@adeu/core 1.13.0 → 1.15.0

package/dist/index.cjs

@@ -1333,7 +1333,12 @@ ${header}`;
           this._add_virtual_text(prefix, current, item);
           current += prefix.length;
         }
-        current = this._map_paragraph_content(item, current, style_cache, default_pstyle);
+        current = this._map_paragraph_content(
+          item,
+          current,
+          style_cache,
+          default_pstyle
+        );
         is_first_para = false;
         previous_item = item;
       } else if (item instanceof Table) {
@@ -1406,7 +1411,13 @@ ${header}`;
   }
   _map_paragraph_content(paragraph, start_offset, style_cache, default_pstyle) {
     let current = start_offset;
-    const span = { start: current, end: current, text: "", run: null, paragraph };
+    const span = {
+      start: current,
+      end: current,
+      text: "",
+      run: null,
+      paragraph
+    };
     this.spans.push(span);
     const active_ids = /* @__PURE__ */ new Set();
     const active_ins = {};
@@ -1424,7 +1435,7 @@ ${header}`;
         this._add_virtual_text(s_tok, current, paragraph);
         current += s_tok.length;
       }
-      for (const [kind, txt, r_obj, i_id, d_id] of pending_runs) {
+      for (const [kind, txt, r_obj, i_id, d_id, c_ids] of pending_runs) {
         if (kind === "virtual") {
           this._add_virtual_text(txt, current, paragraph, active_hyperlink_id);
         } else {
@@ -1436,7 +1447,8 @@ ${header}`;
             paragraph,
             ins_id: i_id || void 0,
             del_id: d_id || void 0,
-            hyperlink_id: active_hyperlink_id || void 0
+            hyperlink_id: active_hyperlink_id || void 0,
+            comment_ids: c_ids.length > 0 ? c_ids : void 0
           };
           this.spans.push(s);
           this._text_chunks.push(txt);
@@ -1450,8 +1462,16 @@ ${header}`;
       pending_runs = [];
     };
     const items = Array.from(iter_paragraph_content(paragraph));
-    const is_heading = is_heading_paragraph(paragraph, style_cache, default_pstyle);
-    const native_heading = is_native_heading(paragraph, style_cache, default_pstyle);
+    const is_heading = is_heading_paragraph(
+      paragraph,
+      style_cache,
+      default_pstyle
+    );
+    const native_heading = is_native_heading(
+      paragraph,
+      style_cache,
+      default_pstyle
+    );
     let leading_strip_active = is_heading;
     for (let i = 0; i < items.length; i++) {
       const item = items[i];
@@ -1486,7 +1506,12 @@ ${header}`;
         const curr_ins_id = Object.keys(active_ins).pop() || null;
         const curr_del_id = Object.keys(active_del).pop() || null;
         if (full_seg_text && !(this.clean_view && curr_del_id) && !(this.original_view && curr_ins_id)) {
-          const new_wrappers = this.clean_view || this.original_view ? ["", ""] : this._get_wrappers(curr_ins_id, curr_del_id, active_ids, active_fmt);
+          const new_wrappers = this.clean_view || this.original_view ? ["", ""] : this._get_wrappers(
+            curr_ins_id,
+            curr_del_id,
+            active_ids,
+            active_fmt
+          );
           const new_style = [prefix, suffix];
           if (pending_runs.length > 0 && new_wrappers[0] === current_wrappers[0] && new_wrappers[1] === current_wrappers[1]) {
             let skip_leading_prefix = false;
@@ -1494,36 +1519,58 @@ ${header}`;
               pending_runs.pop();
               skip_leading_prefix = true;
             }
+            const curr_comment_ids = Array.from(active_ids);
             for (const [kind, txt, r_obj] of run_parts) {
               if (skip_leading_prefix && kind === "virtual" && txt === new_style[0]) {
                 skip_leading_prefix = false;
                 continue;
               }
-              pending_runs.push([kind, txt, r_obj, curr_ins_id, curr_del_id]);
+              pending_runs.push([
+                kind,
+                txt,
+                r_obj,
+                curr_ins_id,
+                curr_del_id,
+                curr_comment_ids
+              ]);
             }
             current_style = new_style;
           } else {
             flush_pending_runs();
             current_wrappers = new_wrappers;
             current_style = new_style;
+            const curr_comment_ids = Array.from(active_ids);
             for (const [kind, txt, r_obj] of run_parts) {
-              pending_runs.push([kind, txt, r_obj, curr_ins_id, curr_del_id]);
+              pending_runs.push([
+                kind,
+                txt,
+                r_obj,
+                curr_ins_id,
+                curr_del_id,
+                curr_comment_ids
+              ]);
             }
           }
         }
         if (!this.clean_view && !this.original_view) {
           const has_meta = Object.keys(active_ins).length > 0 || Object.keys(active_del).length > 0 || active_ids.size > 0 || Object.keys(active_fmt).length > 0;
           if (has_meta) {
-            deferred_meta_states.push([{ ...active_ins }, { ...active_del }, new Set(active_ids), { ...active_fmt }]);
+            deferred_meta_states.push([
+              { ...active_ins },
+              { ...active_del },
+              new Set(active_ids),
+              { ...active_fmt }
+            ]);
           }
           let should_defer = false;
-          const is_redline = curr_ins_id !== null || curr_del_id !== null || Object.keys(active_fmt).length > 0;
-          if (is_redline) {
+          const has_any_meta = curr_ins_id !== null || curr_del_id !== null || Object.keys(active_fmt).length > 0 || active_ids.size > 0;
+          if (has_any_meta) {
             let j = i + 1;
-            let next_is_redline = false;
+            let next_has_meta = false;
             let temp_ins_count = Object.keys(active_ins).length;
             let temp_del_count = Object.keys(active_del).length;
             let temp_fmt_count = Object.keys(active_fmt).length;
+            const temp_comment_ids = new Set(active_ids);
             while (j < items.length) {
               const next_item = items[j];
               if (next_item instanceof Run) {
@@ -1531,22 +1578,27 @@ ${header}`;
                   j++;
                   continue;
                 }
-                if (temp_ins_count > 0 || temp_del_count > 0 || temp_fmt_count > 0) {
-                  next_is_redline = true;
+                if (temp_ins_count > 0 || temp_del_count > 0 || temp_fmt_count > 0 || temp_comment_ids.size > 0) {
+                  next_has_meta = true;
                 }
                 break;
               } else {
                 const ev = next_item;
                 if (ev.type === "ins_start") temp_ins_count++;
-                else if (ev.type === "ins_end") temp_ins_count = Math.max(0, temp_ins_count - 1);
+                else if (ev.type === "ins_end")
+                  temp_ins_count = Math.max(0, temp_ins_count - 1);
                 else if (ev.type === "del_start") temp_del_count++;
-                else if (ev.type === "del_end") temp_del_count = Math.max(0, temp_del_count - 1);
+                else if (ev.type === "del_end")
+                  temp_del_count = Math.max(0, temp_del_count - 1);
                 else if (ev.type === "fmt_start") temp_fmt_count++;
-                else if (ev.type === "fmt_end") temp_fmt_count = Math.max(0, temp_fmt_count - 1);
+                else if (ev.type === "fmt_end")
+                  temp_fmt_count = Math.max(0, temp_fmt_count - 1);
+                else if (ev.type === "start") temp_comment_ids.add(ev.id);
+                else if (ev.type === "end") temp_comment_ids.delete(ev.id);
               }
               j++;
             }
-            if (next_is_redline) should_defer = true;
+            if (next_has_meta) should_defer = true;
           }
           if (!should_defer && deferred_meta_states.length > 0) {
             const meta_block = this._build_merged_meta_block(deferred_meta_states);
@@ -1635,7 +1687,8 @@ ${header}`;
   _get_wrappers(ins_id, del_id, active_ids, active_fmt) {
     if (del_id) return ["{--", "--}"];
     if (ins_id) return ["{++", "++}"];
-    if (active_ids.size > 0 || Object.keys(active_fmt).length > 0) return ["{==", "==}"];
+    if (active_ids.size > 0 || Object.keys(active_fmt).length > 0)
+      return ["{==", "==}"];
     return ["", ""];
   }
   _build_merged_meta_block(states_list) {
@@ -1643,7 +1696,9 @@ ${header}`;
     const comment_lines = [];
     const seen_sigs = /* @__PURE__ */ new Set();
     for (const [ins_map, del_map, comments_set, fmt_map] of states_list) {
-      for (const [uid, meta] of Object.entries(ins_map)) {
+      for (const [uid, meta] of Object.entries(
+        ins_map
+      )) {
         const sig = `Chg:${uid}`;
         if (!seen_sigs.has(sig)) {
           const auth = meta.author || "Unknown";
@@ -1651,7 +1706,9 @@ ${header}`;
           seen_sigs.add(sig);
         }
       }
-      for (const [uid, meta] of Object.entries(del_map)) {
+      for (const [uid, meta] of Object.entries(
+        del_map
+      )) {
         const sig = `Chg:${uid}`;
         if (!seen_sigs.has(sig)) {
           const auth = meta.author || "Unknown";
@@ -1659,7 +1716,9 @@ ${header}`;
           seen_sigs.add(sig);
         }
       }
-      for (const [uid, meta] of Object.entries(fmt_map)) {
+      for (const [uid, meta] of Object.entries(
+        fmt_map
+      )) {
         const sig = `Chg:${uid}`;
         if (!seen_sigs.has(sig)) {
           const auth = meta.author || "Unknown";
@@ -1733,7 +1792,16 @@ ${header}`;
     if (remaining) parts.push(escapeRegExp(remaining));
     return parts.join("");
   }
-  find_match_index(target_text) {
+  find_match_index(target_text, is_regex = false) {
+    if (is_regex) {
+      try {
+        const pattern = new RegExp(target_text);
+        const match = pattern.exec(this.full_text);
+        if (match) return [match.index, match[0].length];
+      } catch (e) {
+      }
+      return [-1, 0];
+    }
     let start_idx = this.full_text.indexOf(target_text);
     if (start_idx !== -1) return [start_idx, target_text.length];
     const norm_full = this._replace_smart_quotes(this.full_text);
@@ -1753,22 +1821,41 @@ ${header}`;
     }
     return [-1, 0];
   }
-  find_all_match_indices(target_text) {
+  find_all_match_indices(target_text, is_regex = false) {
     if (!target_text) return [];
+    if (is_regex) {
+      try {
+        const pattern = new RegExp(target_text, "g");
+        const matches2 = [...this.full_text.matchAll(pattern)];
+        if (matches2.length > 0)
+          return matches2.map((m) => [m.index, m[0].length]);
+      } catch (e) {
+      }
+      return [];
+    }
     const escapeRegExp = (str) => str.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
-    let matches = [...this.full_text.matchAll(new RegExp(escapeRegExp(target_text), "g"))];
+    let matches = [
+      ...this.full_text.matchAll(new RegExp(escapeRegExp(target_text), "g"))
+    ];
     if (matches.length > 0) return matches.map((m) => [m.index, m[0].length]);
     const norm_full = this._replace_smart_quotes(this.full_text);
     const norm_target = this._replace_smart_quotes(target_text);
-    matches = [...norm_full.matchAll(new RegExp(escapeRegExp(norm_target), "g"))];
+    matches = [
+      ...norm_full.matchAll(new RegExp(escapeRegExp(norm_target), "g"))
+    ];
     if (matches.length > 0) return matches.map((m) => [m.index, m[0].length]);
     const stripped_target = this._strip_markdown_formatting(target_text);
-    matches = [...this.full_text.matchAll(new RegExp(escapeRegExp(stripped_target), "g"))];
+    matches = [
+      ...this.full_text.matchAll(
+        new RegExp(escapeRegExp(stripped_target), "g")
+      )
+    ];
     if (matches.length > 0) return matches.map((m) => [m.index, m[0].length]);
     try {
       const pattern = new RegExp(this._make_fuzzy_regex(target_text), "g");
       matches = [...this.full_text.matchAll(pattern)];
-      if (matches.length > 0) return matches.map((m) => [m.index, m[0].length]);
+      if (matches.length > 0)
+        return matches.map((m) => [m.index, m[0].length]);
     } catch (e) {
     }
     return [];
@@ -1779,14 +1866,22 @@ ${header}`;
     return this._resolve_runs_at_range(start_idx, start_idx + length);
   }
   find_target_runs_by_index(start_index, length, rebuild_map = true) {
-    return this._resolve_runs_at_range(start_index, start_index + length, rebuild_map);
+    return this._resolve_runs_at_range(
+      start_index,
+      start_index + length,
+      rebuild_map
+    );
   }
   get_virtual_spans_in_range(start_index, length) {
     const end_index = start_index + length;
-    return this.spans.filter((s) => s.run === null && s.text === "\n\n" && s.start >= start_index && s.end <= end_index);
+    return this.spans.filter(
+      (s) => s.run === null && s.text === "\n\n" && s.start >= start_index && s.end <= end_index
+    );
   }
   _resolve_runs_at_range(start_idx, end_idx, rebuild_map = true) {
-    const affected_spans = this.spans.filter((s) => s.end > start_idx && s.start < end_idx);
+    const affected_spans = this.spans.filter(
+      (s) => s.end > start_idx && s.start < end_idx
+    );
     if (affected_spans.length === 0) return [];
     const working_runs = affected_spans.filter((s) => s.run !== null).map((s) => s.run);
     if (working_runs.length === 0) return [];
@@ -1797,7 +1892,10 @@ ${header}`;
       const local_start = start_idx - first_real_span.start;
       if (local_start > 0) {
         const idx_in_working = 0;
-        const [, right_run] = this._split_run_at_index(working_runs[idx_in_working], local_start);
+        const [, right_run] = this._split_run_at_index(
+          working_runs[idx_in_working],
+          local_start
+        );
         working_runs[idx_in_working] = right_run;
         dom_modified = true;
         start_split_adjustment = local_start;
@@ -1834,7 +1932,9 @@ ${header}`;
         if (preceding[i].paragraph) return [null, preceding[i].paragraph];
       }
     }
-    const containing = this.spans.filter((s) => s.start < index && index < s.end);
+    const containing = this.spans.filter(
+      (s) => s.start < index && index < s.end
+    );
     if (containing.length > 0) {
       const span = containing[0];
       if (span.run === null) {
@@ -1857,10 +1957,12 @@ ${header}`;
     const preceding_gap = this.spans.filter((s) => s.end < index);
     if (preceding_gap.length > 0) {
       for (let i = preceding_gap.length - 1; i >= 0; i--) {
-        if (preceding_gap[i].run) return [preceding_gap[i].run, preceding_gap[i].paragraph];
+        if (preceding_gap[i].run)
+          return [preceding_gap[i].run, preceding_gap[i].paragraph];
       }
       for (let i = preceding_gap.length - 1; i >= 0; i--) {
-        if (preceding_gap[i].paragraph) return [null, preceding_gap[i].paragraph];
+        if (preceding_gap[i].paragraph)
+          return [null, preceding_gap[i].paragraph];
       }
     }
     return [null, null];
@@ -1873,7 +1975,10 @@ ${header}`;
     const new_r_element = run._element.cloneNode(true);
     this._set_run_text_elements(new_r_element, right_text);
     if (run._element.parentNode) {
-      run._element.parentNode.insertBefore(new_r_element, run._element.nextSibling);
+      run._element.parentNode.insertBefore(
+        new_r_element,
+        run._element.nextSibling
+      );
     }
     const new_run = new Run(new_r_element, run._parent);
     return [run, new_run];
@@ -1900,7 +2005,9 @@ ${header}`;
     }
   }
   get_context_at_range(start_idx, end_idx) {
-    const real_spans = this.spans.filter((s) => s.run && s.end > start_idx && s.start < end_idx);
+    const real_spans = this.spans.filter(
+      (s) => s.run && s.end > start_idx && s.start < end_idx
+    );
     if (real_spans.length > 0) return real_spans[0];
     return null;
   }
@@ -2252,6 +2359,193 @@ function create_word_patch_diff(original_text, modified_text, original_path = "O
   return output.join("\n");
 }
 
+// src/pagination.ts
+var PAGE_TARGET_CHARS = 19e3;
+var APPENDIX_MARKER = "<!-- READONLY_BOUNDARY_START -->";
+var _CRITIC_TOKENS = {
+  "{++": "++}",
+  "{--": "--}",
+  "{==": "==}",
+  "{>>": "<<}"
+};
+var _CHG_ID_PATTERN = /\bChg:(\d+)\b/g;
+function split_structural_appendix(markdown) {
+  if (!markdown) return ["", ""];
+  const idx = markdown.indexOf(APPENDIX_MARKER);
+  if (idx === -1) return [markdown, ""];
+  const line_start = markdown.lastIndexOf("\n", idx) + 1;
+  const body = markdown.substring(0, line_start).trimEnd();
+  const appendix = markdown.substring(line_start);
+  return [body, appendix];
+}
+function paginate(markdown_body, structural_appendix = "") {
+  if (!markdown_body) {
+    const appendix_clean = structural_appendix ? structural_appendix.trim() : "";
+    const content = appendix_clean;
+    return {
+      pages: [{
+        page: 1,
+        total_pages: 1,
+        has_next: false,
+        has_prev: false,
+        tracked_change_count: _count_tracked_changes(content),
+        page_content: content
+      }],
+      total_pages: 1,
+      body_pages: [""],
+      body_page_offsets: [0]
+    };
+  }
+  const block_records = _tokenize_into_atomic_blocks(markdown_body);
+  const [body_pages, body_page_offsets] = _assemble_pages(block_records);
+  let final_pages;
+  if (structural_appendix && structural_appendix.trim()) {
+    const appendix = structural_appendix.trim();
+    final_pages = body_pages.map((bp) => bp ? `${bp}
+
+${appendix}` : appendix);
+  } else {
+    final_pages = [...body_pages];
+  }
+  const total = final_pages.length;
+  const page_infos = final_pages.map((content, i) => ({
+    page: i + 1,
+    total_pages: total,
+    has_next: i + 1 < total,
+    has_prev: i + 1 > 1,
+    tracked_change_count: _count_tracked_changes(content),
+    page_content: content
+  }));
+  return {
+    pages: page_infos,
+    total_pages: total,
+    body_pages,
+    body_page_offsets
+  };
+}
+function _tokenize_into_atomic_blocks(markdown_body) {
+  const raw_blocks = _split_on_safe_paragraph_breaks(markdown_body);
+  return _merge_footnote_sections(raw_blocks);
+}
+function _split_on_safe_paragraph_breaks(text) {
+  const counters = { "++}": 0, "--}": 0, "==}": 0, "<<}": 0 };
+  const blocks = [];
+  let block_start = 0;
+  let i = 0;
+  const n = text.length;
+  while (i < n) {
+    let matched_open = false;
+    for (const [open_tok, close_tok] of Object.entries(_CRITIC_TOKENS)) {
+      if (text.startsWith(open_tok, i)) {
+        counters[close_tok]++;
+        i += open_tok.length;
+        matched_open = true;
+        break;
+      }
+    }
+    if (matched_open) continue;
+    let matched_close = false;
+    for (const close_tok of Object.values(_CRITIC_TOKENS)) {
+      if (text.startsWith(close_tok, i)) {
+        if (counters[close_tok] > 0) counters[close_tok]--;
+        i += close_tok.length;
+        matched_close = true;
+        break;
+      }
+    }
+    if (matched_close) continue;
+    if (text[i] === "\n" && i + 1 < n && text[i + 1] === "\n") {
+      if (Object.values(counters).every((c) => c === 0)) {
+        const block_text = text.substring(block_start, i);
+        if (block_text) blocks.push([block_text, block_start]);
+        let j = i;
+        while (j < n && text[j] === "\n") j++;
+        i = j;
+        block_start = i;
+        continue;
+      }
+    }
+    i++;
+  }
+  if (block_start < n) {
+    const block_text = text.substring(block_start, n);
+    if (block_text) blocks.push([block_text, block_start]);
+  }
+  return blocks;
+}
+function _merge_footnote_sections(blocks) {
+  if (!blocks.length) return blocks;
+  const merged = [];
+  let i = 0;
+  while (i < blocks.length) {
+    const [block_text, block_offset] = blocks[i];
+    const stripped = block_text.trimStart();
+    const is_section_header = stripped.startsWith("## Footnotes") || stripped.startsWith("## Endnotes");
+    if (!is_section_header) {
+      merged.push([block_text, block_offset]);
+      i++;
+      continue;
+    }
+    let accumulated_text = block_text;
+    let j = i + 1;
+    while (j < blocks.length) {
+      const [next_text] = blocks[j];
+      const next_stripped = next_text.trimStart();
+      if (next_stripped.startsWith("[^fn-") || next_stripped.startsWith("[^en-")) {
+        accumulated_text = `${accumulated_text}
+
+${next_text}`;
+        j++;
+      } else {
+        break;
+      }
+    }
+    merged.push([accumulated_text, block_offset]);
+    i = j;
+  }
+  return merged;
+}
+function _assemble_pages(block_records) {
+  if (!block_records.length) return [[""], [0]];
+  const pages = [];
+  const page_starts = [];
+  let current_blocks = [];
+  let current_size = 0;
+  let current_start = -1;
+  const flush_current = () => {
+    if (current_blocks.length > 0) {
+      pages.push(current_blocks.join("\n\n"));
+      page_starts.push(current_start);
+    }
+    current_blocks = [];
+    current_size = 0;
+    current_start = -1;
+  };
+  for (const [block_text, block_offset] of block_records) {
+    const block_size = block_text.length;
+    const added_size = block_size + (current_blocks.length > 0 ? 2 : 0);
+    if (current_blocks.length > 0 && current_size + added_size > PAGE_TARGET_CHARS) {
+      flush_current();
+    }
+    if (current_blocks.length === 0 && block_size > PAGE_TARGET_CHARS) {
+      pages.push(block_text);
+      page_starts.push(block_offset);
+      continue;
+    }
+    if (current_blocks.length === 0) current_start = block_offset;
+    current_blocks.push(block_text);
+    current_size += current_size > 0 ? added_size : block_size;
+  }
+  flush_current();
+  if (!pages.length) return [[""], [0]];
+  return [pages, page_starts];
+}
+function _count_tracked_changes(page_content) {
+  const matches = [...page_content.matchAll(_CHG_ID_PATTERN)];
+  const distinct = new Set(matches.map((m) => m[1]));
+  return distinct.size;
+}
+
 // src/markup.ts
 var AMBIGUITY_EXAMPLES_CAP = 5;
 var AMBIGUITY_CONTEXT_CHARS = 50;
@@ -2731,7 +3025,8 @@ var RedlineEngine = class {
       start_idx + length,
       start_idx + length + 30
     );
-    const critic_markup = `${context_before}{--${target_text}--}{++${new_text}++}${context_after}`;
+    const insertion = new_text ? `{++${new_text}++}` : "";
+    const critic_markup = `${context_before}{--${target_text}--}${insertion}${context_after}`;
     let clean_text = critic_markup;
     clean_text = clean_text.replace(/\{>>.*?<<\}/gs, "");
