officeparser 7.7.0 → 7.8.0

package/dist/sbom.cdx.json

@@ -3,9 +3,9 @@
   "bomFormat": "CycloneDX",
   "specVersion": "1.6",
   "version": 1,
-  "serialNumber": "urn:uuid:394bb8c0-cf83-4786-b6f3-bf5883961cd2",
+  "serialNumber": "urn:uuid:f41b2f13-0241-44b2-bd4b-1ae51d458463",
   "metadata": {
-    "timestamp": "2026-08-18T07:53:54.343Z",
+    "timestamp": "2026-08-18T21:56:11.512Z",
     "tools": {
       "components": [
         {
@@ -98,8 +98,8 @@
     "component": {
       "type": "application",
       "name": "officeparser",
-      "version": "7.7.0",
-      "bom-ref": "officeparser@7.7.0",
+      "version": "7.8.0",
+      "bom-ref": "officeparser@7.8.0",
       "author": "Harsh Ankur",
       "description": "A robust, strictly-typed Node.js and Browser library for parsing office files (.docx, .pptx, .xlsx, .odt, .odp, .ods, .pdf, .rtf, .csv, .md, .html, .epub) and generating high-fidelity outputs in Markdown, HTML, CSV, RTF, PDF, EPUB, and RAG-focused chunks.",
       "licenses": [
@@ -110,7 +110,7 @@
           }
         }
       ],
-      "purl": "pkg:npm/officeparser@7.7.0?vcs_url=git%2Bhttps%3A%2F%2Fgithub.com%2Fharshankur%2FofficeParser.git",
+      "purl": "pkg:npm/officeparser@7.8.0?vcs_url=git%2Bhttps%3A%2F%2Fgithub.com%2Fharshankur%2FofficeParser.git",
       "externalReferences": [
         {
           "url": "git+https://github.com/harshankur/officeParser.git",
@@ -146,7 +146,7 @@
       "name": "xmldom",
       "group": "@xmldom",
       "version": "0.9.10",
-      "bom-ref": "officeparser@7.7.0|@xmldom/xmldom@0.9.10",
+      "bom-ref": "officeparser@7.8.0|@xmldom/xmldom@0.9.10",
       "description": "A pure JavaScript W3C standard-based (XML DOM Level 2 Core) DOMParser and XMLSerializer module.",
       "licenses": [
         {
@@ -200,7 +200,7 @@
       "type": "library",
       "name": "fflate",
       "version": "0.8.3",
-      "bom-ref": "officeparser@7.7.0|fflate@0.8.3",
+      "bom-ref": "officeparser@7.8.0|fflate@0.8.3",
       "author": "Arjun Barrett",
       "description": "High performance (de)compression in an 8kB package",
       "licenses": [
@@ -251,7 +251,7 @@
       "type": "library",
       "name": "file-type",
       "version": "22.0.1",
-      "bom-ref": "officeparser@7.7.0|file-type@22.0.1",
+      "bom-ref": "officeparser@7.8.0|file-type@22.0.1",
       "author": "Sindre Sorhus",
       "description": "Detect the file type of a file, stream, or data",
       "licenses": [
@@ -306,7 +306,7 @@
       "type": "library",
       "name": "pdfjs-dist",
       "version": "6.1.200",
-      "bom-ref": "officeparser@7.7.0|pdfjs-dist@6.1.200",
+      "bom-ref": "officeparser@7.8.0|pdfjs-dist@6.1.200",
       "description": "Generic build of Mozilla's PDF.js library.",
       "licenses": [
         {
@@ -360,7 +360,7 @@
       "type": "library",
       "name": "tesseract.js",
       "version": "7.0.0",
-      "bom-ref": "officeparser@7.7.0|tesseract.js@7.0.0",
+      "bom-ref": "officeparser@7.8.0|tesseract.js@7.0.0",
       "description": "Pure Javascript Multilingual OCR",
       "licenses": [
         {
@@ -411,7 +411,7 @@
       "name": "inflate",
       "group": "@tokenizer",
       "version": "0.4.1",
-      "bom-ref": "officeparser@7.7.0|@tokenizer/inflate@0.4.1",
+      "bom-ref": "officeparser@7.8.0|@tokenizer/inflate@0.4.1",
       "author": "Borewit",
