@taloswork/ose-shared 0.1.12 → 0.1.14

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

@@ -24,7 +24,12 @@ function loadMupdf() {
   if (!_mupdf) {
     const deps = process.env.TALOS_SKILL_DEPS;
     const spec = deps ? (0, import_node_url.pathToFileURL)((0, import_node_path.join)(deps, "mupdf", "dist", "mupdf.js")).href : "mupdf";
-    _mupdf = import(spec);
+    const loading = import(spec);
+    _mupdf = loading;
+    loading.catch(() => {
+      if (_mupdf === loading)
+        _mupdf = null;
+    });
   }
   return _mupdf;
 }

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

@@ -24,7 +24,12 @@ function loadMupdf() {
   if (!_mupdf) {
     const deps = process.env.TALOS_SKILL_DEPS;
     const spec = deps ? (0, import_node_url.pathToFileURL)((0, import_node_path.join)(deps, "mupdf", "dist", "mupdf.js")).href : "mupdf";
-    _mupdf = import(spec);
+    const loading = import(spec);
+    _mupdf = loading;
+    loading.catch(() => {
+      if (_mupdf === loading)
+        _mupdf = null;
+    });
   }
   return _mupdf;
 }

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

@@ -9790,9 +9790,9 @@ var require_load = __commonJS({
 var require_lib3 = __commonJS({
   "node_modules/jszip/lib/index.js"(exports2, module2) {
     "use strict";
-    function JSZip4() {
-      if (!(this instanceof JSZip4)) {
-        return new JSZip4();
+    function JSZip6() {
+      if (!(this instanceof JSZip6)) {
+        return new JSZip6();
       }
       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 JSZip4();
+        var newObj = new JSZip6();
         for (var i in this) {
           if (typeof this[i] !== "function") {
             newObj[i] = this[i];
@@ -9810,16 +9810,16 @@ var require_lib3 = __commonJS({
         return newObj;
       };
     }
-    JSZip4.prototype = require_object();
-    JSZip4.prototype.loadAsync = require_load();
-    JSZip4.support = require_support();
-    JSZip4.defaults = require_defaults();
-    JSZip4.version = "3.10.1";
-    JSZip4.loadAsync = function(content, options) {
-      return new JSZip4().loadAsync(content, options);
+    JSZip6.prototype = require_object();
+    JSZip6.prototype.loadAsync = require_load();
+    JSZip6.support = require_support();
+    JSZip6.defaults = require_defaults();
+    JSZip6.version = "3.10.1";
+    JSZip6.loadAsync = function(content, options) {
+      return new JSZip6().loadAsync(content, options);
     };
-    JSZip4.external = require_external();
-    module2.exports = JSZip4;
+    JSZip6.external = require_external();
+    module2.exports = JSZip6;
   }
 });
 
@@ -27608,6 +27608,7 @@ __export(pdf_extract_exports, {
   extractPdfText: () => extractPdfText,
   extractPdfTextLines: () => extractPdfTextLines,
   pdfNeedsPassword: () => pdfNeedsPassword,
+  pngPixels: () => pngPixels,
   renderPdfPages: () => renderPdfPages,
   scanPdfSignals: () => scanPdfSignals,
   textMathHint: () => textMathHint,
@@ -27617,7 +27618,12 @@ function loadMupdf() {
   if (!_mupdf) {
     const deps = process.env.TALOS_SKILL_DEPS;
     const spec = deps ? (0, import_node_url.pathToFileURL)((0, import_node_path.join)(deps, "mupdf", "dist", "mupdf.js")).href : "mupdf";
-    _mupdf = import(spec);
+    const loading = import(spec);
+    _mupdf = loading;
+    loading.catch(() => {
+      if (_mupdf === loading)
+        _mupdf = null;
+    });
   }
   return _mupdf;
 }
@@ -27901,6 +27907,13 @@ async function renderPdfPages(bytes, pages, scale = 2, maxEdge) {
   }
   return out;
 }
+async function pngPixels(png) {
+  const mupdf = await loadMupdf();
+  const pix = new mupdf.Image(png).toPixmap();
+  const width = pix.getWidth(), height = pix.getHeight();
+  const channels = pix.getStride() / width;
+  return { width, height, channels, data: pix.getPixels() };
+}
 async function extractPdfImages(bytes, pages) {
   const { doc } = await openDoc(bytes);
   const total = doc.countPages();
@@ -27973,8 +27986,74 @@ var import_promises2 = require("node:fs/promises");
 var import_node_path3 = require("node:path");
 
