feat: Parse URLs from policyRef

package.json

@@ -50,6 +50,7 @@
     "@types/geojson": "^7946.0.12",
     "ajv": "^8.12.0",
     "ajv-formats": "^2.1.1",
+    "cheerio": "1.0.0-rc.12",
     "copyfiles": "^2.4.1",
     "docx": "^8.2.4",
     "eslint": "^8.51.0",
@@ -67,10 +68,10 @@
     "lodash.omit": "^4.5.0",
     "lodash.set": "^4.3.2",
     "lodash.startcase": "^4.4.0",
+    "marked": "^9.1.5",
     "prettier": "^3.0.3",
     "react": "^18.2.0",
     "react-dom": "^18.2.0",
-    "striptags": "^3.2.0",
     "type-fest": "^4.4.0",
     "uuid": "^9.0.1",
     "zod": "^3.22.4"
@@ -89,12 +90,12 @@
     "@types/lodash.set": "^4.3.7",
     "@types/lodash.startcase": "^4.4.8",
     "@types/node": "^18.16.19",
+    "@types/react": "^18.2.17",
     "@types/react-dom": "^18.2.14",
     "@types/uuid": "^9.0.6",
-    "esbuild": "^0.19.5",
-    "@types/react": "^18.2.17",
     "@typescript-eslint/eslint-plugin": "^6.2.0",
     "@typescript-eslint/parser": "^6.8.0",
+    "esbuild": "^0.19.5",
     "esbuild-jest": "^0.5.0",
     "eslint": "^8.46.0",
     "eslint-plugin-simple-import-sort": "^10.0.0",

src/export/bops/index.ts

@@ -1,6 +1,7 @@
+import { load } from "cheerio";
 import isEmpty from "lodash.isempty";
 import isNil from "lodash.isnil";
-import striptags from "striptags";
+import { marked } from "marked";
 
 import { Passport } from "../../models/passport";
 import { getResultData } from "../../models/result";
@@ -264,10 +265,7 @@ export function formatProposalDetails({
       }
       if (crumb.autoAnswered) metadata.auto_answered = true;
       if (node.data?.policyRef) {
-        metadata.policy_refs = [
-          // remove html tags
-          { text: striptags(node.data?.policyRef as string) },
-        ];
+        metadata.policy_refs = parsePolicyRefs(node.data.policyRefs as string);
       }
       return metadata;
     })();
@@ -286,6 +284,17 @@ export function formatProposalDetails({
   };
 }
 
+export const parsePolicyRefs = (markdownOrHTML: string) => {
+  const htmlString = marked.parse(markdownOrHTML);
+  const $ = load(htmlString);
+  const policyRefs = $("a").map((_index, el) => ({
+    text: $(el).prop("textContent")?.trim() ?? undefined,
+    url: $(el).attr("href"),
+  }));
+
+  return policyRefs.toArray();
+};
+
 export function computeBOPSParams({
   sessionId,
   flow,