Data: 33 Special Procedures reports to the 81st General Assembly

.zenodo.json

@@ -3,7 +3,7 @@
   "upload_type": "software",
   "access_right": "open",
   "license": "PolyForm-Noncommercial-1.0.0",
-  "description": "Paragraph-level research environment for UN human rights law. UNHRDB combines Treaty Body General Comments and Recommendations, individual-communication jurisprudence, and thematic Special Procedures reports in a source-first interface with faceted search, citation export, document workspaces, and extractive retrieval. The software and curated datasets have separate licences; see LICENSE, LICENSE-DATA, and NOTICE in the repository. The v2.1.0 release (3 September 2026) covers 6,111 documents and 323,065 paragraphs: 187 General Comments and Recommendations, 4,333 individual-communication decisions from 8 treaty bodies, and 1,591 thematic Special Procedures reports from all 46 mandates. The repository holds the published site and its front-end code; the data-processing pipeline is kept in a separate, private repository.",
+  "description": "Paragraph-level research environment for UN human rights law. UNHRDB combines Treaty Body General Comments and Recommendations, individual-communication jurisprudence, and thematic Special Procedures reports in a source-first interface with faceted search, citation export, document workspaces, and extractive retrieval. The software and curated datasets have separate licences; see LICENSE, LICENSE-DATA, and NOTICE in the repository. The v2.1.0 release (3 September 2026) covers 6,152 documents and 326,457 paragraphs: 188 General Comments and Recommendations, 4,333 individual-communication decisions from 8 treaty bodies, and 1,631 thematic Special Procedures reports from all 46 mandates. The repository holds the published site and its front-end code; the data-processing pipeline is kept in a separate, private repository.",
   "creators": [
     {
       "name": "Szoszkiewicz, Łukasz",

README.md

@@ -21,14 +21,14 @@ original document symbol and context.
 
 ## Coverage
 
-The September 2026 public build contains 323,065 searchable paragraphs across 6,111
+The October 2026 public build contains 326,457 searchable paragraphs across 6,152
 documents:
 
 | Collection | Coverage |
 |---|---:|
-| Treaty Body General Comments and Recommendations | 187 documents |
+| Treaty Body General Comments and Recommendations | 188 documents |
 | Individual-communication jurisprudence | 4,333 cases from 8 mechanisms |
-| Thematic Special Procedures reports | 1,591 reports from 46 mandates |
+| Thematic Special Procedures reports | 1,631 reports from 46 mandates |
 | Recommendations to States (UHRI, live) | 267,942 records from 70 mechanisms to 199 States, 2006–2026 |
 
 The Recommendations scope is not part of the static corpus: concluding

build_static_pages.py

@@ -125,6 +125,10 @@ def provenance(d) -> str:
     text does not come from the Word file on UN Documents."""
     if d.get("type") != "sp" or d.get("textSource") == "documents.un.org docx":
         return ""
+    if d.get("textSource") == "documents.un.org pdf (docx pending)":
+        return ('<aside class="note warn">A recent report: UN Documents has not yet published its Word file, so the text '
+                'comes from the PDF. Footnotes and headings may be incomplete until it is rebuilt from the Word file; '
+                'check quotations against the official text.</aside>')
     if d.get("textSource") == "documents.un.org pdf (layout)":
         msg = ("From PDF: no Word file exists for this report, so its text was re-cut from the PDF layout. "
                "In such reports about one paragraph in nine may still show a slip, often OCR in older scans.")

docs/assets/app.js

@@ -1293,12 +1293,20 @@ function documentStatusDetails(doc) {
 }
 
 // Special Procedures reports whose text does not come from the Word file on
-// UN Documents: 60 re-cut from the PDF layout, 21 still as first extracted.
+// UN Documents: re-cut from the PDF layout, still as first extracted, or recent
+// reports whose Word file is not out yet (rebuilt once it is).
 // The shares are from the automated integrity check (September 2026).
 function spTextBannerHtml(doc, sourceUrl) {
   if (doc?.type !== 'sp' || doc.textSource === 'documents.un.org docx') return '';
-  const recut = doc.textSource === 'documents.un.org pdf (layout)';
   const pdf = sourceUrl ? `<a href="${escape(sourceUrl)}" target="_blank" rel="noopener">original PDF</a>` : 'original PDF';
+  if (doc.textSource === 'documents.un.org pdf (docx pending)') {
+    return `<aside class="docs-reader-ocr-banner ocr-banner-cleaned sp-text-banner" role="note">
+      <span class="ocr-banner-tag">FROM PDF</span>
+      <div class="ocr-banner-body"><strong>A recent report: UN Documents has not yet published its Word file, so the text comes from the PDF.</strong>
+        Footnotes and headings may be incomplete until it is rebuilt from the Word file; check quotations against the ${pdf}.</div>
+    </aside>`;
+  }
+  const recut = doc.textSource === 'documents.un.org pdf (layout)';
   return `<aside class="docs-reader-ocr-banner ocr-banner-${recut ? 'cleaned' : 'raw'} sp-text-banner" role="note">
       <span class="ocr-banner-tag">${recut ? 'FROM PDF' : 'PDF · UNREVIEWED'}</span>
       <div class="ocr-banner-body">${recut