Scann fictional document collection, revised designs

The collection contains 50 PDFs in five categories, with ten documents per category.
The PDFs contain 62 pages. All 50 original data objects are unchanged.

What changed

Suppliers, merchants, hotels and venues have separate visual identities, contact
details and reference formats. Layouts vary in typography, alignment, tables,
letterheads and use of colour. Documents from Mossfield Community Studio retain
the same organisation identity, with separate minutes, brief and report formats.
Personal inventories and study material use simpler office and academic layouts.

The four receipts are still 80 mm wide. Their lengths now follow their content
instead of leaving unused paper below the printed text. Other paper dimensions
are unchanged. Hotel reservations and reading summaries each fit on one page.
Itineraries keep one day per page. Worksheets have separate answer-key pages.

Dataset identifiers are no longer printed on the pages. They remain in the YAML,
filenames and PDF keywords. Visible references follow the fictional issuer's own
format. The presentation object in each YAML record stores the added identity
details separately from the unchanged data object.

Files

categories.yaml defines the five categories and the required fields for each type.
dataset.yaml contains all 50 records, PDF paths, page dimensions and page counts.
pdfs/ contains the individual documents, grouped by category.
text/ contains text extracted from every finished PDF.
validation.json records the automated checks against the original collection.
preview.jpg shows one example of every document type at a common physical scale.
invoice-designs.jpg compares the three invoice issuers at the same physical scale.
before-after.jpg compares the original and revised Cedar invoice.

Use and limitations

All people, organisations, addresses and reservations are invented for this
collection. City names are real. Contact domains are reserved example.com names.
Tax identifiers marked DEMO are deliberately invalid. There are no bank account
details, valid tickets or official identity documents. Fictional reading sources
are teaching examples, not published research.

Money is in EUR. Unit prices exclude tax. Tax rates are illustrative and do not
represent advice about local tax rules. Decimal monetary values remain strings
in the YAML. Dates use YYYY-MM-DD. All documents are in English.

The PDFs are clean digital documents with selectable text. They do not include
simulated wrinkles, camera distortion, scan noise or fake handwriting. Financial
documents and reservations retain small specimen notices. The originals remain
available separately in the first collection.

Validation

The checks compare every data object to the original, verify required fields,
check monetary values against extracted text, reconcile invoice arithmetic,
open and rasterize every PDF page, check text boundaries and content-block
collisions, and confirm dataset identifiers occur only in metadata, not page text.
All final pages were reviewed in rendered contact sheets. Rendering and text
extraction are not a substitute for an OCR test through the app.