@adeu/core 1.23.0 → 1.25.0

package/dist/index.js

@@ -2787,7 +2787,8 @@ function _row_ops_for_table(table_o, table_m, opcodes, warnings) {
             type: "modify",
             target_text: o_txt,
             new_text: m_txt,
-            comment: "Diff: Table row modified"
+            comment: "Diff: Table row modified",
+            _is_table_edit: true
           });
         }
       }
@@ -2889,14 +2890,18 @@ function generate_structured_edits(text_orig, struct_orig, text_mod, struct_mod)
         if (row_opcodes !== null) {
           edits.push(..._row_ops_for_table(t_o, t_m, row_opcodes, warnings));
         } else {
-          const tbl_edits = generate_edits_from_text(
-            text_orig.substring(t_o.start, t_o.end),
-            text_mod.substring(t_m.start, t_m.end)
-          );
-          for (const e of tbl_edits) {
-            e._match_start_index = (e._match_start_index || 0) + t_o.start;
+          for (let k = 0; k < t_o.rows.length; k++) {
+            const o_txt = t_o.rows[k].cells.join(" | ");
+            const m_txt = t_m.rows[k].cells.join(" | ");
+            if (o_txt === m_txt) continue;
+            edits.push({
+              type: "modify",
+              target_text: o_txt,
+              new_text: m_txt,
+              comment: "Diff: Table row modified",
+              _is_table_edit: true
+            });
           }
-          edits.push(...tbl_edits);
         }
       }
     }
@@ -2914,7 +2919,7 @@ function generate_structured_edits(text_orig, struct_orig, text_mod, struct_mod)
   };
   let ambiguous_anchor_warned = false;
   for (const e of edits) {
-    if ((e.type === "insert_row" || e.type === "delete_row") && !ambiguous_anchor_warned) {
+    if ((e.type === "insert_row" || e.type === "delete_row" || e._is_table_edit) && !ambiguous_anchor_warned) {
       if (e.target_text && countOccurrences(text_orig, e.target_text) > 1) {
         warnings.push(
           `The row anchor "${e.target_text.substring(0, 60)}" appears more than once in the document. Applying this diff from its JSON output may be rejected as ambiguous \u2014 make the anchor rows unique, or apply the row changes with explicit insert_row/delete_row edits.`
@@ -4666,7 +4671,7 @@ var RedlineEngine = class _RedlineEngine {
    *
    * Does NOT attach comments; callers handle that.
    */
-  _track_insert_multiline(text, anchor_run, anchor_paragraph, reuse_id) {
+  _track_insert_multiline(text, anchor_run, anchor_paragraph, reuse_id, positional_anchor_el = null) {
     if (!text) {
       return {
         first_node: null,
@@ -4687,7 +4692,29 @@ var RedlineEngine = class _RedlineEngine {
       }
       current_p = walker;
     }
-    while (lines.length > 1 && lines[lines.length - 1] === "") {
+    const suffix_nodes = [];
+    const pos_source = positional_anchor_el && positional_anchor_el.parentNode ? positional_anchor_el : anchor_run !== null && anchor_run._element.parentNode ? anchor_run._element : null;
+    if (current_p !== null && pos_source !== null) {
+      let pos_anchor = pos_source;
+      while (pos_anchor && pos_anchor.parentNode !== current_p) {
+        pos_anchor = pos_anchor.parentNode;
+        if (pos_anchor === current_p) {
+          pos_anchor = null;
+          break;
+        }
+      }
+      if (pos_anchor) {
+        const relocatable = /* @__PURE__ */ new Set(["w:r", "w:ins", "w:del"]);
+        let nxt = pos_anchor.nextSibling;
+        while (nxt) {
+          if (nxt.nodeType === 1 && relocatable.has(nxt.tagName)) {
+            suffix_nodes.push(nxt);
+          }
+          nxt = nxt.nextSibling;
+        }
+      }
+    }
+    while (lines.length > 1 && lines[lines.length - 1] === "" && suffix_nodes.length === 0) {
       lines.pop();
     }
     if (lines.length === 0) {
@@ -4785,6 +4812,12 @@ var RedlineEngine = class _RedlineEngine {
         first_node = new_p;
       }
     }
+    if (!block_mode && last_p && suffix_nodes.length > 0) {
+      for (const node of suffix_nodes) {
+        node.parentNode?.removeChild(node);
+        last_p.appendChild(node);
+      }
+    }
     return { first_node, last_p, last_ins, used_block_mode: block_mode };
   }
   /**
@@ -4842,7 +4875,7 @@ var RedlineEngine = class _RedlineEngine {
     if (stripped_text.startsWith("* ") || stripped_text.startsWith("- ")) {
       return [stripped_text.substring(2).trim(), "List Paragraph"];
     }
-    const match = stripped_text.match(/^\d+\.\s+/);
+    const match = stripped_text.match(/^1\.\s+/);
     if (match) {
       return [stripped_text.substring(match[0].length).trim(), "List Number"];
     }
@@ -5724,14 +5757,17 @@ var RedlineEngine = class _RedlineEngine {
         resolved_edits.push([edit, edit.new_text || null]);
       } else if (edit.type === "insert_row" || edit.type === "delete_row") {
         let matches = this.mapper.find_all_match_indices(edit.target_text);
+        let resolved_mapper = this.mapper;
         if (matches.length === 0) {
           if (!this.clean_mapper) {
