statistics.py

"""
statistics.py
Project statistics
"""

from __future__ import annotations

from pathlib import Path

from utils import (
    estimate_tokens,
    count_lines,
    human_size,
    language_from_extension,
)


def build_statistics(files: list[Path], contents: list[tuple[Path, str]]) -> dict:
    total_size = 0
    total_lines = 0
    total_tokens = 0

    languages = {}

    largest = None
    smallest = None

    for path, text in contents:

        try:
            size = path.stat().st_size
        except Exception:
            size = len(text.encode())

        total_size += size
        total_lines += count_lines(text)
        total_tokens += estimate_tokens(text)

        lang = language_from_extension(path.suffix)

        languages[lang] = languages.get(lang, 0) + 1

        if largest is None or size > largest[1]:
            largest = (path.name, size)

        if smallest is None or size < smallest[1]:
            smallest = (path.name, size)

    return {
        "files": len(files),
        "lines": total_lines,
        "tokens": total_tokens,
        "size": total_size,
        "size_human": human_size(total_size),
        "languages": languages,
        "largest": largest,
        "smallest": smallest,
        "average_size": total_size // max(len(files), 1),
    }


def statistics_markdown(stats: dict) -> str:

    langs = "\n".join(
        f"- {k}: {v}"
        for k, v in sorted(stats["languages"].items())
    )

    return f"""
## Project Statistics

Files: {stats["files"]}

Lines: {stats["lines"]}

Approx Tokens: {stats["tokens"]}

Project Size: {stats["size_human"]}

Average File Size: {human_size(stats["average_size"])}

Largest File:
{stats["largest"]}

Smallest File:
{stats["smallest"]}

### Languages

{langs}
""".strip()