       "description": "Tokenized zip support",
       "licenses": [
@@ -466,7 +466,7 @@
       "type": "library",
       "name": "strtok3",
       "version": "10.3.5",
-      "bom-ref": "officeparser@7.7.0|strtok3@10.3.5",
+      "bom-ref": "officeparser@7.8.0|strtok3@10.3.5",
       "author": "Borewit",
       "description": "A promise based streaming tokenizer",
       "licenses": [
@@ -521,7 +521,7 @@
       "type": "library",
       "name": "token-types",
       "version": "6.1.2",
-      "bom-ref": "officeparser@7.7.0|token-types@6.1.2",
+      "bom-ref": "officeparser@7.8.0|token-types@6.1.2",
       "author": "Borewit",
       "description": "Common token types for decoding and encoding numeric and string values",
       "licenses": [
@@ -576,7 +576,7 @@
       "type": "library",
       "name": "uint8array-extras",
       "version": "1.5.0",
-      "bom-ref": "officeparser@7.7.0|uint8array-extras@1.5.0",
+      "bom-ref": "officeparser@7.8.0|uint8array-extras@1.5.0",
       "author": "Sindre Sorhus",
       "description": "Useful utilities for working with Uint8Array (and Buffer)",
       "licenses": [
@@ -632,7 +632,7 @@
       "name": "canvas",
       "group": "@napi-rs",
       "version": "1.0.1",
-      "bom-ref": "officeparser@7.7.0|@napi-rs/canvas@1.0.1",
+      "bom-ref": "officeparser@7.8.0|@napi-rs/canvas@1.0.1",
       "description": "Canvas for Node.js with skia backend",
       "scope": "optional",
       "licenses": [
@@ -687,7 +687,7 @@
       "type": "library",
       "name": "bmp-js",
       "version": "0.1.0",
-      "bom-ref": "officeparser@7.7.0|bmp-js@0.1.0",
+      "bom-ref": "officeparser@7.8.0|bmp-js@0.1.0",
       "author": "shaozilee",
       "description": "A pure javascript BMP encoder and decoder",
       "licenses": [
@@ -738,7 +738,7 @@
       "type": "library",
       "name": "idb-keyval",
       "version": "6.2.2",
-      "bom-ref": "officeparser@7.7.0|idb-keyval@6.2.2",
+      "bom-ref": "officeparser@7.8.0|idb-keyval@6.2.2",
       "author": "Jake Archibald",
       "description": "A super-simple-small keyval store built on top of IndexedDB",
       "licenses": [
@@ -789,7 +789,7 @@
       "type": "library",
       "name": "is-url",
       "version": "1.2.4",
-      "bom-ref": "officeparser@7.7.0|is-url@1.2.4",
+      "bom-ref": "officeparser@7.8.0|is-url@1.2.4",
       "description": "Check whether a string is a URL.",
       "licenses": [
         {
@@ -839,7 +839,7 @@
       "type": "library",
       "name": "node-fetch",
       "version": "2.7.0",
-      "bom-ref": "officeparser@7.7.0|node-fetch@2.7.0",
+      "bom-ref": "officeparser@7.8.0|node-fetch@2.7.0",
       "author": "David Frank",
       "description": "A light-weight module that brings window.fetch to node.js",
       "licenses": [
@@ -894,7 +894,7 @@
       "type": "library",
       "name": "opencollective-postinstall",
       "version": "2.0.3",
-      "bom-ref": "officeparser@7.7.0|opencollective-postinstall@2.0.3",
+      "bom-ref": "officeparser@7.8.0|opencollective-postinstall@2.0.3",
       "author": "Xavier Damman",
       "description": "Lightweight npm postinstall message to invite people to donate to your collective",
       "licenses": [
@@ -945,7 +945,7 @@
       "type": "library",
       "name": "regenerator-runtime",
       "version": "0.13.11",
