officeparser 7.5.1 → 7.6.2

package/dist/sbom.cdx.json

@@ -3,21 +3,21 @@
   "bomFormat": "CycloneDX",
   "specVersion": "1.6",
   "version": 1,
-  "serialNumber": "urn:uuid:30a1956c-acfa-497a-bf1c-b6d265b11bc0",
+  "serialNumber": "urn:uuid:a62832bf-64dc-4267-ae37-c4ec9958e934",
   "metadata": {
-    "timestamp": "2026-07-30T21:00:33.153Z",
+    "timestamp": "2026-08-15T20:48:28.765Z",
     "tools": {
       "components": [
         {
           "type": "application",
           "name": "npm",
-          "version": "11.16.0"
+          "version": "11.17.0"
         },
         {
           "type": "application",
           "name": "cyclonedx-npm",
           "group": "@cyclonedx",
-          "version": "6.0.0",
+          "version": "6.0.1",
           "author": "Jan Kowalleck",
           "description": "Create CycloneDX Software Bill of Materials (SBOM) from NPM projects.",
           "licenses": [
@@ -49,7 +49,7 @@
           "type": "library",
           "name": "cyclonedx-library",
           "group": "@cyclonedx",
-          "version": "10.1.0",
+          "version": "10.1.1",
           "author": "Jan Kowalleck",
           "description": "Core functionality of CycloneDX for JavaScript (Node.js or WebBrowser).",
           "licenses": [
@@ -82,8 +82,8 @@
     "component": {
       "type": "application",
       "name": "officeparser",
-      "version": "7.5.1",
-      "bom-ref": "officeparser@7.5.1",
+      "version": "7.6.2",
+      "bom-ref": "officeparser@7.6.2",
       "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": [
@@ -94,7 +94,7 @@
           }
         }
       ],
-      "purl": "pkg:npm/officeparser@7.5.1?vcs_url=git%2Bhttps%3A%2F%2Fgithub.com%2Fharshankur%2FofficeParser.git",
+      "purl": "pkg:npm/officeparser@7.6.2?vcs_url=git%2Bhttps%3A%2F%2Fgithub.com%2Fharshankur%2FofficeParser.git",
       "externalReferences": [
         {
           "url": "git+https://github.com/harshankur/officeParser.git",
@@ -126,7 +126,7 @@
       "name": "xmldom",
       "group": "@xmldom",
       "version": "0.9.10",
-      "bom-ref": "officeparser@7.5.1|@xmldom/xmldom@0.9.10",
+      "bom-ref": "officeparser@7.6.2|@xmldom/xmldom@0.9.10",
       "description": "A pure JavaScript W3C standard-based (XML DOM Level 2 Core) DOMParser and XMLSerializer module.",
       "licenses": [
         {
@@ -176,7 +176,7 @@
       "type": "library",
       "name": "fflate",
       "version": "0.8.3",
-      "bom-ref": "officeparser@7.5.1|fflate@0.8.3",
+      "bom-ref": "officeparser@7.6.2|fflate@0.8.3",
       "author": "Arjun Barrett",
       "description": "High performance (de)compression in an 8kB package",
       "licenses": [
@@ -227,7 +227,7 @@
       "type": "library",
       "name": "file-type",
       "version": "22.0.1",
-      "bom-ref": "officeparser@7.5.1|file-type@22.0.1",
+      "bom-ref": "officeparser@7.6.2|file-type@22.0.1",
       "author": "Sindre Sorhus",
       "description": "Detect the file type of a file, stream, or data",
       "licenses": [
@@ -278,7 +278,7 @@
       "type": "library",
       "name": "pdfjs-dist",
       "version": "6.1.200",
