remove files not maintained by zotero

zot_rm_unmaintained_files.py

#!/usr/bin/env python
# coding: utf-8

from __future__ import print_function

import configparser
import re
import shutil
import sqlite3
import sys

try:
    from pathlib import Path
except ImportError:
    from pathlib2 import Path

if sys.version_info.major == 2:
    reload(sys)
    sys.setdefaultencoding('UTF8')


def get_zotfile_dest_and_zotero_data_dirs():
    '''
    Get the Zotero data dir and the Zotfile destination dir in PosixPath type
    '''
    profile_dirs = {
        'darwin': Path.home() / 'Library/Application Support/Zotero',
        'linux': Path.home() / '.zotero/zotero',
        'linux2': Path.home() / '.zotero/zotero',
        'win32': Path.home() / 'AppData/Roaming/Zotero/Zotero'
    }
    profile_dir = profile_dirs[sys.platform]

    config = configparser.ConfigParser()
    config.read('{}'.format(profile_dir / 'profiles.ini'))
    configs_loc = profile_dir / config['Profile0']['Path'] / 'prefs.js'
    configs = configs_loc.read_text()

    zotero_data_pat = re.compile(
        r'user_pref\("extensions.zotero.dataDir",\ "(?P<zotero_data>.+)"\);')
    zotero_data_dir = Path(zotero_data_pat.search(
        configs).group('zotero_data'))
    zotfile_dest_pat = re.compile(
        r'user_pref\("extensions.zotfile.dest_dir",\ "(?P<zotfile_dest>.+)"\);')
    zotfile_dest_dir = Path(
        zotfile_dest_pat.search(configs).group('zotfile_dest'))

    return zotero_data_dir, zotfile_dest_dir


def get_unmaintained_files(zotero_data_dir,
                           zotfile_dest_dir,
                           case_sensitive='auto'):
    '''
    Get a list of atthchment in PosixPath type that unmaintained in the Zotero
    Args:
        zotero_data_dir(PosixPath): Zotero data dir
        zotfile_dest_dir(PosixPath): Zotfile destination dir
        case_sensitive(bool or str): wether the os is case sensitive, 
                                     default set linux as True, and rest as False
    '''
    attachments_local = set(p.as_posix() for p in zotfile_dest_dir.glob('**/*')
                            if p.is_file() and p.name[0] != '.')

    con = sqlite3.connect('{}'.format(zotero_data_dir / 'zotero.sqlite'))
    with con:
        cur = con.cursor()
        cur.execute('SELECT path FROM itemAttachments WHERE linkMode = 2')
        attachments_zotero = set([
            p.as_posix() for p in [
                zotfile_dest_dir / p[0].replace('attachments:', '', 1)
                for p in cur.fetchall()
            ]
        ])

    if sys.platform == 'darwin':
        import unicodedata
        attachments_zotero = set(
            list(attachments_zotero) +
            [unicodedata.normalize('NFD', p) for p in attachments_zotero])
    if case_sensitive == 'auto':
        case_sensitive = {
            'darwin': False,
            'linux': True,
            'linux2': True,
            'win32': False
        }[sys.platform]
    if not case_sensitive:
        attachments_local = set([p.lower() for p in attachments_local])
        attachments_zotero = set([p.lower() for p in attachments_zotero])
    attachments_to_remove = attachments_local - attachments_zotero

    return [Path(p) for p in attachments_to_remove]


def remove_unmaintained(attachments_to_remove):
    '''
    Remove the unmaintained files in PosixPath type, and clear empty dirs
    '''
    [p.unlink() for p in attachments_to_remove]
    empty_dirs = [
        p for p in zotfile_dest_dir.glob('**/*') if (not p.is_file()) and (
            not len([f for f in list(p.iterdir()) if f.name[0] != '.']))
    ]
    [shutil.rmtree(p.as_posix(), ignore_errors=True) for p in empty_dirs]


if __name__ == '__main__':
    zotero_data_dir, zotfile_dest_dir = get_zotfile_dest_and_zotero_data_dirs()
    attachments_to_remove = get_unmaintained_files(zotero_data_dir,
                                                   zotfile_dest_dir)

    try:
        import click
        print('The following files are no longer managed by Zotero:')
        print('\n'.join(['  {}'.format(p) for p in attachments_to_remove]))
        if click.confirm('Do you want remove them?', default=True):
            remove_unmaintained(attachments_to_remove)
    except ImportError:
        print(
            'The following files no longer managed by Zotero will be removed:')
        print('\n'.join(['  {}'.format(p) for p in attachments_to_remove]))
        remove_unmaintained(attachments_to_remove)

(base) C:\Users\sunsh\Desktop>zot_rm_unmaintained_files.py
The following files are no longer managed by Zotero:

Do you want remove them? [Y/n]: Y
知乎留言转到这里，执行上述操作后，同步盘内重复文件并未删除，坚果云默认路径，有中文。
Miniconda3 Windows 64-bit，Python3.7

更新1：修改为纯英文路径后将原来的文件复制过去（保证有重复的文件），在zotero中更改路径同步后再执行脚本，与之前相同，重复的文件还在。

更新2：原因搞清楚了，相同文件名不同后缀的pdf不会被删除（例如abc，abc1，abc2...），之前从endnote导入到zotero里有些文献重复关联了相同的pdf，改名后就可以正常删除，希望后续能更新脚本。

abc abc1 abc2 不叫相同文件名不同后缀，这已经是不同文件名了。
”有些文献关联相同 pdf“这个说法不妥，zotero 或者 zotfile 都很难做到这个，除非手动。按照你的描述，更像是因导入产生的重复条目。
造成abc，abc1，abc2 的在我所见有一种情况，就是zotfile 在同一目录下产生多个同名附件，原因可能是同一条目多附件，也可能是相同目录下的不同条目生成的重名附件。而zotfile在处理这种重名的时候，本来就有bug。

请问zotero5没有profiles.ini怎么办？

我写这个脚本的时候，zotero 早就是5以后的版本了，脚本找不到文件可能是用户中文目录的问题。

有没有考虑增加删除zotero数据库无效attachment链接？

有没有考虑增加删除zotero数据库无效attachment链接？

嗯，这个脚本反过来比就行，问题是你怎么会产生大量无效的链接附件呢？
如果是你不小心删除了，或者挪地方了，不会想着先抢救下吗？

emmm, Zotero 7 用了 zotero-attanger ？应该config换了，有空我再更新脚本吧。。。