Extract links and email addresses

Choose what kind of address you need

Start with the purpose of the list. If you are building a list of web references from a meeting note, choose Links. If you need to review addresses written in an event export, choose Emails. Choose Links and emails only when it is useful to keep both kinds in one list. The output is one match per line. It is not an address book with a separate column for the person, organization or source paragraph.

The editor recognizes visible HTTP and HTTPS addresses and common email forms in text. It does not visit the websites, send email or verify that an inbox exists. A result that looks like an address may still contain a mistake or point to a page that is no longer available. If your workflow requires a verification step, keep it separate and check whether you are authorized to use the contacts or open the links.

If you copied a rich-text document, an address may be attached to words like “read the guide” without appearing in the plain-text paste. The tool cannot recover a destination that the source did not include in the clipboard text. Copy the visible URL separately or use the source application's export. The plain-text converter explains what survives the move from formatted text.

Test a duplicate, parentheses and trailing punctuation

Paste the following example and choose Links and emails. Leave Remove duplicates on. The second appearance of the reference should not create a second copy in the result. The closing parenthesis and period around the first link belong to the sentence, not to the destination.

Guide: (https://example.com/setup).
    Questions: [email protected]
    Shared again: https://example.com/setup

The expected output is the setup URL followed by the email address. If you select Emails, only [email protected] remains. If you turn duplicate removal off, the setup URL appears again in its original position. This makes a useful test when you need to decide whether the result is a clean inventory of addresses or a record of repeated mentions.

Automatic punctuation handling is a practical compromise. A raw URL followed immediately by a period in prose usually does not include that period, but an address can intentionally end with punctuation. Balanced parentheses within a path can also be meaningful. When the exact ending matters, compare the extracted value with the original. An editor can recognize patterns, but it cannot know which characters the author meant to include in every context.

Decide whether repeated appearances are meaningful

When a document repeats the same support URL in a heading and a footer, you may want just one item in the output. Duplicate removal retains the first occurrence. For links, the extracted text is the comparison key. The tool does not decide that two different URLs lead to the same destination. A trailing slash, a different query value or an added fragment can make two separate results even if a browser would eventually show a similar page.

For email addresses, the domain is compared without regard to case, while the part before the at sign is preserved as supplied for matching. Do not assume that a modified spelling or a plus alias represents the same inbox. Mail services can apply different conventions. If you are preparing a list for review, preserve the extracted addresses and ask the source owner how duplicates should be handled.

Turn duplicate removal off if every occurrence matters. For example, the same documentation link may appear in several separately reviewed notes. A list of unique values will not tell you how often each appeared or which section contained it. Keep the original notes if you need context, and label the output as a unique list or an occurrence list so the recipient understands what was retained.

Keep source context when a bare address is not enough

A bare URL can be difficult to interpret. Two paths may have nearly identical names while pointing to different product editions or projects. If the source includes a helpful title, date or owner, keep those details in a separate source document or table. This utility extracts addresses only. It does not attach the neighboring label, resolve a redirect or reconstruct the sentence in which the match appeared.

For a mixed list, one useful workflow is to extract just one address type, review a few matches and then repeat for the other type. Save separate outputs named for their purpose, such as references and contact candidates. This makes it harder to mistake an email address for a web source when handing data to another tool. Do not remove a record simply because its URL contains a parameter you do not recognize; check whether it identifies a page state or a required resource.

If you have old and new versions of the same document, extract matching address types from both and use Compare two lists to see which values appear only in one version. Use Compare two texts when you need to know where an address moved or how the surrounding description changed.

Recognize what visible extraction can miss

A domain without a scheme is not necessarily recognized as a link by this utility. A relative path such as /docs/start is not a complete destination because it needs a host. An email address written as “notes at example dot com” is deliberately not in a normal address format. The editor does not infer a missing scheme, repair obfuscated addresses or identify every unusual internationalized form. Use the original structured data or an export when completeness matters.

A long pasted URL can also break across physical lines. If one part is on a new line, the extractor cannot know that it should be joined to the first. Check the source and copy the entire address. For prose copied from a PDF, read the PDF text cleanup guide before attempting to join wrapped material. Do not join a whole list of separate URLs into a single string just to fix one broken address.

HTML source is another special case. An ampersand written as a character reference inside markup may need interpretation in its HTML context. An extracted sequence from raw source is not the same as the final address the browser uses. Use the HTML entity decoder for a known reference, and compare it with the actual address before using it.

Review the result before sharing or importing it

Start with a small known source and check a match at the beginning and the end. Inspect a link with query punctuation and an address that appears twice. If a result is missing, find out whether it was a visible address, a formatted hyperlink target, or a pattern that falls outside the supported syntax. A clean-looking output is not evidence that every possible address was detected.

When you only want addresses matching a literal suffix or a substring, use Filter matching lines on the extracted list. Be precise about what that filter means: a search for a domain name as a substring might match part of a longer address. Compare the result with a few expected matches and a few expected exclusions before taking further action.

Before passing the output to another person, state the source, the extraction mode and whether duplicates were removed. Check that sharing those addresses is appropriate, especially if they were taken from a private message or internal document. Keep the result in plain text when another tool expects one address per row. If it needs names or other attributes, use a structured export rather than trying to reconstruct missing fields from the extracted lines alone.

Reconcile addresses against a known source list

For a migration, run Links on the old instructions and on the new draft. Compare the two outputs by membership. A changed URL in the draft should appear once in the old-only view and once in the new-only view. Inspect the original sentences before accepting the substitution: two references can point to different editions of the same manual. If the domain stayed the same but a query value changed, keep both full addresses long enough to check that difference. Do not shorten or normalize them automatically.

When handing a list to someone who will check individual destinations, include the relevant source section and the intended reason for keeping each address. An extract without context can flag differences but cannot tell which destinations the new document actually needs.