@adeu/core 1.26.0 → 1.28.0

package/src/sanitize/report.ts

@@ -31,20 +31,22 @@ export class SanitizeReport {
       const lower = line.toLowerCase();
       if (lower.includes("tracked change") || lower.includes("insertion") || lower.includes("deletion") || lower.includes("accepted")) {
         this.change_lines.push(line);
-      } else if (lower.includes("comment") || lower.includes("[open]") || lower.includes("[resolved]")) {
-        if (lower.includes("kept") || lower.includes("visible")) {
-          this.kept_comment_lines.push(line);
-        } else {
-          this.removed_comment_lines.push(line);
-        }
       } else if (
         lower.includes("author") || lower.includes("template") || lower.includes("company") ||
         lower.includes("manager") || lower.includes("metadata") || lower.includes("timestamp") ||
         lower.includes("custom xml") || lower.includes("custom propert") || lower.includes("identifier") ||
+        lower.includes("document variable") ||
         lower.includes("language") || lower.includes("version") ||
-        lower.includes("last modified by") || lower.includes("revision count") || lower.includes("last printed")
+        lower.includes("last modified by") || lower.includes("revision count") || lower.includes("last printed") ||
+        lower.includes("description/comments")
       ) {
         this.metadata_lines.push(line);
+      } else if (lower.includes("comment") || lower.includes("[open]") || lower.includes("[resolved]")) {
+        if (lower.includes("kept") || lower.includes("visible")) {
+          this.kept_comment_lines.push(line);
+        } else {
+          this.removed_comment_lines.push(line);
+        }
       } else if (lower.includes("hyperlink") || lower.includes("warning")) {
         this.warnings.push(line);
       } else {

package/src/sanitize/transforms.ts

@@ -634,6 +634,45 @@ export function strip_custom_properties(doc: DocumentObject): string[] {
   return [`Custom document properties: ${count} removed (docProps/custom.xml)`].concat(lines);
 }
 
+/**
+ * Remove Word document variables (<w:docVars>/<w:docVar>) from
+ * word/settings.xml. Document variables are invisible in Word's UI but are a
+ * standard carrier for matter references, DMS identifiers, template state and
+ * integration tokens — a package that keeps them can never be reported CLEAN
+ * (QA 2026-07-19 ADEU-QA-001). Variables are disclosed by NAME only: their
+ * values are exactly the secrets a sanitize report must not echo.
+ */
+export function strip_document_variables(doc: DocumentObject): string[] {
+  const settingsPart = doc.pkg.getPartByPath('word/settings.xml');
+  if (!settingsPart) return [];
+
+  const names: string[] = [];
+  let removedAny = false;
+
+  for (const container of findDescendantsByLocalName(settingsPart._element, 'docVars')) {
+    for (const child of Array.from(container.childNodes)) {
+      const el = child as Element;
+      const tag = el.tagName || '';
+      if (tag === 'docVar' || tag.endsWith(':docVar')) {
+        names.push(el.getAttribute('w:name') || el.getAttribute('name') || '(unnamed)');
+      }
+    }
+    container.parentNode?.removeChild(container);
+    removedAny = true;
+  }
+
+  // Defensive: stray <w:docVar> elements outside a <w:docVars> container.
+  for (const stray of findDescendantsByLocalName(settingsPart._element, 'docVar')) {
+    names.push(stray.getAttribute('w:name') || stray.getAttribute('name') || '(unnamed)');
+    stray.parentNode?.removeChild(stray);
+    removedAny = true;
+  }
+
+  if (!removedAny) return [];
+  const shown = Array.from(new Set(names)).sort().join(', ') || '(unnamed)';
+  return [`Document variables: ${names.length} removed (${shown})`];
+}
+
 export function strip_image_alt_text(doc: DocumentObject): string[] {
   let count = 0;
   for (const docPr of findDescendantsByLocalName(doc.element, 'docPr')) {

