@coco-vexel/coco-office-oojson 0.0.0-stage → 1.0.0

package/src/ooxml-drawing-to-json.mjs

@@ -0,0 +1,209 @@
+const LINE_PRESETS = new Set([
+  "line", "straightConnector1",
+  "bentConnector2", "bentConnector3", "bentConnector4", "bentConnector5",
+  "curvedConnector2", "curvedConnector3", "curvedConnector4", "curvedConnector5",
+]);
+
+function findNode(node, tag) {
+  if (!node || typeof node !== "object") return null;
+  if (node.tag === tag) return node;
+  for (const child of node.kids || []) {
+    const found = findNode(child, tag);
+    if (found) return found;
+  }
+  return null;
+}
+
+function nodeText(node) {
+  if (!node || typeof node !== "object") return "";
+  if (node.text != null) return String(node.text);
+  return (node.kids || []).map(nodeText).join("");
+}
+
+function numberValue(value) {
+  const n = Number(value);
+  return Number.isFinite(n) ? n : undefined;
+}
+
+function boolValue(value, fallback = false) {
+  if (value == null || value === "") return fallback;
+  return ["1", "true", "t", "on"].includes(String(value).toLowerCase());
+}
+
+function point(node, tag) {
+  const value = findNode(node, tag);
+  return { x: numberValue(value?.attrs?.x) ?? 0, y: numberValue(value?.attrs?.y) ?? 0 };
+}
+
+function extent(node, tag) {
+  const value = findNode(node, tag);
+  return { cx: numberValue(value?.attrs?.cx) ?? 0, cy: numberValue(value?.attrs?.cy) ?? 0 };
+}
+
+function position(container, tag, defaultRelativeFrom) {
+  const value = findNode(container, tag);
+  return {
+    relativeFrom: String(value?.attrs?.relativeFrom ?? defaultRelativeFrom),
+    posOffset: numberValue(nodeText(findNode(value, "wp:posOffset"))) ?? 0,
+    percent: false,
+  };
+}
+
+function solidFill(node) {
+  const hex = String(findNode(node, "a:srgbClr")?.attrs?.val ?? "").trim();
+  if (!/^[0-9a-f]{6}$/i.test(hex)) return null;
+  const rgba = {
+    red: Number.parseInt(hex.slice(0, 2), 16),
+    green: Number.parseInt(hex.slice(2, 4), 16),
+    blue: Number.parseInt(hex.slice(4, 6), 16),
+    alpha: 255,
+  };
+  return {
+    fill: {
+      color: { rgba, color: { rgba, type: "srgb" }, mods: [], type: "uniColor" },
+      type: "solid",
+    },
+    transparent: null,
+    type: "fill",
+    fillType: "solid",
+  };
+}
+
+function presetGeometry(preset) {
+  return {
+    ahLst: { ahPolar: [], ahXY: [] },
+    avLst: {},
+    adjLst: {},
+    cnxLst: [
+      { pos: { x: "l", y: "t" }, ang: "cd4" },
+      { pos: { x: "r", y: "b" }, ang: "_3cd4" },
+    ],
+    gdLst: [],
+    pathLst: [
+      {
+        commands: [
+          { id: "moveTo", pt: { x: "l", y: "t" } },
+          { id: "lnTo", pt: { x: "r", y: "b" } },
+        ],
+        extrusionOk: false,
+        fill: "norm",
+        stroke: true,
+      },
+    ],
+    rect: null,
+    preset,
+  };
+}
+
+function lineToJson(drawing) {
+  const preset = String(findNode(drawing, "a:prstGeom")?.attrs?.prst ?? "");
+  if (!LINE_PRESETS.has(preset)) return null;
+  const anchor = findNode(drawing, "wp:anchor");
+  const inline = findNode(drawing, "wp:inline");
+  const container = anchor || inline;
+  if (!container) return null;
+
+  const docPr = findNode(container, "wp:docPr");
+  const effect = findNode(container, "wp:effectExtent");
+  const outerExtent = extent(container, "wp:extent");
