fe59fa9785
- Add #page=N URL fragment routing for shareable inspector links - Support browser back/forward navigation via hashchange event - Persist overlay toggle state in localStorage with error handling - Add isUpdatingFragment flag to prevent double-render on hash updates - Update thumbnail click handler to rely on updateFragment() - Clamp out-of-range page numbers with console warnings - Default to page 0 for invalid/non-numeric page numbers - Add vector fixture provenance entries Acceptance criteria: - URL #page=14 on load → starts on page 14 ✓ - Navigate via next button → URL updates to #page=15 ✓ - Browser back button → URL and view update correctly ✓ - Bookmark with #page=14 → reopens to page 14 ✓ - Overlay toggles persist across page refresh ✓ - Out-of-range #page=999 → clamps to last page ✓ - Invalid #page=abc → defaults to page 0 ✓ Closes pdftract-47e42 Verification: notes/pdftract-47e42.md
27 lines
723 B
Markdown
27 lines
723 B
Markdown
# Multi-Page Academic Paper - CER Test Fixture
|
|
|
|
## Purpose
|
|
This fixture is used for Character Error Rate (CER) testing in the vector PDF corpus.
|
|
|
|
## Files
|
|
- `source.pdf` - Clean vector PDF with embedded text
|
|
- `ground_truth.txt` - Exact text content for CER comparison
|
|
- `README.md` - This file
|
|
|
|
## Content
|
|
Abstract
|
|
This paper presents a comprehensive study of distributed systems.
|
|
Page 1 of 3
|
|
|
|
Introduction
|
|
Distributed systems form the backbone of modern cloud computing.
|
|
We explore consistency models and ...
|
|
|
|
## Expected CER
|
|
Target: < 0.5% character error rate when extracted by pdftract.
|
|
|
|
## Metadata
|
|
- Title: Multi-Page Academic Paper
|
|
- Author: Dr. Smith
|
|
- Creator: LaTeX
|
|
- Generated by: generate_vector_cer_corpus.py
|