@ansonlai/docx-redline-js 0.2.0 → 0.4.0

package/pipeline/list-generation.js

@@ -30,9 +30,10 @@ export async function executeListGeneration(options) {
         originalRunModel = [],
         originalText = '',
         generateRedlines = true,
-        author = 'AI',
-        font = null,
-        numberingService
+        author = 'AI',
+        font = null,
+        revisionIdAllocator = null,
+        numberingService
     } = options;
 
     const normalizedListText = normalizeCompositeListMarkers(cleanText);
@@ -72,13 +73,18 @@ export async function executeListGeneration(options) {
             if (tableData.headers.length > 0 || tableData.rows.length > 0) {
                 if (generateRedlines && results.length === 0 && deletionRuns.length > 0) {
                     results.push(serializeToOoxml(deletionRuns, null, [], {
-                        author,
-                        generateRedlines,
-                        font
-                    }));
+                        author,
+                        generateRedlines,
+                        font,
+                        revisionIdAllocator
+                    }));
                 }
 
-                results.push(generateTableOoxml(tableData, { generateRedlines, author }));
+                results.push(generateTableOoxml(tableData, {
+                    generateRedlines,
+                    author,
+                    revisionIdAllocator
+                }));
                 i = tableBlock.endIndex;
                 continue;
             }
@@ -92,9 +98,10 @@ export async function executeListGeneration(options) {
             numberingContext,
             numberingService,
             generateRedlines,
-            author,
-            font,
-            deletionRuns
+            author,
+            font,
+            revisionIdAllocator,
+            deletionRuns
         );
         results.push(entry.ooxml);
     }
@@ -226,9 +233,10 @@ function buildListEntry(
     numberingContext,
     numberingService,
     generateRedlines,
-    author,
-    font,
-    deletionRuns
+    author,
+    font,
+    revisionIdAllocator,
+    deletionRuns
 ) {
     let pPrXml = '';
     let segmentText = '';
@@ -272,9 +280,10 @@ function buildListEntry(
 
     return {
         ooxml: serializeToOoxml(runModel, pPrXml, formatHints, {
-            author,
-            generateRedlines,
-            font
-        })
+            author,
+            generateRedlines,
+            font,
+            revisionIdAllocator
+        })
     };
 }

package/pipeline/patching.js

@@ -182,9 +182,8 @@ export function applyPatches(splitModel, diffOps, options) {
 
 function processInsertionOperation(context) {
     const {
-        insertOp,
-        splitModel,
-        styleLookup,
+        insertOp,
+        styleLookup,
         patchedModel,
         state,
         options,

package/pipeline/pipeline.js

@@ -16,7 +16,7 @@ import { detectNumberingContext } from './ingestion.js';
 import { generateTableOoxml } from '../services/table-reconciliation.js';
 import { executeListGeneration, detectIndentationStep } from './list-generation.js';
 import { detectContentType, parseListItems, parseTable } from './content-analysis.js';
-import { createParser } from '../adapters/xml-adapter.js';
+import { parseOoxmlSafe } from '../adapters/xml-adapter.js';
 import { log, error as logError } from '../adapters/logger.js';
 import { getFirstElementByTagNS, getXmlParseError } from '../core/xml-query.js';
 import { getPlatform } from '../adapters/config.js';
@@ -54,6 +54,7 @@ export class ReconciliationPipeline {
         this.validationMode = options.validationMode ?? 'auto';
         this.numberingService = options.numberingService || new NumberingService();
         this.font = options.font || null;
+        this.revisionIdAllocator = options.revisionIdAllocator || null;
         this.platform = options.platform ?? getPlatform();
         this.isWebPlatform = options.isWebPlatform ?? isWebPlatform(this.platform);
         this.enableEventLoopYielding = options.enableEventLoopYielding ?? this.isWebPlatform;
@@ -75,10 +76,20 @@ export class ReconciliationPipeline {
 
