✨ Allow logging to disk (#237)

* WIP logging updates

Signed-off-by: John Matthews <jwmatthews@gmail.com>

* Logging to disk via podman compose working

Signed-off-by: John Matthews <jwmatthews@gmail.com>

* Tweak to log levels

Signed-off-by: John Matthews <jwmatthews@gmail.com>

* Cleanup tweaks

Signed-off-by: John Matthews <jwmatthews@gmail.com>

* Ensure 'load-data' from podman compose up can log to disk

Signed-off-by: John Matthews <jwmatthews@gmail.com>

---------

Signed-off-by: John Matthews <jwmatthews@gmail.com>

compose.yaml

@@ -4,7 +4,8 @@ x-common-variables: &kai-variables
   DEMO_MODE: "False"
   HUB_URL: ${HUB_URL:-https://tackle-konveyor-tackle.apps.example.com/hub}
   IMPORTER_ARGS:
-  LOGLEVEL: info
+  LOG_LEVEL: info
+  LOG_DIR: "/podman_compose/logs"
   NUM_WORKERS: 8
   USE_HUB_IMPORTER: ${USE_HUB_IMPORTER:-False}
 
@@ -27,7 +28,7 @@ services:
       OPENAI_API_KEY:
     image: ${IMAGE:-quay.io/konveyor/kai}:${TAG:-stable}
     volumes:
-      - ${PWD}:/podman_compose:ro,z
+      - ${PWD}:/podman_compose:rw,z
     ports:
       - "8080:8080"
     depends_on:
@@ -39,7 +40,7 @@ services:
       MODE: importer
     image: ${IMAGE:-quay.io/konveyor/kai}:${TAG:-stable}
     volumes:
-      - ${PWD}:/podman_compose:ro,z
+      - ${PWD}:/podman_compose:rw,z
     depends_on:
       - kai_db
     profiles:

example/run_demo.py

@@ -1,6 +1,7 @@
 #!/usr/bin/env python
 
 import json
+import logging
 import os
 import sys
 import time
@@ -12,8 +13,10 @@
 
 # Ensure that we have 'kai' in our import path
 sys.path.append("../../kai")
-from kai import Report
-from kai.kai_logging import KAI_LOG
+from kai.kai_logging import formatter
+from kai.report import Report
+
+KAI_LOG = logging.getLogger(__name__)
 
 SERVER_URL = "http://0.0.0.0:8080"
 APP_NAME = "coolstore"
@@ -249,7 +252,11 @@ def run_demo(report):
 
 
 if __name__ == "__main__":
-    KAI_LOG.setLevel("info".upper())
+    console_handler = logging.StreamHandler()
+    console_handler.setFormatter(formatter)
+    KAI_LOG.addHandler(console_handler)
+    KAI_LOG.setLevel("DEBUG")
+
     start = time.time()
     coolstore_analysis_dir = "./analysis/coolstore/output.yaml"
     r = Report.load_report_from_file(coolstore_analysis_dir)

