@file-type/xml 0.1.1 → 0.2.1

package/README.md

@@ -1,25 +1,42 @@
-[![NPM version](https://img.shields.io/npm/v/@file-type/xml.svg)](https://npmjs.org/package/@file-type/xml)
-[![Node.js CI](https://github.com/Borewit/file-type-xml/actions/workflows/nodejs-ci.yml/badge.svg)](https://github.com/Borewit/file-type-xml/actions/workflows/nodejs-ci.yml)
-# @file-type/xml
-
-Detector plugin for [file-type](https://github.com/sindresorhus/file-type) for XML files.
-
-## Installation
-
-```bash
-npm install @file-type/xml
-```
-
-### Usage
-
-The following example shows how add the XML detector to [file-type](https://github.com/sindresorhus/file-type).
-```js
-import {NodeFileTypeParser} from 'file-type';
-import {detectXml} from '@file-type/xml';
-
-const parser = new NodeFileTypeParser({customDetectors: [detectXml]});
-const fileType = await parser.fromFile('example.kml');
-console.log(fileType);
-```
-
-
+[![NPM version](https://img.shields.io/npm/v/@file-type/xml.svg)](https://npmjs.org/package/@file-type/xml)
+[![Node.js CI](https://github.com/Borewit/file-type-xml/actions/workflows/nodejs-ci.yml/badge.svg)](https://github.com/Borewit/file-type-xml/actions/workflows/nodejs-ci.yml)
+# @file-type/xml
+
+Detector plugin for [file-type](https://github.com/sindresorhus/file-type) for XML files.
+
+## Installation
+
+```bash
+npm install @file-type/xml
+```
+
+## Usage
+
+The following example shows how add the XML detector to [file-type](https://github.com/sindresorhus/file-type).
+```js
+import {NodeFileTypeParser} from 'file-type';
+import {detectXml} from '@file-type/xml';
+
+const parser = new NodeFileTypeParser({customDetectors: [detectXml]});
+const fileType = await parser.fromFile('example.kml');
+console.log(fileType);
+```
+
+You can also use the XML detector outside file-type:
+```js
+import {XmlTextDetector} from 'index.js';
+
+xmlTextDetector.write('<svg xmlns="http://www.w3.org/2000/svg"><path fill="#00CD9F"/></svg>');
+const fileType = xmlTextDetector.fileType;
+console.log(JSON.stringify(fileType)); // Outputs: {"ext":"svg","mime":"image/svg+xml"}
+```
+
+## Support file formats
+
+- [XML](https://en.wikipedia.org/wiki/XML) (default for XML, unless more specific format was detected)
+- [GML (Geography Markup Language)](https://en.wikipedia.org/wiki/Geography_Markup_Language)
+- [KML (Keyhole Markup Language)](https://en.wikipedia.org/wiki/XHTML)
+- [MusicXML, Uncompressed](https://en.wikipedia.org/wiki/MusicXML)
+- [RSS (RDF Site Summary or Really Simple Syndication)](https://en.wikipedia.org/wiki/RSS)
+- [SVG: (Scalable Vector Graphics)](https://en.wikipedia.org/wiki/SVG)
+- [XHTML](https://en.wikipedia.org/wiki/XHTML)

package/lib/index.js

@@ -15,13 +15,17 @@ function startsWith(array, prefix) {
 
 function isXml(array) {
   if (startsWith(array,[60, 63, 120, 109, 108, 32])) {
-     return {xml: true,  encoding: 'utf-8'}
+     return {xml: true,  encoding: 'utf-8', offset: 0}
   } else if (startsWith(array,[0xEF, 0xBB, 0xBF, 60, 63, 120, 109, 108, 32])) { // UTF-8 BOM
-    return {xml: true,  encoding: 'utf-8'}
+    return {xml: true,  encoding: 'utf-8', offset: 3}
   } else if (startsWith(array,[0xFE, 0xFF, 0, 60,  0, 63, 0, 120, 0, 109, 0, 108, 0, 32 ])) {
-    return {xml: true,  encoding: 'utf-16be'}
+    return {xml: true,  encoding: 'utf-16be', offset: 2}
   } else if (startsWith(array,[0xFF, 0xFE, 60,  0, 63, 0, 120, 0, 109, 0, 108, 0, 32, 0 ])) {
-    return {xml: true,  encoding: 'utf-16le'}
+    return {xml: true,  encoding: 'utf-16le', offset: 2}
+  } else if (startsWith(array,[0, 60,  0, 63, 0, 120, 0, 109, 0, 108, 0, 32 ])) {
+    return {xml: true,  encoding: 'utf-16be', offset: 0}
+  } else if (startsWith(array,[60,  0, 63, 0, 120, 0, 109, 0, 108, 0, 32, 0 ])) {
+    return {xml: true,  encoding: 'utf-16le', offset: 0}
   }
   return {xml: false, encoding: undefined}
 }
@@ -58,7 +62,8 @@ const namespaceMapping = {
 }
 
