@taloswork/ose-shared 0.1.11 → 0.1.13

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

@@ -9790,9 +9790,9 @@ var require_load = __commonJS({
 var require_lib3 = __commonJS({
   "node_modules/jszip/lib/index.js"(exports2, module2) {
     "use strict";
-    function JSZip2() {
-      if (!(this instanceof JSZip2)) {
-        return new JSZip2();
+    function JSZip3() {
+      if (!(this instanceof JSZip3)) {
+        return new JSZip3();
       }
       if (arguments.length) {
         throw new Error("The constructor with parameters has been removed in JSZip 3.0, please check the upgrade guide.");
@@ -9801,7 +9801,7 @@ var require_lib3 = __commonJS({
       this.comment = null;
       this.root = "";
       this.clone = function() {
-        var newObj = new JSZip2();
+        var newObj = new JSZip3();
         for (var i in this) {
           if (typeof this[i] !== "function") {
             newObj[i] = this[i];
@@ -9810,16 +9810,16 @@ var require_lib3 = __commonJS({
         return newObj;
       };
     }
-    JSZip2.prototype = require_object();
-    JSZip2.prototype.loadAsync = require_load();
-    JSZip2.support = require_support();
-    JSZip2.defaults = require_defaults();
-    JSZip2.version = "3.10.1";
-    JSZip2.loadAsync = function(content, options) {
-      return new JSZip2().loadAsync(content, options);
+    JSZip3.prototype = require_object();
+    JSZip3.prototype.loadAsync = require_load();
+    JSZip3.support = require_support();
+    JSZip3.defaults = require_defaults();
+    JSZip3.version = "3.10.1";
+    JSZip3.loadAsync = function(content, options) {
+      return new JSZip3().loadAsync(content, options);
     };
-    JSZip2.external = require_external();
-    module2.exports = JSZip2;
+    JSZip3.external = require_external();
+    module2.exports = JSZip3;
   }
 });
 
@@ -9827,11 +9827,11 @@ var require_lib3 = __commonJS({
 var require_pptxgen_cjs = __commonJS({
   "node_modules/pptxgenjs/dist/pptxgen.cjs.js"(exports2, module2) {
     "use strict";
-    var JSZip2 = require_lib3();
+    var JSZip3 = require_lib3();
     function _interopDefaultLegacy(e) {
       return e && typeof e === "object" && "default" in e ? e : { "default": e };
     }
-    var JSZip__default = /* @__PURE__ */ _interopDefaultLegacy(JSZip2);
+    var JSZip__default = /* @__PURE__ */ _interopDefaultLegacy(JSZip3);
     var __assign2 = function() {
       __assign2 = Object.assign || function __assign3(t) {
         for (var s, i = 1, n = arguments.length; i < n; i++) {
@@ -85256,7 +85256,7 @@ var require_browser_buffer_encode = __commonJS({
 var require_zip_stream = __commonJS({
   "node_modules/exceljs/lib/utils/zip-stream.js"(exports2, module2) {
     var events = require("events");
-    var JSZip2 = require_lib3();
+    var JSZip3 = require_lib3();
     var StreamBuf = require_stream_buf();
     var { stringToBuffer } = require_browser_buffer_encode();
     var ZipWriter = class extends events.EventEmitter {
@@ -85269,7 +85269,7 @@ var require_zip_stream = __commonJS({
           },
           options
         );
-        this.zip = new JSZip2();
+        this.zip = new JSZip3();
         this.stream = new StreamBuf();
       }
       append(data, options) {
@@ -96949,7 +96949,7 @@ var require_theme1 = __commonJS({
 var require_xlsx2 = __commonJS({
   "node_modules/exceljs/lib/xlsx/xlsx.js"(exports2, module2) {
     var fs = require("fs");
-    var JSZip2 = require_lib3();
+    var JSZip3 = require_lib3();
     var { PassThrough } = require_readable2();
     var ZipStream = require_zip_stream();
     var StreamBuf = require_stream_buf();
@@ -97188,7 +97188,7 @@ var require_xlsx2 = __commonJS({
           tables: {},
           vmlDrawings: {}
         };
-        const zip2 = await JSZip2.loadAsync(buffer);
+        const zip2 = await JSZip3.loadAsync(buffer);
         for (const entry of Object.values(zip2.files)) {
           if (!entry.dir) {
             let entryName = entry.name;
@@ -105437,8 +105437,8 @@ var require_graceful_fs = __commonJS({
         }
       }
       var fs$writeFile = fs2.writeFile;
-      fs2.writeFile = writeFile;
-      function writeFile(path, data, options, cb) {
+      fs2.writeFile = writeFile2;
+      function writeFile2(path, data, options, cb) {
         if (typeof options === "function")
           cb = options, options = null;
         return go$writeFile(path, data, options, cb);
@@ -141451,7 +141451,7 @@ var require_excel = __commonJS({
 var import_node_module = __toESM(require("node:module"), 1);
 var import_node_path = require("node:path");
 var import_pptxgenjs = __toESM(require_pptxgen_cjs(), 1);
-var import_jszip = __toESM(require_lib3(), 1);
+var import_jszip2 = __toESM(require_lib3(), 1);
 var import_pizzip = __toESM(require_js(), 1);
 var pdfLib = __toESM(require_cjs(), 1);
 