+  const xfrm = findNode(drawing, "a:xfrm");
+  const innerExtent = extent(xfrm, "a:ext");
+  const line = findNode(drawing, "a:ln");
+  const simplePos = findNode(container, "wp:simplePos");
+  const id = numberValue(docPr?.attrs?.id) ?? 1;
+  const metadata = {
+    id,
+    name: String(docPr?.attrs?.name ?? `Line ${id}`),
+    descr: docPr?.attrs?.descr == null ? null : String(docPr.attrs.descr),
+    title: docPr?.attrs?.title == null ? null : String(docPr.attrs.title),
+    hidden: null,
+  };
+  const shapeMetadata = {
+    id: -1,
+    name: "",
+    descr: null,
+    title: null,
+    hidden: null,
+  };
+  const connectorLocks = findNode(drawing, "a:cxnSpLocks");
+
+  return {
+    type: "paraDrawing",
+    docPr: metadata,
+    effectExtent: {
+      l: numberValue(effect?.attrs?.l) ?? 0, t: numberValue(effect?.attrs?.t) ?? 0,
+      r: numberValue(effect?.attrs?.r) ?? 0, b: numberValue(effect?.attrs?.b) ?? 0,
+    },
+    extent: outerExtent,
+    graphic: {
+      bWordShape: true,
+      extX: innerExtent.cx || outerExtent.cx,
+      extY: innerExtent.cy || outerExtent.cy,
+      nvCxnSpPr: {
+        cNvPr: shapeMetadata,
+        nvPr: { isPhoto: null, userDrawn: null },
+        cNvCxnSpPr: {
+          cxnSpLocks: {
+            noChangeShapeType: boolValue(
+              connectorLocks?.attrs?.noChangeShapeType,
+              true,
+            ),
+          },
+        },
+      },
+      spPr: {
+        custGeom: presetGeometry(preset), effectDag: null, effectLst: null,
+        ln: {
+          fill: solidFill(line),
+          lineJoin: findNode(line, "a:round") ? { type: "round", lim: null } : null,
+          algn: null, cap: line?.attrs?.cap ?? null, cmpd: line?.attrs?.cmpd ?? null,
+          w: numberValue(line?.attrs?.w) ?? 0,
+          headEnd: null,
+          prstDash: String(findNode(line, "a:prstDash")?.attrs?.val ?? "solid"),
+          tailEnd: null, type: "stroke",
+        },
+        fill: {
+          fill: { type: "noFill" }, transparent: null, type: "fill", fillType: "noFill",
+        },
+        xfrm: {
+          ext: { cx: innerExtent.cx || outerExtent.cx, cy: innerExtent.cy },
+          off: point(xfrm, "a:off"),
+          flipH: xfrm?.attrs?.flipH == null ? null : boolValue(xfrm.attrs.flipH),
+          flipV: xfrm?.attrs?.flipV == null ? null : boolValue(xfrm.attrs.flipV),
+          rot: numberValue(xfrm?.attrs?.rot) ?? null,
+          chOffX: null, chOffY: null, chExtX: null, chExtY: null,
+        },
+        bwMode: 0,
+      },
+      content: null,
+      modelId: null,
+      type: "connectShape",
+      id: String(id),
+    },
+    cNvGraphicFramePr: { graphicFrameLocks: {} },
+    positionH: position(container, "wp:positionH", "column"),
+    positionV: position(container, "wp:positionV", "paragraph"),
+    simplePos: {
+      x: numberValue(simplePos?.attrs?.x) ?? 0,
+      y: numberValue(simplePos?.attrs?.y) ?? 0,
+      use: boolValue(container.attrs?.simplePos, false),
+    },
+    distB: numberValue(container.attrs?.distB) ?? 0,
+    distL: numberValue(container.attrs?.distL) ?? 0,
+    distR: numberValue(container.attrs?.distR) ?? 0,
+    distT: numberValue(container.attrs?.distT) ?? 0,
+    allowOverlap: boolValue(container.attrs?.allowOverlap, true),
