view create_or_update_organism.py @ 17:1a2404ca43f9 draft default tip

planemo upload for repository https://github.com/galaxy-genome-annotation/galaxy-tools/tree/master/tools/apollo commit 12282c16658b37858f49944796fd95515ef0fc0b
author gga
date Wed, 11 Jan 2023 11:47:13 +0000
parents 3323060ba9a6
children
line wrap: on
line source

#!/usr/bin/env python
from __future__ import print_function

import argparse
import json
import logging
import os
import re
import shutil
import stat
import subprocess
import sys
import tarfile
import tempfile
import time
from pathlib import Path

from apollo import accessible_organisms
from apollo.util import GuessOrg, OrgOrGuess

from arrow.apollo import get_apollo_instance

from webapollo import UserObj, handle_credentials

logging.basicConfig(level=logging.INFO)
log = logging.getLogger(__name__)


def IsBlatEnabled():
    if 'BLAT_ENABLED' not in os.environ:
        return False
    value = os.environ['BLAT_ENABLED']
    if value.lower() in ('true', 't', '1'):
        return True
    else:
        return False


def IsOrgCNSuffixEnabled():
    if 'GALAXY_APOLLO_ORG_SUFFIX' not in os.environ:
        return False
    value = os.environ['GALAXY_APOLLO_ORG_SUFFIX'].lower()
    if value in ('id', 'email'):
        return value

    return False


def IsRemote():
    return 'GALAXY_SHARED_DIR' not in os.environ or len(os.environ['GALAXY_SHARED_DIR'].lower().strip()) == 0


def zip_data_dir(dataset_data_dir, tar):
    tar.add(dataset_data_dir, arcname='./', recursive=False)
    for r, d, f in os.walk(dataset_data_dir):
        for file_name in f:
            abs_file = os.path.join(r, file_name)
            rel_dir = os.path.relpath(r, dataset_data_dir)
            rel_file = os.path.join(rel_dir, file_name)
            if not rel_file.startswith('./'):
                rel_file = './' + rel_file
            if os.path.islink(abs_file):
                target = Path(abs_file).resolve().absolute().as_posix()
                if re.match(r'.*/_metadata_files/[0-9]+/metadata_[0-9]+.dat', target):
                    # This is a metadata file generated by galaxy, symlink would certainly be dead on remote host, resolve it
                    abs_file = target
            tar.add(abs_file, arcname=rel_file)


if __name__ == '__main__':
    parser = argparse.ArgumentParser(description='Create or update an organism in an Apollo instance')
    parser.add_argument('jbrowse_src', help='Source JBrowse Data Directory')
    parser.add_argument('jbrowse', help='Destination JBrowse Data Directory')
    parser.add_argument('email', help='User Email')
    OrgOrGuess(parser)
    parser.add_argument('--genus', help='Organism Genus')
    parser.add_argument('--species', help='Organism Species')
    parser.add_argument('--public', action='store_true', help='Make organism public')
    parser.add_argument('--group', help='Give access to a user group')
    parser.add_argument('--remove_old_directory', action='store_true', help='Remove old directory')
    parser.add_argument('--no_reload_sequences', action='store_true', help='Disable update genome sequence')
    parser.add_argument('--userid', help='User unique id')
    args = parser.parse_args()
    CHUNK_SIZE = 2**20
    blat_db = None

    path_fasta = args.jbrowse_src + '/seq/genome.fasta'

    # Cleanup if existing
    if not IsRemote():
        if os.path.exists(args.jbrowse):
            shutil.rmtree(args.jbrowse)
        # Copy files
        shutil.copytree(args.jbrowse_src, args.jbrowse, symlinks=True)

        path_2bit = args.jbrowse + '/seq/genome.2bit'
    else:
        twobittemp = tempfile.NamedTemporaryFile(prefix="genome.2bit")
        path_2bit = twobittemp.name
        os.chmod(path_2bit, stat.S_IRUSR | stat.S_IWUSR | stat.S_IRGRP | stat.S_IROTH)

    # Convert fasta if existing
    if IsBlatEnabled() and os.path.exists(path_fasta):
        arg = ['faToTwoBit', path_fasta, path_2bit]
        proc = subprocess.Popen(args=arg, shell=False, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
        out, err = proc.communicate()
        if proc.returncode:
            print("Error building index:", file=sys.stderr)
            sys.stderr.write(err)
            sys.exit(proc.returncode)
        if not IsRemote():
            # No need to send this in remote mode, it will be in the archive
            blat_db = path_2bit

    wa = get_apollo_instance()