-      "bom-ref": "officeparser@7.7.0|regenerator-runtime@0.13.11",
+      "bom-ref": "officeparser@7.8.0|regenerator-runtime@0.13.11",
       "author": "Ben Newman",
       "description": "Runtime for Regenerator-compiled generator and async functions.",
       "licenses": [
@@ -996,7 +996,7 @@
       "type": "library",
       "name": "tesseract.js-core",
       "version": "7.0.0",
-      "bom-ref": "officeparser@7.7.0|tesseract.js-core@7.0.0",
+      "bom-ref": "officeparser@7.8.0|tesseract.js-core@7.0.0",
       "author": "antimatter15",
       "description": "Tesseract C++ API in Pure Javascript",
       "licenses": [
@@ -1047,7 +1047,7 @@
       "type": "library",
       "name": "wasm-feature-detect",
       "version": "1.8.0",
-      "bom-ref": "officeparser@7.7.0|wasm-feature-detect@1.8.0",
+      "bom-ref": "officeparser@7.8.0|wasm-feature-detect@1.8.0",
       "author": "Surma",
       "description": "A small library to detect which features of WebAssembly are supported in your current browser.",
       "licenses": [
@@ -1098,7 +1098,7 @@
       "type": "library",
       "name": "zlibjs",
       "version": "0.3.1",
-      "bom-ref": "officeparser@7.7.0|zlibjs@0.3.1",
+      "bom-ref": "officeparser@7.8.0|zlibjs@0.3.1",
       "author": "Yuta Imaya",
       "description": "zlib, gzip and zip implementation in JavaScript",
       "licenses": [
@@ -1153,7 +1153,7 @@
       "type": "library",
       "name": "debug",
       "version": "4.4.3",
-      "bom-ref": "officeparser@7.7.0|debug@4.4.3",
+      "bom-ref": "officeparser@7.8.0|debug@4.4.3",
       "author": "Josh Junon",
       "description": "Lightweight debugging utility for Node.js and the browser",
       "licenses": [
@@ -1209,7 +1209,7 @@
       "name": "token",
       "group": "@tokenizer",
       "version": "0.3.0",
-      "bom-ref": "officeparser@7.7.0|@tokenizer/token@0.3.0",
+      "bom-ref": "officeparser@7.8.0|@tokenizer/token@0.3.0",
       "author": "Borewit",
       "description": "TypeScript definition for strtok3 token",
       "licenses": [
@@ -1261,7 +1261,7 @@
       "name": "text-codec",
       "group": "@borewit",
       "version": "0.2.2",
-      "bom-ref": "officeparser@7.7.0|@borewit/text-codec@0.2.2",
+      "bom-ref": "officeparser@7.8.0|@borewit/text-codec@0.2.2",
       "author": "Borewit",
       "description": "Text Decoder",
       "licenses": [
@@ -1312,7 +1312,7 @@
       "type": "library",
       "name": "ieee754",
       "version": "1.2.1",
-      "bom-ref": "officeparser@7.7.0|ieee754@1.2.1",
+      "bom-ref": "officeparser@7.8.0|ieee754@1.2.1",
       "author": "Feross Aboukhadijeh",
       "description": "Read/write IEEE754 floating point numbers from/to a Buffer or array-like object",
       "licenses": [
@@ -1364,7 +1364,7 @@
       "name": "canvas-linux-x64-gnu",
       "group": "@napi-rs",
       "version": "1.0.1",
-      "bom-ref": "officeparser@7.7.0|@napi-rs/canvas-linux-x64-gnu@1.0.1",
+      "bom-ref": "officeparser@7.8.0|@napi-rs/canvas-linux-x64-gnu@1.0.1",
       "description": "Canvas for Node.js with skia backend",
       "scope": "optional",
       "licenses": [
@@ -1419,7 +1419,7 @@
       "type": "library",
       "name": "whatwg-url",
       "version": "5.0.0",
-      "bom-ref": "officeparser@7.7.0|whatwg-url@5.0.0",
+      "bom-ref": "officeparser@7.8.0|whatwg-url@5.0.0",
       "author": "Sebastian Mayr",
