ファイルと画像認識処理の改善 (#5690)

* dimensions制限とリファクタ

* comment

* 不要な変更削除

* use fromFile など

* Add probe-image-size.d.ts

* えーCRLFで作るなよ…

* Update src/@types/probe-image-size.d.ts

Co-Authored-By: Acid Chicken (硫酸鶏) <root@acid-chicken.com>

* fix d.ts

* Update src/@types/probe-image-size.d.ts

Co-Authored-By: Acid Chicken (硫酸鶏) <root@acid-chicken.com>

* Update src/@types/probe-image-size.d.ts

Co-Authored-By: Acid Chicken (硫酸鶏) <root@acid-chicken.com>

* fix

Co-authored-by: Acid Chicken (硫酸鶏) <root@acid-chicken.com>

4 files changed, 205 insertions, 47 deletions

diff --git a/src/misc/check-svg.ts b/src/misc/check-svg.ts
deleted file mode 100644
index 8ddeefede9..0000000000
--- a/src/misc/check-svg.ts
+++ /dev/null
@@ -1,12 +0,0 @@
-import * as fs from 'fs';
-import isSvg from 'is-svg';
-
-export default function(path: string) {
-	try {
-		const size = fs.statSync(path).size;
-		if (size > 1 * 1024 * 1024) return false;
-		return isSvg(fs.readFileSync(path));
-	} catch {
-		return false;
-	}
-}
diff --git a/src/misc/detect-mine.ts b/src/misc/detect-mine.ts
deleted file mode 100644
index f47f127353..0000000000
--- a/src/misc/detect-mine.ts
+++ /dev/null
@@ -1,31 +0,0 @@
-import * as fs from 'fs';
-import checkSvg from '../misc/check-svg';
-const FileType = require('file-type');
-
-export async function detectMine(path: string) {
-	return new Promise<[string, string | null]>((res, rej) => {
-		const readable = fs.createReadStream(path);
-		readable
-			.on('error', rej)
-			.once('data', async (buffer: Buffer) => {
-				readable.destroy();
-				const type = await FileType.fromBuffer(buffer);
-				if (type) {
-					if (type.mime == 'application/xml' && checkSvg(path)) {
-						res(['image/svg+xml', 'svg']);
-					} else {
-						res([type.mime, type.ext]);
-					}
-				} else if (checkSvg(path)) {
-					res(['image/svg+xml', 'svg']);
-				} else {
-					// 種類が同定できなかったら application/octet-stream にする
-					res(['application/octet-stream', null]);
-				}
-			})
-			.on('end', () => {
-				// maybe 0 bytes
-				res(['application/octet-stream', null]);
-			});
-	});
-}
diff --git a/src/misc/detect-url-mine.ts b/src/misc/detect-url-mime.ts
index eef64cfc56..8d71cd0137 100644
--- a/src/misc/detect-url-mine.ts
+++ b/src/misc/detect-url-mime.ts
@@ -1,14 +1,14 @@
 import { createTemp } from './create-temp';
 import { downloadUrl } from './donwload-url';
-import { detectMine } from './detect-mine';
+import { detectType } from './get-file-info';
 
-export async function detectUrlMine(url: string) {
+export async function detectUrlMime(url: string) {
 	const [path, cleanup] = await createTemp();
 