-      "bom-ref": "officeparser@7.5.1|pdfjs-dist@6.1.200",
+      "bom-ref": "officeparser@7.6.2|pdfjs-dist@6.1.200",
       "description": "Generic build of Mozilla's PDF.js library.",
       "licenses": [
         {
@@ -328,7 +328,7 @@
       "type": "library",
       "name": "tesseract.js",
       "version": "7.0.0",
-      "bom-ref": "officeparser@7.5.1|tesseract.js@7.0.0",
+      "bom-ref": "officeparser@7.6.2|tesseract.js@7.0.0",
       "description": "Pure Javascript Multilingual OCR",
       "licenses": [
         {
@@ -379,7 +379,7 @@
       "name": "inflate",
       "group": "@tokenizer",
       "version": "0.4.1",
-      "bom-ref": "officeparser@7.5.1|@tokenizer/inflate@0.4.1",
+      "bom-ref": "officeparser@7.6.2|@tokenizer/inflate@0.4.1",
       "author": "Borewit",
       "description": "Tokenized zip support",
       "licenses": [
@@ -430,7 +430,7 @@
       "type": "library",
       "name": "strtok3",
       "version": "10.3.5",
-      "bom-ref": "officeparser@7.5.1|strtok3@10.3.5",
+      "bom-ref": "officeparser@7.6.2|strtok3@10.3.5",
       "author": "Borewit",
       "description": "A promise based streaming tokenizer",
       "licenses": [
@@ -481,7 +481,7 @@
       "type": "library",
       "name": "token-types",
       "version": "6.1.2",
-      "bom-ref": "officeparser@7.5.1|token-types@6.1.2",
+      "bom-ref": "officeparser@7.6.2|token-types@6.1.2",
       "author": "Borewit",
       "description": "Common token types for decoding and encoding numeric and string values",
       "licenses": [
@@ -532,7 +532,7 @@
       "type": "library",
       "name": "uint8array-extras",
       "version": "1.5.0",
-      "bom-ref": "officeparser@7.5.1|uint8array-extras@1.5.0",
+      "bom-ref": "officeparser@7.6.2|uint8array-extras@1.5.0",
       "author": "Sindre Sorhus",
       "description": "Useful utilities for working with Uint8Array (and Buffer)",
       "licenses": [
@@ -584,7 +584,7 @@
       "name": "canvas",
       "group": "@napi-rs",
       "version": "1.0.1",
-      "bom-ref": "officeparser@7.5.1|@napi-rs/canvas@1.0.1",
+      "bom-ref": "officeparser@7.6.2|@napi-rs/canvas@1.0.1",
       "description": "Canvas for Node.js with skia backend",
       "scope": "optional",
       "licenses": [
@@ -635,7 +635,7 @@
       "type": "library",
       "name": "bmp-js",
       "version": "0.1.0",
-      "bom-ref": "officeparser@7.5.1|bmp-js@0.1.0",
+      "bom-ref": "officeparser@7.6.2|bmp-js@0.1.0",
       "author": "shaozilee",
       "description": "A pure javascript BMP encoder and decoder",
       "licenses": [
@@ -686,7 +686,7 @@
       "type": "library",
       "name": "idb-keyval",
       "version": "6.2.2",
-      "bom-ref": "officeparser@7.5.1|idb-keyval@6.2.2",
+      "bom-ref": "officeparser@7.6.2|idb-keyval@6.2.2",
       "author": "Jake Archibald",
       "description": "A super-simple-small keyval store built on top of IndexedDB",
       "licenses": [
@@ -737,7 +737,7 @@
       "type": "library",
       "name": "is-url",
       "version": "1.2.4",
-      "bom-ref": "officeparser@7.5.1|is-url@1.2.4",
+      "bom-ref": "officeparser@7.6.2|is-url@1.2.4",
