Add Twitter/X syndication API support for image extraction

Twitter/X blocks regular scrapers and doesn't provide og:image meta tags.
This uses the cdn.syndication.twimg.com API to fetch tweet data including
photos, video posters, and card thumbnails.

Note: The API may return 404 for protected/deleted tweets or accounts
with restricted access - in those cases we fall back gracefully.

internal/services/metadata.go

@@ -11,6 +11,7 @@ import (
 	"net/http"
 	"net/url"
 	"path"
+	"regexp"
 	"strings"
 	"time"
 
@@ -21,6 +22,8 @@ import (
 	"git.soup.land/soup/lookbook/internal/data/item"
 )
 
+var tweetURLPattern = regexp.MustCompile(`^https?://(?:www\.)?(?:twitter\.com|x\.com)/([^/]+)/status/(\d+)`)
+
 const thumbWidth = 480
 
 func CreateItemFromURL(ctx context.Context, db *sql.DB, sourceURL string) (item.Row, error) {
@@ -66,6 +69,11 @@ type Metadata struct {
 }
 
 func FetchMetadata(ctx context.Context, sourceURL string) (Metadata, error) {
+	// Check if this is a Twitter/X URL and use syndication API
+	if meta, ok := fetchTwitterMetadata(ctx, sourceURL); ok {
+		return meta, nil
+	}
+
 	resp, err := fetchURL(ctx, sourceURL)
 	if err != nil {
 		return Metadata{}, err
@@ -129,6 +137,120 @@ type oEmbedResponse struct {
 	ProviderName string `json:"provider_name"`
 }
 
+// Twitter/X syndication API response structures
+type twitterSyndicationResponse struct {
+	Text   string              `json:"text"`
+	User   twitterUser         `json:"user"`
+	Photos []twitterPhoto      `json:"photos"`
+	Video  *twitterVideo       `json:"video"`
+	Card   *twitterCard        `json:"card"`
+	Media  []twitterMediaEntry `json:"mediaDetails"`
+}
+
+type twitterUser struct {
+	Name       string `json:"name"`
+	ScreenName string `json:"screen_name"`
+}
+
+type twitterPhoto struct {
+	URL string `json:"url"`
+}
+
+type twitterVideo struct {
+	Poster string `json:"poster"`
+}
+
+type twitterCard struct {
+	ThumbnailImageOriginal string `json:"thumbnail_image_original"`
+}
+
+type twitterMediaEntry struct {
+	MediaURLHTTPS string `json:"media_url_https"`
+	Type          string `json:"type"`
+}
+
+// fetchTwitterMetadata attempts to fetch metadata from Twitter's syndication API
+// Returns the metadata and true if successful, or empty metadata and false if not a Twitter URL or fetch failed
+func fetchTwitterMetadata(ctx context.Context, sourceURL string) (Metadata, bool) {
+	matches := tweetURLPattern.FindStringSubmatch(sourceURL)
+	if matches == nil {
+		return Metadata{}, false
+	}
+
+	username := matches[1]
+	tweetID := matches[2]
+
+	syndicationURL := fmt.Sprintf("https://cdn.syndication.twimg.com/tweet-result?id=%s&token=0", tweetID)
+
+	req, err := http.NewRequestWithContext(ctx, http.MethodGet, syndicationURL, nil)
+	if err != nil {
+		return Metadata{}, false
+	}
+
+	// The syndication API requires specific headers
+	req.Header.Set("User-Agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36")
+	req.Header.Set("Referer", "https://platform.twitter.com/")
+
+	client := &http.Client{Timeout: 12 * time.Second}
+	resp, err := client.Do(req)
+	if err != nil {
+		return Metadata{}, false
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != 200 {
+		return Metadata{}, false
+	}
+
+	// Check content type - if it's HTML, it's an error page
+	contentType := resp.Header.Get("Content-Type")
+	if !strings.Contains(contentType, "application/json") {
+		return Metadata{}, false
+	}
+
+	var tweet twitterSyndicationResponse
+	if err := json.NewDecoder(resp.Body).Decode(&tweet); err != nil {
+		return Metadata{}, false
+	}
+
+	meta := Metadata{
+		Title:       truncateText(tweet.Text, 200),
+		Description: fmt.Sprintf("@%s", username),
+		SiteName:    "X",
+	}
+
+	// Try to get image URL from various sources
+	// Priority: photos > video poster > card thumbnail > mediaDetails
+	if len(tweet.Photos) > 0 {
+		meta.ImageURL = tweet.Photos[0].URL
+	} else if tweet.Video != nil && tweet.Video.Poster != "" {
+		meta.ImageURL = tweet.Video.Poster
+	} else if tweet.Card != nil && tweet.Card.ThumbnailImageOriginal != "" {
+		meta.ImageURL = tweet.Card.ThumbnailImageOriginal
+	} else if len(tweet.Media) > 0 {
+		meta.ImageURL = tweet.Media[0].MediaURLHTTPS
+	}
+
+	// If we got user info, use it for a better description
+	if tweet.User.Name != "" {
+		meta.Description = fmt.Sprintf("%s (@%s)", tweet.User.Name, tweet.User.ScreenName)
+	}
+
+	return meta, true
+}
+
+func truncateText(s string, maxLen int) string {
+	if len(s) <= maxLen {
+		return s
+	}
+	// Try to break at a word boundary
+	truncated := s[:maxLen]
+	if lastSpace := strings.LastIndex(truncated, " "); lastSpace > maxLen/2 {
+		truncated = truncated[:lastSpace]
+	}
+	return truncated + "…"
+}
+
 func fetchOEmbed(ctx context.Context, sourceURL string) (oEmbedResponse, error) {
 	oembedURL := fmt.Sprintf("https://noembed.com/embed?url=%s", url.QueryEscape(sourceURL))
 	resp, err := fetchURL(ctx, oembedURL)