EPUB samples: a validation and repair lab

Name: EPUB validation and repair lab
Creator: Epublys
License: https://creativecommons.org/publicdomain/zero/1.0/

A passing check is useful only if you know what it checked. Start with a small original EPUB, change one thing, and compare the reports. These 9 files are free to download, reuse and modify.

Choose the control for reading and conversion. The other files are deliberately faulty teaching examples. Do not publish them as finished books.

Download the complete lab Browse individual files

Original text and artwork · CC0 reuse · No account required

Try one fault at a time

Download the valid control. Open it in your reader, or use it to try a conversion without uploading a personal book.
Pick one faulty variant and inspect its source change below.
Compare EPUBCheck 5.4.0 with the Epublys structural checker. Start with Deep scan enabled and Auto-fix off so you can read the input diagnostics.
Optionally run the file again with Auto-fix on. When a repair is available, the online tool returns that EPUB instead of the input report. Inspect and recheck the output. Each online run uses the normal allowance; downloading the recorded examples below does not.

The sample files

Results below describe these exact bytes. Each file after the control intentionally breaks a format rule. EPUBCheck and Epublys are measured separately: a checker can miss a deliberate fault. The raw results manifest includes diagnostic locations, the lighter ZIP check, repair outcomes and file hashes.

Sample	EPUBCheck 5.4.0	Epublys structural scan
Complete two-chapter EPUB	No error reported	No error reported
Wrong mimetype text	Errors	Error reported
Mimetype is not first	Errors	Error reported
Missing package language	Errors	Warning reported
Missing container.xml	Errors	Error reported
Missing chapter file	Errors	Error reported
Broken image reference	Errors	Warning reported
Duplicate ID inside one chapter	Errors	Fault not detected
Compressed mimetype only	No error reported	Fault not detected

CONTROL FILE

Complete two-chapter EPUB

Original prose, navigation, linked chapters and an SVG diagram; no deliberate fault.

EPUBCheck 5.4.0: No error reported
Epublys structural check: No error reported
Automatic repair: No repair needed

Download valid sample3,269 bytes · EPUB

Source change, diagnostics and checksum

This is the unchanged control. Start here before trying the faulty variants.

EPUBCheck output

No diagnostic messages.

Epublys output before repair

valid: true; fixCount: 0. A true structural verdict is not proof of EPUB conformance.

No warnings or errors reported by this structural scan.

Input SHA-256: 7dc8def3275e27dd212a9d71ccfa95744066335ce5d7c32c99427e6da6703133

INTENTIONAL FAULT

Wrong mimetype text

The first uncompressed mimetype entry has the wrong value.

EPUBCheck 5.4.0: Reports errors
Epublys structural check: Error reported
Automatic repair: A repaired copy is available below

Download faulty sample3,259 bytes · EPUB

Source change, diagnostics and checksum

Changed location: mimetype. One intentional change can produce several diagnostic messages.

Before

application/epub+zip

After

text/plain

EPUBCheck output

PKG-007 (ERROR): The content of the mimetype file must be the string "application/epub+zip"

Epublys output before repair

valid: false; fixCount: 1. A true structural verdict is not proof of EPUB conformance.

error: Invalid mimetype: "text/plain"

Recorded repair

Download the repaired EPUB (3,269 bytes). EPUBCheck 5.4.0 passes this repaired copy.

Repair SHA-256: 7dc8def3275e27dd212a9d71ccfa95744066335ce5d7c32c99427e6da6703133

Input SHA-256: 774cbd47acb74912933abaeca2f1ddd165daf73c927582f252769759c81d490d

INTENTIONAL FAULT

Mimetype is not first

The correct uncompressed mimetype entry is moved to the end of the ZIP.

EPUBCheck 5.4.0: Reports errors
Epublys structural check: Error reported
Automatic repair: A repaired copy is available below

Download faulty sample3,269 bytes · EPUB

Source change, diagnostics and checksum

Changed location: mimetype. One intentional change can produce several diagnostic messages.

Before

first ZIP entry

After

last ZIP entry

EPUBCheck output

PKG-006 (ERROR): Mimetype file entry is missing or is not the first file in the archive.