@@ -192429,9 +192429,147 @@ for (const pluginName of Object.getOwnPropertyNames(plugin_exports)) {
 
 // apps/admin/data/skills.lib/document/bin-src/doc-run.mjs
 var import_exceljs = __toESM(require_excel(), 1);
+
+// packages/core/dist/document/pptx-finalize.js
+var import_jszip = __toESM(require_lib3(), 1);
+var import_promises = require("node:fs/promises");
+async function finalizePptx(bytes) {
+  return stripDanglingOverrides(await dedupeParagraphProps(await normalizeChartParts(bytes)));
+}
+function dedupeParagraphPropsXml(xml) {
+  return xml.replace(/<a:p>([\s\S]*?)<\/a:p>/g, (para, inner) => {
+    const first2 = inner.search(/<a:(?:r|br|fld)\b/);
+    if (first2 < 0)
+      return para;
+    const tail = inner.slice(first2).replace(/<a:pPr\b[^>]*\/>|<a:pPr\b[^>]*>[\s\S]*?<\/a:pPr>/g, "");
+    return `<a:p>${inner.slice(0, first2)}${tail}</a:p>`;
+  });
+}
+async function dedupeParagraphProps(bytes) {
+  const zip2 = await import_jszip.default.loadAsync(Buffer.from(bytes));
+  let changed = false;
+  for (const name of Object.keys(zip2.files).filter((n) => /^ppt\/slides\/slide\d+\.xml$/.test(n))) {
+    const xml = await zip2.file(name).async("string");
+    const next = dedupeParagraphPropsXml(xml);
+    if (next !== xml) {
+      zip2.file(name, next);
+      changed = true;
+    }
+  }
+  if (!changed)
+    return bytes;
+  return new Uint8Array(await zip2.generateAsync({ type: "nodebuffer", compression: "DEFLATE" }));
+}
+async function normalizeChartParts(bytes) {
+  const zip2 = await import_jszip.default.loadAsync(Buffer.from(bytes));
+  const pres = zip2.file("ppt/presentation.xml");
+  if (pres) {
+    let xml = await pres.async("string");
+    const notes = xml.match(/<p:notesMasterIdLst>[\s\S]*?<\/p:notesMasterIdLst>/);
+    const sldIdx = xml.indexOf("<p:sldIdLst>");
+    if (notes && sldIdx >= 0 && xml.indexOf(notes[0]) > sldIdx) {
+      xml = xml.replace(notes[0], "");
+      xml = xml.replace("<p:sldIdLst>", `${notes[0]}<p:sldIdLst>`);
+      zip2.file("ppt/presentation.xml", xml);
+    }
+  }
+  const parts = Object.keys(zip2.files).filter((n) => /^(ppt|word|xl)\/charts\/chart\d+\.xml$/.test(n));
+  if (!parts.length && !pres)
+    return bytes;
+  for (const name of parts) {
+    let xml = await zip2.file(name).async("string");
+    const isLineArea = /<c:(line|area)Chart>/.test(xml);
+    for (const ct of ["lineChart", "areaChart"]) {
+      xml = xml.replace(new RegExp(`<c:${ct}>(?!\\s*<c:grouping)`, "g"), `<c:${ct}><c:grouping val="standard"/>`);
+    }
+    if (isLineArea) {
+      xml = xml.replace(/<c:invertIfNegative[^>]*\/>/g, "");
+      xml = xml.replace(/<c:ser>[\s\S]*?<\/c:ser>/g, (ser) => {
+        const m = ser.match(/<c:marker>[\s\S]*?<\/c:marker>|<c:marker\s*\/>/);
+        const dpos = ser.indexOf("<c:dLbls");
+        if (!m || dpos < 0 || ser.indexOf(m[0]) < dpos)
+          return ser;
+        const without = ser.replace(m[0], "");
+        const at = without.indexOf("<c:dLbls");
+        return without.slice(0, at) + m[0] + without.slice(at);
+      });
+    }
+    xml = xml.replace(/<c:ser>[\s\S]*?<\/c:ser>/g, (ser) => {
+      const dPts = ser.match(/<c:dPt>[\s\S]*?<\/c:dPt>/g);
+      if (!dPts)
+        return ser;
+      const stripped = ser.replace(/<c:dPt>[\s\S]*?<\/c:dPt>/g, "");
+      for (const anchor of ["<c:dLbls", "<c:trendline", "<c:errBars", "<c:cat>", "<c:val>", "</c:ser>"]) {
+        const pos = stripped.indexOf(anchor);
+        if (pos >= 0)
