include a workflow ID for each deposition form open in the browser, c…

biosimdb_interface/__init__.py

@@ -12,6 +12,8 @@
 from dotenv import load_dotenv
 from flask import Flask
 
+from .form.workflows import WorkflowStore
+
 load_dotenv()
 # UPLOAD_FOLDER = "/tmp"
 
@@ -45,6 +47,7 @@ def create_app(test_config=None):
 
     # App and Invenio OAuth2 configuration — values loaded from .env
     app.config.from_mapping(
+        WORKFLOW_TTL_SECONDS=int(os.getenv("WORKFLOW_TTL_SECONDS", "14400")),
         UPLOAD_FOLDER=os.getenv("UPLOAD_FOLDER", "/tmp"),  # App specific
         CLIENT_ID=os.getenv("CLIENT_ID", ""),
         CLIENT_SECRET=os.getenv("CLIENT_SECRET", ""),
@@ -55,7 +58,7 @@ def create_app(test_config=None):
         API_BASE=os.getenv("API_BASE", ""),
         REDIRECT_URI=os.getenv("REDIRECT_URI", ""),
         SCOPES=os.getenv("SCOPES", "").strip(),
-    )  # invenio app configs
+    )
 
     if test_config is None:
         # load the instance config, if it exists, when not testing
@@ -64,6 +67,10 @@ def create_app(test_config=None):
         # load the test config if passed in
         app.config.from_mapping(test_config)
 
+    app.extensions["workflow_store"] = WorkflowStore(
+        ttl_seconds=app.config["WORKFLOW_TTL_SECONDS"]
+    )
+
     @app.context_processor
     def inject_base_url():
         return {

biosimdb_interface/form/extract.py

@@ -10,12 +10,12 @@
 import os
 
 from biosim_extractor.metadata.populatemetadata import MetadataPopulator
-from flask import jsonify, request, session
+from flask import current_app, jsonify, request
 from werkzeug.utils import secure_filename
 
 from . import form_bp
-from .upload import cache_extracted_files, cleanup_tmpdir
-from .utils import make_upload_tmpdir
+from .upload import cache_extracted_files
+from .workflows import WorkflowNotFound
 
 
 def extract_files_validate(top_file, traj_file):
@@ -67,13 +67,19 @@ def extract_metadata():
         - ``{"simulation_metadata": ..., "validation_errors": [...]}`` if schema validation fails.
         - ``{"error": "..."}`` with status 400 if files are missing, or 500 on unexpected error.
     """
-    # clear the existing tmpdir from previous extraction
-    tmpdir = session.get("submission_tmpdir")
-    if tmpdir:
-        cleanup_tmpdir(tmpdir)
-        session.pop("submission_tmpdir", None)
-    tmpdir = make_upload_tmpdir("biosimdb_submission_")
-    session["submission_tmpdir"] = tmpdir
+    workflow_id = request.form.get("workflow_id")
+    topology = request.files.get("topology")
+    trajectories = request.files.getlist("trajectory[]")
+
+    if not topology or not trajectories:
+        return jsonify({"error": "Simulation files are missing."}), 400
+
+    try:
+        workflow = current_app.extensions["workflow_store"].reset(workflow_id)
+    except WorkflowNotFound as exc:
+        return jsonify({"error": str(exc)}), 400
+
+    tmpdir = workflow.tmpdir
 
     try:
         topology = request.files.get("topology")
@@ -96,9 +102,7 @@ def extract_metadata():
 
         result, validation_errors = extract_files_validate(topo_path, traj_files)
 
-        if len(validation_errors) > 0:
-            cleanup_tmpdir(tmpdir)
-            session.pop("submission_tmpdir", None)
+        if validation_errors:
             return jsonify(
                 {
                     "simulation_metadata": result,
@@ -114,6 +118,7 @@ def extract_metadata():
             )
 
     except Exception as e:
+        current_app.extensions["workflow_store"].delete(workflow_id)
         print(f"ERROR: {e}")
         import traceback
 
@@ -123,9 +128,6 @@ def extract_metadata():
 
 @form_bp.route("/clear_extraction", methods=["POST"])
 def clear_extraction():
-    """Discard extracted files and reset the pending submission tmpdir."""
-    tmpdir = session.get("submission_tmpdir")
-    cleanup_tmpdir(tmpdir)
-    for key in ("submission_tmpdir", "topo_path", "traj_files"):
-        session.pop(key, None)
+    """Delete only the current tab's extracted workflow files."""
+    current_app.extensions["workflow_store"].delete(request.form.get("workflow_id"))
     return jsonify({"success": True})

biosimdb_interface/form/upload.py

@@ -36,18 +36,28 @@ def _load_pending_file_meta(tmpdir):
         return json.load(f)
 