     clean_text = clean_text.replace(/\{--.*?--\}/gs, "");
@@ -2749,6 +3044,37 @@ var RedlineEngine = class {
     }
     return maxId;
   }
+  _get_heading_path_and_page(start_idx, text, page_offsets) {
+    let page = 1;
+    for (let i = 0; i < page_offsets.length; i++) {
+      if (start_idx >= page_offsets[i]) {
+        page = i + 1;
+      } else {
+        break;
+      }
+    }
+    const textBefore = text.substring(0, start_idx);
+    const lines = textBefore.split("\n");
+    const path = [];
+    let current_level = 999;
+    for (let i = lines.length - 1; i >= 0; i--) {
+      const line = lines[i];
+      const m = line.match(/^(#{1,6})\s+(.*)/);
+      if (m) {
+        const level = m[1].length;
+        if (level < current_level) {
+          let cleanHeading = m[2].replace(/\*\*|__|[*_]/g, "").replace(/\{#[^}]+\}/g, "").trim();
+          if (cleanHeading.length > 80) {
+            cleanHeading = cleanHeading.substring(0, 80) + "...";
+          }
+          path.unshift(cleanHeading);
+          current_level = level;
+          if (level === 1) break;
+        }
+      }
+    }
+    return [path.join(" > "), page];
+  }
   accept_all_revisions() {
     const parts_to_process = [this.doc.element];
     for (const part of this.doc.pkg.parts) {
@@ -3447,13 +3773,15 @@ var RedlineEngine = class {
     for (let i = 0; i < edits.length; i++) {
       const edit = edits[i];
       if (!edit.target_text) continue;
-      let matches = this.mapper.find_all_match_indices(edit.target_text);
+      const is_regex = edit.regex || false;
+      const match_mode = edit.match_mode || "strict";
+      let matches = this.mapper.find_all_match_indices(edit.target_text, is_regex);
       let activeText = this.mapper.full_text;
       let target_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);
+        matches = this.clean_mapper.find_all_match_indices(edit.target_text, is_regex);
         if (matches.length > 0) {
           activeText = this.clean_mapper.full_text;
           target_mapper = this.clean_mapper;
@@ -3475,7 +3803,7 @@ var RedlineEngine = class {
         if (!this.original_mapper) {
           this.original_mapper = new DocumentMapper(this.doc, false, true);
         }
-        const orig_matches = this.original_mapper.find_all_match_indices(edit.target_text);
+        const orig_matches = this.original_mapper.find_all_match_indices(edit.target_text, is_regex);
         if (orig_matches.length > 0) {
           is_deleted_text = true;
           for (const [start, length] of orig_matches) {
@@ -3512,7 +3840,7 @@ var RedlineEngine = class {
   "${edit.target_text}"`
           );
         }
-      } else if (matches.length > 1) {
+      } else if (matches.length > 1 && match_mode === "strict") {
         const positions = matches.map(([start, length]) => [
           start,
           start + length
@@ -3570,6 +3898,14 @@ var RedlineEngine = class {
               if (auth && auth !== this.author) nestedAuthors.add(auth);
             }
           }
+          if (s.comment_ids) {
+            for (const cid of s.comment_ids) {
+              const c_data = this.mapper.comments_map[cid];
+              if (c_data && c_data.author && c_data.author !== this.author) {
+                nestedAuthors.add(c_data.author);
+              }
+            }
+          }
         }
         if (nestedAuthors.size > 0) {
           errors.push(
@@ -3618,6 +3954,22 @@ var RedlineEngine = class {
     return errors;
   }
   process_batch(changes, dry_run = false) {
+    if (Array.isArray(changes)) {
+      changes = changes.map((item) => {
+        if (typeof item === "string") {
+          try {
+            const parsed = JSON.parse(item);
+            if (parsed !== null && typeof parsed === "object") {
+              return parsed;
+            }
+            return item;
+          } catch {
+            return item;
+          }
+        }
+        return item;
+      });
+    }
     if (dry_run) {
       const baselines = /* @__PURE__ */ new Map();
       for (const part of this.doc.pkg.parts) {
@@ -3684,6 +4036,9 @@ var RedlineEngine = class {
         if (this.clean_mapper) this.clean_mapper["_build_map"]();
       }
     }
+    const [body_text] = split_structural_appendix(this.mapper.full_text);
+    const pag_res = paginate(body_text, "");
+    const page_offsets = pag_res.body_page_offsets;
     const edits_reports = [];
     let applied_edits = 0;
     let skipped_edits = 0;
@@ -3707,9 +4062,8 @@ var RedlineEngine = class {
             });
             continue;
           }
-          const res = this.apply_edits([edit]);
-          const applied = res[0];
-          if (applied > 0) {
+          const res = this.apply_edits([edit], page_offsets);
+          if (edit._applied_status) {
             applied_edits++;
             const previews = this._build_edit_context_previews(edit);
             edits_reports.push({
@@ -3719,7 +4073,11 @@ var RedlineEngine = class {
               warning,
               error: null,
               critic_markup: previews[0],
-              clean_text: previews[1]
+              clean_text: previews[1],
+              pages: edit._pages || [],
+              heading_path: edit._heading_path || "",
+              occurrences_modified: edit._occurrences_modified || 0,
+              match_mode: edit.match_mode || "strict"
             });
           } else {
             skipped_edits++;
@@ -3741,9 +4099,9 @@ var RedlineEngine = class {
           throw new BatchValidationError(errors);
         }
         const cloned_edits = edits.map((e) => JSON.parse(JSON.stringify(e)));
-        const res = this.apply_edits(cloned_edits);
-        applied_edits = res[0];
-        skipped_edits = res[1];
+        const res = this.apply_edits(cloned_edits, page_offsets);
+        applied_edits = cloned_edits.filter((e) => e._applied_status).length;
+        skipped_edits = cloned_edits.length - applied_edits;
         for (const edit of cloned_edits) {
           const success = edit._applied_status || false;
           const error_msg = edit._error_msg || null;
@@ -3764,7 +4122,11 @@ var RedlineEngine = class {
             warning,
             error: error_msg,
             critic_markup,
-            clean_text
+            clean_text,
+            pages: edit._pages || [],
+            heading_path: edit._heading_path || "",
+            occurrences_modified: edit._occurrences_modified || 0,
+            match_mode: edit.match_mode || "strict"
           });
         }
       }
@@ -3780,9 +4142,13 @@ var RedlineEngine = class {
       version: "1.10.0"
     };
   }
-  apply_edits(edits) {
+  apply_edits(edits, page_offsets = []) {
     let applied = 0;
     let skipped = 0;
+    if (!page_offsets || page_offsets.length === 0) {
+      const [body_text] = split_structural_appendix(this.mapper.full_text);
+      page_offsets = paginate(body_text, "").body_page_offsets;
+    }
     const resolved_edits = [];
     for (const edit of edits) {
       edit._applied_status = false;
@@ -3884,6 +4250,19 @@ var RedlineEngine = class {
         const parent = edit._parent_edit_ref;
         if (parent) {
           parent._applied_status = true;
+          parent._occurrences_modified = (parent._occurrences_modified || 0) + 1;
+          const [path, page] = this._get_heading_path_and_page(start, this.mapper.full_text, page_offsets);
+          const pages = parent._pages || [];
+          if (!pages.includes(page)) pages.unshift(page);
+          parent._pages = pages;
+          parent._heading_path = path;
+        } else {
+          edit._occurrences_modified = (edit._occurrences_modified || 0) + 1;
+          const [path, page] = this._get_heading_path_and_page(start, this.mapper.full_text, page_offsets);
+          const pages = edit._pages || [];
+          if (!pages.includes(page)) pages.unshift(page);
+          edit._pages = pages;
+          edit._heading_path = path;
         }
       } else {
         skipped++;
@@ -4036,96 +4415,192 @@ var RedlineEngine = class {
     }
     return false;
   }
-  /**
-   * Returns the first match of `target_text` in the raw mapper that is NOT
-   * entirely contained within a tracked deletion (<w:del>). Tracked-deleted
-   * copies are not live, editable text, so an edit must resolve to a live
-   * occurrence even when a dead copy appears earlier in the document
-   * (BUG-23-5). Falls back to the plain first match when no live copy is
-   * found (e.g. fuzzy/normalized matches the span filter cannot align).
-   */
-  _first_live_match(target_text) {
-    const all = this.mapper.find_all_match_indices(target_text);
-    if (all.length <= 1) {
-      return this.mapper.find_match_index(target_text);
-    }
-    for (const [start, length] of all) {
-      const realSpans = this.mapper.spans.filter(
-        (s) => s.run !== null && s.end > start && s.start < start + length
-      );
-      if (realSpans.length === 0) return [start, length];
-      if (realSpans.some((s) => !s.del_id)) return [start, length];
-    }
-    return this.mapper.find_match_index(target_text);
-  }
   _pre_resolve_heuristic_edit(edit) {
     if (!edit.target_text) return null;
-    let [start_idx, match_len] = this._first_live_match(edit.target_text);
+    const is_regex = edit.regex || false;
+    const match_mode = edit.match_mode || "strict";
+    let matches = this.mapper.find_all_match_indices(edit.target_text, is_regex);
     let use_clean_map = false;
-    if (start_idx === -1) {
+    if (matches.length === 0) {
       if (!this.clean_mapper)
         this.clean_mapper = new DocumentMapper(this.doc, true);
-      [start_idx, match_len] = this.clean_mapper.find_match_index(
-        edit.target_text
-      );
-      if (start_idx !== -1) use_clean_map = true;
+      matches = this.clean_mapper.find_all_match_indices(edit.target_text, is_regex);
+      if (matches.length > 0) use_clean_map = true;
       else return null;
     }
     const active_mapper = use_clean_map ? this.clean_mapper : this.mapper;
-    const effective_new_text = edit.new_text || "";
-    const actual_doc_text = this.mapper.full_text.substring(
-      start_idx,
-      start_idx + match_len
-    );
-    if (actual_doc_text === effective_new_text || edit.target_text === effective_new_text) {
-      return {
-        type: "modify",
-        target_text: actual_doc_text,
-        new_text: actual_doc_text,
-        comment: edit.comment,
-        _match_start_index: start_idx,
-        _internal_op: "COMMENT_ONLY",
-        _active_mapper_ref: active_mapper
-      };
-    }
-    let effective_op = "";
-    let final_target = "";
-    let final_new = "";
-    let effective_start_idx = start_idx;
-    if (effective_new_text.startsWith(actual_doc_text)) {
-      effective_op = "INSERTION";
-      final_new = effective_new_text.substring(actual_doc_text.length);
-      effective_start_idx = start_idx + match_len;
-    } else {
-      const [prefix_len, suffix_len] = trim_common_context(
-        actual_doc_text,
-        effective_new_text
+    let live_matches = [];
+    for (const [s, match_len] of matches) {
+      const realSpans = active_mapper.spans.filter(
+        (span) => span.run !== null && span.end > s && span.start < s + match_len
       );
-      const t_end = actual_doc_text.length - suffix_len;
-      const n_end = effective_new_text.length - suffix_len;
-      final_target = actual_doc_text.substring(prefix_len, t_end);
-      final_new = effective_new_text.substring(prefix_len, n_end);
-      effective_start_idx = start_idx + prefix_len;
-      if (!final_target && final_new) effective_op = "INSERTION";
-      else if (final_target && !final_new) effective_op = "DELETION";
-      else if (final_target && final_new) effective_op = "MODIFICATION";
-      else effective_op = "COMMENT_ONLY";
+      if (realSpans.length === 0 || realSpans.some((span) => !span.del_id)) {
+        live_matches.push([s, match_len]);
+      }
+    }
+    if (live_matches.length === 0) return null;
+    if (match_mode === "strict" || match_mode === "first") {
+      live_matches = live_matches.slice(0, 1);
+    }
+    const all_sub_edits = [];
+    for (const [start_idx, match_len] of live_matches) {
+      const actual_doc_text = active_mapper.full_text.substring(
+        start_idx,
+        start_idx + match_len
+      );
+      let current_effective_new_text = edit.new_text || "";
+      if (is_regex && current_effective_new_text) {
+        try {
+          current_effective_new_text = actual_doc_text.replace(new RegExp(edit.target_text), current_effective_new_text);
+        } catch (e) {
+        }
+      }
+      const [edit_target_clean, edit_target_style] = this._parse_markdown_style(edit.target_text);
+      const [edit_new_clean, edit_new_style] = this._parse_markdown_style(current_effective_new_text);
+      if (edit_target_style !== edit_new_style) {
+        const [actual_clean] = this._parse_markdown_style(actual_doc_text);
+        const final_target2 = actual_clean;
+        const final_new2 = edit_new_clean;
+        const style_op = final_target2 === final_new2 ? "STYLE_ONLY" : "STYLE_AND_TEXT";
+        const prefix_offset = actual_doc_text.indexOf(actual_clean);
+        const effective_start_idx2 = start_idx + (prefix_offset !== -1 ? prefix_offset : 0);
+        const resolved_style = edit_new_style !== null ? edit_new_style : "Normal";
+        all_sub_edits.push({
+          type: "modify",
+          target_text: final_target2,
+          new_text: final_new2,
+          comment: edit.comment,
+          _match_start_index: effective_start_idx2,
+          _internal_op: style_op,
+          _new_style: resolved_style,
+          _active_mapper_ref: active_mapper
+        });
+        continue;
+      }
+      if (actual_doc_text === current_effective_new_text || edit.target_text === current_effective_new_text) {
+        all_sub_edits.push({
+          type: "modify",
+          target_text: actual_doc_text,
+          new_text: actual_doc_text,
+          comment: edit.comment,
+          _match_start_index: start_idx,
+          _internal_op: "COMMENT_ONLY",
+          _active_mapper_ref: active_mapper
+        });
+        continue;
+      }
+      let effective_op = "";
+      let final_target = "";
+      let final_new = "";
+      let effective_start_idx = start_idx;
+      if (current_effective_new_text.startsWith(actual_doc_text)) {
+        effective_op = "INSERTION";
+        final_new = current_effective_new_text.substring(actual_doc_text.length);
+        effective_start_idx = start_idx + match_len;
+      } else {
+        const [prefix_len, suffix_len] = trim_common_context(
+          actual_doc_text,
+          current_effective_new_text
+        );
+        const t_end = actual_doc_text.length - suffix_len;
+        const n_end = current_effective_new_text.length - suffix_len;
+        final_target = actual_doc_text.substring(prefix_len, t_end);
+        final_new = current_effective_new_text.substring(prefix_len, n_end);
+        effective_start_idx = start_idx + prefix_len;
+        if (!final_target && final_new) effective_op = "INSERTION";
+        else if (final_target && !final_new) effective_op = "DELETION";
+        else if (final_target && final_new) effective_op = "MODIFICATION";
+        else effective_op = "COMMENT_ONLY";
+      }
+      all_sub_edits.push({
+        type: "modify",
+        target_text: final_target,
+        new_text: final_new,
+        comment: edit.comment,
+        _match_start_index: effective_start_idx,
+        _internal_op: effective_op,
+        _active_mapper_ref: active_mapper
+      });
+    }
+    if (all_sub_edits.length === 0) return null;
+    if (match_mode === "all" || all_sub_edits.length > 1) return all_sub_edits;
+    return all_sub_edits[0];
+  }
+  _apply_single_edit_indexed(edit, orig_new, rebuild_map) {
+    let op = edit._internal_op;
+    const active_mapper = edit._active_mapper_ref || this.mapper;
+    const start_idx = edit._resolved_start_idx !== void 0 && edit._resolved_start_idx !== null ? edit._resolved_start_idx : edit._match_start_index || 0;
+    const length = edit.target_text ? edit.target_text.length : 0;
+    if (op === "STYLE_ONLY" || op === "STYLE_AND_TEXT") {
+      const [anchor_run, anchor_para] = active_mapper.get_insertion_anchor(
+        start_idx,
+        rebuild_map
+      );
+      let target_para_el = null;
+      if (anchor_para) {
+        target_para_el = anchor_para._element;
+      } else if (anchor_run) {
+        let walker = anchor_run._element;
+        while (walker && walker.tagName !== "w:p") {
+          walker = walker.parentNode;
+        }
+        target_para_el = walker;
+      }
+      if (target_para_el && edit._new_style) {
+        this._set_paragraph_style(target_para_el, edit._new_style);
+      }
+      if (op === "STYLE_ONLY") {
+        if (edit.comment) {
+          const target_runs2 = active_mapper.find_target_runs_by_index(
+            start_idx,
+            length,
+            rebuild_map
+          );
+          if (target_runs2.length > 0) {
+            const first_el = target_runs2[0]._element;
+            const last_el = target_runs2[target_runs2.length - 1]._element;
+            let start_p = first_el;
+            while (start_p && start_p.tagName !== "w:p")
+              start_p = start_p.parentNode;
+            let end_p = last_el;
+            while (end_p && end_p.tagName !== "w:p")
+              end_p = end_p.parentNode;
+            if (start_p && end_p) {
+              const ascend_to_paragraph_child = (el, p) => {
+                let cur = el;
+                while (cur.parentNode && cur.parentNode !== p) {
+                  cur = cur.parentNode;
+                }
+                return cur;
+              };
+              const first_anchor = ascend_to_paragraph_child(first_el, start_p);
+              const last_anchor = ascend_to_paragraph_child(last_el, end_p);
+              if (start_p === end_p) {
+                this._attach_comment(start_p, first_anchor, last_anchor, edit.comment);
+              } else {
+                this._attach_comment_spanning(
+                  start_p,
+                  first_anchor,
+                  end_p,
+                  last_anchor,
+                  edit.comment
+                );
+              }
+            }
+          }
+        }
+        return true;
+      }
+      if (edit.target_text && edit.new_text) {
+        op = "MODIFICATION";
+      } else if (!edit.target_text && edit.new_text) {
+        op = "INSERTION";
+      } else if (edit.target_text && !edit.new_text) {
+        op = "DELETION";
+      } else {
+        op = "COMMENT_ONLY";
+      }
     }
-    return {
-      type: "modify",
-      target_text: final_target,
-      new_text: final_new,
-      comment: edit.comment,
-      _match_start_index: effective_start_idx,
-      _internal_op: effective_op,
-      _active_mapper_ref: active_mapper
-    };
-  }
-  _apply_single_edit_indexed(edit, orig_new, rebuild_map) {
-    let op = edit._internal_op;
-    const active_mapper = edit._active_mapper_ref || this.mapper;
-    const start_idx = edit._resolved_start_idx !== void 0 && edit._resolved_start_idx !== null ? edit._resolved_start_idx : edit._match_start_index || 0;
-    const length = edit.target_text ? edit.target_text.length : 0;
     const del_id = ["DELETION", "MODIFICATION"].includes(op) ? this._getNextId() : null;
     const ins_id = ["INSERTION", "MODIFICATION"].includes(op) ? this._getNextId() : null;
     if (op === "COMMENT_ONLY") {
@@ -4508,1863 +4983,1814 @@ var RedlineEngine = class {
   }
 };
 
-// src/pagination.ts
-var PAGE_TARGET_CHARS = 19e3;
-var APPENDIX_MARKER = "<!-- READONLY_BOUNDARY_START -->";
-var _CRITIC_TOKENS = {
-  "{++": "++}",
-  "{--": "--}",
-  "{==": "==}",
-  "{>>": "<<}"
-};
-var _CHG_ID_PATTERN = /\bChg:(\d+)\b/g;
-function split_structural_appendix(markdown) {
-  if (!markdown) return ["", ""];
-  const idx = markdown.indexOf(APPENDIX_MARKER);
-  if (idx === -1) return [markdown, ""];
-  const line_start = markdown.lastIndexOf("\n", idx) + 1;
-  const body = markdown.substring(0, line_start).trimEnd();
-  const appendix = markdown.substring(line_start);
-  return [body, appendix];
-}
-function paginate(markdown_body, structural_appendix = "") {
-  if (!markdown_body) {
-    const appendix_clean = structural_appendix ? structural_appendix.trim() : "";
-    const content = appendix_clean;
-    return {
-      pages: [{
-        page: 1,
-        total_pages: 1,
-        has_next: false,
-        has_prev: false,
-        tracked_change_count: _count_tracked_changes(content),
-        page_content: content
-      }],
-      total_pages: 1,
-      body_pages: [""],
-      body_page_offsets: [0]
-    };
-  }
-  const block_records = _tokenize_into_atomic_blocks(markdown_body);
-  const [body_pages, body_page_offsets] = _assemble_pages(block_records);
-  let final_pages;
-  if (structural_appendix && structural_appendix.trim()) {
-    const appendix = structural_appendix.trim();
-    final_pages = body_pages.map((bp) => bp ? `${bp}
-
-${appendix}` : appendix);
-  } else {
-    final_pages = [...body_pages];
+// src/sanitize/transforms.ts
+function findDescendantsByLocalName(element, localName) {
+  const result = [];
+  const all = element.getElementsByTagName("*");
+  for (let i = 0; i < all.length; i++) {
+    const tag = all[i].tagName;
+    if (tag === localName || tag.endsWith(":" + localName)) {
+      result.push(all[i]);
+    }
   }
-  const total = final_pages.length;
-  const page_infos = final_pages.map((content, i) => ({
-    page: i + 1,
-    total_pages: total,
-    has_next: i + 1 < total,
-    has_prev: i + 1 > 1,
-    tracked_change_count: _count_tracked_changes(content),
-    page_content: content
-  }));
-  return {
-    pages: page_infos,
-    total_pages: total,
-    body_pages,
-    body_page_offsets
-  };
-}
-function _tokenize_into_atomic_blocks(markdown_body) {
-  const raw_blocks = _split_on_safe_paragraph_breaks(markdown_body);
-  return _merge_footnote_sections(raw_blocks);
+  return result;
 }
-function _split_on_safe_paragraph_breaks(text) {
-  const counters = { "++}": 0, "--}": 0, "==}": 0, "<<}": 0 };
-  const blocks = [];
-  let block_start = 0;
-  let i = 0;
-  const n = text.length;
-  while (i < n) {
-    let matched_open = false;
-    for (const [open_tok, close_tok] of Object.entries(_CRITIC_TOKENS)) {
