[pull] dev from ethereum:dev

.claude/skills/intl-review/references/ethglossary-usage.md

@@ -19,14 +19,22 @@ This doc focuses on the review-specific patterns (severity mapping, what to flag
 
 This is the determinism backbone. Reviewers don't argue terminology with the pipeline — both pipeline and reviewer defer to ETHGlossary. If a glossary entry looks wrong during review, flag it in the report; don't patch the locale to disagree with it.
 
+## Authority hierarchy — and what to do for items the glossary doesn't cover
+
+ETHGlossary is the source of truth for **term translations AND for transliteration/calque/keep-Latin guidance**. Apply it in this strict order; do not substitute your own instinct:
+
+1. **Term IS in ETHGlossary** → its per-term `script_rule` is the *only* authority for transliterate / calque / keep-latin / always-latin. Query it (`/filter` per file, or `/translations/{lang}/{termId}`). A deviation is CRITICAL.
+2. **Term is NOT in ETHGlossary** (author names, brand-new products, etc.) → apply the script-aware fallback in `known-patterns.md` §1: **transliterate** into non-Latin target scripts, **keep as-is** for Latin scripts.
+3. **Never infer a "default" `script_rule` for an unlisted term.** An absent entry means "use the fallback," **not** "keep Latin." Flagging a correctly-transliterated non-Latin author name (e.g. `te` "మారియో హావెల్" for "Mario Havel") as "should be Latin" is a **false positive** — the kind of fabricated critical that wastes reviewer time. When in doubt, query the API; if the term isn't there, the fallback decides, not you.
+
 ## How to query
 
 **Preferred — per-file `/filter`:**
 
 ```bash
 curl -sf -X POST "$GLOSSARY_API_URL/filter" \
   -H "Content-Type: application/json" \
-  -d "$(jq -n --arg text "$ENGLISH_SOURCE" --arg lang "$LANG" '{text: $text, language: $lang}')"
+  -d "$(jq -n --arg content "$ENGLISH_SOURCE" --arg lang "$LANG" '{content: $content, language: $lang}')"
 ```
 
 Returns only the glossary terms that appear in the English source for the file being reviewed, with translations sorted by occurrence. Avoids pulling hundreds of irrelevant terms into review context.

.claude/translation-review/known-patterns.md

@@ -1,7 +1,7 @@
 # Known Translation Patterns & Issues
 
 > This is a living document. Updated after each language review.
-> Last updated: 2026-03-16 (updated with Gemini-confirmed transliteration policy from Hindi PR #17101)
+> Last updated: 2026-06-09 (PR #18375: added MDX duplicated-closer / dropped-`>` breakers, the duplicate ghost-heading migration artifact, and the ETHGlossary authority hierarchy for unlisted terms)
 
 ## Issue Categories
 
@@ -70,6 +70,12 @@ flag phonetic transliterations.
 
 **Transliteration authority:** ETHGlossary (https://ethglossary.visual-20-hoists.workers.dev) is the canonical source for term translations, including per-language transliterated forms for non-Latin scripts. The pipeline queries ETHGlossary directly; reviewers verify against the per-term `script_rule` returned by the API. The previous local bank at `.claude/translation-review/transliterations/` has been removed as of ETHGlossary v0.3.0.
 
+**Authority hierarchy — terms ETHGlossary covers vs. items it doesn't (READ THIS before flagging a transliteration/calque/keep-Latin "error"):**
+
+1. **For any term ETHGlossary covers, its per-term `script_rule` is the ONLY authority** for the transliterate / calque / keep-latin / always-latin decision. Query the API (`/filter` per file, or `/translations/{lang}/{termId}`); never assume.
+2. **For items ETHGlossary does NOT cover** (author names, brand-new product names not yet in the glossary, etc.), apply the script-aware fallback above: **transliterate** into non-Latin target scripts, **keep as-is** for Latin scripts.
+3. **Never infer a "default" `script_rule` for an unlisted term.** An absent glossary entry means "fall back to the script-aware policy," **NOT** "keep Latin." Example caught in PR #18375 review: a `te` author name "Mario Havel" rendered as "మారియో హావెల్" is **CORRECT** per the fallback — a reviewer flagging it as "should stay Latin" by assuming an `always_latin` default was a **false positive**. When ETHGlossary and a reviewer's instinct disagree, ETHGlossary (or, for unlisted items, this documented fallback) wins — there is one source of truth.
+
 ### 2. Cross-Script Contamination (CRITICAL)
 
 Crowdin translation memory leaks content from other language translations.
@@ -82,16 +88,28 @@ Crowdin translation memory leaks content from other language translations.
 
