#!/usr/bin/env python3

r"""
Read duplicates, report on them and/or hardlink them together.

The input format looks like:
dup1\\0dup2\\0dup3\\0\\0dup4\\0dup5\\0\\0

In this example, dup1 == dup2 == dup3, and dup4 == dup5.
"""

import os
import sys

import readline0


def usage(retval: int) -> None:
    """Output a usage message and exit."""
    if retval == 0:
        file_ = sys.stdout
    else:
        file_ = sys.stderr

    print(f"Usage: {sys.argv[0]} --report --hardlinks", file=file_)
    print(file=file_)
    print("Report and/or hardlink duplicate files.  For already-hardlinked duplicates, we only report one.", file=file_)

    sys.exit(retval)


class Options:
    """Parse and hold command line options."""

    def __init__(self) -> None:
        self.report = False
        self.hardlinks = False

        while sys.argv[1:]:
            match sys.argv[1]:
                case "--help" | "-h":
                    usage(0)
                case "--report":
                    self.report = True
                case "--hardlinks":
                    self.hardlinks = True
                case _:
                    print(f"{sys.argv[0]}: unrecognized option: {sys.argv[1]}", file=sys.stderr)
                    usage(1)
            del sys.argv[1]

        preflight_good = True

        if self.report + self.hardlinks < 1:
            print(f"{sys.argv[0]}: you must specify at least one of --report and --hardlinks", file=sys.stderr)
            preflight_good = False

        if not preflight_good:
            print(f"{sys.argv[0]}: one or more items in preflight check failed", file=sys.stderr)
            usage(1)


def get_st_nlink(filename: str) -> int:
    """Look up the number of links this file has."""
    s = os.stat(filename)
    return s.st_nlink


def main() -> None:
    """Start the ball rolling."""
    options = Options()

    first_time = True

    for equal_files in readline0.readline0(file_=0, separator=b"\0\0"):
        filenames = equal_files.split(b"\0")
        if options.hardlinks:
            # If we're hardlinking, we need to be careful not to split any groups of pre-existing hardlinks.  So we pick
            # the file with the highest hard link count to link the others to.
            # BTW, this could be done in O(n) time, even though this currently does it in O(nlogn) - where n is the
            # number of identical files.
            list_ = [(get_st_nlink(filename), filename) for filename in filenames]
            list_.sort(reverse=True)
            filenames = [tuple_[1] for tuple_ in list_]
        if first_time:
            first_time = False
        else:
            print()
        filename0 = filenames[0]
        if options.report:
            print(filename0)
        for filename in filenames[1:]:
            if options.report:
                print("   ", filename)
            if options.hardlinks:
                # Hardlink the current filename.
                # We do this temp_file stuff to gain atomicity.
                temp_file = filename0 + b".temp"
                os.link(filename0, temp_file)
                os.rename(temp_file, filename)


if __name__ == "__main__":
    main()