-      if (text.startsWith(open_tok, i)) {
-        counters[close_tok]++;
-        i += open_tok.length;
-        matched_open = true;
-        break;
+function coalesce_runs(doc) {
+  let count = 0;
+  function areRunsIdentical(rPr1, rPr2) {
+    const xml1 = rPr1 ? rPr1.toString() : "";
+    const xml2 = rPr2 ? rPr2.toString() : "";
+    return xml1 === xml2;
+  }
+  function hasSpecialContent(run) {
+    const safeTags = ["w:t", "w:tab", "w:br", "w:cr", "w:delText", "w:rPr"];
+    for (let i = 0; i < run.childNodes.length; i++) {
+      const child = run.childNodes[i];
+      if (child.nodeType === 1) {
+        const tag = child.tagName;
+        if (!safeTags.includes(tag)) return true;
       }
     }
-    if (matched_open) continue;
-    let matched_close = false;
-    for (const close_tok of Object.values(_CRITIC_TOKENS)) {
-      if (text.startsWith(close_tok, i)) {
-        if (counters[close_tok] > 0) counters[close_tok]--;
-        i += close_tok.length;
-        matched_close = true;
-        break;
+    return false;
+  }
+  function coalesceContainer(container) {
+    const children = Array.from(container.childNodes).filter((n) => n.nodeType === 1);
+    let i = 0;
+    while (i < children.length - 1) {
+      const curr = children[i];
+      const nxt = children[i + 1];
+      if (curr.tagName === "w:r" && nxt.tagName === "w:r") {
+        if (!hasSpecialContent(curr) && !hasSpecialContent(nxt)) {
+          const rPr1 = findChild(curr, "w:rPr");
+          const rPr2 = findChild(nxt, "w:rPr");
+          if (areRunsIdentical(rPr1, rPr2)) {
+            let last_t = null;
+            for (let c = 0; c < curr.childNodes.length; c++) {
+              const child = curr.childNodes[c];
+              if (child.nodeType === 1 && (child.tagName === "w:t" || child.tagName === "w:delText")) {
+                last_t = child;
+              }
+            }
+            const nxtChildren = Array.from(nxt.childNodes).filter((n) => n.nodeType === 1);
+            for (const child of nxtChildren) {
+              if (child.tagName === "w:rPr") continue;
+              if ((child.tagName === "w:t" || child.tagName === "w:delText") && last_t && last_t.tagName === child.tagName) {
+                const t1 = last_t.textContent || "";
+                const t2 = child.textContent || "";
+                const combined = t1 + t2;
+                last_t.textContent = combined;
+                if (combined.trim() !== combined) {
+                  last_t.setAttribute("xml:space", "preserve");
+                }
+              } else {
+                curr.appendChild(child);
+                if (child.tagName === "w:t" || child.tagName === "w:delText") {
+                  last_t = child;
+                }
+              }
+            }
+            container.removeChild(nxt);
+            children.splice(i + 1, 1);
+            count++;
+            continue;
+          }
+        }
       }
-    }
-    if (matched_close) continue;
-    if (text[i] === "\n" && i + 1 < n && text[i + 1] === "\n") {
-      if (Object.values(counters).every((c) => c === 0)) {
-        const block_text = text.substring(block_start, i);
-        if (block_text) blocks.push([block_text, block_start]);
-        let j = i;
-        while (j < n && text[j] === "\n") j++;
-        i = j;
-        block_start = i;
-        continue;
+      if (["w:ins", "w:del", "w:hyperlink", "w:sdt", "w:smartTag", "w:fldSimple", "w:sdtContent"].includes(curr.tagName)) {
+        coalesceContainer(curr);
       }
-    }
-    i++;
-  }
-  if (block_start < n) {
-    const block_text = text.substring(block_start, n);
-    if (block_text) blocks.push([block_text, block_start]);
-  }
-  return blocks;
-}
-function _merge_footnote_sections(blocks) {
-  if (!blocks.length) return blocks;
-  const merged = [];
-  let i = 0;
-  while (i < blocks.length) {
-    const [block_text, block_offset] = blocks[i];
-    const stripped = block_text.trimStart();
-    const is_section_header = stripped.startsWith("## Footnotes") || stripped.startsWith("## Endnotes");
-    if (!is_section_header) {
-      merged.push([block_text, block_offset]);
       i++;
-      continue;
     }
-    let accumulated_text = block_text;
-    let j = i + 1;
-    while (j < blocks.length) {
-      const [next_text] = blocks[j];
-      const next_stripped = next_text.trimStart();
-      if (next_stripped.startsWith("[^fn-") || next_stripped.startsWith("[^en-")) {
-        accumulated_text = `${accumulated_text}
-
-${next_text}`;
-        j++;
-      } else {
-        break;
+    if (children.length > 0) {
+      const last = children[children.length - 1];
+      if (["w:ins", "w:del", "w:hyperlink", "w:sdt", "w:smartTag", "w:fldSimple", "w:sdtContent"].includes(last.tagName)) {
+        coalesceContainer(last);
       }
     }
-    merged.push([accumulated_text, block_offset]);
-    i = j;
   }
-  return merged;
+  const paragraphs = findAllDescendants(doc.element, "w:p");
+  for (const p of paragraphs) coalesceContainer(p);
+  return count ? [`Adjacent identical runs coalesced: ${count}`] : [];
 }
-function _assemble_pages(block_records) {
-  if (!block_records.length) return [[""], [0]];
-  const pages = [];
-  const page_starts = [];
-  let current_blocks = [];
-  let current_size = 0;
-  let current_start = -1;
-  const flush_current = () => {
-    if (current_blocks.length > 0) {
-      pages.push(current_blocks.join("\n\n"));
-      page_starts.push(current_start);
+function strip_rsid(doc) {
+  let count = 0;
+  const rsidAttrs = ["w:rsidR", "w:rsidRPr", "w:rsidRDefault", "w:rsidP", "w:rsidDel", "w:rsidSect", "w:rsidTr"];
+  const all = doc.element.getElementsByTagName("*");
+  for (let i = 0; i < all.length; i++) {
+    for (const attr of rsidAttrs) {
+      if (all[i].hasAttribute(attr)) {
+        all[i].removeAttribute(attr);
+        count++;
+      }
     }
-    current_blocks = [];
-    current_size = 0;
-    current_start = -1;
-  };
-  for (const [block_text, block_offset] of block_records) {
-    const block_size = block_text.length;
-    const added_size = block_size + (current_blocks.length > 0 ? 2 : 0);
-    if (current_blocks.length > 0 && current_size + added_size > PAGE_TARGET_CHARS) {
-      flush_current();
+  }
+  const rsidsElements = findAllDescendants(doc.element, "w:rsids");
+  for (const el of rsidsElements) {
+    if (el.parentNode) {
+      el.parentNode.removeChild(el);
+      count++;
     }
-    if (current_blocks.length === 0 && block_size > PAGE_TARGET_CHARS) {
-      pages.push(block_text);
-      page_starts.push(block_offset);
-      continue;
+  }
+  return count ? [`rsid attributes: ${count} removed`] : [];
+}
+function strip_para_ids(doc) {
+  let count = 0;
+  const attrs = ["w14:paraId", "w14:textId"];
+  const all = doc.element.getElementsByTagName("*");
+  for (let i = 0; i < all.length; i++) {
+    for (const attr of attrs) {
+      if (all[i].hasAttribute(attr)) {
+        all[i].removeAttribute(attr);
+        count++;
+      }
     }
-    if (current_blocks.length === 0) current_start = block_offset;
-    current_blocks.push(block_text);
-    current_size += current_size > 0 ? added_size : block_size;
   }
-  flush_current();
-  if (!pages.length) return [[""], [0]];
-  return [pages, page_starts];
+  return count ? [`Paragraph/text IDs: ${count} removed`] : [];
 }
-function _count_tracked_changes(page_content) {
-  const matches = [...page_content.matchAll(_CHG_ID_PATTERN)];
-  const distinct = new Set(matches.map((m) => m[1]));
-  return distinct.size;
+function strip_proof_errors(doc) {
+  const elements = findAllDescendants(doc.element, "w:proofErr");
+  elements.forEach((el) => el.parentNode?.removeChild(el));
+  return elements.length ? [`Spell check markers: ${elements.length} removed`] : [];
 }
-
-// src/domain.ts
-function boundedLevenshtein(a, b, maxDist = 2) {
-  if (a === b) return 0;
-  if (Math.abs(a.length - b.length) > maxDist) return maxDist + 1;
-  if (a.length === 0) return b.length <= maxDist ? b.length : maxDist + 1;
-  if (b.length === 0) return a.length <= maxDist ? a.length : maxDist + 1;
-  if (a.length > b.length) {
-    const temp = a;
-    a = b;
-    b = temp;
+function strip_empty_properties(doc) {
+  let count = 0;
+  for (const tag of ["w:rPr", "w:pPr"]) {
+    const elements = findAllDescendants(doc.element, tag);
+    for (const el of elements) {
+      if (el.childNodes.length === 0 || el.childNodes.length === 1 && el.childNodes[0].nodeType === 3 && !el.childNodes[0].textContent?.trim()) {
+        el.parentNode?.removeChild(el);
+        count++;
+      }
+    }
   }
-  let row = Array.from({ length: a.length + 1 }, (_, i) => i);
-  for (let i = 1; i <= b.length; i++) {
-    const newRow = [i];
-    let minInRow = i;
-    for (let j = 1; j <= a.length; j++) {
-      const cost = a[j - 1] === b[i - 1] ? 0 : 1;
-      const val = Math.min(
-        row[j] + 1,
-        newRow[j - 1] + 1,
-        row[j - 1] + cost
-      );
-      newRow.push(val);
-      if (val < minInRow) minInRow = val;
+  return count ? [`Empty property elements: ${count} removed`] : [];
+}
+function strip_hidden_text(doc) {
+  let count = 0;
+  const elements = findAllDescendants(doc.element, "w:rPr");
+  for (const rPr of elements) {
+    if (findChild(rPr, "w:vanish") || findChild(rPr, "w:webHidden")) {
+      const run = rPr.parentNode;
+      if (run && run.tagName === "w:r" && run.parentNode) {
+        run.parentNode.removeChild(run);
+        count++;
+      }
     }
-    if (minInRow > maxDist) return maxDist + 1;
-    row = newRow;
   }
-  return row[a.length] <= maxDist ? row[a.length] : maxDist + 1;
+  return count ? [`Hidden text runs: ${count} removed`] : [];
 }
-function _get_paragraph_text(p) {
-  let text = "";
-  const runs = findAllDescendants(p._element, "w:r");
-  for (const r of runs) {
-    text += get_run_text(new Run(r, p));
+function count_tracked_changes(doc) {
+  const ins = findAllDescendants(doc.element, "w:ins").length;
+  const del = findAllDescendants(doc.element, "w:del").length;
+  const fmt = findAllDescendants(doc.element, "w:rPrChange").length + findAllDescendants(doc.element, "w:pPrChange").length + findAllDescendants(doc.element, "w:sectPrChange").length;
+  return [ins, del, fmt];
+}
+function get_track_change_authors(doc) {
+  const authors = /* @__PURE__ */ new Set();
+  for (const tag of ["w:ins", "w:del", "w:rPrChange", "w:pPrChange", "w:sectPrChange"]) {
+    for (const el of findAllDescendants(doc.element, tag)) {
+      const author = el.getAttribute("w:author");
+      if (author) authors.add(author);
+    }
   }
-  return text;
+  return authors;
 }
-function extract_all_domain_metadata(doc, base_text) {
-  const definitions = {};
-  const duplicates = /* @__PURE__ */ new Set();
-  const raw_anchors = {};
-  const raw_references = [];
-  const leading_re = /^(?:[\d.\-()a-zA-Z]+\s*)?["“]([A-Z][A-Za-z0-9\s\-&'’]{1,60})["”]/;
-  const inline_re = /\([^)]*?["“]([A-Z][A-Za-z0-9\s\-&'’]{1,60})["”][^)]*?\)/g;
-  for (const item of iter_block_items(doc)) {
-    if (!(item instanceof Paragraph)) continue;
-    const text = _get_paragraph_text(item).trim();
-    if (!text) continue;
-    const extracted_terms = [];
-    const leading_match = text.match(leading_re);
-    if (leading_match) extracted_terms.push(leading_match[1].trim());
-    const inline_matches = text.matchAll(inline_re);
-    for (const m of inline_matches) {
-      extracted_terms.push(m[1].trim());
+function _getElementText(el) {
+  const texts = [];
+  const ts = findAllDescendants(el, "w:t");
+  for (const t of ts) if (t.textContent) texts.push(t.textContent);
+  const dts = findAllDescendants(el, "w:delText");
+  for (const dt of dts) if (dt.textContent) texts.push(dt.textContent);
+  return texts.join("");
+}
+function _truncate(text, maxLen = 60) {
+  const clean = text.replace(/\n/g, " ").trim();
+  if (clean.length <= maxLen) return clean;
+  return clean.substring(0, maxLen - 3) + "...";
+}
+function accept_all_tracked_changes(doc) {
+  const lines = [];
+  const insEls = findAllDescendants(doc.element, "w:ins");
+  const delEls = findAllDescendants(doc.element, "w:del");
+  for (const ins of insEls) {
+    const text = _getElementText(ins).trim();
+    if (text) lines.push(`  Accepted insertion: "${_truncate(text, 60)}"`);
+  }
+  for (const del of delEls) {
+    const text = _getElementText(del).trim();
+    if (text) lines.push(`  Accepted deletion of: "${_truncate(text, 60)}"`);
+  }
+  const engine = new RedlineEngine(doc);
+  engine.accept_all_revisions();
+  for (const tag of ["w:rPrChange", "w:pPrChange", "w:sectPrChange"]) {
+    for (const el of findAllDescendants(doc.element, tag)) {
+      el.parentNode?.removeChild(el);
     }
-    for (const term of extracted_terms) {
-      if (definitions[term]) duplicates.add(term);
-      else definitions[term] = { count: 0 };
+  }
+  const total = insEls.length + delEls.length;
+  if (total) {
+    return [`Tracked changes auto-accepted: ${total}`].concat(lines);
+  }
+  return [];
+}
+function get_comments_summary(doc) {
+  const data = extract_comments_data(doc.pkg);
+  const comments = [];
+  let openCount = 0;
+  let resolvedCount = 0;
+  for (const [cId, info] of Object.entries(data)) {
+    if (info.resolved) resolvedCount++;
+    else openCount++;
+    comments.push({ id: cId, ...info });
+  }
+  return { total: comments.length, open: openCount, resolved: resolvedCount, comments };
+}
+function remove_all_comments(doc) {
+  const data = extract_comments_data(doc.pkg);
+  const keys = Object.keys(data);
+  if (keys.length === 0) return [];
+  const lines = [];
+  const cm = new CommentsManager(doc);
+  for (const [cId, info] of Object.entries(data)) {
+    const status = info.resolved ? "[Resolved]" : "[Open]";
+    lines.push(`  ${status} "${_truncate(info.text || "", 60)}" (${info.author || "Unknown"})`);
+    cm.deleteComment(cId);
+  }
+  for (const tag of ["w:commentRangeStart", "w:commentRangeEnd"]) {
+    for (const el of findAllDescendants(doc.element, tag)) {
+      el.parentNode?.removeChild(el);
     }
-    const short_text = text.length > 60 ? text.substring(0, 60) + "..." : text;
-    const nodes = findAllDescendants(item._element, "*");
-    for (const node of nodes) {
-      if (node.tagName === "w:bookmarkStart") {
-        const b_name = node.getAttribute("w:name");
-        if (b_name && (!b_name.startsWith("_") || b_name.startsWith("_Ref"))) {
-          if (!raw_anchors[b_name]) {
-            raw_anchors[b_name] = { anchored_to: short_text, referenced_from: [] };
-          }
+  }
+  const refs = findAllDescendants(doc.element, "w:commentReference");
+  for (const ref of refs) {
+    const parent = ref.parentNode;
+    if (parent) {
+      if (parent.tagName === "w:r" || parent.tagName.endsWith(":r")) {
+        const nonRprChildren = Array.from(parent.childNodes).filter(
+          (c) => c.nodeType === 1 && c.tagName !== "w:rPr" && c.tagName !== "rPr"
+        );
+        if (nonRprChildren.length <= 1) {
+          parent.parentNode?.removeChild(parent);
+        } else {
+          parent.removeChild(ref);
         }
+      } else {
+        parent.removeChild(ref);
       }
-      let target = null;
-      if (node.tagName === "w:fldSimple") {
-        const instr = node.getAttribute("w:instr") || "";
-        const parts = instr.trim().split(/\s+/);
-        if (parts.length > 1 && parts[0] === "REF") target = parts[1];
-      } else if (node.tagName === "w:instrText") {
-        const instr = node.textContent || "";
-        const parts = instr.trim().split(/\s+/);
-        if (parts.length > 1 && parts[0] === "REF") target = parts[1];
-      }
-      if (target) raw_references.push([target, short_text]);
-    }
-  }
-  for (const [target, ref_text] of raw_references) {
-    if (raw_anchors[target]) {
-      raw_anchors[target].referenced_from.push(ref_text);
     }
   }
-  const diagnostics = [];
-  const def_keys = Object.keys(definitions);
-  if (def_keys.length > 0) {
-    const sorted_terms = def_keys.sort((a, b) => b.length - a.length);
-    const escapeRegExp = (str) => str.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
-    const alt = sorted_terms.map(escapeRegExp).join("|");
-    const usage_pattern = new RegExp(`(?<!["\u201C])\\b(${alt})\\b(?![\u201D"])`, "g");
-    for (const m of base_text.matchAll(usage_pattern)) {
-      const matched_term = m[1];
-      if (definitions[matched_term]) definitions[matched_term].count++;
+  const resolvedCount = Object.values(data).filter((c) => c.resolved).length;
+  const openCount = Object.values(data).filter((c) => !c.resolved).length;
+  return [`Comments removed: ${keys.length} (${resolvedCount} resolved, ${openCount} open)`].concat(lines);
+}
+function eject_comment_parts(doc) {
+  const pkg = doc.pkg;
+  const comment_partnames = /* @__PURE__ */ new Set();
+  for (const part of pkg.parts) {
+    if (part.partname.toLowerCase().includes("comments")) {
+      comment_partnames.add(part.partname);
+      const withSlash = part.partname.startsWith("/") ? part.partname : "/" + part.partname;
+      const withoutSlash = part.partname.startsWith("/") ? part.partname.substring(1) : part.partname;
+      comment_partnames.add(withSlash);
+      comment_partnames.add(withoutSlash);
     }
-    for (const term of def_keys) {
-      if (definitions[term].count === 0) {
-        delete definitions[term];
-        duplicates.delete(term);
+  }
+  if (comment_partnames.size === 0) return;
+  for (const part of pkg.parts) {
+    if (part.partname.endsWith(".rels")) {
+      const rels = findAllDescendants(part._element, "Relationship");
+      const toRemove = [];
+      for (const rel of rels) {
+        const target = rel.getAttribute("Target") || "";
+        if (target.toLowerCase().includes("comments")) {
+          toRemove.push(rel);
+          const sourcePath = part.partname.replace("/_rels/", "/").replace(".rels", "");
+          const sourcePart = pkg.getPartByPath(sourcePath);
+          if (sourcePart) {
+            const relId = rel.getAttribute("Id");
+            if (relId) sourcePart.rels.delete(relId);
+          }
+        }
+      }
+      for (const relEl of toRemove) {
+        relEl.parentNode?.removeChild(relEl);
       }
     }
   }
-  for (const term of duplicates) {
-    diagnostics.push(`[Error] Duplicate Definition: '${term}' is defined multiple times.`);
-  }
-  const stop_words = /* @__PURE__ */ new Set([
-    "The",
-    "This",
-    "That",
-    "Such",
-    "A",
-    "An",
-    "Any",
-    "All",
-    "Some",
-    "No",
-    "Every",
-    "Each",
-    "As",
-    "In",
-    "Of",
-    "For",
-    "To",
-    "On",
-    "By",
-    "With"
-  ]);
-  const all_cap_pattern = /\b[A-Z][a-zA-Z]*(?:\s+[A-Z][a-zA-Z]*)*\b/g;
-  const all_caps = new Set(base_text.match(all_cap_pattern) || []);
-  const valid_terms = new Set(Object.keys(definitions));
-  const terms_by_first_letter = {};
-  for (const term of valid_terms) {
-    const fl = term[0].toLowerCase();
-    if (!terms_by_first_letter[fl]) terms_by_first_letter[fl] = [];
-    terms_by_first_letter[fl].push(term);
-  }
-  const candidates_by_term = {};
-  for (const raw_candidate of all_caps) {
-    let candidate = raw_candidate.trim();
-    const words = candidate.split(/\s+/);
-    while (words.length > 0) {
-      const first = words[0];
-      const title = first.charAt(0).toUpperCase() + first.slice(1).toLowerCase();
-      if (stop_words.has(title)) words.shift();
-      else break;
-    }
-    candidate = words.join(" ");
-    if (candidate.length < 4) continue;
-    if (valid_terms.has(candidate)) continue;
-    const first_letter = candidate[0].toLowerCase();
-    let candidate_terms = terms_by_first_letter[first_letter] || [];
-    if (candidate.length > 5) {
-      for (const [k, v] of Object.entries(terms_by_first_letter)) {
-        if (k !== first_letter) candidate_terms = candidate_terms.concat(v);
+  const ctPart = pkg.getPartByPath("[Content_Types].xml");
+  if (ctPart) {
+    const overrides = findAllDescendants(ctPart._element, "Override");
+    const toRemove = [];
+    for (const override of overrides) {
+      const partName = override.getAttribute("PartName") || "";
+      if (comment_partnames.has(partName) || partName.toLowerCase().includes("comments")) {
+        toRemove.push(override);
       }
     }
-    for (const term of candidate_terms) {
-      if (Math.abs(candidate.length - term.length) > 2) continue;
-      if (candidate === term + "s" || candidate === term + "es") continue;
-      if (term === candidate + "s" || term === candidate + "es") continue;
-      const dist = boundedLevenshtein(candidate, term, 2);
-      if (dist === 0 || dist > 2) continue;
-      if (term.length <= 5) {
-        if (dist > 1) continue;
-        if (candidate[0].toLowerCase() !== term[0].toLowerCase()) continue;
-      }
-      if (!candidates_by_term[term]) candidates_by_term[term] = [];
-      if (!candidates_by_term[term].includes(candidate)) candidates_by_term[term].push(candidate);
+    for (const overrideEl of toRemove) {
+      overrideEl.parentNode?.removeChild(overrideEl);
     }
   }
-  for (const [term, candidates] of Object.entries(candidates_by_term)) {
-    candidates.sort();
-    const c_str = candidates.map((c) => `'${c}'`).join(", ");
-    diagnostics.push(`[Info] Possible Typos for '${term}': Found ${c_str}`);
-  }
-  function diag_sort_key(msg) {
-    if (msg.startsWith("[Error]")) return 0;
-    if (msg.startsWith("[Warning]")) return 1;
-    return 2;
+  pkg.parts = pkg.parts.filter((p) => !p.partname.toLowerCase().includes("comments"));
+  for (const key of Object.keys(pkg.unzipped)) {
+    if (key.toLowerCase().includes("comments")) {
+      delete pkg.unzipped[key];
+    }
   }
-  diagnostics.sort((a, b) => {
-    const keyA = diag_sort_key(a);
-    const keyB = diag_sort_key(b);
-    if (keyA !== keyB) return keyA - keyB;
-    return a.localeCompare(b);
-  });
-  return [definitions, diagnostics, raw_anchors];
 }
-function build_structural_appendix(doc, base_text) {
-  const [defs, diagnostics, anchors] = extract_all_domain_metadata(doc, base_text);
-  const lines = [
-    "\n\n---",
-    "",
-    "<!-- READONLY_BOUNDARY_START -->",
-    "# Document Structure (Read-Only)",
-    "The content below is metadata describing the document's reference structure. Do not include this section in any tracked changes or edits \u2014 it is for your context only and will be discarded on write."
