Spaces:

vsagar100
/

codevista

Sleeping

Update app.py

80e7d63 verified about 2 years ago

4.85 kB

# app.py

# Import necessary modules from lib
from lib.code_reviewer import CodeReviewer, ReviewManager
import streamlit as st
import os
import glob
import json
from datasets import DatasetDict

# Sample Training Data for Ansible Code Standards
sample_data = [
    {
        "prompt": "Review the following Ansible script thoroughly according to best practices:\n\n- Avoid using hard-coded values\n- Ensure idempotency\n- Proper error handling\n\nCode:\n---\n- name: Install nginx\n  hosts: web\n  tasks:\n    - name: Install nginx package\n      apt:\n        name: nginx\n        state: present",
        "response": "Ensure that the playbook includes proper error handling for the package installation. Consider using a retry mechanism and notify handlers if required. Avoid hard-coding package names where possible."  
    },
    {
        "prompt": "Review the following Ansible script thoroughly according to best practices:\n\n- Avoid using hard-coded values\n- Ensure idempotency\n- Use variables\n\nCode:\n---\n- name: Create a directory\n  hosts: all\n  tasks:\n    - name: Create /var/www directory\n      file:\n        path: /var/www\n        state: directory",
        "response": "It is recommended to use variables for paths such as '/var/www' to ensure flexibility. The task is idempotent, which is good, but consider adding more context to describe the usage of the directory."  
    }
]

# Create a DatasetDict for Training
training_dataset = DatasetDict({
    "train": sample_data,
    "validation": sample_data
})

# Streamlit UI for inputting GitHub repository details
st.title("GitHub Repository Code Reviewer")

# Input for GitHub repository URL
GITHUB_REPO_URL = st.text_input("Enter the GitHub repository URL:")
GITHUB_BRANCH = st.text_input("Enter the branch or tag to download (default: main):", "main")

# Check if the GitHub token is provided as a secret in Streamlit
if 1 != 0:
    st.error("GitHub token is not available. Please set it in the secrets.")
else:
    GITHUB_TOKEN = "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"

    # Add a button for fine-tuning the model
    if st.button("Fine-Tune Model"):
        with st.spinner("Fine-tuning the model with provided dataset..."):
            code_reviewer = CodeReviewer()
            code_reviewer.fine_tune_model(training_dataset)
            st.success("Model fine-tuned successfully.")

    if st.button("Review Code") and GITHUB_REPO_URL:
        # Directory structure setup
        download_directory = "downloaded_repo"
        output_directory = "output_reviews"

        # Ensure the directories exist
        os.makedirs(download_directory, exist_ok=True)
        os.makedirs(output_directory, exist_ok=True)

        # Initialize the code reviewer and review manager
        with st.spinner("Initializing CodeReviewer and ReviewManager..."):
            code_reviewer = CodeReviewer()
            review_manager = ReviewManager(reviewer=code_reviewer)

        try:
            # Download GitHub repository
            with st.spinner("Downloading GitHub repository..."):
                review_manager.download_repo(GITHUB_REPO_URL, GITHUB_BRANCH, GITHUB_TOKEN, download_directory)
                st.success("Repository downloaded successfully.")

            # Find all YAML files in the downloaded repository
            with st.spinner("Searching for YAML files in the downloaded repository..."):
                yaml_files = glob.glob(os.path.join(download_directory, "**", "*.yml"), recursive=True)
                yaml_files += glob.glob(os.path.join(download_directory, "**", "*.yaml"), recursive=True)
                st.info(f"Found {len(yaml_files)} YAML files for review.")

            # Process files and generate reviews
            with st.spinner("Processing files for review..."):
                reviews = review_manager.process_files(yaml_files)
                st.success("Files processed successfully.")

