@ansonlai/docx-redline-js 0.4.0 → 0.5.1

package/node/index.js

@@ -0,0 +1,2 @@
+export { DocxDocument, openDocx, computePackageRevisionToken } from './docx-document.js';
+export { executeCli, runCli } from './cli.js';

package/node/zip-archive.js

@@ -0,0 +1,52 @@
+import { deflateRawSync, inflateRawSync } from 'node:zlib';
+
+const CRC_TABLE = (() => { const table = new Uint32Array(256); for (let n = 0; n < 256; n++) { let c = n; for (let k = 0; k < 8; k++) c = c & 1 ? 0xedb88320 ^ (c >>> 1) : c >>> 1; table[n] = c >>> 0; } return table; })();
+function crc32(buffer) { let crc = 0xffffffff; for (const byte of buffer) crc = CRC_TABLE[(crc ^ byte) & 0xff] ^ (crc >>> 8); return (crc ^ 0xffffffff) >>> 0; }
+
+export function unzipDocx(input) {
+    const source = Buffer.from(input);
+    let eocd = -1;
+    for (let offset = source.length - 22; offset >= Math.max(0, source.length - 0xffff - 22); offset--) if (source.readUInt32LE(offset) === 0x06054b50) { eocd = offset; break; }
+    if (eocd < 0) throw new Error('Invalid DOCX: ZIP end record not found.');
+    const count = source.readUInt16LE(eocd + 10); const centralOffset = source.readUInt32LE(eocd + 16);
+    if (count === 0xffff || centralOffset === 0xffffffff) throw new Error('Unsupported DOCX: ZIP64 archives are not supported.');
+    const entries = new Map(); let offset = centralOffset;
+    for (let index = 0; index < count; index++) {
+        if (source.readUInt32LE(offset) !== 0x02014b50) throw new Error('Invalid DOCX: central directory is corrupt.');
+        const flags = source.readUInt16LE(offset + 8); const method = source.readUInt16LE(offset + 10);
+        if (flags & 1) throw new Error('Unsupported DOCX: encrypted ZIP entries are not supported.');
+        const compressedSize = source.readUInt32LE(offset + 20); const nameLength = source.readUInt16LE(offset + 28);
+        const extraLength = source.readUInt16LE(offset + 30); const commentLength = source.readUInt16LE(offset + 32); const localOffset = source.readUInt32LE(offset + 42);
+        const name = source.toString('utf8', offset + 46, offset + 46 + nameLength);
+        if (source.readUInt32LE(localOffset) !== 0x04034b50) throw new Error('Invalid DOCX: local entry is corrupt.');
+        const localNameLength = source.readUInt16LE(localOffset + 26); const localExtraLength = source.readUInt16LE(localOffset + 28);
+        const start = localOffset + 30 + localNameLength + localExtraLength; const compressed = source.subarray(start, start + compressedSize);
+        if (method !== 0 && method !== 8) throw new Error(`Unsupported DOCX compression method: ${method}.`);
+        entries.set(name, method === 8 ? inflateRawSync(compressed) : Buffer.from(compressed));
+        offset += 46 + nameLength + extraLength + commentLength;
+    }
+    return entries;
+}
+
+export function zipDocx(entries) {
+    const local = []; const central = []; let offset = 0;
+    for (const [name, raw] of entries) {
+        const nameBytes = Buffer.from(name); const data = Buffer.from(raw); const compressed = deflateRawSync(data, { level: 9 });
+        const method = compressed.length < data.length ? 8 : 0; const payload = method === 8 ? compressed : data; const crc = crc32(data);
+        const lh = Buffer.alloc(30); lh.writeUInt32LE(0x04034b50); lh.writeUInt16LE(20,4); lh.writeUInt16LE(method,8); lh.writeUInt16LE(0x5c21,12); lh.writeUInt32LE(crc,14); lh.writeUInt32LE(payload.length,18); lh.writeUInt32LE(data.length,22); lh.writeUInt16LE(nameBytes.length,26);
+        local.push(lh, nameBytes, payload);
+        const ch = Buffer.alloc(46); ch.writeUInt32LE(0x02014b50); ch.writeUInt16LE(20,4); ch.writeUInt16LE(20,6); ch.writeUInt16LE(method,10); ch.writeUInt16LE(0x5c21,14); ch.writeUInt32LE(crc,16); ch.writeUInt32LE(payload.length,20); ch.writeUInt32LE(data.length,24); ch.writeUInt16LE(nameBytes.length,28); ch.writeUInt32LE(offset,42);
+        central.push(ch, nameBytes); offset += lh.length + nameBytes.length + payload.length;
+    }
+    const directory = Buffer.concat(central); const end = Buffer.alloc(22); end.writeUInt32LE(0x06054b50); end.writeUInt16LE(entries.size,8); end.writeUInt16LE(entries.size,10); end.writeUInt32LE(directory.length,12); end.writeUInt32LE(offset,16);
+    return Buffer.concat([...local, directory, end]);
+}
+
+export class MemoryZip {
+    constructor(entries) { this.entries = entries; }
+    file(path, value) {
+        if (value !== undefined) { this.entries.set(path, Buffer.isBuffer(value) ? Buffer.from(value) : Buffer.from(String(value))); return this; }
+        const data = this.entries.get(path); if (!data) return null;
+        return { async: async type => type === 'string' ? data.toString('utf8') : Buffer.from(data) };
+    }
+}

