Upload folder using huggingface_hub

home.py

@@ -33,6 +33,43 @@ load_dotenv(override=True)
 # ----- Data shaping helpers -----
 
 
+def get_status_icon(status):
+    if status == "completed":
+        status = "✅"
+    if status == "inprogress":
+        status = "⌛"
+    return status
+
+
+# Example: flatten your MongoDB document
+def flatten_run_stats(docs):
+    rows = []
+    for doc in docs:
+        latest_milestone_status = get_status_icon(
+            doc.get("milestones", [{}])[-1].get("status", "")
+        )
+        rows.append(
+            {
+                "id": str(doc.get("_id", "")),
+                "upload_date": doc.get("created_at", ""),
+                # "user_id": str(doc.get("user_id", "")),
+                # "patient_id": str(doc.get("patient_id", "")),
+                "source_files": ", ".join(doc.get("source_file_names", [])),
+                "action": doc.get("action", ""),
+                "status": get_status_icon(doc.get("status")),
+                "progress": f"{doc.get("steps_completed", 0)}/{doc.get("steps_total", 0)}",
+                # For milestones, show latest milestone name & status
+                "latest_milestone": doc.get("milestones", [{}])[-1].get(
+                    "milestone", ""
+                ),
+                "milestone_status": latest_milestone_status,
+            }
+        )
+    df = pd.DataFrame(rows)
+    df = df.fillna("-")  # 👈 normalize missing values
+    return df
+
+
 def _fmt(dt: Any) -> str:
     if isinstance(dt, (datetime,)):
         return dt.strftime("%Y-%m-%d %H:%M")
@@ -243,6 +280,7 @@ def load_patient_bundle(
     - test_names (for dropdown)
     - meta dict with patient basics
     - final_reports list
+    - run_stats list
     """
     if not patient_id:
         return pd.DataFrame(), [], {}, []
@@ -263,6 +301,7 @@ def load_patient_bundle(
     reports = db.get_reports_by_patient(patient_id)
     trends = db.get_trends_by_patient(patient_id)
     finals = db.get_final_reports_by_patient(patient_id)
+    run_stats = db.get_run_stats_by_patient(patient_id)
 
     reports_df = flatten_reports_v2(reports)
     test_names = trends_index(trends)
@@ -274,7 +313,7 @@ def load_patient_bundle(
         "Created": _fmt(patient.get("created_at")),
     }
     selected_patient_info = f'**{patient.get("name", "-")}** | **{patient.get("gender", "-")}** | Born: {patient.get("dob", "-")} | Created: {_fmt(patient.get("created_at"))}'
-    return reports_df, test_names, meta, selected_patient_info, finals
+    return reports_df, test_names, meta, selected_patient_info, finals, run_stats
 
 
 def load_trend_figure(patient_id: str, test_name: str) -> Figure:
@@ -497,7 +536,7 @@ with gr.Blocks(
                         # df_row is a dict representing the selected row
                         if df_selected_row is None:
                             return gr.update(visible=False), None
-                        report_id = df_selected_row[0] # first column is report_id
+                        report_id = df_selected_row[0]  # first column is report_id
                         report_file_name = download_final_report(report_id)
                         report_html_content = get_db().get_final_report_html(report_id)
                         # Return modal visibility + the file to bind to download button
@@ -522,9 +561,13 @@ with gr.Blocks(
                         inputs=[report_id_in],
                         outputs=download_btn,
                     )
-
                 with gr.Tab("📄 Upload History"):
                     gr.Markdown("### here is your upload history")
+                    run_stats_df = gr.DataFrame(
+                        row_count=(0, "dynamic"),
+                        wrap=True,
+                        interactive=False,
+                    )
 
     # ---- Events ----
     def on_load(email):
@@ -538,7 +581,7 @@ with gr.Blocks(
     )
 
     def on_patient_select(patient_id):
-        reports_df_val, test_names, meta, selected_patient_info, finals = (
+        reports_df_val, test_names, meta, selected_patient_info, finals, run_stats = (
             load_patient_bundle(patient_id)
         )
         return (
@@ -550,12 +593,23 @@ with gr.Blocks(
                 headers=["final_report_id", "generated_at"],
                 value=format_final_reports(finals),
             ),
+            gr.update(
+                headers=["user_id", "patient_id"],
+                value=flatten_run_stats(run_stats),
+            ),
         )
 
     patient_list.change(
         on_patient_select,
         inputs=[patient_list],
-        outputs=[meta_box, selected_patient_info, reports_df, test_dd, final_df],
+        outputs=[
+            meta_box,
+            selected_patient_info,
+            reports_df,
+            test_dd,
+            final_df,
+            run_stats_df,
+        ],
     )
 
     def on_test_change(patient_id, test_name):

modules/db.py

@@ -8,6 +8,7 @@ from dotenv import load_dotenv
 from modules.models import StandardizedReport
 import gridfs
 
+
 class SheamiDB:
     def __init__(self, uri: str, db_name: str = "sheami"):
         """Initialize connection to MongoDB Atlas (or local Mongo)."""
@@ -374,7 +375,7 @@ class SheamiDB:
         # print("updated/inserted", updated, "trends")
         return updated
 
-    def aggregate_trends_snapshot(self, patient_id : str):
+    def aggregate_trends_snapshot(self, patient_id: str):
         # fetch trends for this patient
         trend_docs = self.get_trends_by_patient(patient_id)
 
@@ -386,15 +387,18 @@ class SheamiDB:
                 continue
             # last point = most recent measurement
             last_point = max(td, key=lambda x: x.get("date"))
-            snapshots.append({
-                "test_name": t.get("test_name", ""),
-                "latest_value": last_point.get("value"),
-                "latest_date": last_point.get("date"),
-                "unit": last_point.get("unit", ""),
-                "reference_range": last_point.get("reference_range", ""),
-            })        
+            snapshots.append(
+                {
+                    "test_name": t.get("test_name", ""),
+                    "latest_value": last_point.get("value"),
+                    "latest_date": last_point.get("date"),
+                    "unit": last_point.get("unit", ""),
+                    "reference_range": last_point.get("reference_range", ""),
+                }
+            )
         return snapshots
-   # ---------------------------
+
+    # ---------------------------
     # FINAL REPORT FUNCTIONS
     # ---------------------------
     def add_final_report_v2(
@@ -436,12 +440,19 @@ class SheamiDB:
         if not doc or not doc.get("pdf_file_id"):
             return None
         return self.fs.get(doc["pdf_file_id"]).read()
-    
+
     def get_final_report_html(self, report_id: str) -> bytes | None:
         """
         Fetch the HTML for a given final_report (if stored).
         """
         doc = self.final_reports.find_one({"_id": ObjectId(report_id)})
         if not doc or not doc.get("summary"):
-            return "<html></html>"#empty tag
-        return doc.get("summary")
+            return "<html></html>"  # empty tag
+        return doc.get("summary")
+
+    def get_run_stats_by_patient(self, patient_id: str) -> list:
+        return list(
+            self.run_stats.find({"patient_id": ObjectId(patient_id)}).sort(
+                "created_at", -1
+            )  # -1 for descending order
+        )