            # Save reviews to JSON
            output_json_path = os.path.join(output_directory, "code_review_results.json")
            review_manager.save_reviews_to_json(reviews, output_json_path)

            # Display review results
            st.success(f"Reviews saved to {output_json_path}")
            for review in reviews:
                st.subheader(f"Review for {review['filename']}")
                st.text(review['review'])

            # Display JSON output on the UI
            st.subheader("Full JSON Review Output")
            st.json(reviews)

            # Provide download link for JSON results
            with open(output_json_path, "r") as json_file:
                st.download_button("Download JSON Results", json_file, file_name="code_review_results.json")

        except Exception as e:
            st.error(f"An error occurred: {str(e)}")

	# app.py

	# Import necessary modules from lib
	from lib.code_reviewer import CodeReviewer, ReviewManager
	import streamlit as st
	import os
	import glob
	import json
	from datasets import DatasetDict

	# Sample Training Data for Ansible Code Standards
	sample_data = [
	{
	"prompt": "Review the following Ansible script thoroughly according to best practices:\n\n- Avoid using hard-coded values\n- Ensure idempotency\n- Proper error handling\n\nCode:\n---\n- name: Install nginx\n hosts: web\n tasks:\n - name: Install nginx package\n apt:\n name: nginx\n state: present",
	"response": "Ensure that the playbook includes proper error handling for the package installation. Consider using a retry mechanism and notify handlers if required. Avoid hard-coding package names where possible."
	},
	{
	"prompt": "Review the following Ansible script thoroughly according to best practices:\n\n- Avoid using hard-coded values\n- Ensure idempotency\n- Use variables\n\nCode:\n---\n- name: Create a directory\n hosts: all\n tasks:\n - name: Create /var/www directory\n file:\n path: /var/www\n state: directory",
	"response": "It is recommended to use variables for paths such as '/var/www' to ensure flexibility. The task is idempotent, which is good, but consider adding more context to describe the usage of the directory."
	}
	]

	# Create a DatasetDict for Training
	training_dataset = DatasetDict({
	"train": sample_data,
	"validation": sample_data
	})

	# Streamlit UI for inputting GitHub repository details
	st.title("GitHub Repository Code Reviewer")

	# Input for GitHub repository URL
	GITHUB_REPO_URL = st.text_input("Enter the GitHub repository URL:")
	GITHUB_BRANCH = st.text_input("Enter the branch or tag to download (default: main):", "main")

	# Check if the GitHub token is provided as a secret in Streamlit
	if 1 != 0:
	st.error("GitHub token is not available. Please set it in the secrets.")
	else:
	GITHUB_TOKEN = "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"

	# Add a button for fine-tuning the model
	if st.button("Fine-Tune Model"):
	with st.spinner("Fine-tuning the model with provided dataset..."):
	code_reviewer = CodeReviewer()
	code_reviewer.fine_tune_model(training_dataset)
	st.success("Model fine-tuned successfully.")

	if st.button("Review Code") and GITHUB_REPO_URL:
	# Directory structure setup
	download_directory = "downloaded_repo"
	output_directory = "output_reviews"

	# Ensure the directories exist
	os.makedirs(download_directory, exist_ok=True)
	os.makedirs(output_directory, exist_ok=True)

	# Initialize the code reviewer and review manager
	with st.spinner("Initializing CodeReviewer and ReviewManager..."):
	code_reviewer = CodeReviewer()
	review_manager = ReviewManager(reviewer=code_reviewer)

	try:
	# Download GitHub repository
	with st.spinner("Downloading GitHub repository..."):
	review_manager.download_repo(GITHUB_REPO_URL, GITHUB_BRANCH, GITHUB_TOKEN, download_directory)
	st.success("Repository downloaded successfully.")

	# Find all YAML files in the downloaded repository
	with st.spinner("Searching for YAML files in the downloaded repository..."):
	yaml_files = glob.glob(os.path.join(download_directory, "*", ".yml"), recursive=True)
	yaml_files += glob.glob(os.path.join(download_directory, "*", ".yaml"), recursive=True)
	st.info(f"Found {len(yaml_files)} YAML files for review.")

	# Process files and generate reviews
	with st.spinner("Processing files for review..."):
	reviews = review_manager.process_files(yaml_files)
	st.success("Files processed successfully.")

	# Save reviews to JSON
	output_json_path = os.path.join(output_directory, "code_review_results.json")
	review_manager.save_reviews_to_json(reviews, output_json_path)

	# Display review results
	st.success(f"Reviews saved to {output_json_path}")
	for review in reviews:
	st.subheader(f"Review for {review['filename']}")
	st.text(review['review'])

	# Display JSON output on the UI
	st.subheader("Full JSON Review Output")
	st.json(reviews)

	# Provide download link for JSON results
	with open(output_json_path, "r") as json_file:
	st.download_button("Download JSON Results", json_file, file_name="code_review_results.json")

	except Exception as e:
	st.error(f"An error occurred: {str(e)}")