#202 : save identifiers and expose it on API

codegouvfr · Feb 19, 2025 · f19cb4c · f19cb4c
1 parent 6fe0d97
commit f19cb4c
Show file tree

Hide file tree

Showing 12 changed files with 118 additions and 3 deletions.
diff --git a/api/src/core/adapters/dbApi/kysely/createPgSoftwareExternalDataRepository.ts b/api/src/core/adapters/dbApi/kysely/createPgSoftwareExternalDataRepository.ts
@@ -12,6 +12,7 @@ export const createPgSoftwareExternalDataRepository = (db: Kysely<Database>): So
             applicationCategories: JSON.stringify(softwareExternalData.applicationCategories),
             programmingLanguages: JSON.stringify(softwareExternalData.programmingLanguages),
             referencePublications: JSON.stringify(softwareExternalData.referencePublications),
+            identifiers: JSON.stringify(softwareExternalData.identifiers),
             description: JSON.stringify(softwareExternalData.description)
         };
 

diff --git a/api/src/core/adapters/dbApi/kysely/createPgSoftwareRepository.ts b/api/src/core/adapters/dbApi/kysely/createPgSoftwareRepository.ts
@@ -194,6 +194,7 @@ export const createPgSoftwareRepository = (db: Kysely<Database>): SoftwareReposi
                         keywords: software?.keywords ?? softwareExternalData?.keywords ?? [],
                         programmingLanguages: softwareExternalData?.programmingLanguages ?? [],
                         referencePublications: softwareExternalData?.referencePublications,
+                        identifiers: softwareExternalData?.identifiers,
                         applicationCategories: software.categories.concat(
                             softwareExternalData?.applicationCategories ?? []
                         ),
@@ -298,7 +299,8 @@ export const createPgSoftwareRepository = (db: Kysely<Database>): SoftwareReposi
                             ),
                             categories: undefined, // merged in applicationCategories, set to undefined to remove it
                             programmingLanguages: softwareExternalData?.programmingLanguages ?? [],
-                            referencePublications: softwareExternalData?.referencePublications
+                            referencePublications: softwareExternalData?.referencePublications,
+                            identifiers: softwareExternalData?.identifiers
                         });
                     }
                 );
@@ -434,6 +436,7 @@ const makeGetSoftwareBuilder = (db: Kysely<Database>) =>
                     programmingLanguages: ref("ext.programmingLanguages"),
                     applicationCategories: ref("ext.applicationCategories"),
                     referencePublications: ref("ext.referencePublications"),
+                    identifiers: ref("ext.identifiers"),
                     keywords: ref("ext.keywords"),
                     softwareVersion: ref("ext.softwareVersion"),
                     publicationTime: ref("ext.publicationTime")
@@ -569,6 +572,7 @@ const makeGetSoftwareById =
                     parentWikidataSoftware: parentExternalData,
                     programmingLanguages: softwareExternalData?.programmingLanguages ?? [],
                     referencePublications: softwareExternalData?.referencePublications,
+                    identifiers: softwareExternalData?.identifiers,
                     applicationCategories: filterDuplicate(
                         software.categories.concat(softwareExternalData?.applicationCategories ?? [])
                     ),

diff --git a/api/src/core/adapters/dbApi/kysely/kysely.database.ts b/api/src/core/adapters/dbApi/kysely/kysely.database.ts
@@ -97,6 +97,7 @@ type SoftwareExternalDatasTable = {
     applicationCategories: JSONColumnType<string[]> | null;
     referencePublications: JSONColumnType<SILL.ScholarlyArticle[]> | null;
     publicationTime: Date | null;
+    identifiers: JSONColumnType<SILL.Identification[]> | null;
 };
 
 type SoftwareType =

diff --git a/api/src/core/adapters/dbApi/kysely/migrations/1738152406578_add-identifiers.ts b/api/src/core/adapters/dbApi/kysely/migrations/1738152406578_add-identifiers.ts
@@ -0,0 +1,9 @@
+import { Kysely } from "kysely";
+
+export async function up(db: Kysely<any>): Promise<void> {
+    await db.schema.alterTable("software_external_datas").addColumn("identifiers", "jsonb").execute();
+}
+
+export async function down(db: Kysely<any>): Promise<void> {
+    await db.schema.alterTable("software_external_datas").dropColumn("identifiers").execute();
+}
diff --git a/api/src/core/adapters/dbApi/kysely/pgDbApi.integration.test.ts b/api/src/core/adapters/dbApi/kysely/pgDbApi.integration.test.ts
@@ -55,6 +55,7 @@ const softwareExternalData: SoftwareExternalData = {
     programmingLanguages: ["C++"],
     applicationCategories: ["Software Cat I", "Software Cat II"],
     referencePublications: undefined,
+    identifiers: undefined,
     publicationTime: new Date(1561566581000)
 };
 