-  ];
-  let has_content = false;
-  if (Object.keys(defs).length > 0) {
-    has_content = true;
-    lines.push("\n## Defined Terms");
-    for (const [term, data] of Object.entries(defs)) {
-      lines.push(`- "${term}" \u2014 used ${data.count} times.`);
+function replace_comment_authors(doc, newAuthor) {
+  const cm = new CommentsManager(doc);
+  if (!cm.commentsPart) return [];
+  const original = /* @__PURE__ */ new Set();
+  const comments = findAllDescendants(cm.commentsPart._element, "w:comment");
+  for (const c of comments) {
+    const author = c.getAttribute("w:author");
+    if (author) {
+      original.add(author);
+      c.setAttribute("w:author", newAuthor);
     }
-  }
-  if (diagnostics.length > 0) {
-    has_content = true;
-    lines.push("\n## Semantic Diagnostics");
-    for (const diag of diagnostics) {
-      lines.push(`- ${diag}`);
+    if (c.hasAttribute("w:initials")) {
+      const initials = newAuthor.split(" ").filter(Boolean).map((p) => p[0]).join("").toUpperCase();
+      c.setAttribute("w:initials", initials);
     }
   }
-  if (Object.keys(anchors).length > 0) {
-    has_content = true;
-    lines.push("\n## Named Anchors");
-    for (const [b_name, data] of Object.entries(anchors)) {
-      lines.push(`- ${b_name} \u2192 Anchored to: "${data.anchored_to}"`);
-      for (const ref of data.referenced_from) {
-        lines.push(`  - Referenced from: "${ref}"`);
+  return original.size ? [`Comment authors replaced: ${Array.from(original).sort().join(", ")} \u2192 "${newAuthor}"`] : [];
+}
+function replace_change_authors(doc, newAuthor) {
+  const original = /* @__PURE__ */ new Set();
+  for (const tag of ["w:ins", "w:del", "w:rPrChange", "w:pPrChange"]) {
+    for (const el of findAllDescendants(doc.element, tag)) {
+      const author = el.getAttribute("w:author");
+      if (author) {
+        original.add(author);
+        el.setAttribute("w:author", newAuthor);
       }
     }
   }
-  if (has_content) {
-    return lines.join("\n");
-  }
-  return "";
-}
-
-// src/ingest.ts
-async function extractTextFromBuffer(buffer, cleanView = false) {
-  const doc = await DocumentObject.load(buffer);
-  return _extractTextFromDoc(doc, cleanView);
+  return original.size ? [`Track change authors replaced: ${Array.from(original).sort().join(", ")} \u2192 "${newAuthor}"`] : [];
 }
-function _extractTextFromDoc(doc, cleanView = false, includeAppendix = true, return_paragraph_offsets = false) {
-  const comments_map = extract_comments_data(doc.pkg);
-  const full_text = [];
-  const paragraph_offsets = /* @__PURE__ */ new Map();
-  let cursor = 0;
-  for (const part of iter_document_parts(doc)) {
-    const part_cursor = full_text.length > 0 ? cursor + 2 : cursor;
-    const part_text = _extract_blocks(part, comments_map, cleanView, part_cursor, return_paragraph_offsets ? paragraph_offsets : void 0);
-    if (part_text) {
-      if (full_text.length > 0) cursor += 2;
-      full_text.push(part_text);
-      cursor += part_text.length;
+function normalize_change_dates(doc) {
+  let count = 0;
+  const fixed = "2025-01-01T00:00:00Z";
+  for (const tag of ["w:ins", "w:del", "w:rPrChange", "w:pPrChange"]) {
+    for (const el of findAllDescendants(doc.element, tag)) {
+      if (el.hasAttribute("w:date")) {
+        el.setAttribute("w:date", fixed);
+        count++;
+      }
     }
   }
-  let base_text = full_text.join("\n\n");
-  if (includeAppendix) {
-    const appendix = build_structural_appendix(doc, base_text);
-    if (appendix) base_text += appendix;
-  }
-  if (return_paragraph_offsets) {
-    return { text: base_text, paragraph_offsets };
-  }
-  return base_text;
+  return count ? [`Track change timestamps: ${count} normalized`] : [];
 }
-function _extract_blocks(container, comments_map, cleanView, cursor, paragraph_offsets) {
-  const part = container.part || container;
-  const [style_cache, default_pstyle] = _get_style_cache(part);
-  const blocks = [];
-  let local_cursor = cursor;
-  let is_first_block = true;
-  let is_first_para = true;
-  if (container.constructor && container.constructor.name === "NotesPart") {
-    const header = container.note_type === "fn" ? "## Footnotes" : "## Endnotes";
-    const sep = `---
-${header}`;
-    blocks.push(sep);
-    local_cursor += sep.length;
-    is_first_block = false;
-  }
-  for (const item of iter_block_items(container)) {
-    if (!is_first_block) local_cursor += 2;
-    const block_start = local_cursor;
-    if (item.constructor.name === "FootnoteItem") {
-      const fn_text = _extract_blocks(item, comments_map, cleanView, block_start, paragraph_offsets);
-      if (fn_text) {
-        blocks.push(fn_text);
-        local_cursor = block_start + fn_text.length;
-        is_first_block = false;
-      } else if (!is_first_block) {
-        local_cursor -= 2;
-      }
-    } else if (item instanceof Paragraph) {
-      let prefix = get_paragraph_prefix(item, style_cache, default_pstyle);
-      if (is_first_para && container.constructor.name === "FootnoteItem") {
-        prefix = `[^${container.note_type}-${container.id}]: ` + prefix;
+function scrub_doc_properties(doc) {
+  const lines = [];
+  const corePart = doc.pkg.getPartByPath("docProps/core.xml");
+  if (corePart) {
+    const creators = findDescendantsByLocalName(corePart._element, "creator");
+    creators.forEach((c) => {
+      if (c.textContent) {
+        lines.push(`Author: ${c.textContent}`);
+        c.textContent = "";
       }
-      const p_text = build_paragraph_text(item, comments_map, cleanView, style_cache, default_pstyle);
-      const full_block = prefix + p_text;
-      blocks.push(full_block);
-      if (paragraph_offsets) {
-        paragraph_offsets.set(item._element, [block_start, full_block.length]);
+    });
+    const modifiers = findDescendantsByLocalName(corePart._element, "lastModifiedBy");
+    modifiers.forEach((c) => {
+      if (c.textContent) {
+        lines.push(`Last modified by: ${c.textContent}`);
+        c.textContent = "";
       }
-      local_cursor = block_start + full_block.length;
-      is_first_para = false;
-      is_first_block = false;
-    } else if (item instanceof Table) {
-      const table_text = extract_table(item, comments_map, cleanView, block_start, paragraph_offsets);
-      if (table_text) {
-        blocks.push(table_text);
-        local_cursor = block_start + table_text.length;
-        is_first_block = false;
-      } else if (!is_first_block) {
-        local_cursor -= 2;
+    });
+    const revisions = findDescendantsByLocalName(corePart._element, "revision");
+    revisions.forEach((c) => {
+      if (c.textContent && parseInt(c.textContent) > 1) {
+        lines.push(`Revision count: ${c.textContent} \u2192 1`);
+        c.textContent = "1";
       }
-      is_first_para = false;
+    });
+  }
+  const appPart = doc.pkg.getPartByPath("docProps/app.xml");
+  if (appPart) {
+    const docEl = appPart._element;
+    const intFields = ["TotalTime", "Words", "Characters", "Paragraphs", "Lines", "CharactersWithSpaces"];
+    for (const f of intFields) {
+      findDescendantsByLocalName(docEl, f).forEach((el) => {
+        if (el.textContent && el.textContent !== "0") {
+          if (f === "TotalTime") lines.push(`Total editing time: ${el.textContent} minutes`);
+          el.textContent = "0";
+        }
+      });
+    }
+    const strFields = ["Template", "Manager", "Company"];
+    for (const f of strFields) {
+      findDescendantsByLocalName(docEl, f).forEach((el) => {
+        if (el.textContent) {
+          lines.push(`${f}: ${el.textContent}`);
+          el.textContent = "";
+        }
+      });
     }
   }
-  return blocks.join("\n\n");
+  return lines.length ? ["Metadata scrubbed:", ...lines.map((l) => `  ${l}`)] : [];
 }
-function extract_table(table, comments_map, cleanView, cursor, paragraph_offsets) {
-  const rows_text = [];
-  let rows_processed = 0;
-  let local_cursor = cursor;
-  for (const row of table.rows) {
-    const cell_texts = [];
-    const seen_cells = /* @__PURE__ */ new Set();
-    const trPr = findChild(row._element, "w:trPr");
-    const ins = trPr ? findChild(trPr, "w:ins") : null;
-    const del_node = trPr ? findChild(trPr, "w:del") : null;
-    if (cleanView && del_node) continue;
-    const row_start = local_cursor + (rows_processed > 0 ? 1 : 0);
-    const wrapper_prefix_len = !cleanView && ins ? 4 : !cleanView && del_node ? 4 : 0;
-    let cell_cursor = row_start + wrapper_prefix_len;
-    let first_cell = true;
-    for (const cell of row.cells) {
-      if (seen_cells.has(cell)) continue;
-      seen_cells.add(cell);
-      if (!first_cell) cell_cursor += 3;
-      const cell_content = _extract_blocks(cell, comments_map, cleanView, cell_cursor, paragraph_offsets);
-      cell_texts.push(cell_content);
-      cell_cursor += cell_content.length;
-      first_cell = false;
+function scrub_timestamps(doc) {
+  let modified = false;
+  const epoch = "1970-01-01T00:00:00Z";
+  const corePart = doc.pkg.getPartByPath("docProps/core.xml");
+  if (corePart) {
+    for (const tag of ["created", "modified", "lastPrinted"]) {
+      findDescendantsByLocalName(corePart._element, tag).forEach((el) => {
+        if (el.textContent && el.textContent !== epoch) {
+          el.textContent = epoch;
+          modified = true;
+        }
+      });
     }
-    let row_str = cell_texts.join(" | ");
-    if (!cleanView) {
-      if (ins) row_str = `{++ ${row_str} |Chg:${ins.getAttribute("w:id")}++}`;
-      else if (del_node) row_str = `{-- ${row_str} |Chg:${del_node.getAttribute("w:id")}--}`;
+  }
+  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) => {
+    const toRemove = [];
+    for (const rel of findAllDescendants(relsPart._element, "Relationship")) {
+      const target = rel.getAttribute("Target");
+      if (target && target.includes("customXml")) toRemove.push(rel);
     }
-    rows_text.push(row_str);
-    local_cursor = row_start + row_str.length;
-    rows_processed++;
+    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);
+  for (const sdtPr of findAllDescendants(doc.element, "w:sdtPr")) {
+    findChildren(sdtPr, "w:dataBinding").forEach((b) => sdtPr.removeChild(b));
   }
-  return rows_text.join("\n");
+  return [`Custom XML parts: ${customParts.length} removed`];
 }
-function build_paragraph_text(paragraph, comments_map, cleanView, style_cache, default_pstyle) {
-  const parts = [];
-  const active_ins = {};
-  const active_del = {};
-  const active_comments = /* @__PURE__ */ new Set();
-  const active_fmt = {};
-  const deferred_meta_states = [];
-  let pending_text = "";
-  let current_wrappers = ["", ""];
-  let current_style = ["", ""];
-  const items = Array.from(iter_paragraph_content(paragraph));
-  const is_heading = is_heading_paragraph(paragraph, style_cache, default_pstyle);
-  const native_heading = is_native_heading(paragraph, style_cache, default_pstyle);
-  let leading_strip_active = is_heading;
-  for (let i = 0; i < items.length; i++) {
-    const item = items[i];
-    if (item instanceof Run) {
-      const [prefix, suffix] = get_run_style_markers(item, native_heading);
-      const text = get_run_text(item);
-      if (cleanView && Object.keys(active_del).length > 0) continue;
-      if (leading_strip_active) {
-        if (!text || !text.trim()) continue;
-        leading_strip_active = false;
-      }
-      const seg = apply_formatting_to_segments(text, prefix, suffix);
-      if (seg) {
-        const new_wrappers = cleanView ? ["", ""] : _get_wrappers(active_ins, active_del, active_comments, active_fmt);
-        const new_style = [prefix, suffix];
-        if (pending_text && new_wrappers[0] === current_wrappers[0] && new_wrappers[1] === current_wrappers[1]) {
-          if (new_style[0] === current_style[0] && new_style[1] === current_style[1] && current_style[0] !== "" && pending_text.endsWith(current_style[1]) && seg.startsWith(new_style[0])) {
-            pending_text = pending_text.slice(0, -current_style[1].length) + seg.slice(new_style[0].length);
-          } else {
-            pending_text += seg;
-          }
-          current_style = new_style;
-        } else {
-          if (pending_text) parts.push(`${current_wrappers[0]}${pending_text}${current_wrappers[1]}`);
-          pending_text = seg;
-          current_wrappers = new_wrappers;
-          current_style = new_style;
-        }
-        if (!cleanView) {
-          const has_meta = Object.keys(active_ins).length > 0 || Object.keys(active_del).length > 0 || active_comments.size > 0 || Object.keys(active_fmt).length > 0;
-          if (has_meta) {
-            deferred_meta_states.push([{ ...active_ins }, { ...active_del }, new Set(active_comments), { ...active_fmt }]);
-          }
-          let should_defer = false;
-          const is_redline = Object.keys(active_ins).length > 0 || Object.keys(active_del).length > 0 || Object.keys(active_fmt).length > 0;
-          if (is_redline) {
-            let j = i + 1;
-            let next_is_redline = false;
-            let temp_ins = Object.keys(active_ins).length;
-            let temp_del = Object.keys(active_del).length;
-            let temp_fmt = Object.keys(active_fmt).length;
-            while (j < items.length) {
-              const next_item = items[j];
-              if (next_item instanceof Run) {
-                if (!get_run_text(next_item)) {
-                  j++;
-                  continue;
-                }
-                if (temp_ins > 0 || temp_del > 0 || temp_fmt > 0) next_is_redline = true;
-                break;
-              } else {
-                const ev = next_item;
-                if (ev.type === "ins_start") temp_ins++;
-                else if (ev.type === "ins_end") temp_ins = Math.max(0, temp_ins - 1);
-                else if (ev.type === "del_start") temp_del++;
-                else if (ev.type === "del_end") temp_del = Math.max(0, temp_del - 1);
-                else if (ev.type === "fmt_start") temp_fmt++;
-                else if (ev.type === "fmt_end") temp_fmt = Math.max(0, temp_fmt - 1);
-              }
-              j++;
-            }
-            if (next_is_redline) should_defer = true;
-          }
-          if (!should_defer && deferred_meta_states.length > 0) {
-            const meta_block = _build_merged_meta_block(deferred_meta_states, comments_map);
-            if (meta_block) {
-              if (pending_text) {
-                parts.push(`${current_wrappers[0]}${pending_text}${current_wrappers[1]}`);
-                pending_text = "";
-                current_wrappers = ["", ""];
-                current_style = ["", ""];
-              }
-              parts.push(`{>>${meta_block}<<}`);
-            }
-            deferred_meta_states.length = 0;
-          }
-        }
+function strip_image_alt_text(doc) {
+  let count = 0;
+  for (const docPr of findDescendantsByLocalName(doc.element, "docPr")) {
+    const descr = docPr.getAttribute("descr");
+    if (descr) {
+      const isShort = descr.length < 10;
+      const isFile = descr.includes(".") && descr.length < 60;
+      if (isShort || isFile) {
+        docPr.removeAttribute("descr");
+        count++;
       }
-    } else {
-      const ev = item;
-      leading_strip_active = false;
-      if (!["ins_start", "ins_end", "del_start", "del_end", "fmt_start", "fmt_end"].includes(ev.type)) {
-        if (pending_text) {
-          parts.push(`${current_wrappers[0]}${pending_text}${current_wrappers[1]}`);
-          pending_text = "";
-          current_wrappers = ["", ""];
-          current_style = ["", ""];
+    }
+  }
+  return count ? [`Image alt text: ${count} auto-generated descriptions removed`] : [];
+}
+function audit_hyperlinks(doc) {
+  const internal = ["sharepoint.com", "onedrive.com", ".internal", "intranet", "localhost", "10.", "192.168.", "172.16."];
+  const warnings = [];
+  const docRels = doc.pkg.getOrCreateRelsPart(doc.part.partname);
+  for (const rel of findAllDescendants(docRels._element, "Relationship")) {
+    if (rel.getAttribute("TargetMode") === "External") {
+      const url = rel.getAttribute("Target") || "";
+      for (const pattern of internal) {
+        if (url.toLowerCase().includes(pattern.toLowerCase())) {
+          warnings.push(`Hyperlink targets internal URL: ${_truncate(url, 80)}`);
+          break;
         }
       }
-      if (ev.type === "start") active_comments.add(ev.id);
-      else if (ev.type === "end") active_comments.delete(ev.id);
-      else if (ev.type === "ins_start") active_ins[ev.id] = ev;
-      else if (ev.type === "ins_end") delete active_ins[ev.id];
-      else if (ev.type === "del_start") active_del[ev.id] = ev;
-      else if (ev.type === "del_end") delete active_del[ev.id];
-      else if (ev.type === "fmt_start") active_fmt[ev.id] = ev;
-      else if (ev.type === "fmt_end") delete active_fmt[ev.id];
-      else if (ev.type === "footnote" || ev.type === "endnote") {
-        if (pending_text) {
-          parts.push(`${current_wrappers[0]}${pending_text}${current_wrappers[1]}`);
-          pending_text = "";
-          current_wrappers = ["", ""];
-          current_style = ["", ""];
-        }
-        parts.push(`[^${ev.type === "footnote" ? "fn" : "en"}-${ev.id}]`);
-      } else if (ev.type === "hyperlink_start") {
-        if (pending_text) {
-          parts.push(`${current_wrappers[0]}${pending_text}${current_wrappers[1]}`);
-          pending_text = "";
-          current_wrappers = ["", ""];
-          current_style = ["", ""];
-        }
-        parts.push("[");
-      } else if (ev.type === "hyperlink_end") {
-        if (pending_text) {
-          parts.push(`${current_wrappers[0]}${pending_text}${current_wrappers[1]}`);
-          pending_text = "";
-          current_wrappers = ["", ""];
-          current_style = ["", ""];
-        }
-        parts.push(`](${ev.date})`);
-      } else if (ev.type === "xref_start") {
-        if (pending_text) {
-          parts.push(`${current_wrappers[0]}${pending_text}${current_wrappers[1]}`);
-          pending_text = "";
-          current_wrappers = ["", ""];
-          current_style = ["", ""];
-        }
-        parts.push("[~");
-      } else if (ev.type === "xref_end") {
-        if (pending_text) {
-          parts.push(`${current_wrappers[0]}${pending_text}${current_wrappers[1]}`);
-          pending_text = "";
-          current_wrappers = ["", ""];
-          current_style = ["", ""];
-        }
-        parts.push(`~](#${ev.id})`);
-      } else if (ev.type === "bookmark") {
-        if (pending_text) {
-          parts.push(`${current_wrappers[0]}${pending_text}${current_wrappers[1]}`);
-          pending_text = "";
-          current_wrappers = ["", ""];
-          current_style = ["", ""];
-        }
-        parts.push(`{#${ev.id}}`);
-      }
     }
   }
-  if (pending_text) parts.push(`${current_wrappers[0]}${pending_text}${current_wrappers[1]}`);
-  if (deferred_meta_states.length > 0) {
-    const meta_block = _build_merged_meta_block(deferred_meta_states, comments_map);
-    if (meta_block) parts.push(`{>>${meta_block}<<}`);
+  return warnings;
+}
+
+// src/domain.ts
+function boundedLevenshtein(a, b, maxDist = 2) {
+  if (a === b) return 0;
+  if (Math.abs(a.length - b.length) > maxDist) return maxDist + 1;
+  if (a.length === 0) return b.length <= maxDist ? b.length : maxDist + 1;
+  if (b.length === 0) return a.length <= maxDist ? a.length : maxDist + 1;
+  if (a.length > b.length) {
+    const temp = a;
+    a = b;
+    b = temp;
   }
-  return parts.join("");
+  let row = Array.from({ length: a.length + 1 }, (_, i) => i);
+  for (let i = 1; i <= b.length; i++) {
+    const newRow = [i];
+    let minInRow = i;
+    for (let j = 1; j <= a.length; j++) {
+      const cost = a[j - 1] === b[i - 1] ? 0 : 1;
+      const val = Math.min(row[j] + 1, newRow[j - 1] + 1, row[j - 1] + cost);
+      newRow.push(val);
+      if (val < minInRow) minInRow = val;
+    }
+    if (minInRow > maxDist) return maxDist + 1;
+    row = newRow;
+  }
+  return row[a.length] <= maxDist ? row[a.length] : maxDist + 1;
 }
-function _get_wrappers(ins, del, comments, fmt) {
-  if (Object.keys(del).length > 0) return ["{--", "--}"];
-  if (Object.keys(ins).length > 0) return ["{++", "++}"];
-  if (comments.size > 0 || Object.keys(fmt).length > 0) return ["{==", "==}"];
-  return ["", ""];
+function _get_paragraph_text(p) {
+  let text = "";
+  const runs = findAllDescendants(p._element, "w:r");
+  for (const r of runs) {
+    text += get_run_text(new Run(r, p));
+  }
+  return text;
 }
-function _build_merged_meta_block(states_list, comments_map) {
-  const change_lines = [];
-  const comment_lines = [];
-  const seen_sigs = /* @__PURE__ */ new Set();
-  for (const [ins_map, del_map, comments_set, fmt_map] of states_list) {
-    let render_comment2 = function(cid) {
-      if (!comments_map[cid]) return;
-      const sig = `Com:${cid}`;
-      if (seen_sigs.has(sig)) return;
-      const data = comments_map[cid];
-      let header = `[${sig}] ${data.author}`;
-      if (data.date) header += ` @ ${data.date}`;
-      if (data.resolved) header += `(RESOLVED)`;
-      comment_lines.push(`${header}: ${data.text}`);
-      seen_sigs.add(sig);
-      if (children_map[cid]) {
-        const children = children_map[cid].sort((a, b) => (comments_map[a]?.date || "").localeCompare(comments_map[b]?.date || ""));
-        for (const child_id of children) {
-          render_comment2(child_id);
+function extract_all_domain_metadata(doc, base_text) {
