Update app.py

app.py

@@ -55,6 +55,7 @@ TARGET_CHUNK_TOKENS = 1200
 PROMPT_RESERVE = 100
 MEDICAL_SECTION_HEADER = "=== MEDICAL SECTION ==="
 
+
 def log_system_usage(tag=""):
     try:
         cpu = psutil.cpu_percent(interval=1)
@@ -70,17 +71,21 @@ def log_system_usage(tag=""):
     except Exception as e:
         print(f"[{tag}] GPU/CPU monitor failed: {e}")
 
+
 def sanitize_utf8(text: str) -> str:
     return text.encode("utf-8", "ignore").decode("utf-8")
 
+
 def file_hash(path: str) -> str:
     with open(path, "rb") as f:
         return hashlib.md5(f.read()).hexdigest()
 
+
 def count_tokens(text: str) -> int:
     encoding = tiktoken.get_encoding(TOKENIZER)
     return len(encoding.encode(text))
 
+
 def extract_all_pages_with_token_count(file_path: str) -> Tuple[str, int, int]:
     try:
         text_chunks = []
@@ -100,6 +105,7 @@ def extract_all_pages_with_token_count(file_path: str) -> Tuple[str, int, int]:
     except Exception as e:
         return f"PDF processing error: {str(e)}", 0, 0
 
+
 def convert_file_to_json(file_path: str, file_type: str) -> str:
     try:
         h = file_hash(file_path)
@@ -133,7 +139,7 @@ def convert_file_to_json(file_path: str, file_type: str) -> str:
                 df = pd.read_excel(file_path, engine="openpyxl", header=None, dtype=str)
             except:
                 df = pd.read_excel(file_path, engine="xlrd", header=None, dtype=str)
-            content = df.fillna("").astype(str).values.tolist()
+            content = df.fillna("" ).astype(str).values.tolist()
             result = json.dumps({
                 "filename": os.path.basename(file_path),
                 "rows": content,
@@ -147,6 +153,7 @@ def convert_file_to_json(file_path: str, file_type: str) -> str:
     except Exception as e:
         return json.dumps({"error": f"Error processing {os.path.basename(file_path)}: {str(e)}"})
 
+
 def clean_response(text: str) -> str:
     text = sanitize_utf8(text)
     patterns = [
@@ -159,6 +166,7 @@ def clean_response(text: str) -> str:
         text = re.sub(pat, "", text, flags=re.DOTALL)
     return re.sub(r"\n{3,}", "\n\n", text).strip()
 
+
 def format_final_report(analysis_results: List[str], filename: str) -> str:
     report = [
         "COMPREHENSIVE CLINICAL OVERSIGHT ANALYSIS",
@@ -173,7 +181,11 @@ def format_final_report(analysis_results: List[str], filename: str) -> str:
     for res in analysis_results:
         for sec in sections:
             m = re.search(
-                rf"{re.escape(sec)}:?\s*\n(.+?)(?=\n\*|\n\n|$)",
+                rf"{re.escape(sec)}:?\s*
+(.+?)(?=
+\*|
+
+|$)",
                 res, re.IGNORECASE | re.DOTALL
             )
             if m:
@@ -193,6 +205,7 @@ def format_final_report(analysis_results: List[str], filename: str) -> str:
     report.append("END OF REPORT")
     return "\n".join(report)
 
+
 def split_content_by_tokens(content: str, max_tokens: int) -> List[str]:
     paragraphs = re.split(r"\n\s*\n", content)
     chunks, current, curr_toks = [], [], 0
@@ -217,6 +230,7 @@ def split_content_by_tokens(content: str, max_tokens: int) -> List[str]:
         chunks.append("\n\n".join(current))
     return chunks
 
+
 def init_agent():
     print("🔁 Initializing model...")
     log_system_usage("Before Load")
@@ -239,6 +253,7 @@ def init_agent():
     print("✅ Agent Ready")
     return agent
 
+
 def analyze_complete_document(content: str, filename: str, agent: TxAgent, temperature: float = 0.3) -> str:
     base_prompt = (
         "Analyze for:\n1. Critical\n2. Missed DX\n3. Med issues\n4. Gaps\n5. Follow-up\n\nContent:\n"
@@ -272,6 +287,7 @@ def analyze_complete_document(content: str, filename: str, agent: TxAgent, tempe
             print(f"Error processing chunk {i}: {e}")
     return format_final_report(results, filename)
 
+
 def create_ui(agent):
     with gr.Blocks(title="Clinical Oversight Assistant") as demo:
         gr.Markdown("""
@@ -307,7 +323,7 @@ def create_ui(agent):
             report = analyze_complete_document(combined, "+".join([os.path.basename(f.name) for f in files]), agent, temp)
             file_hash_val = hashlib.md5(combined.encode()).hexdigest()
             path = os.path.join(report_dir, f"{file_hash_val}_report.txt")
-            with open(path, "w") as rd:
+            with open(path, "w", encoding="utf-8") as rd:
                 rd.write(report)
             yield report, path, "✅ Analysis complete!", previews
         send_btn.click(analyze, [file_upload, msg_input, temperature], [report_output, download_output, status, data_preview])
@@ -326,5 +342,6 @@ if __name__ == "__main__":
         server_name="0.0.0.0",
         server_port=7860,
         show_error=True,
-        share=False
+        share=False,
+        allowed_paths=[report_dir]
     )