 ### 3. MDX Syntax Errors (CRITICAL — breaks builds)
 
-Four predictable categories that appear in every import:
+Predictable categories that appear in nearly every import:
 
 | Pattern | Example | Fix |
 |---------|---------|-----|
 | Raw `<` before numbers | `<5GB` in MDX context | Escape to `&lt;5GB` |
 | Missing closing backtick | `` `<contract>.<function>() `` | Add closing backtick |
 | Misplaced backtick exposing JSX | ``(`<> ...` </>`)`` | Fix backtick placement |
 | Orphaned HTML closing tags | `</a>` from sentence restructuring | Remove orphaned tag |
+| Duplicated inner closer over a wrapper | `<ExpandableCard>…<ButtonLink>x</ButtonLink></ButtonLink>` (2nd should close the wrapper) | Restore the wrapper's real closing tag from the English source (`</ExpandableCard>`, `</Callout>`, …) |
+| Dropped `>` in angle-bracket link whose URL has parens | `[t](<https://en.wikipedia.org/wiki/Electra_(star))` | Restore the `>` before the final `)`: `…_(star)>)` |
+
+**Pattern:** The first two are most common. The last two were the entire cause of the failing build in **PR #18375** (77 files, all 24 langs). **Detect deterministically** by compiling each changed file through `@mdx-js/mdx` (the parser `next-mdx-remote` uses) — strip frontmatter and `{#id}` heading anchors first (else every file false-positives on the custom heading-id syntax), and confirm the English sources compile clean as a control before trusting the run.
+
+### 3b. Duplicate "Ghost" Headings (CRITICAL — structural-migration artifact)
+
+When a base-branch change shifts page block structure — e.g. the h1 → `frontmatter.title` migration that removed leading `#` headings — the pipeline's incremental block-matching can mis-align and emit a section **twice**: an anchor-less "ghost" copy (often an older or differently-worded translation, sometimes a different formality register) immediately followed by the correct `{#anchor}` copy. The reader sees the section rendered twice in a row.
+
+**Signature:** a translated `h2`–`h4` heading **without** `{#id}` immediately followed (after blank lines and/or one duplicate paragraph) by a **same-level** heading **with** `{#id}`. English requires `{#id}` on every heading, so any anchor-less translated heading is the tell.
+
+**Fix:** delete the ghost block (the anchor-less heading + its duplicate paragraph) up to the anchored twin; keep the anchored version that matches the English source. Observed in **PR #18375** at **254 occurrences across 69 files** (24 langs × `community/grants`, `contributing/adding-videos`, `roadmap/glamsterdam`).
 
-**Pattern:** These same 4 patterns recur in every Crowdin import. The first two are most common.
+**Detection:** scan changed translated files for `^#{2,4} ` lines lacking `{#` (outside code fences); classify each as ghost-twin (next same-level heading is anchored → safe to delete) vs. lone-missing-anchor (needs the anchor *added* from English) before fixing. **The durable fix belongs in the sanitizer** (collapse adjacent duplicate headings during sanitization) so future structural changes self-heal rather than shipping duplicates.
 
