@ca-plant-list/ca-plant-list 0.4.8 → 0.4.10

package/lib/inat_photo.js

@@ -1,10 +1,4 @@
-import {
-    CC0,
-    CC_BY,
-    CC_BY_NC,
-    COPYRIGHT,
-    Photo,
-} from "./photo.js";
+import { CC0, CC_BY, CC_BY_NC, COPYRIGHT, Photo } from "./photo.js";
 
 class InatPhoto extends Photo {
     /** @type {number} */
@@ -18,17 +12,25 @@ class InatPhoto extends Photo {
      * @param {InatLicenseCode} licenseCode
      * @param {string} attrName
      */
-    constructor( id, ext, licenseCode, attrName ) {
+    constructor(id, ext, licenseCode, attrName) {
         /** @type {typeof COPYRIGHT | typeof CC_BY | typeof CC_BY_NC | typeof CC0} */
         let rights = COPYRIGHT;
-        if ( licenseCode === "cc0" ) rights = CC0;
-        else if ( licenseCode === "cc-by" ) rights = CC_BY;
-        else if ( licenseCode === "cc-by-nc" ) rights = CC_BY_NC;
-        super( null, attrName, rights );
+        if (licenseCode === "cc0") rights = CC0;
+        else if (licenseCode === "cc-by") rights = CC_BY;
+        else if (licenseCode === "cc-by-nc") rights = CC_BY_NC;
+        super(null, attrName, rights);
         this.inatPhotoId = id;
         this.ext = ext;
     }
 
+    getExt() {
+        return this.ext;
+    }
+
+    getId() {
+        return this.inatPhotoId;
+    }
+
     getUrl() {
         return `https://inaturalist-open-data.s3.amazonaws.com/photos/${this.inatPhotoId}/medium.${this.ext}`;
     }
@@ -38,6 +40,4 @@ class InatPhoto extends Photo {
     }
 }
 
-export {
-    InatPhoto
-};
+export { InatPhoto };

package/lib/progressmeter.js

@@ -0,0 +1,29 @@
+import cliProgress from "cli-progress";
+
+export class ProgressMeter {
+    #meter;
+
+    /**
+     * @param {string} label
+     * @param {number} total
+     */
+    constructor(label, total) {
+        this.#meter = new cliProgress.SingleBar({
+            format: `${label} {percentage}% | {value}/{total}{custom}`,
+            hideCursor: true,
+        });
+        this.#meter.start(total, 0, { custom: "" });
+    }
+
+    stop() {
+        this.#meter.stop();
+    }
+
+    /**
+     * @param {number} current
+     * @param {{ custom: string; }} [custom={ custom: "" }]
+     */
+    update(current, custom = { custom: "" }) {
+        this.#meter.update(current, custom);
+    }
+}

package/lib/taxa.js

@@ -48,7 +48,7 @@ class Taxa {
         taxonFactory = (td, g) => new Taxon(td, g),
         extraTaxa = [],
         extraSynonyms = [],
-        includePhotos = true
+        includePhotos = true,
     ) {
         this.#isSubset = inclusionList !== true;
 
@@ -68,7 +68,7 @@ class Taxa {
             extraTaxa,
             inclusionList,
             taxonFactory,
-            showFlowerErrors
+            showFlowerErrors,
         );
 
         // Make sure everything in the inclusionList has been loaded.
@@ -79,12 +79,11 @@ class Taxa {
         }
 
         this.#sortedTaxa = Object.values(this.#taxa).sort((a, b) =>
-            a.getName().localeCompare(b.getName())
+            a.getName().localeCompare(b.getName()),
         );
 
-
-        if ( includePhotos ) {
-            this.#loadInatPhotos( dataDir );
+        if (includePhotos) {
+            this.#loadInatPhotos(dataDir);
         }
 
         const synCSV = CSV.parseFile(dataDir, "synonyms.csv");
@@ -95,22 +94,24 @@ class Taxa {
     /**
      * @param {string} dataDir
      */
-    #loadInatPhotos( dataDir ) {
+    #loadInatPhotos(dataDir) {
         const photosFileName = "inattaxonphotos.csv";
-        if ( fs.existsSync( path.join( dataDir, photosFileName ) ) ) {
+        if (fs.existsSync(path.join(dataDir, photosFileName))) {
             /** @type {InatCsvPhoto[]} */
-            const csvPhotos = CSV.parseFile( dataDir, photosFileName );
-            for ( const csvPhoto of csvPhotos ) {
+            const csvPhotos = CSV.parseFile(dataDir, photosFileName);
+            for (const csvPhoto of csvPhotos) {
                 const taxon = this.getTaxon(csvPhoto.name);
-                if(!taxon) {
+                if (!taxon) {
                     continue;
                 }
-                taxon.addPhoto(new InatPhoto(
-                    csvPhoto.id,
-                    csvPhoto.ext,
-                    csvPhoto.licenseCode,
-                    csvPhoto.attrName
-                ) );
+                taxon.addPhoto(
+                    new InatPhoto(
+                        csvPhoto.id,
+                        csvPhoto.ext,
+                        csvPhoto.licenseCode,
+                        csvPhoto.attrName,
+                    ),
+                );
             }
         }
     }
@@ -121,7 +122,7 @@ class Taxa {
 
     getFlowerColors() {
         return Object.values(this.#flower_colors).filter(
-            (fc) => fc.getTaxa().length > 0
+            (fc) => fc.getTaxa().length > 0,
         );
     }
 
