joplin-mdbooks-website/move_html_to_dir

#!/usr/bin/env python3
"""
Rename files like `foo/bar.html` to `foo/bar/index.html` for prettier URLs.
"""
import re
import sys
from pathlib import Path
from typing import Dict


def convert_relative_to_absolute(path: Path):
    """
    Convert relative links in files to absolute ones.

    Sometimes files contain links relative to their location, e.g.
    `public/test/foo.html` might contain a link to `bar/baz.html`, which makes it hard
    for us to match `public/test/bar/baz.html` that way. This function converts that
    into its full path (`public/test/bar/baz.html`) so we can match it later.
    """

    def replace_wrapper(filename: Path):
        def replace_link(match: re.Match) -> str:
            property, text, suffix = match.groups()
            if "://" in text or text == "/" or text.startswith("#"):
                # Not a valid filename, return it.
                return f"{property}{text}{suffix}"

            if text.startswith("/"):
                filepath = (path / text[1:]).resolve()
            else:
                filepath = (filename.parent / text).resolve()

            if not filepath.exists():
                # Not a valid filename, return it.
                sys.exit(f"Possible broken link in {filename}: {text}")

            replacement_path = filepath.relative_to(path.absolute())
            return f"{property}/{replacement_path}{suffix}"

        return replace_link

    for filename in path.glob("**/*.html"):
        print(f"Converting relative links in {filename}...")
        with filename.open("r+") as f:
            contents = f.read()
            f.truncate(0)
            f.seek(0)
            contents = re.sub(
                r"""
                ((?:href|src|root)\s*=\s*")  # Various tags like href="
                ([^\"]+?)                    # Anything non-quote, non-greedily.
                (
                (?:\#[^\"]*|)                # Either an anchor or nothing.
                ")                           # Ending quote.
                """,
                replace_wrapper(filename),
                contents,
                flags=re.VERBOSE,
            )
            f.write(contents)


def replace_links(path: Path, replacements: Dict[str, str]):
    """
    Convert links in files.

    `path`         - The root of the repo.
    `replacements` - A dictionary of replacement URLs in the form of
                     `{"dir/file.html": "dir/file/"}`.
    """

    for filename in path.glob("**/*.html"):
        print(f"Converting links in {filename}...")
        with filename.open("r+") as f:
            contents = f.read()
            f.truncate(0)
            f.seek(0)
            for source, target in replacements.items():
                contents = contents.replace(source, target)

            # Convert relative links to absolute.
            contents = re.sub(
                r'((?:href|src|root)\s*=\s*")((?:\.\./)+)([^\.])', r"\1/\3", contents
            )

            f.write(contents)


def main(path: Path):
    convert_relative_to_absolute(path)

    replacements: Dict[str, str] = {}
    for p in path.glob("**/*.html"):
        if str(p.parent) == ".":
            # Don't convert top-level files.
            continue

        if p.name == "index.html":
            # Don't convert top-level files.
            continue

        print(f"Renaming {p}...")

        dir_path = p.parent / p.stem
        dir_path.mkdir(parents=True, exist_ok=True)

        new_path = dir_path / "index.html"
        p.rename(new_path)
        # Construct the dictionary of replacements that have been done.
        replacements[str(p.relative_to(path))] = f"{new_path.parent.relative_to(path)}/"

    replace_links(path, replacements)


if __name__ == "__main__":
    main(Path(sys.argv[1]))