@taloswork/ose-shared 0.1.11 → 0.1.12

package/skills/document/bin/doc-convert.cjs

@@ -58096,8 +58096,8 @@ var require_graceful_fs = __commonJS({
         }
       }
       var fs$writeFile = fs2.writeFile;
-      fs2.writeFile = writeFile2;
-      function writeFile2(path, data, options, cb2) {
+      fs2.writeFile = writeFile3;
+      function writeFile3(path, data, options, cb2) {
         if (typeof options === "function")
           cb2 = options, options = null;
         return go$writeFile(path, data, options, cb2);
@@ -59600,7 +59600,7 @@ var require_old = __commonJS({
       splitRootRe = /^[\/]*/;
     }
     var splitRootRe;
-    exports2.realpathSync = function realpathSync(p, cache) {
+    exports2.realpathSync = function realpathSync2(p, cache) {
       p = pathModule.resolve(p);
       if (cache && Object.prototype.hasOwnProperty.call(cache, p)) {
         return cache[p];
@@ -59753,8 +59753,8 @@ var require_fs = __commonJS({
   "node_modules/fs.realpath/index.js"(exports2, module2) {
     module2.exports = realpath;
     realpath.realpath = realpath;
-    realpath.sync = realpathSync;
-    realpath.realpathSync = realpathSync;
+    realpath.sync = realpathSync2;
+    realpath.realpathSync = realpathSync2;
     realpath.monkeypatch = monkeypatch;
     realpath.unmonkeypatch = unmonkeypatch;
     var fs = require("fs");
@@ -59782,7 +59782,7 @@ var require_fs = __commonJS({
         }
       });
     }
-    function realpathSync(p, cache) {
+    function realpathSync2(p, cache) {
       if (ok) {
         return origRealpathSync(p, cache);
       }
@@ -59798,7 +59798,7 @@ var require_fs = __commonJS({
     }
     function monkeypatch() {
       fs.realpath = realpath;
-      fs.realpathSync = realpathSync;
+      fs.realpathSync = realpathSync2;
     }
     function unmonkeypatch() {
       fs.realpath = origRealpath;
@@ -85503,7 +85503,7 @@ var require_dir_writer = __commonJS({
     module2.exports = DirWriter;
     var Writer = require_writer2();
     var inherits2 = require_inherits();
-    var mkdir2 = require_mkdirp();
+    var mkdir3 = require_mkdirp();
     var path = require("path");
     var collect = require_collect();
     inherits2(DirWriter, Writer);
@@ -85519,7 +85519,7 @@ var require_dir_writer = __commonJS({
     }
     DirWriter.prototype._create = function() {
       var self2 = this;
-      mkdir2(self2._path, Writer.dirmode, function(er) {
+      mkdir3(self2._path, Writer.dirmode, function(er) {
         if (er) return self2.error(er);
         self2.ready = true;
         self2.emit("ready");
@@ -85864,7 +85864,7 @@ var require_writer2 = __commonJS({
     var fs = require_graceful_fs();
     var inherits2 = require_inherits();
     var rimraf = require_rimraf();
-    var mkdir2 = require_mkdirp();
+    var mkdir3 = require_mkdirp();
     var path = require("path");
     var umask = process.platform === "win32" ? 0 : process.umask();
     var getType = require_get_type();
@@ -85970,7 +85970,7 @@ var require_writer2 = __commonJS({
       }
     };
     function create2(self2) {
-      mkdir2(path.dirname(self2._path), Writer.dirmode, function(er, made) {
+      mkdir3(path.dirname(self2._path), Writer.dirmode, function(er, made) {
         if (er) return self2.error(er);
         self2._madeDir = made;
         return self2._create();
@@ -103048,6 +103048,10 @@ async function convertToPdfViaSoffice(inputPath, options = {}) {
   const profileDir = await (0, import_promises.mkdtemp)((0, import_node_path.join)((0, import_node_os.tmpdir)(), "talos-lo-profile-"));
   const outDir = await (0, import_promises.mkdtemp)((0, import_node_path.join)((0, import_node_os.tmpdir)(), "talos-lo-out-"));