Epublys output before repair

valid: false; fixCount: 1. A true structural verdict is not proof of EPUB conformance.

error: Misplaced mimetype file: it must be the first entry in the archive

Recorded repair

Download the repaired EPUB (3,451 bytes). EPUBCheck 5.4.0 passes this repaired copy.

Repair SHA-256: bfebf07ba40eeef3309c5d8f5598ba12bc6fd9fadd2c9f4c97fb86bce08b4852

Input SHA-256: 700c943bda93da96e9818e04b8a7942ece61caf7aa3a33d44f67c61a1142fb88

INTENTIONAL FAULT

Missing package language

Only dc:language is removed; the XHTML language attributes remain present.

EPUBCheck 5.4.0: Reports errors
Epublys structural check: Warning reported
Automatic repair: A repaired copy is available below

Download faulty sample3,256 bytes · EPUB

Source change, diagnostics and checksum

Changed location: OEBPS/package.opf. One intentional change can produce several diagnostic messages.

Before

    <dc:language>en</dc:language>

After

EPUBCheck output

RSC-005 (ERROR): Error while parsing file: element "metadata" incomplete; missing required element "dc:language"

Epublys output before repair

valid: true; fixCount: 1. A true structural verdict is not proof of EPUB conformance.

warning: Missing dc:language in metadata

Recorded repair

Download the repaired EPUB (3,358 bytes). EPUBCheck 5.4.0 passes this repaired copy.

Repair SHA-256: 9bbce1901bbafb75ddafeb0cfc3e534c50e6659502ddc12b593ca166d90412fb

Input SHA-256: 5f0f2b1ea7e5884ad6d05c9b4773362a348531f4c07d11c3d1e83f130789e6b7

INTENTIONAL FAULT

Missing container.xml

Only META-INF/container.xml is removed; the package document is still present.

EPUBCheck 5.4.0: Reports errors
Epublys structural check: Error reported
Automatic repair: No repaired copy produced

Download faulty sample2,977 bytes · EPUB

Source change, diagnostics and checksum

Changed location: META-INF/container.xml. One intentional change can produce several diagnostic messages.

Before

entry present

After

entry absent

EPUBCheck output

RSC-002 (FATAL): Required META-INF/container.xml resource could not be found.

Epublys output before repair

valid: false; fixCount: 0. A true structural verdict is not proof of EPUB conformance.

error: Missing META-INF/container.xml
error: Cannot locate OPF package file

Input SHA-256: 04ece61b18efbfd44b4a45bd247ede60524ae364a088051350debb7530bcad3d

INTENTIONAL FAULT

Missing chapter file

The first chapter file is removed, leaving its manifest, spine and navigation references intact.

EPUBCheck 5.4.0: Reports errors
Epublys structural check: Error reported
Automatic repair: No repaired copy produced

Download faulty sample2,612 bytes · EPUB

Source change, diagnostics and checksum

Changed location: OEBPS/chapter-1.xhtml. One intentional change can produce several diagnostic messages.

Before

entry present

After

entry absent

EPUBCheck output

RSC-001 (ERROR): File "OEBPS/chapter-1.xhtml" could not be found.
RSC-012 (ERROR): Fragment identifier is not defined.

Epublys output before repair

valid: false; fixCount: 0. A true structural verdict is not proof of EPUB conformance.

error: Manifest references missing file: chapter-1.xhtml
warning: 1 broken internal link(s) found

Input SHA-256: 7bda1912d76c11492ad4f024adb601c776e62ec2ebadc30ee1ac5a8ceb0fa4eb

INTENTIONAL FAULT

Broken image reference

One img src points to a nonexistent file; the original diagram remains in the archive.

EPUBCheck 5.4.0: Reports errors
Epublys structural check: Warning reported
Automatic repair: No repaired copy produced

Download faulty sample3,274 bytes · EPUB

Source change, diagnostics and checksum

Changed location: OEBPS/chapter-1.xhtml. One intentional change can produce several diagnostic messages.

Before

src="diagram.svg"

After

src="missing-diagram.svg"

EPUBCheck output

RSC-007 (ERROR): Referenced resource "OEBPS/missing-diagram.svg" could not be found in the EPUB.

