upload with more data

2025-10-26 02:28:02 +02:00
parent 62cadbcfa0
commit e51a59a829
5 changed files with 364 additions and 82 deletions
@@ -793,6 +793,8 @@ pub mod schemas {
        pub folder_id: Option<Uuid>,
        #[schema(nullable)]
        pub metadata: Option<Value>,
+        #[schema(nullable)]
+        pub title: Option<String>,
    }

    #[derive(Serialize, Deserialize, ToSchema)]
@@ -6,7 +6,7 @@ use std::{
 use axum::extract::{Json, Multipart, Path, Query, State};
 use axum::http::StatusCode;
 use axum::response::IntoResponse;
-use chrono::{NaiveDateTime, Utc};
+use chrono::{DateTime, NaiveDateTime, Utc};
 use diesel::dsl::exists;
 use diesel::{prelude::*, result::DatabaseErrorKind, select, PgConnection};
 use reqwest::Client;
@@ -293,6 +293,10 @@ struct UploadRequest {
    content_type: Option<String>,
    folder_id: Option<Uuid>,
    metadata: Value,
+    title_override: Option<String>,
+    tag_ids: Vec<Uuid>,
+    correspondents: Vec<CorrespondentAssignmentInput>,
+    issued_at_override: Option<NaiveDateTime>,
 }

 struct UploadOutcome {
@@ -636,6 +640,10 @@ pub async fn upload_document(
    let mut content_type: Option<String> = None;
    let mut folder_id: Option<Uuid> = None;
    let mut metadata: Value = Value::Object(Default::default());
+    let mut tag_ids: Vec<Uuid> = Vec::new();
+    let mut correspondents: Vec<CorrespondentAssignmentInput> = Vec::new();
+    let mut issued_at_override: Option<NaiveDateTime> = None;
+    let mut title_override: Option<String> = None;

    while let Some(field) = multipart.next_field().await.map_err(|err| {
        let msg = format!("invalid multipart data: {err}");
@@ -679,6 +687,70 @@ pub async fn upload_document(
                    AppError::bad_request(msg)
                })?;
            }
+            Some("title") => {
+                let value = field.text().await.map_err(|err| {
+                    let msg = format!("invalid title: {err}");
+                    error!(error = %err, "invalid title payload");
+                    AppError::bad_request(msg)
+                })?;
+                let trimmed = value.trim();
+                if !trimmed.is_empty() {
+                    title_override = Some(trimmed.to_string());
+                }
+            }
+            Some("tag_ids") => {
+                let value = field.text().await.map_err(|err| {
+                    let msg = format!("invalid tag_ids: {err}");
+                    error!(error = %err, "invalid tag_ids payload");
+                    AppError::bad_request(msg)
+                })?;
+                let parsed: Vec<String> = serde_json::from_str(&value).map_err(|err| {
+                    let msg = format!("tag_ids must be a JSON array of UUID strings: {err}");
+                    error!(error = %err, "invalid tag_ids json");
+                    AppError::bad_request(msg)
+                })?;
+                let mut set = HashSet::new();
+                for raw in parsed {
+                    let trimmed = raw.trim();
+                    if trimmed.is_empty() {
+                        continue;
+                    }
+                    let uuid = Uuid::parse_str(trimmed)
+                        .map_err(|_| AppError::bad_request("tag_ids must contain valid UUIDs"))?;
+                    set.insert(uuid);
+                }
+                tag_ids = set.into_iter().collect();
+            }
+            Some("correspondents") => {
+                let value = field.text().await.map_err(|err| {
+                    let msg = format!("invalid correspondents: {err}");
+                    error!(error = %err, "invalid correspondents payload");
+                    AppError::bad_request(msg)
+                })?;
+                correspondents = serde_json::from_str(&value).map_err(|err| {
+                    let msg = format!(
+                        "correspondents must be a JSON array of {{correspondent_id, role}} objects: {err}"
+                    );
+                    error!(error = %err, "invalid correspondents json");
+                    AppError::bad_request(msg)
+                })?;
+            }
+            Some("issued_at") => {
+                let value = field.text().await.map_err(|err| {
+                    let msg = format!("invalid issued_at: {err}");
+                    error!(error = %err, "invalid issued_at payload");
+                    AppError::bad_request(msg)
+                })?;
+                let trimmed = value.trim();
+                if !trimmed.is_empty() {
+                    let parsed = DateTime::parse_from_rfc3339(trimmed).map_err(|err| {
+                        let msg = format!("issued_at must be an RFC3339 timestamp: {err}");
+                        error!(error = %err, "invalid issued_at format");
+                        AppError::bad_request(msg)
+                    })?;
+                    issued_at_override = Some(parsed.naive_utc());
+                }
+            }
            _ => {}
        }
    }
@@ -704,6 +776,10 @@ pub async fn upload_document(
        content_type,
        folder_id,
        metadata,
+        title_override,
+        tag_ids,
+        correspondents,
+        issued_at_override,
    };

