Migrated from using print statements in the application code to using logger module (left prints in files intended to be run as scripts)

pages/040_Test_Reporter.py

@@ -70,7 +70,7 @@ def show_stats(for_test_group: str):
         questions = list(stats[0]['q_and_a'].keys())
         num_archs = len(stats)
         with st.expander("**Request/Response Details**"):
-            print(f'Displaying {len(questions)} questions and {num_archs} architectures')
+            st.write(f'Displaying {len(questions)} questions and {num_archs} architectures')
         for q in questions:
             with st.expander(f"**{q}**"):
                 for i in range(num_archs):

src/architectures.py

@@ -7,6 +7,7 @@ config/architectures.json
 
 import chromadb
 import json
+import logging
 import os
 import regex as re
 import requests
@@ -227,11 +228,11 @@ class LogWorker(Thread):
             if request is None:
                 # There was a period of inactivity so run the timeout functions
                 for func in LogWorker.timeout_functions:
-                    print(f"LogWorker commit running {func.__name__}")
+                    logging.info(f"LogWorker commit running {func.__name__}")
                     try:
                         func()
                     except Exception as e:
-                        print(f"Timeout func {func.__name__} had error {e}")
+                        logging.error(f"Timeout func {func.__name__} had error {e}")
             else:
                 if LogWorker.commit_timer is not None and LogWorker.commit_timer.is_alive():
                     # Cancel the inactivity timer
@@ -250,7 +251,7 @@ class LogWorker(Thread):
                     if LogWorker.commit_count >= LogWorker.commit_after:
                         LogWorker.commit_repo()
                 except Exception as err:
-                    print(f"Request / trace save failed {err}")
+                    logging.error(f"Request / trace save failed {err}")
 
                 # Restart the inactivity timer
                 LogWorker.commit_timer = Timer(LogWorker.commit_time, LogWorker.signal_commit)
@@ -262,14 +263,14 @@ class LogWorker(Thread):
         If the working log file is not download, then get a local copy.
         Add the new record to the local file.
         """
-        print(f"LogWorker logging open record {LogWorker.commit_count + 1}")
+        logging.debug(f"LogWorker logging open record {LogWorker.commit_count + 1}")
         if cls.save_repo is None and not cls.save_repo_load_error:
             try:
                 hf_write_token = hf_api_token(write=True)
                 cls.save_repo = Repository(local_dir=cls.trace_dir, clone_from=cls.save_repo_url, token=hf_write_token)
             except Exception as err:
                 cls.save_repo_load_error = True
-                print(f"Error connecting to the save repo {err} - persistence now disabled")
+                logging.error(f"Error connecting to the save repo {err} - persistence now disabled")
 
         if cls.save_repo is not None:
             with open(cls.trace_file, 'r') as f:
@@ -284,7 +285,7 @@ class LogWorker(Thread):
         If there are any changes in the local file which are not committed to the repo then commit them.
         """
         if cls.commit_count > 0:
-            print(f"LogWorker committing {LogWorker.commit_count} open records")
+            logging.info(f"LogWorker committing {LogWorker.commit_count} open records")
             cls.save_repo.push_to_hub()
             LogWorker.commit_count = 0
 
@@ -293,7 +294,7 @@ class LogWorker(Thread):
         # Signalling this back via the queue and not doing the work here as it would
         # be executed on the Timer thread and may conflict with resources if the main
         # LogWorker starts doing work concurrently.
-        print("LogWorker signalling commit based on time elapsed")
+        logging.debug("LogWorker signalling commit based on time elapsed")
         cls.queue.put((None, None, None, None, None))
 
     @classmethod
@@ -350,7 +351,7 @@ class Architecture:
             cls.save_repo.push_to_hub()
         except Exception as err:
             cls.save_repo_load_error = True
-            print(f"Error connecting to the save repo {err} - persistence now disabled")
+            logging.error(f"Error connecting to the save repo {err} - persistence now disabled")
 
     @classmethod
     def get_trace_records(cls) -> List[Dict]:
@@ -363,7 +364,7 @@ class Architecture:
                 cls.save_repo = Repository(local_dir=cls.trace_dir, clone_from=cls.save_repo_url, token=hf_write_token)
             except Exception as err:
                 cls.save_repo_load_error = True
-                print(f"Error connecting to the save repo {err} - persistence now disabled")
+                logging.error(f"Error connecting to the save repo {err} - persistence now disabled")
                 return []
         with open(cls.trace_file, 'r') as f:
             test_json = json.load(f)
@@ -436,7 +437,7 @@ class Architecture:
         :param request: The architecture request to pass down the pipeline
         :return: The trace record for this invocation of the architecture
         """
-        print(f'{self.name} processing query "{request.request}"')
+        logging.info(f'{self.name} processing query "{request.request}"')
         trace = ArchitectureTrace()
         for component in self.steps:
             trace.start_trace(name=component.__class__.__name__)

src/common.py

@@ -1,3 +1,4 @@
+import logging
 import os
 import requests
 import streamlit as st
@@ -7,6 +8,8 @@ from time import time
 from typing import List, Tuple
 
 
+logging.basicConfig(level=logging.INFO, format='%(asctime)s %(levelname)s: %(message)s', datefmt='%Y-%m-%d %I:%M:%S')
+
 data_dir = os.path.join(os.path.dirname(__file__), '..', 'data')
 img_dir = os.path.join(os.path.dirname(__file__), '..', 'img')
 config_dir = os.path.join(os.path.dirname(__file__), '..', 'config')

src/testing.py

@@ -1,6 +1,7 @@
 from __future__ import annotations  # For self-referencing annotations
 
 import json
+import logging
 import os
 import shutil
 import sqlite3
@@ -37,7 +38,7 @@ class ArchitectureTestWorker(Thread):
                 if arch is None:  # None passed to signal end of test requests
                     running = False
                 else:
-                    print(f'{self.worker_name} running "{request.request}" through {arch}')
+                    logging.info(f'{self.worker_name} running "{request.request}" through {arch}')
                     architecture = Architecture.get_architecture(arch)
                     architecture(request, trace_tags=self.trace_tags, trace_comment=self.trace_comment)
             finally:
@@ -338,7 +339,7 @@ class TestGroup:
         into the class variable - for efficiency do not reload unless requested
         """
         if cls.force_load_all not in LogWorker.timeout_functions:
-            print("TestGroup adding forced refresh to LogWorker timeout")
+            logging.info("TestGroup adding forced refresh to LogWorker timeout")
             LogWorker.timeout_functions.append(TestGroup.force_load_all)
 
         if cls.all is None or reload:
@@ -429,7 +430,7 @@ def move_test_records_to_db(hf_hub_token: str) -> None:
                     con.execute(sql)
             con.commit()
         else:
-            print(f"Warning TestGroup {test_group.test_group} was not added to the DB as it already existed there")
+            logging.warning(f"TestGroup {test_group.test_group} was not added to the DB as it already existed there")
 
     def load_all_test_groups_to_db(con: sqlite3.Connection) -> None:
         """
@@ -446,7 +447,7 @@ def move_test_records_to_db(hf_hub_token: str) -> None:
     conn = get_local_db()
     load_all_test_groups_to_db(conn)
     Architecture.wipe_trace(hf_hub_token)
-    print("REMINDER: need to commit the local sqlite file to make it available to the server")
+    logging.info("REMINDER: need to commit the local sqlite file to make it available to the server")
 
 
 if __name__ == "__main__":