logdetective 0.11.1__tar.gz → 1.0.0__tar.gz

{logdetective-0.11.1 → logdetective-1.0.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: logdetective
-Version: 0.11.1
+Version: 1.0.0
 Summary: Log using LLM AI to search for build/test failures and provide ideas for fixing these.
 License: Apache-2.0
 Author: Jiri Podivin
@@ -48,14 +48,16 @@ Log Detective
 
 [PyPI Releases]: https://pypi.org/project/logdetective/#history
 
-A Python tool to analyze logs using a Language Model (LLM) and Drain template miner.
+A tool, service and RHEL process integration to analyze logs using a Large Language Model (LLM) and a [Drain template miner](https://github.com/logpai/Drain3).
+
+The service that explains logs is available here: https://logdetective.com/explain
 
 Note: if you are looking for code of website logdetective.com it is in [github.com/fedora-copr/logdetective-website](https://github.com/fedora-copr/logdetective-website).
 
 Installation
 ------------
 
-**Fedora 40+**
+**Fedora 41+**
 
     dnf install logdetective
 
@@ -70,11 +72,12 @@ First, ensure that the necessary dependencies for the `llama-cpp-python` project
 
 Then, install the `logdetective` project using pip:
 
-    # then install logdetective project
     pip install logdetective
 
 **Local repository install**
 
+Clone this repository and install with pip:
+
     pip install .
 
 Usage
@@ -111,14 +114,14 @@ Example of altered prompts:
      logdetective https://kojipkgs.fedoraproject.org//work/tasks/3367/131313367/build.log --prompts ~/my-prompts.yml
 
 
-Note that streaming with some models (notably Meta-Llama-3 is broken) is broken and can be workarounded by `no-stream` option:
+Note that streaming with some models (notably Meta-Llama-3 is broken) is broken and can be worked around by `no-stream` option:
 
     logdetective https://example.com/logs.txt --model QuantFactory/Meta-Llama-3-8B-Instruct-GGUF --no-stream
 
 
 Real Example
 ------------
-Let's have a look at a real world example. Log Detective can work with any logs though we optimize it for build logs.
+Let's have a look at a real world example. Log Detective can work with any logs though we optimize it for RPM build logs.
 
 We're going to analyze a failed build of a python-based library that happened in Fedora Koji buildsystem:
 ```
@@ -184,8 +187,13 @@ Contributing
 ------------
 
 Contributions are welcome! Please submit a pull request if you have any improvements or new features to add. Make sure your changes pass all existing tests before submitting.
+For bigger code changes, please consult us first by creating an issue.
+
+We are always looking for more annotated snippets that will increase the quality of Log Detective's results. The contributions happen in our website: https://logdetective.com/
+
+Log Detective performs several inference queries while evaluating a log file. Prompts are stored in a separate file (more info below: https://github.com/fedora-copr/logdetective?tab=readme-ov-file#system-prompts). If you have an idea for improvements to our prompts, please open a PR and we'd happy to test it out.
 
-To develop logdetective, you should fork this repository, clone your fork, and install dependencies using pip:
+To develop Log Detective, you should fork this repository, clone your fork, and install dependencies using pip:
 
     git clone https://github.com/yourusername/logdetective.git
     cd logdetective

{logdetective-0.11.1 → logdetective-1.0.0}/README.md

@@ -5,14 +5,16 @@ Log Detective
 
 [PyPI Releases]: https://pypi.org/project/logdetective/#history
 
-A Python tool to analyze logs using a Language Model (LLM) and Drain template miner.
+A tool, service and RHEL process integration to analyze logs using a Large Language Model (LLM) and a [Drain template miner](https://github.com/logpai/Drain3).
+
+The service that explains logs is available here: https://logdetective.com/explain
 
 Note: if you are looking for code of website logdetective.com it is in [github.com/fedora-copr/logdetective-website](https://github.com/fedora-copr/logdetective-website).
 
 Installation
 ------------
 
-**Fedora 40+**
+**Fedora 41+**
 
     dnf install logdetective
 
@@ -27,11 +29,12 @@ First, ensure that the necessary dependencies for the `llama-cpp-python` project
 
 Then, install the `logdetective` project using pip:
 
-    # then install logdetective project
     pip install logdetective
 
 **Local repository install**
 
+Clone this repository and install with pip:
+
     pip install .
 
 Usage
@@ -68,14 +71,14 @@ Example of altered prompts:
      logdetective https://kojipkgs.fedoraproject.org//work/tasks/3367/131313367/build.log --prompts ~/my-prompts.yml
 
 
-Note that streaming with some models (notably Meta-Llama-3 is broken) is broken and can be workarounded by `no-stream` option:
+Note that streaming with some models (notably Meta-Llama-3 is broken) is broken and can be worked around by `no-stream` option:
 
     logdetective https://example.com/logs.txt --model QuantFactory/Meta-Llama-3-8B-Instruct-GGUF --no-stream
 
 
 Real Example
 ------------
-Let's have a look at a real world example. Log Detective can work with any logs though we optimize it for build logs.
+Let's have a look at a real world example. Log Detective can work with any logs though we optimize it for RPM build logs.
 
 We're going to analyze a failed build of a python-based library that happened in Fedora Koji buildsystem:
 ```
@@ -141,8 +144,13 @@ Contributing
 ------------
 
 Contributions are welcome! Please submit a pull request if you have any improvements or new features to add. Make sure your changes pass all existing tests before submitting.
+For bigger code changes, please consult us first by creating an issue.
+
+We are always looking for more annotated snippets that will increase the quality of Log Detective's results. The contributions happen in our website: https://logdetective.com/
+
+Log Detective performs several inference queries while evaluating a log file. Prompts are stored in a separate file (more info below: https://github.com/fedora-copr/logdetective?tab=readme-ov-file#system-prompts). If you have an idea for improvements to our prompts, please open a PR and we'd happy to test it out.
 
