plaintext.go RawHistoryFolder

1// SPDX-License-Identifier: AGPL-3.0-or-later
2
3package web
4
5import (
6	"net/url"
7	"regexp"
8	"strings"
9)
10
11// TextPart is a run of plain text, or a link found in it.
12type TextPart struct {
13	Text string
14	// Href is the target of a link, "" for plain text. Only http and https
15	// URLs are ever linked.
16	Href string
17}
18
19// reURL finds the candidates for a link in plain text.
20var reURL = regexp.MustCompile(`https?://[^\s<>"'` + "`" + `]+`)
21
22// linkTrail is what ends a sentence rather than an address.
23const linkTrail = ".,;:!?)]}"
24
25// plainParts splits a user text (a description, a comment) into runs: the text
26// stays text, escaped by the template with its line breaks kept, and the
27// http(s) URLs in it become links (ADR 0017: "text is plain everywhere"). The
28// invisible characters of the text are shown as markers.
29func plainParts(s string) []TextPart {
30	s = strings.ReplaceAll(strings.ReplaceAll(s, "\r\n", "\n"), "\r", "\n")
31	var out []TextPart
32	last := 0
33	for _, loc := range reURL.FindAllStringIndex(s, -1) {
34		raw := s[loc[0]:loc[1]]
35		// A closing bracket belongs to the address only when it has an opening
36		// one in it: "(see https://x.test/a)" ends before the ")".
37		end := len(raw)
38		for end > 0 && strings.ContainsRune(linkTrail, rune(raw[end-1])) {
39			if raw[end-1] == ')' && strings.Count(raw[:end], "(") >= strings.Count(raw[:end], ")") {
40				break
41			}
42			end--
43		}
44		raw = raw[:end]
45		u, err := url.Parse(raw)
46		if err != nil || (u.Scheme != "http" && u.Scheme != "https") || u.Hostname() == "" {
47			continue
48		}
49		if loc[0] > last {
50			out = append(out, TextPart{Text: visibleText(s[last:loc[0]])})
51		}
52		out = append(out, TextPart{Text: visibleText(raw), Href: u.String()})
53		last = loc[0] + len(raw)
54	}
55	if last < len(s) {
56		out = append(out, TextPart{Text: visibleText(s[last:])})
57	}
58	return out
59}