pptx2js 0.4.0 → 0.4.3

package/lib/placeholder.js

@@ -2,7 +2,7 @@
  * 母版 / 版式占位符继承（design.html §4.1）
  * 优先级：slide → slideLayout → slideMaster
  */
-const { asArray, attr, child } = require('./xml-utils');
+const { attr, child, children, documentRoot } = require('./xml-utils');
 
 const REL_SLIDE_LAYOUT =
   'http://schemas.openxmlformats.org/officeDocument/2006/relationships/slideLayout';
@@ -69,10 +69,7 @@ function resolveRelByType(relIndex, sourcePath, typeFragment) {
 function getSpTree(parsed, partPath) {
   const doc = parsed[partPath];
   if (!doc) return null;
-  const root =
-    child(doc, 'p:sld') ??
-    child(doc, 'p:sldLayout') ??
-    child(doc, 'p:sldMaster');
+  const root = documentRoot(doc, 'p:sld', 'p:sldLayout', 'p:sldMaster');
   if (!root) return null;
   return child(child(root, 'p:cSld'), 'p:spTree');
 }
@@ -112,12 +109,11 @@ function findSpByPlaceholder(spTree, key) {
  * @returns {object[]}
  */
 function* walkAllSp(spTree) {
-  for (const sp of asArray(spTree['p:sp'])) {
+  for (const sp of children(spTree, 'p:sp')) {
     yield sp;
   }
-  for (const grp of asArray(spTree['p:grpSp'])) {
-    const inner = child(grp, 'p:spTree');
-    if (inner) yield* walkAllSp(inner);
+  for (const grp of children(spTree, 'p:grpSp')) {
+    yield* walkAllSp(grp);
   }
 }
 
@@ -152,10 +148,12 @@ function resolvePlaceholderSps(slideSp, inheritance) {
 function mergeXfrm(slideXfrm, layoutXfrm, masterXfrm) {
   const layers = [slideXfrm, layoutXfrm, masterXfrm];
   const off = mergeXfrmPart(
+    'a:off',
     layers.map((x) => child(x, 'a:off')),
     ['x', 'y']
   );
   const ext = mergeXfrmPart(
+    'a:ext',
     layers.map((x) => child(x, 'a:ext')),
     ['cx', 'cy'],
     { skipZero: true }
@@ -163,18 +161,24 @@ function mergeXfrm(slideXfrm, layoutXfrm, masterXfrm) {
 
   if (!off && !ext) return null;
   return {
-    'a:off': off ?? { $: { x: '0', y: '0' } },
-    'a:ext': ext ?? { $: { cx: '0', cy: '0' } },
+    tag: 'a:xfrm',
+    attrs: {},
+    children: [
+      off ?? makeXfrmPart('a:off', { x: '0', y: '0' }),
+      ext ?? makeXfrmPart('a:ext', { cx: '0', cy: '0' }),
+    ],
+    text: '',
   };
 }
 
 /**
+ * @param {string} tag
  * @param {(object|null|undefined)[]} parts
  * @param {string[]} attrNames
  * @param {{ skipZero?: boolean }} [opts]
- * @returns {object|null}
+ * @returns {import('./xml-parser').OONode|null}
  */
-function mergeXfrmPart(parts, attrNames, opts = {}) {
+function mergeXfrmPart(tag, parts, attrNames, opts = {}) {
   const values = {};
   let hasAny = false;
 
@@ -193,7 +197,29 @@ function mergeXfrmPart(parts, attrNames, opts = {}) {
   }
 
   if (!hasAny) return null;
-  return { $: values };
+  return makeXfrmPart(tag, values);
+}
+
+/**
+ * @param {string} tag
+ * @param {Record<string, string>} attrs
+ * @returns {import('./xml-parser').OONode}
+ */
+function makeXfrmPart(tag, attrs) {
+  return { tag, attrs, children: [], text: '' };
+}
+
+/**
+ * @param {import('./xml-parser').OONode} node
+ * @returns {import('./xml-parser').OONode}
+ */
+function cloneNode(node) {
+  return {
+    tag: node.tag,
+    attrs: { ...node.attrs },
+    children: node.children.map(cloneNode),
+    text: node.text,
+  };
 }
 