+    behindDoc: boolValue(container.attrs?.behindDoc, false),
+    hidden: null,
+    layoutInCell: boolValue(container.attrs?.layoutInCell, true),
+    locked: boolValue(container.attrs?.locked, false),
+    relativeHeight: numberValue(container.attrs?.relativeHeight) ?? 0,
+    wrapType: "none",
+    drawingType: anchor ? "anchor" : "inline",
+  };
+}
+
+export function ooxmlDrawingToJson(node) {
+  const drawing = node?.tag === "w:drawing" ? node : findNode(node, "w:drawing");
+  return drawing ? lineToJson(drawing) : null;
+}

package/src/ooxml-to-json.mjs

@@ -0,0 +1,305 @@
+// 工具②：OOXML(WordprocessingML body) → ONLYOFFICE-JSON。
+// 写链路用：agent 改完的 OOXML → 本工具转 ONLYOFFICE-JSON → FromJSON+RemoveAll+InsertContent 灌回编辑器。
+// FromJSON 严格：段落必须有 changes:[]、顶层必须有 paraPr。
+// 无损保证：每个属性袋/元素 = { ...oo:x 兜底, ...本侧模型解析 }，模型键覆盖兜底（反映 agent 编辑）。
+import { parseXml, kid, kidsOf } from "./xml.mjs";
+import { colorToJson, highlightToJson, mergeAutoRgb } from "./color.mjs";
+import { readExtra } from "./extra.mjs";
+import { ooxmlDrawingToJson } from "./ooxml-drawing-to-json.mjs";
+
+// 数值用 Number 而非 parseInt：ONLYOFFICE 宽度等可为小数(如 tcW.w=22.36)，parseInt 会截断丢精度。
+const a2i = (v) => (v == null || v === "" ? null : Number(v));
+
+// 缺口收集：解析时凡遇本侧未解析、且未被兜底属性覆盖的元素即记入（理论上应为空）。
+export class Gaps {
+  constructor() { this.counts = new Map(); }
+  add(key) { this.counts.set(key, (this.counts.get(key) || 0) + 1); }
+  merge(other) { for (const [k, v] of other.counts) this.counts.set(k, (this.counts.get(k) || 0) + v); }
+  top(n = 50) { return [...this.counts.entries()].sort((a, b) => b[1] - a[1]).slice(0, n); }
+  get size() { return this.counts.size; }
+}
+
+function xmlShd(node) {
+  if (!node) return undefined;
+  const o = { val: node.attrs["w:val"], color: mergeAutoRgb(colorToJson(node.attrs["w:color"]), node.attrs["oo:crgb"]) };
+  if (node.attrs["w:fill"]) o.fill = mergeAutoRgb(colorToJson(node.attrs["w:fill"]), node.attrs["oo:frgb"]);
+  return { ...readExtra(node), ...o }; // themeColor/themeFill 等经 oo:x 兜底
+}
+function xmlPpToJson(pPrNode, gaps) {
+  const o = { bFromDocument: true, type: "paraPr" };
+  if (!pPrNode) return o;
+  const g = (t) => kid(pPrNode, t);
+  if (g("w:pStyle")) o.pStyle = g("w:pStyle").attrs["w:val"];
+  for (const t of ["widowControl", "keepNext", "keepLines", "pageBreakBefore", "contextualSpacing", "suppressLineNumbers"])
+    if (g("w:" + t)) o[t] = true;
+  if (g("w:shd")) o.shd = xmlShd(g("w:shd"));
+  const sp = g("w:spacing");
+  if (sp) {
+    o.spacing = {};
+    if (sp.attrs["w:before"]) o.spacing.before = a2i(sp.attrs["w:before"]);
+    if (sp.attrs["w:after"]) o.spacing.after = a2i(sp.attrs["w:after"]);
+    if (sp.attrs["w:line"]) o.spacing.line = a2i(sp.attrs["w:line"]);
+    if (sp.attrs["w:lineRule"]) o.spacing.lineRule = sp.attrs["w:lineRule"];
+    if (sp.attrs["w:beforeAutospacing"]) o.spacing.beforeAutoSpacing = sp.attrs["w:beforeAutospacing"] === "1" ? "on" : "off";
+    if (sp.attrs["w:afterAutospacing"]) o.spacing.afterAutoSpacing = sp.attrs["w:afterAutospacing"] === "1" ? "on" : "off";
+  }
+  const ind = g("w:ind");
+  if (ind) {
+    o.ind = {};
+    if (ind.attrs["w:left"]) o.ind.left = a2i(ind.attrs["w:left"]);
+    if (ind.attrs["w:right"]) o.ind.right = a2i(ind.attrs["w:right"]);
