Skip uploads of existing documents and support checksum preflight in importer

backend/src/openapi.rs

@@ -13,6 +13,7 @@ use uuid::Uuid;
         doc::me,
         doc::select_tenant,
         doc::list_documents,
+        doc::check_document,
         doc::upload_document,
         doc::get_document,
         doc::update_document,
@@ -81,6 +82,8 @@ use uuid::Uuid;
             schemas::ReanalyzeResponse,
             schemas::DocumentAssetRequestParams,
             schemas::AssetObjectsQuery,
+            schemas::DocumentCheckQuery,
+            schemas::DocumentCheckResponse,
             schemas::UploadDocumentForm,
             schemas::CreateFolderRequest,
             schemas::EnsureFolderPathRequest,
@@ -189,11 +192,24 @@ mod doc {
         post,
         path = "/api/documents",
         request_body = UploadDocumentForm,
-        responses((status = 201, description = "Document uploaded", body = DocumentDetailResponse)),
+        responses(
+            (status = 201, description = "Document created", body = DocumentDetailResponse),
+            (status = 200, description = "Existing document reused", body = DocumentDetailResponse),
+            (status = 204, description = "Upload skipped because the document already exists")
+        ),
         tag = "Documents"
     )]
     pub(super) fn upload_document() {}
 
+    #[utoipa::path(
+        get,
+        path = "/api/documents/check",
+        params(DocumentCheckQuery),
+        responses((status = 200, description = "Checksum lookup", body = DocumentCheckResponse)),
+        tag = "Documents"
+    )]
+    pub(super) fn check_document() {}
+
     #[utoipa::path(
         get,
         path = "/api/documents/{id}",
@@ -785,6 +801,29 @@ pub mod schemas {
         pub limit: Option<i32>,
     }
 
+    #[derive(Serialize, Deserialize, IntoParams, ToSchema)]
+    #[into_params(parameter_in = Query)]
+    pub struct DocumentCheckQuery {
+        pub checksum: String,
+    }
+
+    #[derive(Serialize, Deserialize, ToSchema)]
+    pub struct DocumentCheckResponse {
+        pub exists: bool,
+        #[schema(nullable)]
+        pub document_id: Option<Uuid>,
+        #[schema(nullable)]
+        pub title: Option<String>,
+        #[schema(nullable)]
+        pub filename: Option<String>,
+        #[schema(nullable)]
+        pub version_id: Option<Uuid>,
+        #[schema(nullable)]
+        pub version_number: Option<i32>,
+        #[schema(nullable)]
+        pub uploaded_at: Option<String>,
+    }
+
     #[derive(Serialize, Deserialize, ToSchema)]
     pub struct UploadDocumentForm {
         #[schema(value_type = String, format = Binary)]
@@ -795,6 +834,14 @@ pub mod schemas {
         pub metadata: Option<Value>,
         #[schema(nullable)]
         pub title: Option<String>,
+        #[schema(nullable, value_type = Vec<Uuid>)]
+        pub tag_ids: Option<Vec<Uuid>>,
+        #[schema(nullable, value_type = Vec<CorrespondentAssignment>)]
+        pub correspondents: Option<Vec<CorrespondentAssignment>>,
+        #[schema(nullable, example = "2024-01-01T00:00:00Z")]
+        pub issued_at: Option<String>,
+        #[schema(nullable)]
+        pub skip_existing: Option<bool>,
     }
 
     #[derive(Serialize, Deserialize, ToSchema)]