zotxform

#! /usr/bin/env python3
#
# zotxform
# BW custom Zotero locale transform.
#
# AUTHORS: Ian-Mathew Hornburg ‹imhornburg@gmail.com›
#          Optima Language Analytics LLC
# LICENSE: [TBD]
#
# Copyright (c) 2014–2015, the authors.
# All rights reserved.
# --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- #


from shutil import copytree, rmtree
from sys import exit
import argparse
import json
import logging
import os
import os.path as path
import zipfile as z


CONFIG_PATH = 'zotxform.json'
TEMPDIR_PATH = 'zotxform_temp'
FIREFOX_MANIFEST_PATH = 'chrome.manifest'
FIREFOX_MANIFEST_LOCALE = '\n\nlocale   zotero     bw-US      chrome/locale/bw-US/zotero/'
STANDALONE_WIN_DFLT_ARCHIVE_PATH = "C:\\Program Files (x86)\\Zotero Standalone\\zotero.jar"
STANDALONE_WIN_DFLT_MANIFEST_PATH = "C:\\Program Files (x86)\\Zotero Standalone\\chrome.manifest"
STANDALONE_MANIFEST_LOCALE = '\n\nlocale   zotero     bw-US      jar:zotero.jar!/chrome/locale/bw-US/zotero/'
replacement_table = {
    'itemTypes.patent                     = Patent': 'itemTypes.patent                     = Class',
    'itemFields.series               = Series': 'itemFields.series               = Series/Type',
    'itemFields.assignee                  = Assignee': 'itemFields.assignee                  = Level',
    'itemFields.patentNumber              = Patent Number': 'itemFields.patentNumber              = Lecture Number',
    'itemFields.priorityNumbers      = Priority Numbers': 'itemFields.priorityNumbers      = Course Title',
    'itemFields.issueDate            = Issue Date': 'itemFields.issueDate            = Class Date',
    'itemFields.references           = References': 'itemFields.references           = Primary Instructor',
    'itemFields.legalStatus               = Legal Status': 'itemFields.legalStatus               = Program',
    'itemFields.country                   = Country': 'itemFields.country                   = Department',
    'itemFields.applicationNumber         = Application Number': 'itemFields.applicationNumber         = Course Number',
    'itemFields.proceedingsTitle          = Proceedings Title': 'itemFields.proceedingsTitle          = Session/Proceedings',
    'itemFields.issuingAuthority          = Issuing Authority': 'itemFields.issuingAuthority          = Material Type',
    'itemFields.filingDate           = Filing Date': 'itemFields.filingDate           = Term',
    'creatorTypes.editor                  = Editor': 'creatorTypes.editor                  = Editor/Chair',
    'creatorTypes.inventor           = Inventor': 'creatorTypes.inventor           = Instructor',
    'creatorTypes.attorneyAgent      = Attorney/Agent': 'creatorTypes.attorneyAgent      = Notes By',
}


def transform(args):
    def check_file(fn):
        fn_exists = path.exists(fn) and path.isfile(fn)
        if not fn_exists:
            raise OSError("The specified file does not exist or cannot be accessed.")
            sys.exit(1)


    if args['firefox']:
        archive_location = {}

        if args['file']:
            # Ensure that we have a full absolute path on the passed filename
            fn_archive = path.abspath(args['file'])
            check_file(fn_archive)

            # Overwrite any existing configuration
            with open(CONFIG_PATH, encoding='utf_8', mode='w') as config_file:
                archive_location['firefox'] = fn_archive
                json.dump(archive_location, config_file)

        else:
            try:
                # Check for an existing config file
                with open(CONFIG_PATH, encoding='utf_8', mode='r') as config_file:
                    archive_location = json.load(config_file)
            except:
                raise FileNotFoundError("Could not find zotxform.json. Please specify a Zotero Firefox archive to be patched.")

    if args['firefox']:
        fn_archive = path.abspath(archive_location.get('firefox'))
        fn_manifest = FIREFOX_MANIFEST_PATH

    if args['standalone']:
        # If we're Standalone, just use the hardcoded defaults
        fn_archive = path.abspath(STANDALONE_WIN_DFLT_ARCHIVE_PATH)
        fn_manifest = path.abspath(STANDALONE_WIN_DFLT_MANIFEST_PATH)


    try:
        with z.ZipFile(fn_archive, mode='r') as archive:
            os.mkdir(TEMPDIR_PATH)
            os.chdir(TEMPDIR_PATH)

            # We wanna preserve the exact order of the files in the archive,
            # since for some applications, order matters (e.g., needing to have
            # the chrome.manifest first). So everything that’s going into the
            # archive needs to be explicitly enumerated in this list, starting
            # with the existing contents.
            file_list = archive.namelist()

            if args['firefox']:
                if FIREFOX_MANIFEST_PATH in archive.namelist():
                    logger.info("Manifest found.")
                else:
                    raise FileNotFoundError("chrome.manifest file not found.")

            if args['standalone']:
                if os.path.isfile(fn_manifest):
                    logger.info("Manifest found.")
                else:
                    raise FileNotFoundError("chrome.manifest file not found.")