   try {
+    if (isMacAppSoffice(soffice)) {
+      await (0, import_promises.mkdir)((0, import_node_path.join)(profileDir, "user"), { recursive: true });
+      await (0, import_promises.writeFile)((0, import_node_path.join)(profileDir, "user", "registrymodifications.xcu"), macFontReplacementXcu(missingWebFonts()));
+    }
     await runSoffice(soffice, [
       "--headless",
       "--norestore",
@@ -103073,6 +103077,83 @@ async function convertToPdfViaSoffice(inputPath, options = {}) {
     });
   }
 }
+function isMacAppSoffice(bin, platform = process.platform) {
+  if (platform !== "darwin")
+    return false;
+  let real = bin;
+  try {
+    real = (0, import_node_fs.realpathSync)(bin);
+  } catch {
+  }
+  return /\.app\/Contents\/MacOS\/soffice(\.bin)?$/.test(real);
+}
+var MAC_CJK_FONT_REPLACEMENT_XCU = `<?xml version="1.0" encoding="UTF-8"?>
+<oor:items xmlns:oor="http://openoffice.org/2001/registry" xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
+<item oor:path="/org.openoffice.Office.Common/Font/Substitution"><prop oor:name="Replacement" oor:op="fuse"><value>true</value></prop></item>
+<item oor:path="/org.openoffice.Office.Common/Font/Substitution/FontPairs"><node oor:name="_0" oor:op="replace"><prop oor:name="Always" oor:op="fuse"><value>true</value></prop><prop oor:name="OnScreenOnly" oor:op="fuse"><value>false</value></prop><prop oor:name="ReplaceFont" oor:op="fuse"><value>Microsoft YaHei</value></prop><prop oor:name="SubstituteFont" oor:op="fuse"><value>\u51AC\u9752\u9ED1\u4F53\u7B80\u4F53\u4E2D\u6587</value></prop></node><node oor:name="_1" oor:op="replace"><prop oor:name="Always" oor:op="fuse"><value>true</value></prop><prop oor:name="OnScreenOnly" oor:op="fuse"><value>false</value></prop><prop oor:name="ReplaceFont" oor:op="fuse"><value>Microsoft JhengHei</value></prop><prop oor:name="SubstituteFont" oor:op="fuse"><value>\u9ED1\u4F53-\u7E41</value></prop></node></item>
+</oor:items>
+`;
+var SANS = { group: "sans", mac: "Arial", cloud: "Liberation Sans" };
+var SERIF_TIMES = { group: "serif", mac: "Times New Roman", cloud: "Liberation Serif" };
+var SERIF_GEORGIA = { group: "serif", mac: "Georgia", cloud: "Liberation Serif" };
+var CONDENSED = { group: "display", mac: "Arial Narrow", cloud: "Liberation Sans Narrow" };
+var WEB_FONT_SUBSTITUTES = {
+  // sans — the neutral grotesques
+  "Inter": SANS,
+  "Jost": SANS,
+  "Public Sans": SANS,
+  "Manrope": SANS,
+  "Karla": SANS,
+  "Sora": SANS,
+  "IBM Plex Sans": SANS,
+  "Archivo": SANS,
+  "Space Grotesk": SANS,
+  "Quicksand": SANS,
+  "Lexend": SANS,
+  "Nunito": SANS,
+  // serif
+  "Playfair Display": SERIF_TIMES,
+  "Bodoni Moda": SERIF_TIMES,
+  "Zilla Slab": SERIF_TIMES,
+  "Fraunces": SERIF_TIMES,
+  "Cormorant Garamond": SERIF_TIMES,
+  "Lora": SERIF_GEORGIA,
+  "Marcellus": SERIF_GEORGIA,
+  "DM Serif Display": SERIF_GEORGIA,
+  // display
+  "Anton": CONDENSED,
+  "Big Shoulders Display": CONDENSED,
+  "Syne": { group: "display", mac: "Arial", cloud: "Liberation Sans" },
+  "Chakra Petch": { group: "display", mac: "Arial", cloud: "Liberation Sans" },
+  "Gochi Hand": { group: "display", mac: "Noteworthy", cloud: "Liberation Sans" }
+};
+var WEB_GROTESQUE_FAMILIES = Object.keys(WEB_FONT_SUBSTITUTES).filter((f) => WEB_FONT_SUBSTITUTES[f].group === "sans");
+var MAC_FONT_DIRS = ["/System/Library/Fonts", "/System/Library/Fonts/Supplemental", "/Library/Fonts"];
+function missingWebFonts(fontFiles = macFontFiles()) {
+  const names = fontFiles.map((f) => (0, import_node_path.basename)(f).toLowerCase());
+  return Object.keys(WEB_FONT_SUBSTITUTES).filter((family) => {