 /**
@@ -213,35 +239,137 @@ function getEffectiveXfrm(slideSp, inheritance) {
 }
 
 /**
- * 合并 txBody 默认样式：用 layout/master 的首段/首 run 补全 slide 层缺失的 rPr
+ * @param {import('./xml-parser').OONode} slide p:sld
+ * @returns {boolean}
+ */
+function shouldShowMasterShapes(slide) {
+  const v = attr(slide, 'showMasterSp');
+  if (v === '0' || v === 'false') return false;
+  return true;
+}
+
+/**
+ * 从 txBody 的 lstStyle / endParaRPr 取默认 run 样式
+ * @param {object|null|undefined} txBody
+ * @returns {import('./xml-parser').OONode|null}
+ */
+function getDefaultRPrFromTxBody(txBody) {
+  if (!txBody) return null;
+
+  const lstStyle = child(txBody, 'a:lstStyle');
+  if (lstStyle) {
+    for (let lvl = 1; lvl <= 9; lvl++) {
+      const lvlPPr = child(lstStyle, `a:lvl${lvl}pPr`);
+      const defRPr = lvlPPr && child(lvlPPr, 'a:defRPr');
+      if (defRPr) return defRPr;
+    }
+    const defPPr = child(lstStyle, 'a:defPPr');
+    const defRPr = defPPr && child(defPPr, 'a:defRPr');
+    if (defRPr) return defRPr;
+  }
+
+  const firstP = children(txBody, 'a:p')[0];
+  if (firstP) {
+    const endRPr = child(firstP, 'a:endParaRPr');
+    if (endRPr) return endRPr;
+    const firstR = children(firstP, 'a:r')[0];
+    if (firstR) {
+      const rPr = child(firstR, 'a:rPr');
+      if (rPr) return rPr;
+    }
+  }
+  return null;
+}
+
+/**
+ * @param {import('./xml-parser').OONode|null|undefined} slideRPr
+ * @param {import('./xml-parser').OONode} fallbackRPr
+ * @returns {import('./xml-parser').OONode}
+ */
+function mergeRPrAttrs(slideRPr, fallbackRPr) {
+  if (!fallbackRPr) return slideRPr;
+  if (!slideRPr) return cloneNode(fallbackRPr);
+
+  const merged = cloneNode(slideRPr);
+  for (const [k, v] of Object.entries(fallbackRPr.attrs)) {
+    if (merged.attrs[k] == null || merged.attrs[k] === '') {
+      merged.attrs[k] = v;
+    }
+  }
+  for (const fc of fallbackRPr.children) {
+    if (!child(merged, fc.tag)) {
+      merged.children.push(cloneNode(fc));
+    }
+  }
+  return merged;
+}
+
+/**
+ * 合并 txBody：补全 lstStyle 与 layout/master 的 defRPr
  * @param {object} slideTxBody
  * @param {object|null} layoutSp
  * @param {object|null} masterSp
  * @returns {object}
  */
 function mergeTxBody(slideTxBody, layoutSp, masterSp) {
-  const slideBody = slideTxBody;
   const fallbackBody =
     (layoutSp && child(layoutSp, 'p:txBody')) ||
     (masterSp && child(masterSp, 'p:txBody'));
-  if (!fallbackBody) return slideBody;
 
-  const slideParas = asArray(slideBody['a:p']);
-  const fallbackPara = asArray(fallbackBody['a:p'])[0];
-  const fallbackRPr = fallbackPara && child(asArray(fallbackPara['a:r'])[0], 'a:rPr');
+  const merged = cloneNode(slideTxBody);
+
+  if (fallbackBody) {
+    const fallbackLst = child(fallbackBody, 'a:lstStyle');
+    if (fallbackLst && !child(merged, 'a:lstStyle')) {
+      merged.children.unshift(cloneNode(fallbackLst));
+    }
+  }
 
-  if (!fallbackRPr) return slideBody;
+  const fallbackRPr = fallbackBody
+    ? getDefaultRPrFromTxBody(fallbackBody)
+    : null;
+  if (!fallbackRPr) return merged;
 
