[FIX] DocumentAPI to query by params and return multiple documents & fix PDF file fetching

extralit-frontend/CHANGELOG.md

@@ -14,6 +14,10 @@ These are the section headers that we use:
 * "Security" in case of vulnerabilities.
 -->
 
+## [Extralit] [0.6.1](https://github.com/extralit/extralit/compare/v0.5.0...v0.6.1)
+### Changed
+- Refactored the frontend to use a single fetchDocument method that queries documents by any identifier and workspace, replacing the previous fetchDocumentByID and fetchDocumentByPubmedID methods. The view model and use case now expect and handle the new API response format
+
 ## [Extralit] [0.6.0](https://github.com/extralit/extralit/compare/v0.4.1...v0.6.0)
 
 ### Added

extralit-frontend/components/base/base-pdf-viewer/PDFViewer.vue

@@ -19,7 +19,7 @@
   </div>
 </template>
 
-<script>
+<script lang="ts">
 import { PDFView } from "@jonnytran/vue-pdf-viewer";
 
 export default {
@@ -44,8 +44,8 @@ export default {
   },
 
   watch: {
-    pageNumber(newPageNumber) {
-      this.currentPageNumber = newPageNumber;
+    pageNumber(newPageNumber: Number | String) {
+      this.currentPageNumber = newPageNumber ? Number(newPageNumber) : 1;
     },
   },
 
@@ -56,21 +56,6 @@ export default {
     };
   },
 
-  mounted() {
-    window.addEventListener("hashchange", this.onHashChange);
-    this.onHashChange(); // Call on component mount to handle initial hash
-  },
-
-  methods: {
-    onHashChange() {
-      const hash = window.location.hash.substring(1); // Remove the '#' from the hash
-      const [key, value] = hash.split(".");
-      if (key === "page_number" && !isNaN(value)) {
-        this.currentPageNumber = Number(value);
-      }
-    },
-  },
-
   errorCaptured(err, component, info) {
     this.error = err;
     console.error(`Error caught from ${component}: ${err}`);

extralit-frontend/components/features/annotation/container/mode/useDocumentViewModel.ts

@@ -6,10 +6,12 @@
 import { useDataset } from "@/v1/infrastructure/storage/DatasetStorage";
 import { waitForAsyncValue } from "@/v1/infrastructure/services/useWait";
 import { useNotifications } from "~/v1/infrastructure/services/useNotifications";
+import { useWorkspaces } from "~/v1/infrastructure/storage/WorkspaceStorage";
 
 export const useDocumentViewModel = (props: { record: any }) => {
   const notification = useNotifications();
   const getDocument = useResolve(GetDocumentByIdUseCase);
+  const { state: workspaces } = useWorkspaces();
   const { state: dataset } = useDataset();
   const { state: document, set: setDocument, clear: clearDocument } = useDocument();
   const isLoading = ref(false);
@@ -17,44 +19,54 @@
   const hasDocumentLoaded = computed(() => {
     return document.id !== null;
   });
   const hasDocument = computed(() => {
     return (
-      props.record.metadata === null || props.record.metadata?.doc_id != null || props.record.metadata?.pmid != null
+      props.record.metadata === null || props.record.metadata?.reference != null || props.record.metadata?.pmid != null
     );
   });
 
-  const fetchDocumentByID = async (id: string) => {
+  const fetchDocument = async (metadata: any) => {
     try {
-      await getDocument.setDocumentByID(id);
-    } catch (e) {
-      notification.notify({
-        message: `Error fetching document with ID ${id}`,
-        type: "danger",
-      });
-      clearDocument();
-    }
-  };
+      await waitForAsyncValue(() => workspaces.selectedWorkspace?.id);
 
-  const fetchDocumentByPubmedID = async (pmid: string) => {
-    try {
-      await getDocument.setDocumentByPubmedID(pmid);
+      const params: {
+        workspace_id: string;
+        doc_id?: string;
+        pmid?: string;
+        doi?: string;
+        reference?: string;
+      } = { workspace_id: workspaces.selectedWorkspace!.id };
+
+      if (metadata?.reference) params.reference = metadata.reference;
+      if (metadata?.doc_id) params.doc_id = metadata.doc_id;
+      if (metadata?.pmid) params.pmid = metadata.pmid;
+      if (metadata?.doi) params.doi = metadata.doi;
+
+      // Ensure at least one identifier is provided
+      if (Object.keys(params).length === 0) {
+        throw new Error("No valid document identifier found in metadata");
+      }
+
+      await getDocument.setDocument(params);
     } catch (e) {
+      const identifier = metadata?.pmid || metadata?.doi || metadata?.doc_id || metadata?.reference || "unknown";
+      console.error(`Error fetching document with identifier "${identifier}":`, e);
       notification.notify({
-        message: `Error fetching document with pmid "${pmid}"`,
+        message: `Error fetching document with identifier "${identifier}"`,
         type: "danger",
       });
       clearDocument();
     }
   };
 
   const updateDocument = async (metadata: any) => {
-    if (metadata?.pmid != null) {
-      if (document.pmid !== metadata.pmid) {
-        fetchDocumentByPubmedID(metadata.pmid);
-      }
+    if (metadata?.pmid != null && document.pmid !== metadata.pmid) {
+      fetchDocument(metadata);
+    } else if (metadata?.doi != null && document.doi !== metadata.doi) {
+      fetchDocument(metadata);
     } else if (metadata?.doc_id != null && document.id !== metadata.doc_id) {
-      fetchDocumentByID(metadata.doc_id);
-    } else if (!metadata?.pmid && !metadata?.doc_id && hasDocumentLoaded.value) {
+      fetchDocument(metadata);
+    } else if (!metadata?.pmid && !metadata?.doi && !metadata?.doc_id && hasDocumentLoaded.value) {
       clearDocument();
     }
 