@@ -83,6 +84,7 @@ const similarSoftwareExternalData: SoftwareExternalData = {
     programmingLanguages: ["Python3"],
     applicationCategories: ["Software Cat I", "Software Cat II"],
     referencePublications: undefined,
+    identifiers: undefined,
     publicationTime: new Date(1561566581000)
 };
 
@@ -446,6 +448,7 @@ describe("pgDbApi", () => {
                     keywords: JSON.stringify(softExtData.keywords),
                     applicationCategories: JSON.stringify(softExtData.applicationCategories),
                     programmingLanguages: JSON.stringify(softExtData.programmingLanguages),
+                    identifiers: JSON.stringify(softExtData.identifiers),
                     referencePublications: JSON.stringify(softExtData.referencePublications)
                 }))
             )

diff --git a/api/src/core/adapters/fetchExternalData.test.ts b/api/src/core/adapters/fetchExternalData.test.ts
@@ -195,6 +195,7 @@ describe("fetches software extra data (from different providers)", () => {
                     websiteUrl: "https://create-react-app.dev/",
                     programmingLanguages: [],
                     referencePublications: null,
+                    identifiers: null,
                     softwareVersion: "5.0.1",
                     publicationTime: new Date("2022-04-12T00:00:00.000Z")
                 },
@@ -224,6 +225,7 @@ describe("fetches software extra data (from different providers)", () => {
                     websiteUrl: "https://vitejs.dev/",
                     programmingLanguages: ["JavaScript"],
                     referencePublications: null,
+                    identifiers: null,
                     softwareVersion: expect.any(String),
                     publicationTime: expect.any(Date)
                 }
@@ -279,6 +281,7 @@ describe("fetches software extra data (from different providers)", () => {
                     sourceUrl: "https://github.com/apache/httpd",
                     websiteUrl: "https://httpd.apache.org/",
                     referencePublications: null,
+                    identifiers: null,
                     programmingLanguages: ["C"],
                     softwareVersion: "2.5.0-alpha",
                     publicationTime: new Date("2017-11-08T00:00:00.000Z")

diff --git a/api/src/core/adapters/hal/getHalSoftware.test.ts b/api/src/core/adapters/hal/getHalSoftware.test.ts
@@ -39,6 +39,33 @@ describe("HAL", () => {
                 "programmingLanguages": undefined,
                 "applicationCategories": ["Computer Science [cs]"],
                 "referencePublications": undefined,
+                "identifiers": [
+                    {
+                        "@type": "PropertyValue",
+                        "subjectOf": {
+                            "@type": "Website",
+                            "additionalType": "HAL",
+                            "name": "HAL instance",
+                            "url": new URL("https://hal.science/")
+                        },
+                        "url": new URL("https://hal.science/hal-01715545"),
+                        "value": "hal-01715545"
+                    },
+                    {
+                        "@type": "PropertyValue",
+                        "subjectOf": {
+                            "@type": "Website",
+                            "additionalType": "SWH",
+                            "name": "Software Heritage instance",
+                            "url": new URL("https://www.softwareheritage.org/")
+                        },
+                        "url": new URL(
+                            "https://archive.softwareheritage.org/swh:1:dir:424f2533fe51aa8a49d891f8413dd089995cc851;origin=https://hal.archives-ouvertes.fr/hal-01715545;visit=swh:1:snp:9f3237e88d818d975a63da2d5e04d9ad38b42581;anchor=swh:1:rev:8b71800feca2e28cc0f7f78d248e49244b554875;path=/"
+                        ),
+                        "value":
+                            "swh:1:dir:424f2533fe51aa8a49d891f8413dd089995cc851;origin=https://hal.archives-ouvertes.fr/hal-01715545;visit=swh:1:snp:9f3237e88d818d975a63da2d5e04d9ad38b42581;anchor=swh:1:rev:8b71800feca2e28cc0f7f78d248e49244b554875;path=/"
+                    }
+                ],
                 "publicationTime": new Date(1521545908000)
             });
         });

diff --git a/api/src/core/adapters/hal/getHalSoftwareExternalData.ts b/api/src/core/adapters/hal/getHalSoftwareExternalData.ts
@@ -99,6 +99,27 @@ const codeMetaToReferencePublication = (HALReferencePublication: string[] | Obje
     return undefined;
 };
 