-To develop logdetective, you should fork this repository, clone your fork, and install dependencies using pip:
+To develop Log Detective, you should fork this repository, clone your fork, and install dependencies using pip:
 
     git clone https://github.com/yourusername/logdetective.git
     cd logdetective

{logdetective-0.11.1 → logdetective-1.0.0}/logdetective/server/gitlab.py

@@ -93,7 +93,7 @@ async def process_gitlab_job_event(
     preprocessed_log.close()
 
     # check if this project is on the opt-in list for posting comments.
-    if project.name not in SERVER_CONFIG.general.packages:
+    if not is_eligible_package(project.name):
         LOG.info("Not publishing comment for unrecognized package %s", project.name)
         return
 
@@ -111,6 +111,31 @@ async def process_gitlab_job_event(
     return staged_response
 
 
+def is_eligible_package(project_name: str):
+    """Check whether the provided package name is eligible for posting
+    comments to the merge request"""
+
+    # First check the allow-list. If it's not allowed, we deny.
+    allowed = False
+    for pattern in SERVER_CONFIG.general.packages:
+        print(f"include {pattern}")
+        if re.search(pattern, project_name):
+            allowed = True
+            break
+    if not allowed:
+        # The project did not match any of the permitted regular expressions
+        return False
+
+    # Next, check the deny-list. If it was allowed before, but denied here, we deny.
+    for pattern in SERVER_CONFIG.general.excluded_packages:
+        print(f"Exclude {pattern}")
+        if re.search(pattern, project_name):
+            return False
+
+    # It was allowed and not denied, so return True to indicate it is eligible
+    return True
+
+
 class LogsTooLargeError(RuntimeError):
     """The log archive exceeds the configured maximum size"""
 
@@ -174,16 +199,17 @@ async def retrieve_and_preprocess_koji_logs(
             with artifacts_zip.open(zipinfo.filename) as task_failed_log:
                 contents = task_failed_log.read().decode("utf-8")
                 match = FAILURE_LOG_REGEX.search(contents)
-                if not match:
-                    LOG.error(
-                        "task_failed.log does not indicate which log contains the failure."
-                    )
-                    raise SyntaxError(
+                if match:
+                    failure_log_name = match.group(1)
+                    failed_arches[architecture] = PurePath(path.parent, failure_log_name)
+                else:
+                    LOG.info(
                         "task_failed.log does not indicate which log contains the failure."
                     )
-                failure_log_name = match.group(1)
-
-            failed_arches[architecture] = PurePath(path.parent, failure_log_name)
+                    # The best thing we can do at this point is return the
+                    # task_failed.log, since it will probably contain the most
+                    # relevant information
+                    failed_arches[architecture] = path
 
     if not failed_arches:
         # No failed task found in the sub-tasks.

{logdetective-0.11.1 → logdetective-1.0.0}/logdetective/server/llm.py

@@ -1,6 +1,7 @@
 import os
 import asyncio
 import json
+import random
 from typing import List, Tuple, Dict, Any, Union
 
 import backoff
@@ -102,7 +103,7 @@ def should_we_giveup(exc: aiohttp.ClientResponseError) -> bool:
     > a truthy value if the exception should not be retried
     """
     LOG.info("Should we give up on retrying error %s", exc)
-    return exc.status < 500
+    return exc.status < 400
 
 
 def we_give_up(details: backoff._typing.Details):
@@ -110,6 +111,7 @@ def we_give_up(details: backoff._typing.Details):
     retries didn't work (or we got a different exc)
     we give up and raise proper 500 for our API endpoint
     """
+    LOG.error("Last exception: %s", details["exception"])
     LOG.error("Inference error: %s", details["args"])
     raise HTTPException(500, "Request to the inference API failed")
 
@@ -117,7 +119,8 @@ def we_give_up(details: backoff._typing.Details):
 @backoff.on_exception(
     lambda: backoff.constant([10, 30, 120]),
     aiohttp.ClientResponseError,
-    max_tries=3,
+    max_tries=4,  # 4 tries and 3 retries
+    jitter=lambda wait_gen_value: random.uniform(wait_gen_value, wait_gen_value + 30),
     giveup=should_we_giveup,
     raise_on_giveup=False,
     on_giveup=we_give_up,

{logdetective-0.11.1 → logdetective-1.0.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "logdetective"
-version = "0.11.1"
+version = "1.0.0"
 description = "Log using LLM AI to search for build/test failures and provide ideas for fixing these."
 authors = ["Jiri Podivin <jpodivin@gmail.com>"]
 license = "Apache-2.0"