@ansonlai/docx-redline-js 0.8.4 → 0.10.0

package/pipeline/ingestion-paragraph.js

@@ -5,8 +5,9 @@
  */
 
 import { NS_W, RunKind, ContainerKind } from '../core/types.js';
+import { getCurrentParagraphProperties } from '../core/current-properties.js';
 import { appendParagraphBoundary, advanceOffsetForParagraphBoundary } from '../core/paragraph-offset-policy.js';
-import { parseOoxmlSafe, serializeXml } from '../adapters/xml-adapter.js';
+import { parseOoxmlSafe, serializeXml } from '../adapters/xml-adapter.js';
 import { warn, error as logError } from '../adapters/logger.js';
 import {
     getElementsByTagNS,
@@ -21,29 +22,29 @@ let containerIdCounter = 0;
 /**
  * Ingests OOXML paragraph content and builds a run-aware text model.
  *
- * @param {string} ooxmlString - OOXML input
- * @param {{ xmlDoc?: Document|null }} [options={}] - Optional pre-parsed XML document
- * @returns {import('../core/types.js').IngestionResult}
- */
-export function ingestOoxml(ooxmlString, options = {}) {
-    const runModel = [];
-    let acceptedText = '';
-    const preParsedDoc = options.xmlDoc || null;
-
-    if (!ooxmlString && !preParsedDoc) {
-        return { runModel, acceptedText, pPr: null };
-    }
-
-    try {
-        const parsed = preParsedDoc ? { doc: preParsedDoc, error: null } : parseOoxmlSafe(ooxmlString, 'application/xml');
-        const doc = parsed.doc;
-        if (parsed.error || !doc) {
-            logError('OOXML parse error:', parsed.error?.message);
-            return { runModel, acceptedText, pPr: null, error: parsed.error };
-        }
-
-        const parseError = getXmlParseError(doc);
-        if (parseError) {
+ * @param {string} ooxmlString - OOXML input
+ * @param {{ xmlDoc?: Document|null }} [options={}] - Optional pre-parsed XML document
+ * @returns {import('../core/types.js').IngestionResult}
+ */
+export function ingestOoxml(ooxmlString, options = {}) {
+    const runModel = [];
+    let acceptedText = '';
+    const preParsedDoc = options.xmlDoc || null;
+
+    if (!ooxmlString && !preParsedDoc) {
+        return { runModel, acceptedText, pPr: null };
+    }
+
+    try {
+        const parsed = preParsedDoc ? { doc: preParsedDoc, error: null } : parseOoxmlSafe(ooxmlString, 'application/xml');
+        const doc = parsed.doc;
+        if (parsed.error || !doc) {
+            logError('OOXML parse error:', parsed.error?.message);
+            return { runModel, acceptedText, pPr: null, error: parsed.error };
+        }
+
+        const parseError = getXmlParseError(doc);
+        if (parseError) {
             logError('OOXML parse error:', parseError.textContent);
             return { runModel, acceptedText, pPr: null };
         }
@@ -81,52 +82,46 @@ export function ingestParagraphElement(paragraphElement) {
  * @returns {Object|null}
  */
 export function detectNumberingContext(pElement) {
-    const pPr = getFirstElementByTagNS(pElement, NS_W, 'pPr');
-    if (!pPr) return null;
-
-    const numPr = getFirstElementByTagNS(pPr, NS_W, 'numPr');
-    if (!numPr) return null;
-
-    const numIdEl = getFirstElementByTagNS(numPr, NS_W, 'numId');
-    const ilvlEl = getFirstElementByTagNS(numPr, NS_W, 'ilvl');
-    if (!numIdEl) return null;
+    // Current numPr only: a w:pPrChange snapshot must not make a historical list binding look current.
+    const { numPr } = getCurrentParagraphProperties(pElement);
+    if (!numPr || !numPr.numIdElement) return null;
 
-    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+$/.test(numId) || Number.parseInt(numId, 10) === 0) return null;
-    const type = numId === '1' ? 'bullet' : (numId === '2' ? 'numbered' : 'unknown');
+    const numId = numPr.numId;
+    // In WordprocessingML, numId="0" explicitly removes numbering inherited
+    // from a paragraph style. It is not a reusable list definition.
+    if (!/^\d+$/.test(numId) || Number.parseInt(numId, 10) === 0) return null;
+    const type = numId === '1' ? 'bullet' : (numId === '2' ? 'numbered' : 'unknown');
 
     return {
         numId,
-        ilvl: parseInt(ilvlEl?.getAttribute('w:val') || '0', 10),
+        ilvl: parseInt(numPr.ilvl || '0', 10),
         type
     };
 }
 
-function ingestParagraphElements(paragraphs, options = {}) {
+function ingestParagraphElements(paragraphs, options = {}) {
     const includeParagraphBoundaries = options.includeParagraphBoundaries ?? true;
     const runModel = [];
     let acceptedText = '';
     let currentOffset = 0;
     let firstPPr = null;
 