+  const definitions = {};
+  const duplicates = /* @__PURE__ */ new Set();
+  const raw_anchors = {};
+  const raw_references = [];
+  const leading_re = /^(?:[\d.\-()a-zA-Z]+\s*)?["“]([A-Z][A-Za-z0-9\s\-&'’]{1,60})["”]/;
+  const inline_re = /\([^)]*?["“]([A-Z][A-Za-z0-9\s\-&'’]{1,60})["”][^)]*?\)/g;
+  for (const item of iter_block_items(doc)) {
+    if (!(item instanceof Paragraph)) continue;
+    const text = _get_paragraph_text(item).trim();
+    if (!text) continue;
+    const extracted_terms = [];
+    const leading_match = text.match(leading_re);
+    if (leading_match) extracted_terms.push(leading_match[1].trim());
+    const inline_matches = text.matchAll(inline_re);
+    for (const m of inline_matches) {
+      extracted_terms.push(m[1].trim());
+    }
+    for (const term of extracted_terms) {
+      if (definitions[term]) duplicates.add(term);
+      else definitions[term] = { count: 0 };
+    }
+    const short_text = text.length > 60 ? text.substring(0, 60) + "..." : text;
+    const nodes = findAllDescendants(item._element, "*");
+    for (const node of nodes) {
+      if (node.tagName === "w:bookmarkStart") {
+        const b_name = node.getAttribute("w:name");
+        if (b_name && (!b_name.startsWith("_") || b_name.startsWith("_Ref"))) {
+          if (!raw_anchors[b_name]) {
+            raw_anchors[b_name] = {
+              anchored_to: short_text,
+              referenced_from: []
+            };
+          }
         }
       }
-    };
-    var render_comment = render_comment2;
-    for (const [uid, meta] of Object.entries(ins_map)) {
-      const sig = `Chg:${uid}`;
-      if (!seen_sigs.has(sig)) {
-        change_lines.push(`[${sig} insert] ${meta.author || "Unknown"}`);
-        seen_sigs.add(sig);
+      let target = null;
+      if (node.tagName === "w:fldSimple") {
+        const instr = node.getAttribute("w:instr") || "";
+        const parts = instr.trim().split(/\s+/);
+        if (parts.length > 1 && parts[0] === "REF") target = parts[1];
+      } else if (node.tagName === "w:instrText") {
+        const instr = node.textContent || "";
+        const parts = instr.trim().split(/\s+/);
+        if (parts.length > 1 && parts[0] === "REF") target = parts[1];
       }
+      if (target) raw_references.push([target, short_text]);
     }
-    for (const [uid, meta] of Object.entries(del_map)) {
-      const sig = `Chg:${uid}`;
-      if (!seen_sigs.has(sig)) {
-        change_lines.push(`[${sig} delete] ${meta.author || "Unknown"}`);
-        seen_sigs.add(sig);
-      }
+  }
+  for (const [target, ref_text] of raw_references) {
+    if (raw_anchors[target]) {
+      raw_anchors[target].referenced_from.push(ref_text);
     }
-    for (const [uid, meta] of Object.entries(fmt_map)) {
-      const sig = `Chg:${uid}`;
-      if (!seen_sigs.has(sig)) {
-        change_lines.push(`[${sig} format] ${meta.author || "Unknown"}`);
-        seen_sigs.add(sig);
+  }
+  const diagnostics = [];
+  const def_keys = Object.keys(definitions);
+  if (def_keys.length > 0) {
+    const sorted_terms = def_keys.sort((a, b) => b.length - a.length);
+    const escapeRegExp = (str) => str.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
+    const alt = sorted_terms.map(escapeRegExp).join("|");
+    const usage_pattern = new RegExp(`(?<!["\u201C])\\b(${alt})\\b(?![\u201D"])`, "g");
+    for (const m of base_text.matchAll(usage_pattern)) {
+      const matched_term = m[1];
+      if (definitions[matched_term]) definitions[matched_term].count++;
+    }
+    for (const term of def_keys) {
+      if (definitions[term].count === 0) {
+        delete definitions[term];
+        duplicates.delete(term);
       }
     }
-    const children_map = {};
-    for (const [c_id, data] of Object.entries(comments_map)) {
-      const p_id = data.parent_id;
-      if (p_id) {
-        if (!children_map[p_id]) children_map[p_id] = [];
-        children_map[p_id].push(c_id);
+  }
+  for (const term of duplicates) {
+    diagnostics.push(
+      `[Error] Duplicate Definition: '${term}' is defined multiple times.`
+    );
+  }
+  const stop_words = /* @__PURE__ */ new Set([
+    "The",
+    "This",
+    "That",
+    "Such",
+    "A",
+    "An",
+    "Any",
+    "All",
+    "Some",
+    "No",
+    "Every",
+    "Each",
+    "As",
+    "In",
+    "Of",
+    "For",
+    "To",
+    "On",
+    "By",
+    "With"
+  ]);
+  const all_cap_pattern = /\b[A-Z][a-zA-Z]*(?:\s+[A-Z][a-zA-Z]*)*\b/g;
+  const all_caps = new Set(base_text.match(all_cap_pattern) || []);
+  const valid_terms = new Set(Object.keys(definitions));
+  const terms_by_first_letter = {};
+  for (const term of valid_terms) {
+    const fl = term[0].toLowerCase();
+    if (!terms_by_first_letter[fl]) terms_by_first_letter[fl] = [];
+    terms_by_first_letter[fl].push(term);
+  }
+  const candidates_by_term = {};
+  for (const raw_candidate of all_caps) {
+    let candidate = raw_candidate.trim();
+    const words = candidate.split(/\s+/);
+    while (words.length > 0) {
+      const first = words[0];
+      const title = first.charAt(0).toUpperCase() + first.slice(1).toLowerCase();
+      if (stop_words.has(title)) words.shift();
+      else break;
+    }
+    candidate = words.join(" ");
+    if (candidate.length < 4) continue;
+    if (valid_terms.has(candidate)) continue;
+    const first_letter = candidate[0].toLowerCase();
+    let candidate_terms = terms_by_first_letter[first_letter] || [];
+    if (candidate.length > 5) {
+      for (const [k, v] of Object.entries(terms_by_first_letter)) {
+        if (k !== first_letter) candidate_terms = candidate_terms.concat(v);
       }
     }
-    const sorted_ids = Array.from(comments_set).sort();
-    for (const c_id of sorted_ids) {
-      render_comment2(c_id);
+    for (const term of candidate_terms) {
+      if (Math.abs(candidate.length - term.length) > 2) continue;
+      if (candidate === term + "s" || candidate === term + "es") continue;
+      if (term === candidate + "s" || term === candidate + "es") continue;
+      const dist = boundedLevenshtein(candidate, term, 2);
+      if (dist === 0 || dist > 2) continue;
+      if (term.length <= 5) {
+        if (dist > 1) continue;
+        if (candidate[0].toLowerCase() !== term[0].toLowerCase()) continue;
+      }
+      if (!candidates_by_term[term]) candidates_by_term[term] = [];
+      if (!candidates_by_term[term].includes(candidate))
+        candidates_by_term[term].push(candidate);
     }
   }
-  return [...change_lines, ...comment_lines].join("\n");
-}
-
-// src/outline.ts
-var _HEADING_PREFIX_RE = /^(#{1,6}) /;
-var _HEURISTIC_MIN_WORDS = 3;
-function extract_outline(doc, projected_body, body_pages, body_page_offsets, paragraph_offsets = null) {
-  if (body_pages.length !== body_page_offsets.length) {
-    throw new Error("body_pages and body_page_offsets length mismatch");
+  for (const [term, candidates] of Object.entries(candidates_by_term)) {
+    candidates.sort();
+    const c_str = candidates.map((c) => `'${c}'`).join(", ");
+    diagnostics.push(`[Info] Possible Typos for '${term}': Found ${c_str}`);
   }
-  if (paragraph_offsets) {
-    return _extract_outline_fast(doc, projected_body, body_page_offsets, paragraph_offsets);
+  function diag_sort_key(msg) {
+    if (msg.startsWith("[Error]")) return 0;
+    if (msg.startsWith("[Warning]")) return 1;
+    return 2;
   }
-  const comments_map = extract_comments_data(doc.pkg);
-  const block_records = _walk_doc_body(doc, comments_map);
-  const heading_indices = [];
-  for (let idx = 0; idx < block_records.length; idx++) {
-    const rec = block_records[idx];
-    if (!(rec.is_paragraph && _is_heading(rec.item))) continue;
-    if (!_heading_passes_quality_filter(rec.item, comments_map)) continue;
-    heading_indices.push(idx);
+  diagnostics.sort((a, b) => {
+    const keyA = diag_sort_key(a);
+    const keyB = diag_sort_key(b);
+    if (keyA !== keyB) return keyA - keyB;
+    return a.localeCompare(b);
+  });
+  return [definitions, diagnostics, raw_anchors];
+}
+function extract_document_settings_warnings(doc) {
+  const warnings = [];
+  const settingsPart = doc.pkg.getPartByPath("word/settings.xml");
+  if (!settingsPart) return warnings;
+  const isTruthy = (el) => {
+    if (!el.hasAttribute("w:val")) return true;
+    const val = (el.getAttribute("w:val") || "").toLowerCase();
+    return val !== "0" && val !== "false" && val !== "off";
+  };
+  const removePersonal = findDescendantsByLocalName(
+    settingsPart._element,
+    "removePersonalInformation"
+  );
+  if (removePersonal.length > 0 && isTruthy(removePersonal[0])) {
+    warnings.push(
+      "[Warning] Privacy flag `removePersonalInformation` is enabled in word/settings.xml. Microsoft Word will strip the `w:author`, `w:initials`, and `w:date` attributes from every tracked change and comment the next time this document is opened and saved. Edits made by this agent will lose attribution, breaking audit trails and any multi-turn workflow that relies on identifying prior edits."
+    );
   }
-  if (heading_indices.length === 0) return [];
-  const nodes = [];
-  for (let h_pos = 0; h_pos < heading_indices.length; h_pos++) {
-    const rec_idx = heading_indices[h_pos];
-    const rec = block_records[rec_idx];
-    const paragraph = rec.item;
-    const level = _heading_level(paragraph);
-    const text = _heading_text(paragraph, comments_map);
-    const style = _determine_heading_style(paragraph);
-    const owned_end = _find_owned_end(
-      block_records,
-      heading_indices,
-      h_pos,
-      level
+  const removeDateTime = findDescendantsByLocalName(
+    settingsPart._element,
+    "removeDateAndTime"
+  );
+  if (removeDateTime.length > 0 && isTruthy(removeDateTime[0])) {
+    warnings.push(
+      "[Warning] Privacy flag `removeDateAndTime` is enabled in word/settings.xml. Microsoft Word will strip the `w:date` attribute from every tracked change and comment the next time this document is opened and saved. Timestamps on this agent's edits will be lost on the next Word save."
     );
-    const owned_blocks = block_records.slice(rec_idx + 1, owned_end);
-    const has_table = _direct_has_table(block_records, rec_idx + 1, owned_end);
-    const footnote_ids = _collect_footnote_ids(owned_blocks);
-    const page_num = _offset_to_page(rec.start_offset, body_page_offsets);
-    nodes.push({ level, text, page: page_num, style, has_table, footnote_ids });
   }
-  return nodes;
+  return warnings;
 }
-function _direct_has_table(block_records, range_start, range_end) {
-  for (let idx = range_start; idx < range_end; idx++) {
-    const rec = block_records[idx];
-    if (rec.is_paragraph && _is_heading(rec.item)) return false;
-    if (rec.is_table) return true;
+function build_structural_appendix(doc, base_text) {
+  const [defs, diagnostics, anchors] = extract_all_domain_metadata(
+    doc,
+    base_text
+  );
+  const settings_warnings = extract_document_settings_warnings(doc);
+  const lines = [
+    "\n\n---",
+    "",
+    "<!-- READONLY_BOUNDARY_START -->",
+    "# Document Structure (Read-Only)",
+    "The content below is metadata describing the document's reference structure. Do not include this section in any tracked changes or edits \u2014 it is for your context only and will be discarded on write."
+  ];
+  let has_content = false;
+  if (settings_warnings.length > 0) {
+    has_content = true;
+    lines.push("\n## Document Settings");
+    for (const warning of settings_warnings) {
+      lines.push(`- ${warning}`);
+    }
   }
-  return false;
-}
-function _walk_doc_body(doc, comments_map) {
-  const parts = Array.from(iter_document_parts(doc));
-  let body_start_offset = 0;
-  let body_part = null;
-  for (const part of parts) {
-    if (part === doc) {
-      body_part = part;
-      break;
+  if (Object.keys(defs).length > 0) {
+    has_content = true;
+    lines.push("\n## Defined Terms");
+    for (const [term, data] of Object.entries(defs)) {
+      lines.push(`- "${term}" \u2014 used ${data.count} times.`);
     }
-    const part_text = _project_part(part, comments_map);
+  }
+  if (diagnostics.length > 0) {
+    has_content = true;
+    lines.push("\n## Semantic Diagnostics");
+    for (const diag of diagnostics) {
+      lines.push(`- ${diag}`);
+    }
+  }
+  if (Object.keys(anchors).length > 0) {
+    has_content = true;
+    lines.push("\n## Named Anchors");
+    for (const [b_name, data] of Object.entries(anchors)) {
+      lines.push(`- ${b_name} \u2192 Anchored to: "${data.anchored_to}"`);
+      for (const ref of data.referenced_from) {
+        lines.push(`  - Referenced from: "${ref}"`);
+      }
+    }
+  }
+  if (has_content) {
+    return lines.join("\n");
+  }
+  return "";
+}
+
+// src/ingest.ts
+async function extractTextFromBuffer(buffer, cleanView = false) {
+  const doc = await DocumentObject.load(buffer);
+  return _extractTextFromDoc(doc, cleanView);
+}
+function _extractTextFromDoc(doc, cleanView = false, includeAppendix = true, return_paragraph_offsets = false) {
+  const comments_map = extract_comments_data(doc.pkg);
+  const full_text = [];
+  const paragraph_offsets = /* @__PURE__ */ new Map();
+  let cursor = 0;
+  for (const part of iter_document_parts(doc)) {
+    const part_cursor = full_text.length > 0 ? cursor + 2 : cursor;
+    const part_text = _extract_blocks(
+      part,
+      comments_map,
+      cleanView,
+      part_cursor,
+      return_paragraph_offsets ? paragraph_offsets : void 0
+    );
     if (part_text) {
-      if (body_start_offset > 0) body_start_offset += 2;
-      body_start_offset += part_text.length;
+      if (full_text.length > 0) cursor += 2;
+      full_text.push(part_text);
+      cursor += part_text.length;
     }
   }
-  if (!body_part) {
-    body_part = doc;
-    body_start_offset = 0;
-  } else {
-    if (body_start_offset > 0) body_start_offset += 2;
+  let base_text = full_text.join("\n\n");
+  if (includeAppendix) {
+    const appendix = build_structural_appendix(doc, base_text);
+    if (appendix) base_text += appendix;
   }
-  const records = [];
-  let cursor = body_start_offset;
+  if (return_paragraph_offsets) {
+    return { text: base_text, paragraph_offsets };
+  }
+  return base_text;
+}
+function _extract_blocks(container, comments_map, cleanView, cursor, paragraph_offsets) {
+  const part = container.part || container;
+  const [style_cache, default_pstyle] = _get_style_cache(part);
+  const blocks = [];
+  let local_cursor = cursor;
   let is_first_block = true;
-  for (const item of iter_block_items(body_part)) {
-    if (item instanceof Paragraph) {
-      const prefix = get_paragraph_prefix(item);
-      const p_text = build_paragraph_text(item, comments_map, false);
-      const block_len = (prefix + p_text).length;
-      if (!is_first_block) cursor += 2;
-      records.push({
+  let is_first_para = true;
+  if (container.constructor && container.constructor.name === "NotesPart") {
+    const header = container.note_type === "fn" ? "## Footnotes" : "## Endnotes";
+    const sep = `---
+${header}`;
+    blocks.push(sep);
+    local_cursor += sep.length;
+    is_first_block = false;
+  }
+  for (const item of iter_block_items(container)) {
+    if (!is_first_block) local_cursor += 2;
+    const block_start = local_cursor;
+    if (item.constructor.name === "FootnoteItem") {
+      const fn_text = _extract_blocks(
         item,
-        is_paragraph: true,
-        is_table: false,
-        start_offset: cursor,
-        projected_length: block_len
-      });
-      cursor += block_len;
+        comments_map,
+        cleanView,
+        block_start,
+        paragraph_offsets
+      );
+      if (fn_text) {
+        blocks.push(fn_text);
+        local_cursor = block_start + fn_text.length;
+        is_first_block = false;
+      } else if (!is_first_block) {
+        local_cursor -= 2;
+      }
+    } else if (item instanceof Paragraph) {
+      let prefix = get_paragraph_prefix(item, style_cache, default_pstyle);
+      if (is_first_para && container.constructor.name === "FootnoteItem") {
+        prefix = `[^${container.note_type}-${container.id}]: ` + prefix;
+      }
+      const p_text = build_paragraph_text(
+        item,
+        comments_map,
+        cleanView,
+        style_cache,
+        default_pstyle
+      );
+      const full_block = prefix + p_text;
+      blocks.push(full_block);
+      if (paragraph_offsets) {
+        paragraph_offsets.set(item._element, [block_start, full_block.length]);
+      }
+      local_cursor = block_start + full_block.length;
+      is_first_para = false;
       is_first_block = false;
     } else if (item instanceof Table) {
-      const table_text = extract_table(item, comments_map, false, 0);
-      const block_len = table_text ? table_text.length : 0;
-      if (!is_first_block) cursor += 2;
-      const table_start = cursor;
-      records.push({
+      const table_text = extract_table(
         item,
-        is_paragraph: false,
-        is_table: true,
-        start_offset: table_start,
-        projected_length: block_len
-      });
-      _record_table_inner_blocks_lite(item, table_start, records, comments_map);
-      cursor += block_len;
-      is_first_block = false;
+        comments_map,
+        cleanView,
+        block_start,
+        paragraph_offsets
+      );
+      if (table_text) {
+        blocks.push(table_text);
+        local_cursor = block_start + table_text.length;
+        is_first_block = false;
+      } else if (!is_first_block) {
+        local_cursor -= 2;
+      }
+      is_first_para = false;
     }
   }
-  return records;
+  return blocks.join("\n\n");
 }
-function _compute_inner_block_offset(table, target_paragraph, table_start_offset, comments_map) {
-  const target_el = target_paragraph._element;
-  let cursor = table_start_offset;
+function extract_table(table, comments_map, cleanView, cursor, paragraph_offsets) {
+  const rows_text = [];
   let rows_processed = 0;
+  let local_cursor = cursor;
   for (const row of table.rows) {
-    if (rows_processed > 0) cursor += 1;
+    const cell_texts = [];
     const seen_cells = /* @__PURE__ */ new Set();
-    let cells_in_row = 0;
+    const trPr = findChild(row._element, "w:trPr");
+    const ins = trPr ? findChild(trPr, "w:ins") : null;
+    const del_node = trPr ? findChild(trPr, "w:del") : null;
+    if (cleanView && del_node) continue;
+    const row_start = local_cursor + (rows_processed > 0 ? 1 : 0);
+    const wrapper_prefix_len = !cleanView && ins ? 4 : !cleanView && del_node ? 4 : 0;
+    let cell_cursor = row_start + wrapper_prefix_len;
+    let first_cell = true;
     for (const cell of row.cells) {
       if (seen_cells.has(cell)) continue;
       seen_cells.add(cell);
-      if (cells_in_row > 0) cursor += 3;
-      const [new_cursor, found] = _walk_cell_for_offset(
+      if (!first_cell) cell_cursor += 3;
+      const cell_content = _extract_blocks(
         cell,
-        target_el,
-        cursor,
-        comments_map
+        comments_map,
+        cleanView,
+        cell_cursor,
+        paragraph_offsets
       );
-      if (found) return new_cursor;
-      cursor = new_cursor;
-      cells_in_row++;
+      cell_texts.push(cell_content);
+      cell_cursor += cell_content.length;
+      first_cell = false;
+    }
+    let row_str = cell_texts.join(" | ");
+    if (!cleanView) {
+      if (ins) row_str = `{++ ${row_str} |Chg:${ins.getAttribute("w:id")}++}`;
+      else if (del_node)
+        row_str = `{-- ${row_str} |Chg:${del_node.getAttribute("w:id")}--}`;
     }
+    rows_text.push(row_str);
+    local_cursor = row_start + row_str.length;
     rows_processed++;
   }
-  return table_start_offset;
+  return rows_text.join("\n");
 }
-function _walk_cell_for_offset(cell, target_el, cell_start_cursor, comments_map) {
-  let cursor = cell_start_cursor;
-  let is_first_block = true;
-  for (const inner_item of iter_block_items(cell)) {
-    if (!is_first_block) cursor += 2;
-    if (inner_item instanceof Paragraph) {
-      if (inner_item._element === target_el) return [cursor, true];
-      const prefix = get_paragraph_prefix(inner_item);
-      const p_text = build_paragraph_text(inner_item, comments_map, false);
-      cursor += (prefix + p_text).length;
-    } else if (inner_item instanceof Table) {
-      const nested_offset = _compute_inner_block_offset(
-        inner_item,
-        new Paragraph(target_el, null),
-        cursor,
-        comments_map
-      );
-      if (nested_offset !== cursor) {
-        if (_element_is_descendant(target_el, inner_item._element))
-          return [nested_offset, true];
-      }
-      const table_text = extract_table(inner_item, comments_map, false, 0);
-      cursor += table_text ? table_text.length : 0;
-    }
-    is_first_block = false;
-  }
-  return [cursor, false];
-}
-function _element_is_descendant(target_el, ancestor_el) {
-  let cur = target_el.parentNode;
-  while (cur) {
-    if (cur === ancestor_el) return true;
-    cur = cur.parentNode;
-  }
-  return false;
-}
-function _record_table_inner_blocks_lite(table, inherited_offset, records, comments_map) {
-  const seen_cells = /* @__PURE__ */ new Set();
-  for (const row of table.rows) {
-    for (const cell of row.cells) {
-      if (seen_cells.has(cell)) continue;
-      seen_cells.add(cell);
-      for (const inner_item of iter_block_items(cell)) {
-        if (inner_item instanceof Paragraph) {
-          const true_offset = _is_heading(inner_item) ? _compute_inner_block_offset(