@@ -107,8 +119,6 @@
 
   return {
     document,
-    fetchDocumentByID,
-    fetchDocumentByPubmedID,
     fetchDocumentSegments,
     focusDocumentPageNumber,
     clearDocument,

extralit-frontend/v1/domain/entities/document/Document.ts

@@ -6,19 +6,19 @@ export class Segment {
 		public readonly header: string | null,
 		public readonly page_number: number | null,
 		public readonly type: string | null,
-	) {}
+	) { }
 
 	public static getDescription(segment: Segment): string {
-			let segmentDescription = Object.entries(segment)
-					.filter(([key, value]) => !['header', 'doc_id'].includes(key))
-					.map(([key, value]) => `${key}: ${value}`)
-					.join('\n');
-			return segmentDescription;
+		let segmentDescription = Object.entries(segment)
+			.filter(([key, value]) => !['header', 'doc_id'].includes(key))
+			.map(([key, value]) => `${key}: ${value}`)
+			.join('\n');
+		return segmentDescription;
 	}
 }
 
 export interface Segments {
-  items: Segment[];
+	items: Segment[];
 }
 
 export class Document {
@@ -27,6 +27,7 @@ export class Document {
 		public readonly url?: string,
 		public readonly file_name?: string,
 		public readonly pmid?: string,
+		public readonly doi?: string,
 		public readonly page_number?: number | string,
 		public reference?: string,
 		public segments?: Segment[],

extralit-frontend/v1/domain/usecases/get-document-by-id-use-case.ts

@@ -8,16 +8,26 @@ export class GetDocumentByIdUseCase {
     private readonly documentStorage: IDocumentStorage
   ) {}
 
-  async setDocumentByID(id: string) {
-    const document = await this.documentRepository.getDocumentById(id);
+  async setDocument(params: {
+    workspace_id: string;
+    doc_id?: string;
+    pmid?: string;
+    doi?: string;
+    reference?: string;
+  }) {
+    const documents = await this.documentRepository.getDocuments(params);
 
-    this.documentStorage.set(document);
-  }
+    if (documents.length === 0) {
+      throw new Error("No documents found with the provided criteria");
+    }
 
-  async setDocumentByPubmedID(id: string) {
-    const document = await this.documentRepository.getDocumentByPubmedID(id);
+    // For now, we'll use the first document found
+    // In the future, you might want to add logic to handle multiple documents
+    if (documents.length > 1) {
+      console.warn(`Multiple documents found (${documents.length}). Using the first one.`);
+    }
 
-    this.documentStorage.set(document);
+    this.documentStorage.set(documents[0]);
   }
 
   async setSegments(workspace: string, reference: string): Promise<Segment[]> {

extralit-frontend/v1/infrastructure/repositories/DocumentRepository.ts

@@ -3,25 +3,29 @@ import { Document, Segment, Segments } from "@/v1/domain/entities/document/Docum
 
 const DOCUMENT_API_ERRORS = {
   ERROR_FETCHING_DOCUMENT: "ERROR_FETCHING_DOCUMENT",
+  ERROR_FETCHING_SEGMENTS: "ERROR_FETCHING_SEGMENTS",
 };
 
 export class DocumentRepository {
   constructor(private readonly axios: NuxtAxiosInstance) {}
 
-  async getDocumentByPubmedID(pmid: string): Promise<Document> {
+  async getDocuments(params: {
+    workspace_id: string;
+    doc_id?: string;
+    pmid?: string;
+    doi?: string;
+    reference?: string;
+  }): Promise<Document[]> {
     try {
-      const { data } = await this.axios.get<Document>(`/v1/documents/by-pmid/${pmid}`);
-      return data;
-    } catch (error) {
-      throw {
-        response: DOCUMENT_API_ERRORS.ERROR_FETCHING_DOCUMENT,
-      };
-    }
-  }
+      const queryParams = Object.fromEntries(Object.entries(params).filter(([_, value]) => value !== undefined));
 
-  async getDocumentById(id: string): Promise<Document> {
-    try {
-      const { data } = await this.axios.get<Document>(`/v1/documents/by-id/${id}`);
+      if (Object.keys(queryParams).length === 0) {
+        throw new Error("At least one identifier parameter must be provided");
+      }
+
+      const { data } = await this.axios.get<Document[]>("/v1/documents", {
+        params: queryParams,
+      });
       return data;
     } catch (error) {
       throw {
@@ -39,7 +43,7 @@ export class DocumentRepository {
       return data.items;
     } catch (error) {
       throw {
-        response: DOCUMENT_API_ERRORS.ERROR_FETCHING_DOCUMENT,
+        response: DOCUMENT_API_ERRORS.ERROR_FETCHING_SEGMENTS,
       };
     }
   }

extralit-server/CHANGELOG.md

@@ -14,6 +14,11 @@ These are the section headers that we use:
 * "Security" in case of vulnerabilities.
 -->
 
+## [Extralit] [0.6.1](https://github.com/extralit/extralit/compare/v0.5.0...v0.6.1)
+
+### Changed
+- Replaced separate /documents/by-id/{id} and /documents/by-pmid/{pmid} endpoints with a single /documents endpoint that accepts workspace_id and one or more identifiers (id, pmid, doi, reference), returning a list of matching documents
+
 ## [Extralit] [0.6.0](https://github.com/extralit/extralit/compare/v0.5.0...v0.6.0)
 
 ### Added