Merge pull request #4 from cedadev/origin/ioana.circu-logging

ioana.circu - logging

Author: lr154lrose

.gitignore

@@ -3,4 +3,4 @@ settings.json
 local/
 localcache/
 *pyc
-build/
+build/

dirconfig

@@ -1,4 +1,6 @@
 # Directory containing new flights to be pushed to Elasticsearch
-/home/badc/software/datasets/flight-finder/add_records/
+/home/users/icircu/flight-pipeline/add/
 # Directory for moving written flights - write DELETE to remove pushed flight records from the local system after pushing.
-/home/badc/software/datasets/flight-finder/stac-flightfinder-items/
+/home/users/icircu/test-flights/
+# Logging File
+/home/users/icircu/test-flights/logging

flight_update.py

@@ -8,6 +8,7 @@
 import importlib
 
 import logging
+from flightpipe.logger import logger
 
 import argparse
 
@@ -18,7 +19,6 @@
 
 settings_file = 'settings.json'
 
-
 def openConfig():
     """
     Function to open configuration file and initialise paths to relevant directories.
@@ -31,14 +31,15 @@ def openConfig():
 
     if VERB:
         print('> (1/6) Opening Config File')
-        logging.info('> (1/6) Opening Config File')
+        logger.info('> (1/6) Opening Config File')
 
     f = open('dirconfig','r')
     content = f.readlines()
     f.close()
     try:
-        return content[1].replace('\n',''), content[3].replace('\n',''), content[5].replace('\n','')
+        return content[1].replace('\n',''), content[3].replace('\n','')
     except IndexError:
+        logger.error('One or both paths missing from the dirconfig file')
         print('Error: One or both paths missing from the dirconfig file - please fill these in')
         return '',''
 
@@ -73,7 +74,7 @@ def addFlights(rootdir, archive, repush=False):
     # ES client to determine array of ids
     if VERB:
         print('> (2/6) Setting up ES Flight Client')
-        logging.info('> (2/6) Setting up ES Flight Client')
+        logger.info('> (2/6) Setting up ES Flight Client')
     if repush:
         files_list = os.listdir(archive)
         fclient = ESFlightClient(archive, settings_file)
@@ -87,21 +88,21 @@ def addFlights(rootdir, archive, repush=False):
     # Push new flights to index
     if VERB:
         print('> (4/6) Identified {} flights'.format(len(checked_list)))
-        logging.info('> (4/6) Identified {} flights'.format(len(checked_list)))
+        logger.info('> (4/6) Identified {} flights'.format(len(checked_list)))
     if len(checked_list) > 0:
         fclient.push_flights(checked_list)
         if VERB:
             print('> (5/6) Pushed flights to ES Index')
-            logging.info('> (5/6) Pushed flights to ES Index')
+            logger.info('> (5/6) Pushed flights to ES Index')
         if not repush:
             moveOldFiles(rootdir, archive, checked_list)
         if VERB:
             print('> (6/6) Removed local files from push directory')
-            logging.info('> (6/6) Removed local files from push directory')
+            logger.info('> (6/6) Removed local files from push directory')
     else:
         if VERB:
             print('> Exiting flight pipeline')
-            logging.info('> Exiting flight pipeline')
+            logger.info('> Exiting flight pipeline')
 
     # Move old records into an archive directory
 
@@ -136,27 +137,19 @@ def main():
     REPUSH = False
 
     if args.mode == 'add':
-        root, archive, log_file = openConfig()
-
-        if log_file == '':
-            print("Error: Please fill in the third directory in dirconfig file")
-
-        # Set up logging config
-        logging.basicConfig(
-            level=logging.DEBUG,  # Capture all levels
-            format='%(asctime)s - %(levelname)s - %(message)s',  # timestamp, level, message
-            handlers=[
-                logging.FileHandler(log_file),  # Write output to file
-                logging.StreamHandler()  # If logging to console
-            ]
-        )
+        logger.debug("Mode set to add")
+        root, archive = openConfig()
+
+        logger.debug("Root directory set to %s", root)
+        logger.debug("Archive set to %s", archive)
+
         if archive == '':
             print('Error: Please fill in second directory in dirconfig file')
-            logging.error("Error: Second directory in dirconfig file missing")
+            logger.error("Second directory in dirconfig file missing")
             sys.exit()
         elif root == '':
             print('Error: Please fill in first directory in dirconfig file')
-            logging.error("Error: First directory in dirconfig file missing")
+            logger.error("First directory in dirconfig file missing")
             sys.exit()
         else:
             addFlights(root, archive, repush=REPUSH)
@@ -171,14 +164,18 @@ def main():
         """
 
     elif args.mode == 'update':
+        logger.debug("Mode set to update")
         updateFlights(args.update)
 
     elif args.mode == 'add_moles':
+        logger.debug("Mode set to add moles")
         updateFlights('moles')
 
     elif args.mode == 'reindex':
+        logger.debug("Mode set to reindex")
         reindex(args.new_index)
     else:
+        logger.error("Mode unrecognised - ", args.mode)
         print('Error: Mode unrecognised - ', args.mode)
         sys.exit()
 

flightpipe/flight_client.py

@@ -11,7 +11,10 @@
 from elasticsearch import Elasticsearch
 from elasticsearch.helpers import bulk
 
-from ceda_elastic_py import SimpleClient, gen_id
+from flightpipe.logger import logger
+
+from flightpipe.simple_client import SimpleClient, gen_id
+from flightpipe.logger import setup_logging
 
 from datetime import datetime
 
@@ -22,6 +25,7 @@
 urllib3.disable_warnings()
 
 def resolve_link(path, ):
+    logger.debug("Debug: Resolving link for path %s", path)
     mpath = str(path)
 
     uuid = None
@@ -32,26 +36,31 @@ def resolve_link(path, ):
             r = json.loads(resp)
             if r['results']:
                 uuid = r['results'][0]['uuid']
+                logger.debug("Debug: Reslolving link, found uuid %s", str(uuid))
         except:
             print(f'Unsuccessful link retrieval for {path} - proceeding without')
         path = '/'.join(path.split('/')[:-1])
 
     if not uuid:
+        logger.error("Error: Recursive path search failed for %s", mpath)
         print(f'Recursive path search failed for: {mpath}')
 
     return uuid
 
 class ESFlightClient(SimpleClient):
     """
     Connects to an elasticsearch instance and exports the
-    documents to elasticsearch."""
+    documents to elasticsearch.
+    """
 
     def __init__(self, rootdir, es_config='settings.json'):
         self.rootdir = rootdir
+        logger.info("Info: Initialising ES Flight Client")
 
-        super().__init__("stac-flightfinder-items", es_config=es_config)
+        super().__init__("stac-flightfinder-items-test", es_config=es_config)
 
         with open('stac_template.json') as f:
+            logger.info("Info: Reading stac templace JSON file")
             self.required_keys = json.load(f).keys()
 
     def push_flights(self, file_list):
@@ -63,10 +72,13 @@ def push_flights(self, file_list):
         elif isinstance(file_list[0], dict):
             flight_list = file_list
         else:
+            logger.error("Error: Flight file not found %s", str(file_list[0]))
             raise FileNotFoundError(file_list[0])
+        logger.info("Info: Flights to be pushed %s", str(flight_list))
         self.push_records(flight_list)
 
     def preprocess_records(self, file_list):
+        logger.debug("Debug: Processing following records - %s", file_list)
 
         def set_defaults(refs):
             collection = refs['collection']
@@ -116,6 +128,7 @@ def set_defaults(refs):
                 if rq not in source:
                     missing.append(rq)
             if len(missing) > 0:
+                logger.error("Error: File is missing entries - %s", str(missing))
                 raise TypeError(f"File {file} is missing entries:{missing}")
 
             source['last_update'] = datetime.strftime(datetime.now(),'%Y-%m-%d %H:%M:%S')
@@ -126,6 +139,7 @@ def set_defaults(refs):
         return records
 
     def obtain_field(self, id, fieldnames):
+        logger.info("Info: Performing query to obtain the following fields: %s", str(fieldnames))
         search = {
             "_source": fieldnames,
             "query": {
@@ -140,11 +154,14 @@ def obtain_field(self, id, fieldnames):
             body=search)
 
         try:
+            logger.info("Info: Found following fields: %s", str(resp['hits']['hits'][0]))
             return resp['hits']['hits'][0]
         except IndexError: # No entry found
+            logger.error("Error: No entry found.")
             return None
 
     def add_field(self, id, data, fieldname):
+        logger.debug("Debug: Update mapping for id - %s", str(id))
         # Update mapping
         self.es.update(index=self.index, doc_type='_doc', id=id, body={'doc':{fieldname:data}})
 
@@ -227,6 +244,7 @@ def check_ptcode(self, ptcode):
         return 100
 
     def reindex(self, new_index):
+        logger.debug("Debug: Reindex for source %s and destination %s", self.index, new_index)
 
         self.es.reindex({
             "source":{

flightpipe/logger.py

@@ -0,0 +1,50 @@
+import logging
+
+def setup_logging(enable_logging=True, console_logging=True):
+    """
+    Sets up logging configuration. If `enable_logging` is False, no logging will occur.
+    
+    :param enable_logging: Flag to enable/disable logging.
+    """
+    file = "dirconfig"
+
+    with open(file) as f: # 'r' is default if not specified.
+        content = [r.strip() for r in f.readlines()] # Removes the '\n' from all lines
+
+    log_file = content[5].replace('\n','')
+
+    if log_file == '':
+        print("Error: Please fill in the third directory in dirconfig file")
+
+    handlers = [
+            logging.FileHandler(log_file),  # Write output to file
+        ]
+
+    if console_logging:
+        handlers.append(logging.StreamHandler())   # Logs to the console if enabled
+
+
+    if enable_logging:
+        logging.basicConfig(
+            level=logging.DEBUG, # Capture all levels
+            format='%(asctime)s - %(levelname)s - %(message)s',
+            handlers=handlers
+        )
+    else:
+        # Disable logging by setting a null handler
+        logging.basicConfig(level=logging.CRITICAL)
+        #NOTSET for no alerts at all
+
+
+enable_logging = True
+
+# Set up logging with a flag (True to enable logging, False to disable logging)
+setup_logging(enable_logging)  # Change to False to disable logging
+
+logger = logging.getLogger(__name__)
+
+
+
+__author__    = "Ioana Circu"
+__contact__   = "[email protected]"
+__copyright__ = "Copyright 2025 United Kingdom Research and Innovation"