             this.clean_mapper = new DocumentMapper(this.doc, true);
           }
           matches = this.clean_mapper.find_all_match_indices(edit.target_text);
+          resolved_mapper = this.clean_mapper;
         }
         if (matches.length > 0) {
           edit._resolved_start_idx = matches[0][0];
+          edit._active_mapper_ref = resolved_mapper;
           resolved_edits.push([edit, null]);
         } else {
           skipped++;
@@ -5980,7 +6016,8 @@ var RedlineEngine = class _RedlineEngine {
   }
   _apply_table_edit(edit, rebuild_map) {
     const start_idx = edit._resolved_start_idx !== void 0 && edit._resolved_start_idx !== null ? edit._resolved_start_idx : edit._match_start_index || 0;
-    const [anchor_run, anchor_para] = this.mapper.get_insertion_anchor(
+    const active_mapper = edit._active_mapper_ref || this.mapper;
+    const [anchor_run, anchor_para] = active_mapper.get_insertion_anchor(
       start_idx,
       rebuild_map
     );
@@ -6161,9 +6198,14 @@ var RedlineEngine = class _RedlineEngine {
       if (overlaps_virtual_pipe) {
         const actual_cells = actual_doc_text.split("|");
         const new_cells = current_effective_new_text.split("|");
-        if (actual_cells.length === new_cells.length && actual_cells.length > 1) {
+        if (actual_cells.length !== new_cells.length) {
+          throw new BatchValidationError([
+            `Target text spans ${actual_cells.length} table cells, but replacement provides ${new_cells.length}. To modify text without altering table structure (rows or columns), ensure the replacement contains the exact same number of '|' separators (e.g., replace with 'CellC | ' to empty the second cell).`
+          ]);
+        }
+        if (actual_cells.length > 1) {
           const sub_edits2 = [];
-          let search_offset = start_idx;
+          let cell_start_in_target = 0;
           let target_comment_idx = 0;
           for (let idx = 0; idx < actual_cells.length; idx++) {
             if (actual_cells[idx].trim() !== new_cells[idx].trim()) {
@@ -6176,13 +6218,7 @@ var RedlineEngine = class _RedlineEngine {
             const n_cell = new_cells[cell_idx];
             const a_clean = a_cell.trim();
             const n_clean = n_cell.trim();
-            let actual_start = search_offset;
-            if (a_clean) {
-              actual_start = this.mapper.full_text.indexOf(a_clean, search_offset);
-              if (actual_start === -1 || actual_start > search_offset + 10) {
-                actual_start = search_offset;
-              }
-            }
+            const actual_start = start_idx + cell_start_in_target + (a_clean ? a_cell.indexOf(a_clean) : 0);
             const should_attach_comment = edit.comment !== null && edit.comment !== void 0 && cell_idx === target_comment_idx;
             if (a_clean !== n_clean || should_attach_comment) {
               const cell_sub_edits = this._word_diff_sub_edits(
@@ -6199,24 +6235,12 @@ var RedlineEngine = class _RedlineEngine {
                 sub_edits2.push(se);
               }
             }
-            if (a_clean) {
-              search_offset = actual_start + a_clean.length;
-            }
-            const next_pipe = this.mapper.full_text.indexOf(" | ", search_offset);
-            if (next_pipe !== -1 && next_pipe <= search_offset + 10) {
-              search_offset = next_pipe + 3;
-            } else {
-              search_offset += a_cell.length + 1;
-            }
+            cell_start_in_target += a_cell.length + 1;
           }
           for (const sub of sub_edits2) {
             all_sub_edits.push(sub);
           }
           continue;
-        } else {
-          throw new BatchValidationError([
-            `Target text spans ${actual_cells.length} table cells, but replacement provides ${new_cells.length}. To modify text without altering table structure (rows or columns), ensure the replacement contains the exact same number of '|' separators (e.g., replace with 'CellC | ' to empty the second cell).`
-          ]);
         }
       }
       let has_markdown = false;
@@ -6310,6 +6334,19 @@ var RedlineEngine = class _RedlineEngine {
           continue;
         }
       }
+      if (!final_target && final_new) {
+        all_sub_edits.push({
+          type: "modify",
+          target_text: "",
+          new_text: final_new,
+          comment: edit.comment,
+          _resolved_start_idx: effective_start_idx,
+          _match_start_index: effective_start_idx,
+          _internal_op: "INSERTION",
+          _active_mapper_ref: active_mapper
+        });
+        continue;
+      }
       const sub_edits = this._word_diff_sub_edits(
         actual_doc_text,
         current_effective_new_text,
@@ -6751,7 +6788,10 @@ var RedlineEngine = class _RedlineEngine {
         edit.new_text,
         style_source_run,
         mod_anchor_para,
-        ins_id
+        ins_id,
+        // The insertion physically follows the deletion block; the style
+        // run was detached when the deletion cloned it into <w:del>.
