@ansonlai/docx-redline-js 0.4.0 → 0.5.0

package/adapters/config.js

@@ -1,43 +1,45 @@
-/**
- * Configurable runtime defaults for the reconciliation core.
- * Callers can set these during bootstrap.
- */
-
-let _defaultAuthor = 'Author';
-let _platform = 'Unknown';
-
-/**
- * Set the default track-change author for revision metadata.
- *
- * @param {string} author
- */
-export function setDefaultAuthor(author) {
-    _defaultAuthor = typeof author === 'string' && author.trim() ? author.trim() : 'Author';
-}
-
-/**
- * Get the current default track-change author.
- *
- * @returns {string}
- */
-export function getDefaultAuthor() {
-    return _defaultAuthor;
-}
-
-/**
- * Set the platform identifier (e.g. 'Win32', 'Mac', 'OfficeOnline').
- *
- * @param {string} platform
- */
-export function setPlatform(platform) {
-    _platform = typeof platform === 'string' && platform.trim() ? platform.trim() : 'Unknown';
-}
-
-/**
- * Get the current platform identifier.
- *
- * @returns {string}
- */
-export function getPlatform() {
-    return _platform;
-}
+/**
+ * Configurable runtime defaults for the reconciliation core.
+ * Callers can set these during bootstrap.
+ */
+
+const fallbackAuthor = () => (typeof process !== 'undefined' && process.env?.DOCX_REDLINE_AUTHOR) || 'AI Redliner';
+
+let _defaultAuthor = fallbackAuthor();
+let _platform = 'Unknown';
+
+/**
+ * Set the default track-change author for revision metadata.
+ *
+ * @param {string} author
+ */
+export function setDefaultAuthor(author) {
+    _defaultAuthor = typeof author === 'string' && author.trim() ? author.trim() : fallbackAuthor();
+}
+
+/**
+ * Get the current default track-change author.
+ *
+ * @returns {string}
+ */
+export function getDefaultAuthor() {
+    return _defaultAuthor;
+}
+
+/**
+ * Set the platform identifier (e.g. 'Win32', 'Mac', 'OfficeOnline').
+ *
+ * @param {string} platform
+ */
+export function setPlatform(platform) {
+    _platform = typeof platform === 'string' && platform.trim() ? platform.trim() : 'Unknown';
+}
+
+/**
+ * Get the current platform identifier.
+ *
+ * @returns {string}
+ */
+export function getPlatform() {
+    return _platform;
+}

package/bin/docx-redline.js

@@ -0,0 +1,3 @@
+#!/usr/bin/env node
+import { runCli } from '../node/cli.js';
+process.exitCode = await runCli();

package/core/list-targeting.js

@@ -7,6 +7,7 @@ import {
     getParagraphText,
     normalizeWhitespaceForTargeting
 } from './paragraph-targeting.js';
+import { parseListItem, stripListMarker } from '../pipeline/list-markers.js';
 
 function getFirstDescendantByLocalName(node, localName) {
     if (!node || typeof node.getElementsByTagNameNS !== 'function') return null;
@@ -25,60 +26,50 @@ function readValAttribute(element) {
     return element.getAttribute('w:val') || element.getAttribute('val') || null;
 }
 
-function parseOutlineLevelFromMarker(marker) {
-    const normalized = String(marker || '').trim();
-    if (!/^\d+(?:\.\d+)+\.?$/.test(normalized)) return null;
-    const parts = normalized.replace(/\.$/, '').split('.');
-    return Math.max(0, parts.length - 1);
-}
-
-const REDUNDANT_LIST_PREFIX_REGEX = /^(?:(?:\d+(?:\.\d+)*\.?|\((?:\d+|[a-zA-Z]|[ivxlcIVXLC]+)\)|[a-zA-Z]\.|[ivxlcIVXLC]+\.|[-*+\u2022]))\s+/;
-
-/**
- * Strips redundant manual list markers from the start of list item text.
- *
- * Useful when model output contains doubled markers like:
- * - "2.1. - Item text"
- * - "- 2.1. Item text"
- *
- * @param {string} text - Candidate list item text
- * @returns {string}
- */
-export function stripRedundantLeadingListMarkers(text) {
-    let value = String(text || '').trim();
-    let passes = 0;
-    while (passes < 4 && REDUNDANT_LIST_PREFIX_REGEX.test(value)) {
-        value = value.replace(REDUNDANT_LIST_PREFIX_REGEX, '').trimStart();
-        passes++;
-    }
-    return value.trim();
-}
-
-function parseModifiedListItems(modifiedText) {
-    const rawLines = String(modifiedText || '').split(/\r?\n/g);
-    const items = [];
-    let hasListMarkers = false;
+/**
+ * Strips redundant manual list markers from the start of list item text.
+ *
+ * Useful when model output contains doubled markers like:
+ * - "2.1. - Item text"
+ * - "- 2.1. Item text"
+ *
+ * @param {string} text - Candidate list item text
+ * @returns {string}
+ */
+export function stripRedundantLeadingListMarkers(text) {
+    let value = String(text || '').trim();
+    let passes = 0;
+    while (passes < 4) {
+        const stripped = stripListMarker(value);
+        if (stripped === value) break;
+        value = stripped.trimStart();
+        passes++;
+    }
+    return value.trim();
+}
+
+function parseModifiedListItems(modifiedText) {
+    const rawLines = String(modifiedText || '').split(/\r?\n/g);
+    const items = [];
+    let hasListMarkers = false;
 
     for (const rawLine of rawLines) {
         const line = rawLine.trimEnd();
         if (!line.trim()) continue;
 
