PDF Text Preflight Lab: Python Example
Explore six artificial PDF cases before deciding what needs OCR. The browser viewer shows recorded text/image observations and the exact Python classifier. It does not parse user PDFs or execute Python in the browser.
Free source download: preflight.py, six-page-fixture.pdf, artificial images, expected results, requirements.txt and MIT license. Tested on Windows with Python 3.12.14 and pypdf 6.10.0; an isolated replay with only pypdf and the standard library reproduced all six routes.
The mixed-review rule deliberately flags a decorative logo. Blank and vector-only pages share a review route. These are six artificial pages, not production accuracy, field-extraction or speed benchmarks. No OCR engine was run.
Need a tutorial checked? Briefcraft Lab offers a $35 pilot: one Python tutorial up to 1,000 words and one entry-point script, annotated findings, corrected instructions where feasible, a command transcript, results and one in-scope correction. Feasibility, payment terms and delivery date are agreed first.
The complete three-page writing sample is available separately on request.
Contact: briefcraftlab26@proton.me
Additional PDF-generation example
The separate invoice-pdf-public-example-oct17.zip demonstrates a synthetic CSV-to-invoice workflow. It includes Python source, 15 regression tests, a sample CSV, a rendered PDF/PNG and pinned dependencies. The example total is USD 174.00, with a manually entered paid amount of USD 50.00 and balance of USD 124.00. It uses integer cents, validates input, preserves existing output files and supports repeated headers for longer invoices.
This download is an inspection/evaluation example with rights reserved; it is separate from the MIT-licensed preflight sample above. It uses ASCII descriptions and USD only. It has no live payment verification, student database, tax logic, recurring schedule or reminders. The invoice is synthetic and does not request payment. For a custom PDF or CSV workflow, email a short description, an artificial input example and the required output; scope, price and timing are agreed before work.
Download
Install instructions
Free CLI sample: download pdf-preflight-cli-sample-oct15.zip and extract it to a folder.
Use Python 3.12. In a terminal opened in that extracted folder, run:
python -m pip install -r requirements.txt
python preflight.py six-page-fixture.pdf
Compare the printed per-page observations with results.json. The pinned dependency is pypdf 6.10.0. Tested on Windows with Python 3.12.14.
The on-page viewer needs no installation. Click Run tool, choose an artificial case and scroll inside the viewer, or use its fullscreen button. It displays stored results; it does not upload or parse your PDFs, execute Python, or perform OCR.
The six artificial pages are a small reproducible example, not a production accuracy or speed benchmark. PDF loading and text-extraction errors propagate to the caller; image-enumeration failures return parse_review. MIT license included.
Optional invoice example: extract invoice-pdf-public-example-oct17.zip separately. On Python 3.12, run python -m pip install -r requirements.txt, then python invoice_preview.py synthetic-items.csv new-invoice.pdf --paid 50.00. Existing output paths are rejected. Run python -m unittest discover -p "test_*.py" -v to replay 15 tests. See its README for limits and evaluation rights. The invoice example is not run by the browser viewer.
Leave a comment
Log in with itch.io to leave a comment.