         try {
             // Stage 1: Ingest OOXML
-            const doc = options.xmlDoc || (() => {
-                const parser = createParser();
-                return parser.parseFromString(originalOoxml, 'application/xml');
-            })();
+            const parsed = options.xmlDoc
+                ? { doc: options.xmlDoc, error: null, warnings: [] }
+                : parseOoxmlSafe(originalOoxml, 'application/xml');
+            if (parsed.error || !parsed.doc) {
+                return {
+                    ooxml: originalOoxml,
+                    isValid: false,
+                    status: 'error',
+                    error: parsed.error,
+                    warnings: parsed.warnings || []
+                };
+            }
+            warnings.push(...(parsed.warnings || []));
+            const doc = parsed.doc;
             const pElement = getFirstElementByTagNS(doc, '*', 'p');
 
             const { runModel, acceptedText, pPr } = ingestOoxml(originalOoxml, { xmlDoc: doc });
@@ -122,19 +133,28 @@ export class ReconciliationPipeline {
             }
             await this.maybeYield(runModel.length, Math.max(acceptedText.length, cleanText.length));
 
-            // Count actual paragraph elements ingested
-            const paragraphCount = runModel.filter(r => r.kind === RunKind.PARAGRAPH_START).length;
-
+            // Count actual paragraph elements ingested
+            const paragraphCount = runModel.filter(r => r.kind === RunKind.PARAGRAPH_START).length;
+            const sourceIsMarkedMultilineList = isListTargetStrict(acceptedText) || isListTargetLoose(acceptedText);
+            const canPatchExistingMarkedList = isTargetList
+                && sourceIsMarkedMultilineList
+                && paragraphCount > 1
+                && acceptedText !== cleanText;
+
             log(`[Reconcile] isTargetList: ${isTargetList}, paragraphCount: ${paragraphCount}`);
-
-            // If target is a list, always use list generation logic
-            // This handles both expansion (1 para -> N items) and conversion (N paras -> M items)
-            if (isTargetList) {
-                log('[Reconcile] 🎯 ENTERING LIST GENERATION PATH');
-                log(`[Reconcile] cleanText preview: ${cleanText.substring(0, 100)}...`);
-                log(`[Reconcile] acceptedText preview: ${acceptedText.substring(0, 100)}...`);
-                return this.executeListGeneration(cleanText, numberingContext, runModel);
-            }
+
+            // Preserve already-marked multi-paragraph lists during ordinary text edits.
+            // Rebuilding the whole block would discard unchanged run formatting and
+            // place all source deletion runs in the first generated paragraph, which
+            // makes rejection merge the original paragraph boundaries.
+            if (canPatchExistingMarkedList) {
+                log('[Reconcile] Existing marked list edit detected; using run-aware patching to preserve formatting and paragraph boundaries.');
+            } else if (isTargetList) {
+                log('[Reconcile] 🎯 ENTERING LIST GENERATION PATH');
+                log(`[Reconcile] cleanText preview: ${cleanText.substring(0, 100)}...`);
+                log(`[Reconcile] acceptedText preview: ${acceptedText.substring(0, 100)}...`);
+                return this.executeListGeneration(cleanText, numberingContext, runModel);
+            }
 
             log(`[Reconcile] Computed ${diffOps.length} diff operations`);
 
@@ -153,9 +173,10 @@ export class ReconciliationPipeline {
             await this.maybeYield(patchedModel.length, Math.max(acceptedText.length, cleanText.length));
 
             // Stage 6: Serialize to OOXML
-            const resultOoxml = serializeToOoxml(patchedModel, pPr, formatHints, {
-                author: this.author,
-                generateRedlines: this.generateRedlines
+            const resultOoxml = serializeToOoxml(patchedModel, pPr, formatHints, {
+                author: this.author,
+                generateRedlines: this.generateRedlines,
+                revisionIdAllocator: this.revisionIdAllocator
             });
 
             // Stage 7: Basic validation
@@ -172,14 +193,15 @@ export class ReconciliationPipeline {
                 warnings
             };
 
-        } catch (error) {
-            logError('[Reconcile] Pipeline error:', error);
-            return {
-                ooxml: originalOoxml,
-                isValid: false,
-                warnings: [`Pipeline error: ${error.message}`]
-            };
-        }
+        } catch (error) {
+            logError('[Reconcile] Pipeline error:', error);
+            return {
+                ooxml: originalOoxml,
+                isValid: false,
+                warnings: [`Pipeline error: ${error.message}`],
+                error: error?.code ? { code: error.code, message: error.message } : undefined
+            };
+        }
     }
 