       "description": "Check whether a string is a URL.",
       "licenses": [
         {
@@ -787,7 +787,7 @@
       "type": "library",
       "name": "node-fetch",
       "version": "2.7.0",
-      "bom-ref": "officeparser@7.5.1|node-fetch@2.7.0",
+      "bom-ref": "officeparser@7.6.2|node-fetch@2.7.0",
       "author": "David Frank",
       "description": "A light-weight module that brings window.fetch to node.js",
       "licenses": [
@@ -838,7 +838,7 @@
       "type": "library",
       "name": "opencollective-postinstall",
       "version": "2.0.3",
-      "bom-ref": "officeparser@7.5.1|opencollective-postinstall@2.0.3",
+      "bom-ref": "officeparser@7.6.2|opencollective-postinstall@2.0.3",
       "author": "Xavier Damman",
       "description": "Lightweight npm postinstall message to invite people to donate to your collective",
       "licenses": [
@@ -889,7 +889,7 @@
       "type": "library",
       "name": "regenerator-runtime",
       "version": "0.13.11",
-      "bom-ref": "officeparser@7.5.1|regenerator-runtime@0.13.11",
+      "bom-ref": "officeparser@7.6.2|regenerator-runtime@0.13.11",
       "author": "Ben Newman",
       "description": "Runtime for Regenerator-compiled generator and async functions.",
       "licenses": [
@@ -940,7 +940,7 @@
       "type": "library",
       "name": "tesseract.js-core",
       "version": "7.0.0",
-      "bom-ref": "officeparser@7.5.1|tesseract.js-core@7.0.0",
+      "bom-ref": "officeparser@7.6.2|tesseract.js-core@7.0.0",
       "author": "antimatter15",
       "description": "Tesseract C++ API in Pure Javascript",
       "licenses": [
@@ -991,7 +991,7 @@
       "type": "library",
       "name": "wasm-feature-detect",
       "version": "1.8.0",
-      "bom-ref": "officeparser@7.5.1|wasm-feature-detect@1.8.0",
+      "bom-ref": "officeparser@7.6.2|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": [
@@ -1042,7 +1042,7 @@
       "type": "library",
       "name": "zlibjs",
       "version": "0.3.1",
-      "bom-ref": "officeparser@7.5.1|zlibjs@0.3.1",
+      "bom-ref": "officeparser@7.6.2|zlibjs@0.3.1",
       "author": "Yuta Imaya",
       "description": "zlib, gzip and zip implementation in JavaScript",
       "licenses": [
@@ -1093,7 +1093,7 @@
       "type": "library",
       "name": "debug",
       "version": "4.4.3",
-      "bom-ref": "officeparser@7.5.1|debug@4.4.3",
+      "bom-ref": "officeparser@7.6.2|debug@4.4.3",
       "author": "Josh Junon",
       "description": "Lightweight debugging utility for Node.js and the browser",
       "licenses": [
@@ -1145,7 +1145,7 @@
       "name": "token",
       "group": "@tokenizer",
       "version": "0.3.0",
-      "bom-ref": "officeparser@7.5.1|@tokenizer/token@0.3.0",
+      "bom-ref": "officeparser@7.6.2|@tokenizer/token@0.3.0",
       "author": "Borewit",
       "description": "TypeScript definition for strtok3 token",
       "licenses": [
@@ -1197,7 +1197,7 @@
       "name": "text-codec",
       "group": "@borewit",
       "version": "0.2.2",
-      "bom-ref": "officeparser@7.5.1|@borewit/text-codec@0.2.2",
+      "bom-ref": "officeparser@7.6.2|@borewit/text-codec@0.2.2",
