wacz-validator-corpus
A collection of WACZ specimens for exercising wacz-validator validation rules.
It is not a collection of good archives
Section titled “It is not a collection of good archives”| count | |
|---|---|
| Deliberately broken | 28 |
| Passing everything | 2 (good, good-webrecorder) |
This provides WACZ-format sample data to wacz-validator. For 28 of the 29 specimens, producing no findings is a failure.
One of those 28 is not broken so much as loyal to a different spec:
datapackage-schema-only is a valid Data Package v2 descriptor, and it fails
only because WACZ 1.1 requires the profile that v2 removed.
Broken is not the same as invalid
Section titled “Broken is not the same as invalid”datapackage-frictionless-bad-name.wacz reports valid: true. Its
resources[0].name is DATA.warc.gz instead of data.warc.gz — uppercase,
which the Frictionless v1 schema forbids — but that is a warning, and warnings
do not overturn validity.
So “has a finding” and “is invalid” are separate axes. The catalogue records both.
What is here
Section titled “What is here”fixtures/ every specimen (Git LFS)manifest.json what wacz-validator actually reports for each onescripts/ check-manifest.mjs manifest ↔ fixtures consistency (no dependencies)docs-site/ this siteNeither the specimens nor the manifest are written by hand — see Regenerating.
Where to go next
Section titled “Where to go next”- Getting started — obtaining the archives (they are in Git LFS)
- Catalogue — every specimen and what it violates
- Using it — reaching for the corpus while writing a rule
- manifest — the schema, and what
defaultProfilemeans - Regenerating — how specimens are produced, and what gets destroyed
- Checks — the three checks and what each one covers