlineWithUnicodeEscape(s template.HTML, locale translation.Locale) DiffInline {
+	status, content := charset.EscapeControlHTML(string(s), locale)
 	return DiffInline{EscapeStatus: status, Content: template.HTML(content)}
 }
 
 // DiffInlineWithHighlightCode makes a DiffInline with code highlight and hidden unicode characters escaped
-func DiffInlineWithHighlightCode(fileName, language, code string) DiffInline {
-	status, content := charset.EscapeControlString(highlight.Code(fileName, language, code))
+func DiffInlineWithHighlightCode(fileName, language, code string, locale translation.Locale) DiffInline {
+	status, content := charset.EscapeControlHTML(highlight.Code(fileName, language, code), locale)
 	return DiffInline{EscapeStatus: status, Content: template.HTML(content)}
 }
 
 // GetComputedInlineDiffFor computes inline diff for the given line.
-func (diffSection *DiffSection) GetComputedInlineDiffFor(diffLine *DiffLine) DiffInline {
+func (diffSection *DiffSection) GetComputedInlineDiffFor(diffLine *DiffLine, locale translation.Locale) DiffInline {
 	if setting.Git.DisableDiffHighlight {
-		return getLineContent(diffLine.Content[1:])
+		return getLineContent(diffLine.Content[1:], locale)
 	}
 
 	var (
@@ -303,26 +304,26 @@ func (diffSection *DiffSection) GetComputedInlineDiffFor(diffLine *DiffLine) Dif
 	// try to find equivalent diff line. ignore, otherwise
 	switch diffLine.Type {
 	case DiffLineSection:
-		return getLineContent(diffLine.Content[1:])
+		return getLineContent(diffLine.Content[1:], locale)
 	case DiffLineAdd:
 		compareDiffLine = diffSection.GetLine(DiffLineDel, diffLine.RightIdx)
 		if compareDiffLine == nil {
-			return DiffInlineWithHighlightCode(diffSection.FileName, language, diffLine.Content[1:])
+			return DiffInlineWithHighlightCode(diffSection.FileName, language, diffLine.Content[1:], locale)
 		}
 		diff1 = compareDiffLine.Content
 		diff2 = diffLine.Content
 	case DiffLineDel:
 		compareDiffLine = diffSection.GetLine(DiffLineAdd, diffLine.LeftIdx)
 		if compareDiffLine == nil {
-			return DiffInlineWithHighlightCode(diffSection.FileName, language, diffLine.Content[1:])
+			return DiffInlineWithHighlightCode(diffSection.FileName, language, diffLine.Content[1:], locale)
 		}
 		diff1 = diffLine.Content
 		diff2 = compareDiffLine.Content
 	default:
 		if strings.IndexByte(" +-", diffLine.Content[0]) > -1 {
-			return DiffInlineWithHighlightCode(diffSection.FileName, language, diffLine.Content[1:])
+			return DiffInlineWithHighlightCode(diffSection.FileName, language, diffLine.Content[1:], locale)
 		}
-		return DiffInlineWithHighlightCode(diffSection.FileName, language, diffLine.Content)
+		return DiffInlineWithHighlightCode(diffSection.FileName, language, diffLine.Content, locale)
 	}
 
 	hcd := newHighlightCodeDiff()
@@ -330,7 +331,7 @@ func (diffSection *DiffSection) GetComputedInlineDiffFor(diffLine *DiffLine) Dif
 	// it seems that Gitea doesn't need the line wrapper of Chroma, so do not add them back
 	// if the line wrappers are still needed in the future, it can be added back by "diffToHTML(hcd.lineWrapperTags. ...)"
 	diffHTML := diffToHTML(nil, diffRecord, diffLine.Type)
-	return DiffInlineWithUnicodeEscape(template.HTML(diffHTML))
+	return DiffInlineWithUnicodeEscape(template.HTML(diffHTML), locale)
 }
 
 // DiffFile represents a file diff.