       "description": "An implementation of the WHATWG URL Standard's URL API and parsing machinery",
       "licenses": [
@@ -1470,7 +1470,7 @@
       "type": "library",
       "name": "ms",
       "version": "2.1.3",
-      "bom-ref": "officeparser@7.7.0|ms@2.1.3",
+      "bom-ref": "officeparser@7.8.0|ms@2.1.3",
       "description": "Tiny millisecond conversion utility",
       "licenses": [
         {
@@ -1520,7 +1520,7 @@
       "type": "library",
       "name": "tr46",
       "version": "0.0.3",
-      "bom-ref": "officeparser@7.7.0|tr46@0.0.3",
+      "bom-ref": "officeparser@7.8.0|tr46@0.0.3",
       "author": "Sebastian Mayr",
       "description": "An implementation of the Unicode TR46 spec",
       "licenses": [
@@ -1571,7 +1571,7 @@
       "type": "library",
       "name": "webidl-conversions",
       "version": "3.0.1",
-      "bom-ref": "officeparser@7.7.0|webidl-conversions@3.0.1",
+      "bom-ref": "officeparser@7.8.0|webidl-conversions@3.0.1",
       "author": "Domenic Denicola",
       "description": "Implements the WebIDL algorithms for converting to and from JavaScript values",
       "licenses": [
@@ -1621,143 +1621,143 @@
   ],
   "dependencies": [
     {
-      "ref": "officeparser@7.7.0",
+      "ref": "officeparser@7.8.0",
       "dependsOn": [
-        "officeparser@7.7.0|@xmldom/xmldom@0.9.10",
-        "officeparser@7.7.0|fflate@0.8.3",
-        "officeparser@7.7.0|file-type@22.0.1",
-        "officeparser@7.7.0|pdfjs-dist@6.1.200",
-        "officeparser@7.7.0|tesseract.js@7.0.0"
+        "officeparser@7.8.0|@xmldom/xmldom@0.9.10",
+        "officeparser@7.8.0|fflate@0.8.3",
+        "officeparser@7.8.0|file-type@22.0.1",
+        "officeparser@7.8.0|pdfjs-dist@6.1.200",
+        "officeparser@7.8.0|tesseract.js@7.0.0"
       ]
     },
     {
-      "ref": "officeparser@7.7.0|@xmldom/xmldom@0.9.10"
+      "ref": "officeparser@7.8.0|@xmldom/xmldom@0.9.10"
     },
     {
-      "ref": "officeparser@7.7.0|fflate@0.8.3"
+      "ref": "officeparser@7.8.0|fflate@0.8.3"
     },
     {
-      "ref": "officeparser@7.7.0|file-type@22.0.1",
+      "ref": "officeparser@7.8.0|file-type@22.0.1",
       "dependsOn": [
-        "officeparser@7.7.0|@tokenizer/inflate@0.4.1",
-        "officeparser@7.7.0|strtok3@10.3.5",
-        "officeparser@7.7.0|token-types@6.1.2",
-        "officeparser@7.7.0|uint8array-extras@1.5.0"
+        "officeparser@7.8.0|@tokenizer/inflate@0.4.1",
+        "officeparser@7.8.0|strtok3@10.3.5",
+        "officeparser@7.8.0|token-types@6.1.2",
+        "officeparser@7.8.0|uint8array-extras@1.5.0"
       ]
     },
     {
-      "ref": "officeparser@7.7.0|pdfjs-dist@6.1.200",
+      "ref": "officeparser@7.8.0|pdfjs-dist@6.1.200",
       "dependsOn": [
-        "officeparser@7.7.0|@napi-rs/canvas@1.0.1"
+        "officeparser@7.8.0|@napi-rs/canvas@1.0.1"
       ]
     },
     {
-      "ref": "officeparser@7.7.0|tesseract.js@7.0.0",
+      "ref": "officeparser@7.8.0|tesseract.js@7.0.0",
       "dependsOn": [
-        "officeparser@7.7.0|bmp-js@0.1.0",
-        "officeparser@7.7.0|idb-keyval@6.2.2",
-        "officeparser@7.7.0|is-url@1.2.4",
-        "officeparser@7.7.0|node-fetch@2.7.0",
-        "officeparser@7.7.0|opencollective-postinstall@2.0.3",
-        "officeparser@7.7.0|regenerator-runtime@0.13.11",
-        "officeparser@7.7.0|tesseract.js-core@7.0.0",