Epublys output before repair

valid: true; fixCount: 0. A true structural verdict is not proof of EPUB conformance.

warning: 1 broken image reference(s) found

Input SHA-256: 2a23463586bd75a20aebbf53f28f3d9940195f866c859de4bb0d29da8f7b2b5f

INTENTIONAL FAULT

Duplicate ID inside one chapter

Two paragraphs in the same XHTML document have the same ID. IDs in separate documents are a different case.

EPUBCheck 5.4.0: Reports errors
Epublys structural check: Fault not detected
Automatic repair: No repaired copy produced

Download faulty sample3,264 bytes · EPUB

Source change, diagnostics and checksum

Changed location: OEBPS/chapter-1.xhtml. One intentional change can produce several diagnostic messages.

Before

id="second-note"

After

id="first-note"

EPUBCheck output

RSC-005 (ERROR): Error while parsing file: Duplicate ID "first-note"

Epublys output before repair

valid: true; fixCount: 0. A true structural verdict is not proof of EPUB conformance.

No warnings or errors reported by this structural scan.

Input SHA-256: 73cbd030afc71af42cbf2214565efa7aca80921d26e9418aec621fe52a12c6cc

INTENTIONAL FAULT

Compressed mimetype only

Only the first mimetype entry changes from STORE to DEFLATE. Its text and position remain correct.

EPUBCheck 5.4.0: No error reported
Epublys structural check: Fault not detected
Automatic repair: No repaired copy produced

Download faulty sample3,271 bytes · EPUB

Source change, diagnostics and checksum

Changed location: mimetype. One intentional change can produce several diagnostic messages.

Before

STORE (0)

After

DEFLATE (8)

EPUBCheck output

No diagnostic messages.

Epublys output before repair

valid: true; fixCount: 0. A true structural verdict is not proof of EPUB conformance.

No warnings or errors reported by this structural scan.

Input SHA-256: 561551f8acfdd30a6e75068d14614cd8cd5c46013f1402b5684e427d67dbd2ed

What these results do not prove

The examples deliberately expose checker limits. In this recorded run, duplicate IDs inside XHTML and a compressed-only mimetype defect escape Epublys's detailed scan. Its lighter ZIP check rejects the compressed mimetype. EPUBCheck 5.4.0 also reports no error for that compressed-only file, although the OCF mimetype rule requires it to be stored without ZIP compression. A reference-checker pass is not a guarantee that every format rule was checked.

The missing-language repair inserts en. That matches the English control in this collection; it is not a language detector. Set the correct language for your own book.

A conforming file can still have poor layout or inaccessible content. Preview it in the destination reading system and perform accessibility checks separately. These tiny fixtures are not a representative ebook benchmark or evidence of typical compression savings.

Source and reproducibility

Collection version 1.0.0, published 2026-10-17. The results record the validator source and options used; software behavior can change after a release. Recorded repair ZIP timestamps are normalized for reproducible checksums without changing entry content, compression or order.

The complete archive includes the original sources, a standalone build package, recorded observations, repaired files and reuse terms. The builder uses JSZip 3.10.2; it does not include Epublys's private application code. Download the source package before running these commands; the script needs the included source and observation files.

unzip epub-validation-lab-source-v1.zip -d epub-validation-lab-source
cd epub-validation-lab-source
npm install --ignore-scripts
node build.mjs --out ./epub-samples
java -jar /path/to/epubcheck.jar ./epub-samples/baseline.epub --failonwarnings

Use the README for the exact build layout and pinned checker instructions. Verify downloads against SHA256SUMS.txt. You can inspect the builder script and collection licence separately.

Reuse and cite the lab

These newly authored sample files and their builder are released under CC0 1.0. You can use them in classes, documentation and software tests. Attribution is welcome, and is not required. The licence applies to this collection, not to the Epublys service or third-party dependencies.

Optional citation: Epublys, EPUB validation and repair lab, version 1.0.0, 2026-10-17, epublys.com/epub-samples.

For broader coverage, use the W3C/IDPF feature samples, the W3C reading-system tests and EPUBTest accessibility books. The EPUB specification defines the format; this lab illustrates a small part of it.