 // packages/core/dist/document/doc-validate.js
-var import_jszip = __toESM(require_lib3(), 1);
+var import_jszip3 = __toESM(require_lib3(), 1);
 var import_pdf_lib = __toESM(require_cjs(), 1);
+
+// packages/core/dist/document/pptx-notes.js
+var import_jszip = __toESM(require_lib3(), 1);
+
+// packages/core/dist/document/pptx-parts.js
+var SLIDE_PART = /^ppt\/slides\/(slide\d+\.xml)$/;
+function relsPathFor(path) {
+  return path.replace(SLIDE_PART, "ppt/slides/_rels/$1.rels");
+}
+function attr(el, name) {
+  return new RegExp(`\\b${name}="([^"]*)"`).exec(el)?.[1];
+}
+async function slidePartsInOrder(zip) {
+  const byNumber = Object.keys(zip.files).filter((p) => SLIDE_PART.test(p)).sort((a, b) => Number(/(\d+)\.xml$/.exec(a)[1]) - Number(/(\d+)\.xml$/.exec(b)[1]));
+  const toParts = (paths) => paths.map((path, i) => ({ position: i + 1, path, relsPath: relsPathFor(path) }));
+  const pres = await zip.file("ppt/presentation.xml")?.async("string");
+  const rels = await zip.file("ppt/_rels/presentation.xml.rels")?.async("string");
+  if (!pres || !rels)
+    return toParts(byNumber);
+  const targetById = /* @__PURE__ */ new Map();
+  for (const [el] of rels.matchAll(/<Relationship\b[^>]*>/g)) {
+    const id = attr(el, "Id");
+    const target = attr(el, "Target");
+    if (!id || !target)
+      continue;
+    targetById.set(id, target.startsWith("/") ? target.slice(1) : `ppt/${target.replace(/^\.\//, "")}`);
+  }
+  const list = /<p:sldIdLst\b[^>]*>([\s\S]*?)<\/p:sldIdLst>/.exec(pres)?.[1] ?? "";
+  const ordered = [...list.matchAll(/<p:sldId\b[^>]*>/g)].map(([el]) => targetById.get(attr(el, "r:id") ?? "")).filter((p) => p !== void 0 && SLIDE_PART.test(p) && zip.file(p) !== null);
+  const complete = ordered.length === byNumber.length && new Set(ordered).size === ordered.length;
+  return toParts(complete ? ordered : byNumber);
+}
+
+// packages/core/dist/document/pptx-notes.js
+var REL_NS = "http://schemas.openxmlformats.org/officeDocument/2006/relationships";
+var NOTES_SLIDE_REL = `${REL_NS}/notesSlide`;
+var NOTES_MASTER_REL = `${REL_NS}/notesMaster`;
+var SLIDE_REL = `${REL_NS}/slide`;
+async function notesKeynoteIssues(zip) {
+  const out = [];
+  for (const p of Object.keys(zip.files).filter((p2) => /^ppt\/notesSlides\/notesSlide\d+\.xml$/.test(p2)).sort()) {
+    const xml = await zip.file(p).async("string");
+    if (!/<p:clrMapOvr\b/.test(xml))
+      out.push(p);
+  }
+  return out;
+}
+async function unlistedNotesMaster(zip) {
+  const relsFile = zip.file("ppt/_rels/presentation.xml.rels");
+  const presFile = zip.file("ppt/presentation.xml");
+  if (!relsFile || !presFile)
+    return null;
+  const rel = new RegExp(`<Relationship\\b[^>]*Type="${NOTES_MASTER_REL}"[^>]*>`).exec(await relsFile.async("string"))?.[0];
+  const rid = rel ? /\bId="([^"]+)"/.exec(rel)?.[1] : void 0;
+  if (!rid)
+    return null;
+  return /<p:notesMasterId\b/.test(await presFile.async("string")) ? null : rid;
+}
+
+// packages/core/dist/document/docx-defaults.js
+var import_jszip2 = __toESM(require_lib3(), 1);
+function hasDefaultParagraphStyle(stylesXml) {
+  return [...stylesXml.matchAll(/<w:style\b[^>]*>/g)].some(([tag]) => /\bw:type="paragraph"/.test(tag) && /\bw:default="(1|true|on)"/.test(tag));
+}
+
+// packages/core/dist/document/doc-validate.js
 function looksLikePdf(bytes) {
   const head = Buffer.from(bytes.subarray(0, 1024)).toString("latin1");
   return head.includes("%PDF-");
@@ -28154,7 +28233,7 @@ async function validateOoxml(bytes) {
   const result = (kind) => ({ ok: errors.length === 0, kind, errors, warnings });
   let zip;
   try {
-    zip = await import_jszip.default.loadAsync(Buffer.from(bytes));
+    zip = await import_jszip3.default.loadAsync(Buffer.from(bytes));
   } catch (e) {
     errors.push({ code: "zip-unreadable", message: `not a readable OOXML zip: ${e.message}` });
     return result("unknown");
@@ -28222,6 +28301,30 @@ async function validateOoxml(bytes) {
       }
     }
   }
+  if (root.kind === "pptx") {
+    const bad = await notesKeynoteIssues(zip);
+    if (bad.length) {
+      errors.push({
+        code: "notes-keynote",
+        part: bad[0],
+        message: `${bad.length} notes slide(s) lack <p:clrMapOvr> (${bad.slice(0, 3).join(", ")}${bad.length > 3 ? ", \u2026" : ""}) \u2014 Keynote refuses the whole file ("\u6587\u4EF6\u683C\u5F0F\u65E0\u6548"). Repair with \`pptx-notes <in.pptx> {} <out.pptx>\`; write notes with pptx-notes, never by hand`
+      });
+    }
+  }
+  if (root.kind === "pptx" && await unlistedNotesMaster(zip)) {
+    errors.push({
+      code: "notes-keynote",
+      part: "ppt/presentation.xml",
+      message: 'the notes master is not listed in <p:notesMasterIdLst> (python-pptx leaves it out) \u2014 Keynote refuses the whole file ("\u6587\u4EF6\u683C\u5F0F\u65E0\u6548"). Repair with `pptx-notes <in.pptx> {} <out.pptx>`'
+    });
+  }
+  if (root.kind === "docx" && has("word/styles.xml") && !hasDefaultParagraphStyle(await text("word/styles.xml"))) {
+    errors.push({
+      code: "docx-default-style",
+      part: "word/styles.xml",
+      message: 'no default paragraph style (w:default="1", usually "Normal") \u2014 Word fills the gap, but Pages lays every unstyled paragraph out like the one before it (a centred title turns the whole body huge and centred). Repair with `docx-defaults <in.docx> <out.docx>`; docx from doc-gen already has it'
+    });
+  }
   return result(root.kind);
 }
 
@@ -28229,13 +28332,18 @@ async function validateOoxml(bytes) {
 var import_node_path2 = require("node:path");
 var import_node_url2 = require("node:url");
 var import_promises = require("node:fs/promises");
-var import_jszip2 = __toESM(require_lib3(), 1);
+var import_jszip4 = __toESM(require_lib3(), 1);
 var _xmllint = null;
 function loadXmllint() {
   if (!_xmllint) {
     const deps = process.env.TALOS_SKILL_DEPS;
     const spec = deps ? (0, import_node_url2.pathToFileURL)((0, import_node_path2.join)(deps, "xmllint-wasm", "index-node.js")).href : "xmllint-wasm";
-    _xmllint = import(spec);
+    const loading = import(spec);
+    _xmllint = loading;
+    loading.catch(() => {
+      if (_xmllint === loading)
+        _xmllint = null;
+    });
   }
   return _xmllint;
 }
@@ -28285,7 +28393,7 @@ function errText(e) {
 }
 async function validateOoxmlStrict(bytes, schemaDir) {
   const [{ validateXML }, schemas] = await Promise.all([loadXmllint(), loadSchemaSet(schemaDir)]);
-  const zip = await import_jszip2.default.loadAsync(Buffer.from(bytes));
+  const zip = await import_jszip4.default.loadAsync(Buffer.from(bytes));
   const checked = [];
   const skipped = [];
   const errors = [];
@@ -28328,7 +28436,7 @@ async function validateOoxmlStrict(bytes, schemaDir) {
 }
 
 // packages/core/dist/document/doc-design-lint.js
-var import_jszip3 = __toESM(require_lib3(), 1);
+var import_jszip5 = __toESM(require_lib3(), 1);
 
 // packages/core/dist/document/ooxml-util.js
 function unescXml(s) {
@@ -28352,35 +28460,6 @@ function unescXml(s) {
   });
 }
 
-// packages/core/dist/document/pptx-parts.js
-var SLIDE_PART = /^ppt\/slides\/(slide\d+\.xml)$/;
-function relsPathFor(path) {
-  return path.replace(SLIDE_PART, "ppt/slides/_rels/$1.rels");
-}
-function attr(el, name) {
-  return new RegExp(`\\b${name}="([^"]*)"`).exec(el)?.[1];
-}
-async function slidePartsInOrder(zip) {
-  const byNumber = Object.keys(zip.files).filter((p) => SLIDE_PART.test(p)).sort((a, b) => Number(/(\d+)\.xml$/.exec(a)[1]) - Number(/(\d+)\.xml$/.exec(b)[1]));
-  const toParts = (paths) => paths.map((path, i) => ({ position: i + 1, path, relsPath: relsPathFor(path) }));
-  const pres = await zip.file("ppt/presentation.xml")?.async("string");
-  const rels = await zip.file("ppt/_rels/presentation.xml.rels")?.async("string");
-  if (!pres || !rels)
-    return toParts(byNumber);
-  const targetById = /* @__PURE__ */ new Map();
-  for (const [el] of rels.matchAll(/<Relationship\b[^>]*>/g)) {
-    const id = attr(el, "Id");
-    const target = attr(el, "Target");
-    if (!id || !target)
-      continue;
-    targetById.set(id, target.startsWith("/") ? target.slice(1) : `ppt/${target.replace(/^\.\//, "")}`);
-  }
-  const list = /<p:sldIdLst\b[^>]*>([\s\S]*?)<\/p:sldIdLst>/.exec(pres)?.[1] ?? "";
-  const ordered = [...list.matchAll(/<p:sldId\b[^>]*>/g)].map(([el]) => targetById.get(attr(el, "r:id") ?? "")).filter((p) => p !== void 0 && SLIDE_PART.test(p) && zip.file(p) !== null);
-  const complete = ordered.length === byNumber.length && new Set(ordered).size === ordered.length;
-  return toParts(complete ? ordered : byNumber);
-}
-
 // packages/core/dist/document/png-decode.js
 var import_node_zlib = require("node:zlib");
 var SIG = [137, 80, 78, 71, 13, 10, 26, 10];
@@ -28477,6 +28556,14 @@ function decodePng(bytes) {
   }
 }
 
+// packages/core/dist/document/readable-type.js
+var MIN_FONT_PT = 10;
+var LABEL_MAX_CHARS = 24;
+var MIN_LABEL_FONT_PT = 8;
+function readableFloorPt(text) {
+  return [...text.trim()].length <= LABEL_MAX_CHARS ? MIN_LABEL_FONT_PT : MIN_FONT_PT;
+}
+
 // packages/core/dist/document/text-slack.js
 var SLACK_CJK_RE = /[\u3000-\u303F\u3400-\u9FFF\uF900-\uFAFF\uFF00-\uFFEF]/;
 var PCT = { single: { cjk: 0.22, latin: 0.16 }, multi: { cjk: 0.1, latin: 0.06 } };
@@ -28672,9 +28759,7 @@ function lintSlideXml(slideXml, slideNumber, masterBg, expectedLang = detectDeck
 var EMU_PER_INCH = 914400;
 var DEFAULT_CANVAS = { w: 12192e3, h: 6858e3 };
 var OFFCANVAS_TOL = EMU_PER_INCH / 10;
-var MIN_FONT_HUNDREDTHS = 1e3;
-var LABEL_MAX_CHARS = 24;
-var MIN_LABEL_FONT_HUNDREDTHS = 800;
+var MIN_FONT_HUNDREDTHS = MIN_FONT_PT * 100;
 var CHART_TEXT_RE = new RegExp(`<p:cNvPr\\b[^>]*\\bname="${CHART_TEXT_NAME}\\b`);
 var OVERLAP_RATIO = 0.3;
 function firstBox(block) {
@@ -28776,7 +28861,7 @@ function lintSlideGeometry(slideXml, slideNumber, canvas = DEFAULT_CANVAS) {
         continue;
       const text = unescXml(raw).trim();
       const sz = /<a:rPr\b[^>]*\bsz="(\d+)"/.exec(run)?.[1];
-      const floor = [...text].length <= LABEL_MAX_CHARS ? MIN_LABEL_FONT_HUNDREDTHS : MIN_FONT_HUNDREDTHS;
+      const floor = readableFloorPt(text) * 100;
       if (sz && +sz < floor && (!smallest || +sz < smallest.sz)) {
         smallest = { sz: +sz, text, floor };
       }
@@ -28951,7 +29036,7 @@ async function backdropRaster(zip, slideXml, relsPath, canvas) {
   return media ? decodePng(await media.async("uint8array")) : void 0;
 }
 async function designLintPptx(bytes) {
-  const zip = await import_jszip3.default.loadAsync(Buffer.from(bytes));
+  const zip = await import_jszip5.default.loadAsync(Buffer.from(bytes));
   const findings = [];
   let masterBg = { colored: false };
   const masterFile = Object.keys(zip.files).find((p) => /^ppt\/slideMasters\/slideMaster\d+\.xml$/.test(p));