Add wikipedia box

Logic/Web/Wikidata.ts

@@ -0,0 +1,89 @@
+import {Utils} from "../../Utils";
+
+
+export interface WikidataResponse {
+
+    id: string,
+    labels: Map<string, string>,
+    descriptions: Map<string, string>,
+    claims: Map<string, Set<string>>,
+    wikisites: Map<string, string>
+    commons: string
+}
+
+/**
+ * Utility functions around wikidata
+ */
+export default class Wikidata {
+
+    private static ParseResponse(entity: any): WikidataResponse {
+        const labels = new Map<string, string>()
+        for (const labelName in entity.labels) {
+            // The labelname is the language code
+            labels.set(labelName, entity.labels[labelName].value)
+        }
+
+        const descr = new Map<string, string>()
+        for (const labelName in entity.descriptions) {
+            // The labelname is the language code
+            descr.set(labelName, entity.descriptions[labelName].value)
+        }
+
+        const sitelinks = new Map<string, string>();
+        for (const labelName in entity.sitelinks) {
+            // labelName is `${language}wiki`
+            const language = labelName.substring(0, labelName.length - 4)
+            const title = entity.sitelinks[labelName].title
+            sitelinks.set(language, title)
+        }
+        
+        const commons = sitelinks.get("commons")
+        sitelinks.delete("commons")
+
+        const claims = new Map<string, Set<string>>();
+        for (const claimId of entity.claims) {
+
+            const claimsList: any[] = entity.claims[claimId]
+            const values = new Set<string>()
+            for (const claim of claimsList) {
+                const value = claim.mainsnak.datavalue.value;
+                values.add(value)
+            }
+            claims.set(claimId, values);
+        }
+
+        return {
+            claims: claims,
+            descriptions: descr,
+            id: entity.id,
+            labels: labels,
+            wikisites: sitelinks,
+            commons: commons
+        }
+    }
+
+    /**
+     * Loads a wikidata page
+     * @returns the entity of the given value
+     */
+    public static async LoadWikidataEntry(value: string | number): Promise<WikidataResponse> {
+        const wikidataUrl = "https://www.wikidata.org/wiki/"
+        if (typeof value === "number") {
+            value = "Q" + value
+        }
+        if (value.startsWith(wikidataUrl)) {
+            value = value.substring(wikidataUrl.length)
+        }
+        if (value.startsWith("http")) {
+            // Probably some random link in the image field - we skip it
+            return undefined
+        }
+        if (!value.startsWith("Q")) {
+            value = "Q" + value
+        }
+        const url = "https://www.wikidata.org/wiki/Special:EntityData/" + value + ".json";
+        const response = await Utils.downloadJson(url)
+        return Wikidata.ParseResponse(response.entities[value]);
+    }
+
+}

Logic/Web/Wikipedia.ts

@@ -2,7 +2,8 @@
  * Some usefull utility functions around the wikipedia API
  */
 import {Utils} from "../../Utils";
-import WikipediaBox from "../../UI/WikipediaBox";
+import {UIEventSource} from "../UIEventSource";
+import Wikidata from "./Wikidata";
 
 export default class Wikipedia {
 
@@ -14,22 +15,36 @@ export default class Wikipedia {
     private static readonly classesToRemove = [
         "shortdescription",
         "sidebar",
-        "infobox",
+        "infobox","infobox_v2",
+        "noprint",
+        "ambox",
         "mw-editsection",
+        "mw-selflink",
         "hatnote" // Often redirects
     ]
 
-    public static async GetArticle(options: {
+    private static readonly _cache = new Map<string, UIEventSource<{ success: string } | { error: any }>>()
+    
+    public static GetArticle(options: {
         pageName: string,
-        language?: "en" | string,
-        section?: number,
+        language?: "en" | string}): UIEventSource<{ success: string } | { error: any }>{
+        const key = (options.language ?? "en")+":"+options.pageName
+        const cached = Wikipedia._cache.get(key)
+        if(cached !== undefined){
+            return cached
+        }
+        const v = UIEventSource.FromPromiseWithErr(Wikipedia.GetArticleAsync(options))
+        Wikipedia._cache.set(key, v)
+        return v;
+    }
+    
+    public static async GetArticleAsync(options: {
+        pageName: string,
+        language?: "en" | string
     }): Promise<string> {
 
-        let section = ""
-        if (options.section !== undefined) {
-            section = "&section=" + options.section
-        }
-        const url = `https://${options.language ?? "en"}.wikipedia.org/w/api.php?action=parse${section}&format=json&origin=*&prop=text&page=` + options.pageName
+        const language = options.language ?? "en"
+        const url = `https://${language}.wikipedia.org/w/api.php?action=parse&format=json&origin=*&prop=text&page=` + options.pageName
         const response = await Utils.downloadJson(url)
         const html = response["parse"]["text"]["*"];
 
@@ -43,7 +58,19 @@ export default class Wikipedia {
                 toRemoveElement.parentElement?.removeChild(toRemoveElement)
             }
         }
-        return content.innerHTML;
+
+        const links = Array.from(content.getElementsByTagName("a"))
+
+        console.log("Links are", links)
+        // Rewrite relative links to absolute links + open them in a new tab
+        links.filter(link => link.getAttribute("href")?.startsWith("/") ?? false).
+        forEach(link => {
+            link.target = '_blank'
+            // note: link.getAttribute("href") gets the textual value, link.href is the rewritten version which'll contain the host for relative paths
+            link.href = `https://${language}.wikipedia.org${link.getAttribute("href")}`;
+        })
+
+        return content.innerHTML
     }
 
 }