text.go RawHistoryFolder

1// SPDX-License-Identifier: AGPL-3.0-or-later
2
3package web
4
5import (
6	"fmt"
7	"strings"
8	"unicode"
9)
10
11// invisible reports whether r is a character that changes how code reads
12// without being seen: the bidirectional controls and the other format
13// characters (Cf: zero-width, joiners, tags, BOM…), the control characters,
14// the line and paragraph separators, and the blank Hangul fillers. A tab is
15// not one, nor is the line break of a commit message.
16func invisible(r rune) bool {
17	switch r {
18	case '\t', '\n':
19		return false
20	case 0x115F, 0x1160, 0x3164, 0xFFA0: // Hangul fillers render blank
21		return true
22	}
23	return unicode.IsControl(r) || unicode.Is(unicode.Cf, r) || unicode.Is(unicode.Zl, r) || unicode.Is(unicode.Zp, r)
24}
25
26// marker is the visible form of an invisible character.
27func marker(r rune) string { return fmt.Sprintf("<U+%04X>", r) }
28
29// visibleText is s with every invisible character replaced by its marker,
30// for the places that show text as a plain string: paths, names, subjects.
31func visibleText(s string) string {
32	if !strings.ContainsFunc(s, invisible) {
33		return s
34	}
35	var b strings.Builder
36	for _, r := range s {
37		if invisible(r) {
38			b.WriteString(marker(r))
39		} else {
40			b.WriteRune(r)
41		}
42	}
43	return b.String()
44}