How can I tell whether a PDF I received holds real text or only scanned images before I archive it?
Run the file through the text extractor and read the summary line, which says how many of the pages have text. Every page without a text layer gets a notice: it looks scanned if it paints an image, or empty if it holds nothing. If most pages are flagged, the file is not searchable, and the tool does no OCR and says so.
Why it matters before archiving
Two files that look identical on screen can be completely different inside. One is an invoice exported by your accounting system, with real characters you can search for an invoice number a year from now. The other is the same invoice printed and scanned, which is only a picture. Archive the second thinking it is the first, and a search of your folder will never find it when you need it.
The one minute check
- Open the PDF text extractor and choose the file.
- For a long file, there is no need to read every page: type a sample in the Pages field, such as 1 5 10.
- Choose Page by page and press Extract text.
- Read the summary line, then each page heading, which shows either a character count or the words no text layer.
Reading the result
Each page ends up in one of three states, sorted by a simple rule: if characters were found it is a text page; if there are no characters but the page paints an image it is most likely a scan; if there is neither, it is blank.
A 12 page file shows the summary: Pages with text: 9 of 12.
The other three pages carry a notice such as: Page 11 has no text layer: it looks like a scanned image.
A white divider page shows: holds neither text nor an image: it looks empty.
If the whole file is scanned, a general warning says none of the selected pages has a text layer.
Cases that can fool you
- A scan whose scanner app added an OCR text layer: the tool treats it as text, because there really are characters in it. Read a sample of the extracted text to make sure the words are right and not gibberish.
- A page with a single line of text, such as a date stamp a system added, over a scanned image: it counts as a text page, but its tiny character count gives it away. A real invoice page has hundreds of characters, not a few dozen.
- Arabic may come out in a reversed order in some files. That does not mean the page is scanned; it means the producing program stored the text in its own way.
If you need a scanned file archived in searchable form, the answer is OCR software, which this tool does not include. If you only want to look at a scanned page as an image to confirm what is on it, convert it with PDF to images.
Common questions
Does the check change the file?
No. The tool only reads; it neither changes the file nor produces a copy of it.
Is there a quicker way?
You can try selecting a word in your PDF reader, but that tests one page. The tool gives you the full picture for several pages at once.
Is the file sent to a server for the check?
No. All the reading happens in your browser.