file-type 8.1.0 → 9.0.0

package/index.js

@@ -151,41 +151,54 @@ module.exports = input => {
 			};
 		}
 
-		// https://github.com/file/file/blob/master/magic/Magdir/msooxml
-		if (check(oxmlContentTypes, {offset: 30}) || check(oxmlRels, {offset: 30})) {
-			const sliced = buf.subarray(4, 4 + 2000);
-			const nextZipHeaderIndex = arr => arr.findIndex((el, i, arr) => arr[i] === 0x50 && arr[i + 1] === 0x4B && arr[i + 2] === 0x3 && arr[i + 3] === 0x4);
-			const header2Pos = nextZipHeaderIndex(sliced);
-
-			if (header2Pos !== -1) {
-				const slicedAgain = buf.subarray(header2Pos + 8, header2Pos + 8 + 1000);
-				const header3Pos = nextZipHeaderIndex(slicedAgain);
-
-				if (header3Pos !== -1) {
-					const offset = 8 + header2Pos + header3Pos + 30;
-
-					if (checkString('word/', {offset})) {
-						return {
-							ext: 'docx',
-							mime: 'application/vnd.openxmlformats-officedocument.wordprocessingml.document'
-						};
-					}
-
-					if (checkString('ppt/', {offset})) {
-						return {
-							ext: 'pptx',
-							mime: 'application/vnd.openxmlformats-officedocument.presentationml.presentation'
-						};
-					}
-
-					if (checkString('xl/', {offset})) {
-						return {
-							ext: 'xlsx',
-							mime: 'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet'
-						};
-					}
+		// The docx, xlsx and pptx file types extend the Office Open XML file format:
+		// https://en.wikipedia.org/wiki/Office_Open_XML_file_formats
+		// We look for:
+		// - one entry named '[Content_Types].xml' or '_rels/.rels',
+		// - one entry indicating specific type of file.
+		// MS Office, OpenOffice and LibreOffice may put the parts in different order, so the check should not rely on it.
+		const findNextZipHeaderIndex = (arr, startAt = 0) => arr.findIndex((el, i, arr) => i >= startAt && arr[i] === 0x50 && arr[i + 1] === 0x4B && arr[i + 2] === 0x3 && arr[i + 3] === 0x4);
+
+		let zipHeaderIndex = 0; // The first zip header was already found at index 0
+		let oxmlFound = false;
+		let type = null;
+
+		do {
+			const offset = zipHeaderIndex + 30;
+
+			if (!oxmlFound) {
+				oxmlFound = (check(oxmlContentTypes, {offset}) || check(oxmlRels, {offset}));
+			}
+
+			if (!type) {
+				if (checkString('word/', {offset})) {
+					type = {
+						ext: 'docx',
+						mime: 'application/vnd.openxmlformats-officedocument.wordprocessingml.document'
+					};
+				} else if (checkString('ppt/', {offset})) {
+					type = {
+						ext: 'pptx',
+						mime: 'application/vnd.openxmlformats-officedocument.presentationml.presentation'
+					};
+				} else if (checkString('xl/', {offset})) {
+					type = {
+						ext: 'xlsx',
+						mime: 'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet'
+					};
 				}
 			}
+
+			if (oxmlFound && type) {
+				return type;
+			}
+
+			zipHeaderIndex = findNextZipHeaderIndex(buf, offset);
+		} while (zipHeaderIndex >= 0);
+
+		// No more zip parts available in the buffer, but maybe we are almost certain about the type?
+		if (type) {
+			return type;
 		}
 	}
 
@@ -312,13 +325,13 @@ module.exports = input => {
 		if (check([0x41, 0x56, 0x49], {offset: 8})) {
 			return {
 				ext: 'avi',
-				mime: 'video/x-msvideo'
+				mime: 'video/vnd.avi'
 			};
 		}
 		if (check([0x57, 0x41, 0x56, 0x45], {offset: 8})) {
 			return {
 				ext: 'wav',
-				mime: 'audio/x-wav'
+				mime: 'audio/vnd.wave'
 			};
 		}
 		// QLCM, QCP file
@@ -398,9 +411,9 @@ module.exports = input => {
 		check([0x66, 0x74, 0x79, 0x70, 0x4D, 0x34, 0x41], {offset: 4}) ||
 		check([0x4D, 0x34, 0x41, 0x20])
 	) {
-		return {
+		return { // MPEG-4 layer 3 (audio)
 			ext: 'm4a',
-			mime: 'audio/m4a'
+			mime: 'audio/mp4' // RFC 4337
 		};
 	}
 
@@ -468,13 +481,20 @@ module.exports = input => {
 		};
 	}
 
-	if (check([0x4D, 0x41, 0x43, 0x20])) {
+	if (check([0x4D, 0x41, 0x43, 0x20])) { // 'MAC '
 		return {
 			ext: 'ape',
 			mime: 'audio/ape'
 		};
 	}
 
+	if (check([0x77, 0x76, 0x70, 0x6B])) { // 'wvpk'
+		return {
+			ext: 'wv',
+			mime: 'audio/wavpack'
+		};
+	}
+
 	if (check([0x23, 0x21, 0x41, 0x4D, 0x52, 0x0A])) {
 		return {
 			ext: 'amr',
@@ -556,7 +576,7 @@ module.exports = input => {
 	) {
 		return {
 			ext: 'eot',
-			mime: 'application/octet-stream'
+			mime: 'application/vnd.ms-fontobject'
 		};
 	}
 

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "file-type",
-	"version": "8.1.0",
+	"version": "9.0.0",
 	"description": "Detect the file type of a Buffer/Uint8Array",
 	"license": "MIT",
 	"repository": "sindresorhus/file-type",

package/readme.md

@@ -76,7 +76,7 @@ Or `null` when no match.
 
 Type: `Buffer` `Uint8Array`
 
-It only needs the first 4100 bytes.
+It only needs the first 4100 bytes. The exception is detection of `docx`, `pptx`, and `xlsx` which potentially requires reading the whole file.
 
 
 ## Supported file types
@@ -164,6 +164,7 @@ It only needs the first 4100 bytes.
 - [`cur`](https://en.wikipedia.org/wiki/ICO_(file_format))
 - [`ktx`](https://www.khronos.org/opengles/sdk/tools/KTX/file_format_spec/)
 - [`ape`](https://en.wikipedia.org/wiki/Monkey%27s_Audio) - Monkey's Audio
+- [`wv`](https://en.wikipedia.org/wiki/WavPack) - WavPack
 
 *SVG isn't included as it requires the whole file to be read, but you can get it [here](https://github.com/sindresorhus/is-svg).*