-  const mergedParas = slideParas.map((p) => {
-    const runs = asArray(p['a:r']).map((r) => {
+  for (const p of children(merged, 'a:p')) {
+    for (const r of children(p, 'a:r')) {
       const rPr = child(r, 'a:rPr');
-      if (rPr) return r;
-      return { ...r, 'a:rPr': JSON.parse(JSON.stringify(fallbackRPr)) };
-    });
-    return { ...p, 'a:r': runs.length === 1 ? runs[0] : runs };
-  });
+      const mergedRPr = mergeRPrAttrs(rPr, fallbackRPr);
+      if (rPr) {
+        const idx = r.children.findIndex((c) => c.tag === 'a:rPr');
+        r.children[idx] = mergedRPr;
+      } else {
+        r.children.unshift(mergedRPr);
+      }
+    }
+    const endRPr = child(p, 'a:endParaRPr');
+    if (endRPr) {
+      const idx = p.children.findIndex((c) => c.tag === 'a:endParaRPr');
+      p.children[idx] = mergeRPrAttrs(endRPr, fallbackRPr);
+    }
+  }
+  return merged;
+}
+
+/**
+ * 按级别读取 lstStyle 中的 defRPr（lvl1pPr → 索引 0）
+ * @param {object|null|undefined} txBody
+ * @returns {Record<number, import('./xml-parser').OONode>}
+ */
+function getLstStyleDefRPrByLevel(txBody) {
+  /** @type {Record<number, import('./xml-parser').OONode>} */
+  const byLevel = {};
+  const lstStyle = txBody && child(txBody, 'a:lstStyle');
+  if (!lstStyle) return byLevel;
 
-  return { ...slideBody, 'a:p': mergedParas.length === 1 ? mergedParas[0] : mergedParas };
+  for (let lvl = 1; lvl <= 9; lvl++) {
+    const lvlPPr = child(lstStyle, `a:lvl${lvl}pPr`);
+    const defRPr = lvlPPr && child(lvlPPr, 'a:defRPr');
+    if (defRPr) byLevel[lvl - 1] = defRPr;
+  }
+  const defPPr = child(lstStyle, 'a:defPPr');
+  const defRPr = defPPr && child(defPPr, 'a:defRPr');
+  if (defRPr && byLevel[0] == null) byLevel[0] = defRPr;
+  return byLevel;
 }
 
 module.exports = {
@@ -254,5 +382,9 @@ module.exports = {
   mergeXfrm,
   getEffectiveXfrm,
   mergeTxBody,
+  mergeRPrAttrs,
+  getDefaultRPrFromTxBody,
+  getLstStyleDefRPrByLevel,
+  shouldShowMasterShapes,
   getSpTree,
 };

package/lib/presentation.js

@@ -1,8 +1,8 @@
 /**
  * presentation.xml：幻灯片列表与版面尺寸
  */
-const { asArray, attr, child } = require('./xml-utils');
-const { emuToInch } = require('./utils/emu');
+const { attr, child, children, documentRoot } = require('./xml-utils');
+const { emuToInch } = require('./bounds');
 
 const PRESENTATION_PATH = 'ppt/presentation.xml';
 
@@ -13,10 +13,10 @@ const PRESENTATION_PATH = 'ppt/presentation.xml';
  */
 function getSlidePaths(parsed, relIndex) {
   const doc = parsed[PRESENTATION_PATH];
-  const pres = child(doc, 'p:presentation');
+  const pres = documentRoot(doc, 'p:presentation');
   if (!pres) return [];
 
-  const sldIds = asArray(child(child(pres, 'p:sldIdLst'), 'p:sldId'));
+  const sldIds = children(child(pres, 'p:sldIdLst'), 'p:sldId');
   const paths = [];
   for (const sldId of sldIds) {
     const relId = attr(sldId, 'r:id');
@@ -32,7 +32,7 @@ function getSlidePaths(parsed, relIndex) {
  * @returns {{ width: number, height: number }}
  */
 function getSlideSizeInches(parsed) {
-  const pres = child(parsed[PRESENTATION_PATH], 'p:presentation');
+  const pres = documentRoot(parsed[PRESENTATION_PATH], 'p:presentation');
   const sldSz = child(pres, 'p:sldSz');
   const cx = parseInt(attr(sldSz, 'cx') ?? '9144000', 10);
   const cy = parseInt(attr(sldSz, 'cy') ?? '6858000', 10);

package/lib/rels.js

@@ -2,7 +2,7 @@
  * OOXML 关系（.rels）解析与路径解析
  */
 const path = require('path');
-const { asArray, attr } = require('./xml-utils');
+const { attr, children, documentRoot } = require('./xml-utils');
 