 
+def load_extracted_files(tmpdir):
+    """Return topology and trajectory paths cached during extraction."""
+    with open(_pending_uploads_path(tmpdir)) as f:
+        saved_files = json.load(f)
+
+    topology = saved_files.get("topology", [])
+    trajectories = saved_files.get("trajectory", [])
+
+    return (
+        topology[0] if topology else None,
+        trajectories,
+    )
+
+
 def cache_extracted_files(tmpdir, saved_files):
-    """Persist saved file paths and computed file metadata for later reuse."""
+    """Persist extracted file paths and file metadata inside one workflow directory."""
     file_meta = files_metadata(saved_files)
 
     with open(_pending_uploads_path(tmpdir), "w") as f:
         json.dump(saved_files, f)
     _save_pending_file_meta(tmpdir, file_meta)
 
-    session["topo_path"] = (
-        saved_files["topology"][0] if saved_files["topology"] else None
-    )
-    session["traj_files"] = saved_files["trajectory"]
     return file_meta
 
 
@@ -191,8 +201,10 @@ def _save_request_files(tmpdir):
     Returns:
         dict[str, list[str]]: Mapping of file role to saved file paths.
     """
-    topo_path = session.get("topo_path")
-    traj_files = session.get("traj_files") or []
+    # topo_path = session.get("topo_path")
+    # traj_files = session.get("traj_files") or []
+
+    topo_path, traj_files = load_extracted_files(tmpdir)
 
     if _paths_are_reusable(tmpdir, topo_path, traj_files):
         return {
@@ -209,10 +221,6 @@ def _save_request_files(tmpdir):
                 file.save(path)
                 saved_files.setdefault(role, []).append(path)
 
-    if saved_files["topology"]:
-        session["topo_path"] = saved_files["topology"][0]
-    session["traj_files"] = saved_files["trajectory"]
-
     return saved_files
 
 
@@ -291,50 +299,19 @@ def cleanup_tmpdir(tmpdir):
         shutil.rmtree(tmpdir, ignore_errors=True)
 
 
-def save_pending_submission(json_form=None):
-    """Persist uploaded files and form payload for post-login submission resume.
-
-    Saves uploaded request files into a temp directory, writes a manifest of
-    allowed upload paths, optionally writes simulation_metadata.json, and stores
-    the form payload as pending_form_data.json.
+def save_pending_submission(json_form, tmpdir):
+    """Persist validated submission data in the given workflow directory."""
+    file_meta = _load_pending_file_meta(tmpdir) or []
 
-    Args:
-    json_form (dict | None): Validated BioSim metadata to persist. When
-    provided, file metadata is attached at json_form["files"] before
-    writing simulation_metadata.json.
-
-    Side Effects:
-    Writes JSON artifacts under tmpdir.
-    Sets session["pending_files_dir"].
-    """
-    tmpdir = session.get("submission_tmpdir")
-    # saved_files, file_meta = _save_files_and_extract_metadata(tmpdir)
-
-    topo_path = session.get("topo_path")
-    traj_files = session.get("traj_files") or []
-    saved_files = {
-        "topology": [topo_path] if topo_path else [],
-        "trajectory": traj_files,
-    }
-
-    file_meta = _load_pending_file_meta(tmpdir)
-    if file_meta is None:
-        # Fallback only if cache missing
-        file_meta = files_metadata(saved_files)
-        _save_pending_file_meta(tmpdir, file_meta)
-
-    # Persist exact user-uploaded paths for later allowlist upload
-    with open(_pending_uploads_path(tmpdir), "w") as f:
-        json.dump(saved_files, f)
-
-    if json_form is not None:
-        json_form["files"] = file_meta
-        json_path = os.path.join(tmpdir, SIM_METADATA_FILENAME)
-        with open(json_path, "w") as f:
-            json.dump(json_form, f, indent=2)
+    json_form["files"] = file_meta
+    with open(os.path.join(tmpdir, SIM_METADATA_FILENAME), "w") as f:
+        json.dump(json_form, f, indent=2)
 
     with open(_pending_form_path(tmpdir), "w") as f:
-        json.dump(request.form.to_dict(flat=False), f)
+        form_values = request.form.to_dict(flat=False)
+        form_values.pop("workflow_id", None)
+
+        json.dump(form_values, f)
 
 
 def prepare_for_invenio(form_data, tmpdir):