package/orchestration/list-markdown.js

@@ -2,23 +2,19 @@
  * Shared list markdown construction/parsing helpers for command adapters.
  */
 
+import {
+    matchListMarker,
+    stripListMarker
+} from '../pipeline/list-markers.js';
+
+export { inferNumberingStyleFromMarker } from '../pipeline/list-markers.js';
+
 /**
  * Infers numbering style from a list marker.
  *
  * @param {string} marker - Marker text
  * @returns {'decimal'|'lowerAlpha'|'upperAlpha'|'lowerRoman'|'upperRoman'}
  */
-export function inferNumberingStyleFromMarker(marker) {
-    const m = (marker || '').trim();
-    if (!m) return 'decimal';
-    if (/^\d+(?:\.\d+)*\.?$/.test(m) || /^\(\d+\)$/.test(m)) return 'decimal';
-    if (/^[ivxlcdm]+\.$/.test(m)) return 'lowerRoman';
-    if (/^[IVXLCDM]{2,}\.$/.test(m)) return 'upperRoman';
-    if (/^[a-z]\.$/.test(m)) return 'lowerAlpha';
-    if (/^[A-Z]\.$/.test(m)) return 'upperAlpha';
-    return 'decimal';
-}
-
 /**
  * Builds list markdown from normalized item+level input.
  *
@@ -60,8 +56,6 @@ export function buildListMarkdown(itemsWithLevels, listType, numberingStyle) {
  */
 export function normalizeListItemsWithLevels(rawItems, options = {}) {
     const indentSpaces = Math.max(1, Number(options.indentSpaces) || 4);
-    const markersRegex = /^((?:\d+(?:\.\d+)*\.?|\((?:\d+|[a-zA-Z]|[ivxlcIVXLC]+)\)|[a-zA-Z]\.|\d+\.|[ivxlcIVXLC]+\.|[-*•])\s*)/;
-
     return (rawItems || []).map((rawItem) => {
         const item = String(rawItem ?? '');
         const indentMatch = item.match(/^(\s*)/);
@@ -70,10 +64,10 @@ export function normalizeListItemsWithLevels(rawItems, options = {}) {
 
         let stripped = item.trim();
         let removedMarker = null;
-        const markerMatch = stripped.match(markersRegex);
+        const markerMatch = matchListMarker(stripped, { allowZeroSpaceAfterMarker: true });
         if (markerMatch) {
-            removedMarker = markerMatch[1].trim() || null;
-            stripped = stripped.replace(markersRegex, '');
+            removedMarker = markerMatch[2].trim() || null;
+            stripped = stripListMarker(stripped, { allowZeroSpaceAfterMarker: true });
         }
 
         return {

package/orchestration/list-parsing.js

@@ -4,7 +4,7 @@
  * Keeps command-layer list parsing aligned with reconciliation marker logic.
  */
 
-import { matchListMarker, stripListMarker } from '../pipeline/list-markers.js';
+import { parseListItem } from '../pipeline/list-markers.js';
 
 /**
  * Parses markdown list-like content into structured items.
@@ -26,19 +26,14 @@ export function parseMarkdownListContent(content) {
     for (const line of lines) {
         if (!line.trim()) continue;
 