-            table,
-            inner_item,
-            inherited_offset,
-            comments_map
-          ) : inherited_offset;
-          records.push({
-            item: inner_item,
-            is_paragraph: true,
-            is_table: false,
-            start_offset: true_offset,
-            projected_length: 0
-          });
-        } else if (inner_item instanceof Table) {
-          records.push({
-            item: inner_item,
-            is_paragraph: false,
-            is_table: true,
-            start_offset: inherited_offset,
-            projected_length: 0
-          });
-          _record_table_inner_blocks_lite(
-            inner_item,
-            inherited_offset,
-            records,
-            comments_map
-          );
-        }
-      }
-    }
-  }
-}
-function _project_part(part, comments_map) {
-  const blocks = [];
-  const c_type = part.constructor.name;
-  if (c_type === "NotesPart") {
-    const header = part.note_type === "fn" ? "## Footnotes" : "## Endnotes";
-    blocks.push(`---
-${header}`);
-  }
-  let is_first_para = true;
-  for (const item of iter_block_items(part)) {
-    if (item.constructor.name === "FootnoteItem") {
-      const fn_text = _project_part(item, comments_map);
-      if (fn_text) blocks.push(fn_text);
-    } else if (item instanceof Paragraph) {
-      let prefix = get_paragraph_prefix(item);
-      if (is_first_para && c_type === "FootnoteItem")
-        prefix = `[^${part.note_type}-${part.id}]: ${prefix}`;
-      const p_text = build_paragraph_text(item, comments_map, false);
-      blocks.push(prefix + p_text);
-      is_first_para = false;
-    } else if (item instanceof Table) {
-      const table_text = extract_table(item, comments_map, false, 0);
-      if (table_text) blocks.push(table_text);
-      is_first_para = false;
-    }
-  }
-  return blocks.join("\n\n");
-}
-function _is_heading(paragraph) {
-  return _HEADING_PREFIX_RE.test(get_paragraph_prefix(paragraph));
-}
-function _heading_passes_quality_filter(paragraph, comments_map) {
-  const style = _determine_heading_style(paragraph);
-  if (style !== "(heuristic)") return true;
-  const text = _heading_text(paragraph, comments_map);
-  if (!text) return false;
-  const word_count = (text.match(/\w+/g) || []).length;
-  return word_count >= _HEURISTIC_MIN_WORDS;
-}
-function _heading_level(paragraph) {
-  const match = _HEADING_PREFIX_RE.exec(get_paragraph_prefix(paragraph));
-  return match ? Math.min(match[1].length, 6) : 1;
-}
-function _heading_text(paragraph, comments_map) {
-  const p_text = build_paragraph_text(paragraph, comments_map, false);
-  let cleaned = _strip_critic_markup(p_text);
-  cleaned = _strip_inline_formatting(cleaned);
-  return cleaned.trim();
-}
-function _strip_critic_markup(text) {
-  if (!text) return "";
-  text = text.replace(/\{--[\s\S]*?--\}/g, "");
-  text = text.replace(/\{>>[\s\S]*?<<\}/g, "");
-  text = text.replace(/\{\+\+([\s\S]*?)\+\+\}/g, "$1");
-  text = text.replace(/\{==([\s\S]*?)==\}/g, "$1");
-  return text;
-}
-function _strip_inline_formatting(text) {
-  if (!text) return "";
-  text = text.replace(/\*\*(.+?)\*\*/g, "$1");
-  text = text.replace(/__(.+?)__/g, "$1");
-  text = text.replace(/(?<!\w)_(\S(?:.*?\S)?)_(?!\w)/g, "$1");
-  return text;
-}
-function _determine_heading_style(paragraph) {
-  const [style_cache, default_pstyle] = _get_style_cache(
-    paragraph._parent.part || paragraph._parent
+function build_paragraph_text(paragraph, comments_map, cleanView, style_cache, default_pstyle) {
+  const parts = [];
+  const active_ins = {};
+  const active_del = {};
+  const active_comments = /* @__PURE__ */ new Set();
+  const active_fmt = {};
+  const deferred_meta_states = [];
+  let pending_text = "";
+  let current_wrappers = ["", ""];
+  let current_style = ["", ""];
+  const items = Array.from(iter_paragraph_content(paragraph));
+  const is_heading = is_heading_paragraph(
+    paragraph,
+    style_cache,
+    default_pstyle
   );
-  const pPr = findChild(paragraph._element, "w:pPr");
-  let style_id = default_pstyle;
-  if (pPr) {
-    const pStyle = findChild(pPr, "w:pStyle");
-    if (pStyle) style_id = pStyle.getAttribute("w:val") || default_pstyle;
-  }
-  let outline_level = null;
-  let outline_level_from_style = false;
-  if (pPr) {
-    const oLvl = findChild(pPr, "w:outlineLvl");
-    if (oLvl && /^\d+$/.test(oLvl.getAttribute("w:val") || "")) {
-      outline_level = parseInt(oLvl.getAttribute("w:val"), 10);
-    }
-  }
-  if (outline_level === null && style_id && style_cache && style_cache[style_id]) {
-    outline_level = style_cache[style_id].outline_level;
-    outline_level_from_style = true;
-  }
-  const style_name = style_id && style_cache && style_cache[style_id] ? style_cache[style_id].name : style_id;
-  let normalized_style_name = style_name;
-  if (normalized_style_name && typeof normalized_style_name === "string") {
-    if (normalized_style_name.toLowerCase().startsWith("heading")) {
-      normalized_style_name = normalized_style_name.replace(/^heading/i, "Heading");
-    } else if (normalized_style_name.toLowerCase() === "title") {
-      normalized_style_name = "Title";
-    }
-  }
-  if (outline_level_from_style && outline_level !== null) {
-    const is_heading_or_title = normalized_style_name && (normalized_style_name.startsWith("Heading") || normalized_style_name === "Title");
-    if (!is_heading_or_title) {
-      outline_level = null;
-    }
-  }
-  if (outline_level !== null && outline_level >= 0 && outline_level <= 8) {
-    if (normalized_style_name && (normalized_style_name.startsWith("Heading") || normalized_style_name === "Title")) {
-      return normalized_style_name;
-    }
-    return "(outline_level)";
-  }
-  if (normalized_style_name && (normalized_style_name.startsWith("Heading") || normalized_style_name === "Title"))
-    return normalized_style_name;
-  if (normalized_style_name && /Heading[ ]?([1-6])(?![0-9])/.test(normalized_style_name))
-    return normalized_style_name;
-  return "(heuristic)";
-}
-function _find_owned_end(block_records, heading_indices, current_h_pos, current_level) {
-  for (let next_h_pos = current_h_pos + 1; next_h_pos < heading_indices.length; next_h_pos++) {
-    const next_idx = heading_indices[next_h_pos];
-    if (_heading_level(block_records[next_idx].item) <= current_level)
-      return next_idx;
-  }
-  return block_records.length;
-}
-function _collect_footnote_ids(owned_blocks) {
-  const seen = /* @__PURE__ */ new Set();
-  const ordered = [];
-  for (const rec of owned_blocks) {
-    if (!rec.is_paragraph) continue;
-    for (const event of iter_paragraph_content(rec.item)) {
-      if (!("type" in event)) continue;
-      let fn_id = "";
-      if (event.type === "footnote") fn_id = `fn-${event.id}`;
-      else if (event.type === "endnote") fn_id = `en-${event.id}`;
-      else continue;
-      if (!seen.has(fn_id)) {
-        seen.add(fn_id);
-        ordered.push(fn_id);
+  const native_heading = is_native_heading(
+    paragraph,
+    style_cache,
+    default_pstyle
+  );
+  let leading_strip_active = is_heading;
+  for (let i = 0; i < items.length; i++) {
+    const item = items[i];
+    if (item instanceof Run) {
+      const [prefix, suffix] = get_run_style_markers(item, native_heading);
+      const text = get_run_text(item);
+      if (cleanView && Object.keys(active_del).length > 0) continue;
+      if (leading_strip_active) {
+        if (!text || !text.trim()) continue;
+        leading_strip_active = false;
       }
-    }
-  }
-  return ordered;
-}
-function _offset_to_page(offset, body_page_offsets) {
-  if (!body_page_offsets || body_page_offsets.length === 0) return 1;
-  let page = 1;
-  for (let i = 0; i < body_page_offsets.length; i++) {
-    if (offset >= body_page_offsets[i]) page = i + 1;
-    else break;
-  }
-  return page;
-}
-function _extract_outline_fast(doc, projected_body, body_page_offsets, paragraph_offsets) {
-  const paragraphs_and_tables = [];
-  const seen_cells = /* @__PURE__ */ new Set();
-  function walk(container) {
-    for (const item of iter_block_items(container)) {
-      const i_type = item.constructor.name;
-      if (i_type === "FootnoteItem") {
-        walk(item);
-      } else if (item instanceof Paragraph) {
-        paragraphs_and_tables.push(["p", item]);
-      } else if (item instanceof Table) {
-        paragraphs_and_tables.push(["t", item]);
-        for (const row of item.rows) {
-          for (const cell of row.cells) {
-            if (seen_cells.has(cell._element)) {
-              continue;
+      const seg = apply_formatting_to_segments(text, prefix, suffix);
+      if (seg) {
+        const new_wrappers = cleanView ? ["", ""] : _get_wrappers(active_ins, active_del, active_comments, active_fmt);
+        const new_style = [prefix, suffix];
+        if (pending_text && new_wrappers[0] === current_wrappers[0] && new_wrappers[1] === current_wrappers[1]) {
+          if (new_style[0] === current_style[0] && new_style[1] === current_style[1] && current_style[0] !== "" && pending_text.endsWith(current_style[1]) && seg.startsWith(new_style[0])) {
+            pending_text = pending_text.slice(0, -current_style[1].length) + seg.slice(new_style[0].length);
+          } else {
+            pending_text += seg;
+          }
+          current_style = new_style;
+        } else {
+          if (pending_text)
+            parts.push(
+              `${current_wrappers[0]}${pending_text}${current_wrappers[1]}`
+            );
+          pending_text = seg;
+          current_wrappers = new_wrappers;
+          current_style = new_style;
+        }
+        if (!cleanView) {
+          const has_meta = Object.keys(active_ins).length > 0 || Object.keys(active_del).length > 0 || active_comments.size > 0 || Object.keys(active_fmt).length > 0;
+          if (has_meta) {
+            deferred_meta_states.push([
+              { ...active_ins },
+              { ...active_del },
+              new Set(active_comments),
+              { ...active_fmt }
+            ]);
+          }
+          let should_defer = false;
+          const has_any_meta = Object.keys(active_ins).length > 0 || Object.keys(active_del).length > 0 || Object.keys(active_fmt).length > 0 || active_comments.size > 0;
+          if (has_any_meta) {
+            let j = i + 1;
+            let next_has_meta = false;
+            let temp_ins = Object.keys(active_ins).length;
+            let temp_del = Object.keys(active_del).length;
+            let temp_fmt = Object.keys(active_fmt).length;
+            const temp_comments = new Set(active_comments);
+            while (j < items.length) {
+              const next_item = items[j];
+              if (next_item instanceof Run) {
+                if (!get_run_text(next_item)) {
+                  j++;
+                  continue;
+                }
+                if (temp_ins > 0 || temp_del > 0 || temp_fmt > 0 || temp_comments.size > 0)
+                  next_has_meta = true;
+                break;
+              } else {
+                const ev = next_item;
+                if (ev.type === "ins_start") temp_ins++;
+                else if (ev.type === "ins_end")
+                  temp_ins = Math.max(0, temp_ins - 1);
+                else if (ev.type === "del_start") temp_del++;
+                else if (ev.type === "del_end")
+                  temp_del = Math.max(0, temp_del - 1);
+                else if (ev.type === "fmt_start") temp_fmt++;
+                else if (ev.type === "fmt_end")
+                  temp_fmt = Math.max(0, temp_fmt - 1);
+                else if (ev.type === "start") temp_comments.add(ev.id);
+                else if (ev.type === "end") temp_comments.delete(ev.id);
+              }
+              j++;
             }
-            seen_cells.add(cell._element);
-            walk(cell);
+            if (next_has_meta) should_defer = true;
+          }
+          if (!should_defer && deferred_meta_states.length > 0) {
+            const meta_block = _build_merged_meta_block(
+              deferred_meta_states,
+              comments_map
+            );
+            if (meta_block) {
+              if (pending_text) {
+                parts.push(
+                  `${current_wrappers[0]}${pending_text}${current_wrappers[1]}`
+                );
+                pending_text = "";
+                current_wrappers = ["", ""];
+                current_style = ["", ""];
+              }
+              parts.push(`{>>${meta_block}<<}`);
+            }
+            deferred_meta_states.length = 0;
           }
         }
       }
-    }
-  }
-  walk(doc);
-  const heading_indices = [];
-  for (let idx = 0; idx < paragraphs_and_tables.length; idx++) {
-    const [kind, item] = paragraphs_and_tables[idx];
-    if (kind !== "p") continue;
-    let hasOffset = false;
-    if (paragraph_offsets instanceof Map) {
-      hasOffset = paragraph_offsets.has(item._element);
     } else {
-      hasOffset = item._element in paragraph_offsets;
-    }
-    if (!hasOffset) {
-      continue;
-    }
-    if (!_is_heading(item)) continue;
-    if (!_heading_passes_quality_filter_fast(item, projected_body, paragraph_offsets)) continue;
-    heading_indices.push(idx);
-  }
-  if (heading_indices.length === 0) return [];
-  const nodes = [];
-  for (let h_pos = 0; h_pos < heading_indices.length; h_pos++) {
-    const item_idx = heading_indices[h_pos];
-    const paragraph = paragraphs_and_tables[item_idx][1];
-    const level = _heading_level(paragraph);
-    const text = _heading_text_fast(paragraph, projected_body, paragraph_offsets);
-    const style = _determine_heading_style(paragraph);
-    let owned_end = item_idx;
-    for (let next_h_pos = h_pos + 1; next_h_pos < heading_indices.length; next_h_pos++) {
-      const next_idx = heading_indices[next_h_pos];
-      const next_paragraph = paragraphs_and_tables[next_idx][1];
-      if (_heading_level(next_paragraph) <= level) {
-        owned_end = next_idx;
-        break;
-      }
-    }
-    if (owned_end === item_idx) {
-      owned_end = paragraphs_and_tables.length;
-    }
-    const owned = paragraphs_and_tables.slice(item_idx + 1, owned_end);
-    let has_table = false;
-    for (const [kind2, item2] of owned) {
-      if (kind2 === "p" && _is_heading(item2)) {
-        break;
+      const ev = item;
+      leading_strip_active = false;
+      if (![
+        "ins_start",
+        "ins_end",
+        "del_start",
+        "del_end",
+        "fmt_start",
+        "fmt_end"
+      ].includes(ev.type)) {
+        if (pending_text) {
+          parts.push(
+            `${current_wrappers[0]}${pending_text}${current_wrappers[1]}`
+          );
+          pending_text = "";
+          current_wrappers = ["", ""];
+          current_style = ["", ""];
+        }
       }
-      if (kind2 === "t") {
-        has_table = true;
-        break;
+      if (ev.type === "start") active_comments.add(ev.id);
+      else if (ev.type === "end") active_comments.delete(ev.id);
+      else if (ev.type === "ins_start") active_ins[ev.id] = ev;
+      else if (ev.type === "ins_end") delete active_ins[ev.id];
+      else if (ev.type === "del_start") active_del[ev.id] = ev;
+      else if (ev.type === "del_end") delete active_del[ev.id];
+      else if (ev.type === "fmt_start") active_fmt[ev.id] = ev;
+      else if (ev.type === "fmt_end") delete active_fmt[ev.id];
+      else if (ev.type === "footnote" || ev.type === "endnote") {
+        if (pending_text) {
+          parts.push(
+            `${current_wrappers[0]}${pending_text}${current_wrappers[1]}`
+          );
+          pending_text = "";
+          current_wrappers = ["", ""];
+          current_style = ["", ""];
+        }
+        parts.push(`[^${ev.type === "footnote" ? "fn" : "en"}-${ev.id}]`);
+      } else if (ev.type === "hyperlink_start") {
+        if (pending_text) {
+          parts.push(
+            `${current_wrappers[0]}${pending_text}${current_wrappers[1]}`
+          );
+          pending_text = "";
+          current_wrappers = ["", ""];
+          current_style = ["", ""];
+        }
+        parts.push("[");
+      } else if (ev.type === "hyperlink_end") {
+        if (pending_text) {
+          parts.push(
+            `${current_wrappers[0]}${pending_text}${current_wrappers[1]}`
+          );
+          pending_text = "";
+          current_wrappers = ["", ""];
+          current_style = ["", ""];
+        }
+        parts.push(`](${ev.date})`);
+      } else if (ev.type === "xref_start") {
+        if (pending_text) {
+          parts.push(
+            `${current_wrappers[0]}${pending_text}${current_wrappers[1]}`
+          );
+          pending_text = "";
+          current_wrappers = ["", ""];
+          current_style = ["", ""];
+        }
+        parts.push("[~");
+      } else if (ev.type === "xref_end") {
+        if (pending_text) {
+          parts.push(
+            `${current_wrappers[0]}${pending_text}${current_wrappers[1]}`
+          );
+          pending_text = "";
+          current_wrappers = ["", ""];
+          current_style = ["", ""];
+        }
+        parts.push(`~](#${ev.id})`);
+      } else if (ev.type === "bookmark") {
+        if (pending_text) {
+          parts.push(
+            `${current_wrappers[0]}${pending_text}${current_wrappers[1]}`
+          );
+          pending_text = "";
+          current_wrappers = ["", ""];
+          current_style = ["", ""];
+        }
+        parts.push(`{#${ev.id}}`);
       }
     }
-    const footnote_ids = _collect_footnote_ids_fast(owned);
-    let para_offset;
-    if (paragraph_offsets instanceof Map) {
-      para_offset = paragraph_offsets.get(paragraph._element);
-    } else {
-      para_offset = paragraph_offsets[paragraph._element];
-    }
-    let page_num = 1;
-    if (para_offset !== void 0) {
-      const [start_offset] = para_offset;
-      page_num = _offset_to_page(start_offset, body_page_offsets);
-    }
-    nodes.push({
-      level,
-      text,
-      page: page_num,
-      style,
-      has_table,
-      footnote_ids
-    });
-  }
-  return nodes;
-}
-function _heading_passes_quality_filter_fast(paragraph, projected_body, paragraph_offsets) {
-  const style = _determine_heading_style(paragraph);
-  if (style !== "(heuristic)") return true;
-  const text = _heading_text_fast(paragraph, projected_body, paragraph_offsets);
-  if (!text) return false;
-  const words = text.match(/\w+/g) || [];
-  return words.length >= _HEURISTIC_MIN_WORDS;
-}
-function _heading_text_fast(paragraph, projected_body, paragraph_offsets) {
-  let offset;
-  if (paragraph_offsets instanceof Map) {
-    offset = paragraph_offsets.get(paragraph._element);
-  } else {
-    offset = paragraph_offsets[paragraph._element];
   }
-  if (offset === void 0) {
-    return "";
+  if (pending_text)
+    parts.push(`${current_wrappers[0]}${pending_text}${current_wrappers[1]}`);
+  if (deferred_meta_states.length > 0) {
+    const meta_block = _build_merged_meta_block(
+      deferred_meta_states,
+      comments_map
+    );
+    if (meta_block) parts.push(`{>>${meta_block}<<}`);
   }
-  const [start, length] = offset;
-  const raw = projected_body.substring(start, start + length);
-  let cleaned = _strip_critic_markup(raw);
-  cleaned = _strip_inline_formatting(cleaned);
-  cleaned = cleaned.replace(/^#+\s+/, "");
-  return cleaned.trim();
+  return parts.join("");
 }
-function _collect_footnote_ids_fast(owned_items) {
-  const seen = /* @__PURE__ */ new Set();
-  const ordered = [];
-  for (const [kind, item] of owned_items) {
-    if (kind !== "p") continue;
-    for (const event of iter_paragraph_content(item)) {
-      if (!("type" in event)) continue;
-      let fn_id = "";
-      if (event.type === "footnote") fn_id = `fn-${event.id}`;
-      else if (event.type === "endnote") fn_id = `en-${event.id}`;
-      else continue;
-      if (!seen.has(fn_id)) {
-        seen.add(fn_id);
-        ordered.push(fn_id);
-      }
-    }
-  }
-  return ordered;
+function _get_wrappers(ins, del, comments, fmt) {
+  if (Object.keys(del).length > 0) return ["{--", "--}"];
+  if (Object.keys(ins).length > 0) return ["{++", "++}"];
+  if (comments.size > 0 || Object.keys(fmt).length > 0) return ["{==", "==}"];
+  return ["", ""];
 }
-
-// src/sanitize/report.ts
-var SanitizeReport = class {
-  filename;
-  mode;
-  author;
-  tracked_changes_found = 0;
-  tracked_changes_accepted = 0;
-  tracked_changes_kept = 0;
-  change_lines = [];
-  comments_removed = 0;
-  comments_kept = 0;
-  removed_comment_lines = [];
-  kept_comment_lines = [];
-  metadata_lines = [];
-  structural_lines = [];
-  warnings = [];
-  status = "clean";
-  blocked_reason = null;
-  constructor(filename, mode = "full", author = null) {
-    this.filename = filename;
-    this.mode = mode;
-    this.author = author;
-  }
-  add_transform_lines(lines) {
-    for (const line of lines) {
-      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("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);
-      } else {
-        this.structural_lines.push(line);
-      }
-    }
-  }
-  render() {
-    const sep = "\u2550".repeat(50);
-    const lines = [sep, `Finalization Report: ${this.filename}`];
-    const flags = [];
-    if (this.mode === "keep-markup") flags.push("--keep-markup");
-    if (this.author) flags.push(`--author "${this.author}"`);
-    if (this.tracked_changes_accepted > 0) flags.push("--accept-all");
-    if (flags.length > 0) lines.push(flags.join(" "));
