pdfbin.net - test PDFs at stable URLs

§0Quick start

Pin a PDF by its flat URL. Bytes are immutable - the same URL returns the same PDF forever.

# fetch a known-corrupt PDF (truncated at xref)
curl -fSL -o sample.pdf \
  https://pdfbin.net/xref-truncated.pdf
# ~700 bytes; your parser should fall back to xref-recovery

import httpx, pytest
from myparser import parse_pdf

def test_xref_recovery():
    pdf = httpx.get("https://pdfbin.net/"
                    "xref-truncated.pdf").content
    with pytest.raises(EOFError):
        parse_pdf(pdf)

const url = "https://pdfbin.net/" +
            "irs-1040-blank.pdf";
const res = await fetch(url);
const buf = await res.arrayBuffer();
// blank IRS 1040, public-domain via irs.gov
renderToCanvas(buf);

resp, err := http.Get("https://pdfbin.net/" +
    "xref-truncated.pdf")
if err != nil { t.Fatal(err) }
defer resp.Body.Close()
// expected: io.ErrUnexpectedEOF from your parser

Catalog

Five views of the same fixture set. A PDF may appear in more than one section - that's the multi-axis catalog at work.

§1By Form Factor

Canonical clean PDFs. Paper sizes (DIN A4, US Letter, JIS B5), page counts (1 to 500), orientations, and byte sizes (1MB to 50MB).

ID	Description	Facets	URL
`clean-100-pages.pdf`	Clean US Letter, 100 portrait pages.		`https://pdfbin.net/clean-100-pages.pdf`
`clean-10mb.pdf`	Clean US Letter PDF padded to approximately 10 MB via an attached random-bytes file (embedded-file feature).	embedded-file	`https://pdfbin.net/clean-10mb.pdf`
`clean-1mb.pdf`	Clean US Letter PDF padded to approximately 1 MB via an attached random-bytes file (embedded-file feature).	embedded-file	`https://pdfbin.net/clean-1mb.pdf`
`clean-25mb.pdf`	Clean US Letter PDF padded to approximately 25 MB via an attached random-bytes file (embedded-file feature).	embedded-file	`https://pdfbin.net/clean-25mb.pdf`
`clean-500-pages.pdf`	Clean US Letter, 500 portrait pages.		`https://pdfbin.net/clean-500-pages.pdf`
`clean-50mb.pdf`	Clean US Letter PDF padded to approximately 50 MB via an attached random-bytes file (embedded-file feature).	embedded-file	`https://pdfbin.net/clean-50mb.pdf`
`clean-a4-1page.pdf`	Clean DIN A4 (210x297mm), single portrait page, body text.		`https://pdfbin.net/clean-a4-1page.pdf`
`clean-a4-3page.pdf`	Clean DIN A4, three portrait pages.		`https://pdfbin.net/clean-a4-3page.pdf`
`clean-jis-b5-1page.pdf`	Clean JIS B5 (182x257mm) - not ISO B5. Single portrait page.		`https://pdfbin.net/clean-jis-b5-1page.pdf`
`clean-letter-1page.pdf`	Clean US Letter (8.5x11in), single portrait page, body text.		`https://pdfbin.net/clean-letter-1page.pdf`
`clean-letter-3page.pdf`	Clean US Letter, three portrait pages.		`https://pdfbin.net/clean-letter-3page.pdf`
`clean-mixed-orientation.pdf`	US Letter, alternating portrait and landscape across 5 pages.		`https://pdfbin.net/clean-mixed-orientation.pdf`
`clean-paper-sizes-mixed.pdf`	Mixed paper sizes: A4, US Letter, and JIS B5 in one document.		`https://pdfbin.net/clean-paper-sizes-mixed.pdf`

§2By Document Shape

PDFs shaped like real-world documents - fax covers, invoices, IRS forms, lab reports, contracts.

ID	Description	Facets	URL
`bank-statement-letter-clean.pdf`	Bank account statement on US Letter. Synthesized.	bank-statement	`https://pdfbin.net/bank-statement-letter-clean.pdf`
`contract-nda-letter-clean.pdf`	Mutual non-disclosure agreement on US Letter. Synthesized.	contract-nda	`https://pdfbin.net/contract-nda-letter-clean.pdf`
`fax-cover-letter-clean.pdf`	Fax cover sheet on US Letter (To/From/Company/Fax/Pages/Date/Re). Synthesized.	fax-cover-sheet	`https://pdfbin.net/fax-cover-letter-clean.pdf`
`fax-cover-letter-scanned-noisy.pdf`	Fax cover sheet rendered as a noisy 300 DPI scan - realistic faxed-document case.	fax-cover-sheetscanned-noisy	`https://pdfbin.net/fax-cover-letter-scanned-noisy.pdf`
`invoice-letter-clean.pdf`	Invoice on US Letter with line items and totals. Synthesized.	invoice	`https://pdfbin.net/invoice-letter-clean.pdf`
`irs-1040-blank.pdf`	Blank IRS Form 1040, imported verbatim from irs.gov. US federal work, public domain.	irs-1040	`https://pdfbin.net/irs-1040-blank.pdf`
`lab-report-letter-clean.pdf`	Diagnostics lab report on US Letter. Synthesized.	lab-report	`https://pdfbin.net/lab-report-letter-clean.pdf`
`receipt-letter-clean.pdf`	Coffee-shop receipt on US Letter. Synthesized.	receipt	`https://pdfbin.net/receipt-letter-clean.pdf`
`receipt-scanned-noisy-300dpi.pdf`	Receipt rendered as a noisy 300 DPI scan - classic crumpled-receipt OCR target.	receiptscanned-noisy	`https://pdfbin.net/receipt-scanned-noisy-300dpi.pdf`

§3By Spec Compliance

PDF spec versions (1.4 / 1.7 / 2.0) and PDF/A conformance levels (1B, 1A, 2B, 3B).