-        const markerMatch = matchListMarker(line, { allowZeroSpaceAfterMarker: false });
-        if (markerMatch) {
-            const indent = markerMatch[1] || '';
-            const marker = markerMatch[2].trim();
-            const text = stripListMarker(line, { allowZeroSpaceAfterMarker: false }).trim();
-            const level = Math.floor(indent.length / 2);
-            const isBullet = /^[-*+\u2022]$/.test(marker);
+        const parsed = parseListItem(line, { allowZeroSpaceAfterMarker: false, indentSpaces: 2 });
+        if (parsed) {
 
             items.push({
-                type: isBullet ? 'bullet' : 'numbered',
-                level,
-                text,
-                marker
+                type: parsed.markerType,
+                level: parsed.level,
+                text: parsed.text.trim(),
+                marker: parsed.marker
             });
             continue;
         }

package/orchestration/list-structural-fallback.js

@@ -11,7 +11,8 @@ import {
     normalizeWhitespaceForTargeting
 } from '../core/paragraph-targeting.js';
 import { getParagraphListInfo } from '../core/list-targeting.js';
-import { ReconciliationPipeline } from '../pipeline/pipeline.js';
+import { executeListGeneration } from '../pipeline/list-generation.js';
+import { NumberingService } from '../services/numbering-service.js';
 import { preprocessMarkdown } from '../pipeline/markdown-processor.js';
 import { parseMarkdownListContent, hasListItems } from './list-parsing.js';
 import { inferNumberingStyleFromMarker } from './list-markdown.js';
@@ -433,7 +434,8 @@ export function buildSingleLineListStructuralFallbackPlan(options = {}) {
  * @param {{
  *   author?: string,
  *   generateRedlines?: boolean,
- *   pipeline?: ReconciliationPipeline,
+ *   revisionIdAllocator?: import('../core/types.js').RevisionIdAllocator|null,
+ *   pipeline?: import('../pipeline/pipeline.js').ReconciliationPipeline,
  *   setAbstractStartOverride?: boolean
  * }} [options={}] - Execution options
  * @returns {Promise<{
@@ -461,14 +463,23 @@ export async function executeSingleLineListStructuralFallback(plan, options = {}
 
     const author = options.author || 'AI';
     const generateRedlines = options.generateRedlines ?? true;
-    const pipeline = options.pipeline || new ReconciliationPipeline({ author, generateRedlines });
-
-    const result = await pipeline.executeListGeneration(
-        plan.listInput,
-        null,
-        null,
-        String(plan.originalText || '')
-    );
+    const result = options.pipeline
+        ? await options.pipeline.executeListGeneration(
+            plan.listInput,
+            null,
+            null,
+            String(plan.originalText || '')
+        )
+        : await executeListGeneration({
+            cleanText: plan.listInput,
+            numberingContext: null,
+            originalRunModel: [],
+            originalText: String(plan.originalText || ''),
+            generateRedlines,
+            author,
+            revisionIdAllocator: options.revisionIdAllocator || null,
+            numberingService: new NumberingService()
+        });
 
     const rawOxml = result?.oxml || result?.ooxml || '';
     const oxml = trimTrailingBlankParagraph(rawOxml);