kai/__init__.py

@@ -1,120 +1 @@
 __version__ = "0.0.1"
-
-from typing import Annotated, List
-
-import typer
-
-from kai.report import Report
-from kai.result import LLMResult
-from kai.service.incident_store.incident_store import IncidentStore
-
-# from typing_extensions import Annotated
-
-
-app = typer.Typer()
-
-# report_app = typer.Typer()
-# result_app = typer.Typer()
-
-# app.add_typer(report_app, name="report", help="Generate a markdown report from a raw analysis yaml.")
-# app.add_typer(result_app, name="result", help="Generate patches for given violations and incidents from an analysis yaml")
-
-
-@app.command()
-def report(analysis_path: str, output_dir: str):
-    """
-    Generate a Markdown report of a given analysis
-    YAML to be read by a human
-    """
-    report = Report.load_report_from_file(analysis_path)
-    r = dict(report)
-    print(f"We have results from {len(r.keys())} RuleSet(s) in {analysis_path}\n")
-    report.write_markdown(output_dir)
-
-
-@app.command()
-def generate(
-    path_to_report: str,
-    path_to_source: str,
-    example_initial_branch: str,
-    example_solved_branch: str,
-    path_to_output: str,
-    limit_rulesets: Annotated[List[str], typer.Option("--ruleset", "-r")] = None,
-    limit_violations: Annotated[List[str], typer.Option("--violation", "-v")] = None,
-    model: Annotated[str, typer.Option("--model", "-m")] = "gpt-3.5-turbo-16k",
-):
-    # model: Annotated[Optional[str], typer.Argument()] = "gpt-3.5-turbo-16k"):
-    """
-    Generate patches for given violations and incidents from an analysis yaml report
-
-    - path_to_report: Path to the analysis yaml report
-
-    - path_to_source: Path to the source code to be patched
-
-    - example_initial_branch: Branch name for the initial state of the source code
-
-    - example_solved_branch: Branch name for the solved state of the source code
-
-    - path_to_output: Path to the output directory for the patches
-
-    - limit_rulesets: Limit to specific rulesets (defaults to 'None', meaning to run all)
-
-    - limit_violations: Limit to specific violations (defaults to 'None', meaning to run all)
-
-    - model: Model name to use for generating the patches (defaults to 'gpt-3.5-turbo-16k', 'gpt-4-1106-preview' is another good option)
-    """
-    print(
-        f"Generating patches for {path_to_report} for example app at {path_to_source}"
-    )
-    print(f"Initial branch: {example_initial_branch}")
-    print(f"Solved branch: {example_solved_branch}")
-    print(f"Output directory: {path_to_output}")
-    print(f"Model: {model}")
-    print(f"Limit to ruleset(s): {limit_rulesets}")
-    print(f"Limit to violation(s): {limit_violations}")
-
-    llmResult = LLMResult(path_to_source, example_initial_branch, example_solved_branch)
-    llmResult.parse_report(path_to_report)
-    llmResult.process(path_to_output, model, limit_rulesets, limit_violations)
-    print(f"Completed processing, output written to {path_to_output}\n")
-
-
-@app.command()
-def load(report_path: str, output_dir: str):
-    """
-    Load the incident store with the given applications
-    write the cached_violations to a file for later use
-    """
-    incident_store = IncidentStore(report_path, output_dir)
-    incident_store.load_incident_store()
-
-
-@app.command()
-def patch(ruleset: str, violation: str, report_path: str, output_dir: str):
-    """
-    Generate patches for a specific violation
-    """
-    print(f"Generating patches for {ruleset} - {violation}")
-    incident_store = IncidentStore(report_path, output_dir)
-    patches = incident_store.get_solved_issue(ruleset, violation)
-    if len(patches) == 0:
-        print(f"No patches found for {ruleset} - {violation}")
-    else:
-        for patch in patches:
-            print(f"Patch: {patch}")
-    return patches
-
-
-@app.command()
-def common(ruleset: str, violation: str, report_path: str, output_dir: str):
-    """
-    Find common violations for a specific violation
-    """
-    print(f"Finding common violations for {ruleset} - {violation}")
-    incident_store = IncidentStore(report_path, output_dir)
-    violations = incident_store.find_common_violations(ruleset, violation)
-    if violations is None:
-        print(f"No common violations found for {ruleset} - {violation}")
-    for violation in violations:
-        print(f"Violation: {violation}")
-    return violations

kai/config.toml

@@ -1,4 +1,6 @@
 log_level = "info"
+file_log_level = "debug"
+log_dir = "$pwd/logs"
 demo_mode = false
 trace_enabled = true
 

kai/embedding_provider.py

@@ -1,3 +1,4 @@
+import logging
 import random
 from abc import ABC, abstractmethod
 from enum import Enum
@@ -9,7 +10,7 @@
 from psycopg2 import sql
 from psycopg2.extensions import connection
 
-from kai.kai_logging import KAI_LOG
+KAI_LOG = logging.getLogger(__name__)
 
 
 class TrimStrategy(Enum):

kai/hub_importer.py

@@ -1,6 +1,7 @@
 #!/usr/bin/env python
 
 import argparse
+import logging
 import os
 import pprint
 import tempfile
@@ -13,11 +14,12 @@
 from git import GitCommandError, Repo
 from pydantic import BaseModel, Field
 
-from kai.kai_logging import KAI_LOG
 from kai.models.kai_config import KaiConfig
 from kai.report import Report
 from kai.service.incident_store import Application, IncidentStore
 
