diff --git a/src/core/tumblr/playlist-parser.test.ts b/src/core/tumblr/playlist-parser.test.ts
index 2ce95d1..27debb3 100644
--- a/src/core/tumblr/playlist-parser.test.ts
+++ b/src/core/tumblr/playlist-parser.test.ts
@@ -141,6 +141,60 @@ And this (https://blog.tumblr.com/post/456/another)!
// Should NOT contain google.com redirect
expect(urls[0]).not.toContain('google.com');
});
+
+ it('should extract both linked and bare plain-text URLs from HTML', () => {
+ const html = `
+
+ Linked
+ https://www.tumblr.com/user2/222/bare-post
+
+ `;
+ const urls = extractTumblrUrls(html);
+ expect(urls).toHaveLength(2);
+ expect(urls[0]).toBe('https://www.tumblr.com/user1/111/linked-post');
+ expect(urls[1]).toBe('https://www.tumblr.com/user2/222/bare-post');
+ });
+
+ it('should extract plain-text URLs from HTML even when no hrefs link to tumblr', () => {
+ const html = `
+
+ Not tumblr
+ https://www.tumblr.com/blogger/333/plain-text-only
+
+ `;
+ const urls = extractTumblrUrls(html);
+ expect(urls).toHaveLength(1);
+ expect(urls[0]).toBe('https://www.tumblr.com/blogger/333/plain-text-only');
+ });
+
+ it('should extract URLs split across span tags', () => {
+ const html = `
+
+ https://www.tumblr.com/user/444/split-span-post
+
+ `;
+ const urls = extractTumblrUrls(html);
+ expect(urls).toHaveLength(1);
+ expect(urls[0]).toBe('https://www.tumblr.com/user/444/split-span-post');
+ });
+
+ it('should preserve document order when linked and bare URLs are interleaved', () => {
+ const html = `
+
+ First
+ https://www.tumblr.com/user2/222/second
+ Third
+ https://www.tumblr.com/user4/444/fourth
+
+ `;
+ const urls = extractTumblrUrls(html);
+ expect(urls).toEqual([
+ 'https://www.tumblr.com/user1/111/first',
+ 'https://www.tumblr.com/user2/222/second',
+ 'https://www.tumblr.com/user3/333/third',
+ 'https://www.tumblr.com/user4/444/fourth',
+ ]);
+ });
});
describe('extractLabelFromUrl', () => {
diff --git a/src/core/tumblr/playlist-parser.ts b/src/core/tumblr/playlist-parser.ts
index fb95654..32330d5 100644
--- a/src/core/tumblr/playlist-parser.ts
+++ b/src/core/tumblr/playlist-parser.ts
@@ -34,69 +34,64 @@ export function extractTumblrUrls(text: string): string[] {
// Pattern for username.tumblr.com URLs: https://username.tumblr.com/post/123456/slug
const subdomainPattern = /https?:\/\/[a-zA-Z0-9_-]+\.tumblr\.com\/post\/\d+[^\s<>"'\]]*/gi;
- // Check if this looks like HTML content
- const isHtml = /]/i.test(text);
if (isHtml) {
- // Extract URLs from anchor tags
- // This handles Google Docs HTML export where URLs are in href attributes
- // Google Docs wraps external links: https://www.google.com/url?q=https://tumblr.com/...
- const hrefPattern = /href=["']([^"']*)["']/gi;
- let hrefMatch;
- while ((hrefMatch = hrefPattern.exec(text)) !== null) {
- let url = hrefMatch[1];
- // Decode HTML entities (Google Docs may encode ampersands)
- url = url.replace(/&/g, '&');
-
- // Unwrap Google redirect URLs
- if (url.includes('google.com/url')) {
- const qMatch = url.match(/[?&]q=([^&]+)/);
- if (qMatch) {
- url = decodeURIComponent(qMatch[1]);
+ // Replace tags whose href resolves to a Tumblr URL with the unwrapped
+ // URL as plain text (in place). Non-Tumblr anchors are left untouched so
+ // their inner content is preserved after the subsequent tag strip.
+ processedText = processedText.replace(
+ /]*href=["']([^"']*)["'][^>]*>[\s\S]*?<\/a>/gi,
+ (_match, href: string) => {
+ let url = href.replace(/&/g, '&');
+
+ // Unwrap Google redirect URLs
+ if (url.includes('google.com/url')) {
+ const qMatch = url.match(/[?&]q=([^&]+)/);
+ if (qMatch) {
+ url = decodeURIComponent(qMatch[1]);
+ }
}
- }
- // Only process Tumblr URLs
- if (!url.includes('tumblr.com')) {
- continue;
- }
+ if (url.includes('tumblr.com')) {
+ return ' ' + url + ' ';
+ }
- const cleaned = cleanUrl(url);
- if (cleaned && !seen.has(cleaned)) {
- seen.add(cleaned);
- urls.push(cleaned);
+ return _match;
}
- }
+ );
+
+ // Strip remaining HTML tags so URLs split across s are reassembled
+ processedText = processedText.replace(/<[^>]+>/g, '');
}
- // For non-HTML content, check for plain text URLs
- // Skip this if we already found URLs from HTML to avoid duplicates from href values
- if (!isHtml || urls.length === 0) {
- for (const line of text.split('\n')) {
- const trimmed = line.trim();
-
- // Try tumblr.com pattern
- const tumblrComMatches = trimmed.match(tumblrComPattern);
- if (tumblrComMatches) {
- for (const match of tumblrComMatches) {
- // Clean up the URL - remove trailing punctuation
- const cleaned = cleanUrl(match);
- if (cleaned && !seen.has(cleaned)) {
- seen.add(cleaned);
- urls.push(cleaned);
- }
+ // Single plain-text pass — document order is preserved
+ for (const line of processedText.split('\n')) {
+ const trimmed = line.trim();
+
+ // Try tumblr.com pattern
+ const tumblrComMatches = trimmed.match(tumblrComPattern);
+ if (tumblrComMatches) {
+ for (const match of tumblrComMatches) {
+ const cleaned = cleanUrl(match);
+ if (cleaned && !seen.has(cleaned)) {
+ seen.add(cleaned);
+ urls.push(cleaned);
}
}
+ }
- // Try subdomain pattern
- const subdomainMatches = trimmed.match(subdomainPattern);
- if (subdomainMatches) {
- for (const match of subdomainMatches) {
- const cleaned = cleanUrl(match);
- if (cleaned && !seen.has(cleaned)) {
- seen.add(cleaned);
- urls.push(cleaned);
- }
+ // Try subdomain pattern
+ const subdomainMatches = trimmed.match(subdomainPattern);
+ if (subdomainMatches) {
+ for (const match of subdomainMatches) {
+ const cleaned = cleanUrl(match);
+ if (cleaned && !seen.has(cleaned)) {
+ seen.add(cleaned);
+ urls.push(cleaned);
}
}
}