Extract Emails from Text – Free Tool

Pulling addresses out of a block of text seems like a single pattern match, and the pattern everyone writes first is wrong in both directions: it misses valid addresses and it captures things that are not addresses at all.

It misses more than people expect because the real specification is permissive. Plus-addressing, the form used for filtering, is valid and is sometimes dropped by naive patterns. Subdomains are valid. Newer top-level domains are long, so a pattern assuming two to four letters after the final dot fails on them. Internationalised addresses using non-Latin characters are valid and almost never matched.

It captures the wrong things just as readily. A version string or a file path with an at sign in it. A social handle written as @name followed by a domain elsewhere in the sentence. An address that has been deliberately obfuscated as name at example dot com, which a human reads and a pattern does not. Trailing punctuation is the most common defect of all: an address at the end of a sentence captures the full stop, and the result is an address that is subtly invalid everywhere it is used.

After extraction, the work is usually deduplication, and the same-list problem applies with a twist. Addresses are effectively case-insensitive, so two spellings are one person and should be merged. Whitespace and invisible characters defeat naive equality, so cleaning before deduplicating is what makes it work.

Worth stating plainly: extracting addresses from a page you do not own, to contact people who did not ask to be contacted, is what anti-spam law exists to prevent, and it is a poor use of anyone's time. The legitimate uses are your own data: pulling recipients out of a thread you are part of, recovering a list from an export, or auditing where addresses appear in a document before sharing it.

The extraction runs in your browser, which for this tool is the whole point, since the input is by definition a list of people's contact details.