package/package.json

@@ -1,102 +1,123 @@
-{
-  "name": "@ansonlai/docx-redline-js",
-  "version": "0.4.0",
-  "description": "Host-independent OOXML reconciliation engine for .docx manipulation with track changes",
-  "license": "MIT",
-  "type": "module",
-  "main": "./index.js",
-  "module": "./index.js",
-  "types": "./index.d.ts",
-  "exports": {
-    ".": {
-      "types": "./index.d.ts",
-      "import": "./index.js",
-      "default": "./index.js"
-    },
-    "./standalone-runner": "./services/standalone-operation-runner.js",
-    "./redline-operation-converter": "./orchestration/redline-operation-converter.js",
-    "./adapters/*": "./adapters/*",
-    "./core/*": "./core/*",
-    "./engine/*": "./engine/*",
-    "./pipeline/*": "./pipeline/*",
-    "./services/*": "./services/*",
-    "./orchestration/*": "./orchestration/*"
-  },
-  "files": [
-    "adapters/",
-    "core/",
-    "engine/",
-    "pipeline/",
-    "services/",
-    "orchestration/",
-    "scripts/",
-    "docs/",
-    "index.js",
-    "index.d.ts",
-    "dist/",
-    "ARCHITECTURE.md",
-    "AGENTS.md",
-    "README.md",
-    "LICENSE"
-  ],
-  "dependencies": {
-    "diff-match-patch": "^1.0.5"
-  },
-  "peerDependencies": {
-    "@xmldom/xmldom": ">=0.8.0"
-  },
-  "peerDependenciesMeta": {
-    "@xmldom/xmldom": {
-      "optional": true
-    }
-  },
-  "devDependencies": {
-    "@xmldom/xmldom": "^0.9.0",
-    "c8": "^12.0.0",
-    "docx-preview": "^0.4.0",
-    "esbuild": "^0.28.1",
-    "eslint": "^10.9.1",
-    "jszip": "^3.10.1",
-    "typescript": "^7.0.2"
-  },
-  "scripts": {
-    "build": "node scripts/build.mjs",
-    "check:types": "tsc -p tsconfig.types.json && node scripts/check-types.mjs",
-    "lint": "eslint index.js adapters core engine pipeline services orchestration scripts tests",
-    "test:coverage": "c8 --reporter=text --reporter=json --reporter=json-summary --reports-dir=coverage node scripts/run-tests.mjs",
-    "coverage:gaps": "node scripts/report-coverage-gaps.mjs",
-    "smoke:word": "powershell -File scripts/word-com-smoke.ps1",
-    "smoke:word:diff": "powershell -File scripts/word-com-differential.ps1",
-    "test:word": "powershell -NoProfile -ExecutionPolicy Bypass -File scripts/word-com-suite.ps1",
-    "test:word:visual": "powershell -NoProfile -ExecutionPolicy Bypass -File scripts/word-com-visual-suite.ps1",
-    "test:corpus:word": "powershell -NoProfile -ExecutionPolicy Bypass -File scripts/word-com-corpus-suite.ps1",
-    "test:corpus:word:visual": "powershell -NoProfile -ExecutionPolicy Bypass -File scripts/word-com-corpus-visual-suite.ps1",
-    "test:visual:inspect": "node scripts/inspect-visual-evidence.mjs",
-    "test:visual:sample": "node scripts/sample-multimodal-visual-check.mjs",
-    "corpus:fetch:superdoc": "node scripts/fetch-superdoc-corpus.mjs",
-    "report:word:coverage": "node scripts/report-word-coverage.mjs",
-    "report:test:dashboard": "node scripts/build-test-dashboard.mjs",
-    "review:word:prepare": "node scripts/prepare-word-review.mjs",
-    "test": "node scripts/run-tests.mjs",
-    "test:isolation": "node tests/no_word_api_index_check.mjs && node tests/core_dependency_graph_check.mjs",
-    "prepublishOnly": "npm run test:isolation && npm run build"
-  },
-  "keywords": [
-    "docx",
-    "ooxml",
-    "reconciliation",
-    "track-changes",
-    "redlines",
-    "word",
-    "office",
-    "document",
-    "xml"
-  ],
-  "repository": {
-    "type": "git",
-    "url": "https://github.com/AnsonLai/docx-redline-js.git"
-  },
-  "engines": {
-    "node": ">=20.0.0"
-  }
-}
+{
+  "name": "@ansonlai/docx-redline-js",
+  "version": "0.5.1",
+  "description": "Host-independent OOXML reconciliation engine for .docx manipulation with track changes",
+  "license": "MIT",
+  "type": "module",
+  "main": "./index.js",
+  "module": "./index.js",
+  "types": "./index.d.ts",
+  "bin": {
+    "docx-redline": "./bin/docx-redline.js"
+  },
+  "exports": {
