Document Contacts & Links Extractor

The extractor applies E.164 pattern formatting to sanitize brackets, hyphens, and country prefixes, outputting standardized dialable telephone numbers.

Because a scanned page is a photograph of text, not text. There are no characters in the file for a parser to match — only pixels that look like characters to a human eye. Run the document through Scanned PDF to Text first, which uses OCR to produce a real text layer, then extract from that. A quick way to tell which kind of PDF you have: if you cannot select a sentence with your cursor in a PDF viewer, there is no text layer.

Two kinds, predictably. Numbers that are the right shape but not phone numbers — invoice references, order IDs, tracking numbers and long dates all match phone patterns in some regions. And addresses that are not people's addresses: noreply senders, mailing-list footers, and the document author's own contact block. The domain blocklist handles the recurring cases such as noreply and unsubscribe. Pattern matching cannot know intent, so treat the output as a reviewed shortlist rather than a finished contact list.

Searching finds one thing at a time in one document. Extraction gives you every match across a set of files in one structured table, de-duplicated and normalised, which is what you need before importing anywhere. From there the practical next step is usually CSV / Excel to VCF, which turns the table into contact cards a phone or mail client will actually import.