-    for (let pIndex = 0; pIndex < paragraphs.length; pIndex++) {
-        const pElement = paragraphs[pIndex];
-        const pPr = getFirstElementByTagNS(pElement, NS_W, 'pPr');
-
-        if (pIndex === 0) {
-            firstPPr = pPr;
-        }
-
-        runModel.push({
-            kind: RunKind.PARAGRAPH_START,
-            // Keep pPr as a DOM reference; serialize only when downstream actually needs string XML.
-            pPrElement: pPr || null,
-            startOffset: currentOffset,
-            endOffset: currentOffset,
-            text: ''
-        });
+    for (let pIndex = 0; pIndex < paragraphs.length; pIndex++) {
+        const pElement = paragraphs[pIndex];
+        const pPr = getFirstElementByTagNS(pElement, NS_W, 'pPr');
+
+        if (pIndex === 0) {
+            firstPPr = pPr;
+        }
+
+        runModel.push({
+            kind: RunKind.PARAGRAPH_START,
+            // Keep pPr as a DOM reference; serialize only when downstream actually needs string XML.
+            pPrElement: pPr || null,
+            startOffset: currentOffset,
+            endOffset: currentOffset,
+            text: ''
+        });
 
         const result = processNodeRecursive(pElement, currentOffset, runModel);
         acceptedText += result.text;
@@ -222,36 +217,36 @@ function getNodeHandlers(runModel) {
         return contentResult;
     });
 
-    handlers.set('del', (child, offset) => {
-        const deletionEntry = processDeletion(child, offset);
-        if (deletionEntry) {
-            runModel.push(deletionEntry);
-        }
-        return { offset, text: '' };
-    });
-
-    handlers.set('moveFrom', (child, offset) => {
-        const deletionEntry = processDeletion(child, offset);
-        if (deletionEntry) {
-            runModel.push(deletionEntry);
-        }
-        return { offset, text: '' };
-    });
-
-    handlers.set('moveTo', (child, offset) => processNodeRecursive(child, offset, runModel));
-
-    for (const markerName of ['moveFromRangeStart', 'moveFromRangeEnd', 'moveToRangeStart', 'moveToRangeEnd']) {
-        handlers.set(markerName, (child, offset) => {
-            runModel.push({
-                kind: RunKind.BOOKMARK,
-                nodeXml: serializeXml(child),
-                startOffset: offset,
-                endOffset: offset,
-                text: ''
-            });
-            return { offset, text: '' };
-        });
-    }
+    handlers.set('del', (child, offset) => {
+        const deletionEntry = processDeletion(child, offset);
+        if (deletionEntry) {
+            runModel.push(deletionEntry);
+        }
+        return { offset, text: '' };
+    });
+
+    handlers.set('moveFrom', (child, offset) => {
+        const deletionEntry = processDeletion(child, offset);
+        if (deletionEntry) {
+            runModel.push(deletionEntry);
+        }
+        return { offset, text: '' };
+    });
+
+    handlers.set('moveTo', (child, offset) => processNodeRecursive(child, offset, runModel));
+
+    for (const markerName of ['moveFromRangeStart', 'moveFromRangeEnd', 'moveToRangeStart', 'moveToRangeEnd']) {
+        handlers.set(markerName, (child, offset) => {
+            runModel.push({
+                kind: RunKind.BOOKMARK,
+                nodeXml: serializeXml(child),
+                startOffset: offset,
+                endOffset: offset,
+                text: ''
+            });
+            return { offset, text: '' };
+        });
+    }
 
     handlers.set('bookmarkStart', (child, offset) => {
         runModel.push({
@@ -332,11 +327,11 @@ 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(':softHyphen') || nodeName === 'softHyphen') {
-            text += '\u00ad';
-        }
+        } else if (nodeName.endsWith(':noBreakHyphen') || nodeName === 'noBreakHyphen') {
+            text += '\u2011';
+        } else if (nodeName.endsWith(':softHyphen') || nodeName === 'softHyphen') {
+            text += '\u00ad';
+        }
     }
 
     if (!text) return null;
@@ -350,13 +345,13 @@ function processRun(runElement, startOffset) {
     };
 }
 
-function processDeletion(delElement, offset) {
-    const author = delElement.getAttribute('w:author') || '';
-
-    // Deleted text is retained as a zero-width deletion model entry for revision-aware callers.
-    // It is intentionally not added to acceptedText by the w:del handler above.
-    let text = '';
-    const delTexts = getElementsByTagNS(delElement, NS_W, 'delText');
+function processDeletion(delElement, offset) {
+    const author = delElement.getAttribute('w:author') || '';
+
+    // Deleted text is retained as a zero-width deletion model entry for revision-aware callers.
+    // It is intentionally not added to acceptedText by the w:del handler above.
+    let text = '';
+    const delTexts = getElementsByTagNS(delElement, NS_W, 'delText');
     for (const delText of delTexts) {
         text += delText.textContent || '';
     }