 /**
  * @typedef {object} Relationship
@@ -51,11 +51,24 @@ function isExternalTarget(rel) {
 }
 
 function resolveTargetPath(ownerPath, target) {
+  const normalized = String(target).replace(/\\/g, '/');
+
+  // PptxGenJS 等写入的包根绝对路径：/ppt/charts/chart1.xml
+  if (normalized.startsWith('/')) {
+    return path.posix.normalize(normalized.slice(1));
+  }
+
   if (!ownerPath) {
-    return path.posix.normalize(target);
+    return path.posix.normalize(normalized);
   }
+
+  // 已是包内绝对路径（无 ..）：ppt/charts/chart1.xml
+  if (normalized.startsWith('ppt/') && !normalized.includes('..')) {
+    return path.posix.normalize(normalized);
+  }
+
   const ownerDir = path.posix.dirname(ownerPath);
-  return path.posix.normalize(path.posix.join(ownerDir, target));
+  return path.posix.normalize(path.posix.join(ownerDir, normalized));
 }
 
 /**
@@ -63,8 +76,8 @@ function resolveTargetPath(ownerPath, target) {
  * @returns {Relationship[]}
  */
 function parseRelationships(relsDoc) {
-  const root = relsDoc?.Relationships ?? relsDoc;
-  const rels = asArray(root?.Relationship);
+  const root = documentRoot(relsDoc, 'Relationships');
+  const rels = children(root, 'Relationship');
   return rels
     .map((rel) => ({
       id: attr(rel, 'Id'),
@@ -84,7 +97,7 @@ function buildRelationIndex(parsed) {
   const bySource = new Map();
 
   for (const [filePath, doc] of Object.entries(parsed)) {
-    if (!filePath.endsWith('.rels')) continue;
+    if (!/\.rels$/i.test(filePath)) continue;
     const owner = relsOwnerPath(filePath);
     const relationships = parseRelationships(doc);
     const map = new Map();

