codegouvfr · guillermau · Jan 28, 2025 · guillermau · Jan 30, 2025
diff --git a/api/src/core/adapters/dbApi/kysely/createPgSoftwareExternalDataRepository.ts b/api/src/core/adapters/dbApi/kysely/createPgSoftwareExternalDataRepository.ts
@@ -11,6 +11,7 @@ export const createPgSoftwareExternalDataRepository = (db: Kysely<Database>): So
             keywords: JSON.stringify(softwareExternalData.keywords),
             applicationCategories: JSON.stringify(softwareExternalData.applicationCategories),
             programmingLanguages: JSON.stringify(softwareExternalData.programmingLanguages),
+            referencePublication: JSON.stringify(softwareExternalData.referencePublication),
             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
@@ -193,6 +193,7 @@ export const createPgSoftwareRepository = (db: Kysely<Database>): SoftwareReposi
                         parentWikidataSoftware: parentExternalData,
                         keywords: software?.keywords ?? softwareExternalData?.keywords ?? [],
                         programmingLanguages: softwareExternalData?.programmingLanguages ?? [],
+                        referencePublication: softwareExternalData?.referencePublication,
                         applicationCategories: software.categories.concat(
                             softwareExternalData?.applicationCategories ?? []
                         ),
@@ -296,7 +297,8 @@ export const createPgSoftwareRepository = (db: Kysely<Database>): SoftwareReposi
                                 softwareExternalData?.applicationCategories ?? []
                             ),
                             categories: undefined, // merged in applicationCategories, set to undefined to remove it
-                            programmingLanguages: softwareExternalData?.programmingLanguages ?? []
+                            programmingLanguages: softwareExternalData?.programmingLanguages ?? [],
+                            referencePublication: softwareExternalData?.referencePublication
                         });
                     }
                 );
@@ -431,6 +433,7 @@ const makeGetSoftwareBuilder = (db: Kysely<Database>) =>
                     documentationUrl: ref("ext.documentationUrl"),
                     programmingLanguages: ref("ext.programmingLanguages"),
                     applicationCategories: ref("ext.applicationCategories"),
+                    referencePublication: ref("ext.referencePublication"),
                     keywords: ref("ext.keywords"),
                     softwareVersion: ref("ext.softwareVersion"),
                     publicationTime: ref("ext.publicationTime")
@@ -565,6 +568,7 @@ const makeGetSoftwareById =
                     testUrl: testUrls[0]?.url,
                     parentWikidataSoftware: parentExternalData,
                     programmingLanguages: softwareExternalData?.programmingLanguages ?? [],
+                    referencePublication: softwareExternalData?.referencePublication,
                     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
@@ -1,4 +1,6 @@
 import { Generated, JSONColumnType } from "kysely";
+// Only allowed import on JSONColumnType
+import { SILL } from "../../../../types/SILL";
 
 // from https://schema.org/Organization
 type SchemaOrganization = {
@@ -93,6 +95,7 @@ type SoftwareExternalDatasTable = {
     keywords: JSONColumnType<string[]> | null;
     programmingLanguages: JSONColumnType<string[]> | null;
     applicationCategories: JSONColumnType<string[]> | null;
+    referencePublication: JSONColumnType<SILL.ScholarlyArticle[]> | null;
     publicationTime: Date | null;
 };
 

diff --git a/api/src/core/adapters/dbApi/kysely/migrations/1737023616811_add-referencePublication.ts b/api/src/core/adapters/dbApi/kysely/migrations/1737023616811_add-referencePublication.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("referencePublication", "jsonb").execute();
+}
+
+export async function down(db: Kysely<any>): Promise<void> {
+    await db.schema.alterTable("software_external_datas").dropColumn("referencePublication").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
@@ -54,6 +54,7 @@ const softwareExternalData: SoftwareExternalData = {
     keywords: ["Usefull", "Daily"],
     programmingLanguages: ["C++"],
     applicationCategories: ["Software Cat I", "Software Cat II"],
+    referencePublication: undefined,
     publicationTime: new Date(1561566581000)
 };
 