-    lines.push(sep);
-    if (this.status === "blocked") {
-      lines.push("");
-      lines.push(`BLOCKED: ${this.blocked_reason}`);
-      lines.push(sep);
-      return lines.join("\n");
-    }
-    if (this.mode === "keep-markup" && (this.tracked_changes_kept > 0 || this.comments_kept > 0)) {
-      lines.push("");
-      lines.push("VISIBLE TO COUNTERPARTY");
-      if (this.tracked_changes_kept > 0) lines.push(`  Tracked changes: ${this.tracked_changes_kept}`);
-      if (this.comments_kept > 0) {
-        lines.push(`  Open comments: ${this.comments_kept}`);
-        for (const cl of this.kept_comment_lines) lines.push(`    ${cl}`);
+function _build_merged_meta_block(states_list, comments_map) {
+  const change_lines = [];
+  const comment_lines = [];
+  const seen_sigs = /* @__PURE__ */ new Set();
+  for (const [ins_map, del_map, comments_set, fmt_map] of states_list) {
+    let render_comment2 = function(cid) {
+      if (!comments_map[cid]) return;
+      const sig = `Com:${cid}`;
+      if (seen_sigs.has(sig)) return;
+      const data = comments_map[cid];
+      let header = `[${sig}] ${data.author}`;
+      if (data.date) header += ` @ ${data.date}`;
+      if (data.resolved) header += `(RESOLVED)`;
+      comment_lines.push(`${header}: ${data.text}`);
+      seen_sigs.add(sig);
+      if (children_map[cid]) {
+        const children = children_map[cid].sort(
+          (a, b) => (comments_map[a]?.date || "").localeCompare(
+            comments_map[b]?.date || ""
+          )
+        );
+        for (const child_id of children) {
+          render_comment2(child_id);
+        }
+      }
+    };
+    var render_comment = render_comment2;
+    for (const [uid, meta] of Object.entries(
+      ins_map
+    )) {
+      const sig = `Chg:${uid}`;
+      if (!seen_sigs.has(sig)) {
+        change_lines.push(`[${sig} insert] ${meta.author || "Unknown"}`);
+        seen_sigs.add(sig);
       }
-      if (this.author) lines.push(`  Author on all markup: "${this.author}"`);
-    }
-    if (this.change_lines.length > 0) {
-      lines.push("");
-      lines.push("TRACKED CHANGES");
-      for (const cl of this.change_lines) lines.push(`  ${cl}`);
-    }
-    if (this.removed_comment_lines.length > 0) {
-      lines.push("");
-      lines.push("COMMENTS (stripped)");
-      for (const cl of this.removed_comment_lines) lines.push(`  ${cl}`);
     }
-    if (this.metadata_lines.length > 0) {
-      lines.push("");
-      lines.push("METADATA");
-      for (const ml of this.metadata_lines) lines.push(`  ${ml}`);
+    for (const [uid, meta] of Object.entries(
+      del_map
+    )) {
+      const sig = `Chg:${uid}`;
+      if (!seen_sigs.has(sig)) {
+        change_lines.push(`[${sig} delete] ${meta.author || "Unknown"}`);
+        seen_sigs.add(sig);
+      }
     }
-    if (this.structural_lines.length > 0) {
-      lines.push("");
-      lines.push("STRUCTURAL & PROTECTION");
-      for (const sl of this.structural_lines) lines.push(`  ${sl}`);
+    for (const [uid, meta] of Object.entries(
+      fmt_map
+    )) {
+      const sig = `Chg:${uid}`;
+      if (!seen_sigs.has(sig)) {
+        change_lines.push(`[${sig} format] ${meta.author || "Unknown"}`);
+        seen_sigs.add(sig);
+      }
     }
-    if (this.warnings.length > 0) {
-      lines.push("");
-      lines.push("WARNINGS");
-      for (const w of this.warnings) lines.push(`  \u26A0 ${w}`);
+    const children_map = {};
+    for (const [c_id, data] of Object.entries(
+      comments_map
+    )) {
+      const p_id = data.parent_id;
+      if (p_id) {
+        if (!children_map[p_id]) children_map[p_id] = [];
+        children_map[p_id].push(c_id);
+      }
     }
-    lines.push("");
-    lines.push(sep);
-    if (this.warnings.length > 0) {
-      lines.push(`Result: CLEAN WITH WARNINGS (${this.warnings.length} warning${this.warnings.length > 1 ? "s" : ""})`);
-    } else {
-      lines.push(`Result: CLEAN (${this.tracked_changes_found} changes resolved, ${this.comments_removed} comments removed)`);
+    const sorted_ids = Array.from(comments_set).sort();
+    for (const c_id of sorted_ids) {
+      render_comment2(c_id);
     }
-    lines.push(sep);
-    return lines.join("\n");
   }
-};
+  return [...change_lines, ...comment_lines].join("\n");
+}
 
-// src/sanitize/transforms.ts
-function findDescendantsByLocalName(element, localName) {
-  const result = [];
-  const all = element.getElementsByTagName("*");
-  for (let i = 0; i < all.length; i++) {
-    const tag = all[i].tagName;
-    if (tag === localName || tag.endsWith(":" + localName)) {
-      result.push(all[i]);
-    }
+// src/outline.ts
+var _HEADING_PREFIX_RE = /^(#{1,6}) /;
+var _HEURISTIC_MIN_WORDS = 3;
+function extract_outline(doc, projected_body, body_pages, body_page_offsets, paragraph_offsets = null) {
+  if (body_pages.length !== body_page_offsets.length) {
+    throw new Error("body_pages and body_page_offsets length mismatch");
   }
-  return result;
-}
-function coalesce_runs(doc) {
-  let count = 0;
-  function areRunsIdentical(rPr1, rPr2) {
-    const xml1 = rPr1 ? rPr1.toString() : "";
-    const xml2 = rPr2 ? rPr2.toString() : "";
-    return xml1 === xml2;
+  if (paragraph_offsets) {
+    return _extract_outline_fast(doc, projected_body, body_page_offsets, paragraph_offsets);
   }
-  function hasSpecialContent(run) {
-    const safeTags = ["w:t", "w:tab", "w:br", "w:cr", "w:delText", "w:rPr"];
-    for (let i = 0; i < run.childNodes.length; i++) {
-      const child = run.childNodes[i];
-      if (child.nodeType === 1) {
-        const tag = child.tagName;
-        if (!safeTags.includes(tag)) return true;
-      }
-    }
-    return false;
+  const comments_map = extract_comments_data(doc.pkg);
+  const block_records = _walk_doc_body(doc, comments_map);
+  const heading_indices = [];
+  for (let idx = 0; idx < block_records.length; idx++) {
+    const rec = block_records[idx];
+    if (!(rec.is_paragraph && _is_heading(rec.item))) continue;
+    if (!_heading_passes_quality_filter(rec.item, comments_map)) continue;
+    heading_indices.push(idx);
   }
-  function coalesceContainer(container) {
-    const children = Array.from(container.childNodes).filter((n) => n.nodeType === 1);
-    let i = 0;
-    while (i < children.length - 1) {
-      const curr = children[i];
-      const nxt = children[i + 1];
-      if (curr.tagName === "w:r" && nxt.tagName === "w:r") {
-        if (!hasSpecialContent(curr) && !hasSpecialContent(nxt)) {
-          const rPr1 = findChild(curr, "w:rPr");
-          const rPr2 = findChild(nxt, "w:rPr");
-          if (areRunsIdentical(rPr1, rPr2)) {
-            let last_t = null;
-            for (let c = 0; c < curr.childNodes.length; c++) {
-              const child = curr.childNodes[c];
-              if (child.nodeType === 1 && (child.tagName === "w:t" || child.tagName === "w:delText")) {
-                last_t = child;
-              }
-            }
-            const nxtChildren = Array.from(nxt.childNodes).filter((n) => n.nodeType === 1);
-            for (const child of nxtChildren) {
-              if (child.tagName === "w:rPr") continue;
-              if ((child.tagName === "w:t" || child.tagName === "w:delText") && last_t && last_t.tagName === child.tagName) {
-                const t1 = last_t.textContent || "";
-                const t2 = child.textContent || "";
-                const combined = t1 + t2;
-                last_t.textContent = combined;
-                if (combined.trim() !== combined) {
-                  last_t.setAttribute("xml:space", "preserve");
-                }
-              } else {
-                curr.appendChild(child);
-                if (child.tagName === "w:t" || child.tagName === "w:delText") {
-                  last_t = child;
-                }
-              }
-            }
-            container.removeChild(nxt);
-            children.splice(i + 1, 1);
-            count++;
-            continue;
-          }
-        }
-      }
-      if (["w:ins", "w:del", "w:hyperlink", "w:sdt", "w:smartTag", "w:fldSimple", "w:sdtContent"].includes(curr.tagName)) {
-        coalesceContainer(curr);
-      }
-      i++;
+  if (heading_indices.length === 0) return [];
+  const nodes = [];
+  for (let h_pos = 0; h_pos < heading_indices.length; h_pos++) {
+    const rec_idx = heading_indices[h_pos];
+    const rec = block_records[rec_idx];
+    const paragraph = rec.item;
+    const level = _heading_level(paragraph);
+    const text = _heading_text(paragraph, comments_map);
+    const style = _determine_heading_style(paragraph);
+    const owned_end = _find_owned_end(
+      block_records,
+      heading_indices,
+      h_pos,
+      level
+    );
+    const owned_blocks = block_records.slice(rec_idx + 1, owned_end);
+    const has_table = _direct_has_table(block_records, rec_idx + 1, owned_end);
+    const footnote_ids = _collect_footnote_ids(owned_blocks);
+    const page_num = _offset_to_page(rec.start_offset, body_page_offsets);
+    nodes.push({ level, text, page: page_num, style, has_table, footnote_ids });
+  }
+  return nodes;
+}
+function _direct_has_table(block_records, range_start, range_end) {
+  for (let idx = range_start; idx < range_end; idx++) {
+    const rec = block_records[idx];
+    if (rec.is_paragraph && _is_heading(rec.item)) return false;
+    if (rec.is_table) return true;
+  }
+  return false;
+}
+function _walk_doc_body(doc, comments_map) {
+  const parts = Array.from(iter_document_parts(doc));
+  let body_start_offset = 0;
+  let body_part = null;
+  for (const part of parts) {
+    if (part === doc) {
+      body_part = part;
+      break;
     }
-    if (children.length > 0) {
-      const last = children[children.length - 1];
-      if (["w:ins", "w:del", "w:hyperlink", "w:sdt", "w:smartTag", "w:fldSimple", "w:sdtContent"].includes(last.tagName)) {
-        coalesceContainer(last);
-      }
+    const part_text = _project_part(part, comments_map);
+    if (part_text) {
+      if (body_start_offset > 0) body_start_offset += 2;
+      body_start_offset += part_text.length;
     }
   }
-  const paragraphs = findAllDescendants(doc.element, "w:p");
-  for (const p of paragraphs) coalesceContainer(p);
-  return count ? [`Adjacent identical runs coalesced: ${count}`] : [];
-}
-function strip_rsid(doc) {
-  let count = 0;
-  const rsidAttrs = ["w:rsidR", "w:rsidRPr", "w:rsidRDefault", "w:rsidP", "w:rsidDel", "w:rsidSect", "w:rsidTr"];
-  const all = doc.element.getElementsByTagName("*");
-  for (let i = 0; i < all.length; i++) {
-    for (const attr of rsidAttrs) {
-      if (all[i].hasAttribute(attr)) {
-        all[i].removeAttribute(attr);
-        count++;
-      }
+  if (!body_part) {
+    body_part = doc;
+    body_start_offset = 0;
+  } else {
+    if (body_start_offset > 0) body_start_offset += 2;
+  }
+  const records = [];
+  let cursor = body_start_offset;
+  let is_first_block = true;
+  for (const item of iter_block_items(body_part)) {
+    if (item instanceof Paragraph) {
+      const prefix = get_paragraph_prefix(item);
+      const p_text = build_paragraph_text(item, comments_map, false);
+      const block_len = (prefix + p_text).length;
+      if (!is_first_block) cursor += 2;
+      records.push({
+        item,
+        is_paragraph: true,
+        is_table: false,
+        start_offset: cursor,
+        projected_length: block_len
+      });
+      cursor += block_len;
+      is_first_block = false;
+    } else if (item instanceof Table) {
+      const table_text = extract_table(item, comments_map, false, 0);
+      const block_len = table_text ? table_text.length : 0;
+      if (!is_first_block) cursor += 2;
+      const table_start = cursor;
+      records.push({
+        item,
+        is_paragraph: false,
+        is_table: true,
+        start_offset: table_start,
+        projected_length: block_len
+      });
+      _record_table_inner_blocks_lite(item, table_start, records, comments_map);
+      cursor += block_len;
+      is_first_block = false;
     }
   }
-  const rsidsElements = findAllDescendants(doc.element, "w:rsids");
-  for (const el of rsidsElements) {
-    if (el.parentNode) {
-      el.parentNode.removeChild(el);
-      count++;
+  return records;
+}
+function _compute_inner_block_offset(table, target_paragraph, table_start_offset, comments_map) {
+  const target_el = target_paragraph._element;
+  let cursor = table_start_offset;
+  let rows_processed = 0;
+  for (const row of table.rows) {
+    if (rows_processed > 0) cursor += 1;
+    const seen_cells = /* @__PURE__ */ new Set();
+    let cells_in_row = 0;
+    for (const cell of row.cells) {
+      if (seen_cells.has(cell)) continue;
+      seen_cells.add(cell);
+      if (cells_in_row > 0) cursor += 3;
+      const [new_cursor, found] = _walk_cell_for_offset(
+        cell,
+        target_el,
+        cursor,
+        comments_map
+      );
+      if (found) return new_cursor;
+      cursor = new_cursor;
+      cells_in_row++;
     }
+    rows_processed++;
   }
-  return count ? [`rsid attributes: ${count} removed`] : [];
+  return table_start_offset;
 }
-function strip_para_ids(doc) {
-  let count = 0;
-  const attrs = ["w14:paraId", "w14:textId"];
-  const all = doc.element.getElementsByTagName("*");
-  for (let i = 0; i < all.length; i++) {
-    for (const attr of attrs) {
-      if (all[i].hasAttribute(attr)) {
-        all[i].removeAttribute(attr);
-        count++;
+function _walk_cell_for_offset(cell, target_el, cell_start_cursor, comments_map) {
+  let cursor = cell_start_cursor;
+  let is_first_block = true;
+  for (const inner_item of iter_block_items(cell)) {
+    if (!is_first_block) cursor += 2;
+    if (inner_item instanceof Paragraph) {
+      if (inner_item._element === target_el) return [cursor, true];
+      const prefix = get_paragraph_prefix(inner_item);
+      const p_text = build_paragraph_text(inner_item, comments_map, false);
+      cursor += (prefix + p_text).length;
+    } else if (inner_item instanceof Table) {
+      const nested_offset = _compute_inner_block_offset(
+        inner_item,
+        new Paragraph(target_el, null),
+        cursor,
+        comments_map
+      );
+      if (nested_offset !== cursor) {
+        if (_element_is_descendant(target_el, inner_item._element))
+          return [nested_offset, true];
       }
+      const table_text = extract_table(inner_item, comments_map, false, 0);
+      cursor += table_text ? table_text.length : 0;
     }
+    is_first_block = false;
   }
-  return count ? [`Paragraph/text IDs: ${count} removed`] : [];
+  return [cursor, false];
 }
-function strip_proof_errors(doc) {
-  const elements = findAllDescendants(doc.element, "w:proofErr");
-  elements.forEach((el) => el.parentNode?.removeChild(el));
-  return elements.length ? [`Spell check markers: ${elements.length} removed`] : [];
+function _element_is_descendant(target_el, ancestor_el) {
+  let cur = target_el.parentNode;
+  while (cur) {
+    if (cur === ancestor_el) return true;
+    cur = cur.parentNode;
+  }
+  return false;
 }
-function strip_empty_properties(doc) {
-  let count = 0;
-  for (const tag of ["w:rPr", "w:pPr"]) {
-    const elements = findAllDescendants(doc.element, tag);
-    for (const el of elements) {
-      if (el.childNodes.length === 0 || el.childNodes.length === 1 && el.childNodes[0].nodeType === 3 && !el.childNodes[0].textContent?.trim()) {
-        el.parentNode?.removeChild(el);
-        count++;
+function _record_table_inner_blocks_lite(table, inherited_offset, records, comments_map) {
+  const seen_cells = /* @__PURE__ */ new Set();
+  for (const row of table.rows) {
+    for (const cell of row.cells) {
+      if (seen_cells.has(cell)) continue;
+      seen_cells.add(cell);
+      for (const inner_item of iter_block_items(cell)) {
+        if (inner_item instanceof Paragraph) {
+          const true_offset = _is_heading(inner_item) ? _compute_inner_block_offset(
+            table,
+            inner_item,
+            inherited_offset,
+            comments_map
+          ) : inherited_offset;
+          records.push({
+            item: inner_item,
+            is_paragraph: true,
+            is_table: false,
+            start_offset: true_offset,
+            projected_length: 0
+          });
+        } else if (inner_item instanceof Table) {
+          records.push({
+            item: inner_item,
+            is_paragraph: false,
+            is_table: true,
+            start_offset: inherited_offset,
+            projected_length: 0
+          });
+          _record_table_inner_blocks_lite(
+            inner_item,
+            inherited_offset,
+            records,
+            comments_map
+          );
+        }
       }
     }
   }
-  return count ? [`Empty property elements: ${count} removed`] : [];
 }
-function strip_hidden_text(doc) {
-  let count = 0;
-  const elements = findAllDescendants(doc.element, "w:rPr");
-  for (const rPr of elements) {
-    if (findChild(rPr, "w:vanish") || findChild(rPr, "w:webHidden")) {
-      const run = rPr.parentNode;
-      if (run && run.tagName === "w:r" && run.parentNode) {
-        run.parentNode.removeChild(run);
-        count++;
-      }
+function _project_part(part, comments_map) {
+  const blocks = [];
+  const c_type = part.constructor.name;
+  if (c_type === "NotesPart") {
+    const header = part.note_type === "fn" ? "## Footnotes" : "## Endnotes";
+    blocks.push(`---
+${header}`);
+  }
+  let is_first_para = true;
+  for (const item of iter_block_items(part)) {
+    if (item.constructor.name === "FootnoteItem") {
+      const fn_text = _project_part(item, comments_map);
+      if (fn_text) blocks.push(fn_text);
+    } else if (item instanceof Paragraph) {
+      let prefix = get_paragraph_prefix(item);
+      if (is_first_para && c_type === "FootnoteItem")
+        prefix = `[^${part.note_type}-${part.id}]: ${prefix}`;
+      const p_text = build_paragraph_text(item, comments_map, false);
+      blocks.push(prefix + p_text);
+      is_first_para = false;
+    } else if (item instanceof Table) {
+      const table_text = extract_table(item, comments_map, false, 0);
+      if (table_text) blocks.push(table_text);
+      is_first_para = false;
     }
   }
-  return count ? [`Hidden text runs: ${count} removed`] : [];
+  return blocks.join("\n\n");
 }
-function count_tracked_changes(doc) {
-  const ins = findAllDescendants(doc.element, "w:ins").length;
-  const del = findAllDescendants(doc.element, "w:del").length;
-  const fmt = findAllDescendants(doc.element, "w:rPrChange").length + findAllDescendants(doc.element, "w:pPrChange").length + findAllDescendants(doc.element, "w:sectPrChange").length;
-  return [ins, del, fmt];
+function _is_heading(paragraph) {
+  return _HEADING_PREFIX_RE.test(get_paragraph_prefix(paragraph));
 }
-function get_track_change_authors(doc) {
-  const authors = /* @__PURE__ */ new Set();
-  for (const tag of ["w:ins", "w:del", "w:rPrChange", "w:pPrChange", "w:sectPrChange"]) {
-    for (const el of findAllDescendants(doc.element, tag)) {
-      const author = el.getAttribute("w:author");
-      if (author) authors.add(author);
-    }
-  }
-  return authors;
+function _heading_passes_quality_filter(paragraph, comments_map) {
+  const style = _determine_heading_style(paragraph);
+  if (style !== "(heuristic)") return true;
+  const text = _heading_text(paragraph, comments_map);
+  if (!text) return false;
+  const word_count = (text.match(/\w+/g) || []).length;
+  return word_count >= _HEURISTIC_MIN_WORDS;
 }
-function _getElementText(el) {
-  const texts = [];
-  const ts = findAllDescendants(el, "w:t");
-  for (const t of ts) if (t.textContent) texts.push(t.textContent);
-  const dts = findAllDescendants(el, "w:delText");
-  for (const dt of dts) if (dt.textContent) texts.push(dt.textContent);
-  return texts.join("");
+function _heading_level(paragraph) {
+  const match = _HEADING_PREFIX_RE.exec(get_paragraph_prefix(paragraph));
+  return match ? Math.min(match[1].length, 6) : 1;
 }
-function _truncate(text, maxLen = 60) {
-  const clean = text.replace(/\n/g, " ").trim();
-  if (clean.length <= maxLen) return clean;
-  return clean.substring(0, maxLen - 3) + "...";
+function _heading_text(paragraph, comments_map) {
+  const p_text = build_paragraph_text(paragraph, comments_map, false);
+  let cleaned = _strip_critic_markup(p_text);
+  cleaned = _strip_inline_formatting(cleaned);
+  return cleaned.trim();
 }
-function accept_all_tracked_changes(doc) {
-  const lines = [];
-  const insEls = findAllDescendants(doc.element, "w:ins");
-  const delEls = findAllDescendants(doc.element, "w:del");
-  for (const ins of insEls) {
-    const text = _getElementText(ins).trim();
-    if (text) lines.push(`  Accepted insertion: "${_truncate(text, 60)}"`);
-  }
-  for (const del of delEls) {
-    const text = _getElementText(del).trim();
-    if (text) lines.push(`  Accepted deletion of: "${_truncate(text, 60)}"`);
-  }
-  const engine = new RedlineEngine(doc);
-  engine.accept_all_revisions();
-  for (const tag of ["w:rPrChange", "w:pPrChange", "w:sectPrChange"]) {
-    for (const el of findAllDescendants(doc.element, tag)) {
-      el.parentNode?.removeChild(el);
-    }
-  }
-  const total = insEls.length + delEls.length;
-  if (total) {
-    return [`Tracked changes auto-accepted: ${total}`].concat(lines);
-  }
-  return [];
+function _strip_critic_markup(text) {
+  if (!text) return "";
+  text = text.replace(/\{--[\s\S]*?--\}/g, "");
+  text = text.replace(/\{>>[\s\S]*?<<\}/g, "");