package/lib/run-utils.js

@@ -0,0 +1,181 @@
+/**
+ * 文本 run 合并与 PptxGenJS 选项压缩（IR → 代码 简化层）
+ */
+
+const DEFAULT_TEXT_COLOR = '000000';
+
+/**
+ * @param {object} a
+ * @param {object} b
+ */
+function optionsEqual(a, b) {
+  return optionsKey(a) === optionsKey(b);
+}
+
+/**
+ * @param {object} opts
+ */
+function optionsKey(opts) {
+  const o = compressRunOptions({ ...opts });
+  return JSON.stringify(o, Object.keys(o).sort());
+}
+
+/**
+ * @param {string} text
+ */
+function isNewlineOnly(text) {
+  return typeof text === 'string' && /^[\r\n\s]*$/.test(text) && /[\r\n]/.test(text);
+}
+
+/**
+ * 剔除 PptxGenJS 默认值，简化 underline
+ * @param {object} opts
+ */
+function compressRunOptions(opts) {
+  if (!opts || typeof opts !== 'object') return {};
+  const copy = { ...opts };
+  delete copy._degraded;
+
+  if (copy.color === DEFAULT_TEXT_COLOR || copy.color === '000000') {
+    delete copy.color;
+  }
+
+  if (copy.underline && typeof copy.underline === 'object') {
+    if ((copy.underline.style ?? 'sng') === 'sng') {
+      copy.underline = true;
+    }
+  }
+
+  return copy;
+}
+
+/**
+ * 合并相邻同样式 run，将独立换行 run 并入前一段文本
+ * @param {Array<{ text: string, options?: object }>} runs
+ */
+function compressTextRuns(runs) {
+  if (!runs?.length) return [];
+
+  /** @type {Array<{ text: string, options: object }>} */
+  const out = [];
+
+  for (const run of runs) {
+    const text = run.text ?? '';
+    const options = compressRunOptions(run.options ?? {});
+
+    if (!text && Object.keys(options).length === 0) continue;
+
+    if (!out.length) {
+      out.push({ text, options });
+      continue;
+    }
+
+    const last = out[out.length - 1];
+
+    if (isNewlineOnly(text)) {
+      const nlOpts = { ...options };
+      delete nlOpts.bullet;
+      if (optionsEqual(last.options, nlOpts) || optionsEqual(last.options, options)) {
+        if (!last.text.endsWith('\n')) {
+          last.text += text.includes('\r') ? '\r\n' : '\n';
+        }
+        continue;
+      }
+    }
+
+    if (optionsEqual(last.options, options)) {
+      last.text += text;
+      continue;
+    }
+
+    out.push({ text, options });
+  }
+
+  while (out.length > 1 && isNewlineOnly(out[out.length - 1].text)) {
+    const tail = out.pop();
+    const prev = out[out.length - 1];
+    if (optionsEqual(prev.options, tail.options) && !prev.text.endsWith('\n')) {
+      prev.text += tail.text.includes('\r') ? '\r\n' : '\n';
+    } else {
+      const nlOpts = { ...tail.options };
+      delete nlOpts.bullet;
+      out.push({ text: tail.text, options: nlOpts });
+      break;
+    }
+  }
+
+  return out.map((run) => {
+    if (!isNewlineOnly(run.text)) return run;
+    const options = { ...run.options };
+    delete options.bullet;
+    return { text: run.text, options };
+  });
+}
+
+/**
+ * 四边相同时合并为 PptxGenJS 简写 border
+ * @param {object|null} border
+ */
+function compressCellBorder(border) {
+  if (!border) return null;
+
+  const sides = ['left', 'right', 'top', 'bottom'];
+  const present = sides.filter((s) => border[s]);
+  if (present.length === 0) return null;
+
+  if (present.length < 4) return border;
+
+  const first = border[present[0]];
+  const allSame = present.every((s) => {
+    const side = border[s];
+    return (
+      (side.color ?? null) === (first.color ?? null) &&
+      (side.pt ?? 1) === (first.pt ?? 1)
+    );
+  });
+
+  if (allSame) {
+    return {
+      type: 'solid',
+      color: first.color,
+      pt: first.pt ?? 1,
+    };
+  }
+
+  return border;
+}
+
+/**
+ * 表格单元格：压缩 runs → 纯文本或内联多 run
+ * @param {Array<{ text: string, options?: object }>} runs
+ */
+function flattenTableCellText(runs) {
+  const merged = compressTextRuns(runs);
+  if (!merged.length) return { text: '', runOptions: {} };
+
+  const allSame =
+    merged.length === 1 ||
+    merged.every((r) => optionsEqual(r.options, merged[0].options));
+
+  if (allSame) {
+    const text = merged.map((r) => r.text).join('').replace(/\r\n/g, '\n');
+    return { text, runOptions: compressRunOptions(merged[0].options) };
+  }
+
+  return {
+    text: merged.map((r) => ({
+      text: String(r.text).replace(/\r\n/g, '\n'),
+      options: compressRunOptions(r.options),
+    })),
+    runOptions: {},
+  };
+}
+
+module.exports = {
+  compressRunOptions,
+  compressTextRuns,
+  compressCellBorder,
+  flattenTableCellText,
+  optionsEqual,
+  isNewlineOnly,
+};

package/lib/smartart.js

@@ -3,7 +3,7 @@
  */
 const { attr, child, textContent } = require('./xml-utils');
 const { getGraphicXfrm } = require('./graphic');
-const { boundsFromXfrm } = require('./utils/bounds');
+const { boundsFromXfrm } = require('./bounds');
 
 /**
  * @param {object} graphicFrame
@@ -74,22 +74,14 @@ function extractSmartArtTexts(graphicFrame, ctx) {
  * @param {string[]} out
  */
 function collectTextNodes(node, out) {
-  if (node == null) return;
-  if (typeof node === 'string') return;
-  if (typeof node !== 'object') return;
-
-  if (node['a:t'] != null) {
-    const t = textContent(node['a:t']);
+  if (!node || typeof node !== 'object') return;
+  if (node.tag === 'a:t') {
+    const t = textContent(node);
     if (t) out.push(t);
+    return;
   }
-
-  for (const [key, val] of Object.entries(node)) {
-    if (key === '$') continue;
-    if (Array.isArray(val)) {
-      for (const item of val) collectTextNodes(item, out);
-    } else if (val && typeof val === 'object') {
-      collectTextNodes(val, out);
-    }
+  for (const c of node.children || []) {
+    collectTextNodes(c, out);
   }
 }
 

