PDF to XML
Convert PDF documents into structured XML format.
Drag & drop your PDF here
or click to select file
User Guide
Upload the document
Select one PDF, up to 10 MB. This converter runs on a server rather than in your browser, so the file is transmitted, converted, and then removed — do not send anything you are not comfortable having leave the machine.
Check that it holds real text
Conversion reads the text layer. A PDF produced by exporting from Word or a reporting tool has one; a PDF that is photographs of paper does not, and will convert to an XML file describing empty pages. Try selecting a sentence in your reader first — if the cursor highlights words, there is a layer to work with.
Run Convert to XML
Press the button and wait. Time scales with page count and with how much is on each page; a short invoice returns quickly, a hundred-page report takes appreciably longer because every block on every page is described in the output.
Open the result before you build on it
Download the .xml and look at it in an editor that can pretty-print. You are checking one thing: whether the element nesting matches how the document is actually organised. That determines how much work your parser has to do, and it is far cheaper to find out now than halfway through writing it.
About the PDF to XML Converter
These two formats disagree about what a document is, and every difficulty in converting between them follows from that.
The mismatch at the centre of it
A PDF stores drawing instructions. It says “place this glyph in this font at these coordinates”, thousands of times per page. It does not record that a run of glyphs is a heading, that three columns of numbers form a table, or even that a sequence of characters is a word — the spaces you see are often just horizontal offsets between letters, not space characters.
XML is the mirror image: almost no appearance, a great deal of structure. Converting one to the other is therefore not a translation but an inference. The converter groups glyphs into words by measuring the gaps, groups words into lines by their vertical position, groups lines into blocks by proximity and alignment, and emits elements describing what it concluded.
Why some documents convert beautifully and others do not
A PDF exported from Word or a reporting tool lays its content down in a tidy reading order, so the inference is easy and the output maps closely onto the document you would describe by eye. A magazine-style layout with pull quotes, sidebars and text flowing around images does not. The converter can only see coordinates, and coordinates alone cannot always tell it whether the block on the right continues the block on the left or is an unrelated aside.
Why a scan produces nothing
A scanned page contains one large image and no glyph instructions at all. There is nothing to group, so the output describes pages with no content. Optical character recognition must run first to create a text layer — and then you are converting the OCR’s reading of the page, with whatever errors it made.
Choosing XML over the alternatives
Take XML when something downstream expects a schema: a publishing pipeline, a records system, an XSLT transform. If you only need the words, PDF to Text is far simpler to work with. If the document is essentially a table, CSV or XLS gives you rows and columns instead of a tree, which loads into a spreadsheet without further work. Files are capped at 10 MB and processed on a server.
Frequently Asked Questions
Does the XML follow a published schema?
No. The output is a structural description generated from the document, not a document type someone standardised. If your pipeline validates against a specific DTD or XSD, plan on a transform step between this output and that schema.
Why did my scanned contract produce empty elements?
Because a scan is an image. There are no characters in the file for the converter to find, only pixels that look like characters to a human eye. Run the pages through OCR first to create a text layer, then convert the OCR’d version.
Are tables preserved as tables?
Partially, and it depends heavily on how the table was drawn. Ruled tables with clear cell boundaries usually survive as recognisable structure. Tables laid out purely with tab stops and whitespace often come through as ordinary text blocks, because visually that is all they are.
Can I convert several PDFs at once?
Not currently — the tool takes one file per run. For a batch, convert them in sequence; each conversion is independent, so nothing is lost by doing them one at a time.
Is formatting like bold and italic recorded?
Font and style information is available to the converter and is reflected where it can be attached to a block, but do not treat this as a faithful styling export. XML here is about structure; if you need the visual result preserved, PDF to HTML keeps more of the presentation.
What happens to images in the document?
Text is what gets described. Embedded pictures are not written into the XML as data, so if you need the artwork as well, run the file through the Extract Images tool separately.