fix(pdf): fix page breaks and element scaling for PDF export (fixes #166)

[ThisIs-Developer]

Description

This PR resolves the issue where text, lists, table rows, and MathJax formulas are cut or sliced horizontally in half across A4 page boundaries during PDF export (Fixes #166).

Key Improvements

1. Expanded Page-Break Detection

• Updated identifyGraphicElements to include page-break sensitive text elements (p, li, blockquote, h1-h6, hr) and MathJax display containers (mjx-container[display="true"], .math-block). These are labeled as 'text' or 'math' to prevent boundary slicing.

2. Dynamic Shift Offset Tracking

• Implemented an accumulatedShift tracking parameter inside insertPageBreaks. As element margins are shifted down to begin at a new page boundary, subsequent element positions are dynamically offset to prevent coordinate overlap calculations from using stale values.

3. Table Splitting & Header Replication

• Added helper functions splitTables and mergeSplitTables which divide tables exceeding a page's height into separate parts.
• Clones and repeats the original thead headers on subsequent pages for continuous reading.
• Inserts precise spacer elements to shift subsequent table rows to the top of the next page instead of letting them cross A4 boundaries.

4. Cascade Stabilization

• Moved the call to handleOversizedElements inside the main applyPageBreaksWithCascade stabilization loop. Margins and scales are now resolved iteratively to guarantee layout stability.

5. Height Slicing & Box Sizing Fixes

• Replaced tempElement.scrollHeight with Math.ceil(tempElement.getBoundingClientRect().height) in html2canvas window properties to handle transformed dimension heights correctly, eliminating empty trailing pages.
• Corrected double-counting of padding in fitExportElementToContent and fitMarkdownExportToContent.
• Cleaned up top-level YAML frontmatter parsing regex (/^\s*---/) so that newlines/whitespace preceding frontmatter do not break parsing.

---

Verification Results

Verified the changes locally using Playwright:

• Lazy loads PDF libraries (jspdf and html2canvas) dynamically.
• Pushes and offsets layout elements cleanly at page boundaries.
• Compiles the PDF and saves the document successfully.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
markdown-viwer	Ready	Preview, Comment	Jun 10, 2026 5:59am

[cloudflare-workers-and-pages]

Deploying markdown-viewer with    Cloudflare Pages

Latest commit:	70915f6
Status:	✅  Deploy successful!
Preview URL:	https://3fc26e46.markdown-viewer.pages.dev
Branch Preview URL:	https://codex-fix-pdf-page-breaks-16.markdown-viewer.pages.dev

View logs

[cloudflare-workers-and-pages]

Deploying markdownviewer with    Cloudflare Pages

Latest commit:	70915f6
Status:	✅  Deploy successful!
Preview URL:	https://cb2bfc07.markdownviewer.pages.dev
Branch Preview URL:	https://codex-fix-pdf-page-breaks-16.markdownviewer.pages.dev

View logs