@v1hz/md2docx 1.4.0 → 1.5.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@v1hz/md2docx",
-  "version": "1.4.0",
+  "version": "1.5.0",
   "description": "基于 pandoc 的 Markdown 转 Word 工具，自动格式化 Markdown，支持自定义样式，AI 友好，用户友好。",
   "keywords": [
     "converter",
@@ -27,6 +27,7 @@
     "src",
     "config/config.json",
     "config/config.schema.json",
+    "config/style.json",
     "README.md",
     "LICENSE"
   ],

package/src/index.ts

@@ -10,6 +10,7 @@ import {
   CONFIG_PATH,
   CONFIG_SCHEMA_PATH,
   PKG_DIR,
+  STYLE_CONFIG,
   TMP_DIR,
   formattedMdPath,
   preprocessDir,
@@ -70,12 +71,14 @@ export async function run(args: string[]): Promise<number> {
     writeFileSync(mdOutput, formattedMd, "utf-8");
 
     if (cfg.pandoc.enabled) {
-      const templatePath = await ensureTemplateDocx();
+      const templatePath = await ensureTemplateDocx(STYLE_CONFIG);
       const configuredName = cfg.pandoc.outputName.replaceAll("{file_name}", baseName);
       const docxOutput = resolve(cli.outputPath ?? configuredName);
       mkdirSync(dirname(docxOutput), { recursive: true });
+      const luaFilter = resolve(PKG_DIR, "config/lua/add-inline-code.lua");
+      const luaFlag = existsSync(luaFilter) ? `--lua-filter=${luaFilter}` : "";
       const result =
-        await $`pandoc ${mdOutput} -o ${docxOutput} --reference-doc=${templatePath}`.nothrow();
+        await $`pandoc ${mdOutput} -o ${docxOutput} --reference-doc=${templatePath} ${luaFlag}`.nothrow();
       if (result.exitCode !== 0) {
         console.error(`pandoc 转换失败 (exit code ${result.exitCode}):`, result.stderr.toString());
         return 1;

package/src/style/extract.ts

@@ -1,4 +1,4 @@
-import { readFileSync } from "node:fs";
+import { readFileSync, writeFileSync } from "node:fs";
 import { DOMParser } from "@xmldom/xmldom";
 import { useNamespaces } from "xpath";
 import PizZip from "pizzip";
@@ -48,6 +48,39 @@ function wBool(el: Element | null, local: string): boolean | null {
   return v === null || !["0", "false", "off", "no"].includes(v.toLowerCase());
 }
 
+// ── 通用映射函数 ──────────────────────────────────────────
+
+/** 将 w:shd 元素映射为 shading 对象 */
+function mapShd(shd: Element | null): Record<string, string> | undefined {
+  if (!shd) return undefined;
+  const sh: Record<string, string> = {};
+  for (const a of [
+    "val",
+    "color",
+    "fill",
+    "themeColor",
+    "themeTint",
+    "themeFill",
+    "themeFillShade",
+    "themeShade",
+  ]) {
+    const v = wAttr(shd, a);
+    if (v !== null) sh[a] = v;
+  }
+  return Object.keys(sh).length > 0 ? sh : undefined;
+}
+
+/** 将 w:border 元素映射为 border 对象 */
+function mapBorder(bdr: Element | null): Record<string, string | number> | undefined {
+  if (!bdr) return undefined;
+  const b: Record<string, string | number> = {};
+  for (const a of ["val", "color", "sz", "space", "themeColor", "themeShade"]) {
+    const v = wAttr(bdr, a);
+    if (v !== null) b[a] = /^\d+$/.test(v) ? Number(v) : v;
+  }
+  return Object.keys(b).length > 0 ? b : undefined;
+}
+
 // ── 映射函数 ─────────────────────────────────────────────
 
 function mapRunProps(rPr: Element | null): Record<string, unknown> {
@@ -58,7 +91,16 @@ function mapRunProps(rPr: Element | null): Record<string, unknown> {
   const rFonts = one(rPr, "w:rFonts");
   if (rFonts) {
     const font: Record<string, string> = {};
-    for (const a of ["ascii", "hAnsi", "cs", "eastAsia"]) {
+    for (const a of [
+      "ascii",
+      "hAnsi",
+      "cs",
+      "eastAsia",
+      "asciiTheme",
+      "hAnsiTheme",
+      "eastAsiaTheme",
+      "cstheme",
+    ]) {
       const v = wAttr(rFonts, a);
       if (v) font[a] = v;
     }
@@ -77,7 +119,7 @@ function mapRunProps(rPr: Element | null): Record<string, unknown> {
     if (v !== null) out.sizeComplexScript = v;
   }
 
