v1.0.4 — Correct RTL for Hebrew lines opening with English

What's fixed

🔤 Hebrew lines that open with an English word now lay out RTL correctly

This was the core-mission bug: dir="auto" picks a line's direction from its first strongly-directional character, so a Hebrew sentence that happens to start with an English term — CGNAT הורג P2P בסלולרי, Floor control צריך שופט אחד — was resolved as LTR and came out scrambled, exactly the mess this app exists to prevent.

Direction is now computed from the line's content: a line containing any Hebrew (or Arabic) character anywhere is RTL; only lines with none at all stay LTR. This applies to paragraphs, headings, list items, blockquotes, and each table cell individually — and to raw (unstyled) text mode too, which previously relied on the same first-character heuristic via unicode-bidi: plaintext.

Text alignment now follows each line's direction as well: Hebrew lines sit on the right, English-only lines on the left, instead of everything being force-aligned right.

Covered by new tests for exactly these cases (mixed lines, headings, mixed table cells) — 62 tests passing.

Install

Run copy-to-rtl-setup-1.0.4.exe — or download copy-to-rtl-win32-x64.zip, extract, and run copy-to-rtl.exe if you prefer no installer. See v1.0.3 for the recent feature additions (paste button, per-code-block copy, toolbar menu).