+          return stripped.slice(0, pos) + dPts.join("") + stripped.slice(pos);
+      }
+      return ser;
+    });
+    const declared = new Set([...xml.matchAll(/<c:(?:catAx|valAx|serAx|dateAx)>\s*<c:axId val="(\d+)"\/>/g)].map((m) => m[1]));
+    xml = xml.replace(/<c:axId val="(\d+)"\/>/g, (m, id) => declared.has(id) ? m : "");
+    zip2.file(name, xml);
+  }
+  const out = await zip2.generateAsync({ type: "nodebuffer", compression: "DEFLATE" });
+  return new Uint8Array(out);
+}
+async function stripDanglingOverrides(bytes) {
+  const zip2 = await import_jszip.default.loadAsync(Buffer.from(bytes));
+  const ctFile = zip2.file("[Content_Types].xml");
+  if (!ctFile)
+    return bytes;
+  const ct = await ctFile.async("string");
+  const cleaned = ct.replace(/<Override\b[^>]*PartName="\/([^"]+)"[^>]*\/>/g, (m, part) => zip2.file(part) ? m : "");
+  if (cleaned === ct)
+    return bytes;
+  zip2.file("[Content_Types].xml", cleaned);
+  return new Uint8Array(await zip2.generateAsync({ type: "nodebuffer", compression: "DEFLATE" }));
+}
+var JSZIP_OUTPUT_TYPES = /* @__PURE__ */ new Set(["nodebuffer", "uint8array", "arraybuffer", "base64", "binarystring", "string", "array", "blob"]);
+function finalizingPptxGenJS(Base) {
+  const finalized = async (self2, compression) => {
+    const raw = await Base.prototype.write.call(self2, { outputType: "nodebuffer", compression });
+    return finalizePptx(new Uint8Array(raw));
+  };
+  const Finalizing = class extends Base {
+    async write(props) {
+      const p = props;
+      const outputType = typeof p === "object" && p?.outputType ? p.outputType : typeof p === "string" && p ? p : null;
+      const compression = typeof p === "object" && p?.compression ? true : false;
+      if (outputType === "STREAM")
+        return this.stream({ compression });
+      const bytes = await finalized(this, compression);
+      const type = outputType ?? "blob";
+      if (!JSZIP_OUTPUT_TYPES.has(type))
+        return Base.prototype.write.call(this, props);
+      if (type === "nodebuffer")
+        return Buffer.from(bytes);
+      if (type === "uint8array")
+        return bytes;
+      const zip2 = await import_jszip.default.loadAsync(Buffer.from(bytes));
+      return zip2.generateAsync({ type, compression: compression ? "DEFLATE" : "STORE" });
+    }
+    async stream(props) {
+      return Buffer.from(await finalized(this, Boolean(props?.compression)));
+    }
+    async writeFile(props) {
+      if (typeof props === "string")
+        console.log("Warning: `writeFile(filename)` is deprecated - please use `WriteFileProps` argument (v3.5.0)");
+      const name = typeof props === "object" && props?.fileName ? props.fileName : typeof props === "string" ? props : "";
+      const fileName = name ? String(name).toLowerCase().endsWith(".pptx") ? String(name) : `${name}.pptx` : "Presentation.pptx";
+      const compression = typeof props === "object" && props?.compression ? true : false;
+      await (0, import_promises.writeFile)(fileName, await finalized(this, compression));
+      return fileName;
+    }
+  };
+  Object.defineProperty(Finalizing, "name", { value: Base.name || "PptxGenJS" });
+  return Finalizing;
+}
+
+// apps/admin/data/skills.lib/document/bin-src/doc-run.mjs
 var REGISTRY = {
-  pptxgenjs: import_pptxgenjs.default,
-  jszip: import_jszip.default,
+  pptxgenjs: finalizingPptxGenJS(import_pptxgenjs.default),
+  jszip: import_jszip2.default,
   pizzip: import_pizzip.default,
   "pdf-lib": pdfLib,
   docx: dist_exports,