    let outcome = match process_upload(&state, request, tenant_id, user_id).await {
@@ -1212,10 +1288,9 @@ pub async fn assign_correspondents(
        return Err(AppError::bad_request("assignments must not be empty"));
    }

-    let (normalized_pairs, correspondents_vec, roles_vec) =
+    let (normalized_pairs, _correspondent_ids, roles_vec) =
        normalize_correspondent_assignments(&payload.assignments)?;
    let replace = payload.replace;
-    let user_id_val = user_id;

    conn.transaction::<(), AppError, _>(|conn| {
        let document: Document = documents::table
@@ -1226,54 +1301,21 @@ pub async fn assign_correspondents(
            return Err(AppError::not_found());
        }

-        if !correspondents_vec.is_empty() {
-            let existing: Vec<Correspondent> = correspondents::table
-                .filter(correspondents::id.eq_any(&correspondents_vec))
-                .filter(correspondents::tenant_id.eq(tenant_id))
-                .load(conn)?;
-            if existing.len() != correspondents_vec.len() {
-                return Err(AppError::bad_request(
-                    "one or more correspondents do not exist",
-                ));
-            }
-        }
-
-        let mut changed = false;
+        let mut deleted = 0;
        if replace {
-            let deleted = diesel::delete(
+            deleted = diesel::delete(
                document_correspondents::table
                    .filter(document_correspondents::document_id.eq(document_id))
                    .filter(document_correspondents::tenant_id.eq(tenant_id))
                    .filter(document_correspondents::role.eq_any(&roles_vec)),
            )
            .execute(conn)?;
-            if deleted > 0 {
-                changed = true;
-            }
        }

-        let new_rows: Vec<NewDocumentCorrespondent> = normalized_pairs
-            .iter()
-            .map(|(correspondent_id, role)| NewDocumentCorrespondent {
-                document_id,
-                correspondent_id: *correspondent_id,
-                role: role.clone(),
-                assigned_by: Some(user_id_val),
-                tenant_id,
-            })
-            .collect();
+        let inserted =
+            insert_document_correspondents(conn, tenant_id, &document, user_id, &normalized_pairs)?;

-        if !new_rows.is_empty() {
-            let inserted = diesel::insert_into(document_correspondents::table)
-                .values(&new_rows)
-                .on_conflict_do_nothing()
-                .execute(conn)?;
-            if inserted > 0 {
-                changed = true;
-            }
-        }
-
-        if changed {
+        if replace && deleted > 0 && inserted == 0 {
            diesel::update(
                documents::table
                    .find(document_id)
@@ -1513,36 +1555,18 @@ pub async fn assign_tags(
        return Err(AppError::bad_request("tag_ids must not be empty"));
    }

-    // Ensure document exists
-    documents::table
+    let document: Document = documents::table
        .find(document_id)
        .filter(documents::tenant_id.eq(tenant_id))
-        .first::<Document>(&mut conn)?;
+        .first(&mut conn)?;

-    // Ensure tags exist
-    let existing_tags: Vec<Tag> = tags::table
-        .filter(tags::id.eq_any(&payload.tag_ids))
-        .filter(tags::tenant_id.eq(tenant_id))
-        .load(&mut conn)?;
-    if existing_tags.len() != payload.tag_ids.len() {
-        return Err(AppError::bad_request("one or more tags do not exist"));
-    }
-
-    let new_tags: Vec<NewDocumentTag> = payload
-        .tag_ids
-        .iter()
-        .map(|tag_id| NewDocumentTag {
-            document_id,
-            tag_id: *tag_id,
-            assigned_by: Some(user_id),
-            tenant_id,
-        })
-        .collect();
-
-    diesel::insert_into(document_tags::table)
-        .values(&new_tags)
-        .on_conflict_do_nothing()
-        .execute(&mut conn)?;
+    assign_tags_internal(
+        &mut conn,
+        tenant_id,
+        &document,
+        &payload.tag_ids,
+        Some(user_id),
+    )?;