-        const markerMatch = line.match(/^(\s*)((?:\d+(?:\.\d+)*\.?|\((?:\d+|[a-zA-Z]|[ivxlcIVXLC]+)\)|[a-zA-Z]\.|[ivxlcIVXLC]+\.|[-*+\u2022]))\s+(.*)$/);
-        if (markerMatch) {
+        const parsed = parseListItem(line, { indentSpaces: 2 });
+        if (parsed) {
             hasListMarkers = true;
-            const marker = markerMatch[2];
-            const markerType = /^[-*+\u2022]$/.test(marker) ? 'bullet' : 'numbered';
-            const level = Math.floor((markerMatch[1] || '').length / 2);
-            items.push({
-                kind: 'list',
-                markerType,
-                level,
-                marker,
-                outlineLevel: markerType === 'numbered' ? parseOutlineLevelFromMarker(marker) : null,
-                text: stripRedundantLeadingListMarkers(markerMatch[3])
-            });
-            continue;
-        }
+            items.push({
+                kind: 'list',
+                markerType: parsed.markerType,
+                level: parsed.level,
+                marker: parsed.marker,
+                outlineLevel: parsed.outlineLevel,
+                text: stripRedundantLeadingListMarkers(parsed.text)
+            });
+            continue;
+        }
 
         items.push({
             kind: 'text',
@@ -99,46 +90,46 @@ function isNormalizedTextEqual(a, b) {
     return normalizeWhitespaceForTargeting(a) === normalizeWhitespaceForTargeting(b);
 }
 
-function resolveInsertionLevel(item, anchorLevel, baselineLevel) {
-    if (Number.isInteger(item?.outlineLevel)) {
-        return Math.max(0, item.outlineLevel);
-    }
-    return Math.max(0, anchorLevel + ((item?.level || 0) - baselineLevel));
-}
-
-function shouldPromoteBulletInsertionsToChildDepth(parsedItems, normalizedTargetText, anchorLevel) {
-    if (!Array.isArray(parsedItems) || parsedItems.length < 2) return false;
-    if (!Number.isInteger(anchorLevel) || anchorLevel < 1) return false;
-
-    const firstItem = parsedItems[0];
-    const trailingListItems = parsedItems.slice(1).filter(item => item.kind === 'list');
-    if (trailingListItems.length === 0) return false;
-    if (trailingListItems.some(item => item.markerType !== 'bullet')) return false;
-    if (trailingListItems.some(item => Number.isInteger(item.outlineLevel))) return false;
-
-    if (firstItem?.kind === 'text') {
-        return isNormalizedTextEqual(firstItem.text, normalizedTargetText);
-    }
-
-    if (firstItem?.kind === 'list' && firstItem.markerType === 'numbered') {
-        const firstLevel = firstItem.level || 0;
-        const alreadyIndented = trailingListItems.some(item => (item.level || 0) > firstLevel);
-        if (alreadyIndented) return false;
-        return isNormalizedTextEqual(firstItem.text, normalizedTargetText);
-    }
-
-    return false;
-}
-
-function promoteBulletInsertionsToChildDepth(entries, anchorLevel) {
-    return entries.map(entry => {
-        const relativeDepth = Math.max(0, (entry.ilvl || 0) - anchorLevel);
-        return {
-            ...entry,
-            ilvl: Math.min(8, anchorLevel + 1 + relativeDepth)
-        };
-    });
-}
+function resolveInsertionLevel(item, anchorLevel, baselineLevel) {
+    if (Number.isInteger(item?.outlineLevel)) {
+        return Math.max(0, item.outlineLevel);
+    }
+    return Math.max(0, anchorLevel + ((item?.level || 0) - baselineLevel));
+}
+
+function shouldPromoteBulletInsertionsToChildDepth(parsedItems, normalizedTargetText, anchorLevel) {
+    if (!Array.isArray(parsedItems) || parsedItems.length < 2) return false;
+    if (!Number.isInteger(anchorLevel) || anchorLevel < 0) return false;
+
+    const firstItem = parsedItems[0];
+    const trailingListItems = parsedItems.slice(1).filter(item => item.kind === 'list');
+    if (trailingListItems.length === 0) return false;
+    if (trailingListItems.some(item => item.markerType !== 'bullet')) return false;
+    if (trailingListItems.some(item => Number.isInteger(item.outlineLevel))) return false;
+
+    if (firstItem?.kind === 'text') {
+        return isNormalizedTextEqual(firstItem.text, normalizedTargetText);
+    }
+
+    if (firstItem?.kind === 'list' && firstItem.markerType === 'numbered') {
+        const firstLevel = firstItem.level || 0;
+        const alreadyIndented = trailingListItems.some(item => (item.level || 0) > firstLevel);
+        if (alreadyIndented) return false;
+        return isNormalizedTextEqual(firstItem.text, normalizedTargetText);
+    }
+
+    return false;
+}
+
+function promoteBulletInsertionsToChildDepth(entries, anchorLevel) {
+    return entries.map(entry => {
+        const relativeDepth = Math.max(0, (entry.ilvl || 0) - anchorLevel);
+        return {
+            ...entry,
+            ilvl: Math.min(8, anchorLevel + 1 + relativeDepth)
+        };
+    });
+}
 