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

@@ -8,6 +8,17 @@ var import_node_path2 = require("node:path");
 // packages/core/dist/document/pdf-extract.js
 var import_node_path = require("node:path");
 var import_node_url = require("node:url");
+
+// packages/core/dist/document/pdf-encrypted.js
+var ENCRYPTED_PDF_MESSAGE = "PDF is password-protected (encrypted); decrypt first: pdf-ops decrypt <in> <password> <out>";
+var PdfEncryptedError = class extends Error {
+  constructor() {
+    super(ENCRYPTED_PDF_MESSAGE);
+    this.name = "PdfEncryptedError";
+  }
+};
+
+// packages/core/dist/document/pdf-extract.js
 var _mupdf = null;
 function loadMupdf() {
   if (!_mupdf) {
@@ -33,6 +44,8 @@ function textTableHint(text) {
 async function openDoc(bytes) {
   const mupdf = await loadMupdf();
   const doc = mupdf.Document.openDocument(bytes, "application/pdf");
+  if (doc.needsPassword())
+    throw new PdfEncryptedError();
   if (doc.countPages() === 0) {
     throw new Error("PDF has no readable pages (corrupt, encrypted, or empty)");
   }

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

@@ -8,6 +8,17 @@ var import_node_path2 = require("node:path");
 // packages/core/dist/document/pdf-extract.js
 var import_node_path = require("node:path");
 var import_node_url = require("node:url");
+
+// packages/core/dist/document/pdf-encrypted.js
+var ENCRYPTED_PDF_MESSAGE = "PDF is password-protected (encrypted); decrypt first: pdf-ops decrypt <in> <password> <out>";
+var PdfEncryptedError = class extends Error {
+  constructor() {
+    super(ENCRYPTED_PDF_MESSAGE);
+    this.name = "PdfEncryptedError";
+  }
+};
+
+// packages/core/dist/document/pdf-extract.js
 var _mupdf = null;
 function loadMupdf() {
   if (!_mupdf) {
@@ -33,6 +44,8 @@ function textTableHint(text) {
 async function openDoc(bytes) {
   const mupdf = await loadMupdf();
   const doc = mupdf.Document.openDocument(bytes, "application/pdf");
+  if (doc.needsPassword())
+    throw new PdfEncryptedError();
   if (doc.countPages() === 0) {
     throw new Error("PDF has no readable pages (corrupt, encrypted, or empty)");
   }