       "author": "Borewit",
       "description": "Text Decoder",
       "licenses": [
@@ -1248,7 +1248,7 @@
       "type": "library",
       "name": "ieee754",
       "version": "1.2.1",
-      "bom-ref": "officeparser@7.5.1|ieee754@1.2.1",
+      "bom-ref": "officeparser@7.6.2|ieee754@1.2.1",
       "author": "Feross Aboukhadijeh",
       "description": "Read/write IEEE754 floating point numbers from/to a Buffer or array-like object",
       "licenses": [
@@ -1300,7 +1300,7 @@
       "name": "canvas-linux-x64-gnu",
       "group": "@napi-rs",
       "version": "1.0.1",
-      "bom-ref": "officeparser@7.5.1|@napi-rs/canvas-linux-x64-gnu@1.0.1",
+      "bom-ref": "officeparser@7.6.2|@napi-rs/canvas-linux-x64-gnu@1.0.1",
       "description": "Canvas for Node.js with skia backend",
       "scope": "optional",
       "licenses": [
@@ -1351,7 +1351,7 @@
       "type": "library",
       "name": "whatwg-url",
       "version": "5.0.0",
-      "bom-ref": "officeparser@7.5.1|whatwg-url@5.0.0",
+      "bom-ref": "officeparser@7.6.2|whatwg-url@5.0.0",
       "author": "Sebastian Mayr",
       "description": "An implementation of the WHATWG URL Standard's URL API and parsing machinery",
       "licenses": [
@@ -1402,7 +1402,7 @@
       "type": "library",
       "name": "ms",
       "version": "2.1.3",
-      "bom-ref": "officeparser@7.5.1|ms@2.1.3",
+      "bom-ref": "officeparser@7.6.2|ms@2.1.3",
       "description": "Tiny millisecond conversion utility",
       "licenses": [
         {
@@ -1452,7 +1452,7 @@
       "type": "library",
       "name": "tr46",
       "version": "0.0.3",
-      "bom-ref": "officeparser@7.5.1|tr46@0.0.3",
+      "bom-ref": "officeparser@7.6.2|tr46@0.0.3",
       "author": "Sebastian Mayr",
       "description": "An implementation of the Unicode TR46 spec",
       "licenses": [
@@ -1503,7 +1503,7 @@
       "type": "library",
       "name": "webidl-conversions",
       "version": "3.0.1",
-      "bom-ref": "officeparser@7.5.1|webidl-conversions@3.0.1",
+      "bom-ref": "officeparser@7.6.2|webidl-conversions@3.0.1",
       "author": "Domenic Denicola",
       "description": "Implements the WebIDL algorithms for converting to and from JavaScript values",
       "licenses": [
@@ -1553,143 +1553,143 @@
   ],
   "dependencies": [
     {
-      "ref": "officeparser@7.5.1",
+      "ref": "officeparser@7.6.2",
       "dependsOn": [
-        "officeparser@7.5.1|@xmldom/xmldom@0.9.10",
-        "officeparser@7.5.1|fflate@0.8.3",
-        "officeparser@7.5.1|file-type@22.0.1",
-        "officeparser@7.5.1|pdfjs-dist@6.1.200",
-        "officeparser@7.5.1|tesseract.js@7.0.0"
+        "officeparser@7.6.2|@xmldom/xmldom@0.9.10",
+        "officeparser@7.6.2|fflate@0.8.3",
+        "officeparser@7.6.2|file-type@22.0.1",
+        "officeparser@7.6.2|pdfjs-dist@6.1.200",
+        "officeparser@7.6.2|tesseract.js@7.0.0"
       ]
     },
     {
-      "ref": "officeparser@7.5.1|@xmldom/xmldom@0.9.10"
+      "ref": "officeparser@7.6.2|@xmldom/xmldom@0.9.10"
     },
     {
-      "ref": "officeparser@7.5.1|fflate@0.8.3"
+      "ref": "officeparser@7.6.2|fflate@0.8.3"