    # User must have an account, if not, create it
    gx_user = UserObj(**wa.users._assert_or_create_user(args.email))
    handle_credentials(gx_user)

    org_cn = GuessOrg(args, wa)
    if isinstance(org_cn, list):
        org_cn = org_cn[0]

    if args.org_raw:
        suffix = IsOrgCNSuffixEnabled()
        if suffix == 'id' and args.userid:
            org_cn += ' (gx%s)' % args.userid
        elif suffix == 'email':
            org_cn += ' (%s)' % args.email

    log.info("Determining if add or update required")
    try:
        org = wa.organisms.show_organism(org_cn)
    except Exception:
        org = None

    if org and 'error' not in org:
        old_directory = org['directory']

        all_orgs = wa.organisms.get_organisms()
        if 'error' in all_orgs:
            all_orgs = []
        all_orgs = [x['commonName'] for x in all_orgs]
        if org_cn not in all_orgs:
            raise Exception("Could not find organism %s" % org_cn)

        orgs = accessible_organisms(gx_user, [org_cn], 'WRITE')
        if not orgs:
            raise Exception("Naming Conflict. You do not have write permission on this organism. Either request permission from the owner, or choose a different name for your organism.")

        log.info("\tUpdating Organism")
        if IsRemote():
            with tempfile.NamedTemporaryFile(suffix='.tar.gz') as archive:
                with tarfile.open(archive.name, mode="w:gz") as tar:
                    dataset_data_dir = args.jbrowse_src
                    zip_data_dir(dataset_data_dir, tar)
                    if IsBlatEnabled():
                        tar.add(path_2bit, arcname="./searchDatabaseData/genome.2bit")
                data = wa.remote.update_organism(
                    org['id'],
                    archive,
                    # mandatory
                    blatdb=blat_db,
                    genus=args.genus,
                    species=args.species,
                    public=args.public,
                    no_reload_sequences=args.no_reload_sequences
                )
        else:
            data = wa.organisms.update_organism(
                org['id'],
                org_cn,
                args.jbrowse,
                # mandatory
                genus=args.genus,
                species=args.species,
                public=args.public,
                blatdb=blat_db,
                no_reload_sequences=args.no_reload_sequences
            )
        time.sleep(2)

        if not IsRemote() and args.remove_old_directory and args.jbrowse != old_directory:
            shutil.rmtree(old_directory)

        data = wa.organisms.show_organism(org_cn)

    else:
        # New organism
        log.info("\tAdding Organism")

        if IsRemote():
            with tempfile.NamedTemporaryFile(suffix='.tar.gz') as archive:
                with tarfile.open(archive.name, mode="w:gz") as tar:
                    dataset_data_dir = args.jbrowse_src
                    zip_data_dir(dataset_data_dir, tar)
                    if IsBlatEnabled():
                        with tempfile.TemporaryDirectory() as empty_dir:
                            os.chmod(empty_dir, stat.S_IRUSR | stat.S_IXUSR | stat.S_IWUSR | stat.S_IRGRP | stat.S_IXGRP | stat.S_IROTH | stat.S_IXOTH)
                            tar.add(empty_dir, arcname="./searchDatabaseData/")
                            tar.add(path_2bit, arcname="./searchDatabaseData/genome.2bit")
                data = wa.remote.add_organism(
                    org_cn,
                    archive,
                    blatdb=blat_db,
                    genus=args.genus,
                    species=args.species,
                    public=args.public,
                    metadata=None
                )
                if isinstance(data, list) and len(data) > 0:
                    data = data[0]
        else:
            data = wa.organisms.add_organism(
                org_cn,
                args.jbrowse,
                blatdb=blat_db,
                genus=args.genus,
                species=args.species,
                public=args.public,
                metadata=None
            )

        # Must sleep before we're ready to handle
        time.sleep(2)
        log.info("Updating permissions for %s on %s", gx_user, org_cn)
        wa.users.update_organism_permissions(
            gx_user.username,
            org_cn,
            write=True,
            export=True,
            read=True,
        )

        # Group access
        if args.group:
            group = wa.groups.get_groups(name=args.group)[0]
            res = wa.groups.update_organism_permissions(group['name'], org_cn,
                                                        administrate=False, write=True, read=True,
                                                        export=True)

    print(json.dumps(data, indent=2))