-  // 颜色
+  // 颜色（仅 val，docx 包不支持主题属性）
   const color = one(rPr, "w:color");
   if (color) {
     const v = wAttr(color, "val");
@@ -140,6 +182,20 @@ function mapRunProps(rPr: Element | null): Record<string, unknown> {
     if (Object.keys(lo).length) out.language = lo;
   }
 
+  // 底纹（背景色）
+  const shd = mapShd(one(rPr, "w:shd"));
+  if (shd) out.shading = shd;
+
+  // 边框
+  const bdr = one(rPr, "w:bdr");
+  if (bdr) {
+    const b = mapBorder(bdr);
+    if (b) out.border = b;
+  }
+
+  // 校对标记
+  if (one(rPr, "w:noProof")) out.noProof = true;
+
   return out;
 }
 
@@ -177,14 +233,30 @@ function mapParagraphProps(pPr: Element | null): Record<string, unknown> {
     }
     const lr = wAttr(spacing, "lineRule");
     if (lr !== null) sp.lineRule = lr;
+    // 也提取 beforeLines / afterLines（Word 有时用这种）
+    for (const tag of ["beforeLines", "afterLines", "beforeAutospacing", "afterAutospacing"]) {
+      const v = wAttr(spacing, tag);
+      if (v !== null) sp[tag] = v;
+    }
     if (Object.keys(sp).length) out.spacing = sp;
   }
 
   // 缩进
   const ind = one(pPr, "w:ind");
   if (ind) {
-    const id: Record<string, number> = {};
-    for (const a of ["left", "right", "firstLine", "hanging", "start", "end"]) {
+    const id: Record<string, number | string> = {};
+    for (const a of [
+      "left",
+      "right",
+      "firstLine",
+      "hanging",
+      "start",
+      "end",
+      "leftChars",
+      "rightChars",
+      "firstLineChars",
+      "hangingChars",
+    ]) {
       const v = wNumber(ind, a);
       if (v !== null) id[a] = v;
     }
@@ -198,6 +270,9 @@ function mapParagraphProps(pPr: Element | null): Record<string, unknown> {
     "pageBreakBefore",
     "widowControl",
     "contextualSpacing",
+    "wordWrap",
+    "suppressLineNumbers",
+    "suppressAutoHyphens",
   ]) {
     const v = wBool(pPr, tag);
     if (v !== null) out[tag] = v;
@@ -210,9 +285,132 @@ function mapParagraphProps(pPr: Element | null): Record<string, unknown> {
     if (v !== null) out.outlineLevel = v;
   }
 
+  // 段落底纹
+  const shd = mapShd(one(pPr, "w:shd"));
+  if (shd) out.shading = shd;
+
+  // 段落边框
+  const pBdr = one(pPr, "w:pBdr");
+  if (pBdr) {
+    const borders: Record<string, unknown> = {};
+    for (const side of ["top", "left", "bottom", "right"]) {
+      const b = mapBorder(one(pBdr, `w:${side}`));
+      if (b) borders[side] = b;
+    }
+    if (Object.keys(borders).length) out.border = borders;
+  }
+
+  // 制表位
+  const tabs = one(pPr, "w:tabs");
+  if (tabs) {
+    const tabList: Record<string, unknown>[] = [];
+    for (const tab of all(tabs, "w:tab")) {
+      const t: Record<string, string | number> = {};
+      const v = wAttr(tab, "val");
+      const pos = wNumber(tab, "pos");
+      if (v) t.val = v;
+      if (pos !== null) t.pos = pos;
+      if (Object.keys(t).length) tabList.push(t);
+    }
+    if (tabList.length) out.tabs = tabList;
+  }
+
   return out;
 }
 