     },
     {
-      "ref": "officeparser@7.5.1|file-type@22.0.1",
+      "ref": "officeparser@7.6.2|file-type@22.0.1",
       "dependsOn": [
-        "officeparser@7.5.1|@tokenizer/inflate@0.4.1",
-        "officeparser@7.5.1|strtok3@10.3.5",
-        "officeparser@7.5.1|token-types@6.1.2",
-        "officeparser@7.5.1|uint8array-extras@1.5.0"
+        "officeparser@7.6.2|@tokenizer/inflate@0.4.1",
+        "officeparser@7.6.2|strtok3@10.3.5",
+        "officeparser@7.6.2|token-types@6.1.2",
+        "officeparser@7.6.2|uint8array-extras@1.5.0"
       ]
     },
     {
-      "ref": "officeparser@7.5.1|pdfjs-dist@6.1.200",
+      "ref": "officeparser@7.6.2|pdfjs-dist@6.1.200",
       "dependsOn": [
-        "officeparser@7.5.1|@napi-rs/canvas@1.0.1"
+        "officeparser@7.6.2|@napi-rs/canvas@1.0.1"
       ]
     },
     {
-      "ref": "officeparser@7.5.1|tesseract.js@7.0.0",
+      "ref": "officeparser@7.6.2|tesseract.js@7.0.0",
       "dependsOn": [
-        "officeparser@7.5.1|bmp-js@0.1.0",
-        "officeparser@7.5.1|idb-keyval@6.2.2",
-        "officeparser@7.5.1|is-url@1.2.4",
-        "officeparser@7.5.1|node-fetch@2.7.0",
-        "officeparser@7.5.1|opencollective-postinstall@2.0.3",
-        "officeparser@7.5.1|regenerator-runtime@0.13.11",
-        "officeparser@7.5.1|tesseract.js-core@7.0.0",
-        "officeparser@7.5.1|wasm-feature-detect@1.8.0",
-        "officeparser@7.5.1|zlibjs@0.3.1"
+        "officeparser@7.6.2|bmp-js@0.1.0",
+        "officeparser@7.6.2|idb-keyval@6.2.2",
+        "officeparser@7.6.2|is-url@1.2.4",
+        "officeparser@7.6.2|node-fetch@2.7.0",
+        "officeparser@7.6.2|opencollective-postinstall@2.0.3",
+        "officeparser@7.6.2|regenerator-runtime@0.13.11",
+        "officeparser@7.6.2|tesseract.js-core@7.0.0",
+        "officeparser@7.6.2|wasm-feature-detect@1.8.0",
+        "officeparser@7.6.2|zlibjs@0.3.1"
       ]
     },
     {
-      "ref": "officeparser@7.5.1|@tokenizer/inflate@0.4.1",
+      "ref": "officeparser@7.6.2|@tokenizer/inflate@0.4.1",
       "dependsOn": [
-        "officeparser@7.5.1|debug@4.4.3",
-        "officeparser@7.5.1|token-types@6.1.2"
+        "officeparser@7.6.2|debug@4.4.3",
+        "officeparser@7.6.2|token-types@6.1.2"
       ]
     },
     {
-      "ref": "officeparser@7.5.1|strtok3@10.3.5",
+      "ref": "officeparser@7.6.2|strtok3@10.3.5",
       "dependsOn": [
-        "officeparser@7.5.1|@tokenizer/token@0.3.0"
+        "officeparser@7.6.2|@tokenizer/token@0.3.0"
       ]
     },
     {
-      "ref": "officeparser@7.5.1|token-types@6.1.2",
+      "ref": "officeparser@7.6.2|token-types@6.1.2",
       "dependsOn": [
-        "officeparser@7.5.1|@borewit/text-codec@0.2.2",
-        "officeparser@7.5.1|@tokenizer/token@0.3.0",
-        "officeparser@7.5.1|ieee754@1.2.1"
+        "officeparser@7.6.2|@borewit/text-codec@0.2.2",
+        "officeparser@7.6.2|@tokenizer/token@0.3.0",
+        "officeparser@7.6.2|ieee754@1.2.1"
       ]
     },
     {
-      "ref": "officeparser@7.5.1|uint8array-extras@1.5.0"
+      "ref": "officeparser@7.6.2|uint8array-extras@1.5.0"