@@ -81,6 +82,7 @@ const similarSoftwareExternalData: SoftwareExternalData = {
     keywords: ["Infra", "Adminsys"],
     programmingLanguages: ["Python3"],
     applicationCategories: ["Software Cat I", "Software Cat II"],
+    referencePublication: undefined,
     publicationTime: new Date(1561566581000)
 };
 
@@ -441,9 +443,10 @@ describe("pgDbApi", () => {
                     developers: JSON.stringify(softExtData.developers),
                     label: JSON.stringify(softExtData.label),
                     description: JSON.stringify(softExtData.description),
-                    keywords: JSON.stringify(softwareExternalData.keywords),
-                    applicationCategories: JSON.stringify(softwareExternalData.applicationCategories),
-                    programmingLanguages: JSON.stringify(softwareExternalData.programmingLanguages)
+                    keywords: JSON.stringify(softExtData.keywords),
+                    applicationCategories: JSON.stringify(softExtData.applicationCategories),
+                    programmingLanguages: JSON.stringify(softExtData.programmingLanguages),
+                    referencePublication: JSON.stringify(softExtData.referencePublication)
                 }))
             )
             .execute();

diff --git a/api/src/core/adapters/fetchExternalData.test.ts b/api/src/core/adapters/fetchExternalData.test.ts
@@ -194,6 +194,7 @@ describe("fetches software extra data (from different providers)", () => {
                     sourceUrl: "https://github.com/facebook/create-react-app",
                     websiteUrl: "https://create-react-app.dev/",
                     programmingLanguages: [],
+                    referencePublication: null,
                     softwareVersion: "5.0.1",
                     publicationTime: new Date("2022-04-12T00:00:00.000Z")
                 },
@@ -213,6 +214,7 @@ describe("fetches software extra data (from different providers)", () => {
                     externalId: "Q111590996",
                     framaLibreId: null,
                     isLibreSoftware: true,
+
                     keywords: [],
                     label: "Vite",
                     license: "MIT licence",
@@ -221,6 +223,7 @@ describe("fetches software extra data (from different providers)", () => {
                     sourceUrl: "https://github.com/vitejs/vite",
                     websiteUrl: "https://vitejs.dev/",
                     programmingLanguages: ["JavaScript"],
+                    referencePublication: null,
                     softwareVersion: expect.any(String),
                     publicationTime: expect.any(Date)
                 }
@@ -278,6 +281,7 @@ describe("fetches software extra data (from different providers)", () => {
                         "//upload.wikimedia.org/wikipedia/commons/thumb/1/10/Apache_HTTP_server_logo_%282019-present%29.svg/220px-Apache_HTTP_server_logo_%282019-present%29.svg.png",
                     sourceUrl: "https://github.com/apache/httpd",
                     websiteUrl: "https://httpd.apache.org/",
+                    referencePublication: 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
@@ -38,6 +38,7 @@ describe("HAL", () => {
                 "keywords": undefined,
                 "programmingLanguages": undefined,
                 "applicationCategories": ["Computer Science [cs]"],
+                "referencePublication": undefined,
                 "publicationTime": new Date(1521545908000)
             });
         });

diff --git a/api/src/core/adapters/hal/getHalSoftwareExternalData.ts b/api/src/core/adapters/hal/getHalSoftwareExternalData.ts
@@ -28,6 +28,72 @@ const buildParentOrganizationTree = async (
     );
 };
 