+    if (ind.attrs["w:firstLine"]) o.ind.firstLine = a2i(ind.attrs["w:firstLine"]);
+  }
+  if (g("w:jc")) o.jc = g("w:jc").attrs["w:val"];
+  return { ...readExtra(pPrNode), ...o };
+}
+function xmlRpToJson(rPrNode, gaps) {
+  const o = { bFromDocument: true, type: "textPr" };
+  if (!rPrNode) return o;
+  const g = (t) => kid(rPrNode, t);
+  const rf = g("w:rFonts");
+  if (rf) { o.rFonts = {}; for (const k of ["ascii", "hAnsi", "eastAsia", "cs", "hint", "asciiTheme", "hAnsiTheme", "eastAsiaTheme", "cstheme"]) if (rf.attrs["w:" + k]) o.rFonts[k] = rf.attrs["w:" + k]; }
+  const tgl = (name) => { const e = g("w:" + name); if (!e) return undefined; const v = e.attrs["w:val"]; return !(v === "false" || v === "0"); };
+  for (const t of ["b", "bCs", "i", "iCs"]) { const v = tgl(t); if (v !== undefined) o[t] = v; }
+  if (g("w:color")) o.color = mergeAutoRgb(colorToJson(g("w:color").attrs["w:val"]), g("w:color").attrs["oo:rgb"]);
+  if (g("w:spacing")) o.spacing = a2i(g("w:spacing").attrs["w:val"]);
+  if (g("w:sz")) o.sz = a2i(g("w:sz").attrs["w:val"]);
+  if (g("w:szCs")) o.szCs = a2i(g("w:szCs").attrs["w:val"]);
+  if (g("w:shd")) o.shd = xmlShd(g("w:shd"));
+  // highlight 不在此解析：OOXML 仅 16 命名色，非精确 RGB 会被最近邻映射丢精度；由 rPr 的 oo:x 兜底精确还原。
+  if (g("w:vertAlign")) o.vertAlign = g("w:vertAlign").attrs["w:val"];
+  const lang = g("w:lang");
+  if (lang) { o.lang = {}; for (const k of ["val", "eastAsia", "bidi"]) if (lang.attrs["w:" + k]) o.lang[k] = lang.attrs["w:" + k]; }
+  return { ...readExtra(rPrNode), ...o };
+}
+function runToJson(rNode, gaps) {
+  const content = [];
+  for (const c of rNode.kids) {
+    // w:delText：被 <w:del> 追踪删除的文字载体（语义同 w:t，只是位于删除修订里）。
+    if (c.tag === "w:t" || c.tag === "w:delText") content.push((c.kids || []).filter((k) => k.text != null).map((k) => k.text).join(""));
+    else if (c.tag === "w:br") { const ty = c.attrs["w:type"]; content.push({ type: "break", breakType: ty === "column" ? "column" : "page" }); }
+    else if (c.tag === "w:tab") content.push({ type: "tab" });
+    else if (c.tag === "oo:ci") { try { content.push(JSON.parse(c.attrs["oo:v"])); } catch { /* 兜底损坏则跳过 */ } }
+    else if (c.tag === "w:drawing" || c.tag === "mc:AlternateContent") {
+      const drawing = ooxmlDrawingToJson(c);
+      if (drawing) content.push(drawing);
+      else gaps && gaps.add("w:r>" + c.tag);
+    } else if (c.tag && c.tag !== "w:rPr") gaps && gaps.add("w:r>" + c.tag);
+  }
+  const ex = readExtra(rNode);
+  return { ...ex, bFromDocument: true, rPr: xmlRpToJson(kid(rNode, "w:rPr"), gaps), content, footnotes: ex.footnotes || [], endnotes: ex.endnotes || [], reviewType: ex.reviewType || "common", type: "run" };
+}
+// <w:ins>/<w:del>（及 move 的 <w:moveTo>/<w:moveFrom>）是 run 的**修订包裹器**：包住一段
+// 被插入/删除的 run。ONLYOFFICE-JSON 不用包裹器，而是给每个 run 打 reviewType + reviewInfo。
+// 这里把包裹器拆开、递归内部 run 并标注——否则 SDK 的 trackChanges（校对）产出的真实
+// <w:ins>/<w:del> 会被当未建模元素整个丢掉 → 插入/删除的文字在写回时全部丢失。
+const REV_TAG_TYPE = { "w:ins": "add", "w:del": "remove", "w:moveTo": "add", "w:moveFrom": "remove" };
+function revInfoFromEl(el) {
+  return {
+    userId: "",
+    author: el.attrs["w:author"] || "",
+    date: el.attrs["w:date"] || "",
+    moveType: "noMove",
+    prevType: -1,
+  };
+}
+/** 把一个修订包裹器里的 run 递归收进 content（每个 run 标 reviewType/reviewInfo）。支持一层嵌套（move）。 */