     },
     {
-      "ref": "officeparser@7.5.1|@napi-rs/canvas@1.0.1",
+      "ref": "officeparser@7.6.2|@napi-rs/canvas@1.0.1",
       "dependsOn": [
-        "officeparser@7.5.1|@napi-rs/canvas-linux-x64-gnu@1.0.1"
+        "officeparser@7.6.2|@napi-rs/canvas-linux-x64-gnu@1.0.1"
       ]
     },
     {
-      "ref": "officeparser@7.5.1|bmp-js@0.1.0"
+      "ref": "officeparser@7.6.2|bmp-js@0.1.0"
     },
     {
-      "ref": "officeparser@7.5.1|idb-keyval@6.2.2"
+      "ref": "officeparser@7.6.2|idb-keyval@6.2.2"
     },
     {
-      "ref": "officeparser@7.5.1|is-url@1.2.4"
+      "ref": "officeparser@7.6.2|is-url@1.2.4"
     },
     {
-      "ref": "officeparser@7.5.1|node-fetch@2.7.0",
+      "ref": "officeparser@7.6.2|node-fetch@2.7.0",
       "dependsOn": [
-        "officeparser@7.5.1|whatwg-url@5.0.0"
+        "officeparser@7.6.2|whatwg-url@5.0.0"
       ]
     },
     {
-      "ref": "officeparser@7.5.1|opencollective-postinstall@2.0.3"
+      "ref": "officeparser@7.6.2|opencollective-postinstall@2.0.3"
     },
     {
-      "ref": "officeparser@7.5.1|regenerator-runtime@0.13.11"
+      "ref": "officeparser@7.6.2|regenerator-runtime@0.13.11"
     },
     {
-      "ref": "officeparser@7.5.1|tesseract.js-core@7.0.0"
+      "ref": "officeparser@7.6.2|tesseract.js-core@7.0.0"
     },
     {
-      "ref": "officeparser@7.5.1|wasm-feature-detect@1.8.0"
+      "ref": "officeparser@7.6.2|wasm-feature-detect@1.8.0"
     },
     {
-      "ref": "officeparser@7.5.1|zlibjs@0.3.1"
+      "ref": "officeparser@7.6.2|zlibjs@0.3.1"
     },
     {
-      "ref": "officeparser@7.5.1|debug@4.4.3",
+      "ref": "officeparser@7.6.2|debug@4.4.3",
       "dependsOn": [
-        "officeparser@7.5.1|ms@2.1.3"
+        "officeparser@7.6.2|ms@2.1.3"
       ]
     },
     {
-      "ref": "officeparser@7.5.1|@tokenizer/token@0.3.0"
+      "ref": "officeparser@7.6.2|@tokenizer/token@0.3.0"
     },
     {
-      "ref": "officeparser@7.5.1|@borewit/text-codec@0.2.2"
+      "ref": "officeparser@7.6.2|@borewit/text-codec@0.2.2"
     },
     {
-      "ref": "officeparser@7.5.1|ieee754@1.2.1"
+      "ref": "officeparser@7.6.2|ieee754@1.2.1"
     },
     {
-      "ref": "officeparser@7.5.1|@napi-rs/canvas-linux-x64-gnu@1.0.1"
+      "ref": "officeparser@7.6.2|@napi-rs/canvas-linux-x64-gnu@1.0.1"
     },
     {
-      "ref": "officeparser@7.5.1|whatwg-url@5.0.0",
+      "ref": "officeparser@7.6.2|whatwg-url@5.0.0",
       "dependsOn": [
-        "officeparser@7.5.1|tr46@0.0.3",
-        "officeparser@7.5.1|webidl-conversions@3.0.1"
+        "officeparser@7.6.2|tr46@0.0.3",
+        "officeparser@7.6.2|webidl-conversions@3.0.1"
       ]
     },
     {
-      "ref": "officeparser@7.5.1|ms@2.1.3"
+      "ref": "officeparser@7.6.2|ms@2.1.3"
     },
     {
-      "ref": "officeparser@7.5.1|tr46@0.0.3"
+      "ref": "officeparser@7.6.2|tr46@0.0.3"
     },
     {
-      "ref": "officeparser@7.5.1|webidl-conversions@3.0.1"
+      "ref": "officeparser@7.6.2|webidl-conversions@3.0.1"
     }
   ]
 }

package/dist/types.d.ts

@@ -401,6 +401,18 @@ export interface HtmlParserConfig {
      * Defaults to false.