+  text = text.replace(/\{\+\+([\s\S]*?)\+\+\}/g, "$1");
+  text = text.replace(/\{==([\s\S]*?)==\}/g, "$1");
+  return text;
 }
-function get_comments_summary(doc) {
-  const data = extract_comments_data(doc.pkg);
-  const comments = [];
-  let openCount = 0;
-  let resolvedCount = 0;
-  for (const [cId, info] of Object.entries(data)) {
-    if (info.resolved) resolvedCount++;
-    else openCount++;
-    comments.push({ id: cId, ...info });
-  }
-  return { total: comments.length, open: openCount, resolved: resolvedCount, comments };
+function _strip_inline_formatting(text) {
+  if (!text) return "";
+  text = text.replace(/\*\*(.+?)\*\*/g, "$1");
+  text = text.replace(/__(.+?)__/g, "$1");
+  text = text.replace(/(?<!\w)_(\S(?:.*?\S)?)_(?!\w)/g, "$1");
+  return text;
 }
-function remove_all_comments(doc) {
-  const data = extract_comments_data(doc.pkg);
-  const keys = Object.keys(data);
-  if (keys.length === 0) return [];
-  const lines = [];
-  const cm = new CommentsManager(doc);
-  for (const [cId, info] of Object.entries(data)) {
-    const status = info.resolved ? "[Resolved]" : "[Open]";
-    lines.push(`  ${status} "${_truncate(info.text || "", 60)}" (${info.author || "Unknown"})`);
-    cm.deleteComment(cId);
-  }
-  for (const tag of ["w:commentRangeStart", "w:commentRangeEnd"]) {
-    for (const el of findAllDescendants(doc.element, tag)) {
-      el.parentNode?.removeChild(el);
-    }
+function _determine_heading_style(paragraph) {
+  const [style_cache, default_pstyle] = _get_style_cache(
+    paragraph._parent.part || paragraph._parent
+  );
+  const pPr = findChild(paragraph._element, "w:pPr");
+  let style_id = default_pstyle;
+  if (pPr) {
+    const pStyle = findChild(pPr, "w:pStyle");
+    if (pStyle) style_id = pStyle.getAttribute("w:val") || default_pstyle;
   }
-  const refs = findAllDescendants(doc.element, "w:commentReference");
-  for (const ref of refs) {
-    const parent = ref.parentNode;
-    if (parent) {
-      if (parent.tagName === "w:r" || parent.tagName.endsWith(":r")) {
-        const nonRprChildren = Array.from(parent.childNodes).filter(
-          (c) => c.nodeType === 1 && c.tagName !== "w:rPr" && c.tagName !== "rPr"
-        );
-        if (nonRprChildren.length <= 1) {
-          parent.parentNode?.removeChild(parent);
-        } else {
-          parent.removeChild(ref);
-        }
-      } else {
-        parent.removeChild(ref);
-      }
+  let outline_level = null;
+  let outline_level_from_style = false;
+  if (pPr) {
+    const oLvl = findChild(pPr, "w:outlineLvl");
+    if (oLvl && /^\d+$/.test(oLvl.getAttribute("w:val") || "")) {
+      outline_level = parseInt(oLvl.getAttribute("w:val"), 10);
     }
   }
-  const resolvedCount = Object.values(data).filter((c) => c.resolved).length;
-  const openCount = Object.values(data).filter((c) => !c.resolved).length;
-  return [`Comments removed: ${keys.length} (${resolvedCount} resolved, ${openCount} open)`].concat(lines);
-}
-function eject_comment_parts(doc) {
-  const pkg = doc.pkg;
-  const comment_partnames = /* @__PURE__ */ new Set();
-  for (const part of pkg.parts) {
-    if (part.partname.toLowerCase().includes("comments")) {
-      comment_partnames.add(part.partname);
-      const withSlash = part.partname.startsWith("/") ? part.partname : "/" + part.partname;
-      const withoutSlash = part.partname.startsWith("/") ? part.partname.substring(1) : part.partname;
-      comment_partnames.add(withSlash);
-      comment_partnames.add(withoutSlash);
-    }
+  if (outline_level === null && style_id && style_cache && style_cache[style_id]) {
+    outline_level = style_cache[style_id].outline_level;
+    outline_level_from_style = true;
   }
-  if (comment_partnames.size === 0) return;
-  for (const part of pkg.parts) {
-    if (part.partname.endsWith(".rels")) {
-      const rels = findAllDescendants(part._element, "Relationship");
-      const toRemove = [];
-      for (const rel of rels) {
-        const target = rel.getAttribute("Target") || "";
-        if (target.toLowerCase().includes("comments")) {
-          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);
-          }
-        }
-      }
-      for (const relEl of toRemove) {
-        relEl.parentNode?.removeChild(relEl);
-      }
+  const style_name = style_id && style_cache && style_cache[style_id] ? style_cache[style_id].name : style_id;
+  let normalized_style_name = style_name;
+  if (normalized_style_name && typeof normalized_style_name === "string") {
+    if (normalized_style_name.toLowerCase().startsWith("heading")) {
+      normalized_style_name = normalized_style_name.replace(/^heading/i, "Heading");
+    } else if (normalized_style_name.toLowerCase() === "title") {
+      normalized_style_name = "Title";
     }
   }
-  const ctPart = pkg.getPartByPath("[Content_Types].xml");
-  if (ctPart) {
-    const overrides = findAllDescendants(ctPart._element, "Override");
-    const toRemove = [];
-    for (const override of overrides) {
-      const partName = override.getAttribute("PartName") || "";
-      if (comment_partnames.has(partName) || partName.toLowerCase().includes("comments")) {
-        toRemove.push(override);
-      }
-    }
-    for (const overrideEl of toRemove) {
-      overrideEl.parentNode?.removeChild(overrideEl);
+  if (outline_level_from_style && outline_level !== null) {
+    const is_heading_or_title = normalized_style_name && (normalized_style_name.startsWith("Heading") || normalized_style_name === "Title");
+    if (!is_heading_or_title) {
+      outline_level = null;
     }
   }
-  pkg.parts = pkg.parts.filter((p) => !p.partname.toLowerCase().includes("comments"));
-  for (const key of Object.keys(pkg.unzipped)) {
-    if (key.toLowerCase().includes("comments")) {
-      delete pkg.unzipped[key];
+  if (outline_level !== null && outline_level >= 0 && outline_level <= 8) {
+    if (normalized_style_name && (normalized_style_name.startsWith("Heading") || normalized_style_name === "Title")) {
+      return normalized_style_name;
     }
+    return "(outline_level)";
   }
+  if (normalized_style_name && (normalized_style_name.startsWith("Heading") || normalized_style_name === "Title"))
+    return normalized_style_name;
+  if (normalized_style_name && /Heading[ ]?([1-6])(?![0-9])/.test(normalized_style_name))
+    return normalized_style_name;
+  return "(heuristic)";
 }
-function replace_comment_authors(doc, newAuthor) {
-  const cm = new CommentsManager(doc);
-  if (!cm.commentsPart) return [];
-  const original = /* @__PURE__ */ new Set();
-  const comments = findAllDescendants(cm.commentsPart._element, "w:comment");
-  for (const c of comments) {
-    const author = c.getAttribute("w:author");
-    if (author) {
-      original.add(author);
-      c.setAttribute("w:author", newAuthor);
-    }
-    if (c.hasAttribute("w:initials")) {
-      const initials = newAuthor.split(" ").filter(Boolean).map((p) => p[0]).join("").toUpperCase();
-      c.setAttribute("w:initials", initials);
-    }
+function _find_owned_end(block_records, heading_indices, current_h_pos, current_level) {
+  for (let next_h_pos = current_h_pos + 1; next_h_pos < heading_indices.length; next_h_pos++) {
+    const next_idx = heading_indices[next_h_pos];
+    if (_heading_level(block_records[next_idx].item) <= current_level)
+      return next_idx;
   }
-  return original.size ? [`Comment authors replaced: ${Array.from(original).sort().join(", ")} \u2192 "${newAuthor}"`] : [];
+  return block_records.length;
 }
-function replace_change_authors(doc, newAuthor) {
-  const original = /* @__PURE__ */ new Set();
-  for (const tag of ["w:ins", "w:del", "w:rPrChange", "w:pPrChange"]) {
-    for (const el of findAllDescendants(doc.element, tag)) {
-      const author = el.getAttribute("w:author");
-      if (author) {
-        original.add(author);
-        el.setAttribute("w:author", newAuthor);
+function _collect_footnote_ids(owned_blocks) {
+  const seen = /* @__PURE__ */ new Set();
+  const ordered = [];
+  for (const rec of owned_blocks) {
+    if (!rec.is_paragraph) continue;
+    for (const event of iter_paragraph_content(rec.item)) {
+      if (!("type" in event)) continue;
+      let fn_id = "";
+      if (event.type === "footnote") fn_id = `fn-${event.id}`;
+      else if (event.type === "endnote") fn_id = `en-${event.id}`;
+      else continue;
+      if (!seen.has(fn_id)) {
+        seen.add(fn_id);
+        ordered.push(fn_id);
       }
     }
   }
-  return original.size ? [`Track change authors replaced: ${Array.from(original).sort().join(", ")} \u2192 "${newAuthor}"`] : [];
+  return ordered;
 }
-function normalize_change_dates(doc) {
-  let count = 0;
-  const fixed = "2025-01-01T00:00:00Z";
-  for (const tag of ["w:ins", "w:del", "w:rPrChange", "w:pPrChange"]) {
-    for (const el of findAllDescendants(doc.element, tag)) {
-      if (el.hasAttribute("w:date")) {
-        el.setAttribute("w:date", fixed);
-        count++;
+function _offset_to_page(offset, body_page_offsets) {
+  if (!body_page_offsets || body_page_offsets.length === 0) return 1;
+  let page = 1;
+  for (let i = 0; i < body_page_offsets.length; i++) {
+    if (offset >= body_page_offsets[i]) page = i + 1;
+    else break;
+  }
+  return page;
+}
+function _extract_outline_fast(doc, projected_body, body_page_offsets, paragraph_offsets) {
+  const paragraphs_and_tables = [];
+  const seen_cells = /* @__PURE__ */ new Set();
+  function walk(container) {
+    for (const item of iter_block_items(container)) {
+      const i_type = item.constructor.name;
+      if (i_type === "FootnoteItem") {
+        walk(item);
+      } else if (item instanceof Paragraph) {
+        paragraphs_and_tables.push(["p", item]);
+      } else if (item instanceof Table) {
+        paragraphs_and_tables.push(["t", item]);
+        for (const row of item.rows) {
+          for (const cell of row.cells) {
+            if (seen_cells.has(cell._element)) {
+              continue;
+            }
+            seen_cells.add(cell._element);
+            walk(cell);
+          }
+        }
       }
     }
   }
-  return count ? [`Track change timestamps: ${count} normalized`] : [];
-}
-function scrub_doc_properties(doc) {
-  const lines = [];
-  const corePart = doc.pkg.getPartByPath("docProps/core.xml");
-  if (corePart) {
-    const creators = findDescendantsByLocalName(corePart._element, "creator");
-    creators.forEach((c) => {
-      if (c.textContent) {
-        lines.push(`Author: ${c.textContent}`);
-        c.textContent = "";
+  walk(doc);
+  const heading_indices = [];
+  for (let idx = 0; idx < paragraphs_and_tables.length; idx++) {
+    const [kind, item] = paragraphs_and_tables[idx];
+    if (kind !== "p") continue;
+    let hasOffset = false;
+    if (paragraph_offsets instanceof Map) {
+      hasOffset = paragraph_offsets.has(item._element);
+    } else {
+      hasOffset = item._element in paragraph_offsets;
+    }
+    if (!hasOffset) {
+      continue;
+    }
+    if (!_is_heading(item)) continue;
+    if (!_heading_passes_quality_filter_fast(item, projected_body, paragraph_offsets)) continue;
+    heading_indices.push(idx);
+  }
+  if (heading_indices.length === 0) return [];
+  const nodes = [];
+  for (let h_pos = 0; h_pos < heading_indices.length; h_pos++) {
+    const item_idx = heading_indices[h_pos];
+    const paragraph = paragraphs_and_tables[item_idx][1];
+    const level = _heading_level(paragraph);
+    const text = _heading_text_fast(paragraph, projected_body, paragraph_offsets);
+    const style = _determine_heading_style(paragraph);
+    let owned_end = item_idx;
+    for (let next_h_pos = h_pos + 1; next_h_pos < heading_indices.length; next_h_pos++) {
+      const next_idx = heading_indices[next_h_pos];
+      const next_paragraph = paragraphs_and_tables[next_idx][1];
+      if (_heading_level(next_paragraph) <= level) {
+        owned_end = next_idx;
+        break;
       }
-    });
-    const modifiers = findDescendantsByLocalName(corePart._element, "lastModifiedBy");
-    modifiers.forEach((c) => {
-      if (c.textContent) {
-        lines.push(`Last modified by: ${c.textContent}`);
-        c.textContent = "";
+    }
+    if (owned_end === item_idx) {
+      owned_end = paragraphs_and_tables.length;
+    }
+    const owned = paragraphs_and_tables.slice(item_idx + 1, owned_end);
+    let has_table = false;
+    for (const [kind2, item2] of owned) {
+      if (kind2 === "p" && _is_heading(item2)) {
+        break;
       }
-    });
-    const revisions = findDescendantsByLocalName(corePart._element, "revision");
-    revisions.forEach((c) => {
-      if (c.textContent && parseInt(c.textContent) > 1) {
-        lines.push(`Revision count: ${c.textContent} \u2192 1`);
-        c.textContent = "1";
+      if (kind2 === "t") {
+        has_table = true;
+        break;
       }
-    });
-  }
-  const appPart = doc.pkg.getPartByPath("docProps/app.xml");
-  if (appPart) {
-    const docEl = appPart._element;
-    const intFields = ["TotalTime", "Words", "Characters", "Paragraphs", "Lines", "CharactersWithSpaces"];
-    for (const f of intFields) {
-      findDescendantsByLocalName(docEl, f).forEach((el) => {
-        if (el.textContent && el.textContent !== "0") {
-          if (f === "TotalTime") lines.push(`Total editing time: ${el.textContent} minutes`);
-          el.textContent = "0";
-        }
-      });
     }
-    const strFields = ["Template", "Manager", "Company"];
-    for (const f of strFields) {
-      findDescendantsByLocalName(docEl, f).forEach((el) => {
-        if (el.textContent) {
-          lines.push(`${f}: ${el.textContent}`);
-          el.textContent = "";
-        }
-      });
+    const footnote_ids = _collect_footnote_ids_fast(owned);
+    let para_offset;
+    if (paragraph_offsets instanceof Map) {
+      para_offset = paragraph_offsets.get(paragraph._element);
+    } else {
+      para_offset = paragraph_offsets[paragraph._element];
     }
-  }
-  return lines.length ? ["Metadata scrubbed:", ...lines.map((l) => `  ${l}`)] : [];
-}
-function scrub_timestamps(doc) {
-  let modified = false;
-  const epoch = "1970-01-01T00:00:00Z";
-  const corePart = doc.pkg.getPartByPath("docProps/core.xml");
-  if (corePart) {
-    for (const tag of ["created", "modified", "lastPrinted"]) {
-      findDescendantsByLocalName(corePart._element, tag).forEach((el) => {
-        if (el.textContent && el.textContent !== epoch) {
-          el.textContent = epoch;
-          modified = true;
-        }
-      });
+    let page_num = 1;
+    if (para_offset !== void 0) {
+      const [start_offset] = para_offset;
+      page_num = _offset_to_page(start_offset, body_page_offsets);
     }
+    nodes.push({
+      level,
+      text,
+      page: page_num,
+      style,
+      has_table,
+      footnote_ids
+    });
   }
-  return modified ? ["Timestamps normalized to epoch"] : [];
+  return nodes;
 }
-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) => {
-    const toRemove = [];
-    for (const rel of findAllDescendants(relsPart._element, "Relationship")) {
-      const target = rel.getAttribute("Target");
-      if (target && target.includes("customXml")) toRemove.push(rel);
-    }
-    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);
-  for (const sdtPr of findAllDescendants(doc.element, "w:sdtPr")) {
-    findChildren(sdtPr, "w:dataBinding").forEach((b) => sdtPr.removeChild(b));
+function _heading_passes_quality_filter_fast(paragraph, projected_body, paragraph_offsets) {
+  const style = _determine_heading_style(paragraph);
+  if (style !== "(heuristic)") return true;
+  const text = _heading_text_fast(paragraph, projected_body, paragraph_offsets);
+  if (!text) return false;
+  const words = text.match(/\w+/g) || [];
+  return words.length >= _HEURISTIC_MIN_WORDS;
+}
+function _heading_text_fast(paragraph, projected_body, paragraph_offsets) {
+  let offset;
+  if (paragraph_offsets instanceof Map) {
+    offset = paragraph_offsets.get(paragraph._element);
+  } else {
+    offset = paragraph_offsets[paragraph._element];
   }
-  return [`Custom XML parts: ${customParts.length} removed`];
+  if (offset === void 0) {
+    return "";
+  }
+  const [start, length] = offset;
+  const raw = projected_body.substring(start, start + length);
+  let cleaned = _strip_critic_markup(raw);
+  cleaned = _strip_inline_formatting(cleaned);
+  cleaned = cleaned.replace(/^#+\s+/, "");
+  return cleaned.trim();
 }
-function strip_image_alt_text(doc) {
-  let count = 0;
-  for (const docPr of findDescendantsByLocalName(doc.element, "docPr")) {
-    const descr = docPr.getAttribute("descr");
-    if (descr) {
-      const isShort = descr.length < 10;
-      const isFile = descr.includes(".") && descr.length < 60;
-      if (isShort || isFile) {
-        docPr.removeAttribute("descr");
-        count++;
+function _collect_footnote_ids_fast(owned_items) {
+  const seen = /* @__PURE__ */ new Set();
+  const ordered = [];
+  for (const [kind, item] of owned_items) {
+    if (kind !== "p") continue;
+    for (const event of iter_paragraph_content(item)) {
+      if (!("type" in event)) continue;
+      let fn_id = "";
+      if (event.type === "footnote") fn_id = `fn-${event.id}`;
+      else if (event.type === "endnote") fn_id = `en-${event.id}`;
+      else continue;
+      if (!seen.has(fn_id)) {
+        seen.add(fn_id);
+        ordered.push(fn_id);
       }
     }
   }
-  return count ? [`Image alt text: ${count} auto-generated descriptions removed`] : [];
+  return ordered;
 }
-function audit_hyperlinks(doc) {
-  const internal = ["sharepoint.com", "onedrive.com", ".internal", "intranet", "localhost", "10.", "192.168.", "172.16."];
-  const warnings = [];
-  const docRels = doc.pkg.getOrCreateRelsPart(doc.part.partname);
-  for (const rel of findAllDescendants(docRels._element, "Relationship")) {
-    if (rel.getAttribute("TargetMode") === "External") {
-      const url = rel.getAttribute("Target") || "";
-      for (const pattern of internal) {
-        if (url.toLowerCase().includes(pattern.toLowerCase())) {
-          warnings.push(`Hyperlink targets internal URL: ${_truncate(url, 80)}`);
-          break;
+
+// src/sanitize/report.ts
+var SanitizeReport = class {
+  filename;
+  mode;
+  author;
+  tracked_changes_found = 0;
+  tracked_changes_accepted = 0;
+  tracked_changes_kept = 0;
+  change_lines = [];
+  comments_removed = 0;
+  comments_kept = 0;
+  removed_comment_lines = [];
+  kept_comment_lines = [];
+  metadata_lines = [];
+  structural_lines = [];
+  warnings = [];
+  status = "clean";
+  blocked_reason = null;
+  constructor(filename, mode = "full", author = null) {
+    this.filename = filename;
+    this.mode = mode;
+    this.author = author;
+  }
+  add_transform_lines(lines) {
+    for (const line of lines) {
+      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("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);
+      } else {
+        this.structural_lines.push(line);
       }
     }
   }
-  return warnings;
-}
+  render() {
+    const sep = "\u2550".repeat(50);
+    const lines = [sep, `Finalization Report: ${this.filename}`];
+    const flags = [];
+    if (this.mode === "keep-markup") flags.push("--keep-markup");
+    if (this.author) flags.push(`--author "${this.author}"`);
+    if (this.tracked_changes_accepted > 0) flags.push("--accept-all");
+    if (flags.length > 0) lines.push(flags.join(" "));
+    lines.push(sep);
+    if (this.status === "blocked") {
+      lines.push("");
+      lines.push(`BLOCKED: ${this.blocked_reason}`);
+      lines.push(sep);
+      return lines.join("\n");
+    }
+    if (this.mode === "keep-markup" && (this.tracked_changes_kept > 0 || this.comments_kept > 0)) {
+      lines.push("");
+      lines.push("VISIBLE TO COUNTERPARTY");
+      if (this.tracked_changes_kept > 0) lines.push(`  Tracked changes: ${this.tracked_changes_kept}`);
+      if (this.comments_kept > 0) {
+        lines.push(`  Open comments: ${this.comments_kept}`);
+        for (const cl of this.kept_comment_lines) lines.push(`    ${cl}`);
+      }
+      if (this.author) lines.push(`  Author on all markup: "${this.author}"`);
+    }
+    if (this.change_lines.length > 0) {
+      lines.push("");
+      lines.push("TRACKED CHANGES");
+      for (const cl of this.change_lines) lines.push(`  ${cl}`);
+    }
+    if (this.removed_comment_lines.length > 0) {
+      lines.push("");
+      lines.push("COMMENTS (stripped)");
+      for (const cl of this.removed_comment_lines) lines.push(`  ${cl}`);
+    }
+    if (this.metadata_lines.length > 0) {
+      lines.push("");
+      lines.push("METADATA");
+      for (const ml of this.metadata_lines) lines.push(`  ${ml}`);
+    }
+    if (this.structural_lines.length > 0) {
+      lines.push("");
+      lines.push("STRUCTURAL & PROTECTION");
+      for (const sl of this.structural_lines) lines.push(`  ${sl}`);
+    }
+    if (this.warnings.length > 0) {
+      lines.push("");
+      lines.push("WARNINGS");
+      for (const w of this.warnings) lines.push(`  \u26A0 ${w}`);
+    }
+    lines.push("");
+    lines.push(sep);
+    if (this.warnings.length > 0) {
+      lines.push(`Result: CLEAN WITH WARNINGS (${this.warnings.length} warning${this.warnings.length > 1 ? "s" : ""})`);
+    } else {
+      lines.push(`Result: CLEAN (${this.tracked_changes_found} changes resolved, ${this.comments_removed} comments removed)`);
+    }
+    lines.push(sep);
+    return lines.join("\n");
+  }
+};
 
 // src/sanitize/core.ts
 async function finalize_document(doc, options) {