rss-tools/sources/telegram/page_title.go (view raw)
| 1 | package telegram |
| 2 | |
| 3 | import ( |
| 4 | "context" |
| 5 | "fmt" |
| 6 | "io" |
| 7 | "net/http" |
| 8 | "strings" |
| 9 | |
| 10 | "github.com/PuerkitoBio/goquery" |
| 11 | "golang.org/x/net/html/charset" |
| 12 | ) |
| 13 | |
| 14 | const maxPageBytes = 2 << 20 // 2 MiB |
| 15 | |
| 16 | func fetchPageTitle(ctx context.Context, get func(context.Context, string) (*http.Response, error), rawURL string) (string, error) { |
| 17 | if get == nil { |
| 18 | return "", fmt.Errorf("missing page getter") |
| 19 | } |
| 20 | |
| 21 | resp, err := get(ctx, rawURL) |
| 22 | if err != nil { |
| 23 | return "", err |
| 24 | } |
| 25 | defer resp.Body.Close() |
| 26 | |
| 27 | if resp.StatusCode < http.StatusOK || resp.StatusCode >= http.StatusMultipleChoices { |
| 28 | return "", fmt.Errorf("unexpected status code: %d", resp.StatusCode) |
| 29 | } |
| 30 | |
| 31 | decoded, err := charset.NewReader(io.LimitReader(resp.Body, maxPageBytes), resp.Header.Get("Content-Type")) |
| 32 | if err != nil { |
| 33 | return "", err |
| 34 | } |
| 35 | |
| 36 | doc, err := goquery.NewDocumentFromReader(decoded) |
| 37 | if err != nil { |
| 38 | return "", err |
| 39 | } |
| 40 | |
| 41 | title := normalizePageTitle(doc.Find("title").First().Text()) |
| 42 | if !isMeaningfulPageTitle(title) { |
| 43 | title = metaPageTitle(doc) |
| 44 | } |
| 45 | if !isMeaningfulPageTitle(title) { |
| 46 | return "", fmt.Errorf("page title is empty") |
| 47 | } |
| 48 | return title, nil |
| 49 | } |
| 50 | |
| 51 | func metaPageTitle(doc *goquery.Document) string { |
| 52 | selectors := []string{ |
| 53 | `meta[property="og:title"]`, |
| 54 | `meta[name="og:title"]`, |
| 55 | `meta[property="twitter:title"]`, |
| 56 | `meta[name="twitter:title"]`, |
| 57 | `meta[itemprop="name"]`, |
| 58 | } |
| 59 | |
| 60 | for _, selector := range selectors { |
| 61 | content, ok := doc.Find(selector).First().Attr("content") |
| 62 | if !ok { |
| 63 | continue |
| 64 | } |
| 65 | title := normalizePageTitle(content) |
| 66 | if isMeaningfulPageTitle(title) { |
| 67 | return title |
| 68 | } |
| 69 | } |
| 70 | return "" |
| 71 | } |
| 72 | |
| 73 | func normalizePageTitle(raw string) string { |
| 74 | return strings.Join(strings.Fields(raw), " ") |
| 75 | } |
| 76 | |
| 77 | func isMeaningfulPageTitle(title string) bool { |
| 78 | switch strings.ToLower(strings.TrimSpace(title)) { |
| 79 | case "", "- youtube", "youtube": |
| 80 | return false |
| 81 | default: |
| 82 | return true |
| 83 | } |
| 84 | } |