challenge_duplicate_file_deletor

dupletor.py

#!/usr/bin/env python3
import sys
from pathlib import Path

# Do noisy debug spouting?:
# dobug = True
dobug = False


class File:
    def __init__(self, path: Path) -> None:
        super().__init__()
        self.name = path.name
        self.head, sep, tail = self.name.partition(' copy')
        self.is_copy = sep != ''
        suffix_base = tail if self.is_copy else self.name
        _, sep, tail = suffix_base.rpartition('.')
        self.suffix = tail if sep != '' else ''
        # Fix the problem that spaces are sorted before dots:
        self.sort_name = self.name.replace(' ', '‾')
        self.duplicates = []

    def __repr__(self):
        status = 'COPY' if self.is_copy else '....'
        return f'<{status}  "{self.name}">'

    def __lt__(self, other):
        return self.sort_name < other.sort_name

    def matches(self, ori_item):
        verity = ori_item.head.startswith(self.head) and \
                 ((self.suffix == '') or (ori_item.suffix == self.suffix))
        return verity

    def add_duplicate(self, duplicate):
        self.duplicates.append(duplicate)


class Dupletor:
    @classmethod
    def spawn(cls):
        cls.spout('spawn Dupletor', debuggy=False)
        # Enable to generate test victims directory and files:
        # cls().generate_victims()
        cls().duplete()

    def duplete(self):
        if len(sys.argv) < 2:
            self.error(f'Usage: python dupletor.py <path>')
            return
        path = Path(sys.argv[1]).resolve()
        if not path.exists():
            self.error(f'"{path}" does not exist')
            return
        if not path.is_dir():
            self.error(f'"{path}" is not a directory')
            return
        path_items = list(path.iterdir())
        alls = [File(item) for item in path_items if not item.is_dir()]
        if len(alls) < 1:
            self.error(f'"{path}" No files found.')
            return
        self.announce(f'duplete "{path}"')
        ####
        alls.sort()
        self.dashout('alls', series=alls)
        originals = []
        rests = []
        dupes = []
        for ofil in alls:
            if ofil.is_copy:
                rests.append(ofil)
            else:
                originals.append(ofil)
        self.dashout('originals', series=originals)
        self.dashout('rest', series=rests)
        ####
        self.dashout('stuff')
        for rest_item in rests:
            self.debug(f' - rest_item: {rest_item}')
            recipient = originals
            for ori_item in originals:
                if rest_item.matches(ori_item):
                    self.debug(f'  -- ori: {ori_item}')
                    ori_item.add_duplicate(rest_item)
                    recipient = dupes
                    break
            recipient.append(rest_item)
        ####
        self.dashout('originals', series=originals, debuggy=False)
        self.dashout('dupes', series=dupes, debuggy=False)
        self.show_relations(originals)
        self.announce('Do some actual logistics ...')

    def show_relations(self, originals):
        self.dashout('relations', debuggy=False)
        for original in originals:
            if original.duplicates:
                self.spout(f' - {original}', debuggy=False)
                for duplicate in original.duplicates:
                    self.debug(f'\t\t-- {duplicate}', debuggy=False)
            else:
                self.debug(f' - {original}', debuggy=False, color_num=34)

    # region utils
    @staticmethod
    def spout(*args, color_num=92, debuggy=True):
        if dobug:
            debuggy = False  # False enables debugging! ;-)
        if debuggy:
            return
        # spout in glorious colors:
        print(f'\033[{color_num}m {" ".join(map(str, args))}\033[0m')

    def error(self, *args, color_num=91, debuggy=False):
        self.spout(*args, color_num=color_num, debuggy=debuggy)

    def announce(self, *args, color_num=93):
        self.spout(*args, color_num=color_num, debuggy=False)

    def debug(self, *args, color_num=90, debuggy=True):
        self.spout(*args, color_num=color_num, debuggy=debuggy)

    def dashout(self, arg, series=None, debuggy=True):
        dash_ = '-' * (80 - len(arg) - 2)
        self.spout(f'{arg}: {dash_}', debuggy=debuggy)
        if series:
            for item in series:
                self.debug(f' - {item}', debuggy=debuggy)

    # endregion utils

    ############################################################

    def generate_victims(self):
        txt = self.generatext
        self.debug(f' - txt: {txt}')
        destin_dir = Path(__file__).parent / 'victims'
        destin_dir.mkdir(parents=True, exist_ok=True)
        lines = txt.splitlines()
        self.debug('-' * 80)
        for line in lines:
            line = line.strip()
            if not line:
                continue
            self.debug(f' - line: "{line}"')
            file = destin_dir / line
            file.touch()

    @property
    def generatext(self) -> str:
        txt = """
faang.png
faang copy.png

index copy.html
index copy 2.html
index copy 3.html

package.abc
package.json
package copy.json
package copy 2.json
package copy 2.xyz

README.md
README copy.md

Screenshot 2023-09-04 at 3.01.45 PM copy
Screenshot 2023-09-04 at 3.01.45 PM copy 2

suffixy 12.34
suffixy 12.34 copy
suffixy 12.34 copy 2

suffixy 12.34 xyz
suffixy 12.34 xyz copy
suffixy 12.34 xyz copy 2

abc
abc copy
abc copy 2

def.aaa.bbb
def.aaa.bbb.ccc
"""
        return txt


if __name__ == '__main__':
    Dupletor.spawn()

Output:

/usr/bin/python3 /opt/projects/challenge_duplicate_file_deletor/py/dupletor.py victims
spawn Dupletor
duplete "/opt/projects/challenge_duplicate_file_deletor/py/victims"
originals: ---------------------------------------------------------------------

• <.... "README.md">
• <.... "abc">
• <.... "def.aaa.bbb">
• <.... "def.aaa.bbb.ccc">
• <.... "faang.png">
• <.... "package.abc">
• <.... "package.json">
• <.... "suffixy 12.34">
• <.... "suffixy 12.34 xyz">
• <COPY "Screenshot 2023-09-04 at 3.01.45 PM copy">
• <COPY "index copy.html">
• <COPY "package copy 2.xyz">
  dupes: -------------------------------------------------------------------------
• <COPY "README copy.md">
• <COPY "Screenshot 2023-09-04 at 3.01.45 PM copy 2">
• <COPY "abc copy">
• <COPY "abc copy 2">
• <COPY "faang copy.png">
• <COPY "index copy 2.html">
• <COPY "index copy 3.html">
• <COPY "package copy.json">
• <COPY "package copy 2.json">
• <COPY "suffixy 12.34 copy">
• <COPY "suffixy 12.34 copy 2">
• <COPY "suffixy 12.34 xyz copy">
• <COPY "suffixy 12.34 xyz copy 2">
  Do some actual logistics ...

Process finished with exit code 0