      */
     preserveAttributes?: boolean;
+    /**
+     * Preserve `<iframe>` embeds that aren't recognized as a known provider (YouTube is always
+     * recognized). By default every non-YouTube iframe is dropped, which is a deliberate security
+     * posture other consumers rely on; set this to opt back in. `true` preserves any iframe; an
+     * array is a hostname allowlist (an entry matches the src's host exactly or as a `.`-suffix,
+     * so `"vimeo.com"` also matches `player.vimeo.com`). Preserved iframes become `embed` nodes
+     * with `embedType: 'iframe'`; on generation the `src` is still scheme-checked (only http/https
+     * survive). This also governs a raw `<iframe>` block encountered in Markdown input.
+     *
+     * Defaults to false.
+     */
+    preserveIframes?: boolean | string[];
 }
 /**
  * Maps an input format string to its corresponding format-specific parser configuration, mirroring
@@ -885,6 +897,18 @@ export interface HtmlGeneratorConfig {
      * Granular injection points for custom HTML, scripts, and styles.
      */
     injections?: HtmlInjectionConfig;
+    /**
+     * Carry each rich node's raw source in a `data-*` attribute, with undelimited text content,
+     * so attribute-driven structured consumers (rich-text editors, custom viewers) can rehydrate
+     * the node from the markup rather than re-parsing the display text. Affects wikilinks
+     * (adds `data-wikilink`/`data-target`/`data-alias`), citations (a `<span class="citation">`
+     * carrying `data-key` instead of `<cite>`), math (the LaTeX in `data-math`, undelimited) and
+     * mermaid (a `<div class="mermaid" data-mermaid>` instead of `<pre><code>`).
+     *
+     * Off by default; the default output is byte-identical to previous releases. The widened
+     * `HtmlParser` reads every shape this emits, so output stays self-round-trippable.
+     */
+    sourceAttributes?: boolean;
 }
 /**
  * Configuration options for PDF generation.
@@ -1100,6 +1124,14 @@ export interface FallbackToHtmlConfig {
     embeds?: boolean;
     /** Multi-line table cell content joined with `<br>` instead of a space. */
     cellLineBreaks?: boolean;
+    /**
+     * Inline text color, highlight, and font size via a `<span style="color:...;background-color:...;
+     * font-size:...">` run, which the Markdown parser reads back. These have no Markdown syntax and
+     * are silently lost otherwise. Unlike the other fields this is **off by default even when
+     * `fallbackToHtml` is `true`**, because it changes default output; enable it explicitly with
+     * `fallbackToHtml: { inlineFormatting: true }`.
+     */
+    inlineFormatting?: boolean;
 }
 /**
  * Configuration options for Markdown generation.
@@ -1371,6 +1403,16 @@ export interface OfficeChunk {
  * Supported file types for parsing.
  */
 export type SupportedFileType = 'docx' | 'pptx' | 'xlsx' | 'odt' | 'odp' | 'ods' | 'pdf' | 'rtf' | 'md' | 'html' | 'csv' | 'epub';
+/**
+ * A structural stand-in for the web `Blob`/`File` so `parseOffice`/`convert` accept them in the
+ * browser without pulling the DOM lib into this package's types. Any object with an
+ * `arrayBuffer()` method qualifies. When `name` is present (as on a `File`) it is used only for
+ * extension-based type detection, never as a filesystem path.
+ */
+export interface BlobLike {
+    arrayBuffer(): Promise<ArrayBuffer>;
+    name?: string;
+}
 /**
  * Types of content nodes in the AST.
  */
@@ -1612,7 +1654,7 @@ export interface TableMetadata {
     /** Unique anchor IDs for internal linking. */
     anchorIds?: string[];
     /**
-     * Layout alignment of the table on the page (e.g. inscript-editor's `CustomTable`).
+     * Layout alignment of the table on the page (e.g. an editor's custom table node).
      * @example 'center'
      */
     align?: 'left' | 'center' | 'right';
@@ -1657,12 +1699,12 @@ export interface ImageMetadata {
     /** Unique anchor IDs for internal linking. */
     anchorIds?: string[];
     /**
-     * Display width of the image (e.g. inscript-editor's `CustomImage`), as a CSS length or percentage.