 function buildListEntriesForInsertion(parsedItems, normalizedTargetText, anchorLevel, defaultMarkerType) {
     const firstItem = parsedItems[0];
@@ -195,7 +186,7 @@ export function getParagraphListInfo(paragraph) {
     if (!numIdEl) return null;
 
     const numId = readValAttribute(numIdEl);
-    if (!numId) return null;
+    if (!numId || numId === '0') return null;
 
     const ilvlEl = getFirstDescendantByLocalName(numPr, 'ilvl');
     const ilvlRaw = readValAttribute(ilvlEl);
@@ -361,7 +352,7 @@ export function synthesizeExpandedListScopeEdit(targetParagraph, modifiedText, o
  * }} [options] - Optional context/log callbacks
  * @returns {{ targetParagraph: Element, numId: string, entries: Array<{ ilvl: number, text: string, markerType: 'bullet'|'numbered' }> }|null}
  */
-export function planListInsertionOnlyEdit(targetParagraph, modifiedText, options = {}) {
+export function planListInsertionOnlyEdit(targetParagraph, modifiedText, options = {}) {
     const onInfo = typeof options.onInfo === 'function' ? options.onInfo : () => {};
     const onWarn = typeof options.onWarn === 'function' ? options.onWarn : () => {};
 
@@ -374,26 +365,26 @@ export function planListInsertionOnlyEdit(targetParagraph, modifiedText, options
     const parsed = parseModifiedListItems(rawModified);
     if (!parsed.hasListMarkers || parsed.items.length < 2) return null;
 
-    const normalizedTargetText = normalizeWhitespaceForTargeting(
-        options.currentParagraphText || getParagraphText(targetParagraph)
-    );
-    const listItemsOnly = parsed.items.filter(item => item.kind === 'list');
-    const defaultMarkerType = listItemsOnly[0]?.markerType || 'bullet';
-    const anchorLevel = Math.max(0, targetListInfo.ilvl);
-    let entries = buildListEntriesForInsertion(parsed.items, normalizedTargetText, anchorLevel, defaultMarkerType);
-
-    if (!entries || entries.length === 0) {
-        onWarn('[List] Could not derive insertion-only entries from multiline list edit.');
-        return null;
-    }
-
-    if (shouldPromoteBulletInsertionsToChildDepth(parsed.items, normalizedTargetText, anchorLevel)) {
-        entries = promoteBulletInsertionsToChildDepth(entries, anchorLevel);
-        onInfo('[List] Promoted bullet insertion to child depth for nested numbered-list intent.');
-    }
-
-    onInfo('[List] Planned insertion-only list redline entries (no block rewrite).');
-    return {
+    const normalizedTargetText = normalizeWhitespaceForTargeting(
+        options.currentParagraphText || getParagraphText(targetParagraph)
+    );
+    const listItemsOnly = parsed.items.filter(item => item.kind === 'list');
+    const defaultMarkerType = listItemsOnly[0]?.markerType || 'bullet';
+    const anchorLevel = Math.max(0, targetListInfo.ilvl);
+    let entries = buildListEntriesForInsertion(parsed.items, normalizedTargetText, anchorLevel, defaultMarkerType);
+
+    if (!entries || entries.length === 0) {
+        onWarn('[List] Could not derive insertion-only entries from multiline list edit.');
+        return null;
+    }
+
+    if (shouldPromoteBulletInsertionsToChildDepth(parsed.items, normalizedTargetText, anchorLevel)) {
+        entries = promoteBulletInsertionsToChildDepth(entries, anchorLevel);
+        onInfo('[List] Promoted bullet insertion to child depth for nested numbered-list intent.');
+    }
+
+    onInfo('[List] Planned insertion-only list redline entries (no block rewrite).');
+    return {
         targetParagraph,
         numId: targetListInfo.numId,
         entries