From a4d17c1395c5942e96e980fdda1f1489874f59d2 Mon Sep 17 00:00:00 2001
From: Simon Legner <Simon.Legner@gmail.com>
Date: Sat, 17 Aug 2024 20:05:33 +0200
Subject: [PATCH] feat(ltData): fetch and extract structured data on Commons

---
 app/api/ltData.ts      | 46 ++++++++++++++++++++++++---
 app/model/index.ts     | 11 +++++++
 app/model/mediainfo.ts | 72 ++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 124 insertions(+), 5 deletions(-)
 create mode 100644 app/model/mediainfo.ts

diff --git a/app/api/ltData.ts b/app/api/ltData.ts
index 70bc165..17e6321 100644
--- a/app/api/ltData.ts
+++ b/app/api/ltData.ts
@@ -1,7 +1,8 @@
 import deepmerge from 'deepmerge';
 import getFilePath from 'wikimedia-commons-file-path';
 
-import {CommonsFile, CommonsTitle, LatLng} from '../model';
+import {CommonsFile, CommonsTitle, LatLng, WikidataProperty} from '../model';
+import {MediaInfo, Statement} from '../model/mediainfo.ts';
 
 export const API_URL = 'https://commons.wikimedia.org/w/api.php';
 const NS_FILE = 6;
@@ -83,6 +84,7 @@ interface Revision {
 
 export interface Slots {
   main: MainSlot;
+  mediainfo: MainSlot;
 }
 
 export interface MainSlot {
@@ -148,12 +150,13 @@ export interface FileDetails {
   author: string;
   timestamp: string;
   url?: string;
-  objectLocation: LatLng;
+  coordinates?: LatLng;
+  objectLocation?: LatLng;
 }
 
 export async function getFileDetails(
   pageid: number,
-  prop = 'categories|imageinfo|revisions',
+  prop = 'categories|imageinfo|revisions|wbentityusage',
   iiprop = 'url|extmetadata'
 ): Promise<FileDetails> {
   const params = {
@@ -163,7 +166,7 @@ export async function getFileDetails(
     iiextmetadatafilter: 'ImageDescription|Artist|DateTimeOriginal',
     iiextmetadatalanguage: document.body.parentElement!.lang,
     ...(prop.includes('categories') ? {clshow: '!hidden'} : {}),
-    ...(prop.includes('revisions') ? {rvslots: 'main', rvprop: 'content'} : {})
+    ...(prop.includes('revisions') ? {rvslots: '*', rvprop: 'content'} : {})
   };
   const data = await $query<ApiResponse<DetailsPage>>(params);
   const page: DetailsPage | undefined = data?.query?.pages?.[pageid];
@@ -177,9 +180,42 @@ export async function getFileDetails(
     author: extmetadata?.Artist?.value,
     timestamp: extmetadata?.DateTimeOriginal?.value,
     ...(iiprop.includes('url') ? {url: page?.imageinfo[0]?.descriptionurl} : {}),
-    objectLocation: extractObjectLocation(page)
+    objectLocation: extractObjectLocation(page),
+    ...extractMediaInfo(page)
   };
 
+  function extractMediaInfo(page: DetailsPage): Partial<FileDetails> {
+    const json = page?.revisions?.[0]?.slots?.mediainfo['*'];
+    if (!json) return {};
+    const mediainfo: MediaInfo = JSON.parse(json);
+    const coordinates = extractMediaInfoLocation(
+      'Location',
+      mediainfo.statements[WikidataProperty['Location']]?.[0]
+    );
+    const objectLocation = extractMediaInfoLocation(
+      'Object location',
+      mediainfo.statements[WikidataProperty['Object location']]?.[0]
+    );
+    return {
+      ...(coordinates ? {coordinates} : {}),
+      ...(objectLocation ? {objectLocation} : {})
+    };
+  }
+
+  function extractMediaInfoLocation(
+    type: LatLng['type'],
+    statement: Statement | undefined
+  ): LatLng | undefined {
+    if (statement?.mainsnak.datavalue?.type !== 'globecoordinate') {
+      return;
+    }
+    return new LatLng(
+      type,
+      statement?.mainsnak.datavalue.value.latitude,
+      statement?.mainsnak.datavalue.value.longitude
+    );
+  }
+
   function extractObjectLocation(page: DetailsPage) {
     try {
       const wikitext: string = page?.revisions?.[0]?.slots?.main['*'] || '';
diff --git a/app/model/index.ts b/app/model/index.ts
index 614fc59..8e3cfa3 100644
--- a/app/model/index.ts
+++ b/app/model/index.ts
@@ -1,4 +1,5 @@
 import {LatLng} from './LatLng';
+
 export {LatLng} from './LatLng';
 
 export interface CommonsFile {
@@ -7,9 +8,19 @@ export interface CommonsFile {
   url: string;
   coordinates: LatLng;
   objectLocation: LatLng;
+
   imageUrl(width?: number): string;
 }
 
 export type CommonsTitle = string;
 
 export type User = string;
+
+export const WikidataProperty = {
+  // coordinate location (P625)
+  '*': 'P625',
+  // coordinates of the point of view (P1259)
+  Location: 'P1259',
+  // coordinates of depicted place (P9149)
+  'Object location': 'P9149'
+};
diff --git a/app/model/mediainfo.ts b/app/model/mediainfo.ts
new file mode 100644
index 0000000..09d0dc5
--- /dev/null
+++ b/app/model/mediainfo.ts
@@ -0,0 +1,72 @@
+export interface MediaInfo {
+  type: 'mediainfo';
+  id: string;
+  labels: unknown[];
+  descriptions: unknown[];
+  statements: Record<string, Statement[]>;
+}
+
+export interface Statement {
+  type: 'statement';
+  mainsnak: Mainsnak;
+  id: string;
+  rank: string;
+  qualifiers: Record<string, Mainsnak[]>;
+}
+
+export interface Mainsnak {
+  snaktype: 'value' | 'somevalue';
+  property: string;
+  hash: string;
+  datavalue?: Datavalue;
+}
+
+export type Datavalue =
+  | {
+      type: 'string';
+      value: string;
+    }
+  | {
+      type: 'wikibase-entityid';
+      value: EntityID;
+    }
+  | {
+      type: 'globecoordinate';
+      value: GlobeCoordinate;
+    }
+  | {
+      type: 'quantity';
+      value: Quanity;
+    }
+  | {
+      type: 'time';
+      value: TimeValue;
+    };
+
+export interface EntityID {
+  'entity-type': string;
+  'numeric-id': number;
+  id: string;
+}
+
+export interface Quanity {
+  amount: string;
+  unit: string;
+}
+
+export interface TimeValue {
+  time: string;
+  timezone: number;
+  before: number;
+  after: number;
+  precision: number;
+  calendarmodel: string;
+}
+
+export interface GlobeCoordinate {
+  latitude: number;
+  longitude: number;
+  altitude: number | null;
+  precision: number;
+  globe: 'http://www.wikidata.org/entity/Q2';
+}