refactor(analyzers): consolidate common functionality into BaseAnalyzer class

Author: sumansaurabh

penify_hook/base_analyzer.py

@@ -0,0 +1,20 @@
+import os
+from git import Repo
+from .api_client import APIClient
+from penify_hook.utils import get_repo_details, recursive_search_git_folder
+
+
+class BaseAnalyzer:
+    
+    def __init__(self, folder_path: str, api_client: APIClient):
+        self.folder_path = folder_path
+        self.repo_path = recursive_search_git_folder(folder_path)
+        self.repo = None
+        self.repo_details = None
+        if self.folder_path:
+            self.repo = Repo(self.repo_path)
+            self.repo_details = get_repo_details(self.repo)
+
+        self.relative_file_path = os.path.relpath(folder_path)
+        self.api_client = api_client
+        self.supported_file_types = set(api_client.get_supported_file_types())

penify_hook/commands/commit_commands.py

@@ -1,4 +1,7 @@
+import os
 import sys
+
+from penify_hook.utils import recursive_search_git_folder
 from ..commit_analyzer import CommitDocGenHook
 from ..api_client import APIClient
 
@@ -13,7 +16,7 @@
 except ImportError:
     JiraClient = None
 
-def commit_code(api_url, gf_path, token, message, open_terminal, 
+def commit_code(api_url, token, message, open_terminal, 
                llm_model=None, llm_api_base=None, llm_api_key=None,
                jira_url=None, jira_user=None, jira_api_token=None):
     """
@@ -56,6 +59,7 @@ def commit_code(api_url, gf_path, token, message, open_terminal,
 
     try:
         # Pass the LLM client and JIRA client to CommitDocGenHook
+        gf_path = recursive_search_git_folder(os.getcwd())
         analyzer = CommitDocGenHook(gf_path, api_client, llm_client, jira_client)
         analyzer.run(message, open_terminal)
     except Exception as e:

penify_hook/commit_analyzer.py

@@ -5,83 +5,16 @@
 from typing import Optional, List
 from git import Repo
 from tqdm import tqdm
+
+from penify_hook.base_analyzer import BaseAnalyzer
 from .api_client import APIClient
 
-class CommitDocGenHook:
+class CommitDocGenHook(BaseAnalyzer):
     def __init__(self, repo_path: str, api_client: APIClient, llm_client=None, jira_client=None):
-        self.repo_path = repo_path
-        self.api_client = api_client
+        super().__init__(repo_path, api_client)
+
         self.llm_client = llm_client  # Add LLM client as an optional parameter
         self.jira_client = jira_client  # Add JIRA client as an optional parameter
-        self.repo = Repo(repo_path)
-        self.supported_file_types = set(self.api_client.get_supported_file_types())
-        self.repo_details = self.get_repo_details()
-
-    def get_repo_details(self):
-        """Get the details of the repository, including the hosting service,
-        organization name, and repository name.
-
-        This method checks the remote URL of the repository to determine whether
-        it is hosted on GitHub, Azure DevOps, Bitbucket, GitLab, or another
-        service. It extracts the organization (or user) name and the repository
-        name from the URL. If the hosting service is not recognized, it will
-        return "Unknown Hosting Service". The method handles potential errors
-        during the extraction process and returns a dictionary with the relevant
-        details.
-
-        Returns:
-            dict: A dictionary containing the organization name, repository name, and
-                hosting service.
-        """
-        remote_url = None
-        hosting_service = "Unknown"
-        org_name = None
-        repo_name = None
-
-        try:
-            # Get the remote URL
-            remote = self.repo.remotes.origin.url
-            remote_url = remote
-
-            # Determine the hosting service based on the URL
-            if "github.com" in remote:
-                hosting_service = "GITHUB"
-                match = re.match(r".*github\.com[:/](.*?)/(.*?)(\.git)?$", remote)
-            elif "dev.azure.com" in remote:
-                hosting_service = "AZUREDEVOPS"
-                match = re.match(r".*dev\.azure\.com/(.*?)/(.*?)/_git/(.*?)(\.git)?$", remote)
-            elif "visualstudio.com" in remote:
-                hosting_service = "AZUREDEVOPS"
-                match = re.match(r".*@(.*?)\.visualstudio\.com/(.*?)/_git/(.*?)(\.git)?$", remote)
-            elif "bitbucket.org" in remote:
-                hosting_service = "BITBUCKET"
-                match = re.match(r".*bitbucket\.org[:/](.*?)/(.*?)(\.git)?$", remote)
-            elif "gitlab.com" in remote:
-                hosting_service = "GITLAB"
-                match = re.match(r".*gitlab\.com[:/](.*?)/(.*?)(\.git)?$", remote)
-            else:
-                hosting_service = "Unknown Hosting Service"
-                match = None
-
-            if match:
-                org_name = match.group(1)
-                repo_name = match.group(2)
-                
-                # For Azure DevOps, adjust the group indices
-                if hosting_service == "AZUREDEVOPS":
-                    repo_name = match.group(3)
-
-        except Exception as e:
-            print(f"Error determining repo details: {e}")
-
-        org_name = org_name if org_name else "UNKNOWN_ORG"
-        repo_name = repo_name if repo_name else "UNKNOWN_REPO"
-
-        return {
-            "organization_name": org_name,
-            "repo_name": repo_name,
-            "vendor": hosting_service
-        }
 
     def get_summary(self, instruction: str):
         """Generate a summary for the commit based on the staged changes.

penify_hook/file_analyzer.py

@@ -4,6 +4,7 @@
 from tqdm import tqdm
 import time
 
+from penify_hook.base_analyzer import BaseAnalyzer
 from penify_hook.utils import get_repo_details, recursive_search_git_folder
 from .api_client import APIClient
 import logging
@@ -16,15 +17,12 @@
 # Set up logger
 logger = logging.getLogger(__name__)
 
