feat: add and run script to get cellxgene collections linked to hca p…

…rojects (#2364) (#2365)

* feat: add and run script to get cellxgene collections linked to hca projects (#2364)

* fix: project url regexp (#2364)

next/package.json

@@ -13,7 +13,8 @@
     "format": "prettier --write .",
     "postbuild": "next-sitemap --config next-sitemap.config.mjs",
     "prepare": "cd .. && husky install next/.husky",
-    "complete-biccn-publications": "esrun ./scripts/complete-biccn-publications.ts"
+    "complete-biccn-publications": "esrun ./scripts/complete-biccn-publications.ts",
+    "get-cellxgene-projects": "esrun ./scripts/get-cellxgene-projects.ts"
   },
   "dependencies": {
     "@clevercanary/data-explorer-ui": "64.0.0",

next/scripts/get-cellxgene-projects.ts

@@ -0,0 +1,42 @@
+import fsp from "fs/promises";
+import got from "got";
+
+interface CellxGeneCollection {
+  collection_id: string;
+  links: { link_url: string }[];
+}
+
+interface CellxGeneProjectMapping {
+  cellxgeneId: string;
+  hcaProjectId: string;
+}
+
+const cellxgeneProjectsFilePath = "./scripts/out/cellxgene-projects.json";
+
+getCellxGeneProjects();
+
+async function getCellxGeneProjects(): Promise<void> {
+  const collections = JSON.parse(
+    (await got("https://api.cellxgene.cziscience.com/curation/v1/collections"))
+      .body
+  ) as CellxGeneCollection[];
+  const cellxgeneProjects: CellxGeneProjectMapping[] = [];
+  for (const collection of collections) {
+    for (const { link_url } of collection.links) {
+      const match =
+        /^https:\/\/(?:explore\.)?data\.humancellatlas\.org(?:\/explore)?\/projects\/([^/?#]+)/.exec(
+          link_url
+        );
+      if (match)
+        cellxgeneProjects.push({
+          cellxgeneId: collection.collection_id,
+          hcaProjectId: match[1],
+        });
+    }
+  }
+  await fsp.writeFile(
+    cellxgeneProjectsFilePath,
+    JSON.stringify(cellxgeneProjects, undefined, 2)
+  );
+  console.log(`Saved to ${cellxgeneProjectsFilePath}`);
+}