ID	Description	Facets	URL
`pdf-1.4-clean.pdf`	Clean PDF saved targeting spec version 1.4.		`https://pdfbin.net/pdf-1.4-clean.pdf`
`pdf-1.7-clean.pdf`	Clean PDF saved targeting spec version 1.7.		`https://pdfbin.net/pdf-1.7-clean.pdf`
`pdf-2.0-clean.pdf`	Clean PDF saved targeting spec version 2.0.		`https://pdfbin.net/pdf-2.0-clean.pdf`
`pdfa-1a-compliant.pdf`	PDF/A-1A compliant document (accessible / tagged variant of -1B).	PDF/A-1A	`https://pdfbin.net/pdfa-1a-compliant.pdf`
`pdfa-1b-compliant.pdf`	PDF/A-1B compliant document (visual appearance preserved).	PDF/A-1B	`https://pdfbin.net/pdfa-1b-compliant.pdf`
`pdfa-2b-compliant.pdf`	PDF/A-2B compliant document (PDF 1.7 features allowed).	PDF/A-2B	`https://pdfbin.net/pdfa-2b-compliant.pdf`
`pdfa-3b-with-attachment.pdf`	PDF/A-3B with an embedded plain-text attachment - the headline PDF/A-3 feature.	PDF/A-3Bembedded-file	`https://pdfbin.net/pdfa-3b-with-attachment.pdf`

§4By Provenance

Digital-native vs scanned variants. Scans vary by DPI, noise, and skew - useful for OCR test suites.

ID	Description	Facets	URL
`fax-cover-letter-scanned-noisy.pdf`	Fax cover sheet rendered as a noisy 300 DPI scan - realistic faxed-document case.	fax-cover-sheetscanned-noisy	`https://pdfbin.net/fax-cover-letter-scanned-noisy.pdf`
`receipt-scanned-noisy-300dpi.pdf`	Receipt rendered as a noisy 300 DPI scan - classic crumpled-receipt OCR target.	receiptscanned-noisy	`https://pdfbin.net/receipt-scanned-noisy-300dpi.pdf`
`scanned-clean-200dpi.pdf`	Clean 200 DPI scan of a generic Letter PDF.	scanned-clean	`https://pdfbin.net/scanned-clean-200dpi.pdf`
`scanned-clean-300dpi.pdf`	Clean 300 DPI scan of a generic Letter PDF.	scanned-clean	`https://pdfbin.net/scanned-clean-300dpi.pdf`
`scanned-noisy-300dpi.pdf`	Noisy 300 DPI scan with high-density speckle noise.	scanned-noisy	`https://pdfbin.net/scanned-noisy-300dpi.pdf`
`scanned-skewed-3deg.pdf`	Clean 300 DPI scan rotated 3 degrees.	scanned-skewed	`https://pdfbin.net/scanned-skewed-3deg.pdf`
`scanned-skewed-noisy.pdf`	Noisy 3-degree skewed 300 DPI scan - hardest realistic case.	scanned-noisy-skewed	`https://pdfbin.net/scanned-skewed-noisy.pdf`

§5By Failure Mode

PDFs damaged in known, named ways. Each PDF's facet record names the exact corruption.

ID	Description	Facets	URL
`byte-flipped-mid-stream.pdf`	One byte XOR-flipped mid-content-stream. Likely renders pages with garbage.	corrupt-byte-flipped	`https://pdfbin.net/byte-flipped-mid-stream.pdf`
`eof-missing.pdf`	%%EOF marker stripped. Parsers that key on it cannot find the end.	corrupt-eof-missing	`https://pdfbin.net/eof-missing.pdf`
`header-truncated.pdf`	First line (%PDF-1.7) removed. Parsers that key on the header fail to detect a PDF.	corrupt-header-truncated	`https://pdfbin.net/header-truncated.pdf`
`object-generation-mismatch.pdf`	First object's header says generation 1; xref says generation 0.	corrupt-object-generation-mismatch	`https://pdfbin.net/object-generation-mismatch.pdf`
`stream-length-mismatch.pdf`	One stream object's /Length is overstated by 99 bytes.	corrupt-stream-length-mismatch	`https://pdfbin.net/stream-length-mismatch.pdf`
`trailer-missing.pdf`	Trailer dictionary removed; startxref present but points to nothing useful.	corrupt-trailer-missing	`https://pdfbin.net/trailer-missing.pdf`
`xref-truncated.pdf`	PDF byte-truncated at the start of the xref table. Parsers without xref-recovery will fail.	corrupt-xref-truncated	`https://pdfbin.net/xref-truncated.pdf`