+function pushRevisionRuns(el, reviewType, reviewInfo, content, gaps) {
+  for (const c of el.kids) {
+    if (c.tag === "w:r") content.push({ ...runToJson(c, gaps), reviewType, reviewInfo });
+    else if (REV_TAG_TYPE[c.tag]) pushRevisionRuns(c, REV_TAG_TYPE[c.tag], revInfoFromEl(c), content, gaps);
+    else if (c.tag) gaps && gaps.add(el.tag + ">" + c.tag);
+  }
+}
+function hyperlinkToJson(node, gaps, rels) {
+  const content = [];
+  for (const c of node.kids || []) {
+    if (c.tag === "w:r") content.push(runToJson(c, gaps));
+    else if (REV_TAG_TYPE[c.tag]) pushRevisionRuns(c, REV_TAG_TYPE[c.tag], revInfoFromEl(c), content, gaps);
+    else if (c.tag === "oo:pi") { try { content.push(JSON.parse(c.attrs["oo:v"])); } catch { /* skip invalid fallback payload */ } }
+    else if (c.tag) gaps && gaps.add("w:hyperlink>" + c.tag);
+  }
+  const relId = node.attrs["r:id"];
+  const rel = relId && rels && typeof rels.get === "function" ? rels.get(relId) : null;
+  const value = rel?.target || node.attrs["w:anchor"] || "";
+  const tooltip = node.attrs["w:tooltip"] || "";
+  return { ...readExtra(node), ...(tooltip ? { tooltip } : {}), ...(value ? { value } : {}), content, type: "hyperlink" };
+}
+
+function applyParagraphTextPr(item, paragraphRPr) {
+  if (!item || !paragraphRPr) return item;
+  if (item.type === "run") {
+    return {
+      ...item,
+      rPr: {
+        ...(item.rPr || {}),
+        ...paragraphRPr,
+        rFonts: {
+          ...(item.rPr?.rFonts || {}),
+          ...(paragraphRPr.rFonts || {}),
+        },
+      },
+    };
+  }
+  if (item.type === "hyperlink" && Array.isArray(item.content)) {
+    return {
+      ...item,
+      content: item.content.map((child) => applyParagraphTextPr(child, paragraphRPr)),
+    };
+  }
+  return item;
+}
+
+function paraToJson(pNode, gaps, rels) {
+  const pPrNode = kid(pNode, "w:pPr");
+  // 段落 mark rPr 在 <w:pPr><w:rPr> 里：既是段落 rPr，也是末尾 endRun 的 rPr
+  const markRpr = xmlRpToJson(pPrNode && kid(pPrNode, "w:rPr"), gaps);
+  // 按文档顺序走内容项：run / oo:pi(commentRangeStart 等未建模项)
+  const content = [];
+  for (const c of pNode.kids) {
+    if (c.tag === "w:r") content.push(runToJson(c, gaps));
+    else if (c.tag === "w:hyperlink") content.push(hyperlinkToJson(c, gaps, rels));
+    else if (REV_TAG_TYPE[c.tag]) pushRevisionRuns(c, REV_TAG_TYPE[c.tag], revInfoFromEl(c), content, gaps);
+    else if (c.tag === "oo:pi") { try { content.push(JSON.parse(c.attrs["oo:v"])); } catch { /* 兜底损坏则跳过 */ } }
+    else if (c.tag === "w:commentRangeStart") { content.push({ type: "commentRangeStart", id: Number(c.attrs["w:id"]) }); }
+    else if (c.tag === "w:commentRangeEnd") { content.push({ type: "commentRangeEnd", id: Number(c.attrs["w:id"]) }); }
+    else if (c.tag === "w:commentReference") { content.push({ type: "commentReference", id: Number(c.attrs["w:id"]) }); }
+    else if (c.tag && c.tag !== "w:pPr") gaps && gaps.add("w:p>" + c.tag);
+  }
+  // 补回段落标记 endRun（ONLYOFFICE 每段 content 末尾都有）；oo:er 还原其修订等非默认字段。
+  // WPS stores effective paragraph text formatting in pPr/rPr while leaving stale,
+  // partial values on individual runs. ONLYOFFICE FromJSON does not apply the
+  // paragraph-level textPr to those runs, so materialize the effective properties.
