@adeu/core 1.10.0 → 1.11.2

package/dist/index.cjs

@@ -34,6 +34,7 @@ __export(index_exports, {
   DocumentMapper: () => DocumentMapper,
   DocumentObject: () => DocumentObject,
   RedlineEngine: () => RedlineEngine,
+  _extractTextFromDoc: () => _extractTextFromDoc,
   apply_edits_to_markdown: () => apply_edits_to_markdown,
   create_unified_diff: () => create_unified_diff,
   create_word_patch_diff: () => create_word_patch_diff,
@@ -1266,14 +1267,16 @@ function* iter_paragraph_content(paragraph) {
 var DocumentMapper = class {
   doc;
   clean_view;
+  original_view;
   comments_map;
   full_text = "";
   spans = [];
   appendix_start_index = -1;
   _text_chunks = [];
-  constructor(doc, clean_view = false) {
+  constructor(doc, clean_view = false, original_view = false) {
     this.doc = doc;
     this.clean_view = clean_view;
+    this.original_view = original_view;
     this.comments_map = extract_comments_data(doc.pkg);
     this._build_map();
   }
@@ -1355,14 +1358,15 @@ ${header}`;
       const ins = trPr ? findChild(trPr, "w:ins") : null;
       const del_node = trPr ? findChild(trPr, "w:del") : null;
       if (this.clean_view && del_node) continue;
+      if (this.original_view && ins) continue;
       if (rows_processed > 0) {
         this._add_virtual_text("\n", current, null);
         current += 1;
       }
-      if (ins && !this.clean_view) {
+      if (ins && !this.clean_view && !this.original_view) {
         this._add_virtual_text("{++ ", current, null);
         current += 4;
-      } else if (del_node && !this.clean_view) {
+      } else if (del_node && !this.clean_view && !this.original_view) {
         this._add_virtual_text("{-- ", current, null);
         current += 4;
       }
@@ -1378,11 +1382,11 @@ ${header}`;
         current = this._map_blocks(cell, current);
         cells_processed += 1;
       }
-      if (ins && !this.clean_view) {
+      if (ins && !this.clean_view && !this.original_view) {
         const suffix = ` |Chg:${ins.getAttribute("w:id")}++}`;
         this._add_virtual_text(suffix, current, null);
         current += suffix.length;
-      } else if (del_node && !this.clean_view) {
+      } else if (del_node && !this.clean_view && !this.original_view) {
         const suffix = ` |Chg:${del_node.getAttribute("w:id")}--}`;
         this._add_virtual_text(suffix, current, null);
         current += suffix.length;
@@ -1476,11 +1480,13 @@ ${header}`;
         }
         if (this.clean_view && Object.keys(active_del).length > 0) {
         }
+        if (this.original_view && Object.keys(active_ins).length > 0) {
+        }
         const full_seg_text = run_parts.map((x) => x[1]).join("");
         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)) {
-          const new_wrappers = this.clean_view ? ["", ""] : this._get_wrappers(curr_ins_id, curr_del_id, active_ids, active_fmt);
+        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_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;
@@ -1505,7 +1511,7 @@ ${header}`;
             }
           }
         }
-        if (!this.clean_view) {
+        if (!this.clean_view && !this.original_view) {
           const has_meta = Object.keys(active_ins).length > 0 || Object.keys(active_del).length > 0 || active_ids.size > 0 || Object.keys(active_fmt).length > 0;
           if (has_meta) {
             deferred_meta_states.push([{ ...active_ins }, { ...active_del }, new Set(active_ids), { ...active_fmt }]);
@@ -2681,6 +2687,7 @@ var RedlineEngine = class {
   mapper;
   comments_manager;
   clean_mapper = null;
+  original_mapper = null;
   skipped_details = [];
   constructor(doc, author = "Adeu AI (TS)") {
     this.doc = doc;
@@ -2720,7 +2727,10 @@ var RedlineEngine = class {
     if (!full_text) return [null, null];
     const before_start = Math.max(0, start_idx - 30);
     const context_before = full_text.substring(before_start, start_idx);
-    const context_after = full_text.substring(start_idx + length, start_idx + length + 30);
+    const context_after = full_text.substring(
+      start_idx + length,
+      start_idx + length + 30
+    );
     const critic_markup = `${context_before}{--${target_text}--}{++${new_text}++}${context_after}`;
     let clean_text = critic_markup;
     clean_text = clean_text.replace(/\{>>.*?<<\}/gs, "");
@@ -2892,7 +2902,9 @@ var RedlineEngine = class {
           overrideEl.parentNode?.removeChild(overrideEl);
         }
       }
-      pkg.parts = pkg.parts.filter((p) => !p.partname.toLowerCase().includes("comments"));
+      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];
@@ -3437,13 +3449,17 @@ var RedlineEngine = class {
       if (!edit.target_text) continue;
       let matches = this.mapper.find_all_match_indices(edit.target_text);
       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);
-        if (matches.length > 0) activeText = this.clean_mapper.full_text;