+     * Display width of the image (e.g. an editor's custom image node), as a CSS length or percentage.
      * @example "50%"
      */
     width?: string;
     /**
-     * Layout alignment of the image (e.g. inscript-editor's `CustomImage`).
+     * Layout alignment of the image (e.g. an editor's custom image node).
      * @example 'center'
      */
     align?: 'left' | 'center' | 'right';
@@ -1672,14 +1714,19 @@ export interface ImageMetadata {
  * Markdown has no native syntax for this - see `MarkdownGenerator`'s `embed` case.
  */
 export interface EmbedMetadata {
-    /** The kind of embed. Only 'youtube' is supported today; the shape is generic for future providers. */
-    embedType: 'youtube';
-    /** The provider-specific video ID (e.g. the 11-character YouTube video ID). */
-    videoId: string;
-    /** The original/canonical URL of the embedded media, if known. */
+    /**
+     * The kind of embed. 'youtube' is recognized from a `data-youtube-video` wrapper or a YouTube
+     * iframe; 'iframe' is a generic preserved iframe (opt-in via `HtmlParserConfig.preserveIframes`).
+     */
+    embedType: 'youtube' | 'iframe';
+    /** The provider-specific video ID (e.g. the 11-character YouTube video ID). Absent for generic iframes. */
+    videoId?: string;
+    /** The original/canonical URL of the embedded media, if known. For a generic iframe, its `src`. */
     url?: string;
     /** Display width, as a CSS length or percentage. */
     width?: string;
+    /** Display height, as a CSS length or percentage (generic iframes). */
+    height?: string;
     /** Layout alignment of the embed. */
     align?: 'left' | 'center' | 'right';
 }
@@ -1763,6 +1810,14 @@ export interface NoteMetadata {
     anchorIds?: string[];
     /** The slide number this note is associated with (used in PowerPoint). */
     slideNumber?: number;
+    /**
+     * True for a footnote/endnote definition that no reference points at (an "orphan").
+     * The Markdown parser sets this when it recovers a `[^id]: ...` definition with no matching
+     * `[^id]` reference so the definition is preserved rather than dropped. Generators route such
+     * notes into their footnotes section without a citation marker, and the HTML generator omits
+     * the (otherwise dangling) back-link.
+     */
+    unreferenced?: boolean;
 }
 /**
  * Metadata for break nodes.
@@ -1777,8 +1832,11 @@ export interface BreakMetadata {
      * - 'lastRenderedPage': The editing application has inserted a soft break on the last save.
      * - 'textWrapping' (default, assumed when not specified): The next text will be placed on the next line.
      * - 'carriageReturn': An explicit carriage return (w:cr) equivalent to a hard line break.
+     * - 'thematic': A thematic break (Markdown `---`/`***`/`___`, HTML `<hr>`) - a horizontal
+     *   rule separating sections, distinct from a page break. Emitted as `---` in Markdown and
+     *   `<hr>` in HTML.
      */
-    breakType: 'column' | 'page' | 'lastRenderedPage' | 'textWrapping' | 'carriageReturn';
+    breakType: 'column' | 'page' | 'lastRenderedPage' | 'textWrapping' | 'carriageReturn' | 'thematic';
     /**
      * Specifies the location which shall be used as the next available line when breakType
      * has a value of 'textWrapping'. Should be ignored for other break types.
@@ -1800,7 +1858,7 @@ export interface CodeMetadata {
     /**
      * When set, this node is a LaTeX math expression rather than a code block. `node.text`
      * holds the bare LaTeX (delimiters excluded); 'inline' round-trips as `$...$`,
-     * 'block' as `$$...$$`. Matches inscript-editor's math node (Roadmap Step 11.5).
+     * 'block' as `$$...$$`. Matches attribute-driven editors' math nodes.
      */
     math?: 'inline' | 'block';
 }