+        last_del
       );
       if (result.first_node) {
         const is_inline_first = result.first_node.tagName === "w:ins";
@@ -7298,7 +7338,10 @@ function scrub_doc_properties(doc) {
       ["keywords", "Keywords"],
       ["subject", "Subject"],
       ["contentStatus", "Content status"],
-      ["description", "Description/comments"]
+      ["description", "Description/comments"],
+      ["identifier", "Identifier"],
+      ["language", "Language"],
+      ["version", "Version"]
     ];
     for (const [local, label] of leakFields) {
       findDescendantsByLocalName(corePart._element, local).forEach((c) => {
@@ -7356,27 +7399,73 @@ function scrub_timestamps(doc) {
   }
   return modified ? ["Timestamps normalized to epoch"] : [];
 }
-function strip_custom_xml(doc) {
-  const customParts = doc.pkg.parts.filter((p) => p.partname.includes("/customXml"));
-  if (customParts.length === 0) return [];
-  const partnames = new Set(customParts.map((p) => p.partname));
-  doc.pkg.parts = doc.pkg.parts.filter((p) => !partnames.has(p.partname));
-  const removeRelationsTo = (relsPart) => {
+function ejectPackageMembers(doc, matcher, relTargetMatcher) {
+  const pkg = doc.pkg;
+  const normalized = (p) => p.startsWith("/") ? p.substring(1) : p;
+  for (const part of pkg.parts) {
+    if (!part.partname.endsWith(".rels")) continue;
     const toRemove = [];
-    for (const rel of findAllDescendants(relsPart._element, "Relationship")) {
-      const target = rel.getAttribute("Target");
-      if (target && target.includes("customXml")) toRemove.push(rel);
+    for (const rel of findAllDescendants(part._element, "Relationship")) {
+      const target = rel.getAttribute("Target") || "";
+      if (relTargetMatcher(target)) {
+        toRemove.push(rel);
+        const sourcePath = part.partname.replace("/_rels/", "/").replace(".rels", "");
+        const sourcePart = pkg.getPartByPath(sourcePath);
+        if (sourcePart) {
+          const relId = rel.getAttribute("Id");
+          if (relId) sourcePart.rels.delete(relId);
+        }
+      }
     }
     toRemove.forEach((r) => r.parentNode?.removeChild(r));
-  };
-  const rootRels = doc.pkg.getPartByPath("_rels/.rels");
-  if (rootRels) removeRelationsTo(rootRels);
-  const docRels = doc.pkg.getOrCreateRelsPart(doc.part.partname);
-  if (docRels) removeRelationsTo(docRels);
+  }
+  const ctPart = pkg.getPartByPath("[Content_Types].xml");
+  if (ctPart) {
+    const toRemove = [];
+    for (const override of findAllDescendants(ctPart._element, "Override")) {
+      const partName = override.getAttribute("PartName") || "";
+      if (matcher(normalized(partName))) toRemove.push(override);
+    }
+    toRemove.forEach((o) => o.parentNode?.removeChild(o));
+  }
+  pkg.parts = pkg.parts.filter((p) => !matcher(normalized(p.partname)));
+  for (const key of Object.keys(pkg.unzipped)) {
+    if (matcher(normalized(key))) delete pkg.unzipped[key];
+  }
+}
+function strip_custom_xml(doc) {
+  const isCustomXml = (path) => path.startsWith("customXml/");
+  const customParts = doc.pkg.parts.filter(
+    (p) => isCustomXml(p.partname.startsWith("/") ? p.partname.substring(1) : p.partname)
+  );
+  const leakedMembers = Object.keys(doc.pkg.unzipped).filter(isCustomXml);
+  if (customParts.length === 0 && leakedMembers.length === 0) return [];
+  ejectPackageMembers(doc, isCustomXml, (target) => target.includes("customXml"));
   for (const sdtPr of findAllDescendants(doc.element, "w:sdtPr")) {
     findChildren(sdtPr, "w:dataBinding").forEach((b) => sdtPr.removeChild(b));
   }
-  return [`Custom XML parts: ${customParts.length} removed`];
+  return [`Custom XML parts: ${Math.max(customParts.length, leakedMembers.length)} removed`];
+}
+var CUSTOM_PROPS_PATH = "docProps/custom.xml";
+function strip_custom_properties(doc) {
+  const part = doc.pkg.getPartByPath(CUSTOM_PROPS_PATH);
+  const leaked = Object.keys(doc.pkg.unzipped).includes(CUSTOM_PROPS_PATH);