+    const stem = family.replace(/\s+/g, "").toLowerCase();
+    return !names.some((n) => n.startsWith(stem) && /^[-_. \[]/.test(n.slice(stem.length)));
+  });
+}
+function macFontFiles() {
+  const dirs = [...MAC_FONT_DIRS, (0, import_node_path.join)(process.env.HOME ?? "", "Library/Fonts")];
+  const out = [];
+  for (const d of dirs) {
+    try {
+      out.push(...(0, import_node_fs.readdirSync)(d));
+    } catch {
+    }
+  }
+  return out;
+}
+var fontPair = (i, from, to) => `<node oor:name="_${i}" oor:op="replace"><prop oor:name="Always" oor:op="fuse"><value>true</value></prop><prop oor:name="OnScreenOnly" oor:op="fuse"><value>false</value></prop><prop oor:name="ReplaceFont" oor:op="fuse"><value>${from}</value></prop><prop oor:name="SubstituteFont" oor:op="fuse"><value>${to}</value></prop></node>`;
+function macFontReplacementXcu(missing) {
+  if (!missing.length)
+    return MAC_CJK_FONT_REPLACEMENT_XCU;
+  const extra = missing.map((f, i) => fontPair(i + 2, f, WEB_FONT_SUBSTITUTES[f]?.mac ?? "Arial")).join("");
+  return MAC_CJK_FONT_REPLACEMENT_XCU.replace("</node></item>", `</node>${extra}</item>`);
+}
 function runSoffice(bin, args, timeoutMs) {
   return new Promise((resolve, reject2) => {
     const child = (0, import_node_child_process.spawn)(bin, args, { stdio: ["ignore", "pipe", "pipe"] });
@@ -103180,6 +103261,11 @@ async function fontManifest() {
   }
   return _fontManifest;
 }
+var _faceB64 = /* @__PURE__ */ new Map();
+async function faceBase64(path) {
+  if (!_faceB64.has(path)) _faceB64.set(path, (await (0, import_promises2.readFile)(path).catch(() => null))?.toString("base64") ?? null);
+  return _faceB64.get(path);
+}
 async function faceCss(entries, range2, doc) {
   const { dir } = await fontManifest();
   const rules = [];
@@ -103188,7 +103274,7 @@ async function faceCss(entries, range2, doc) {
     if (!families.length) continue;
     const path = (0, import_node_path2.join)(dir, entry.file);
     if (!(0, import_node_fs2.existsSync)(path)) continue;
-    const b64 = (await (0, import_promises2.readFile)(path).catch(() => null))?.toString("base64");
+    const b64 = await faceBase64(path);
     if (!b64) continue;
     for (const family of families) {
       rules.push(
@@ -103198,22 +103284,21 @@ async function faceCss(entries, range2, doc) {
   }
   return rules.join("");
 }
-var _cjkFontCss;
+var _faceCssBySet = /* @__PURE__ */ new Map();
+async function cachedFaceCss(group2, entries, range2, doc) {
+  const named = (entries ?? []).flatMap((e) => (e.families ?? []).filter((f) => doc.includes(f)));
+  const key = `${group2}|${named.join("|")}`;
+  if (!_faceCssBySet.has(key)) _faceCssBySet.set(key, await faceCss(entries, range2, doc));
+  return _faceCssBySet.get(key);
+}
 async function cjkFontFaceCss(doc) {
-  if (_cjkFontCss === void 0) {
-    const m = await fontManifest();
-    _cjkFontCss = await faceCss(m.cjk, m.cjkUnicodeRange, doc);
-  }
-  return _cjkFontCss;
+  if (process.env.TALOS_CJK_FONTS === "system") return "";
+  const m = await fontManifest();
+  return cachedFaceCss("cjk", m.cjk, m.cjkUnicodeRange, doc);
 }
-var _latinFontCss;
 async function latinFontFaceCss(doc) {
-  if (_latinFontCss === void 0) {
-    const m = await fontManifest();
-    const wanted = (m.latin ?? []).filter((e) => (e.families ?? []).some((f) => doc.includes(f)));
-    _latinFontCss = await faceCss(wanted, void 0);
-  }
-  return _latinFontCss;
+  const m = await fontManifest();
+  return cachedFaceCss("latin", m.latin, void 0, doc);
 }
 function injectStyle(html, css) {
   const tag = `<style>${css}</style>`;
@@ -103300,32 +103385,29 @@ async function renderViaSoffice(input, cachePath) {
