analyze_red_knot_diagnostics.py

# /// script
# requires-python = ">=3.13"
# dependencies = []
# ///

import re
import sys
from collections import defaultdict
from dataclasses import dataclass
from pathlib import Path


@dataclass
class Diagnostic:
    level: str
    lint_name: str

    path: str
    line: int
    column: int

    message: str


def parse_diagnostic_message(line: str) -> Diagnostic | None:
    pattern = (
        r"^(?P<level>error|warning)\[lint:(?P<lint_name>.+?)\] "
        r"(?P<path>.+?):(?P<line>\d+):(?P<column>\d+): "
        r"(?P<message>.+)$"
    )

    if match := re.match(pattern, line):
        return Diagnostic(
            level=match.group("level"),
            lint_name=match.group("lint_name"),
            path=match.group("path"),
            line=int(match.group("line")),
            column=int(match.group("column")),
            message=match.group("message"),
        )
    return None


def parse_diagnostics(content: str) -> list[Diagnostic]:
    messages = []
    for line in content.splitlines():
        line = line.strip()
        if not line:
            continue

        if message := parse_diagnostic_message(line):
            messages.append(message)
    return messages


def print_statistics(messages: list[Diagnostic]) -> None:
    level_stats = defaultdict(int)
    lint_stats = defaultdict(int)
    file_stats = defaultdict(int)

    for msg in messages:
        level_stats[msg.level] += 1
        lint_stats[msg.lint_name] += 1
        file_stats[msg.path] += 1

    print("Statistics:")

    print(f"  Number of diagnostics: {len(messages)}")
    print(f"  Number of files with diagnostics: {len(file_stats)}")

    print("\nDiagnostics by level:")
    for level, count in level_stats.items():
        print(f"  {level.title()}: {count}")

    print("\nDiagnostics by file (top 5):")
    top_files = sorted(file_stats.items(), key=lambda x: x[1], reverse=True)[:5]
    for path, count in top_files:
        print(f"  {path}: {count}")

    print("\nDiagnostics by lint name:")
    for lint_name, count in sorted(
        lint_stats.items(), key=lambda x: x[1], reverse=True,
    ):
        print(f"  {lint_name}: {count}")


def main() -> None:
    import argparse

    parser = argparse.ArgumentParser(
        description="Parse Red Knot output and show statistics."
    )
    parser.add_argument("path", type=Path, help="Text file with Red Knot output", nargs="?")
    args = parser.parse_args()

    if args.path:
        if not args.path.exists():
            print(f"Error: File not found: {args.path}")
            sys.exit(1)

        content = args.path.read_text()
    else:
        content = sys.stdin.read()

    diagnostics = parse_diagnostics(content)
    print_statistics(diagnostics)


if __name__ == "__main__":
    main()