@@ -209,10 +210,10 @@ class Taxa {
                     const color = this.#flower_colors[colorName];
                     if (!color) {
                         throw new Error(
-                            "flower color \"" +
+                            'flower color "' +
                                 colorName +
-                                "\" not found for " +
-                                name
+                                '" not found for ' +
+                                name,
                         );
                     }
                     color.addTaxon(taxon);
@@ -229,7 +230,7 @@ class Taxa {
                     ) {
                         this.#errorLog.log(
                             name,
-                            "does not have all flower info"
+                            "does not have all flower info",
                         );
                     }
                 } else {

package/lib/util.js

@@ -3,8 +3,9 @@
  * https://github.com/you-dont-need/You-Dont-Need-Lodash-Underscore?tab=readme-ov-file#_chunk
  * @param {any[]} input
  * @param {number} size
+ * @returns {any[][]}
  */
-export function chunk( input, size ) {
+export function chunk(input, size) {
     return input.reduce((arr, item, idx) => {
         return idx % size === 0
             ? [...arr, [item]]
@@ -15,6 +16,6 @@ export function chunk( input, size ) {
 /**
  * @param {number} time
  */
-export async function sleep( time ) {
-    return new Promise( resolve => setTimeout( resolve, time ) );
+export async function sleep(time) {
+    return new Promise((resolve) => setTimeout(resolve, time));
 }

package/lib/utils/inat-tools.js

@@ -0,0 +1,90 @@
+import { ProgressMeter } from "../progressmeter.js";
+import { chunk, sleep } from "../util.js";
+
+const ALLOWED_LICENSE_CODES = ["cc0", "cc-by", "cc-by-nc"];
+
+/**
+ * @param {Taxon[]} taxa
+ * @return {Promise<InatApiTaxon[]>}
+ */
+async function fetchInatTaxa(taxa) {
+    const inatTaxonIDs = taxa.map((taxon) => taxon.getINatID()).filter(Boolean);
+    const url = `https://api.inaturalist.org/v2/taxa/${inatTaxonIDs.join(",")}?fields=(taxon_photos:(photo:(medium_url:!t,attribution:!t,license_code:!t)))`;
+    const resp = await fetch(url);
+    if (!resp.ok) {
+        const error = await resp.text();
+        throw new Error(`Failed to fetch taxa from iNat: ${error}`);
+    }
+    const json = await resp.json();
+    return json.results;
+}
+
+/**
+ * @param {Taxon[]} taxaToUpdate
+ * @returns {Promise<Map<string,InatPhotoInfo[]>>}
+ */
+export async function getTaxonPhotos(taxaToUpdate) {
+    /** @type {Map<string,string>} */
+    const idMap = new Map();
+
+    for (const taxon of taxaToUpdate) {
+        if (taxon.getINatID()) {
+            idMap.set(taxon.getINatID(), taxon.getName());
+        }
+    }
+
+    /** @type {Map<string,InatPhotoInfo[]>} */
+    const photos = new Map();
+
+    const meter = new ProgressMeter("retrieving taxa", taxaToUpdate.length);
+    let taxaRetrieved = 0;
+
+    for (const batch of chunk(taxaToUpdate, 30)) {
+        const inatTaxa = await fetchInatTaxa(batch);
+        for (const iNatTaxon of inatTaxa) {
+            const iNatTaxonPhotos = iNatTaxon.taxon_photos
+                .filter((tp) =>
+                    ALLOWED_LICENSE_CODES.includes(tp.photo.license_code),
+                )
+                .slice(0, 5);
+
+            const taxonName = idMap.get(iNatTaxon.id.toString());
+            if (!taxonName) {
+                throw new Error(`iNat id ${iNatTaxon.id} not found`);
+            }
+            /** @type {InatPhotoInfo[]} */
+            const taxonPhotos = [];
+            for (const taxonPhoto of iNatTaxonPhotos) {
+                const ext = taxonPhoto.photo.medium_url.split(".").at(-1);
+                if (!ext) {
+                    continue;
+                }
+                /** @type {InatPhotoInfo} */
+                const obj = {
+                    id: taxonPhoto.photo.id.toString(),
+                    ext: ext,
+                    licenseCode: taxonPhoto.photo.license_code,
+                    attrName:
+                        // Photographers retain copyright for most CC licenses,
+                        // except CC0, so attribution is a bit different
+                        taxonPhoto.photo.attribution.match(
+                            /\(c\) (.*?),/,
+                        )?.[1] ||
+                        taxonPhoto.photo.attribution.match(
+                            /uploaded by (.*)/,
+                        )?.[1],
+                };
+
+                taxonPhotos.push(obj);
+            }
+            photos.set(taxonName, taxonPhotos);
+        }
+        taxaRetrieved += batch.length;
+        meter.update(taxaRetrieved);
+        await sleep(1_100);
+    }
+
+    meter.stop();
+
+    return photos;
+}