Switch Unicode Escaping to a VSCode-like system (go-gitea#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 go-gitea#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)
+			}
+		})
+	}
+}