-        "officeparser@7.7.0|wasm-feature-detect@1.8.0",
-        "officeparser@7.7.0|zlibjs@0.3.1"
+        "officeparser@7.8.0|bmp-js@0.1.0",
+        "officeparser@7.8.0|idb-keyval@6.2.2",
+        "officeparser@7.8.0|is-url@1.2.4",
+        "officeparser@7.8.0|node-fetch@2.7.0",
+        "officeparser@7.8.0|opencollective-postinstall@2.0.3",
+        "officeparser@7.8.0|regenerator-runtime@0.13.11",
+        "officeparser@7.8.0|tesseract.js-core@7.0.0",
+        "officeparser@7.8.0|wasm-feature-detect@1.8.0",
+        "officeparser@7.8.0|zlibjs@0.3.1"
       ]
     },
     {
-      "ref": "officeparser@7.7.0|@tokenizer/inflate@0.4.1",
+      "ref": "officeparser@7.8.0|@tokenizer/inflate@0.4.1",
       "dependsOn": [
-        "officeparser@7.7.0|debug@4.4.3",
-        "officeparser@7.7.0|token-types@6.1.2"
+        "officeparser@7.8.0|debug@4.4.3",
+        "officeparser@7.8.0|token-types@6.1.2"
       ]
     },
     {
-      "ref": "officeparser@7.7.0|strtok3@10.3.5",
+      "ref": "officeparser@7.8.0|strtok3@10.3.5",
       "dependsOn": [
-        "officeparser@7.7.0|@tokenizer/token@0.3.0"
+        "officeparser@7.8.0|@tokenizer/token@0.3.0"
       ]
     },
     {
-      "ref": "officeparser@7.7.0|token-types@6.1.2",
+      "ref": "officeparser@7.8.0|token-types@6.1.2",
       "dependsOn": [
-        "officeparser@7.7.0|@borewit/text-codec@0.2.2",
-        "officeparser@7.7.0|@tokenizer/token@0.3.0",
-        "officeparser@7.7.0|ieee754@1.2.1"
+        "officeparser@7.8.0|@borewit/text-codec@0.2.2",
+        "officeparser@7.8.0|@tokenizer/token@0.3.0",
+        "officeparser@7.8.0|ieee754@1.2.1"
       ]
     },
     {
-      "ref": "officeparser@7.7.0|uint8array-extras@1.5.0"
+      "ref": "officeparser@7.8.0|uint8array-extras@1.5.0"
     },
     {
-      "ref": "officeparser@7.7.0|@napi-rs/canvas@1.0.1",
+      "ref": "officeparser@7.8.0|@napi-rs/canvas@1.0.1",
       "dependsOn": [
-        "officeparser@7.7.0|@napi-rs/canvas-linux-x64-gnu@1.0.1"
+        "officeparser@7.8.0|@napi-rs/canvas-linux-x64-gnu@1.0.1"
       ]
     },
     {
-      "ref": "officeparser@7.7.0|bmp-js@0.1.0"
+      "ref": "officeparser@7.8.0|bmp-js@0.1.0"
     },
     {
-      "ref": "officeparser@7.7.0|idb-keyval@6.2.2"
+      "ref": "officeparser@7.8.0|idb-keyval@6.2.2"
     },
     {
-      "ref": "officeparser@7.7.0|is-url@1.2.4"
+      "ref": "officeparser@7.8.0|is-url@1.2.4"
     },
     {
-      "ref": "officeparser@7.7.0|node-fetch@2.7.0",
+      "ref": "officeparser@7.8.0|node-fetch@2.7.0",
       "dependsOn": [
-        "officeparser@7.7.0|whatwg-url@5.0.0"
+        "officeparser@7.8.0|whatwg-url@5.0.0"
       ]
     },
     {
-      "ref": "officeparser@7.7.0|opencollective-postinstall@2.0.3"
+      "ref": "officeparser@7.8.0|opencollective-postinstall@2.0.3"
     },
     {
-      "ref": "officeparser@7.7.0|regenerator-runtime@0.13.11"
+      "ref": "officeparser@7.8.0|regenerator-runtime@0.13.11"
     },
     {
-      "ref": "officeparser@7.7.0|tesseract.js-core@7.0.0"
+      "ref": "officeparser@7.8.0|tesseract.js-core@7.0.0"
     },
     {
-      "ref": "officeparser@7.7.0|wasm-feature-detect@1.8.0"
+      "ref": "officeparser@7.8.0|wasm-feature-detect@1.8.0"
     },
     {
-      "ref": "officeparser@7.7.0|zlibjs@0.3.1"
+      "ref": "officeparser@7.8.0|zlibjs@0.3.1"