 ### 4. Semantic Inversions (CRITICAL)
 

.claude/translation-review/localization-rules-by-language-group.md

@@ -2,7 +2,7 @@
 
 > **Source:** Gemini (2026-03-18), confirmed by project maintainer
 > **Applies to:** All 13 non-Latin-script locales in the transliteration pipeline
-> **Used by:** sanitizer (post_import_sanitize.ts), transliteration script (transliterate.ts), review agents
+> **Used by:** sanitizer (`src/scripts/intl-pipeline/intl-sanitizer.ts`), ETHGlossary integration, review agents
 
 ## Quick Reference Table
 
@@ -141,7 +141,7 @@ acts as a strong RTL character that anchors the directionality:
 
 ## Impact on Sanitizer and Transliteration Scripts
 
-### Sanitizer (post_import_sanitize.ts)
+### Sanitizer (`src/scripts/intl-pipeline/intl-sanitizer.ts`)
 
 Current behavior and needed changes:
 

.manifests/src/intl/ar/common.json/source.json

@@ -1,11 +1,11 @@
 {
   "version": 1,
   "sourceFile": "src/intl/en/common.json",
-  "generatedAt": "2026-06-02T09:58:27.177Z",
-  "rootHash": "f6744960f9ba",
+  "generatedAt": "2026-06-08T11:40:15.805Z",
+  "rootHash": "8859281c9894",
   "tree": {
-    "contentHash": "d0599bca3752",
-    "anchorHash": "9073bb8b8520",
+    "contentHash": "37cbcc725ee7",
+    "anchorHash": "04ad04fb0db4",
     "children": {
       "about-ethereum-org": {
         "contentHash": "9aafb3686977",
@@ -174,6 +174,42 @@
         "contentHash": "e21f935f11d7",
         "anchorHash": "e3b0c44298fc"
       },
+      "copy-page": {
+        "contentHash": "62d9b9487f59",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-aria-label": {
+        "contentHash": "c1188f922b2d",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-options": {
+        "contentHash": "66c1ae393307",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-markdown-title": {
+        "contentHash": "17d36a505f97",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-markdown-description": {
+        "contentHash": "28cded2269c4",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-chatgpt-title": {
+        "contentHash": "c27a47a5c24f",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-chatgpt-description": {
+        "contentHash": "babab9a22ef4",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-claude-title": {
+        "contentHash": "9512e37d4a15",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-claude-description": {
+        "contentHash": "babab9a22ef4",
+        "anchorHash": "e3b0c44298fc"
+      },
       "danksharding": {
         "contentHash": "d5c7c6294c3d",
         "anchorHash": "e3b0c44298fc"
@@ -2241,6 +2277,15 @@
       "cookie-policy",
       "copied",
       "copy",
+      "copy-page",
+      "copy-page-aria-label",
+      "copy-page-options",
+      "copy-page-markdown-title",
+      "copy-page-markdown-description",
+      "copy-page-chatgpt-title",
+      "copy-page-chatgpt-description",
+      "copy-page-claude-title",
+      "copy-page-claude-description",
       "danksharding",
       "dao-page",
       "dark-mode",
@@ -2739,5 +2784,5 @@
       "region-sint-eustatius-netherlands"
     ]
   },
-  "sourceCommitSha": "9c8ec781d111c9802b6d9677f581d21a46bedcb4"
+  "sourceCommitSha": "e7089af701ba3acee7872235b2cc7706c10efc91"
 }

