Fix: canonicalize wikimedia links, see #2367, fix attribution

src/Logic/ImageProviders/WikimediaImageProvider.ts

@@ -25,12 +25,36 @@ export class WikimediaImageProvider extends ImageProvider {
         super()
     }
 
-    private static ExtractFileName(url: string) {
+    /**
+     * Replaces (multiple) spaces to underscores.
+     * Will remove a "File:"-prefix
+     *
+     * WikimediaImageProvider.makeCanonical("Some File.jpg") // => "Some_File.jpg"
+     *
+     * // Double spaces
+     * WikimediaImageProvider.makeCanonical("Some      File.jpg") // => "Some_File.jpg"
+     * WikimediaImageProvider.makeCanonical("Some+File.jpg") // => "Some+File.jpg"
+     *
+     * // Remove File: prefix
+     */
+    private static makeCanonical(filename: string): string {
+        if (filename.startsWith("File:")) {
+            filename = filename.substring(5)
+        }
+        return filename.trim().replace(/\s+/g, "_")
+    }
+
+    /**
+     *
+     * WikimediaImageProvider.extractFileName("https://commons.wikimedia.org/wiki/File:Somefile.jpg") // => "Somefile.jpg"
+     * WikimediaImageProvider.extractFileName("https://commons.wikimedia.org/wiki/File:S%C3%A8vres%20-%20square_madame_de_Pompadour_-_bo%C3%AEte_%C3%A0_livres.jpg?uselang=en") // => "Sèvres_-_square_madame_de_Pompadour_-_boîte_à_livres.jpg"
+     */
+    private static extractFileName(url: string) {
         if (!url.startsWith("http")) {
             return url
         }
-        const path = new URL(url).pathname
+        const path = decodeURIComponent(new URL(url).pathname)
-        return path.substring(path.lastIndexOf("/") + 1)
+        return WikimediaImageProvider.makeCanonical(path.substring(path.lastIndexOf("/") + 1))
     }
 
     private static PrepareUrl(value: string, useHd = false): string {
@@ -98,6 +122,15 @@ export class WikimediaImageProvider extends ImageProvider {
         return this.UrlForImage("File:" + value)
     }
 
+    /**
+     *
+     * @param key
+     * @param value
+     * @constructor
+     *
+     * const result = await WikimediaImageProvider.singleton.ExtractUrls("wikimedia_commons", "File:Sèvres_-_square_madame_de_Pompadour_-_boîte_à_livres.jpg")
+     * result[0].url_hd // => "https://commons.wikimedia.org/wiki/Special:FilePath/File%3AS%C3%A8vres_-_square_madame_de_Pompadour_-_bo%C3%AEte_%C3%A0_livres.jpg"
+     */
     public async ExtractUrls(key: string, value: string): undefined | Promise<ProvidedImage[]> {
         const hasCommonsPrefix = WikimediaImageProvider.startsWithCommonsPrefix(value)
         if (key !== undefined && key !== this.commons_key && !hasCommonsPrefix) {
@@ -123,7 +156,8 @@ export class WikimediaImageProvider extends ImageProvider {
     }
 
     public async DownloadAttribution(img: { url: string }): Promise<LicenseInfo> {
-        const filename = WikimediaImageProvider.ExtractFileName(img.url)
+        const filename = "File:" + WikimediaImageProvider.extractFileName(img.url)
+        console.log("Downloading attribution for", filename, img.url)
         if (filename === "") {
             return undefined
         }
@@ -145,23 +179,19 @@ export class WikimediaImageProvider extends ImageProvider {
             pageInfo = pages.at(-1)
         }
         if (pageInfo === undefined) {
+            console.warn("No attribution found for wikimedia image:", filename)
             return undefined
         }
 
         const license = (pageInfo.imageinfo ?? [])[0]?.extmetadata
         if (license === undefined) {
             console.warn(
-                "The file",
+                "The file", filename, "has no usable metedata or license attached... Please fix the license info file yourself!"
-                filename,
-                "has no usable metedata or license attached... Please fix the license info file yourself!"
             )
             return undefined
         }
 
-        let title = pageInfo.title
+        let title = WikimediaImageProvider.makeCanonical(pageInfo.title)
-        if (title.startsWith("File:")) {
-            title = title.substr("File:".length)
-        }
         if (title.endsWith(".jpg") || title.endsWith(".png")) {
             title = title.substring(0, title.length - 4)
         }
@@ -180,9 +210,7 @@ export class WikimediaImageProvider extends ImageProvider {
     }
 
     private UrlForImage(image: string): ProvidedImage {
-        if (!image.startsWith("File:")) {
+        image = "File:" + WikimediaImageProvider.makeCanonical(image)
-            image = "File:" + image
-        }
         return {
             url: WikimediaImageProvider.PrepareUrl(image),
             url_hd: WikimediaImageProvider.PrepareUrl(image, true),