fix: address review 4139493870 — remove XXX placeholders + support custom elements in leakage scanner

- cross-reference-map.md: drop "Stage 1: Government Proposition" subgraph
  containing `Prop 2025/26:XXX` placeholders (real IDs are not reliably
  linked in the betänkande payloads). Added an explanatory note; pipeline
  now starts at the committee-report stage. PR description's
  "0 AI_MUST_REPLACE markers" claim is now accurate.

- scripts/detect-swedish-leakage.ts: `parseTagAt` tag-name regex now
  accepts `-` and `:` so custom/hyphenated elements such as
  `<my-component lang="sv">` are correctly recognised and depth-tracked
  by `stripLangTaggedBlocks`.

- Added regression test for custom-element suppression. 39/39 leakage
  tests pass, 35/35 analysis-quality tests pass.

Agent-Logs-Url: https://github.com/Hack23/riksdagsmonitor/sessions/5f3e3f35-8c48-4e67-9f48-2b63c5ef592d

Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>

Author: Copilot

analysis/daily/2026-04-20/committeeReports/cross-reference-map.md

@@ -100,45 +100,29 @@ mindmap
 ```mermaid
 graph TB
     subgraph "📋 Legislative Pipeline — Committee Reports 2026-04-20"
-        subgraph "Stage 1: Government Proposition"
-            P1["Prop 2025/26:XXX<br/>(KU33 underlying prop)"]
-            P2["Prop 2025/26:XXX<br/>(KU32 underlying prop)"]
-            P3["Prop 2025/26:XXX<br/>(CU27 underlying prop)"]
-            P4["Prop 2025/26:XXX<br/>(CU28 underlying prop)"]
-            P5["Prop 2025/26:XXX<br/>(CU22 underlying prop)"]
-            P6["Riksrevisionen report<br/>(CU42 source)"]
-        end
-        
-        subgraph "Stage 2: Committee Report (Current)"
+        subgraph "Stage 1: Committee Report (Current)"
             B1["HD01KU33 Betänkande<br/>Vilande first reading ✅"]
             B2["HD01KU32 Betänkande<br/>Vilande first reading ✅"]
             B3["HD01CU27 Betänkande<br/>Adopted ✅"]
             B4["HD01CU28 Betänkande<br/>Adopted ✅"]
             B5["HD01CU22 Betänkande<br/>Adopted ✅"]
             B6["HD01CU42 Betänkande<br/>SOU referral ✅"]
         end
-        
-        subgraph "Stage 3: Chamber Decision"
+
+        subgraph "Stage 2: Chamber Decision"
             V1{{"🗳️ 2026 Election<br/>Sept 14"}}
             V2["Second reading<br/>(if Tidö wins)"]
             E1["Effective July 1, 2026"]
             E2["Registry 2029-2030"]
             E3["Authority Q1-Q2 2027"]
             S1["SOU investigation"]
         end
-        
-        P1 --> B1
-        P2 --> B2
-        P3 --> B3
-        P4 --> B4
-        P5 --> B5
-        P6 --> B6
-        
+
         B1 --> V1
         B2 --> V1
         V1 -->|"Coalition wins"| V2
         V1 -->|"S-bloc wins"| X["❌ Blocked"]
-        
+
         B3 --> E1
         B4 --> E2
         B5 --> E3
@@ -154,6 +138,8 @@ graph TB
     style S1 fill:#9E9E9E,color:#FFFFFF
 ```
 
+> **Note on proposition-stage nodes:** The underlying government propositions that preceded these betänkanden are not included as separate nodes because the specific proposition IDs (`Prop 2025/26:NN`) are not reliably linked from the betänkande payloads retrieved via `get_dokument_innehall`. Upstream proposition provenance is captured in each per-document analysis where the source was identified, and in the `data-download-manifest.md` MCP call trail.
+
 ---
 
 ## 🗳️ Political Alignment Matrix

scripts/detect-swedish-leakage.ts

@@ -255,7 +255,7 @@ function parseTagAt(html: string, startIndex: number): ParsedTag | null {
 
   const isClosing = inner.startsWith('/');
   const nameSource = isClosing ? inner.slice(1).trimStart() : inner;
-  const nameMatch = /^([a-zA-Z][a-zA-Z0-9]*)/.exec(nameSource);
+  const nameMatch = /^([a-zA-Z][a-zA-Z0-9:-]*)/.exec(nameSource);
   const tagName = nameMatch ? nameMatch[1].toLowerCase() : null;
   const isSelfClosing = !isClosing && /\/\s*>$/.test(rawTag);
 

tests/detect-swedish-leakage.test.ts

@@ -337,5 +337,16 @@ describe('Swedish Leakage Detector', () => {
       expect(terms).toContain('utskottet');
       expect(terms).toContain('propositionen');
     });
+
+    it('should strip custom-element (hyphenated) tags with lang="sv"', () => {
+      // Custom elements have hyphenated tag names (e.g. <my-quote>).
+      const html =
+        '<p>Article intro.</p>' +
+        '<my-quote lang="sv">riksdagen antog propositionen och utskottet</my-quote>' +
+        '<p>Article continues.</p>';
+      const report = detectSwedishLeakage(html, 'en');
+      expect(report.leakedTerms).toEqual([]);
+      expect(report.score).toBe(0);
+    });
   });
 });