.manifests/src/intl/bn/common.json/source.json

@@ -1,11 +1,11 @@
 {
   "version": 1,
   "sourceFile": "src/intl/en/common.json",
-  "generatedAt": "2026-06-02T09:58:58.348Z",
-  "rootHash": "f6744960f9ba",
+  "generatedAt": "2026-06-08T11:40:42.465Z",
+  "rootHash": "8859281c9894",
   "tree": {
-    "contentHash": "d0599bca3752",
-    "anchorHash": "9073bb8b8520",
+    "contentHash": "37cbcc725ee7",
+    "anchorHash": "04ad04fb0db4",
     "children": {
       "about-ethereum-org": {
         "contentHash": "9aafb3686977",
@@ -174,6 +174,42 @@
         "contentHash": "e21f935f11d7",
         "anchorHash": "e3b0c44298fc"
       },
+      "copy-page": {
+        "contentHash": "62d9b9487f59",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-aria-label": {
+        "contentHash": "c1188f922b2d",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-options": {
+        "contentHash": "66c1ae393307",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-markdown-title": {
+        "contentHash": "17d36a505f97",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-markdown-description": {
+        "contentHash": "28cded2269c4",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-chatgpt-title": {
+        "contentHash": "c27a47a5c24f",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-chatgpt-description": {
+        "contentHash": "babab9a22ef4",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-claude-title": {
+        "contentHash": "9512e37d4a15",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-claude-description": {
+        "contentHash": "babab9a22ef4",
+        "anchorHash": "e3b0c44298fc"
+      },
       "danksharding": {
         "contentHash": "d5c7c6294c3d",
         "anchorHash": "e3b0c44298fc"
@@ -2241,6 +2277,15 @@
       "cookie-policy",
       "copied",
       "copy",
+      "copy-page",
+      "copy-page-aria-label",
+      "copy-page-options",
+      "copy-page-markdown-title",
+      "copy-page-markdown-description",
+      "copy-page-chatgpt-title",
+      "copy-page-chatgpt-description",
+      "copy-page-claude-title",
+      "copy-page-claude-description",
       "danksharding",
       "dao-page",
       "dark-mode",
@@ -2739,5 +2784,5 @@
       "region-sint-eustatius-netherlands"
     ]
   },
-  "sourceCommitSha": "9c8ec781d111c9802b6d9677f581d21a46bedcb4"
+  "sourceCommitSha": "e7089af701ba3acee7872235b2cc7706c10efc91"
 }

.manifests/src/intl/cs/common.json/source.json

@@ -1,11 +1,11 @@
 {
   "version": 1,
   "sourceFile": "src/intl/en/common.json",
-  "generatedAt": "2026-06-02T09:58:20.388Z",
-  "rootHash": "f6744960f9ba",
+  "generatedAt": "2026-06-08T11:40:07.313Z",
+  "rootHash": "8859281c9894",
   "tree": {
-    "contentHash": "d0599bca3752",
-    "anchorHash": "9073bb8b8520",
+    "contentHash": "37cbcc725ee7",
+    "anchorHash": "04ad04fb0db4",
     "children": {
       "about-ethereum-org": {
         "contentHash": "9aafb3686977",
@@ -174,6 +174,42 @@
         "contentHash": "e21f935f11d7",
         "anchorHash": "e3b0c44298fc"
       },
+      "copy-page": {
+        "contentHash": "62d9b9487f59",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-aria-label": {
+        "contentHash": "c1188f922b2d",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-options": {
+        "contentHash": "66c1ae393307",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-markdown-title": {
+        "contentHash": "17d36a505f97",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-markdown-description": {
+        "contentHash": "28cded2269c4",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-chatgpt-title": {
+        "contentHash": "c27a47a5c24f",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-chatgpt-description": {
+        "contentHash": "babab9a22ef4",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-claude-title": {
+        "contentHash": "9512e37d4a15",
+        "anchorHash": "e3b0c44298fc"
+      },
+      "copy-page-claude-description": {
+        "contentHash": "babab9a22ef4",
+        "anchorHash": "e3b0c44298fc"
+      },
       "danksharding": {
         "contentHash": "d5c7c6294c3d",
         "anchorHash": "e3b0c44298fc"
@@ -2241,6 +2277,15 @@
       "cookie-policy",
       "copied",
       "copy",
+      "copy-page",
+      "copy-page-aria-label",
+      "copy-page-options",
+      "copy-page-markdown-title",
+      "copy-page-markdown-description",
+      "copy-page-chatgpt-title",
+      "copy-page-chatgpt-description",
+      "copy-page-claude-title",
+      "copy-page-claude-description",
       "danksharding",
       "dao-page",
       "dark-mode",
@@ -2739,5 +2784,5 @@
       "region-sint-eustatius-netherlands"
     ]
   },
-  "sourceCommitSha": "9c8ec781d111c9802b6d9677f581d21a46bedcb4"
+  "sourceCommitSha": "e7089af701ba3acee7872235b2cc7706c10efc91"
 }