Base64 to PDF Decoder

Base64 is a text encoding of binary data, and it turns up wherever a PDF has to travel through a channel that only carries text: an API response that returns a document as a JSON field, a database column holding an attachment, an email or a support ticket, a log, or a data URL in a web page. Tools like this site's Image to Base64 Converter produce those strings; this tool reverses the process for PDFs specifically.

Paste a base64 payload — on its own, or as a full data:application/pdf;base64,… URL — and the tool decodes it back to the exact bytes it encodes. Newlines and spaces inside the string are ignored, and a missing trailing padding character is restored, because base64 copied out of an email or a wrapped console output is usually broken across lines. When the payload carries a data:…;base64, prefix, the prefix is stripped and the payload read from it.

A base64 string can decode to anything, so the tool checks the result twice before it hands it over. First it looks for the PDF header, the characters %PDF- that every PDF file begins with. Then it opens the bytes with pdf-lib, the same open-source library the site's other PDF tools use, which parses the file's internal structure. Only if both checks pass does it report the page count and offer the download. That two-step check is why the tool can tell you the bytes are "not a PDF" (a PNG signature, say) separately from a PDF that is "damaged or incomplete".

The download is the decoded bytes, byte for byte — nothing is re-saved, re-compressed or rebuilt, so the PDF you get back is identical to the file that was encoded. Decoding happens locally in your browser; the string is never sent to a server, which makes it safe for PDFs that came out of private or production systems. This tool decodes and validates a PDF. It does not repair one: if the file is genuinely damaged, it says so and explains why, rather than pretending to fix it or handing you a corrupt download.

A base64 PDF can be a bare payload or a data URL:

data:application/pdf;base64,<payload>

1. An optional data:…;base64, prefix is stripped, and the payload read from it. 2. All whitespace — spaces, tabs and the newlines a wrapped string carries — is removed. 3. The payload is checked against the base64 alphabet and decoded, 4 characters (3 bytes) at a time. A missing trailing `=` is restored; a length one short of a multiple of 4 cannot come from valid base64 and is refused. 4. The decoded bytes are checked for the %PDF- header in the first 1,024 bytes — the window PDF readers themselves search, to tolerate a little leading junk. 5. The bytes are opened with pdf-lib, which parses the cross-reference table and page tree, so a file that merely starts with %PDF- but is truncated is catchable as damaged. 6. The decoded file is held in memory, the page count is read from the parsed document, and the download is offered.

Base64 inflates binary data by about a third as it encodes it into text, so the reverse is true when you decode: the PDF you download is roughly three-quarters the length of the string you pasted. The reported byte length is the true decoded size — the size of the file you get, not the size of the string.