     /**
@@ -194,8 +216,13 @@ export class ReconciliationPipeline {
         try {
             // Check for well-formed XML by wrapping in namespace container
             const wrappedXml = `<root xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main" xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships">${ooxml}</root>`;
-            const parser = createParser();
-            const doc = parser.parseFromString(wrappedXml, 'application/xml');
+            const parsed = parseOoxmlSafe(wrappedXml, 'application/xml');
+            const doc = parsed.doc;
+
+            if (parsed.error || !doc) {
+                errors.push('Generated OOXML is not well-formed XML: ' + (parsed.error?.message || 'parse error'));
+                return { isValid: false, errors };
+            }
 
             const parseError = getXmlParseError(doc);
             if (parseError) {
@@ -276,9 +303,10 @@ export class ReconciliationPipeline {
             originalRunModel,
             originalText,
             generateRedlines: this.generateRedlines,
-            author: this.author,
-            font: this.font,
-            numberingService: this.numberingService
+            author: this.author,
+            font: this.font,
+            revisionIdAllocator: this.revisionIdAllocator,
+            numberingService: this.numberingService
         });
     }
 
@@ -309,9 +337,10 @@ export class ReconciliationPipeline {
             };
         }
 
-        const tableOoxml = generateTableOoxml(tableData, {
-            generateRedlines: this.generateRedlines,
-            author: this.author
+        const tableOoxml = generateTableOoxml(tableData, {
+            generateRedlines: this.generateRedlines,
+            author: this.author,
+            revisionIdAllocator: this.revisionIdAllocator
         });
 
         return {

package/pipeline/serialization.js

@@ -24,7 +24,7 @@ const XMLNS_ATTR_REGEX = /\s+xmlns:[^=]+="[^"]*"/g;
  */
 export function serializeToOoxml(patchedModel, pPr, formatHints = [], options = {}) {
     const serializationOptions = normalizeSerializationOptions(options);
-    const { author, generateRedlines } = serializationOptions;
+    const { generateRedlines } = serializationOptions;
     const paragraphs = [];
     let currentPPrXml = '';
     let currentPPrElement = null;
@@ -140,7 +140,8 @@ function normalizeSerializationOptions(options) {
         return {
             author: getDefaultAuthor(),
             generateRedlines: true,
-            font: options
+            font: options,
+            revisionIdAllocator: null
         };
     }
 
@@ -152,7 +153,8 @@ function normalizeSerializationOptions(options) {
     return {
         author: resolvedAuthor,
         generateRedlines: normalized.generateRedlines ?? true,
-        font: normalized.font ?? null
+        font: normalized.font ?? null,
+        revisionIdAllocator: normalized.revisionIdAllocator ?? null
     };
 }
 
@@ -228,7 +230,10 @@ function buildSimpleRun(text, rPrXml) {
  * @returns {string}
  */
 function buildDeletionXml(item, options = {}) {
-    const metadata = createRevisionMetadata(options.author ?? getDefaultAuthor());
+    const metadata = createRevisionMetadata(
+        options.author ?? getDefaultAuthor(),
+        options.revisionIdAllocator
+    );
     const font = options.font ?? null;
     let rPr = item.rPrXml ? stripNamespaceDeclarations(item.rPrXml) : '';
 
@@ -250,7 +255,10 @@ function buildDeletionXml(item, options = {}) {
  * @returns {string}
  */
 function buildInsertionXml(item, formatHints, options = {}) {
-    const metadata = createRevisionMetadata(options.author ?? getDefaultAuthor());
+    const metadata = createRevisionMetadata(
+        options.author ?? getDefaultAuthor(),
+        options.revisionIdAllocator
+    );
     const font = options.font ?? null;
 