+    ".": {
+      "types": "./index.d.ts",
+      "import": "./index.js",
+      "default": "./index.js"
+    },
+    "./standalone-runner": {
+      "types": "./services/standalone-operation-runner.d.ts",
+      "import": "./services/standalone-operation-runner.js",
+      "default": "./services/standalone-operation-runner.js"
+    },
+    "./node": {
+      "types": "./node/index.d.ts",
+      "import": "./node/index.js",
+      "default": "./node/index.js"
+    },
+    "./redline-operation-converter": "./orchestration/redline-operation-converter.js",
+    "./adapters/*": "./adapters/*",
+    "./core/*": "./core/*",
+    "./engine/*": "./engine/*",
+    "./pipeline/*": "./pipeline/*",
+    "./services/*": "./services/*",
+    "./orchestration/*": "./orchestration/*"
+  },
+  "files": [
+    "adapters/",
+    "core/",
+    "engine/",
+    "pipeline/",
+    "services/",
+    "node/",
+    "bin/",
+    "orchestration/",
+    "scripts/",
+    "docs/",
+    "index.js",
+    "index.d.ts",
+    "dist/",
+    "ARCHITECTURE.md",
+    "AGENTS.md",
+    "CHANGELOG.md",
+    "README.md",
+    "LICENSE"
+  ],
+  "dependencies": {
+    "diff-match-patch": "^1.0.5"
+  },
+  "peerDependencies": {
+    "@xmldom/xmldom": ">=0.8.0"
+  },
+  "peerDependenciesMeta": {
+    "@xmldom/xmldom": {
+      "optional": true
+    }
+  },
+  "devDependencies": {
+    "@xmldom/xmldom": "^0.9.0",
+    "c8": "^12.0.0",
+    "docx-preview": "^0.4.0",
+    "esbuild": "^0.28.1",
+    "eslint": "^10.9.1",
+    "jszip": "^3.10.1",
+    "typescript": "^7.0.2"
+  },
+  "scripts": {
+    "build": "node scripts/build.mjs",
+    "check:types": "tsc -p tsconfig.types.json && node scripts/check-types.mjs",
+    "lint": "eslint index.js adapters core engine pipeline services orchestration scripts tests",
+    "test:coverage": "c8 --reporter=text --reporter=json --reporter=json-summary --reports-dir=coverage node scripts/run-tests.mjs",
+    "coverage:gaps": "node scripts/report-coverage-gaps.mjs",
+    "benchmark:session": "node scripts/benchmark-operation-session.mjs",
+    "benchmark:targeting": "node scripts/benchmark-targeting-hot-paths.mjs",
+    "benchmark:tests": "node scripts/benchmark-test-runner.mjs",
+    "profile:routes": "node scripts/profile-route-selection.mjs",
+    "smoke:word": "powershell -File scripts/word-com-smoke.ps1",
+    "smoke:word:diff": "powershell -File scripts/word-com-differential.ps1",
+    "test:word": "powershell -NoProfile -ExecutionPolicy Bypass -File scripts/word-com-suite.ps1",
+    "test:word:visual": "powershell -NoProfile -ExecutionPolicy Bypass -File scripts/word-com-visual-suite.ps1",
+    "test:corpus:word": "powershell -NoProfile -ExecutionPolicy Bypass -File scripts/word-com-corpus-suite.ps1",
+    "test:corpus:word:visual": "powershell -NoProfile -ExecutionPolicy Bypass -File scripts/word-com-corpus-visual-suite.ps1",
+    "test:visual:inspect": "node scripts/inspect-visual-evidence.mjs",
+    "test:visual:sample": "node scripts/sample-multimodal-visual-check.mjs",
+    "fixtures:reredline": "node scripts/export-reredline-stress-fixtures.mjs",
+    "corpus:fetch:superdoc": "node scripts/fetch-superdoc-corpus.mjs",
+    "report:word:coverage": "node scripts/report-word-coverage.mjs",
+    "report:test:dashboard": "node scripts/build-test-dashboard.mjs",
+    "review:word:prepare": "node scripts/prepare-word-review.mjs",
+    "docx-redline": "node bin/docx-redline.js",
+    "test": "node scripts/run-tests.mjs",
+    "test:isolation": "node tests/no_word_api_index_check.mjs && node tests/core_dependency_graph_check.mjs",
+    "prepublishOnly": "npm run test:isolation && npm run build"
+  },
+  "keywords": [
+    "docx",
+    "ooxml",
+    "reconciliation",
+    "track-changes",
+    "redlines",
+    "word",
+    "office",
+    "document",
+    "xml"
+  ],
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/AnsonLai/docx-redline-js.git"
+  },
+  "engines": {
+    "node": ">=20.0.0"
+  }
+}