+  for (let i = 0; i < content.length; i += 1) {
+    content[i] = applyParagraphTextPr(content[i], markRpr);
+  }
+  const er = readExtra(pNode, "oo:er");
+  content.push({ bFromDocument: true, footnotes: [], endnotes: [], reviewType: "common", ...er, rPr: markRpr, content: [], type: "endRun" });
+  const ex = readExtra(pNode);
+  return {
+    ...ex,
+    bFromDocument: true,
+    pPr: xmlPpToJson(pPrNode, gaps),
+    rPr: markRpr,
+    content,
+    changes: ex.changes || [],
+    type: "paragraph",
+  };
+}
+
+// 块级内容数组(body 或单元格)：w:p→段落、w:tbl→表格、oo:blk→未建模块(blockLvlSdt 等)无损还原。
+function blocksFromKids(kids, gaps, rels) {
+  const out = [];
+  for (const c of kids || []) {
+    if (c.tag === "w:p") out.push(paraToJson(c, gaps, rels));
+    else if (c.tag === "w:tbl") out.push(tableToJson(c, gaps, rels));
+    else if (c.tag === "oo:blk") { try { out.push(JSON.parse(c.attrs["oo:v"])); } catch { /* 兜底损坏则跳过 */ } }
+    else if (c.tag && c.tag !== "w:sectPr" && c.tag !== "w:tcPr" && c.tag !== "w:trPr" && c.tag !== "w:tblPr" && c.tag !== "w:tblGrid") gaps && gaps.add("blk>" + c.tag);
+  }
+  return out;
+}
+
+// ── 表格 ──
+const BDR_SIDES = ["top", "left", "bottom", "right", "insideH", "insideV", "start", "end", "between", "bar"];
+function xmlBdrEl(node) {
+  const o = { color: mergeAutoRgb(colorToJson(node.attrs["w:color"]), node.attrs["oo:rgb"]), sz: a2i(node.attrs["w:sz"]) || 0, space: a2i(node.attrs["w:space"]) || 0, value: node.attrs["w:val"] };
+  return { ...readExtra(node), ...o }; // themeColor 等经 oo:x 兜底
+}
+function xmlBorders(node) {
+  if (!node) return undefined;
+  const o = {};
+  for (const c of node.kids) { const s = (c.tag || "").replace(/^w:/, ""); if (BDR_SIDES.includes(s)) o[s] = xmlBdrEl(c); }
+  return o; // 空元素 → {}（保留"显式无边框"语义，不丢键）
+}
+function xmlWidth(node) {
+  if (!node) return undefined;
+  return { type: node.attrs["w:type"] || "auto", w: a2i(node.attrs["w:w"]) || 0 };
+}
+function xmlMar(node) {
+  if (!node) return undefined;
+  const o = {};
+  for (const c of node.kids) { const s = (c.tag || "").replace(/^w:/, ""); if (["top", "left", "bottom", "right", "start", "end"].includes(s)) o[s] = xmlWidth(c); }
+  return Object.keys(o).length ? o : undefined;
+}
+function tblPrToJson(node, gaps) {
+  const o = { type: "tablePr" };
+  if (!node) return o;
+  const g = (t) => kid(node, t);
+  if (g("w:tblStyle")) o.tblStyle = g("w:tblStyle").attrs["w:val"];
+  if (g("w:tblW")) o.tblW = xmlWidth(g("w:tblW"));
+  if (g("w:jc")) o.jc = g("w:jc").attrs["w:val"];
+  if (g("w:tblBorders")) o.tblBorders = xmlBorders(g("w:tblBorders"));
+  if (g("w:tblCellMar")) o.tblCellMar = xmlMar(g("w:tblCellMar"));
+  if (g("w:tblLayout")) o.tblLayout = g("w:tblLayout").attrs["w:type"];
+  if (g("w:tblLook")) { const a = g("w:tblLook").attrs; o.tblLook = {}; for (const k of ["firstColumn", "firstRow", "lastColumn", "lastRow", "noHBand", "noVBand"]) if (a["w:" + k] != null) o.tblLook[k] = a["w:" + k] === "1" || a["w:" + k] === "true"; }
