Default MIME image type: docs & refactoring

danburzo · Aug 11, 2024 · 6c2ba67 · 6c2ba67
1 parent d7b655b
commit 6c2ba67
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 30 deletions.
diff --git a/src/inline-images.js b/src/inline-images.js
@@ -1,5 +1,5 @@
 import { parseSrcset, stringifySrcset } from 'srcset';
-import { getMimetypeFromURL, imageMimetypes } from './util/file-mimetype.js';
+import { getMimetypeFromURL, isImageURL } from './util/file-mimetype.js';
 import fetchBase64 from './util/fetch-base64.js';
 
 export default async function inlineImages(doc, fetchOptions = {}, out) {
@@ -9,17 +9,16 @@ export default async function inlineImages(doc, fetchOptions = {}, out) {
 	let src_promises = Array.from(
 		doc.querySelectorAll('picture source[src], img[src]')
 	).map(async el => {
-		let mime = getMimetypeFromURL(el.src, doc);
 		/*
 			For web pages using atypical URLs for images
 			let’s just use a generic MIME type and hope it works.
 			
 			For an example, see:
 			https://github.com/danburzo/percollate/issues/174
 		*/
-		if (!mime || !imageMimetypes.has(mime)) {
-			mime = 'image';
-		}
+		let mime = isImageURL(el.src, doc)
+			? getMimetypeFromURL(el.src, doc)
+			: 'image';
 		if (out) {
 			out.write(el.src + '\n');
 		}
@@ -51,18 +50,16 @@ export default async function inlineImages(doc, fetchOptions = {}, out) {
 				stringifySrcset(
 					await Promise.all(
 						items.map(async item => {
-							let mime = getMimetypeFromURL(item.url, doc);
-
 							/*
 								For web pages using atypical URLs for images
 								let’s just use a generic MIME type and hope it works.
 								
 								For an example, see:
 								https://github.com/danburzo/percollate/issues/174
 							*/
-							if (!mime || !imageMimetypes.has(mime)) {
-								mime = 'image';
-							}
+							let mime = isImageURL(item.url, doc)
+								? getMimetypeFromURL(item.url, doc)
+								: 'image';
 							let data = await fetchBase64(
 								item.url,
 								fetchOptions

diff --git a/src/remote-resources.js b/src/remote-resources.js
@@ -2,8 +2,8 @@ import { randomUUID as uuid } from 'node:crypto';
 import { parseSrcset, stringifySrcset } from 'srcset';
 import {
 	getMimetypeFromURL,
-	imageMimetypes,
-	extForMimetype
+	extForMimetype,
+	isImageURL
 } from './util/file-mimetype.js';
 import { getUrlOrigin } from './util/url-origin.js';
 
@@ -15,13 +15,17 @@ export default function remoteResources(doc) {
 		and return a uniquely generated file name instead.
 	 */
 	function collectAndReplace(src) {
-		let ext;
-		let mime = getMimetypeFromURL(src);
-		if (mime && imageMimetypes.has(mime)) {
-			ext = extForMimetype(mime);
-		} else {
+		/*
+			If image URLs don’t have an extension with which
+			to figure out the image format, use the generic
+			`image` MIME media type and the `.image` extension
+			for EPUB remote resources.
+		*/
+		let mime = 'image',
 			ext = '.image';
-			mime = 'image';
+		if (isImageURL(src, doc)) {
+			mime = getMimetypeFromURL(src, doc);
+			ext = extForMimetype(mime);
 		}
 		if (!srcs.has(src)) {
 			srcs.set(src, {

diff --git a/src/util/file-mimetype.js b/src/util/file-mimetype.js
@@ -1,5 +1,16 @@
 import mimetype from 'mimetype';
 
+const IMAGE_MIMETYPES = new Set([
+	'image/avif',
+	'image/bmp',
+	'image/gif',
+	'image/jpeg',
+	'image/png',
+	'image/svg+xml',
+	'image/tiff',
+	'image/webp'
+]);
+
 /*
 	Add newer image formats to the MIME type database.
  */
@@ -25,16 +36,5 @@ export function getMimetypeFromURL(src, doc) {
 }
 
 export function isImageURL(src, doc) {
-	return imageMimetypes.has(getMimetypeFromURL(src, doc));
+	return IMAGE_MIMETYPES.has(getMimetypeFromURL(src, doc));
 }
-
-export const imageMimetypes = new Set([
-	'image/avif',
-	'image/bmp',
-	'image/gif',
-	'image/jpeg',
-	'image/png',
-	'image/svg+xml',
-	'image/tiff',
-	'image/webp'
-]);