 /**
- * Maps the root element name to corresponding file-type
+ * Maps the root element name to corresponding file-type.
+ * Used for Non-namespaced XML
  * @type {{rss: {ext: string, mime: string}}}
  */
 const rootNameMapping = {
@@ -69,50 +74,81 @@ const rootNameMapping = {
   'score-partwise': {
     ext: 'musicxml',
     mime: 'application/vnd.recordare.musicxml+xml',
-  }
+  },
+  svg: {
+    ext: 'svg',
+    mime: 'image/svg+xml',
+  },
 }
 
-export const detectXml = async tokenizer => {
-
-  const buffer = new Uint8Array(512);
-
-  // Increase sample size from 12 to 256.
-  await tokenizer.peekBuffer(buffer, {length: 128, mayBeLess: true});
-
-  const {xml, encoding} = isXml(buffer);
-
-  if (xml) {
-    let fileType;
-
-    const parser = sax.parser(true);
+export class XmlTextDetector {
 
-    let firstTag = true;
-    let onEnd = false;
+  constructor(options) {
+    this.options = options ?? {};
+    this.firstTag = true;
+    this.onEnd = false;
+    this.parser = sax.parser(true);
+    this.nesting = 0;
 
-    parser.onerror = e => {
-      onEnd = true;
+    this.parser.onerror = e => {
+      if (e.message.startsWith('Invalid character entity')) { // Allow entity reference
+        return;
+      }
+      this.fileType = undefined;
+      this.onEnd = true;
     };
-    parser.onopentag = node => {
-      if (!firstTag) {
+    this.parser.onopentag = node => {
+      ++this.nesting;
+      if (!this.firstTag || this.onEnd) {
         return;
       }
-      firstTag = false;
+      this.firstTag = false;
       const nsNode = extractNsElement(node);
       if (nsNode.ns) {
         // Resolve file-type boot root element namespace
-        fileType = namespaceMapping[nsNode.ns];
+        this.fileType = namespaceMapping[nsNode.ns.toLowerCase()];
       } else {
         // Fall back on element name if there is no namespace
-        fileType = rootNameMapping[nsNode.name];
+        this.fileType = rootNameMapping[nsNode.name?.toLowerCase()];
       }
 
-      if (fileType) {
-        onEnd = true;
+      if (this.fileType && !this.options.fullScan) {
+        this.onEnd = true;
       }
     };
-    parser.onend = () => {
-      onEnd = true;
-    };
+
+    this.parser.onclosetag = () => {
+      --this.nesting;
+    }
+  }
+
+  write(text) {
+    this.parser.write(text);
+  }
+
+  close() {
+    this.parser.close();
+    this.onEnd = true;
+  }
+
+  isValid() {
+    return this.nesting === 0;
+  }
+}
+
+export const detectXml = async tokenizer => {
+
+  const buffer = new Uint8Array(512);
+
+  // Increase sample size from 12 to 256.
+  await tokenizer.peekBuffer(buffer, {length: 128, mayBeLess: true});
+
+  const {xml, encoding, offset} = isXml(buffer);
+
+  if (xml) {
+    await tokenizer.ignore(offset);
+
+    const xmlTextDetector = new XmlTextDetector();
 
     const textDecoder = new TextDecoder(encoding);
 
@@ -120,17 +156,17 @@ export const detectXml = async tokenizer => {
       const len = await tokenizer.readBuffer(buffer, {mayBeLess: true});
       const portion = buffer.subarray(0, len);
       const text = textDecoder.decode(portion);
-      parser.write(text);
+      xmlTextDetector.write(text);
       if (len < buffer.length) {
-        parser.close();
-        onEnd = true;
+        xmlTextDetector.close();
       }
-    } while(!onEnd)
+    } while(!xmlTextDetector.onEnd)
 
-    return fileType ?? {
+    return xmlTextDetector.fileType ?? {
       ext: 'xml',
       mime: 'application/xml',
     }
   }
 
-};
+};
+

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@file-type/xml",
-  "version": "0.1.1",
+  "version": "0.2.1",
   "description": "XML detection plugin",
   "type": "module",
   "exports": "./lib/index.js",
@@ -9,13 +9,18 @@
   },
   "keywords": [
     "file-type",
+    "detect",
     "detection",
     "detector",
     "XML",
+    "signature",
+    "namespace",
     "SVG",
     "XHTML",
     "RSS",
-    "KML"
+    "KML",
+    "GML",
+    "MusicXML"
   ],
   "dependencies": {
     "sax": "^1.4.1",