From f8965637974544d6b53bfc78892e8e38378ae954 Mon Sep 17 00:00:00 2001
From: Soazic Guillaume Bourdat <guillaume.bourdat@univ-grenoble-alpes.fr>
Date: Mon, 26 Jan 2026 10:57:59 +0100
Subject: [PATCH] feat: add get all ids on HAL see: #515

---
 api/src/core/adapters/hal/HalAPI/index.ts | 19 +++++++++++++++++++
 api/src/core/usecases/importFromSource.ts |  2 +-
 2 files changed, 20 insertions(+), 1 deletion(-)

diff --git a/api/src/core/adapters/hal/HalAPI/index.ts b/api/src/core/adapters/hal/HalAPI/index.ts
index 12ee993f0..29bc340eb 100644
--- a/api/src/core/adapters/hal/HalAPI/index.ts
+++ b/api/src/core/adapters/hal/HalAPI/index.ts
@@ -8,6 +8,8 @@ import { HAL } from "./types/HAL";
 
 const HAL_API_TIMEOUT = 60000;
 
+type HALID = Pick<HAL.API.Software, "docid">;
+
 export type HALAPIGateway = {
     software: {
         getById: (halDocid: string) => Promise<HAL.API.Software | undefined>;
@@ -15,6 +17,7 @@ export type HALAPIGateway = {
             queryString?: string | undefined;
             SWHFilter?: boolean | undefined;
         }) => Promise<HAL.API.Software[]>;
+        getAllIds: (params: { queryString?: string | undefined; SWHFilter?: boolean | undefined }) => Promise<HALID[]>;
         getCodemetaByUrl: (urlSoftwareDoc: string) => Promise<HAL.SoftwareApplication | undefined>;
     };
     domain: {
@@ -115,6 +118,22 @@ export const makeHalAPIGateway = (source?: Source): HALAPIGateway => {
                 const json = await getHalApiRequest<HAL.API.Software>(url);
                 return json?.response?.docs ?? [];
             },
+            getAllIds: async (params: { queryString?: string; SWHFilter?: boolean }) => {
+                const { queryString, SWHFilter } = params;
+                let url = `https://api.archives-ouvertes.fr/search/?fq=docType_s:SOFTWARE&rows=10000&fl:docid`;
+
+                if (queryString) {
+                    url = url + `&q=${encodeURIComponent(queryString)}`;
+                }
+
+                // Filter only software who have an swhidId to filter clean data on https://hal.science, TODO remove and set it as an option to be generic
+                if (SWHFilter) {
+                    url = url + `&fq=swhidId_s:["" TO *]`;
+                }
+
+                const json = await getHalApiRequest<HAL.API.Software>(url);
+                return json?.response?.docs ?? [];
+            },
             getCodemetaByUrl: (urlSoftwareDoc: string) => {
                 const url = `${urlSoftwareDoc}/codemeta`;
                 return getHalRequest<HAL.SoftwareApplication>(url);
diff --git a/api/src/core/usecases/importFromSource.ts b/api/src/core/usecases/importFromSource.ts
index 794cabe51..76b713fd5 100644
--- a/api/src/core/usecases/importFromSource.ts
+++ b/api/src/core/usecases/importFromSource.ts
@@ -63,7 +63,7 @@ const resolveAllIdsAccordingToSource = async (source: Source): Promise<string[]>
     switch (source.kind) {
         case "HAL":
             const halAPIGateway = makeHalAPIGateway(source);
-            return (await halAPIGateway.software.getAll({ SWHFilter: true })).map(doc => doc.docid);
+            return (await halAPIGateway.software.getAllIds({ SWHFilter: true })).map(doc => doc.docid);
         case "Zenodo":
             const zenodoAPI = makeZenodoApi(source);
             return (await zenodoAPI.records.getAllSoftware()).hits.hits.map(item => item.id.toString());