package/src/utils/docx.ts

@@ -468,6 +468,19 @@ export function get_paragraph_prefix(
   }
 
   if (!style_name || style_name === "Normal") {
+    let is_inside_tc = false;
+    let curr: Element | null = paragraph._element;
+    while (curr) {
+      if (curr.tagName === "w:tc") {
+        is_inside_tc = true;
+        break;
+      }
+      curr = curr.parentNode as Element | null;
+    }
+    if (is_inside_tc) {
+      return "";
+    }
+
     const text = paragraph.text.trim();
     if (text && text.length < 100 && text === text.toUpperCase()) {
       let is_bold = false;
@@ -581,6 +594,80 @@ export function split_boundary_whitespace(
   ];
 }
 
+/**
+ * Maps each tracked-change id in a merged meta bubble to the OTHER ids of its
+ * resolution group, rendered ready for the bubble line suffix
+ * (`uid -> "Chg:2"` / `uid -> "Chg:2, Chg:3"`).
+ *
+ * A replacement is stored as a contiguous same-author <w:del> + <w:ins> pair
+ * with two distinct w:id values, but both engines resolve such a group as ONE
+ * unit — accepting or rejecting either side decides the whole replacement.
+ * Projecting the two ids side by side with no linkage implied they were
+ * independently resolvable (QA 2026-07-19 ADEU-QA-004); every meta-block
+ * builder (Python/Node ingest + mapper) uses this map to annotate grouped
+ * lines as `(pairs with Chg:N)`.
+ *
+ * Grouping mirrors the engines' `_get_paired_nodes` walk: consecutive ins/del
+ * ids in bubble order group while the author stays the same; any state
+ * carrying NO active ins/del (a comment- or format-only run between changes —
+ * physical text separating the elements) breaks the group. `states_list`
+ * entries are [ins_map, del_map, comments_set, fmt_map] snapshots in document
+ * order, as accumulated by the ingest/mapper state machines.
+ */
+export function compute_change_pair_map(
+  states_list: any[],
+): Record<string, string> {
+  const groups: Array<Array<[string, string]>> = [];
+  let current: Array<[string, string]> = [];
+  const seen_ids = new Set<string>();
+
+  for (const [ins_map, del_map] of states_list) {
+    const ins_entries = Object.entries(ins_map as Record<string, any>);
+    const del_entries = Object.entries(del_map as Record<string, any>);
+    if (ins_entries.length === 0 && del_entries.length === 0) {
+      // A run with only comment/format meta sits between the tracked
+      // elements: they are not siblings, the engine will not group them.
+      if (current.length > 0) {
+        groups.push(current);
+        current = [];
+      }
+      continue;
+    }
+    const state_new: Array<[string, string]> = [];
+    for (const [uid, meta] of ins_entries) {
+      if (!seen_ids.has(uid)) {
+        state_new.push([uid, (meta && meta.author) || "Unknown"]);
+      }
+    }
+    for (const [uid, meta] of del_entries) {
+      if (!seen_ids.has(uid)) {
+        state_new.push([uid, (meta && meta.author) || "Unknown"]);
+      }
+    }
+    for (const [uid, author] of state_new) {
+      seen_ids.add(uid);
+      if (current.length > 0 && current[current.length - 1][1] !== author) {
+        groups.push(current);
+        current = [];
+      }
+      current.push([uid, author]);
+    }
+  }
+  if (current.length > 0) groups.push(current);
+
+  const pair_map: Record<string, string> = {};
+  for (const group of groups) {
+    if (group.length < 2) continue;
+    for (const [uid] of group) {
+      pair_map[uid] = group
+        .filter(([u]) => u !== uid)
+        .map(([u]) => `Chg:${u}`)
+        .join(", ");
+    }
+  }
+  return pair_map;
+}
+
 export function apply_formatting_to_segments(
   text: string,
   prefix: string,