Split a PDF wherever a phrase appears

A batch print is one PDF containing three hundred documents, and the thing that marks the start of each is the words on the page. Type those words and every page containing them begins a new file. Pages before the first match stay together in a file of their own, so a covering sheet is not glued to the first invoice, and the result tells you how many pages matched — which is how you know the phrase was the right one.

It reads the page’s text layer. A scan or a photograph has none, and the job stops with a message rather than handing back one big file; the OCR tool reports which pages are affected.

Split by a phrase

No account. Your upload is deleted after 2 hours, the result after 24 hours.

How it works

  1. Find a phrase only first pages have. Open two of the documents inside the file and look for a line that appears on the first page of each and on no other page. A phrase repeated in a footer is the trap: every page would then start a new file.
  2. Type it in and run. Capital letters are ignored unless you ask for them to count, and a phrase that wraps onto two lines in the PDF still matches, because both sides are flattened to single spaces before they are compared.
  3. Check the count before you use the files. The result says how many pages matched. Three hundred matches in a three-hundred-page file means you matched something on every page; three where you expected sixty means the phrase is not the one.

What the options do

The form above lists the options this tool is running today, straight from the server. This is what they are for.

Text that starts a new file
The words that begin each document. Whitespace is flattened on both sides before matching, so a phrase broken across two lines still counts and a heading spread out with extra spaces does not defeat it. It is a plain substring search rather than a pattern: you cannot ask for ‘any six-digit number’, so pick the fixed label next to it — ‘Account number’, ‘Invoice number’, ‘Policy no’ — instead. Text is read with position sorting on, so a two-column page does not interleave its columns in the middle of your phrase.
Match capital letters exactly
Off by default, so INVOICE, Invoice and invoice all count as the same words. Turn it on when the capitals are what identify the page — a statement where ‘TOTAL’ in capitals only ever appears in the closing summary, for instance, while ‘total’ turns up in the notes on every page.

What this tool does not do

  • It reads the page’s text layer. A scan or a photograph has none, and the job stops with a message rather than handing back one big file; the OCR tool reports which pages are affected.
  • The phrase decides where the cuts go, not what the files are called. Parts are numbered in page order — split by bookmarks is the tool that names parts after their contents.
  • It cannot start a new file part-way down a page. The boundary is always a page boundary.
  • It matches text, not patterns. There are no wildcards, no regular expressions and no ‘any number’.
  • The parts are new documents, so bookmarks are not carried over.

What happens to your file

The file is uploaded because Split PDF by text runs on our server rather than in the browser tab. It is kept for at most 2 hours and the result you download for 24 hours, then a scheduled job deletes both — a job has to be able to read the file while it runs, and you have to be able to retry a failed conversion without re-uploading. Neither needs longer.

Nothing about the contents is logged, and no one reads your documents. The retention page lists the same numbers the code enforces, and the privacy policy says who can reach the storage they sit in.

Questions

How do I split a file of 200 invoices into 200 files?
Find the words that appear on the first page of every invoice — usually a label such as ‘Invoice number’ or ‘Tax invoice’ — and type them in. Every page containing them starts a new file, so as long as the label is not repeated in a footer you get one file per invoice.
It says no page contains my phrase, but I can see it on the page.
Two likely reasons. The document is a scan, so the page is a picture and there is no text to search; the OCR tool will confirm that. Or the phrase differs by a character you cannot see, such as a non-breaking space or a different kind of hyphen — try a shorter fragment of it.
Can I name each file after the customer or the invoice number?
Not here — the parts are numbered in page order. Split by bookmarks is the tool that names parts after their content, so if the batch has an outline, use that instead.
What happens to the pages before the first match?
They become the first file, and a note tells you where the first match was. That is deliberate: a covering letter or a control sheet at the front of a batch print is not part of the first document.
Does it match across a line break?
Yes. The page’s text and your phrase are both flattened to single spaces before they are compared, so a heading that wraps in the middle of your phrase still counts. It does not match across a page break, because the boundary is always the page.