URL Extractor
Turn a block of text into a plain list of HTTP and HTTPS URL candidates. Keep their original spelling or explicitly choose native URL serialization, filter by hostname and export the complete list without opening any links.
Text stays in this tab and its local worker. No URL requests, DNS lookups, previews or reachability checks.Collect links without opening them
Paste text or upload a file, choose your duplicate and filter settings, then select Extract URLs. Review the list below before copying or downloading.
Up to 1 MiB of UTF-8 text or file content. Large result sets are rejected rather than partly exported. Processing stops after 2.5 seconds; shorten the text if needed.
Use full http:// or https:// links, without login credentials. Bare www addresses, relative paths and other schemes are not supported. Surrounding sentence punctuation is removed under the extraction policy explained below. No link is opened or checked online.
Advanced settings
The original extractor above keeps its original file, filter, spelling and export behavior. These optional tasks use their own sources below.
How to use URL Extractor
Use the original extractor for explicit HTTP/HTTPS links in pasted text or a UTF-8 text file. Its original spelling, duplicate, sort and hostname controls remain. Copy TXT and complete TXT/CSV downloads include every retained URL beyond the 200-row preview. Clear releases working data; Cancel stops the current read or worker.
Open Advanced settings and enable additional extraction for bare-www detection, relative links, tracking previews, source positions, labels, groups and comparison reports. The additional sources and action stay visible when you close settings. Choose plain text tokens or HTML anchors. HTML is parsed without rendering, scripts, uploads or network requests.
Bare-www detection explicitly assumes HTTPS. Relative extraction needs a complete HTTP/HTTPS base without credentials. Plain text recognizes /, ./, ../, ? and # prefixes; arbitrary bare filenames are skipped. HTML href attributes can use ordinary relative filenames. HTML base tags are ignored. Labels use decoded text and image alt text with collapsed whitespace; they do not represent a browser’s visual layout or computed accessible name.
Tracking preview shows before/after URLs and removed parameter names. Apply is a separate checkbox that changes exported/comparison URLs. Removing tracking can change signed links or application behavior. Remaining raw query pairs keep their encoding and order. Original candidates remain in the full report.
Review results as retained links, hostname groups, rejected candidates, filtered rows or removed duplicates. Positions are source line/UTF-16 column and offset; HTML positions point to the href attribute start. Download complete source TXT/CSV, hostnames only or a rejected report. Hostnames are full normalized hosts without ports, not registered-domain results.
To compare, enable a second source and extract both under the same settings. Review only-A, only-B and shared final output strings with occurrence counts after filtering and duplicate handling. Complete comparison CSV and the full JSON report include every entry beyond the visible page. Bounds refuse the whole result rather than truncate it. Clipboard denial provides selected complete text for manual copying.
Example: review links from two project notes
Paste your first notes into source A. Enable bare-www detection only if addresses such as www.example.com/docs should assume HTTPS. For links such as /docs/next, enable relative extraction and provide your own site’s complete base URL.
Preview tracking removal, then review each original and cleaned URL. Enable Apply only when you want cleaned URLs in downloads. Add a second set of notes to source B and compare. A shared link means its final output string appears in both selected lists; it does not prove the pages have the same content.
About URL candidates, HTML labels and tracking
A supported URL candidate has accepted syntax, not verified reachability, ownership, safety or permission to contact it. Credentials, unsupported schemes, backslashes and malformed percent escapes are rejected. Native serialization can normalize hosts, default ports and path segments, while many query/path/fragment differences remain distinct.
Text extraction uses the original whole-token punctuation policy. Surrounding quotes and surplus closing brackets are removed; balanced path brackets remain. Terminal punctuation may be treated as prose, so encode a meaningful final character when needed. HTML extraction reads anchor hrefs and labels through the installed inert HTML parser; no DOM is created and no resources are loaded.
Tracking removal matches utm_* and fbclid, gclid, dclid, msclkid, mc_cid and mc_eid parameter names without regard to case. It is an explicit editorial choice, not a safety check or universal URL equivalence rule. Encoded path-prefix filtering is literal and case-sensitive: /docs also matches /docs-old.
Input stays in memory. Reports may contain private tokens in candidate URLs; choose where to save them. No inputs are automatically stored. Additional extraction has a shared 8-second worker deadline and complete-result bounds.
Stars and feedback
Loading…
What should we improve next?
Help shape the next update to URL extractor. Tell us what would make it better for you.
Prefer email? feedback@tooltulip.com
Tool reference #180