     // Build the inner run content with format hints

package/scripts/build-test-dashboard.mjs

@@ -0,0 +1,43 @@
+import { execFileSync } from 'child_process';
+import { existsSync, readFileSync } from 'fs';
+import { join, resolve } from 'path';
+
+const repoRoot = process.cwd();
+const syntheticDir = join(repoRoot, 'tmp', 'dashboard-docx');
+const corpusSourceDir = join(repoRoot, 'tmp', 'superdoc-corpus');
+const corpusFixturesArgIndex = process.argv.indexOf('--corpus-fixtures-dir');
+if (corpusFixturesArgIndex >= 0 && !process.argv[corpusFixturesArgIndex + 1]) {
+    throw new Error('--corpus-fixtures-dir requires a path');
+}
+const suppliedCorpusFixturesDir = corpusFixturesArgIndex >= 0;
+const corpusFixturesDir = suppliedCorpusFixturesDir
+    ? resolve(repoRoot, process.argv[corpusFixturesArgIndex + 1])
+    : join(repoRoot, 'tmp', 'superdoc-word-fixtures');
+const manifest = JSON.parse(readFileSync(
+    join(repoRoot, 'tests', 'corpus', 'superdoc-english-legal-administrative.json'),
+    'utf8'
+));
+const run = (script, args = []) => execFileSync(process.execPath, [script, ...args], {
+    cwd: repoRoot,
+    stdio: 'inherit'
+});
+
+run('scripts/export-validation-fixtures.mjs', ['--output-dir', syntheticDir]);
+
+const corpusReady = manifest.documents.every(item =>
+    existsSync(join(corpusSourceDir, `${item.id}.docx`))
+);
+if (corpusReady && suppliedCorpusFixturesDir) {
+    if (!existsSync(join(corpusFixturesDir, 'suite.json'))) {
+        throw new Error(`Supplied corpus fixture directory has no suite.json: ${corpusFixturesDir}`);
+    }
+} else if (corpusReady) {
+    run('scripts/prepare-superdoc-word-corpus.mjs');
+} else {
+    console.warn('Real-document corpus is not downloaded; embedding synthetic DOCX previews only.');
+    console.warn('Run npm run test:corpus:word once to fetch and validate the pinned corpus.');
+}
+
+const args = ['scripts/generate-test-dashboard.mjs', '--fixtures-dir', syntheticDir];
+if (corpusReady) args.push('--corpus-fixtures-dir', corpusFixturesDir);
+run(args[0], args.slice(1));

package/scripts/check-types.mjs

@@ -1,29 +1,21 @@
-import { readFileSync } from 'fs';
+import { readFileSync } from 'node:fs';
 
-const dts = readFileSync(new URL('../index.d.ts', import.meta.url), 'utf8');
-
-const requiredSnippets = [
-  'export interface RedlineOptions',
-  'export interface RedlineResult',
-  'export function applyRedlineToOxml',
-  'export function acceptTrackedChangesInOoxml',
-  'export function rejectTrackedChangesInOoxml',
-  'export function deleteCommentsByAuthorInOoxml',
-  'export function validateRedlineOoxml'
-];
-
-for (const snippet of requiredSnippets) {
-  if (!dts.includes(snippet)) {
-    throw new Error(`Missing declaration snippet: ${snippet}`);
-  }
+const declarationPath = new URL('../index.d.ts', import.meta.url);
+const declarationText = readFileSync(declarationPath, 'utf8');
+const declaredRuntimeNames = new Set();
+const declarationPattern = /^export\s+(?:declare\s+)?(?:function|class|const|let|var|enum)\s+([A-Za-z_$][\w$]*)/gm;
+for (const match of declarationText.matchAll(declarationPattern)) {
+    declaredRuntimeNames.add(match[1]);
 }
 
-let balance = 0;
-for (const char of dts) {
-  if (char === '{') balance += 1;
-  if (char === '}') balance -= 1;
-  if (balance < 0) throw new Error('index.d.ts has unbalanced braces');
+const runtimeModule = await import('../index.js');
+const runtimeNames = Object.keys(runtimeModule).sort();
+const missingDeclarations = runtimeNames.filter(name => !declaredRuntimeNames.has(name));
+
+if (missingDeclarations.length > 0) {
+    throw new Error(
+        `Runtime exports missing from index.d.ts:\n${missingDeclarations.map(name => `- ${name}`).join('\n')}`
+    );
 }
-if (balance !== 0) throw new Error('index.d.ts has unbalanced braces');
 
-console.log('PASS: index.d.ts declaration smoke check');
+console.log(`PASS: ${runtimeNames.length} runtime exports have declarations`);