            # Extract and end scope, closing archive
            archive.extractall()

        # Append custom locale line to manifest
        with open(fn_manifest, encoding='utf_8', mode='r') as manifest:
            s = manifest.read()

        if args['firefox']:
            if FIREFOX_MANIFEST_LOCALE in s:
                raise ValueError("Manifest already patched.")
            else:
                s = s.join([s, FIREFOX_MANIFEST_LOCALE])

        if args['standalone']:
            if STANDALONE_MANIFEST_LOCALE in s:
                raise ValueError("Manifest already patched.")
            s = s.join([s, STANDALONE_MANIFEST_LOCALE])

        logger.info("Writing patched manifest.")
        with open(fn_manifest, encoding='utf_8', mode='w') as manifest:
            manifest.write(s)

        # Copy the contents of en-US locale to our new custom locale directory
        new_locale_dir = path.join('chrome', 'locale', 'bw-US')
        copytree(path.join('chrome', 'locale', 'en-US'), new_locale_dir)
        file_list.append(new_locale_dir)

        # [TO-DO] Make this cleaner, more clear; better comprehensions
        # Walk the directory tree, grabbing all nodes along the way, and
        # append this list to our archive list.
        for e in [path.join(dirpath, f) for dirpath, dirname, filename in os.walk(new_locale_dir) for f in filename]:
            file_list.append(e)

        # Read in properties file, patch, write
        with open(path.join(new_locale_dir, 'zotero', 'zotero.properties'), encoding='utf_8', mode='r') as f:
            properties = f.read()

        for target in replacement_table.keys():
            properties = properties.replace(target, replacement_table.get(target))

        with open(path.join(new_locale_dir, 'zotero', 'zotero.properties'), encoding='utf_8', mode='w') as f:
            f.write(properties)

        # Truncate and write new archive
        with z.ZipFile(fn_archive, compression=z.ZIP_DEFLATED, mode='w') as archive:
            for e in file_list:
                archive.write(e)
                if 'bw-US' in e or e == FIREFOX_MANIFEST_PATH:
                    print(e)

        logger.info("Done.")

    finally:
        os.chdir(os.pardir)
        rmtree(TEMPDIR_PATH)


# Command-line argument parsing - --- --- --- --- --- --- --- --- --- --- --- #
parser = argparse.ArgumentParser(
    add_help=True,
    description="Custom Zotero locale transform.",
    epilog="Copyright (c) 2014–2015, Optima Language Analytics LLC. All rights reserved."
)
parser.add_argument('-v', '--verbose', action='store_true', help='increase verbosity')
parser.add_argument('-f', '--firefox', action='store_true', help='patch Zotero Firefox')
parser.add_argument('-s', '--standalone', action='store_true', help='patch Zotero Standalone')
parser.add_argument('file', nargs='?', default=False, help="A valid Zotero Firefox settings file to be patched. If none given, zotxform will attempt to use the last valid archive location.")


# “main()” -- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- #
if __name__ == '__main__':
    logging.basicConfig(format="[zotxform] {levelname} :: {message}", style='{')
    logger = logging.getLogger('zotxform')
    args = vars(parser.parse_args())


    if args['verbose']:
        logger.setLevel('INFO')
    del args['verbose']

    logger.info("zotxform v0.612")

    # If no action flag is given, warn and bail.
    if not any(args.values()):
        logger.warn("At least one argument is required.")
        exit(1)

    if args['firefox'] and args['standalone']:
        logger.warn("Specify only one type of Zotero settings file to patch.")
        exit(1)

    if args['file'] and args['standalone']:
        logger.warn("Cannot specify file. Zotero Standalone patch uses default files located at " + STANDALONE_WIN_DFLT_ARCHIVE_PATH + " and " + STANDALONE_WIN_DFLT_MANIFEST_PATH + ".")
        exit(1)

    transform(args)