+KAI_LOG = logging.getLogger(__name__)
+
 
 # BaseModel that also acts as a dict
 class KaiBaseModel(BaseModel):

kai/kai_logging.py

@@ -1,10 +1,63 @@
 import logging
+import os
 
-KAI_LOG = logging.getLogger(__name__)
-console_handler = logging.StreamHandler()
+from kai.models.kai_config import KaiConfig
+
+parent_log = logging.getLogger("kai")
+
+# console_handler = logging.StreamHandler()
 # formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
 formatter = logging.Formatter(
-    "%(levelname)s - %(asctime)s - [%(filename)20s:%(lineno)-4s - %(funcName)20s()] - %(message)s"
+    "%(levelname)s - %(asctime)s - %(name)s - [%(filename)20s:%(lineno)-4s - %(funcName)20s()] - %(message)s"
 )
-console_handler.setFormatter(formatter)
-KAI_LOG.addHandler(console_handler)
+
+
+def process_log_dir_replacements(log_dir: str):
+    ##
+    # We want to replace $pwd with the location of the Kai project directory,
+    # this is needed to help with specifying from configuration
+    ##
+    if log_dir.startswith("$pwd"):
+        log_dir = log_dir.replace(
+            "$pwd", os.path.join(os.path.dirname(os.path.realpath(__file__)), "../")
+        )
+    return log_dir
+
+
+def setup_console_handler(logger, log_level: str = "INFO"):
+    console_handler = logging.StreamHandler()
+    console_handler.setLevel(log_level)
+    console_handler.setFormatter(formatter)
+    logger.addHandler(console_handler)
+    print(f"Console logging for '{parent_log.name}' is set to level '{log_level}'")
+
+
+def setup_file_handler(
+    logger, log_file_name: str, log_dir: str, log_level: str = "DEBUG"
+):
+    # Ensure any needed log directories exist
+    log_dir = process_log_dir_replacements(log_dir)
+    log_file_path = os.path.join(log_dir, log_file_name)
+    if log_dir.startswith("$pwd"):
+        log_dir = os.path.join(os.getcwd(), log_dir[5:])
+    os.makedirs(os.path.dirname(log_file_path), exist_ok=True)
+
+    file_handler = logging.FileHandler(log_file_path)
+    file_handler.setLevel(log_level)
+    file_handler.setFormatter(formatter)
+    logger.addHandler(file_handler)
+    print(
+        f"File logging for '{logger.name}' is set to level '{log_level}' writing to file: '{log_file_path}'"
+    )
+
+
+def initLogging(console_log_level, file_log_level, log_dir, log_file="kai_server.log"):
+    setup_console_handler(parent_log, console_log_level)
+    setup_file_handler(parent_log, log_file, log_dir, file_log_level)
+    # Attempt to set the parent log level to
+    # most persmissive and allow child loggers to control what is filtered or not
+    parent_log.setLevel("DEBUG")
+
+
+def initLoggingFromConfig(config: KaiConfig):
+    initLogging(config.log_level.upper(), config.file_log_level.upper(), config.log_dir)

kai/llm_io_handler.py

@@ -1,4 +1,5 @@
 import itertools
+import logging
 import os
 import time
 import traceback
@@ -16,7 +17,6 @@
 )
 
 from kai.constants import PATH_TEMPLATES
-from kai.kai_logging import KAI_LOG
 from kai.model_provider import ModelProvider
 from kai.models.file_solution import guess_language, parse_file_solution_content
 from kai.service.incident_store.incident_store import IncidentStore
@@ -25,6 +25,8 @@
 LLM_RETRIES = 5
 LLM_RETRY_DELAY = 10
 
+KAI_LOG = logging.getLogger(__name__)
+
 
 def get_prompt(
     model_provider: ModelProvider,

kai/models/file_solution.py

@@ -1,10 +1,11 @@
+import logging
 import re
 
 from pydantic import BaseModel
 from pygments import lexers
 from pygments.util import ClassNotFound
 
-from kai.kai_logging import KAI_LOG
+KAI_LOG = logging.getLogger(__name__)
 
 
 class FileSolutionContent(BaseModel):

kai/models/kai_config.py

@@ -92,6 +92,10 @@ class KaiConfigModels(BaseModel):
 
 class KaiConfig(BaseModel):
     log_level: str = "info"
+    file_log_level: str = "info"
+    log_dir: str = os.path.join(
+        os.path.dirname(os.path.realpath(__file__)), "../../logs"
+    )
     demo_mode: bool = False
     trace_enabled: bool = False
 

kai/report.py

@@ -2,13 +2,14 @@
 
 import hashlib
 import json
+import logging
 import os
 import shutil
 from io import StringIO
 
 import yaml
 
-from kai.kai_logging import KAI_LOG
+KAI_LOG = logging.getLogger(__name__)
 
 
 class Report:

kai/result.py

@@ -1,17 +1,19 @@
 __all__ = ["LLMResult"]
 
+import logging
 import os
 
 from langchain.chains import LLMChain
 from langchain.prompts import PromptTemplate
 from langchain_community.chat_models import ChatOpenAI
 
 from kai.constants import PATH_TEMPLATES
-from kai.kai_logging import KAI_LOG
 
 from .report import Report
 from .scm import GitDiff
 
+KAI_LOG = logging.getLogger(__name__)
+
 
 class LLMResult:
     """The intent of this class is to help us form several Prompt examples using a single application

kai/scm.py

@@ -1,9 +1,10 @@
 __all__ = ["GitDiff"]
 
+import logging
 
 from git import BadName, Repo
 
-from kai.kai_logging import KAI_LOG
+KAI_LOG = logging.getLogger(__name__)
 
 
 class GitDiff: