Upload folder using huggingface_hub

.gitignore

@@ -14,3 +14,4 @@ output/
 data/jg_2/
 app.py
 data/vasudevan/
+data/Srinivas/

graph.py

@@ -309,6 +309,22 @@ async def fn_unit_normalizer_node(state: SheamiState):
     return state
 
 
+async def fn_db_update_node(state: SheamiState):
+    ## add parsed reports
+    report_id_list = await get_db().add_report_v2(
+        patient_id=state["patient_id"],
+        reports=state["standardized_reports"],
+        run_id=state["run_id"],
+    )
+    state["report_id_list"] = report_id_list
+
+    logger.info("report_id_list = %s", report_id_list)
+    for report_id in report_id_list.split(","):
+        await get_db().aggregate_trends_from_report(state["patient_id"], report_id)
+
+    return state
+
+
 async def fn_trends_aggregator_node(state: SheamiState):
     logger.info("%s| Aggregating Trends : started", state["thread_id"])
     state["messages"].append("Aggregating Trends : started")
@@ -434,6 +450,8 @@ async def fn_interpreter_node(state: SheamiState):
         indent=1,
     )
 
+    # logger.info("llm_input = %s", llm_input)
+
     # 1. LLM narrative
     messages = [
         SystemMessage(
@@ -444,12 +462,63 @@ async def fn_interpreter_node(state: SheamiState):
                 "2. The names of the reports used to summarize this information."
                 "3. Patient summary (patient id, name, age, sex if available)"
                 "4. Test window (mention the from and to dates)"
-                "5. Trend summaries (tables with Test Name, Latest Value, Highest Value, Lowest Value, Unit, Reference Range, Trend Direction and Inference) "
-                "6. Clinical insights. "
-                "For inference column, use ✅ for normal, ▲ for high, and ▼ for low. "
-                "For trend direction, use appropriate unicode  icons like up arrow (improving trend) , down arrow (worsening trend) or checkmark if determined normal"
-                "Format tables in proper <table> with <tr>, <th>, <td>. "
-                "Do not include charts, they will be programmatically added."
+                """
+                5. Trend summaries
+                Generate tables with the following columns:
+
+                - Test Name
+                - Latest Value 1, Latest Value 2, Latest Value 3 (use a hyphen "–" if a value is missing)
+                - Unit
+                - Reference Range
+                - Inference (latest value only): ✅ if within normal range, ▲ if above normal (high), ▼ if below normal (low)
+                - Trend Direction (across last 3 values): ⬆️ if values are rising, ⬇️ if values are falling, ➖ (or ✅) if stable/normal
+                """
+                "6. Clinical insights. \n"
+                "\nImportant Rules:\n"
+                "- Format tables in proper <table> with <tr>, <th>, <td>. "
+                "- Do not include charts, they will be programmatically added."
+                """
+5. Trend summaries
+Generate HTML tables with the following structure and formatting rules:
+
+Columns:
+- Test Name
+- Latest Value 1, Latest Value 2, Latest Value 3 (use a hyphen "–" if a value is missing)
+- Unit
+- Reference Range
+- Inference (latest value only): ✅ if within normal range, ▲ if above normal (high), ▼ if below normal (low)
+- Trend Direction (across last 3 values): ⬆️ if values are rising, ⬇️ if values are falling, ➖ (or ✅) if stable/normal
+
+Formatting requirements:
+- The HTML will be shown in a UI (`gr.HTML`) and also rendered to PDF via WeasyPrint.
+- The table must ALWAYS fit within 100% of the container width. Do not allow horizontal scrolling, clipping, or overlapping columns.
+- Use `table-layout: fixed;` and `<colgroup>` with percentage widths that sum to 100%.
+- Allow text wrapping inside cells so narrow columns still display all content.
+- Example CSS to embed at the top of the HTML:
+
+<style>
+  table { width: 100%; border-collapse: collapse; table-layout: fixed; }
+  col { }
+  th, td {
+    font-size: 11px;
+    padding: 4px 6px;
+    white-space: normal;
+    word-break: break-word;
+  }
+</style>
+
+- Example `<colgroup>` (adjust if needed):
+<colgroup>
+  <col style="width:20%"> <!-- Test Name -->
+  <col style="width:8%">  <!-- Latest Value 1 -->
+  <col style="width:8%">  <!-- Latest Value 2 -->
+  <col style="width:8%">  <!-- Latest Value 3 -->
+  <col style="width:8%">  <!-- Unit -->
+  <col style="width:16%"> <!-- Reference Range -->
+  <col style="width:16%"> <!-- Inference -->
+  <col style="width:16%"> <!-- Trend Direction -->
+</colgroup>
+                """
             )
         ),
         HumanMessage(content=llm_input),
@@ -544,16 +613,6 @@ async def fn_final_cleanup_node(state: SheamiState):
 
     await get_db().update_run_stats(run_id=state["run_id"], status="completed")
 
-    ## add parsed reports
-    report_id_list = await get_db().add_report_v2(
-        patient_id=state["patient_id"], reports=state["standardized_reports"]
-    )
-    state["report_id_list"] = report_id_list
-
-    logger.info("report_id_list = %s", report_id_list)
-    for report_id in report_id_list.split(","):
-        await get_db().aggregate_trends_from_report(state["patient_id"], report_id)
-
     # add final report
     # Save PDF along with metadata
     with open(pdf_path, "rb") as f:
@@ -642,6 +701,7 @@ def create_graph(user_email: str, patient_id: str, thread_id: str):
     workflow.add_node("increment_index", fn_increment_index_node)
     workflow.add_node("testname_standardizer", fn_testname_standardizer_node)
     workflow.add_node("unit_normalizer", fn_unit_normalizer_node)
+    workflow.add_node("db_update_node", fn_db_update_node)
     workflow.add_node("trends", fn_trends_aggregator_node)
     workflow.add_node("interpreter", fn_interpreter_node)
 
@@ -672,7 +732,8 @@ def create_graph(user_email: str, patient_id: str, thread_id: str):
     workflow.add_edge("testname_standardizer_notifier", "testname_standardizer")
     workflow.add_edge("testname_standardizer", "unit_normalizer_notifier")
     workflow.add_edge("unit_normalizer_notifier", "unit_normalizer")
-    workflow.add_edge("unit_normalizer", "trends_notifier")
+    workflow.add_edge("unit_normalizer", "db_update_node")
+    workflow.add_edge("db_update_node", "trends_notifier")
     workflow.add_edge("trends_notifier", "trends")
     workflow.add_edge("trends", "interpreter_notifier")
     workflow.add_edge("interpreter_notifier", "interpreter")

modules/db.py

@@ -93,7 +93,7 @@ class SheamiDB:
     # REPORT FUNCTIONS
     # ---------------------------
     async def add_report_v2(
-        self, patient_id: str, reports: list[StandardizedReport]
+        self, patient_id: str, reports: list[StandardizedReport], run_id: str
     ) -> str:
         inserted_ids: list[ObjectId] = []
         for parsed_data in reports:
@@ -102,6 +102,7 @@ class SheamiDB:
                 "uploaded_at": datetime.now(timezone.utc),
                 "file_name": parsed_data.original_report_file_name,
                 "parsed_data_v2": parsed_data.model_dump(),
+                "run_id" : ObjectId(run_id),
             }
             result = await self.reports.insert_one(report)
             inserted_ids.append(result.inserted_id)