templates/repo/blame.tmpl

@@ -55,7 +55,11 @@
 								<span id="L{{$row.RowNumber}}" data-line-number="{{$row.RowNumber}}"></span>
 							</td>
 							{{if $.EscapeStatus.Escaped}}
-								<td class="lines-escape">{{if $row.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{$.locale.Tr "repo.line_unicode"}}"></a>{{end}}</td>
+								<td class="lines-escape">
+									{{if $row.EscapeStatus.Escaped}}
+										<a href="" class="toggle-escape-button" title="{{template "repo/diff/escape_title" dict "diff" $row "locale" $.locale}}"></a>
+									{{end}}
+								</td>
 							{{end}}
 							<td rel="L{{$row.RowNumber}}" rel="L{{$row.RowNumber}}" class="lines-code blame-code chroma">
 								<code class="code-inner pl-3">{{$row.Code}}</code>

templates/repo/diff/blob_excerpt.tmpl

@@ -19,20 +19,25 @@
 					</a>
 				{{end}}
 			</td>
-			<td colspan="5" class="lines-code lines-code-old ">{{$inlineDiff := $.section.GetComputedInlineDiffFor $line}}<code {{if $inlineDiff.EscapeStatus.Escaped}}class="code-inner has-escaped" title="{{$.locale.Tr "repo.line_unicode"}}"{{else}}class="code-inner"{{end}}>{{$inlineDiff.Content}}</code></td>
+			<td colspan="5" class="lines-code lines-code-old ">{{$inlineDiff := $.section.GetComputedInlineDiffFor $line $.locale}}{{/*
+				*/}}{{template "repo/diff/section_code" dict "diff" $inlineDiff "locale" $.locale}}</td>
 		{{else}}
-			{{$inlineDiff := $.section.GetComputedInlineDiffFor $line}}
+			{{$inlineDiff := $.section.GetComputedInlineDiffFor $line $.locale}}
 			<td class="lines-num lines-num-old" data-line-num="{{if $line.LeftIdx}}{{$line.LeftIdx}}{{end}}"><span rel="{{if $line.LeftIdx}}diff-{{$.FileNameHash}}L{{$line.LeftIdx}}{{end}}"></span></td>
-			<td class="blob-excerpt lines-escape lines-escape-old">{{if and $line.LeftIdx $inlineDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{$.locale.Tr "repo.line_unicode"}}"></a>{{end}}</td>
+			<td class="blob-excerpt lines-escape lines-escape-old">{{if and $line.LeftIdx $inlineDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{template "repo/diff/escape_title" dict "diff" $inlineDiff "locale" $.locale}}"></a>{{end}}</td>
 			<td class="blob-excerpt lines-type-marker lines-type-marker-old">{{if $line.LeftIdx}}<span class="mono" data-type-marker=""></span>{{end}}</td>
 			<td class="blob-excerpt lines-code lines-code-old halfwidth">{{/*
-				*/}}<code {{if and $line.LeftIdx $inlineDiff.EscapeStatus.Escaped}}class="code-inner has-escaped" title="{{$.locale.Tr "repo.line_unicode"}}"{{else}}class="code-inner"{{end}}>{{if $line.LeftIdx}}{{$inlineDiff.Content}}{{end}}</code>{{/*
+				*/}}{{if $line.LeftIdx}}{{template "repo/diff/section_code" dict "diff" $inlineDiff "locale" $.locale}}{{else}}{{/*
+					*/}}<code class="code-inner"></code>{{/*
+				*/}}{{end}}{{/*
 			*/}}</td>
 			<td class="lines-num lines-num-new" data-line-num="{{if $line.RightIdx}}{{$line.RightIdx}}{{end}}"><span rel="{{if $line.RightIdx}}diff-{{$.FileNameHash}}R{{$line.RightIdx}}{{end}}"></span></td>
-			<td class="blob-excerpt lines-escape lines-escape-new">{{if and $line.RightIdx $inlineDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{$.locale.Tr "repo.line_unicode"}}"></a>{{end}}</td>