package/lib/table.js

@@ -1,11 +1,12 @@
 /**
  * 表格提取（design.html §4.5）
  */
-const { asArray, attr, child, textContent } = require('./xml-utils');
+const { attr, child, children } = require('./xml-utils');
 const { getGraphicXfrm } = require('./graphic');
-const { boundsFromXfrm } = require('./utils/bounds');
-const { resolveFillColor } = require('./utils/color');
-const { emuToInch } = require('./utils/emu');
+const { boundsFromXfrm, emuToInch } = require('./bounds');
+const { resolveFillColor } = require('./color');
+const { compressCellBorder, flattenTableCellText } = require('./run-utils');
+const { extractTextRuns } = require('./text-utils');
 
 /**
  * @param {object} graphicFrame
@@ -49,23 +50,42 @@ function extractTable(graphicFrame, ctx) {
  */
 function extractTableRows(tbl, scheme) {
   const rows = [];
-  for (const tr of asArray(tbl['a:tr'])) {
+  for (const tr of children(tbl, 'a:tr')) {
     const cells = [];
-    for (const tc of asArray(tr['a:tc'])) {
-      const text = extractCellText(tc);
+    for (const tc of children(tr, 'a:tc')) {
       const tcPr = child(tc, 'a:tcPr');
+      const txBody = child(tc, 'a:txBody');
+      let text = '';
+      /** @type {object} */
+      let runOptions = {};
+      if (txBody) {
+        const flat = flattenTableCellText(extractTextRuns(txBody, scheme, null));
+        text = flat.text;
+        runOptions = flat.runOptions;
+      }
+
       const fill = tcPr ? resolveFillColor(tcPr, scheme).color : null;
+      const borderRaw = extractCellBorderSides(tcPr, scheme);
+      const borderColors = borderRaw
+        ? Object.values(borderRaw)
+            .map((b) => b.color)
+            .filter(Boolean)
+        : [];
+      const border = compressCellBorder(borderRaw);
+      const fillIsJustBorderColor =
+        fill &&
+        borderColors.length > 0 &&
+        borderColors.every((c) => c === fill);
+
       const merge = {};
       const rowSpan = attr(tcPr, 'rowSpan');
       const gridSpan = attr(tcPr, 'gridSpan');
       if (rowSpan && rowSpan !== '1') merge.rowspan = parseInt(rowSpan, 10);
       if (gridSpan && gridSpan !== '1') merge.colspan = parseInt(gridSpan, 10);
 
-      const cell = { text, options: {} };
-      if (fill) cell.options.fill = { color: fill };
+      const cell = { text, options: { ...runOptions } };
+      if (fill && !fillIsJustBorderColor) cell.options.fill = fill;
       if (Object.keys(merge).length) Object.assign(cell.options, merge);
-
-      const border = extractCellBorder(tcPr, scheme);
       if (border) cell.options.border = border;
 
       cells.push(cell);
@@ -79,7 +99,7 @@ function extractTableRows(tbl, scheme) {
  * @param {object|null|undefined} tcPr
  * @param {Record<string, string>} scheme
  */
-function extractCellBorder(tcPr, scheme) {
+function extractCellBorderSides(tcPr, scheme) {
   if (!tcPr) return null;
   const border = {};
   const sides = {
@@ -106,28 +126,13 @@ function extractCellBorder(tcPr, scheme) {
   return hasAny ? border : null;
 }
 
-/**
- * @param {object} tc
- */
-function extractCellText(tc) {
-  const txBody = child(tc, 'a:txBody');
-  if (!txBody) return '';
-  const parts = [];
-  for (const p of asArray(txBody['a:p'])) {
-    for (const r of asArray(p['a:r'])) {
-      parts.push(textContent(r['a:t']));
-    }
-  }
-  return parts.join('');
-}
-
 /**
  * @param {object} tbl
  * @returns {number[]|undefined}
  */
 function extractColWidths(tbl) {
   const grid = child(tbl, 'a:tblGrid');
-  const cols = asArray(child(grid, 'a:gridCol'));
+  const cols = children(grid, 'a:gridCol');
   if (!cols.length) return undefined;
   return cols.map((col) => {
     const w = parseInt(attr(col, 'w') ?? '0', 10);