+        if (matches.length > 0) {
+          activeText = this.clean_mapper.full_text;
+          target_mapper = this.clean_mapper;
+        }
       }
-      if (activeText === this.mapper.full_text && matches.length > 1) {
+      if (activeText === this.mapper.full_text && matches.length > 0) {
         const liveMatches = matches.filter(([start, length]) => {
           const realSpans = this.mapper.spans.filter(
             (s) => s.run !== null && s.end > start && s.start < start + length
@@ -3451,13 +3467,51 @@ var RedlineEngine = class {
           if (realSpans.length === 0) return true;
           return realSpans.some((s) => !s.del_id);
         });
-        if (liveMatches.length > 0) matches = liveMatches;
+        matches = liveMatches;
       }
+      let is_deleted_text = false;
+      const deleted_authors = /* @__PURE__ */ new Set();
       if (matches.length === 0) {
-        errors.push(
-          `- Edit ${i + 1} Failed: Target text not found in document:
+        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);
+        if (orig_matches.length > 0) {
+          is_deleted_text = true;
+          for (const [start, length] of orig_matches) {
+            const spans = this.original_mapper.spans.filter(
+              (s) => s.end > start && s.start < start + length
+            );
+            for (const s of spans) {
+              if (s.run !== null) {
+                let parent = s.run._element;
+                while (parent) {
+                  if (parent.nodeType === 1 && parent.tagName === "w:del") {
+                    const auth = parent.getAttribute("w:author");
+                    if (auth) {
+                      deleted_authors.add(auth);
+                    }
+                    break;
+                  }
+                  parent = parent.parentNode;
+                }
+              }
+            }
+          }
+        }
+      }
+      if (matches.length === 0) {
+        if (is_deleted_text) {
+          const author_phrase = deleted_authors.size > 0 ? `by ${Array.from(deleted_authors).sort().join(", ")}` : "by an existing revision";
+          errors.push(
+            `- Edit ${i + 1} Failed: Target text matches text inside a tracked deletion ${author_phrase}. Reject/accept that change first or target the active replacement text instead.`
+          );
+        } else {
+          errors.push(
+            `- Edit ${i + 1} Failed: Target text not found in document:
   "${edit.target_text}"`
-        );
+          );
+        }
       } else if (matches.length > 1) {
         const positions = matches.map(([start, length]) => [
           start,
@@ -3478,7 +3532,10 @@ var RedlineEngine = class {
         const [pfx, sfx] = trim_common_context(matched, edit.new_text || "");
         const t_end = matched.length - sfx;
         const final_target = matched.substring(pfx, t_end);
-        const final_new = (edit.new_text || "").substring(pfx, (edit.new_text || "").length - sfx);
+        const final_new = (edit.new_text || "").substring(
+          pfx,
+          (edit.new_text || "").length - sfx
+        );
         if (final_target.includes("\n\n")) {
           if (final_new.includes("\n\n")) {
             const parts = matched.split("\n\n");
@@ -3634,7 +3691,9 @@ var RedlineEngine = class {
       if (dry_run_mode) {
         for (const edit of edits) {
           const single_errors = this.validate_edits([edit]);
-          const warning = this._check_punctuation_warning(edit.target_text || "");
+          const warning = this._check_punctuation_warning(
+            edit.target_text || ""
+          );
           if (single_errors.length > 0) {
             skipped_edits++;
             edits_reports.push({
@@ -3688,7 +3747,9 @@ var RedlineEngine = class {
         for (const edit of cloned_edits) {
           const success = edit._applied_status || false;
           const error_msg = edit._error_msg || null;
-          const warning = this._check_punctuation_warning(edit.target_text || "");
+          const warning = this._check_punctuation_warning(
+            edit.target_text || ""
+          );
           let critic_markup = null;
           let clean_text = null;
           if (success) {
@@ -3716,7 +3777,7 @@ var RedlineEngine = class {
       skipped_details: this.skipped_details,
       edits: edits_reports,
       engine: "node",
-      version: "1.9.0"
+      version: "1.10.0"
     };
   }
   apply_edits(edits) {
@@ -4239,7 +4300,10 @@ var RedlineEngine = class {
             if (result.first_node.tagName === "w:p") {
               first_anchor_target = findAllDescendants(result.first_node, "w:ins")[0] || result.first_node;
             }
-            const anchor = ascend_to_paragraph_child(first_anchor_target, host_p);
+            const anchor = ascend_to_paragraph_child(
