Graciously handle multiple entries in wikidata for fetching images and showing articles, verious bug fixes

This commit is contained in:
Pieter Vander Vennet 2021-10-07 22:06:47 +02:00
parent 8d52ef1106
commit a996ba2a7c
14 changed files with 231 additions and 90 deletions

View file

@ -1,17 +1,20 @@
import {UIEventSource} from "../UIEventSource";
import BaseUIElement from "../../UI/BaseUIElement";
import {LicenseInfo} from "./LicenseInfo";
import {Utils} from "../../Utils";
export interface ProvidedImage {
url: string, key: string, provider: ImageProvider
url: string,
key: string,
provider: ImageProvider
}
export default abstract class ImageProvider {
public abstract readonly defaultKeyPrefixes : string[] = ["mapillary", "image"]
public abstract readonly defaultKeyPrefixes: string[] = ["mapillary", "image"]
private _cache = new Map<string, UIEventSource<LicenseInfo>>()
GetAttributionFor(url: string): UIEventSource<LicenseInfo> {
const cached = this._cache.get(url);
if (cached !== undefined) {
@ -31,43 +34,49 @@ export default abstract class ImageProvider {
*/
public GetRelevantUrls(allTags: UIEventSource<any>, options?: {
prefixes?: string[]
}):UIEventSource<ProvidedImage[]> {
}): UIEventSource<ProvidedImage[]> {
const prefixes = options?.prefixes ?? this.defaultKeyPrefixes
if(prefixes === undefined){
throw "The image provider"+this.constructor.name+" doesn't define `defaultKeyPrefixes`"
if (prefixes === undefined) {
throw "The image provider" + this.constructor.name + " doesn't define `defaultKeyPrefixes`"
}
const relevantUrls = new UIEventSource<{ url: string; key: string; provider: ImageProvider }[]>([])
const seenValues = new Set<string>()
const self = this
const self =this
allTags.addCallbackAndRunD(tags => {
for (const key in tags) {
if(!prefixes.some(prefix => key.startsWith(prefix))){
if (!prefixes.some(prefix => key.startsWith(prefix))) {
continue
}
const value = tags[key]
if(seenValues.has(value)){
continue
}
seenValues.add(value)
this.ExtractUrls(key, value).then(promises => {
for (const promise of promises ?? []) {
if(promise === undefined){
continue
}
promise.then(providedImage => {
if(providedImage === undefined){
return
}
relevantUrls.data.push(providedImage)
relevantUrls.ping()
})
const values = Utils.NoEmpty(tags[key]?.split(";")?.map(v => v.trim()) ?? [])
for (const value of values) {
if (seenValues.has(value)) {
continue
}
})
seenValues.add(value)
this.ExtractUrls(key, value).then(promises => {
console.log("Got ", promises.length, "promises for", value,"by",self.constructor.name)
for (const promise of promises ?? []) {
if (promise === undefined) {
continue
}
promise.then(providedImage => {
if (providedImage === undefined) {
return
}
relevantUrls.data.push(providedImage)
relevantUrls.ping()
})
}
})
}
}
})
return relevantUrls
}
public abstract ExtractUrls(key: string, value: string) : Promise<Promise<ProvidedImage>[]>;
public abstract ExtractUrls(key: string, value: string): Promise<Promise<ProvidedImage>[]>;
}

View file

@ -1,4 +1,5 @@
export class LicenseInfo {
title: string = ""
artist: string = "";
license: string = "";
licenseShortName: string = "";

View file

@ -65,12 +65,26 @@ export class WikimediaImageProvider extends ImageProvider {
"&format=json&origin=*";
const data = await Utils.downloadJson(url)
const licenseInfo = new LicenseInfo();
const license = (data.query.pages[-1].imageinfo ?? [])[0]?.extmetadata;
const pageInfo = data.query.pages[-1]
if(pageInfo === undefined){
return undefined;
}
const license = (pageInfo.imageinfo ?? [])[0]?.extmetadata;
if (license === undefined) {
console.warn("The file", filename ,"has no usable metedata or license attached... Please fix the license info file yourself!")
return undefined;
}
let title = 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)
}
licenseInfo.title = title
licenseInfo.artist = license.Artist?.value;
licenseInfo.license = license.License?.value;
licenseInfo.copyrighted = license.Copyrighted?.value;

View file

@ -47,7 +47,7 @@ export default class Wikidata {
const claimsList: any[] = entity.claims[claimId]
const values = new Set<string>()
for (const claim of claimsList) {
const value = claim.mainsnak?.datavalueq?.value;
const value = claim.mainsnak?.datavalue?.value;
if(value !== undefined){
values.add(value)
}

View file

@ -22,6 +22,10 @@ export default class Wikipedia {
"mw-selflink",
"hatnote" // Often redirects
]
private static readonly idsToRemove = [
"sjabloon_zie"
]
private static readonly _cache = new Map<string, UIEventSource<{ success: string } | { error: any }>>()
@ -59,6 +63,13 @@ export default class Wikipedia {
}
}
for (const forbiddenId of Wikipedia.idsToRemove) {
const toRemove = content.querySelector("#"+forbiddenId)
toRemove?.parentElement?.removeChild(toRemove)
}
const links = Array.from(content.getElementsByTagName("a"))
// Rewrite relative links to absolute links + open them in a new tab