rss-tools/sources/telegram/links.go (view raw)
Oleksandr Smirnov
Oleksandr Smirnov
olexsmir@gmail.com telegram: turn links into links; return youtube links in youtube official feed format, 1 month ago
olexsmir@gmail.com telegram: turn links into links; return youtube links in youtube official feed format, 1 month ago
| 1 | package telegram |
| 2 | |
| 3 | import ( |
| 4 | "fmt" |
| 5 | "html" |
| 6 | "net/url" |
| 7 | "regexp" |
| 8 | "strings" |
| 9 | |
| 10 | "olexsmir.xyz/rss-tools/app" |
| 11 | ) |
| 12 | |
| 13 | var ( |
| 14 | linkRe = regexp.MustCompile(`https?://[^\s<>"']+`) |
| 15 | youtubeIDRe = regexp.MustCompile(`^[A-Za-z0-9_-]{11}$`) |
| 16 | trailingPunctRe = regexp.MustCompile(`[.,!?:;)]+$`) |
| 17 | ) |
| 18 | |
| 19 | type foundLink struct { |
| 20 | start int |
| 21 | end int |
| 22 | raw string |
| 23 | } |
| 24 | |
| 25 | func findLinks(text string) []foundLink { |
| 26 | indexes := linkRe.FindAllStringIndex(text, -1) |
| 27 | links := make([]foundLink, 0, len(indexes)) |
| 28 | for _, idx := range indexes { |
| 29 | start, end := idx[0], idx[1] |
| 30 | candidate := text[start:end] |
| 31 | trimmed := trailingPunctRe.ReplaceAllString(candidate, "") |
| 32 | if trimmed == "" { |
| 33 | continue |
| 34 | } |
| 35 | trimmedEnd := start + len(trimmed) |
| 36 | if !isHTTPURL(trimmed) { |
| 37 | continue |
| 38 | } |
| 39 | links = append(links, foundLink{ |
| 40 | start: start, |
| 41 | end: trimmedEnd, |
| 42 | raw: trimmed, |
| 43 | }) |
| 44 | } |
| 45 | return links |
| 46 | } |
| 47 | |
| 48 | func isHTTPURL(raw string) bool { |
| 49 | u, err := url.Parse(raw) |
| 50 | if err != nil || u.Host == "" { |
| 51 | return false |
| 52 | } |
| 53 | return u.Scheme == "http" || u.Scheme == "https" |
| 54 | } |
| 55 | |
| 56 | func linkifyMessageText(text string) (string, []string) { |
| 57 | links := findLinks(text) |
| 58 | if len(links) == 0 { |
| 59 | return html.EscapeString(text), nil |
| 60 | } |
| 61 | |
| 62 | var b strings.Builder |
| 63 | urls := make([]string, 0, len(links)) |
| 64 | last := 0 |
| 65 | for _, l := range links { |
| 66 | if l.start < last { |
| 67 | continue |
| 68 | } |
| 69 | b.WriteString(html.EscapeString(text[last:l.start])) |
| 70 | escaped := html.EscapeString(l.raw) |
| 71 | fmt.Fprintf(&b, `<a href="%s">%s</a>`, escaped, escaped) |
| 72 | urls = append(urls, l.raw) |
| 73 | last = l.end |
| 74 | } |
| 75 | b.WriteString(html.EscapeString(text[last:])) |
| 76 | return b.String(), urls |
| 77 | } |
| 78 | |
| 79 | func messageLinks(text string) []string { |
| 80 | links := findLinks(text) |
| 81 | out := make([]string, 0, len(links)) |
| 82 | seen := make(map[string]struct{}, len(links)) |
| 83 | for _, link := range links { |
| 84 | if _, ok := seen[link.raw]; ok { |
| 85 | continue |
| 86 | } |
| 87 | seen[link.raw] = struct{}{} |
| 88 | out = append(out, link.raw) |
| 89 | } |
| 90 | return out |
| 91 | } |
| 92 | |
| 93 | func feedLinks(urls []string) []app.FeedLink { |
| 94 | links := make([]app.FeedLink, 0, len(urls)) |
| 95 | for _, u := range urls { |
| 96 | links = append(links, app.FeedLink{ |
| 97 | Rel: "alternate", |
| 98 | Type: "text/html", |
| 99 | Href: u, |
| 100 | }) |
| 101 | } |
| 102 | return links |
| 103 | } |
| 104 | |
| 105 | func youtubeCanonicalLink(raw string) (string, string, bool) { |
| 106 | u, err := url.Parse(raw) |
| 107 | if err != nil { |
| 108 | return "", "", false |
| 109 | } |
| 110 | host := strings.ToLower(u.Hostname()) |
| 111 | host = strings.TrimPrefix(host, "www.") |
| 112 | host = strings.TrimPrefix(host, "m.") |
| 113 | |
| 114 | videoID := "" |
| 115 | switch host { |
| 116 | case "youtube.com", "youtube-nocookie.com": |
| 117 | path := strings.TrimSuffix(u.Path, "/") |
| 118 | switch path { |
| 119 | case "/watch": |
| 120 | videoID = u.Query().Get("v") |
| 121 | default: |
| 122 | if afterShort, okShort := strings.CutPrefix(path, "/shorts/"); okShort { |
| 123 | videoID = afterShort |
| 124 | } else if afterLive, okLive := strings.CutPrefix(path, "/live/"); okLive { |
| 125 | videoID = afterLive |
| 126 | } |
| 127 | } |
| 128 | case "youtu.be": |
| 129 | videoID = strings.Trim(u.Path, "/") |
| 130 | default: |
| 131 | return "", "", false |
| 132 | } |
| 133 | |
| 134 | if !youtubeIDRe.MatchString(videoID) { |
| 135 | return "", "", false |
| 136 | } |
| 137 | |
| 138 | canonical := "https://www.youtube.com/watch?v=" + videoID |
| 139 | return canonical, videoID, true |
| 140 | } |
| 141 | |
| 142 | func normalizeLinks(rawLinks []string) []string { |
| 143 | out := make([]string, 0, len(rawLinks)) |
| 144 | seen := make(map[string]struct{}, len(rawLinks)) |
| 145 | for _, raw := range rawLinks { |
| 146 | normalized := raw |
| 147 | if canonical, _, ok := youtubeCanonicalLink(raw); ok { |
| 148 | normalized = canonical |
| 149 | } |
| 150 | if _, ok := seen[normalized]; ok { |
| 151 | continue |
| 152 | } |
| 153 | seen[normalized] = struct{}{} |
| 154 | out = append(out, normalized) |
| 155 | } |
| 156 | return out |
| 157 | } |
| 158 | |
| 159 | func firstYouTubeVideoID(urls []string) (string, bool) { |
| 160 | for _, u := range urls { |
| 161 | _, videoID, ok := youtubeCanonicalLink(u) |
| 162 | if ok { |
| 163 | return videoID, true |
| 164 | } |
| 165 | } |
| 166 | return "", false |
| 167 | } |