Improve parsing TIFF files

Detect DNG files in a reliable matter. Fixes: #480
sindresorhus · Aug 13, 2021 · 8861872 · 8861872
1 parent c037ba7
commit 8861872
Showing 1 changed file with 48 additions and 41 deletions.
diff --git a/core.js b/core.js
@@ -68,7 +68,7 @@ export async function fileTypeFromTokenizer(tokenizer) {
 }
 
 async function _fromTokenizer(tokenizer) {
-	let buffer = Buffer.alloc(minimumBytes);
+	const buffer = Buffer.alloc(minimumBytes);
 	const bytesRead = 12;
 	const check = (header, options) => _check(buffer, header, options);
 	const checkString = (header, options) => check(stringToBytes(header), options);
@@ -617,58 +617,65 @@ async function _fromTokenizer(tokenizer) {
 		};
 	}
 
-	// TIFF, little-endian type
-	if (check([0x49, 0x49, 0x2A, 0x0])) {
-		if (checkString('CR', {offset: 8})) {
+	async function readTiffTag() {
+		const tagId = await tokenizer.readToken(Token.UINT16_LE);
+		await tokenizer.readToken(Token.UINT16_LE); // Tag data
+		await tokenizer.readToken(Token.UINT32_LE); // Nr of values
+		if (tagId === 50_706) { // Tag = DNGVersion
 			return {
-				ext: 'cr2',
-				mime: 'image/x-canon-cr2',
+				ext: 'dng',
+				mime: 'image/x-adobe-dng',
 			};
 		}
+	}
 
-		if (check([0x1C, 0x00, 0xFE, 0x00], {offset: 8}) || check([0x1F, 0x00, 0x0B, 0x00], {offset: 8})) {
-			return {
-				ext: 'nef',
-				mime: 'image/x-nikon-nef',
-			};
+	async function readTiffIFD() {
+		const numberOfTags = await tokenizer.readToken(Token.UINT16_LE);
+		for (let n = 0; n < numberOfTags; ++n) {
+			const fileType = await readTiffTag();
+			if (fileType) {
+				return fileType;
+			}
 		}
+	}
 
-		if (
-			check([0x08, 0x00, 0x00, 0x00], {offset: 4})
-				&& (check([0x2D, 0x00, 0xFE, 0x00], {offset: 8})
-					|| check([0x27, 0x00, 0xFE, 0x00], {offset: 8}))
-		) {
-			return {
-				ext: 'dng',
-				mime: 'image/x-adobe-dng',
+	// TIFF, little-endian type
+	if (check([0x49, 0x49])) {
+		const version = Token.UINT16_LE.get(buffer, 2);
+
+		if (version === 42) {
+			// TIFF file header
+
+			if (checkString('CR', {offset: 8})) {
+				return {
+					ext: 'cr2',
+					mime: 'image/x-canon-cr2',
+				};
+			}
+
+			if (check([0x1C, 0x00, 0xFE, 0x00], {offset: 8}) || check([0x1F, 0x00, 0x0B, 0x00], {offset: 8})) {
+				return {
+					ext: 'nef',
+					mime: 'image/x-nikon-nef',
+				};
+			}
+
+			const ifdOffset = Token.UINT16_LE.get(buffer, 4);
+			await tokenizer.ignore(ifdOffset);
+			const fileType = await readTiffIFD();
+			return fileType ? fileType : {
+				ext: 'tif',
+				mime: 'image/tiff',
 			};
 		}
 
-		buffer = Buffer.alloc(24);
-		await tokenizer.peekBuffer(buffer);
-		if (
-			(check([0x10, 0xFB, 0x86, 0x01], {offset: 4}) || check([0x08, 0x00, 0x00, 0x00], {offset: 4}))
-			// This pattern differentiates ARW from other TIFF-ish file types:
-			&& check([0x00, 0xFE, 0x00, 0x04, 0x00, 0x01, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x03, 0x01], {offset: 9})
-		) {
+		if (version === 43) {
+			// Big TIFF file header
 			return {
-				ext: 'arw',
-				mime: 'image/x-sony-arw',
+				ext: 'tif',
+				mime: 'image/tiff',
 			};
 		}
-
-		return {
-			ext: 'tif',
-			mime: 'image/tiff',
-		};
-	}
-
-	// TIFF, big-endian type
-	if (check([0x4D, 0x4D, 0x0, 0x2A])) {
-		return {
-			ext: 'tif',
-			mime: 'image/tiff',
-		};
 	}
 
 	if (checkString('MAC ')) {