Create tool for retrieving object children. (#88)

- Includes indexing adjustments for more accurate hierarchical representation.

Author: demiankatz

api/src/routes/edit.ts

@@ -2,9 +2,32 @@ import express = require("express");
 import Config from "../models/Config";
 const edit = express.Router();
 import { requireToken } from "./auth";
+import { pidSanitizer } from "./sanitize";
+import Solr from "../services/Solr";
 
 edit.get("/models", requireToken, function (req, res) {
     res.json({ CollectionModels: Config.getInstance().collectionModels, DataModels: Config.getInstance().dataModels });
 });
 
+async function getChildren(req, res) {
+    const query =
+        (req.params.pid ?? "").length > 0
+            ? `fedora_parent_id_str_mv:"${req.params.pid.replace('"', "")}"`
+            : "-fedora_parent_id_str_mv:*";
+    const config = Config.getInstance();
+    const solr = Solr.getInstance();
+    const rows = parseInt(req.query.rows ?? "100000").toString();
+    const start = parseInt(req.query.start ?? "0").toString();
+    const result = await solr.query(config.solrCore, query, { fl: "id,title", rows, start });
+    if (result.statusCode !== 200) {
+        res.status(result.statusCode ?? 500).send("Unexpected Solr response code.");
+        return;
+    }
+    const response = result?.body?.response ?? { numFound: 0, start: 0, docs: [] };
+    res.json(response);
+}
+
+edit.get("/object/children", requireToken, getChildren);
+edit.get("/object/children/:pid", requireToken, pidSanitizer, getChildren);
+
 export default edit;

api/src/routes/messenger.ts

@@ -4,11 +4,9 @@ import express = require("express");
 import Config from "../models/Config";
 import { Queue } from "bullmq";
 import { requireToken } from "./auth";
-import { sanitizeParameters } from "./sanitize";
+import { pidSanitizer } from "./sanitize";
 const messenger = express.Router();
 
-const pidSanitizer = sanitizeParameters({ pid: /^[a-zA-Z]+:[0-9]+/ }, /^$/);
-
 messenger.post("/pdfgenerator/:pid", pidSanitizer, requireToken, async function (req, res) {
     const q = new Queue("vudl");
     await q.add("generatepdf", { pid: req.params.pid });

api/src/routes/sanitize.ts

@@ -14,3 +14,5 @@ export function sanitizeParameters(customRules = {}, defaultRule = /^[-.a-zA-Z0-
         next();
     };
 }
+
+export const pidSanitizer = sanitizeParameters({ pid: /^[a-zA-Z]+:[0-9]+/ }, /^$/);

api/src/services/HierarchyCollector.ts

@@ -81,11 +81,9 @@ class HierarchyCollector {
         // Create promises to retrieve parents asynchronously...
         const promises = (result.relations.isMemberOf ?? []).map(async (resource) => {
             const parentPid = resource.substr("info:fedora/".length);
-            if (!this.config.topLevelPids.includes(parentPid)) {
-                // The "false" here skips RDF retrieval:
-                const parent = await this.getHierarchy(parentPid, false);
-                result.addParent(parent);
-            }
+            // The "false" here skips RDF retrieval:
+            const parent = await this.getHierarchy(parentPid, false);
+            result.addParent(parent);
         });
         // Now wait for the promises to complete before we return results, so
         // nothing happens out of order.

api/src/services/Solr.ts

@@ -32,6 +32,18 @@ class Solr {
         return http(method, url, data, options);
     }
 
+    public async query(
+        core: string,
+        solrQuery: string,
+        queryParams: Record<string, string> = {}
+    ): Promise<NeedleResponse> {
+        const formatter = ([key, val]) => {
+            return "&" + encodeURIComponent(key) + "=" + encodeURIComponent(val);
+        };
+        const extras = Object.entries(queryParams).map(formatter).join("");
+        return this._request("get", core + "/select?q=" + encodeURIComponent(solrQuery) + extras);
+    }
+
     public async deleteRecord(core: string, pid: string): Promise<NeedleResponse> {
         // Strip double quotes from PID -- they should never be present, and it protects
         // against malicious query manipulation.

api/src/services/SolrIndexer.ts

@@ -106,19 +106,26 @@ class SolrIndexer {
             fields[dynamic_sequence_field_name] = this.padNumber(seqNum);
         }
 
-        // Process parent data:
+        // Process parent data (note that hierarchyParents makes some special exceptions for VuFind;
+        // fedoraParents exactly maintains the hierarchy as represented in Fedora):
         const hierarchyParents: Array<FedoraData> = [];
+        const fedoraParents: Array<FedoraData> = [];
         const hierarchySequences: Array<string> = [];
         for (const parent of fedoraData.parents) {
-            // If the object is a Data, the parentPID is the Resource it belongs
+            // Fedora parents should directly reflect the repository without any
+            // VuFind-specific filtering or adjustments:
+            fedoraParents.push(parent);
+
+            // If the object is a Data, the VuFind parentPID is the Resource it belongs
             // to (skip the List object):
             if (fedoraData.models.includes("vudl-system:DataModel")) {
                 for (const grandParent of parent.parents) {
                     hierarchyParents.push(grandParent);
                     hierarchySequences.push(this.padNumber(sequenceIndex[grandParent.pid] ?? 0));
                 }
-            } else {
-                // ...else it is the immediate parent (Folder most likely):
+            } else if (!this.config.topLevelPids.includes(pid)) {
+                // ...for non-Data objects, store the immediate parent (Folder most likely)
+                // as long as the current pid is not marked as a top-level one:
                 hierarchyParents.push(parent);
                 hierarchySequences.push(this.padNumber(sequenceIndex[parent.pid] ?? 0));
             }
@@ -134,6 +141,7 @@ class SolrIndexer {
                 }
             }
         }
+        fields.fedora_parent_id_str_mv = fedoraParents.map((parent) => parent.pid);
         if (hierarchyParents.length > 0) {
             // This is what we are collapsing on:
             fields.hierarchy_first_parent_id_str = fedoraData.models.includes("vudl-system:DataModel")