+			<td class="blob-excerpt lines-escape lines-escape-new">{{if and $line.RightIdx $inlineDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{template "repo/diff/escape_title" dict "diff" $inlineDiff "locale" $.locale}}"></a>{{end}}</td>
 			<td class="blob-excerpt lines-type-marker lines-type-marker-new">{{if $line.RightIdx}}<span class="mono" data-type-marker=""></span>{{end}}</td>
 			<td class="blob-excerpt lines-code lines-code-new halfwidth">{{/*
-				*/}}<code {{if and $line.RightIdx $inlineDiff.EscapeStatus.Escaped}}class="code-inner has-escaped" title="{{$.locale.Tr "repo.line_unicode"}}"{{else}}class="code-inner"{{end}}>{{if $line.RightIdx}}{{$inlineDiff.Content}}{{end}}</code>{{/*
+				*/}}{{if $line.RightIdx}}{{template "repo/diff/section_code" dict "diff" $inlineDiff "locale" $.locale}}{{else}}{{/*
+					*/}}<code class="code-inner"></code>{{/*
+				*/}}{{end}}{{/*
 			*/}}</td>
 		{{end}}
 	</tr>
@@ -62,10 +67,10 @@
 			<td class="lines-num lines-num-old" data-line-num="{{if $line.LeftIdx}}{{$line.LeftIdx}}{{end}}"><span rel="{{if $line.LeftIdx}}diff-{{$.FileNameHash}}L{{$line.LeftIdx}}{{end}}"></span></td>
 			<td class="lines-num lines-num-new" data-line-num="{{if $line.RightIdx}}{{$line.RightIdx}}{{end}}"><span rel="{{if $line.RightIdx}}diff-{{$.FileNameHash}}R{{$line.RightIdx}}{{end}}"></span></td>
 		{{end}}
-		{{$inlineDiff := $.section.GetComputedInlineDiffFor $line}}
-		<td class="blob-excerpt lines-escape">{{if $inlineDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{$.locale.Tr "repo.line_unicode"}}"></a>{{end}}</td>
+		{{$inlineDiff := $.section.GetComputedInlineDiffFor $line $.locale}}
+		<td class="blob-excerpt lines-escape">{{if $inlineDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{template "repo/diff/escape_title" dict "diff" $inlineDiff "locale" $.locale}}"></a>{{end}}</td>
 		<td class="blob-excerpt lines-type-marker"><span class="mono" data-type-marker="{{$line.GetLineTypeMarker}}"></span></td>
-		<td class="blob-excerpt lines-code{{if (not $line.RightIdx)}} lines-code-old{{end}}"><code {{if $inlineDiff.EscapeStatus.Escaped}}class="code-inner has-escaped" title="{{$.locale.Tr "repo.line_unicode"}}"{{else}}class="code-inner"{{end}}>{{$inlineDiff.Content}}</code></td>
+		<td class="blob-excerpt lines-code{{if (not $line.RightIdx)}} lines-code-old{{end}}"><code {{if $inlineDiff.EscapeStatus.Escaped}}class="code-inner has-escaped" title="{{template "repo/diff/escape_title" dict "diff" $inlineDiff "locale" $.locale}}"{{else}}class="code-inner"{{end}}>{{$inlineDiff.Content}}</code></td>
 	</tr>
 	{{end}}
 {{end}}

templates/repo/diff/escape_title.tmpl

@@ -0,0 +1,2 @@
+{{if .diff.EscapeStatus.HasInvisible}}{{.locale.Tr "repo.invisible_runes_line"}} {{end}}{{/*
+*/}}{{if .diff.EscapeStatus.HasAmbiguous}}{{.locale.Tr "repo.ambiguous_runes_line"}}{{end}}

templates/repo/diff/section_code.tmpl

@@ -0,0 +1,6 @@
+<code {{if .diff.EscapeStatus.Escaped}}{{/*
+	*/}}class="code-inner has-escaped" {{/*
+	*/}}title="{{template "repo/diff/escape_title" .}}"{{/*
+	*/}}{{else}}{{/*
+	*/}}class="code-inner"{{/*
+	*/}}{{end}}>{{.diff.Content}}</code>