+function mapTableProps(tblPr: Element | null): Record<string, unknown> | undefined {
+  if (!tblPr) return undefined;
+  const out: Record<string, unknown> = {};
+
+  // 表格对齐
+  const jc = one(tblPr, "w:jc");
+  if (jc) {
+    const v = wAttr(jc, "val");
+    if (v) out.alignment = v;
+  }
+
+  // 表格缩进
+  const tblInd = one(tblPr, "w:tblInd");
+  if (tblInd) {
+    const id: Record<string, number | string> = {};
+    const w = wNumber(tblInd, "w");
+    if (w !== null) id.width = w;
+    const t = wAttr(tblInd, "type");
+    if (t !== null) id.type = t;
+    if (Object.keys(id).length) out.indent = id;
+  }
+
+  // 表格边框
+  const tblBorders = one(tblPr, "w:tblBorders");
+  if (tblBorders) {
+    const borders: Record<string, unknown> = {};
+    for (const side of ["top", "left", "bottom", "right", "insideH", "insideV"]) {
+      const b = mapBorder(one(tblBorders, `w:${side}`));
+      if (b) borders[side] = b;
+    }
+    if (Object.keys(borders).length) out.borders = borders;
+  }
+
+  // 单元格边距
+  const tblCellMar = one(tblPr, "w:tblCellMar");
+  if (tblCellMar) {
+    const margins: Record<string, number> = {};
+    for (const side of ["top", "left", "bottom", "right"]) {
+      const v = wNumber(one(tblCellMar, `w:${side}`), "w");
+      if (v !== null) margins[side] = v;
+    }
+    if (Object.keys(margins).length) out.cellMargin = margins;
+  }
+
+  return Object.keys(out).length > 0 ? out : undefined;
+}
+
+function mapTableStylePr(tblStylePr: Element | null): Record<string, unknown> | undefined {
+  if (!tblStylePr) return undefined;
+
+  const type = wAttr(tblStylePr, "type");
+  const out: Record<string, unknown> = {};
+  if (type) out.type = type;
+
+  // rPr
+  const rPr = mapRunProps(one(tblStylePr, "w:rPr"));
+  if (Object.keys(rPr).length) out.run = rPr;
+
+  // pPr（用于 firstRow 段落格式）
+  const pPr = mapParagraphProps(one(tblStylePr, "w:pPr"));
+  if (Object.keys(pPr).length) out.paragraph = pPr;
+
+  // tblPr
+  const tblPr = one(tblStylePr, "w:tblPr");
+  if (tblPr) {
+    const tbl = mapTableProps(tblPr);
+    if (tbl) out.table = tbl;
+  }
+
+  // tcPr
+  const tcPr = one(tblStylePr, "w:tcPr");
+  if (tcPr) {
+    const tc: Record<string, unknown> = {};
+    const tcBorders = one(tcPr, "w:tcBorders");
+    if (tcBorders) {
+      const borders: Record<string, unknown> = {};
+      for (const side of ["top", "left", "bottom", "right", "insideH", "insideV"]) {
+        const b = mapBorder(one(tcBorders, `w:${side}`));
+        if (b) borders[side] = b;
+      }
+      if (Object.keys(borders).length) tc.borders = borders;
+    }
+    const vAlign = one(tcPr, "w:vAlign");
+    if (vAlign) {
+      const v = wAttr(vAlign, "val");
+      if (v) tc.verticalAlign = v;
+    }
+    if (Object.keys(tc).length) out.cell = tc;
+  }
+
+  return Object.keys(out).length > 0 ? out : undefined;
+}
+
 // ── 导出类型 ─────────────────────────────────────────────
 
 export interface StyleEntry {
@@ -227,12 +425,41 @@ export interface StyleEntry {
   uiPriority?: number;
   run?: Record<string, unknown>;
   paragraph?: Record<string, unknown>;
+  table?: Record<string, unknown>;
+  tableStyleParts?: Record<string, unknown>[];
 }
 