+  // tblInd(数字而非宽度对象)/tblOverlap/tblpPr/inline 等 → 由 oo:x 兜底，不在此解析以免写错形态
+  return { ...readExtra(node), ...o };
+}
+function tcPrToJson(node, gaps) {
+  const o = { type: "tableCellPr" };
+  if (!node) return o;
+  const g = (t) => kid(node, t);
+  if (g("w:tcW")) o.tcW = xmlWidth(g("w:tcW"));
+  if (g("w:gridSpan")) o.gridSpan = a2i(g("w:gridSpan").attrs["w:val"]);
+  if (g("w:vMerge")) o.vMerge = g("w:vMerge").attrs["w:val"] || "continue";
+  if (g("w:tcBorders")) o.tcBorders = xmlBorders(g("w:tcBorders"));
+  if (g("w:shd")) o.shd = xmlShd(g("w:shd"));
+  if (g("w:vAlign")) o.vAlign = g("w:vAlign").attrs["w:val"];
+  return { ...readExtra(node), ...o };
+}
+function trPrToJson(node, gaps) {
+  const o = { type: "tableRowPr" };
+  if (!node) return o;
+  const g = (t) => kid(node, t);
+  if (g("w:tblHeader")) o.tblHeader = true;
+  if (g("w:jc")) o.jc = g("w:jc").attrs["w:val"];
+  if (g("w:trHeight")) o.trHeight = { val: a2i(g("w:trHeight").attrs["w:val"]), hRule: g("w:trHeight").attrs["w:hRule"] };
+  return { ...readExtra(node), ...o };
+}
+let _cellId = 1;
+function cellToJson(tc, gaps, rels) {
+  const blocks = blocksFromKids(tc.kids, gaps, rels);
+  const inner = blocks.length ? blocks : [paraToJson({ tag: "w:p", attrs: {}, kids: [] }, gaps, rels)];
+  const ex = readExtra(tc);
+  // cell.content 是 DocumentContent 壳；id 优先用兜底里的原值，缺失才用计数器（agent 新增单元格）
+  return { ...ex, content: { bPresentation: false, content: inner, type: "docContent" }, tcPr: tcPrToJson(kid(tc, "w:tcPr"), gaps), id: ex.id != null ? ex.id : String(_cellId++), type: "tblCell" };
+}
+function rowToJson(tr, gaps, rels) {
+  const ex = readExtra(tr);
+  return {
+    ...ex,
+    content: kidsOf(tr, "w:tc").map((tc) => cellToJson(tc, gaps, rels)),
+    reviewInfo: ex.reviewInfo || { userId: "", author: "", date: "", moveType: "noMove", prevType: -1 },
+    reviewType: ex.reviewType || "common",
+    trPr: trPrToJson(kid(tr, "w:trPr"), gaps),
+    type: "tblRow",
+  };
+}
+function tableToJson(tbl, gaps, rels) {
+  const tblGrid = kidsOf(kid(tbl, "w:tblGrid"), "w:gridCol").map((g) => ({ w: a2i(g.attrs["w:w"]), type: "gridCol" }));
+  const ex = readExtra(tbl);
+  return {
+    ...ex,
+    bPresentation: false,
+    tblGrid,
+    tblPr: tblPrToJson(kid(tbl, "w:tblPr"), gaps),
+    content: kidsOf(tbl, "w:tr").map((tr) => rowToJson(tr, gaps, rels)),
+    changes: ex.changes || [],
+    type: "table",
+  };
+}
+
+/**
+ * OOXML(document.xml) → ONLYOFFICE-JSON 文档对象
+ * @returns { doc, gaps }
+ */
+export function ooxmlToJson(xml, gaps = new Gaps(), options = {}) {
+  _cellId = 1;
+  const docNode = parseXml(xml);
+  const body = kid(docNode, "w:body");
+  const empty = { type: "document", paraPr: { bFromDocument: true, type: "paraPr" }, content: [] };
+  if (!body) return { doc: empty, gaps };
+  const content = blocksFromKids(body.kids, gaps, options.relationships || options.rels);
+  const ex = readExtra(body); // 文档级 paraPr 等
+  return { doc: { type: "document", paraPr: { bFromDocument: true, type: "paraPr" }, ...ex, content }, gaps };
+}