first commit

sftp/file_monitor.py

@@ -0,0 +1,95 @@
+#!/usr/bin/env python3
+"""
+File monitoring and discovery for UPI files.
+Scans SFTP directories for new files across multiple banks.
+
+Filename convention handled:
+    <LOCATION>_DDMMYYYY_<CODE>_UPI.txt.gz
+
+Example:
+    TUMLUK_31052026_8C_UPI.txt.gz
+"""
+
+import re
+from typing import List, Tuple, Dict
+from logging_config import get_logger
+from config import get_config
+from .sftp_client import SFTPClient
+
+logger = get_logger(__name__)
+
+
+class FileMonitor:
+    """Monitors SFTP for new UPI files."""
+
+    def __init__(self, sftp_client: SFTPClient = None):
+        self.config = get_config()
+        self.sftp_client = sftp_client or SFTPClient()
+
+    def scan_for_new_files(self, processed_filenames: List[str]) -> List[Tuple[str, str, str]]:
+        """
+        Scan all bank directories for new UPI files.
+
+        Returns:
+            List of (filename, bankcode, full_remote_path)
+        """
+        new_files: List[Tuple[str, str, str]] = []
+
+        for bank_code in self.config.bank_codes:
+            remote_path = f"{self.config.sftp_base_path}/{bank_code}/UPI"
+
+            # ✅ Pattern for UPI files
+            files = self.sftp_client.list_files(
+                remote_path,
+                pattern='*_UPI.txt.gz'
+            )
+
+            for filename in files:
+                if filename not in processed_filenames:
+                    full_path = f"{remote_path}/{filename}"
+                    new_files.append((filename, bank_code, full_path))
+                    logger.info(f"Found new UPI file: {filename} (bank: {bank_code})")
+                else:
+                    logger.debug(f"Skipping already processed UPI file: {filename}")
+
+        logger.info(f"UPI scan complete: Found {len(new_files)} new files")
+        return new_files
+
+    @staticmethod
+    def parse_filename(filename: str) -> Dict[str, str]:
+        """
+        Parse UPI filename.
+
+        Expected format:
+            <LOCATION>_DDMMYYYY_<CODE>_UPI.txt.gz
+
+        Example:
+            TUMLUK_31052026_8C_UPI.txt.gz
+        """
+
+        pattern = r'^([A-Z]+)_(\d{2})(\d{2})(\d{4})_([A-Z0-9]+)_UPI\.txt\.gz$'
+        match = re.match(pattern, filename, flags=re.IGNORECASE)
+
+        if not match:
+            logger.warning(f"Could not parse UPI filename: {filename}")
+            return {}
+
+        location, day, month, year, code = match.groups()
+
+        return {
+            'filename': filename,
+            'location': location,
+            'day': day,
+            'month': month,
+            'year': year,
+            'code': code,
+            'timestamp': f"{day}/{month}/{year}"
+        }
+
+    def __enter__(self):
+        if not self.sftp_client.sftp:
+            self.sftp_client.connect()
+        return self
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        self.sftp_client.disconnect()