 	try {
 		await downloadUrl(url, path);
-		const [type] = await detectMine(path);
-		return type;
+		const { mime } = await detectType(path);
+		return mime;
 	} finally {
 		cleanup();
 	}
diff --git a/src/misc/get-file-info.ts b/src/misc/get-file-info.ts
new file mode 100644
index 0000000000..5ccb280260
--- /dev/null
+++ b/src/misc/get-file-info.ts
@@ -0,0 +1,201 @@
+import * as fs from 'fs';
+import * as crypto from 'crypto';
+import * as fileType from 'file-type';
+import isSvg from 'is-svg';
+import * as probeImageSize from 'probe-image-size';
+import * as sharp from 'sharp';
+
+export type FileInfo = {
+	size: number;
+	md5: string;
+	type: {
+		mime: string;
+		ext: string | null;
+	};
+	width?: number;
+	height?: number;
+	avgColor?: number[];
+	warnings: string[];
+};
+
+const TYPE_OCTET_STREAM = {
+	mime: 'application/octet-stream',
+	ext: null
+};
+
+const TYPE_SVG = {
+	mime: 'image/svg+xml',
+	ext: 'svg'
+};
+
+/**
+ * Get file information
+ */
+export async function getFileInfo(path: string): Promise<FileInfo> {
+	const warnings = [] as string[];
+
+	const size = await getFileSize(path);
+	const md5 = await calcHash(path);
+
+	let type = await detectType(path);
+
+	// image dimensions
+	let width: number | undefined;
+	let height: number | undefined;
+
+	if (['image/jpeg', 'image/gif', 'image/png', 'image/apng', 'image/webp', 'image/bmp', 'image/tiff', 'image/svg+xml', 'image/vnd.adobe.photoshop'].includes(type.mime)) {
+		const imageSize = await detectImageSize(path).catch(e => {
+			warnings.push(`detectImageSize failed: ${e}`);
+			return undefined;
+		});
+
+		// うまく判定できない画像は octet-stream にする
+		if (!imageSize) {
+			warnings.push(`cannot detect image dimensions`);
+			type = TYPE_OCTET_STREAM;
+		} else if (imageSize.wUnits === 'px') {
+			width = imageSize.width;
+			height = imageSize.height;
+
+			// 制限を超えている画像は octet-stream にする
+			if (imageSize.width > 16383 || imageSize.height > 16383) {
+				warnings.push(`image dimensions exceeds limits`);
+				type = TYPE_OCTET_STREAM;
+			}
+		} else {
+			warnings.push(`unsupported unit type: ${imageSize.wUnits}`);
+		}
+	}
+
+	// average color
+	let avgColor: number[] | undefined;
+
+	if (['image/jpeg', 'image/gif', 'image/png', 'image/apng', 'image/webp', 'image/svg+xml'].includes(type.mime)) {
+		avgColor = await calcAvgColor(path).catch(e => {
+			warnings.push(`calcAvgColor failed: ${e}`);
+			return undefined;
+		});
+	}
+
+	return {
+		size,
+		md5,
+		type,
+		width,
+		height,
+		avgColor,
+		warnings,
+	};
+}
+
+/**
+ * Detect MIME Type and extension
+ */
+export async function detectType(path: string) {
+	// Check 0 byte
+	const fileSize = await getFileSize(path);
+	if (fileSize === 0) {
+		return TYPE_OCTET_STREAM;
+	}
+
+	const type = await fileType.fromFile(path);
+
+	if (type) {
+		// XMLはSVGかもしれない
+		if (type.mime === 'application/xml' && await checkSvg(path)) {
+			return TYPE_SVG;
+		}
+
+		return {
+			mime: type.mime,
+			ext: type.ext
+		};
+	}
+
+	// 種類が不明でもSVGかもしれない
+	if (await checkSvg(path)) {
+		return TYPE_SVG;
+	}
+
+	// それでも種類が不明なら application/octet-stream にする
+	return TYPE_OCTET_STREAM;
+}
+
+/**
+ * Check the file is SVG or not
+ */
+export async function checkSvg(path: string) {
+	try {
+		const size = await getFileSize(path);
+		if (size > 1 * 1024 * 1024) return false;
+		return isSvg(fs.readFileSync(path));
+	} catch {
+		return false;
+	}
+}
+
+/**
+ * Get file size
+ */
+export async function getFileSize(path: string): Promise<number> {
+	return new Promise<number>((res, rej) => {
+		fs.stat(path, (err, stats) => {
+			if (err) return rej(err);
+			res(stats.size);
+		});
+	});
+}
+
+/**
+ * Calculate MD5 hash
+ */
+async function calcHash(path: string): Promise<string> {
+	return new Promise<string>((res, rej) => {
+		const readable = fs.createReadStream(path);
+		const hash = crypto.createHash('md5');
+		const chunks: Buffer[] = [];
+		readable
+			.on('error', rej)
+			.pipe(hash)
+			.on('error', rej)
+			.on('data', chunk => chunks.push(chunk))
+			.on('end', () => {
+				const buffer = Buffer.concat(chunks);
+				res(buffer.toString('hex'));
+			});
+	});
+}
+
+/**
+ * Detect dimensions of image
+ */
+async function detectImageSize(path: string): Promise<{
+	width: number;
+	height: number;
+	wUnits: string;
+	hUnits: string;
+}> {
+	const readable = fs.createReadStream(path);
+	const imageSize = await probeImageSize(readable);
+	readable.destroy();
+	return imageSize;
+}
+
+/**
+ * Calculate average color of image
+ */
+async function calcAvgColor(path: string): Promise<number[]> {
+	const img = sharp(path);
+
+	const info = await (img as any).stats();
+
+	if (info.isOpaque) {
+		const r = Math.round(info.channels[0].mean);
+		const g = Math.round(info.channels[1].mean);
+		const b = Math.round(info.channels[2].mean);
+
+		return [r, g, b];
+	} else {
+		return [255, 255, 255];
+	}
+}