     });
     return { bytes, cached: false };
   } catch (e) {
-    process.stderr.write(`doc-convert: LibreOffice failed (${e.message}); falling back to HTML render
-`);
-    return null;
+    return { failed: `LibreOffice failed (${e.message})` };
   }
 }
 async function renderViaOfficeService(input, raw) {
   const url = process.env.TALOS_OFFICE_URL;
   if (!url) return null;
   try {
-    return await convertViaOfficeService(raw, (0, import_node_path3.basename)(input), url, {
+    return { bytes: await convertViaOfficeService(raw, (0, import_node_path3.basename)(input), url, {
       token: process.env.TALOS_OFFICE_TOKEN,
       timeoutMs: 12e4
-    });
+    }) };
   } catch (e) {
-    process.stderr.write(`doc-convert: office service failed (${e.message}); falling back
-`);
-    return null;
+    return { failed: `office service failed (${e.message})` };
   }
 }
 async function main() {
   const args = process.argv.slice(2);
+  const requireOffice = args.includes("--require-office");
   const cacheIdx = args.indexOf("--cache-dir");
   const cacheDir = cacheIdx >= 0 ? args[cacheIdx + 1] : process.env.TALOS_OFFICE_CACHE_DIR;
-  const consumed = /* @__PURE__ */ new Set([cacheIdx, cacheIdx >= 0 ? cacheIdx + 1 : -1]);
+  const consumed = /* @__PURE__ */ new Set([cacheIdx, cacheIdx >= 0 ? cacheIdx + 1 : -1, args.indexOf("--require-office")]);
   const [input, out] = args.filter((_4, i) => !consumed.has(i));
-  if (!input || !out) die("usage: doc-convert <in.docx|pptx|xlsx> <out.pdf> [--cache-dir <dir>]", 64);
+  if (!input || !out) die("usage: doc-convert <in.docx|pptx|xlsx> <out.pdf> [--cache-dir <dir>] [--require-office]", 64);
   if (cacheIdx >= 0 && !cacheDir) die("--cache-dir requires a directory", 64);
   if (!/\.pdf$/i.test(out)) die("output must be a .pdf path", 64);
   let raw;
@@ -103336,20 +103418,29 @@ async function main() {
   }
   const buffer = raw.buffer.slice(raw.byteOffset, raw.byteOffset + raw.byteLength);
   const cachePath = await faithfulCachePath(cacheDir, input, raw).catch((e) => die(`cannot prepare cache: ${e.message}`, 73));
+  const failures = [];
   const loResult = await renderViaSoffice(input, cachePath);
-  if (loResult) {
+  if (loResult?.bytes) {
     await (0, import_promises3.writeFile)(out, Buffer.from(loResult.bytes));
     process.stdout.write(`wrote ${out} [pdf (libreoffice${loResult.cached ? ", cache hit" : ""})]
 `);
     return;
   }
-  const svcBytes = await renderViaOfficeService(input, raw);
-  if (svcBytes) {
-    await (0, import_promises3.writeFile)(out, Buffer.from(svcBytes));
+  if (loResult?.failed) failures.push(loResult.failed);
+  const svc = await renderViaOfficeService(input, raw);
+  if (svc?.bytes) {
+    await (0, import_promises3.writeFile)(out, Buffer.from(svc.bytes));
     process.stdout.write(`wrote ${out} [pdf (libreoffice service)]
 `);
     return;
   }
+  if (svc?.failed) failures.push(svc.failed);
+  if (requireOffice) {
+    if (failures.length) die(`${failures.join("; ")} \u2014 --require-office does not fall back to a Chromium reconstruction (its layout is not Office's); retry, a timeout under load is the usual cause`, 75);
+    die("--require-office: no LibreOffice engine here (no local soffice, TALOS_OFFICE_URL unset) \u2014 only a Chromium reconstruction is available, which is not an Office render", 69);
+  }
+  for (const f of failures) process.stderr.write(`doc-convert: ${f}; falling back to HTML render
+`);
   if (/\.(xlsx|xls|ods)$/i.test(input)) {
     die("xlsx\u2192PDF \u9700\u8981\u672C\u673A LibreOffice \u6216 TALOS_OFFICE_URL sidecar,\u5F53\u524D\u90FD\u4E0D\u53EF\u7528", 69);
   }