+  if (!part && !leaked) return [];
+  const lines = [];
+  if (part) {
+    for (const prop of findDescendantsByLocalName(part._element, "property")) {
+      const name = prop.getAttribute("name") || "(unnamed)";
+      const value = (prop.textContent || "").trim();
+      lines.push(`  Custom property removed: ${name} = "${_truncate(value, 60)}"`);
+    }
+  }
+  ejectPackageMembers(
+    doc,
+    (path) => path === CUSTOM_PROPS_PATH,
+    (target) => target.includes("docProps/custom.xml")
+  );
+  const count = Math.max(lines.length, 1);
+  return [`Custom document properties: ${count} removed (docProps/custom.xml)`].concat(lines);
 }
 function strip_image_alt_text(doc) {
   let count = 0;
@@ -8730,6 +8819,9 @@ function _collect_footnote_ids_fast(owned_items) {
   return ordered;
 }
 
+// src/sanitize/core.ts
+import { strFromU8 as strFromU82, unzipSync as unzipSync2 } from "fflate";
+
 // src/sanitize/report.ts
 var SanitizeReport = class {
   filename;
@@ -8764,7 +8856,7 @@ var SanitizeReport = class {
         } 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("last modified by") || lower.includes("revision count") || lower.includes("last printed")) {
+      } 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("language") || lower.includes("version") || lower.includes("last modified by") || lower.includes("revision count") || lower.includes("last printed")) {
         this.metadata_lines.push(line);
       } else if (lower.includes("hyperlink") || lower.includes("warning")) {
         this.warnings.push(line);
@@ -8877,6 +8969,7 @@ async function finalize_document(doc, options) {
   report.add_transform_lines(scrub_doc_properties(doc));
   report.add_transform_lines(scrub_timestamps(doc));
   report.add_transform_lines(strip_custom_xml(doc));
+  report.add_transform_lines(strip_custom_properties(doc));
   report.add_transform_lines(strip_image_alt_text(doc));
   const warnings = audit_hyperlinks(doc);
   for (const w of warnings) report.warnings.push(w);
@@ -8924,8 +9017,47 @@ async function finalize_document(doc, options) {
   }
   if (report.warnings.length > 0) report.status = "clean_with_warnings";
   const outBuffer = await doc.save();
+  verifySanitizedPackage(outBuffer);
   return { reportText: report.render(), outBuffer };
 }
+var VERIFIED_CORE_FIELDS = [
+  ["creator", "author (dc:creator)"],
+  ["lastModifiedBy", "last modified by (cp:lastModifiedBy)"],
+  ["identifier", "identifier (dc:identifier)"],
+  ["description", "description (dc:description)"],
+  ["keywords", "keywords (cp:keywords)"],
+  ["category", "category (cp:category)"],
+  ["subject", "subject (dc:subject)"],
+  ["contentStatus", "content status (cp:contentStatus)"],
+  ["language", "language (dc:language)"],
+  ["version", "version (cp:version)"]
+];
+function verifySanitizedPackage(outBuffer) {
+  const unzipped = unzipSync2(new Uint8Array(outBuffer));
+  const names = Object.keys(unzipped);
+  const problems = [];
+  if (names.includes("docProps/custom.xml")) {
+    problems.push("docProps/custom.xml (custom document properties) is still in the package");
+  }
+  if (names.some((n) => n.startsWith("customXml/"))) {
+    problems.push("customXml/* parts are still in the package");
+  }
+  if (names.includes("docProps/core.xml")) {
+    const core = parseXml(strFromU82(unzipped["docProps/core.xml"]));
+    for (const [local, label] of VERIFIED_CORE_FIELDS) {
+      for (const el of findDescendantsByLocalName(core.documentElement, local)) {
+        if ((el.textContent || "").trim()) {
+          problems.push(`core property ${label} still contains a value`);
+        }
+      }
+    }
+  }
+  if (problems.length > 0) {
+    throw new Error(
+      "Sanitize integrity check failed \u2014 the saved package still contains metadata this run claims to remove:\n  - " + problems.join("\n  - ") + "\nRefusing to report a clean document."
+    );
+  }
+}
 
 // src/index.ts
 function identifyEngine() {