     },
     {
-      "ref": "officeparser@7.7.0|debug@4.4.3",
+      "ref": "officeparser@7.8.0|debug@4.4.3",
       "dependsOn": [
-        "officeparser@7.7.0|ms@2.1.3"
+        "officeparser@7.8.0|ms@2.1.3"
       ]
     },
     {
-      "ref": "officeparser@7.7.0|@tokenizer/token@0.3.0"
+      "ref": "officeparser@7.8.0|@tokenizer/token@0.3.0"
     },
     {
-      "ref": "officeparser@7.7.0|@borewit/text-codec@0.2.2"
+      "ref": "officeparser@7.8.0|@borewit/text-codec@0.2.2"
     },
     {
-      "ref": "officeparser@7.7.0|ieee754@1.2.1"
+      "ref": "officeparser@7.8.0|ieee754@1.2.1"
     },
     {
-      "ref": "officeparser@7.7.0|@napi-rs/canvas-linux-x64-gnu@1.0.1"
+      "ref": "officeparser@7.8.0|@napi-rs/canvas-linux-x64-gnu@1.0.1"
     },
     {
-      "ref": "officeparser@7.7.0|whatwg-url@5.0.0",
+      "ref": "officeparser@7.8.0|whatwg-url@5.0.0",
       "dependsOn": [
-        "officeparser@7.7.0|tr46@0.0.3",
-        "officeparser@7.7.0|webidl-conversions@3.0.1"
+        "officeparser@7.8.0|tr46@0.0.3",
+        "officeparser@7.8.0|webidl-conversions@3.0.1"
       ]
     },
     {
-      "ref": "officeparser@7.7.0|ms@2.1.3"
+      "ref": "officeparser@7.8.0|ms@2.1.3"
     },
     {
-      "ref": "officeparser@7.7.0|tr46@0.0.3"
+      "ref": "officeparser@7.8.0|tr46@0.0.3"
     },
     {
-      "ref": "officeparser@7.7.0|webidl-conversions@3.0.1"
+      "ref": "officeparser@7.8.0|webidl-conversions@3.0.1"
     }
   ]
 }

package/dist/types.d.ts

@@ -413,6 +413,18 @@ export interface HtmlParserConfig {
      * Defaults to false.
      */
     preserveIframes?: boolean | string[];
+    /**
+     * Import ambiguous "folk" embed forms in Markdown as embeds: a standalone Obsidian-style image
+     * whose URL is a YouTube link (`![](https://youtube.com/watch?v=ID)`), and the clickable
+     * thumbnail-link (`[![alt](https://img.youtube.com/vi/ID/…)](watch-url)`). Both become a
+     * `embedType: 'youtube'` embed (rendered from the validated id, so it is safe). Off by default:
+     * auto-upgrading an image/link to an embed is a heuristic that could mangle a genuinely-intended
+     * image link, so a consumer opts in. The unambiguous forms (`<div data-youtube-video>`, a bare
+     * YouTube `<iframe>`, the `::youtube` directive) are always recognized, independent of this flag.
+     *
+     * Defaults to false.
+     */
+    embedFolkForms?: boolean;
 }
 /**
  * Maps an input format string to its corresponding format-specific parser configuration, mirroring
@@ -909,6 +921,15 @@ export interface HtmlGeneratorConfig {
      * `HtmlParser` reads every shape this emits, so output stays self-round-trippable.
      */
     sourceAttributes?: boolean;
+    /**
+     * Emit a generic (non-YouTube) iframe embed as a gated placeholder,
+     * `<div data-embed-gated data-embed-src="…" …>`, instead of a live `<iframe>`. The gated shape
+     * never auto-loads its src: an editor renders a click-to-load placeholder from it, and
+     * `HtmlParser` reads it back to the same `embed` node. The src is scheme-checked (`sanitizeUrl`)
+     * on emit. Off by default; the default output (a live `<iframe>`) is unchanged. YouTube embeds
+     * are unaffected (they already render from a validated id).
+     */
+    gatedEmbeds?: boolean;
 }
 /**
  * Configuration options for PDF generation.
