source_display.py

from pathlib import Path


def pdf_source_for_markdown(source: str | None) -> str:
    if not source:
        return ""
    md_path = Path(source)
    if md_path.suffix.lower() != ".md":
        return ""
    pdf_path = md_path.parent.parent / "pdf" / f"{md_path.stem}.pdf"
    return str(pdf_path) if pdf_path.exists() else ""


def source_after_data_pdf(source: str) -> str:
    parts = Path(source).parts
    lowered = [part.lower() for part in parts]
    for index in range(len(lowered) - 1):
        if lowered[index] == "data" and lowered[index + 1] == "pdf":
            return str(Path(*parts[index + 2 :])) if index + 2 < len(parts) else ""
    return source


def display_source_from_metadata(metadata: dict) -> str:
    source = (
        metadata.get("source_display_path")
        or metadata.get("source_pdf_path")
        or metadata.get("source")
    )
    if not source:
        return "Không có nguồn"
    pdf_source = pdf_source_for_markdown(str(source)) or str(source)
    return source_after_data_pdf(pdf_source)