precheck

Precheck: one-stop quality gate

Invocation: /precheck $1 where $1 is a page slug (e.g. technical-diving) or all. Optional --langs (default ru,de).

Run ALL checks, COLLECT every failure, then report a single pass/fail summary. Do NOT stop at the first failure — the goal is a complete picture for the caller (content-loop, design-system, safe-deploy) to fix in one pass.

Languages in scope: en (source) + ru, de.

Checks (run every one, capture output)

1. Encoding guard (MUST be zero). Scan for corruption introduced by bad deploys.

   • Literal ??: Select-String -Path output\*\*.html,themes\gotodive-com\**\*.php,themes\gotodive-com\**\*.html -Pattern '\?\?'
   • Replacement char U+FFFD (�): same scan with pattern \uFFFD.
   • ZERO matches required. This catches the · -> ?? class of bug.

2. Translation structure parity. python scripts/_check-translation-sync.py

   • Compares EN vs ru/de tag skeleton + text-node count. Tolerates abs(diff)<=1.
   • Real drift = tag mismatch OR node delta >1.

3. Duplicated / echoing words (advisory). python scripts/_check-dupe-words.py --langs ru,de

   • Flags consecutive words sharing a >=4-char stem in different forms (the double-translation signature). Expect some false positives; review, don't auto-strip.

4. Sentence fragmentation. python scripts/_check-fragmentation.py --langs ru,de

   • Flags lowercase+period+space+lowercase inside one text node (broken wrap).

5. Broken internal links. node scripts/_check-broken-links.js

6. Sitemap health (when deploying live). python scripts/_check_sitemap.py

   • Every sitemap URL must be 200 AND same-language.

Output contract

Print, per check: PASS/FAIL + the specific failing files/lines. End with a single line: PRECHECK: PASS (all green) or PRECHECK: FAIL (<n> issues).

Remedies (hand off to the caller)

• Encoding ??/mojibake -> re-author the real UTF-8 glyph; deploy ONLY via binary scp; prefer HTML entities (·, →) in theme files.
• Structure drift -> fix EN source, regenerate that slug (do not hand-hack locale tags).
• Double word / fragmentation -> a mid-phrase newline in EN; merge onto one line and regenerate (python scripts/generate_full_page_translations.py --langs ru,de --slugs <slug>).
• Spec-key garbled/Latin -> python scripts/_apply-spec-keys.py --langs ru,de.
• Broken link / sitemap -> fix the source link or slug, redeploy, re-check.