@@ -1082,6 +1103,20 @@ export type CitationSyntax = 'at' | 'none';
 export type WikilinkSyntax = 'double-bracket' | 'none';
 /** `{width=50%}` attribute lists (`'brace'`), or `'none'`. */
 export type AttributeListSyntax = 'brace' | 'none';
+/**
+ * How an `embed` node is written to Markdown:
+ * - `'html'` (default): the single-line `<div data-youtube-video="ID">` / `<iframe src=...>` block
+ *   this library has always emitted. Round-trips through officeParser, but renders as an invisible
+ *   empty box on GitHub.
+ * - `'directive'`: a remark-directive leaf, `::youtube[Label]{id=... width=... align=...}` /
+ *   `::embed[Label]{src=... width=... height=... align=...}`. Round-trips within an editor that
+ *   understands it; renders verbatim (not just the label) on GitHub, so it is an editor format, not
+ *   a GitHub-interop one.
+ * - `'link'`: a plain `[YouTube](url)` / `[Embed](url)`.
+ * - `'thumbnail'`: a YouTube-only clickable thumbnail `[![Label](.../vi/ID/hqdefault.jpg)](watch)`,
+ *   the best GitHub degrade; a non-YouTube embed falls back to `'link'`.
+ */
+export type EmbedSyntax = 'html' | 'directive' | 'link' | 'thumbnail';
 /**
  * @deprecated Legacy flavor names for `MarkdownDialectConfig.admonitions`. Use the syntax names
  * instead: `'github'` -> `'blockquote'`, `'gitlab'` -> `'fence'`, `'pandoc'` -> `'fence-attribute'`.
@@ -1168,6 +1203,13 @@ export interface MarkdownDialectConfig {
     /** Table syntax: native GFM pipe tables, or forced HTML `<table>` (required for strict
      *  CommonMark, which has no table syntax of its own). */
     tables?: 'native' | 'html';
+    /**
+     * How an `embed` node is written to Markdown (`'html'` | `'directive'` | `'link'` |
+     * `'thumbnail'`; see `EmbedSyntax`). This is the authority for embed form. When omitted, the
+     * deprecated `fallbackToHtml.embeds` boolean is honored (`true`/unset maps to `'html'`, `false`
+     * to `'link'`), then the default `'html'`.
+     */
+    embeds?: EmbedSyntax;
 }
 /**
  * Granular control over when the Markdown generator falls back to raw HTML tags for features
@@ -1184,7 +1226,12 @@ export interface FallbackToHtmlConfig {
     anchors?: boolean;
     /** Nested-table and merged-cell (colspan/rowspan) HTML `<table>` fallback. */
     tables?: boolean;
-    /** YouTube embed `<div data-youtube-video>` vs. a plain link. */
+    /**
+     * YouTube embed `<div data-youtube-video>` vs. a plain link.
+     * @deprecated Use `mdConfig.dialect.embeds` (`EmbedSyntax`) instead, which also selects the
+     * `'directive'` and `'thumbnail'` forms. When `dialect.embeds` is unset this boolean is still
+     * honored (`true` maps to `'html'`, `false` to `'link'`); it will be removed in the next major.
+     */
     embeds?: boolean;
     /** Multi-line table cell content joined with `<br>` instead of a space. */
     cellLineBreaks?: boolean;
@@ -1804,10 +1851,13 @@ export interface EmbedMetadata {
     url?: string;
     /** Display width, as a CSS length or percentage. */
     width?: string;
-    /** Display height, as a CSS length or percentage (generic iframes). */
+    /** Display height, as a CSS length or percentage. */
     height?: string;
     /** Layout alignment of the embed. */
     align?: 'left' | 'center' | 'right';
+    /** Human-readable label for the embed (e.g. the `[Label]` of a `::youtube[Label]{...}` leaf
+     *  directive, or a gated embed's caption). Purely descriptive; never a trust or render input. */
+    label?: string;
 }
 /**
  * Metadata for an admonition/alert node (e.g. GitHub's `> [!NOTE]` or GLFM's `:::note`).

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "officeparser",
-  "version": "7.7.0",
+  "version": "7.8.0",
   "description": "A robust, strictly-typed Node.js and Browser library for parsing office files (.docx, .pptx, .xlsx, .odt, .odp, .ods, .pdf, .rtf, .csv, .md, .html, .epub) and generating high-fidelity outputs in Markdown, HTML, CSV, RTF, PDF, EPUB, and RAG-focused chunks.",
   "funding": "https://github.com/sponsors/harshankur",
   "main": "dist/index.js",