    Ok(StatusCode::NO_CONTENT)
 }
@@ -1663,6 +1687,10 @@ async fn process_upload(
        content_type,
        folder_id,
        metadata,
+        title_override,
+        tag_ids,
+        correspondents,
+        issued_at_override,
    } = request;

    if let Some(folder) = folder_id {
@@ -1672,7 +1700,13 @@ async fn process_upload(
    let doc_id = Uuid::new_v4();
    let version_id = Uuid::new_v4();
    let version_number = 1;
-    let stored_filename = original_name.clone();
+    let derived_title = title_override
+        .as_ref()
+        .map(|value| value.trim())
+        .filter(|value| !value.is_empty())
+        .map(|value| value.to_string())
+        .unwrap_or_else(|| derive_document_title(&original_name));
+    let stored_filename = filename_with_retained_extension(&derived_title, &original_name);

    let checksum = Sha256::digest(&bytes);
    let checksum_hex = hex::encode(checksum);
@@ -1694,6 +1728,32 @@ async fn process_upload(
            .optional()?;

        if let Some((mut document, version)) = existing {
+            if let Some(issued_at) = issued_at_override {
+                if document.issued_at != Some(issued_at) {
+                    diesel::update(
+                        documents::table
+                            .find(document.id)
+                            .filter(documents::tenant_id.eq(tenant_id)),
+                    )
+                    .set((
+                        documents::issued_at.eq(Some(issued_at)),
+                        documents::updated_at.eq(Utc::now().naive_utc()),
+                    ))
+                    .execute(&mut conn)?;
+                    document.issued_at = Some(issued_at);
+                }
+            }
+
+            assign_tags_internal(&mut conn, tenant_id, &document, &tag_ids, Some(user_id))?;
+
+            assign_correspondents_internal(
+                &mut conn,
+                tenant_id,
+                &document,
+                user_id,
+                &correspondents,
+            )?;
+
            if document.deleted_at.is_some() {
                let now = Utc::now().naive_utc();
                diesel::update(documents::table.find(document.id))
@@ -1737,7 +1797,7 @@ async fn process_upload(
        }
    }

-    let content_disposition = inline_content_disposition(&original_name);
+    let content_disposition = inline_content_disposition(&stored_filename);

    let storage = state.storage_for_tenant(tenant_id)?;

@@ -1770,8 +1830,8 @@ async fn process_upload(
                content_type: content_type.clone(),
                folder_id,
                current_version_id: version_id,
-                issued_at: None,
-                title: derive_document_title(&original_name),
+                issued_at: issued_at_override,
+                title: derived_title.clone(),
                metadata: metadata_value.clone(),
                tenant_id,
            };
@@ -1802,15 +1862,29 @@ async fn process_upload(
        })?
    };

-    let detail = DocumentDetailResponse {
-        document: to_document_response(
-            state,
-            user_id,
-            document,
-            None,
-            Vec::new(),
-            Some((to_version_response(version.clone(), true), Vec::new())),
-        )?,
+    let detail = {
+        let mut conn = state.db_for_tenant(tenant_id)?;
+
+        assign_tags_internal(&mut conn, tenant_id, &document, &tag_ids, Some(user_id))?;
+
+        assign_correspondents_internal(&mut conn, tenant_id, &document, user_id, &correspondents)?;
+
+        let tags_map = load_tags_for_documents(&mut conn, &[doc_id])?;
+        let mut correspondents_map = load_correspondents_for_documents(&mut conn, &[doc_id])?;
+        let tags = tags_map.get(&doc_id).cloned();
+        let correspondents = correspondents_map.remove(&doc_id).unwrap_or_default();
+        drop(conn);
+
+        DocumentDetailResponse {
+            document: to_document_response(
+                state,
+                user_id,
+                document,
+                tags,
+                correspondents,
+                Some((to_version_response(version.clone(), true), Vec::new())),
+            )?,
+        }
    };

    if let Ok(mut conn) = state.db_for_tenant(tenant_id) {
@@ -1837,6 +1911,136 @@ async fn process_upload(
    })
 }