package/pipeline/content-analysis.js

@@ -3,7 +3,7 @@
  */
 
 import { ContentType } from '../core/types.js';
-import { matchListMarker, stripListMarker } from './list-markers.js';
+import { parseListItem } from './list-markers.js';
 
 /**
  * Parses table from markdown-style table text.
@@ -60,22 +60,17 @@ export function parseListItems(text) {
     const items = [];
 
     lines.forEach(line => {
-        const markerMatch = matchListMarker(line, { allowZeroSpaceAfterMarker: true });
-        if (!markerMatch) return;
-
-        const marker = markerMatch[2].trim();
-        const indent = (line.match(/^(\s*)/)?.[1].length) || 0;
-        const listType = /^[-*+•]/.test(marker) ? 'bullet' : 'numbered';
-        const outlineDepth = (marker.match(/\./g) || []).length;
-        const level = outlineDepth > 1 ? Math.min(8, outlineDepth - 1) : Math.min(8, Math.floor(indent / 2));
-
-        items.push({
-            line,
-            text: stripListMarker(line, { allowZeroSpaceAfterMarker: true }),
-            marker,
-            indent,
-            level,
-            listType
+        const parsed = parseListItem(line, { allowZeroSpaceAfterMarker: true, indentSpaces: 2 });
+        if (!parsed) return;
+        const level = parsed.outlineLevel ?? parsed.level;
+
+        items.push({
+            line,
+            text: parsed.text,
+            marker: parsed.marker,
+            indent: parsed.indent,
+            level,
+            listType: parsed.listType
         });
     });
 

package/pipeline/ingestion-export.js

@@ -11,6 +11,7 @@
 import { parseOoxmlSafe } from '../adapters/xml-adapter.js';
 import { NS_W } from '../core/types.js';
 import { getXmlParseError } from '../core/xml-query.js';
+import { isNodeVisibleInRevisionView, readCanonicalRunText } from '../core/paragraph-text.js';
 
 function hasParserError(doc) {
     if (!doc || !doc.documentElement) return true;
@@ -54,34 +55,6 @@ function getWordAttribute(element, names) {
     return '';
 }
 
-function hasWordAncestorWithin(node, localName, boundary) {
-    let cursor = node?.parentNode || null;
-    while (cursor && cursor !== boundary) {
-        if (cursor.nodeType === 1 && cursor.namespaceURI === NS_W && cursor.localName === localName) {
-            return true;
-        }
-        cursor = cursor.parentNode;
-    }
-    return false;
-}
-
-function readRunText(run) {
-    let text = '';
-    for (const child of Array.from(run?.childNodes || [])) {
-        if (!child || child.nodeType !== 1 || child.namespaceURI !== NS_W) continue;
-        if (child.localName === 't') {
-            text += child.textContent || '';
-        } else if (child.localName === 'tab') {
-            text += '\t';
-        } else if (child.localName === 'br' || child.localName === 'cr') {
-            text += '\n';
-        } else if (child.localName === 'noBreakHyphen') {
-            text += '\u2011';
-        }
-    }
-    return text;
-}
-
 function getRunFormatting(run) {
     const rPr = getDirectWordChild(run, 'rPr');
     if (!rPr) return { bold: false, italic: false };
@@ -99,9 +72,8 @@ function collectParagraphSegments(paragraph) {
     const segments = [];
     const runs = Array.from(paragraph?.getElementsByTagNameNS?.(NS_W, 'r') || []);
     for (const run of runs) {
-        if (hasWordAncestorWithin(run, 'del', paragraph)) continue;
-        if (hasWordAncestorWithin(run, 'moveFrom', paragraph)) continue;
-        const text = readRunText(run);
+        if (!isNodeVisibleInRevisionView(run, paragraph, 'accepted')) continue;
+        const text = readCanonicalRunText(run, { boundary: paragraph, revisionView: 'accepted' });
         if (!text) continue;
         segments.push({
             text,

package/pipeline/ingestion-paragraph.js

@@ -91,8 +91,11 @@ export function detectNumberingContext(pElement) {
     const ilvlEl = getFirstElementByTagNS(numPr, NS_W, 'ilvl');
     if (!numIdEl) return null;
 
-    const numId = numIdEl.getAttribute('w:val');
-    const type = numId === '1' ? 'bullet' : (numId === '2' ? 'numbered' : 'unknown');
+    const numId = numIdEl.getAttribute('w:val');
+    // In WordprocessingML, numId="0" explicitly removes numbering inherited
+    // from a paragraph style. It is not a reusable list definition.
+    if (!/^\d+$/.test(numId) || Number.parseInt(numId, 10) === 0) return null;
+    const type = numId === '1' ? 'bullet' : (numId === '2' ? 'numbered' : 'unknown');
 
     return {
         numId,
@@ -329,9 +332,11 @@ function processRun(runElement, startOffset) {
             text += '\n';
         } else if (nodeName.endsWith(':tab') || nodeName === 'tab') {
             text += '\t';
-        } else if (nodeName.endsWith(':noBreakHyphen') || nodeName === 'noBreakHyphen') {
-            text += '\u2011';
-        }
+        } else if (nodeName.endsWith(':noBreakHyphen') || nodeName === 'noBreakHyphen') {
+            text += '\u2011';
+        } else if (nodeName.endsWith(':softHyphen') || nodeName === 'softHyphen') {
+            text += '\u00ad';
+        }
     }
 
     if (!text) return null;