v1.0.7 — Bugfix Release

v1.0.7 — Bugfix Release

16 bugs discovered and fixed across 8 commits during a full-project code review.

Critical (Data Corruption)

• Markdown escaping: Links, images, and catalog items now escape ], ), \ and other metacharacters
• Double-escaping regression FIXED: Link text no longer double-escaped (text nodes already escaped by tree walker)
• mergeField: Now prefers price-with-currency-symbol over bare numbers and longer duration over truncated values
• CJK word boundary: Chinese view count labels now correctly matched

High (Incorrect Output / Feature Breakage)

• PRICE_RE: European postfix currency symbols now matched
• Nested tables: Inner tr rows no longer leak into outer table output
• Code fences: Adaptive backtick fence length prevents premature termination
• Inline text escaping: Only semantically significant chars escaped (\ ` * _ [ ] ( ))
• Nested list indent: No longer compounds across levels
• Download listener: onChanged now removed on ALL exit paths
• readExpectedTotal: Now captures total count rather than page offset (fixes 'Showing 1-20 of 500 results' → 500 not 1)
• Scroll oscillation: scrollIntoView uses block:'start' to avoid center oscillation in accumulation loop

Medium (Quality / Robustness)

• Defuddle re-injection: 749KB bundle now skipped on repeated clicks
• betterTitle: Prefix match requires 55% overlap floor
• Data URL size check: Uses encodeURIComponent length
• _defuddleInjected guard: Auto-clears if >5000 entries
• collectAttrsPrice: closest scoped to parentElement to avoid self-matching

Full Changelog: ce0f12c..fc20814

ToMarkdown v1.0.6

ToMarkdown v1.0.6

Highlights

• Scroll-accumulate catalogs for virtualized product/media grids (full list across scroll windows).
• Prices, durations, and view counts with multi-currency and structural parsers (not site-specific).
• Safer view extraction: rejects title noise such as character codes (2B) and ages (18 year old); never treats duration seconds (13:22) as view counts; better values win when merging scroll passes.
• Fuller titles when one candidate is a truncated prefix of another.
• Full-page fallback still available for normal articles.

Upgrade path from v1.0.3

If you already use scroll-accumulate, reload the extension. Catalogs should include reliable $price · duration · views lines for most store grids.

Install

1. Download ToMarkdown-v1.0.6.zip and unzip
2. Chrome → chrome://extensions → Developer mode → Load unpacked (or Reload if already loaded from this repo)
3. Path when developing: /Volumes/WD_BLACK/Code/ToMarkdown (or your clone)

Files

• Extension package: ToMarkdown-v1.0.6.zip

ToMarkdown v1.0.3

ToMarkdown v1.0.3

Highlights

• Full-page extraction by default — keep store/grid content (titles, prices, links) instead of Defuddle “main article only” stripping.
• Scroll-accumulate catalogs for virtualized lists (e.g. ManyVids Store/Videos): auto-scroll, dedupe video cards, export a clean numbered list with title / price / duration / views / URL.
• Longer extract timeout (120s) for multi-pass scrolling.
• Metadata still uses Defuddle where available.

Install

1. Download ToMarkdown-v1.0.3.zip
2. Unzip
3. Chrome → chrome://extensions → Developer mode → Load unpacked → select the unzipped folder
4. Or reload if already loaded from this repo path

Files

• Extension package: ToMarkdown-v1.0.3.zip

ToMarkdown v1.0.1

ToMarkdown v1.0.1

Chrome extension that one-click saves a page's main content as AI-friendly Markdown (.md).

Highlights

• One-click save — click the toolbar icon to download
• Smart extraction — powered by Defuddle (same engine as Obsidian Web Clipper)
• Markdown output — headings, lists, code blocks, links preserved
• Local only — no remote upload of page content
• Brand logo — navy double-line M with download arrow, rounded app-icon corners

Install

1. Download ToMarkdown-v1.0.1.zip below and unzip
2. Open Chrome → chrome://extensions
3. Enable Developer mode
4. Load unpacked → select the unzipped folder
5. Pin the icon and click on any page

Notes

• Restricted pages (chrome://, Chrome Web Store, some system pages) cannot be injected
• For file:// pages, enable Allow access to file URLs on the extension details page

License

MIT — fully open source