+fn assign_tags_internal(
+    conn: &mut PgConnection,
+    tenant_id: Uuid,
+    document: &Document,
+    raw_tag_ids: &[Uuid],
+    assigned_by: Option<Uuid>,
+) -> AppResult<usize> {
+    if raw_tag_ids.is_empty() {
+        return Ok(0);
+    }
+
+    let mut tag_ids: Vec<Uuid> = raw_tag_ids.iter().copied().collect();
+    tag_ids.sort_unstable();
+    tag_ids.dedup();
+
+    if tag_ids.is_empty() {
+        return Ok(0);
+    }
+
+    let existing: Vec<Uuid> = tags::table
+        .filter(tags::id.eq_any(&tag_ids))
+        .filter(tags::tenant_id.eq(tenant_id))
+        .select(tags::id)
+        .load(conn)?;
+
+    if existing.len() != tag_ids.len() {
+        return Err(AppError::bad_request("one or more tags do not exist"));
+    }
+
+    let new_tags: Vec<NewDocumentTag> = tag_ids
+        .into_iter()
+        .map(|tag_id| NewDocumentTag {
+            document_id: document.id,
+            tag_id,
+            assigned_by,
+            tenant_id,
+        })
+        .collect();
+
+    if new_tags.is_empty() {
+        return Ok(0);
+    }
+
+    let inserted = diesel::insert_into(document_tags::table)
+        .values(&new_tags)
+        .on_conflict_do_nothing()
+        .execute(conn)?;
+
+    Ok(inserted)
+}
+
+fn assign_correspondents_internal(
+    conn: &mut PgConnection,
+    tenant_id: Uuid,
+    document: &Document,
+    user_id: Uuid,
+    assignments: &[CorrespondentAssignmentInput],
+) -> AppResult<usize> {
+    if assignments.is_empty() {
+        return Ok(0);
+    }
+
+    let (normalized_pairs, _correspondent_ids, _roles) =
+        normalize_correspondent_assignments(assignments)?;
+
+    insert_document_correspondents(conn, tenant_id, document, user_id, &normalized_pairs)
+}
+
+fn insert_document_correspondents(
+    conn: &mut PgConnection,
+    tenant_id: Uuid,
+    document: &Document,
+    user_id: Uuid,
+    normalized_pairs: &[(Uuid, String)],
+) -> AppResult<usize> {
+    if normalized_pairs.is_empty() {
+        return Ok(0);
+    }
+
+    let mut correspondent_ids: Vec<Uuid> = normalized_pairs.iter().map(|(id, _)| *id).collect();
+    correspondent_ids.sort_unstable();
+    correspondent_ids.dedup();
+
+    if !correspondent_ids.is_empty() {
+        let existing: Vec<Uuid> = correspondents::table
+            .filter(correspondents::id.eq_any(&correspondent_ids))
+            .filter(correspondents::tenant_id.eq(tenant_id))
+            .select(correspondents::id)
+            .load(conn)?;
+
+        if existing.len() != correspondent_ids.len() {
+            return Err(AppError::bad_request(
+                "one or more correspondents do not exist",
+            ));
+        }
+    }
+
+    let new_rows: Vec<NewDocumentCorrespondent> = normalized_pairs
+        .iter()
+        .map(|(correspondent_id, role)| NewDocumentCorrespondent {
+            document_id: document.id,
+            correspondent_id: *correspondent_id,
+            role: role.clone(),
+            assigned_by: Some(user_id),
+            tenant_id,
+        })
+        .collect();
+
+    if new_rows.is_empty() {
+        return Ok(0);
+    }
+
+    let inserted = diesel::insert_into(document_correspondents::table)
+        .values(&new_rows)
+        .on_conflict_do_nothing()
+        .execute(conn)?;
+
+    if inserted > 0 {
+        diesel::update(
+            documents::table
+                .find(document.id)
+                .filter(documents::tenant_id.eq(tenant_id)),
+        )
+        .set(documents::updated_at.eq(Utc::now().naive_utc()))
+        .execute(conn)?;
+    }
+
+    Ok(inserted)
+}
+
 fn ensure_folder_exists(state: &AppState, tenant_id: Uuid, folder_id: Uuid) -> AppResult<()> {
    let mut conn = state.db_for_tenant(tenant_id)?;
    let exists: bool = diesel::select(exists(
@@ -478,6 +478,30 @@ impl TestApp {
        data: &[u8],
        folder_id: Option<Uuid>,
        token: &str,
+    ) -> Result<hyper::Response<Body>> {
+        self.upload_document_with_options(
+            path,
+            filename,
+            content_type,
+            data,
+            folder_id,
+            None,
+            None,
+            token,
+        )
+        .await
+    }
+
+    pub async fn upload_document_with_options(
+        &self,
+        path: &str,
+        filename: &str,
+        content_type: &str,
+        data: &[u8],
+        folder_id: Option<Uuid>,
+        title: Option<&str>,
+        metadata_json: Option<&str>,
+        token: &str,
    ) -> Result<hyper::Response<Body>> {
        let boundary = format!("boundary-{}", Uuid::new_v4());
        let mut body = Vec::new();
@@ -500,6 +524,20 @@ impl TestApp {
            body.extend(b"\r\n");
        }