templates/repo/diff/section_split.tmpl

@@ -21,15 +21,17 @@
 								{{svg "octicon-fold"}}
 							</a>
 						{{end}}
-					</td>{{$inlineDiff := $section.GetComputedInlineDiffFor $line}}
-					<td class="lines-escape lines-escape-old">{{if $inlineDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{$.locale.Tr "repo.line_unicode"}}"></a>{{end}}</td>
-					<td colspan="6" class="lines-code lines-code-old "><code {{if $inlineDiff.EscapeStatus.Escaped}}class="code-inner has-escaped" title="{{$.root.locale.Tr "repo.line_unicode"}}"{{else}}class="code-inner"{{end}}>{{$inlineDiff.Content}}</span></td>
+					</td>{{$inlineDiff := $section.GetComputedInlineDiffFor $line $.root.locale}}
+					<td class="lines-escape lines-escape-old">{{if $inlineDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{template "repo/diff/escape_title" dict "diff" $inlineDiff "locale" $.locale}}"></a>{{end}}</td>
+					<td colspan="6" class="lines-code lines-code-old ">{{/*
+						*/}}{{template "repo/diff/section_code" dict "diff" $inlineDiff "locale" $.root.locale}}{{/*
+					*/}}</td>
 				{{else if and (eq .GetType 3) $hasmatch}}{{/* DEL */}}
 					{{$match := index $section.Lines $line.Match}}
-					{{- $leftDiff := ""}}{{if $line.LeftIdx}}{{$leftDiff = $section.GetComputedInlineDiffFor $line}}{{end}}
-					{{- $rightDiff := ""}}{{if $match.RightIdx}}{{$rightDiff = $section.GetComputedInlineDiffFor $match}}{{end}}
+					{{- $leftDiff := ""}}{{if $line.LeftIdx}}{{$leftDiff = $section.GetComputedInlineDiffFor $line $.root.locale}}{{end}}
+					{{- $rightDiff := ""}}{{if $match.RightIdx}}{{$rightDiff = $section.GetComputedInlineDiffFor $match $.root.locale}}{{end}}
 					<td class="lines-num lines-num-old del-code" data-line-num="{{$line.LeftIdx}}"><span rel="diff-{{$file.NameHash}}L{{$line.LeftIdx}}"></span></td>
-					<td class="lines-escape del-code lines-escape-old">{{if $line.LeftIdx}}{{if $leftDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{$.locale.Tr "repo.line_unicode"}}"></a>{{end}}{{end}}</td>
+					<td class="lines-escape del-code lines-escape-old">{{if $line.LeftIdx}}{{if $leftDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{template "repo/diff/escape_title" dict "diff" $leftDiff "locale" $.locale}}"></a>{{end}}{{end}}</td>
 					<td class="lines-type-marker lines-type-marker-old del-code"><span class="mono" data-type-marker="{{$line.GetLineTypeMarker}}"></span></td>
 					<td class="lines-code lines-code-old halfwidth del-code">{{/*
 						*/}}{{if and $.root.SignedUserID $.root.PageIsPullFiles}}{{/*
@@ -38,13 +40,13 @@
 							*/}}</a>{{/*
 						*/}}{{end}}{{/*
 						*/}}{{if $line.LeftIdx}}{{/*
-							*/}}<code {{if $leftDiff.EscapeStatus.Escaped}}class="code-inner has-escaped" title="{{$.root.locale.Tr "repo.line_unicode"}}"{{else}}class="code-inner"{{end}}>{{$leftDiff.Content}}</code>{{/*
+							*/}}{{template "repo/diff/section_code" dict "diff" $leftDiff "locale" $.root.locale}}{{/*
 						*/}}{{else}}{{/*
 						*/}}<code class="code-inner"></code>{{/*
 						*/}}{{end}}{{/*
 					*/}}</td>
 					<td class="lines-num lines-num-new add-code" data-line-num="{{if $match.RightIdx}}{{$match.RightIdx}}{{end}}"><span rel="{{if $match.RightIdx}}diff-{{$file.NameHash}}R{{$match.RightIdx}}{{end}}"></span></td>
-					<td class="lines-escape add-code lines-escape-new">{{if $match.RightIdx}}{{if $rightDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{$.locale.Tr "repo.line_unicode"}}"></a>{{end}}{{end}}</td>
+					<td class="lines-escape add-code lines-escape-new">{{if $match.RightIdx}}{{if $rightDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{template "repo/diff/escape_title" dict "diff" $rightDiff "locale" $.locale}}"></a>{{end}}{{end}}</td>