+const HALSource: SILL.WebSite = {
+    "@type": "Website" as const,
+    name: "HAL instance",
+    url: new URL("https://hal.science"),
+    additionalType: "HAL"
+};
+
+const SWHSource: SILL.WebSite = {
+    "@type": "Website" as const,
+    name: "Software Heritage instance",
+    url: new URL("https://www.softwareheritage.org/"),
+    additionalType: "SWH"
+};
+
+const DOISource: SILL.WebSite = {
+    "@type": "Website" as const,
+    name: "DOI instance",
+    url: new URL("https://www.doi.org"),
+    additionalType: "doi"
+};
+
 export const getHalSoftwareExternalData: GetSoftwareExternalData = memoize(
     async (halDocId): Promise<SoftwareExternalData | undefined> => {
         const halRawSoftware = await fetchHalSoftwareById(halDocId).catch(error => {
@@ -170,6 +191,30 @@ export const getHalSoftwareExternalData: GetSoftwareExternalData = memoize(
             })
         );
 
+        const identifiers: SILL.Identification[] =
+            codemetaSoftware?.identifier?.map(halIdentifier => {
+                const base = {
+                    "@type": "PropertyValue" as const,
+                    value: halIdentifier.value,
+                    url: new URL(halIdentifier.propertyID)
+                };
+                switch (halIdentifier["@type"]) {
+                    case "hal":
+                        return { ...base, subjectOf: HALSource };
+                    case "swhid":
+                        return { ...base, subjectOf: SWHSource };
+                    case "doi":
+                        return { ...base, subjectOf: DOISource };
+                    case "bibcode":
+                    case "cern":
+                    case "prodinra":
+                    case "arxiv":
+                    case "biorxiv":
+                    default:
+                        return base;
+                }
+            }) ?? [];
+
         return {
             externalId: halRawSoftware.docid,
             externalDataOrigin: "HAL",
@@ -197,7 +242,8 @@ export const getHalSoftwareExternalData: GetSoftwareExternalData = memoize(
             publicationTime: halRawSoftware?.releasedDate_tdate
                 ? new Date(halRawSoftware?.releasedDate_tdate)
                 : undefined,
-            referencePublications: codeMetaToReferencePublication(codemetaSoftware.referencePublication)
+            referencePublications: codeMetaToReferencePublication(codemetaSoftware.referencePublication),
+            identifiers: identifiers
         };
     },
     {

diff --git a/api/src/core/adapters/wikidata/getWikidataSoftware.ts b/api/src/core/adapters/wikidata/getWikidataSoftware.ts
@@ -256,7 +256,8 @@ export const getWikidataSoftware: GetSoftwareExternalData = memoize(
             programmingLanguages: programmingLanguageString ? [programmingLanguageString] : [],
             applicationCategories: undefined, // doesn't exit on wiki data
             referencePublications: undefined, // doesn't exit on wiki data
-            publicationTime: publicationTimeDate
+            publicationTime: publicationTimeDate,
+            identifiers: undefined
         };
     },
     {

diff --git a/api/src/core/ports/GetSoftwareExternalData.ts b/api/src/core/ports/GetSoftwareExternalData.ts
@@ -33,6 +33,7 @@ export type SoftwareExternalData = {
     applicationCategories: string[];
     publicationTime: Date;
     referencePublications: SILL.ScholarlyArticle[];
+    identifiers: SILL.Identification[];
 }>;
 
 export type SimilarSoftwareExternalData = Pick<

diff --git a/api/src/core/usecases/readWriteSillData/types.ts b/api/src/core/usecases/readWriteSillData/types.ts
@@ -61,6 +61,7 @@ export type Software = {
     keywords: string[];
     programmingLanguages: string[];
     referencePublications?: SILL.ScholarlyArticle[];
+    identifiers?: SILL.Identification[];
 };
 
 export namespace Software {

diff --git a/api/src/types/SILL.ts b/api/src/types/SILL.ts
@@ -1,4 +1,22 @@
 export namespace SILL {
+    export type Sources = "GitLab" | "HAL" | "WikiData" | "SWH" | "Orcid" | "doi" | "GitHub";
+
+    // from https://schema.org/PropertyValue
+    export type Identification = {
+        "@type": "PropertyValue";
+        value: string;
+        url: URL;
+        subjectOf?: WebSite;
+    };
+
+    // from https://schema.org/WebSite
+    export type WebSite = {
+        "@type": "Website";
+        name: string;
+        url: URL;
+        additionalType?: Sources;
+    };
+
     // from https://schema.org/ScholarlyArticle
     export type ScholarlyArticle = {
         "@id": string;