 export interface ExtractedStyles {
   default?: Record<string, unknown>;
   paragraphStyles?: StyleEntry[];
   characterStyles?: StyleEntry[];
+  tableStyles?: StyleEntry[];
+  /** 表格样式的原始 XML 字符串，用于注入到生成的模板 docx */
+  tableStylesXml?: string;
+}
+
+/** 将 XML 元素序列化为字符串 */
+function serializeElement(el: Element): string {
+  let result = "<" + el.tagName;
+  for (let i = 0; i < el.attributes.length; i++) {
+    const attr = el.attributes[i]!;
+    result += ` ${attr.name}="${attr.value.replace(/"/g, "&quot;")}"`;
+  }
+  if (el.childNodes.length === 0) {
+    result += "/>";
+  } else {
+    result += ">";
+    for (let i = 0; i < el.childNodes.length; i++) {
+      const child = el.childNodes[i]!;
+      if (child.nodeType === 1) {
+        result += serializeElement(child as Element);
+      } else if (child.nodeType === 3) {
+        result += (child as Text).data;
+      }
+    }
+    result += "</" + el.tagName + ">";
+  }
+  return result;
 }
 
 const BUILTIN_PARA: Record<string, string> = {
@@ -272,6 +499,7 @@ export function extractStylesFromDocx(docxPath: string): ExtractedStyles {
   const defaultOverrides: Record<string, unknown> = {};
   const paragraphStyles: StyleEntry[] = [];
   const characterStyles: StyleEntry[] = [];
+  const tableStyles: StyleEntry[] = [];
 
   // 1. docDefaults
   const xmlDocument = doc as unknown as XmlNode;
@@ -293,11 +521,15 @@ export function extractStylesFromDocx(docxPath: string): ExtractedStyles {
   for (const style of all(xmlDocument, "//w:style")) {
     const styleId = wAttr(style, "styleId") ?? "";
     const styleType = wAttr(style, "type") ?? "";
-    if (!styleId || !["paragraph", "character"].includes(styleType)) continue;
+    if (!styleId || !["paragraph", "character", "table"].includes(styleType)) continue;
     const name = wAttr(one(style, "w:name"), "val") ?? styleId;
 
     // 跳过隐式样式和 token
-    if (["Normal", "DefaultParagraphFont", "TableNormal"].includes(styleId)) continue;
+    if (
+      styleType !== "table" &&
+      ["Normal", "DefaultParagraphFont", "TableNormal"].includes(styleId)
+    )
+      continue;
     if (styleId.endsWith("Tok") && styleType === "character") continue;
 
     const entry: StyleEntry = { id: styleId, name };
@@ -334,11 +566,35 @@ export function extractStylesFromDocx(docxPath: string): ExtractedStyles {
       if (v !== null) entry.uiPriority = v;
     }
 
-    // run / paragraph
-    const run = mapRunProps(one(style, "w:rPr"));
-    if (Object.keys(run).length) entry.run = run;
-    const para = mapParagraphProps(one(style, "w:pPr"));
-    if (Object.keys(para).length) entry.paragraph = para;
+    // 段落/字符样式：run + paragraph
+    if (styleType !== "table") {
+      const run = mapRunProps(one(style, "w:rPr"));
+      if (Object.keys(run).length) entry.run = run;
+      const para = mapParagraphProps(one(style, "w:pPr"));
+      if (Object.keys(para).length) entry.paragraph = para;
+    }
+
+    // 表格样式：table + tableStyleParts
+    if (styleType === "table") {
+      const tblPr = one(style, "w:tblPr");
+      if (tblPr) {
+        const tbl = mapTableProps(tblPr);
+        if (tbl) entry.table = tbl;
+      }
+
+      const parts: Record<string, unknown>[] = [];
+      for (const tsp of all(style, "w:tblStylePr")) {
+        const mapped = mapTableStylePr(tsp);
+        if (mapped) parts.push(mapped);
+      }
+      if (parts.length) entry.tableStyleParts = parts;
+
+      // 表格样式的 run（用于单元格默认字体）
+      const run = mapRunProps(one(style, "w:rPr"));
+      if (Object.keys(run).length) entry.run = run;
+      const pPr = mapParagraphProps(one(style, "w:pPr"));
+      if (Object.keys(pPr).length) entry.paragraph = pPr;
+    }
 
     // 分发
     if (styleType === "paragraph") {
@@ -357,9 +613,22 @@ export function extractStylesFromDocx(docxPath: string): ExtractedStyles {
       } else {
         characterStyles.push(entry);
       }
+    } else if (styleType === "table") {
+      tableStyles.push(entry);
     }
   }
 
