Function check_files

scripts/check_article_quality.py:404–451

(args: argparse.Namespace)

def check_files(args: argparse.Namespace) -> list[Failure]:
    paths, use_staged_content = collect_paths(args)
    article_paths = sorted({path for path in paths if is_article_path(path)})
    failures: list[Failure] = []
    root = repo_root()

    for path in article_paths:
        if not use_staged_content and not path.exists():
            continue

        text = read_path(path, staged=use_staged_content)
        metrics = measure(text)
        image_count, image_issues = check_images(
            path,
            text,
            staged=use_staged_content,
            args=args,
            root=root,
        )
        metrics = replace(
            metrics,
            images=image_count,
            broken_images=len(image_issues),
        )
        issues: list[str] = []
        if metrics.chars < args.min_chars:
            issues.append(f"body has {metrics.chars} meaningful chars; expected >= {args.min_chars}")
        if metrics.paragraphs < args.min_paragraphs:
            issues.append(f"has {metrics.paragraphs} developed paragraphs; expected >= {args.min_paragraphs}")
        if metrics.sentences < args.min_sentences:
            issues.append(f"has {metrics.sentences} developed sentences; expected >= {args.min_sentences}")
        if metrics.broken_images > args.max_broken_images:
            issues.append(
                f"has {metrics.broken_images} broken image references; "
                f"expected <= {args.max_broken_images}"
            )

        if issues:
            failures.append(
                Failure(
                    path=path,
                    metrics=metrics,
                    issues=tuple(issues),
                    image_issues=image_issues,
                )
            )

    return failures


def build_parser() -> argparse.ArgumentParser: