Public Test Corpus

Our scanner-compatibility claims are only as good as the evidence they're built on. This page is the corpus — the canonical inputs we regression-test every encoder against, plus the phone-camera behaviour you can audit yourself without taking our word for any of it.

Why this exists: most QR tools publish a compatibility matrix. Ours publishes the corpus behind the matrix. If you're responsible for a scanner fleet — retail, warehouse, pharma, airline, DMV — you can run this corpus through your own readers and verify every claim we make.

Download corpus.zip Bundled fixtures JSON + CSV + methodology + README. ~4 KB.

Loading version…

How to use the corpus

Download the fixtures.

Every test vector lives in /test-vectors/fixtures.json. One JSON object per type with the input fields and the expected decoded substring(s). Fetched live from this site — no stale copies.

Regenerate the QRs.

Take any fixture's fields map, drop the values into the generator for that type, and hit Generate. The resulting QR is bit-for-bit identical to what our CI round-trips against.

Scan with your own fleet.

Point an iPhone Camera, Android Camera, Google Lens, industrial scanner (Zebra, Honeywell, Datalogic), or app-specific reader (a banking app, an authenticator) at the generated QR. Confirm the decoded text contains every expect substring for that fixture.

Report discrepancies.

If a scanner in your fleet returns a decode that doesn't match the fixture, file an issue. We'll reproduce, add your scanner to the public compatibility matrix, and (if it's an encoder bug) ship a fix.

Expected scan results — curated subset

This is a compact view of the full fixture JSON, grouped by scanner behaviour. For the exhaustive list across every type Abundera supports, go to /test-vectors/.

Type	Fixture input	iOS Camera	Android Camera	Pro scanner
URL	`https://example.com/test`	Decodes	Decodes	Decodes
WiFi	`SSID=MyNet PSK=... WPA`	Joins network	Joins network	Decodes payload
WiFi Enterprise (EAP)	`WPA2-EAP PEAP identity=alice@corp`	iOS 17+	Decodes	Decodes
vCard	`FN=Alice ORG=Acme ...`	Add to Contacts	Add to Contacts	Decodes payload
TOTP	`otpauth://totp/...?secret=JBSW...`	Authenticator app reads it	Authenticator app reads it	Decodes URI
eSIM	`LPA:1$rsp.example.com$K2-LM...`	iOS 17.4+ provisions	Android 13+ provisions	Decodes URI
Bitcoin (BIP-21)	`bitcoin:1A1zP1eP...`	Wallet app opens	Wallet app opens	Decodes URI
SEPA (EPC)	`BCD header + name + IBAN + amount`	Raw string only	Raw string only	Banking-app support
Swiss QR-bill	`SPC header + IBAN + address + amount`	Raw string only	Raw string only	Swiss banking apps native
GS1 Digital Link	`https://id.gs1.org/01/09506000134352`	Opens URL	Opens URL	POS firmware parses AIs
Data Matrix	`Arbitrary payload`	No decode	Google Lens only	Native
Aztec	`Arbitrary payload`	No decode	Some Android readers	Native
PDF417	`AAMVA or arbitrary`	No decode	Google Lens partial	Native
AAMVA driver license	`@\n\x1e\rANSI 636014080002DL...`	iOS 17+ license-aware	Android license-aware apps	Native
EAN-13	`0012345678905`	Decodes digits	Decodes digits	Every retail POS
Code 128	`ABC-123`	Decodes	Decodes	Native
Micro QR	`Compact payload`	No decode	No decode	Industrial readers only
rMQR	`Rectangular payload`	No decode	No decode	Some Scandit/Cognex firmware

What "expected scan result" means

Decodes — native phone-camera reader returns the payload without needing a separate app.
Partial — decode depends on a specific app (authenticator, wallet, banking) or a specific OS version.
No decode — native camera returns nothing. Requires a specialist scanner or reader.

Methodology notes

Every fixture round-trips. Our CI generates each code in headless Chromium, decodes it back via the browser's BarcodeDetector API, and asserts the substring match. A green CI means the corpus is internally consistent.
Scanner results in this table are from manual testing on real devices. Not synthesized. iPhone 15 Pro on iOS 18.3, Pixel 8 on Android 14, Zebra DS8100 imager on firmware CAAACS00-002-R06.
We publish negative results. If iPhone Camera refuses to decode a symbology (Data Matrix, Aztec, PDF417), that's a "No decode" in the table — not a rebrand of "native reader".
The corpus is versioned. When we add a type or fix an encoder bug, the corresponding fixture entry ships in the same commit. Check the fixture history for the audit trail.

What's coming

Pre-rendered artifacts. The current ZIP bundles fixtures + methodology + CSV. A future revision will add pre-rendered SVG + PNG of each fixture so you can drop them straight into a scanner-fleet harness. Pending build-time headless-browser step.
Scanner-fleet attestation. Dated log at docs/SCANNER_ATTESTATION.md — grows each time we verify against a new device.
Third-party verification. An independent security lab runs the corpus through a scanner fleet and publishes a signed attestation. Pending (commissioned; see /accessibility/audit/ placeholder).