-class FileAnalyzerGenHook:
+class FileAnalyzerGenHook(BaseAnalyzer):
     def __init__(self, file_path: str, api_client: APIClient):
         self.file_path = file_path
-        self.repo_path = recursive_search_git_folder(file_path)
-        self.repo = Repo(self.repo_path)
-        self.repo_details = get_repo_details(self.repo)
-        self.relative_file_path = os.path.relpath(file_path)
-        self.api_client = api_client
-        self.supported_file_types = set(self.api_client.get_supported_file_types())
+        super().__init__(file_path, api_client)
+        
+
 
     def process_file(self, file_path, pbar):
         """Process a file by reading its content and sending it to an API for

penify_hook/folder_analyzer.py

@@ -1,13 +1,15 @@
 import os
 from git import Repo
+
+from penify_hook.base_analyzer import BaseAnalyzer
 from .api_client import APIClient
 from .file_analyzer import FileAnalyzerGenHook
 from tqdm import tqdm
 
-class FolderAnalyzerGenHook:
+class FolderAnalyzerGenHook(BaseAnalyzer):
     def __init__(self, dir_path: str, api_client: APIClient):
         self.dir_path = dir_path
-        self.api_client = api_client
+        super().__init__(dir_path, api_client)
 
     def list_all_files_in_dir(self, dir_path: str):
         """List all files in a directory and its subdirectories.

penify_hook/git_analyzer.py

@@ -3,6 +3,7 @@
 from git import Repo
 from tqdm import tqdm
 
+from penify_hook.base_analyzer import BaseAnalyzer
 from penify_hook.utils import get_repo_details, recursive_search_git_folder
 from .api_client import APIClient
 import logging
@@ -15,13 +16,9 @@
 # Set up logger
 logger = logging.getLogger(__name__)
 
-class GitDocGenHook:
+class GitDocGenHook(BaseAnalyzer):
     def __init__(self, repo_path: str, api_client: APIClient):
-        self.repo_path = recursive_search_git_folder(repo_path)
-        self.api_client = api_client
-        self.repo = Repo(self.repo_path)
-        self.supported_file_types = set(self.api_client.get_supported_file_types())
-        self.repo_details = get_repo_details(self.repo)
+        super().__init__(repo_path, api_client)
 
     def get_modified_files_in_last_commit(self):
         """Get the list of files modified in the last commit.

penify_hook/main.py

@@ -63,20 +63,21 @@ def main():
     parser.add_argument_group(advanced_title)
 
     # ===== BASIC COMMANDS (No login required) =====
+
+    
+    commit_parser_description = """
+It generates smart commit messages. By default, it will just generate just the Title of the commit message.
+
+1. If you have not configured LLM, it will give an error. You either need to configure LLM or use the API key.
+2. If you have not configured JIRA. It will not enhance the commit message with JIRA issue details.
+3. For more information, visit https://docs.penify.dev/
+"""
 
     # Subcommand: commit
-    commit_parser = subparsers.add_parser("commit", help="Generate smart commit messages using local-LLM(no login required).")
-    commit_parser.add_argument("-gf", "--git_folder_path", help="Path to the folder with git.", default=os.getcwd())
+    commit_parser = subparsers.add_parser("commit", help="Generate smart commit messages using local-LLM(no login required).", description=commit_parser_description, formatter_class=argparse.RawDescriptionHelpFormatter)
     commit_parser.add_argument("-m", "--message", required=False, help="Commit with contextual commit message.", default="N/A")
-    commit_parser.add_argument("-e", "--terminal", required=False, help="Open edit terminal", default="False")
-    # Add LLM options
-    commit_parser.add_argument("--llm", "--llm-model", dest="llm_model", help="LLM model to use")
-    commit_parser.add_argument("--llm-api-base", help="API base URL for the LLM service")
-    commit_parser.add_argument("--llm-api-key", help="API key for the LLM service")
-    # Add JIRA options
-    commit_parser.add_argument("--jira-url", help="JIRA base URL")
-    commit_parser.add_argument("--jira-user", help="JIRA username or email")
-    commit_parser.add_argument("--jira-api-token", help="JIRA API token")
+    commit_parser.add_argument("-e", "--terminal", action="store_true", help="Open edit terminal before committing.")
+    commit_parser.add_argument("-d", "--description", action="store_false", help="It will generate commit message with title and description.")
 
     # Subcommand: config
     config_parser = subparsers.add_parser("config", help="Configure local-LLM and JIRA.")
@@ -153,7 +154,7 @@ def main():
 
     elif args.subcommand == "commit":
         # For commit, token is now optional - some functionality may be limited without it
-        open_terminal = args.terminal.lower() == "true"
+        open_terminal = args.terminal
 
         # Get LLM configuration
         llm_model = args.llm_model
@@ -179,7 +180,7 @@ def main():
             jira_user = jira_user or jira_config.get('username')
             jira_api_token = jira_api_token or jira_config.get('api_token')
 
-        commit_code(API_URL, args.git_folder_path, token, args.message, open_terminal,
+        commit_code(API_URL, token, args.message, open_terminal,
                    llm_model, llm_api_base, llm_api_key,
                    jira_url, jira_user, jira_api_token)
 

penify_hook/utils.py

@@ -90,6 +90,9 @@ def recursive_search_git_folder(folder_path):
     if os.path.isdir(folder_path):
         if '.git' in os.listdir(folder_path):
             return folder_path
+        # reached the root of the filesystem
+        elif folder_path == os.path.dirname(folder_path):
+            return None
         else:
             return recursive_search_git_folder(os.path.dirname(folder_path))