 					<td class="lines-type-marker lines-type-marker-new add-code">{{if $match.RightIdx}}<span class="mono" data-type-marker="{{$match.GetLineTypeMarker}}"></span>{{end}}</td>
 					<td class="lines-code lines-code-new halfwidth add-code">{{/*
 						*/}}{{if and $.root.SignedUserID $.root.PageIsPullFiles}}{{/*
@@ -53,15 +55,15 @@
 							*/}}</a>{{/*
 						*/}}{{end}}{{/*
 						*/}}{{if $match.RightIdx}}{{/*
-							*/}}<code {{if $rightDiff.EscapeStatus.Escaped}}class="code-inner has-escaped" title="{{$.root.locale.Tr "repo.line_unicode"}}"{{else}}class="code-inner"{{end}}>{{$rightDiff.Content}}</code>{{/*
+							*/}}{{template "repo/diff/section_code" dict "diff" $rightDiff "locale" $.root.locale}}{{/*
 						*/}}{{else}}{{/*
 							*/}}<code class="code-inner"></code>{{/*
 						*/}}{{end}}{{/*
 					*/}}</td>
 				{{else}}
-					{{$inlineDiff := $section.GetComputedInlineDiffFor $line}}
+					{{$inlineDiff := $section.GetComputedInlineDiffFor $line $.root.locale}}
 					<td class="lines-num lines-num-old" data-line-num="{{if $line.LeftIdx}}{{$line.LeftIdx}}{{end}}"><span rel="{{if $line.LeftIdx}}diff-{{$file.NameHash}}L{{$line.LeftIdx}}{{end}}"></span></td>
-					<td class="lines-escape lines-escape-old">{{if $line.LeftIdx}}{{if $inlineDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{$.locale.Tr "repo.line_unicode"}}"></a>{{end}}{{end}}</td>
+					<td class="lines-escape lines-escape-old">{{if $line.LeftIdx}}{{if $inlineDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{template "repo/diff/escape_title" dict "diff" $inlineDiff "locale" $.locale}}"></a>{{end}}{{end}}</td>
 					<td class="lines-type-marker lines-type-marker-old">{{if $line.LeftIdx}}<span class="mono" data-type-marker="{{$line.GetLineTypeMarker}}"></span>{{end}}</td>
 					<td class="lines-code lines-code-old halfwidth">{{/*
 						*/}}{{if and $.root.SignedUserID $.root.PageIsPullFiles (not (eq .GetType 2))}}{{/*
@@ -70,13 +72,13 @@
 							*/}}</a>{{/*
 						*/}}{{end}}{{/*
 						*/}}{{if $line.LeftIdx}}{{/*
-							*/}}<code {{if $inlineDiff.EscapeStatus.Escaped}}class="code-inner has-escaped" title="{{$.root.locale.Tr "repo.line_unicode"}}"{{else}}class="code-inner"{{end}}>{{$inlineDiff.Content}}</code>{{/*
+							*/}}{{template "repo/diff/section_code" dict "diff" $inlineDiff "locale" $.root.locale}}{{/*
 						*/}}{{else}}{{/*
 						*/}}<code class="code-inner"></code>{{/*
 						*/}}{{end}}{{/*
 					*/}}</td>
 					<td class="lines-num lines-num-new" data-line-num="{{if $line.RightIdx}}{{$line.RightIdx}}{{end}}"><span rel="{{if $line.RightIdx}}diff-{{$file.NameHash}}R{{$line.RightIdx}}{{end}}"></span></td>
-					<td class="lines-escape lines-escape-new">{{if $line.RightIdx}}{{if $inlineDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{$.locale.Tr "repo.line_unicode"}}"></a>{{end}}{{end}}</td>
+					<td class="lines-escape lines-escape-new">{{if $line.RightIdx}}{{if $inlineDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{template "repo/diff/escape_title" dict "diff" $inlineDiff "locale" $.locale}}"></a>{{end}}{{end}}</td>
 					<td class="lines-type-marker lines-type-marker-new">{{if $line.RightIdx}}<span class="mono" data-type-marker="{{$line.GetLineTypeMarker}}"></span>{{end}}</td>
 					<td class="lines-code lines-code-new halfwidth">{{/*
 						*/}}{{if and $.root.SignedUserID $.root.PageIsPullFiles (not (eq .GetType 3))}}{{/*
@@ -85,7 +87,7 @@
 							*/}}</a>{{/*
 						*/}}{{end}}{{/*
 						*/}}{{if $line.RightIdx}}{{/*
-							*/}}<code {{if $inlineDiff.EscapeStatus.Escaped}}class="code-inner has-escaped" title="{{$.root.locale.Tr "repo.line_unicode"}}"{{else}}class="code-inner"{{end}}>{{$inlineDiff.Content}}</code>{{/*
+							*/}}{{template "repo/diff/section_code" dict "diff" $inlineDiff "locale" $.root.locale}}{{/*
 						*/}}{{else}}{{/*
 						*/}}<code class="code-inner"></code>{{/*
 						*/}}{{end}}{{/*

templates/repo/diff/section_unified.tmpl

@@ -25,12 +25,12 @@
 					<td class="lines-num lines-num-old" data-line-num="{{if $line.LeftIdx}}{{$line.LeftIdx}}{{end}}"><span rel="{{if $line.LeftIdx}}diff-{{$file.NameHash}}L{{$line.LeftIdx}}{{end}}"></span></td>
 					<td class="lines-num lines-num-new" data-line-num="{{if $line.RightIdx}}{{$line.RightIdx}}{{end}}"><span rel="{{if $line.RightIdx}}diff-{{$file.NameHash}}R{{$line.RightIdx}}{{end}}"></span></td>
 				{{end}}
-				{{$inlineDiff := $section.GetComputedInlineDiffFor $line -}}
-				<td class="lines-escape">{{if $inlineDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{$.locale.Tr "repo.line_unicode"}}"></a>{{end}}</td>
+				{{$inlineDiff := $section.GetComputedInlineDiffFor $line $.root.locale -}}
+				<td class="lines-escape">{{if $inlineDiff.EscapeStatus.Escaped}}<a href="" class="toggle-escape-button" title="{{template "repo/diff/escape_title" dict "diff" $inlineDiff "locale" $.locale}}"></a>{{end}}</td>
 				<td class="lines-type-marker"><span class="mono" data-type-marker="{{$line.GetLineTypeMarker}}"></span></td>
 				{{if eq .GetType 4}}
 					<td class="chroma lines-code blob-hunk">{{/*
-						*/}}<code {{if $inlineDiff.EscapeStatus.Escaped}}class="code-inner has-escaped" title="{{$.root.locale.Tr "repo.line_unicode"}}"{{else}}class="code-inner"{{end}}>{{$inlineDiff.Content}}</code>{{/*
+						*/}}{{template "repo/diff/section_code" dict "diff" $inlineDiff "locale" $.root.locale}}{{/*
 					*/}}</td>
 				{{else}}
 					<td class="chroma lines-code{{if (not $line.RightIdx)}} lines-code-old{{end}}">{{/*
@@ -39,7 +39,7 @@
 								*/}}{{svg "octicon-plus"}}{{/*
 							*/}}</a>{{/*
 						*/}}{{end}}{{/*
-						*/}}<code {{if $inlineDiff.EscapeStatus.Escaped}}class="code-inner has-escaped" title="{{$.root.locale.Tr "repo.line_unicode"}}"{{else}}class="code-inner"{{end}}>{{$inlineDiff.Content}}</code>{{/*
+						*/}}{{template "repo/diff/section_code" dict "diff" $inlineDiff "locale" $.root.locale}}{{/*
 					*/}}</td>
 				{{end}}
 			</tr>

templates/repo/unicode_escape_prompt.tmpl

@@ -1,19 +1,22 @@
 {{if .EscapeStatus}}
-	{{if .EscapeStatus.BadBIDI}}
+	{{if .EscapeStatus.HasInvisible}}
 		<div class="ui error message unicode-escape-prompt tl">
 			<span class="close icon hide-panel button" data-panel-closest=".message">{{svg "octicon-x" 16 "close inside"}}</span>
 			<div class="header">
-				{{$.root.locale.Tr "repo.bidi_bad_header"}}
+				{{$.root.locale.Tr "repo.invisible_runes_header"}}
 			</div>
-			<p>{{$.root.locale.Tr "repo.bidi_bad_description" | Str2html}}</p>
+			<p>{{$.root.locale.Tr "repo.invisible_runes_description" | Str2html}}</p>
+			{{if .EscapeStatus.HasAmbiguous}}
+				<p>{{$.root.locale.Tr "repo.ambiguous_runes_description" | Str2html}}</p>
+			{{end}}
 		</div>
-	{{else if .EscapeStatus.HasBIDI}}
+	{{else if .EscapeStatus.HasAmbiguous}}
 		<div class="ui warning message unicode-escape-prompt tl">
 			<span class="close icon hide-panel button" data-panel-closest=".message">{{svg "octicon-x" 16 "close inside"}}</span>
 			<div class="header">
-				{{$.root.locale.Tr "repo.unicode_header"}}
+				{{$.root.locale.Tr "repo.ambiguous_runes_header"}}
 			</div>
-			<p>{{$.root.locale.Tr "repo.unicode_description" | Str2html}}</p>
+			<p>{{$.root.locale.Tr "repo.ambiguous_runes_description" | Str2html}}</p>
 		</div>
 	{{end}}
 {{end}}

templates/repo/view_file.tmpl

@@ -113,7 +113,7 @@
 						<tr>
 							<td id="L{{$line}}" class="lines-num"><span id="L{{$line}}" data-line-number="{{$line}}"></span></td>
 							{{if $.EscapeStatus.Escaped}}
-								<td class="lines-escape">{{if (index $.LineEscapeStatus $idx).Escaped}}<a href="" class="toggle-escape-button" title="{{$.locale.Tr "repo.line_unicode"}}"></a>{{end}}</td>
+								<td class="lines-escape">{{if (index $.LineEscapeStatus $idx).Escaped}}<a href="" class="toggle-escape-button" title="{{if (index $.LineEscapeStatus $idx).HasInvisible}}{{$.locale.Tr "repo.invisible_runes_line"}} {{end}}{{if (index $.LineEscapeStatus $idx).HasAmbiguous}}{{$.locale.Tr "repo.ambiguous_runes_line"}}{{end}}"></a>{{end}}</td>
 							{{end}}
 							<td rel="L{{$line}}" class="lines-code chroma"><code class="code-inner">{{$code | Safe}}</code></td>
 						</tr>

web_src/less/_repository.less

@@ -82,7 +82,11 @@
 
   .broken-code-point {
     font-family: var(--fonts-monospace);
-    color: blue;
+    color: var(--color-blue);
+  }
+
+  .unicode-escaped .ambiguous-code-point {
+    border: 1px var(--color-yellow) solid;
   }
 
   .metas {