+const parseReferencePublication = (key: string, value: string | string[]): SILL.ScholarlyArticle[] => {
+    const arrayValue = typeof value === "string" ? value.split(",") : value;
+    switch (key) {
+        case "hal":
+            return arrayValue.map((halThing): SILL.ScholarlyArticle => {
+                return {
+                    "@id": halThing,
+                    "@type": "ScholarlyArticle",
+                    identifier: {
+                        "@type": "PropertyValue",
+                        value: halThing,
+                        propertyID: "HAL",
+                        url: halThing.includes("https") ? new URL(halThing) : new URL(`https://hal.science/${halThing}`)
+                    }
+                };
+            });
+        case "doi":
+            return arrayValue.map((doi): SILL.ScholarlyArticle => {
+                return {
+                    "@id": doi,
+                    "@type": "ScholarlyArticle",
+                    identifier: {
+                        "@type": "PropertyValue",
+                        value: doi,
+                        propertyID: "doi",
+                        url: doi.includes("https") ? URL.parse(doi) : URL.parse(`https://doi.org/${doi}`)
+                    }
+                };
+            });
+        case "arxiv":
+            return arrayValue.map((arxivId): SILL.ScholarlyArticle => {
+                return {
+                    "@id": arxivId,
+                    "@type": "ScholarlyArticle",
+                    identifier: {
+                        "@type": "PropertyValue",
+                        value: arxivId,
+                        propertyID: "arxiv",
+                        url: arxivId.includes("https")
+                            ? URL.parse(arxivId)
+                            : URL.parse(`https://arxiv.org/abs/${arxivId}`)
+                    }
+                };
+            });
+        default:
+            return [];
+    }
+};
+
+const codeMetaToReferencePublication = (HALReferencePublication: string[] | Object | undefined) => {
+    if (HALReferencePublication) {
+        if (Array.isArray(HALReferencePublication)) {
+            console.error("Issue with HAL data. This data need to be curated", HALReferencePublication);
+            return undefined;
+        }
+
+        return Object.entries(HALReferencePublication).reduce(
+            (publicationArray: SILL.ScholarlyArticle[], [key, value]) => {
+                return publicationArray.concat(parseReferencePublication(key, value));
+            },
+            []
+        );
+    }
+    return undefined;
+};
+
 export const getHalSoftwareExternalData: GetSoftwareExternalData = memoize(
     async (halDocId): Promise<SoftwareExternalData | undefined> => {
         const halRawSoftware = await fetchHalSoftwareById(halDocId).catch(error => {
@@ -125,7 +191,8 @@ export const getHalSoftwareExternalData: GetSoftwareExternalData = memoize(
             applicationCategories: sciencesCategories,
             publicationTime: halRawSoftware?.releasedDate_tdate
                 ? new Date(halRawSoftware?.releasedDate_tdate)
-                : undefined
+                : undefined,
+            referencePublication: codeMetaToReferencePublication(codemetaSoftware.referencePublication)
         };
     },
     {

diff --git a/api/src/core/adapters/hal/types/HAL.ts b/api/src/core/adapters/hal/types/HAL.ts
@@ -34,6 +34,7 @@ export namespace HAL {
         developmentStatus?: string;
         author: Role[]; // Non regular Schema.org / CodeMeta
         contributor?: Person[];
+        referencePublication?: string[] | string | Object;
     };
 
     // HAL implementation of https://codemeta.github.io/terms/

diff --git a/api/src/core/adapters/wikidata/getWikidataSoftware.ts b/api/src/core/adapters/wikidata/getWikidataSoftware.ts
@@ -255,6 +255,7 @@ export const getWikidataSoftware: GetSoftwareExternalData = memoize(
             keywords: getClaimDataValue<"string">("P921"),
             programmingLanguages: programmingLanguageString ? [programmingLanguageString] : [],
             applicationCategories: undefined, // doesn't exit on wiki data
+            referencePublication: undefined, // doesn't exit on wiki data
             publicationTime: publicationTimeDate
         };
     },

diff --git a/api/src/core/ports/GetSoftwareExternalData.ts b/api/src/core/ports/GetSoftwareExternalData.ts
@@ -32,6 +32,7 @@ export type SoftwareExternalData = {
     programmingLanguages: string[];
     applicationCategories: string[];
     publicationTime: Date;
+    referencePublication: SILL.ScholarlyArticle[];
 }>;
 
 export type SimilarSoftwareExternalData = Pick<

diff --git a/api/src/types/SILL.ts b/api/src/types/SILL.ts
@@ -1,4 +1,19 @@
 export namespace SILL {
+    // from https://schema.org/ScholarlyArticle
+    export type ScholarlyArticle = {
+        "@id": string;
+        "@type": "ScholarlyArticle";
+        identifier?: PropertyValue;
+    };
+
+    // from https://schema.org/PropertyValue
+    export type PropertyValue = {
+        "@type"?: "PropertyValue";
+        value?: string;
+        propertyID?: string;
+        url?: URL | null;
+    };
+
     // from https://schema.org/Organization
     export type Organization = {
         "@type": "Organization";