+        if let Some(title_value) = title {
+            body.extend(format!("--{boundary}\r\n").as_bytes());
+            body.extend(b"Content-Disposition: form-data; name=\"title\"\r\n\r\n");
+            body.extend(title_value.as_bytes());
+            body.extend(b"\r\n");
+        }
+
+        if let Some(metadata_value) = metadata_json {
+            body.extend(format!("--{boundary}\r\n").as_bytes());
+            body.extend(b"Content-Disposition: form-data; name=\"metadata\"\r\n\r\n");
+            body.extend(metadata_value.as_bytes());
+            body.extend(b"\r\n");
+        }
+
        body.extend(format!("--{boundary}--\r\n").as_bytes());

        let builder = Request::builder()
@@ -15,6 +15,7 @@ struct DocumentDetail {
 struct DocumentInfo {
    id: Uuid,
    title: String,
+    filename: String,
    original_name: String,
    deleted_at: Option<String>,
    issued_at: Option<String>,
@@ -240,6 +241,43 @@ async fn upload_and_list_document() -> Result<()> {
    Ok(())
 }

+#[tokio::test]
+async fn upload_document_with_custom_title_sets_filename() -> Result<()> {
+    let _lock = acquire_db_lock().await;
+    let app = TestApp::new().await?;
+
+    let password = "passw0rd";
+    app.insert_user("nora", password, "admin").await?;
+    let token = app.login_token("nora", password).await?;
+
+    let file_bytes = b"example contract body".to_vec();
+    let title = "Vendor Contract";
+    let original_filename = "scan.pdf";
+
+    let upload = app
+        .upload_document_with_options(
+            "/api/documents",
+            original_filename,
+            "application/pdf",
+            &file_bytes,
+            None,
+            Some(title),
+            None,
+            &token,
+        )
+        .await?;
+    assert_eq!(upload.status(), StatusCode::CREATED);
+    let body = body_to_vec(upload.into_body()).await?;
+    let detail: DocumentDetail = serde_json::from_slice(&body)?;
+
+    assert_eq!(detail.document.title, title);
+    assert_eq!(detail.document.filename, format!("{title}.pdf"));
+    assert_eq!(detail.document.original_name, original_filename);
+
+    app.cleanup().await?;
+    Ok(())
+}
+
 #[tokio::test]
 async fn duplicate_and_restore_document() -> Result<()> {
    let _lock = acquire_db_lock().await;
@@ -17,7 +17,7 @@ Health
 Documents
 ---------
 - GET  /api/documents - List or search documents. Optional filters: `folder_id` (defaults to root when omitted), `include_deleted`, `include_descendants` (defaults to true when a `folder_id` is provided and no other override is supplied), `query` (Quickwit full-text), `tags` (comma-separated tag UUIDs), and `correspondents` (comma-separated correspondent UUIDs). Each entry includes tags, correspondent assignments, and current version info.
- POST /api/documents - Upload a document via multipart form-data (`file`, optional metadata/folder fields).
+- POST /api/documents - Upload a document via multipart form-data (`file`, optional `title`, `folder_id`, and JSON `metadata`). When `title` is supplied, the stored filename is set to `<title><original_extension>` automatically.
 - POST /api/documents/bulk/move - Move multiple documents to a target folder.
 - POST /api/documents/bulk/tags - Add or remove tags across multiple documents.
 - POST /api/documents/bulk/correspondents - Bulk correspondent actions. Default `action=add` replaces existing assignments for the provided roles before adding the supplied correspondents; `action=remove` drops the specified correspondent/role pairs.