+  // 3. 收集表格样式的原始 XML（用于注入模板）
+  let tableXmlFragments = "";
+  for (const el of all(
+    xmlDocument,
+    "//w:style[@w:type='table' and not(@w:default='1') and @w:customStyle='1']",
+  )) {
+    // 手动序列化 XML 元素
+    const xmlStr = serializeElement(el);
+    tableXmlFragments += xmlStr;
+  }
+
   const result: ExtractedStyles = {};
   if (Object.keys(defaultDoc).length > 0 || Object.keys(defaultOverrides).length > 0) {
     result.default = {
@@ -369,6 +638,16 @@ export function extractStylesFromDocx(docxPath: string): ExtractedStyles {
   }
   if (paragraphStyles.length) result.paragraphStyles = paragraphStyles;
   if (characterStyles.length) result.characterStyles = characterStyles;
+  if (tableStyles.length) result.tableStyles = tableStyles;
+  if (tableXmlFragments) result.tableStylesXml = tableXmlFragments;
 
   return result;
 }
+
+if (import.meta.main) {
+  const result = extractStylesFromDocx("test.docx");
+  writeFileSync("config/style.json", JSON.stringify(result, null, 2));
+  console.log(
+    `Extracted ${result.paragraphStyles?.length ?? 0} paragraph, ${result.characterStyles?.length ?? 0} character, ${result.tableStyles?.length ?? 0} table styles`,
+  );
+}

package/src/style/generate.ts

@@ -1,27 +1,58 @@
 import { readFileSync, mkdirSync } from "fs";
 import { Document, Packer } from "docx";
 import { dirname } from "path";
-import { STYLE_CONFIG, STYLE_TEMPLATE_DOCX } from "../paths";
+import { STYLE_TEMPLATE_DOCX } from "../paths";
+
+/**
+ * 根据样式 JSON 文件生成模板 docx。
+ *
+ * @param styleJsonPath - 样式 JSON 文件路径（如 config/style.json）
+ */
+export async function generateTemplateDocx(styleJsonPath: string): Promise<void> {
+  const raw = JSON.parse(readFileSync(styleJsonPath, "utf-8")) as Record<string, unknown>;
+  const tableStylesXml = raw.tableStylesXml as string | undefined;
+  // 移除 tableStylesXml，docx 包不识别它
+  const { tableStylesXml: _, ...styles } = raw;
 
-export async function generateTemplateDocx(): Promise<void> {
-  const styles = JSON.parse(readFileSync(STYLE_CONFIG, "utf-8"));
   const doc = new Document({
-    styles: styles,
+    styles: styles as Record<string, unknown>,
     sections: [],
   });
   const buf = await Packer.toBuffer(doc);
   mkdirSync(dirname(STYLE_TEMPLATE_DOCX), { recursive: true });
+
+  // 如果有表格样式 XML，注入到生成的 docx 中
+  if (tableStylesXml) {
+    const PizZip = (await import("pizzip")).default;
+    const zip = new PizZip(buf);
+    const stylesFile = zip.file("word/styles.xml");
+    if (stylesFile) {
+      const originalXml = stylesFile.asText();
+      // 在 </w:styles> 之前插入表格样式
+      const injectedXml = originalXml.replace("</w:styles>", tableStylesXml + "</w:styles>");
+      zip.file("word/styles.xml", injectedXml);
+      const modifiedBuf = zip.generate({ type: "nodebuffer" }) as Buffer;
+      await Bun.write(STYLE_TEMPLATE_DOCX, modifiedBuf);
+      return;
+    }
+  }
+
   await Bun.write(STYLE_TEMPLATE_DOCX, buf);
 }
 
 let _generatePromise: Promise<void> | null = null;
 
-export async function ensureTemplateDocx(): Promise<string> {
+/**
+ * 获取模板 docx 路径。如果模板不存在，则自动生成。
+ *
+ * @param styleJsonPath - 样式 JSON 文件路径（如 config/style.json）
+ */
+export async function ensureTemplateDocx(styleJsonPath: string): Promise<string> {
   const exists = await Bun.file(STYLE_TEMPLATE_DOCX).exists();
   if (!exists) {
     if (_generatePromise === null) {
-      _generatePromise = generateTemplateDocx().catch((e) => {
-        _generatePromise = null; // 重置，允许下次重试
+      _generatePromise = generateTemplateDocx(styleJsonPath).catch((e) => {
+        _generatePromise = null;
         throw e;
       });
     }