officeparser 7.5.1 → 7.6.2

package/dist/utils/configUtils.js

@@ -207,7 +207,7 @@ function resolveGeneratorConfig(destination, astConfig, userConfig) {
     // 2. Merge common properties and sub-configs
     if (userConfig) {
         // Extract sub-configs to avoid shallow-overwriting the whole sub-config objects
-        const { htmlConfig, mdConfig, pdfConfig, csvConfig, textConfig, chunksConfig, ...commonProps } = userConfig;
+        const { htmlConfig, mdConfig, pdfConfig, csvConfig, textConfig, rtfConfig, chunksConfig, ...commonProps } = userConfig;
         Object.assign(config, withoutPrototypeKeys(commonProps));
         // Merge sub-configs individually, ignoring undefined properties to preserve defaults
         const mergeSubConfig = (target, source) => {
@@ -254,6 +254,8 @@ function resolveGeneratorConfig(destination, astConfig, userConfig) {
             mergeSubConfig(config.csvConfig, csvConfig);
         if (textConfig)
             mergeSubConfig(config.textConfig, textConfig);
+        if (rtfConfig)
+            mergeSubConfig(config.rtfConfig, rtfConfig);
         if (chunksConfig)
             mergeSubConfig(config.chunksConfig, chunksConfig);
     }

package/dist/utils/sanitize.d.ts

@@ -62,6 +62,15 @@ export declare function sanitizeCssValue(value: string): string;
  * URLs are passed through.
  */
 export declare function sanitizeUrl(url: string): string;
+/**
+ * Decide whether a non-provider `<iframe>` should be preserved, per
+ * `HtmlParserConfig.preserveIframes`. `true` allows any src; an array is a hostname allowlist,
+ * where an entry matches the src's host exactly or as a `.`-suffix (so `"vimeo.com"` also matches
+ * `player.vimeo.com`). A relative or unparseable src is allowed only under `true`. This is a
+ * preservation gate, not a sanitizer - the src is still scheme-checked with `sanitizeUrl` on
+ * generation.
+ */
+export declare function iframeAllowed(src: string, preserve: boolean | string[] | undefined): boolean;
 /**
  * Like sanitizeUrl but for an <img>/<source> src: additionally permits
  * `data:image/*` URIs (embedded document images) while still rejecting

package/dist/utils/sanitize.js

@@ -15,6 +15,7 @@ exports.escapeHtml = escapeHtml;
 exports.escapeXml = escapeXml;
 exports.sanitizeCssValue = sanitizeCssValue;
 exports.sanitizeUrl = sanitizeUrl;
+exports.iframeAllowed = iframeAllowed;
 exports.sanitizeImageUrl = sanitizeImageUrl;
 exports.serializeForInlineScript = serializeForInlineScript;
 exports.csvSafeCell = csvSafeCell;
@@ -149,6 +150,31 @@ function sanitizeUrl(url) {
     // Emit the same normalized string that was validated.
     return escapeHtml(stripped);
 }
+/**
+ * Decide whether a non-provider `<iframe>` should be preserved, per
+ * `HtmlParserConfig.preserveIframes`. `true` allows any src; an array is a hostname allowlist,
+ * where an entry matches the src's host exactly or as a `.`-suffix (so `"vimeo.com"` also matches
+ * `player.vimeo.com`). A relative or unparseable src is allowed only under `true`. This is a
+ * preservation gate, not a sanitizer - the src is still scheme-checked with `sanitizeUrl` on
+ * generation.
+ */
+function iframeAllowed(src, preserve) {
+    if (preserve === true)
+        return true;
+    if (!Array.isArray(preserve) || preserve.length === 0)
+        return false;
+    let host;
+    try {
+        host = new URL(src).hostname.toLowerCase();
+    }
+    catch {
+        return false;
+    }
+    return preserve.some(entry => {
+        const e = String(entry).toLowerCase().trim();
+        return e.length > 0 && (host === e || host.endsWith('.' + e));
+    });
+}
 /**
  * Like sanitizeUrl but for an <img>/<source> src: additionally permits
  * `data:image/*` URIs (embedded document images) while still rejecting

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "officeparser",
-  "version": "7.5.1",
+  "version": "7.6.2",
   "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",
@@ -31,10 +31,10 @@
     "build": "npm run sync:versions && npm run build:node && npm run build:esm-wrapper && npm run build:browser:types && npm run build:browser",
     "build:node": "tsc",
     "build:esm-wrapper": "node scripts/generate-esm-wrapper.js",
-    "build:browser:types": "dts-bundle-generator --no-check -o dist/officeparser.browser.d.ts src/index.ts && cp dist/officeparser.browser.d.ts dist/officeparser.browser.slim.d.ts",
+    "build:browser:types": "dts-bundle-generator --no-check -o dist/officeparser.browser.d.ts src/index.ts && node -e \"require('fs').copyFileSync('dist/officeparser.browser.d.ts','dist/officeparser.browser.slim.d.ts')\"",
     "build:browser": "node build_browser.js && npm run sync:docs",
     "sync:versions": "node scripts/sync-pdfjs-versions.js",
-    "sync:docs": "mkdir -p docs/dist && cp dist/officeparser.browser.iife.js docs/dist/ && cp dist/officeparser.browser.mjs docs/dist/ && cp dist/officeparser.browser.slim.iife.js docs/dist/ && cp dist/officeparser.browser.slim.mjs docs/dist/ && mkdir -p docs/test/files && cp -r test/files/* docs/test/files/",
+    "sync:docs": "node -e \"const fs=require('fs');fs.mkdirSync('docs/dist',{recursive:true});for(const f of ['officeparser.browser.iife.js','officeparser.browser.mjs','officeparser.browser.slim.iife.js','officeparser.browser.slim.mjs'])fs.copyFileSync('dist/'+f,'docs/dist/'+f);fs.mkdirSync('docs/test/files',{recursive:true});fs.cpSync('test/files','docs/test/files',{recursive:true})\"",
     "lint": "eslint src",
     "test": "npm run lint && npm run test:clean && npm run build && npm run test:license && npm run test:artifacts && npm run test:parser && npm run test:exhaustive && npm run test:generator && npm run test:security && npm run test:cli",
     "test:fast": "npm run lint && npm run test:parser:fast && npm run test:exhaustive && npm run test:generator:fast && npm run test:security && npm run test:cli:fast",
@@ -53,8 +53,8 @@
     "test:visualizer": "node test/testVisualizer.js",
     "test:integration": "node test/testIntegration.js",
     "test:license": "npm run sbom && node scripts/validate-licenses.js",
-    "test:clean": "rm -rf test/results test/generator/results test/generator/output test/parser/results test/parser/output test/cli/results",
-    "clean": "rm -rf dist && npm run test:clean",
+    "test:clean": "node -e \"for(const d of ['test/results','test/generator/results','test/generator/output','test/parser/results','test/parser/output','test/cli/results'])require('fs').rmSync(d,{recursive:true,force:true})\"",
+    "clean": "node -e \"require('fs').rmSync('dist',{recursive:true,force:true})\" && npm run test:clean",
     "sbom": "npx --yes @cyclonedx/cyclonedx-npm --output-format json --output-file dist/sbom.cdx.json --omit dev",
     "prepublishOnly": "npm run build",
     "prepare": "husky"