|
| 1 | +#!/usr/bin/env python3 |
| 2 | +import sys |
| 3 | +import subprocess |
| 4 | +import json |
| 5 | +import base64 |
| 6 | +import argparse |
| 7 | +import io |
| 8 | +from ruamel.yaml import YAML |
| 9 | +from typing import Optional |
| 10 | + |
| 11 | +def run_command(command: list[str]) -> tuple[int, str, str]: |
| 12 | + """Runs a shell command and returns its exit code, stdout, and stderr.""" |
| 13 | + res = subprocess.run(command, capture_output=True, text=True, check=False) |
| 14 | + return res.returncode, res.stdout, res.stderr |
| 15 | + |
| 16 | +def get_file_content_from_pr(owner: str, repo: str, sha: str, file_path: str) -> str: |
| 17 | + """Fetches file content for a given commit SHA using the GitHub API. |
| 18 | + The base64 encoding and decoding step is necessary because of how the |
| 19 | + GitHub API is designed to transmit file content. |
| 20 | + """ |
| 21 | + api_path = f"/repos/{owner}/{repo}/contents/{file_path}?ref={sha}" |
| 22 | + print(f"-> Fetching file content for '{file_path}' from commit {sha[:7]}...") |
| 23 | + |
| 24 | + returncode, stdout, stderr = run_command(["gh", "api", api_path]) |
| 25 | + |
| 26 | + if returncode != 0: |
| 27 | + print(f"Exception: Could not fetch file. It may be new or deleted. (stderr: {stderr.strip()})") |
| 28 | + return "" |
| 29 | + |
| 30 | + try: |
| 31 | + content_json = json.loads(stdout) |
| 32 | + encoded_content = content_json.get('content', '') |
| 33 | + if not encoded_content: |
| 34 | + return "" |
| 35 | + |
| 36 | + return base64.b64decode(encoded_content).decode('utf-8') |
| 37 | + except (json.JSONDecodeError, KeyError, TypeError) as e: |
| 38 | + print(f"Exception: Could not parse or decode content. Error: {e}", file=sys.stderr) |
| 39 | + return "" |
| 40 | + |
| 41 | +def parse_yaml_and_get_keys(content: str, keys_to_find: list[str]) -> Optional[str]: |
| 42 | + """ |
| 43 | + Parses a YAML content string, cleans it, and finds the values of ALL |
| 44 | + matching keys from the provided list. Returns a canonical YAML string |
| 45 | + of the found keys and their values for comparison. |
| 46 | + """ |
| 47 | + yaml = YAML(typ='safe') |
| 48 | + |
| 49 | + # The rule.yml has '{{{...}}}' that can't be parsed by YMAL. |
| 50 | + # Pre-process content to remove invalid lines. |
| 51 | + lines = content.splitlines() |
| 52 | + clean_lines = [line for line in lines if not line.strip().startswith('{{{')] |
| 53 | + cleaned_content = "\n".join(clean_lines) |
| 54 | + |
| 55 | + if not cleaned_content.strip(): |
| 56 | + return None |
| 57 | + |
| 58 | + try: |
| 59 | + data = yaml.load(cleaned_content) |
| 60 | + if not isinstance(data, dict): |
| 61 | + return None |
| 62 | + except Exception as e: |
| 63 | + print(f"Exception: Could not parse YAML content. Error: {e}", file=sys.stderr) |
| 64 | + return None |
| 65 | + |
| 66 | + # Create a dictionary of all found keys and their values |
| 67 | + found_data = {} |
| 68 | + for key in keys_to_find: |
| 69 | + if key in data: |
| 70 | + found_data[key] = data.get(key) |
| 71 | + |
| 72 | + if not found_data: |
| 73 | + return None |
| 74 | + |
| 75 | + # For consistent comparison, convert the found data dict back to a YAML string |
| 76 | + string_stream = io.StringIO() |
| 77 | + yaml.dump(found_data, string_stream) |
| 78 | + return string_stream.getvalue().strip() |
| 79 | + |
| 80 | +def main(): |
| 81 | + """Main function to fetch PR files, parse them, and compare keys.""" |
| 82 | + parser = argparse.ArgumentParser( |
| 83 | + description="Compare specific keys in a YAML file between a PR's base and head branches." |
| 84 | + ) |
| 85 | + parser.add_argument("--owner", required=True, help="The owner of the repository.") |
| 86 | + parser.add_argument("--repo", required=True, help="The name of the repository.") |
| 87 | + parser.add_argument("pr_number", type=int, help="The Pull Request number.") |
| 88 | + parser.add_argument("file_path", type=str, help="The full path to the file within the repository.") |
| 89 | + parser.add_argument("keys", nargs='+', help="One or more keys to check in order (e.g., description options).") |
| 90 | + args = parser.parse_args() |
| 91 | + |
| 92 | + print(f"--- Analyzing '{args.file_path}' in PR #{args.pr_number} for keys: {args.keys} ---") |
| 93 | + print(f"Repository: {args.owner}/{args.repo}") |
| 94 | + |
| 95 | + |
| 96 | + # 1. Get the base and head commit SHAs. |
| 97 | + gh_pr_command = [ |
| 98 | + "gh", "pr", "view", str(args.pr_number), "--repo", f"{args.owner}/{args.repo}", "--json", "baseRefOid,headRefOid" |
| 99 | + ] |
| 100 | + returncode, stdout, stderr = run_command(gh_pr_command) |
| 101 | + |
| 102 | + if returncode != 0: |
| 103 | + print(f"\nException: Failed to get PR details. gh stderr: {stderr.strip()}", file=sys.stderr) |
| 104 | + sys.exit(0) |
| 105 | + try: |
| 106 | + shas = json.loads(stdout) |
| 107 | + base_sha = shas['baseRefOid'] |
| 108 | + head_sha = shas['headRefOid'] |
| 109 | + print(f"Base SHA (Before): {base_sha}") |
| 110 | + print(f"Head SHA (After): {head_sha}\n") |
| 111 | + except (json.JSONDecodeError, KeyError) as e: |
| 112 | + print(f"\nException: Could not parse commit SHAs. Error: {e}", file=sys.stderr) |
| 113 | + sys.exit(0) |
| 114 | + |
| 115 | + # 2. Fetch file content for both commits. |
| 116 | + before_content = get_file_content_from_pr(args.owner, args.repo, base_sha, args.file_path) |
| 117 | + after_content = get_file_content_from_pr(args.owner, args.repo, head_sha, args.file_path) |
| 118 | + |
| 119 | + # 3. Parse the content in memory to get the desired values. |
| 120 | + before_value = parse_yaml_and_get_keys(before_content, args.keys) |
| 121 | + after_value = parse_yaml_and_get_keys(after_content, args.keys) |
| 122 | + |
| 123 | + # 4. Compare the results and print the final output. |
| 124 | + print("\n--- Comparison Result ---") |
| 125 | + print(f"Value(s) in base branch:\n---\n{before_value}\n---") |
| 126 | + print(f"Value(s) in PR branch:\n---\n{after_value}\n---") |
| 127 | + |
| 128 | + if before_value == after_value: |
| 129 | + print("\nNo changes detected for the specified keys.") |
| 130 | + sys.exit(0) |
| 131 | + else: |
| 132 | + print("\nChange detected for one of the specified keys!") |
| 133 | + print("\nCHANGE_FOUND=true") |
| 134 | + sys.exit(1) |
| 135 | + |
| 136 | +if __name__ == "__main__": |
| 137 | + main() |
| 138 | + |
| 139 | + |
0 commit comments