+              first_anchor_target,
+              host_p
+            );
             this._attach_comment(host_p, anchor, anchor, edit.comment);
           }
         }
@@ -4251,7 +4315,10 @@ var RedlineEngine = class {
       length,
       rebuild_map
     );
-    const virtual_spans = active_mapper.get_virtual_spans_in_range(start_idx, length);
+    const virtual_spans = active_mapper.get_virtual_spans_in_range(
+      start_idx,
+      length
+    );
     if (target_runs.length === 0 && virtual_spans.length === 0) return false;
     const affected_ps = /* @__PURE__ */ new Set();
     for (const run of target_runs) {
@@ -4334,7 +4401,10 @@ var RedlineEngine = class {
             let pPr = findChild(p1_element, "w:pPr");
             if (!pPr) {
               pPr = p1_element.ownerDocument.createElement("w:pPr");
-              p1_element.insertBefore(pPr, p1_element.firstChild);
+              p1_element.insertBefore(
+                pPr,
+                p1_element.firstChild
+              );
             }
             let rPr = findChild(pPr, "w:rPr");
             if (!rPr) {
@@ -4863,13 +4933,14 @@ async function extractTextFromBuffer(buffer, cleanView = false) {
   const doc = await DocumentObject.load(buffer);
   return _extractTextFromDoc(doc, cleanView);
 }
-function _extractTextFromDoc(doc, cleanView = false, includeAppendix = true) {
+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);
+    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);
@@ -4881,9 +4952,12 @@ function _extractTextFromDoc(doc, cleanView = false, includeAppendix = true) {
     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;
 }
-function _extract_blocks(container, comments_map, cleanView, cursor) {
+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 = [];
@@ -4902,7 +4976,7 @@ ${header}`;
     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);
+      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;
@@ -4918,11 +4992,14 @@ ${header}`;
       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, cleanView, block_start);
+      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;
@@ -4935,7 +5012,7 @@ ${header}`;
   }
   return blocks.join("\n\n");
 }
-function extract_table(table, comments_map, cleanView, cursor) {
+function extract_table(table, comments_map, cleanView, cursor, paragraph_offsets) {
   const rows_text = [];
   let rows_processed = 0;
   let local_cursor = cursor;
@@ -4954,7 +5031,7 @@ function extract_table(table, comments_map, cleanView, cursor) {
       if (seen_cells.has(cell)) continue;
       seen_cells.add(cell);
       if (!first_cell) cell_cursor += 3;
-      const cell_content = _extract_blocks(cell, comments_map, cleanView, cell_cursor);
+      const cell_content = _extract_blocks(cell, comments_map, cleanView, cell_cursor, paragraph_offsets);
       cell_texts.push(cell_content);
       cell_cursor += cell_content.length;
       first_cell = false;
@@ -5211,6 +5288,9 @@ function extract_outline(doc, projected_body, body_pages, body_page_offsets, par
   if (body_pages.length !== body_page_offsets.length) {
     throw new Error("body_pages and body_page_offsets length mismatch");
   }
+  if (paragraph_offsets) {
+    return _extract_outline_fast(doc, projected_body, body_page_offsets, paragraph_offsets);
+  }
   const comments_map = extract_comments_data(doc.pkg);
   const block_records = _walk_doc_body(doc, comments_map);
   const heading_indices = [];
@@ -5486,6 +5566,7 @@ function _determine_heading_style(paragraph) {
     if (pStyle) style_id = pStyle.getAttribute("w:val") || default_pstyle;
   }
   let outline_level = null;
+  let outline_level_from_style = false;
   if (pPr) {
     const oLvl = findChild(pPr, "w:outlineLvl");
     if (oLvl && /^\d+$/.test(oLvl.getAttribute("w:val") || "")) {
@@ -5494,6 +5575,7 @@ function _determine_heading_style(paragraph) {
   }
   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;
@@ -5504,6 +5586,12 @@ function _determine_heading_style(paragraph) {
       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;
@@ -5552,6 +5640,146 @@ function _offset_to_page(offset, body_page_offsets) {
   }
   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);
+          }
+        }
+      }
+    }
+  }
+  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;
+      }
+      if (kind2 === "t") {
+        has_table = true;
+        break;
+      }
+    }
+    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 "";
+  }
+  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 _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;
+}
 
 // src/sanitize/report.ts
 var SanitizeReport = class {
@@ -6239,6 +6467,7 @@ function identifyEngine() {
   DocumentMapper,
   DocumentObject,
   RedlineEngine,
+  _extractTextFromDoc,
   apply_edits_to_markdown,
   create_unified_diff,
   create_word_patch_diff,