comparison hubArchiveCreator.py @ 13:25809f699cb3 draft

planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb

12:747475757cb0

"""

import argparse
import collections
import json
import sys

# Internal dependencies
from Bam import Bam
from BedSimpleRepeats import BedSimpleRepeats
from Bed import Bed
from BigWig import BigWig
from util.Fasta import Fasta
from Gff3 import Gff3
from Gtf import Gtf
from TrackHub import TrackHub


# TODO: Verify each subprocessed dependency is accessible [gff3ToGenePred, genePredToBed, twoBitInfo, faToTwoBit, bedToBigBed, sort


def main(argv):

    parser.add_argument('--user_email', help='Email of the user who launched the Hub Archive Creation')

    parser.add_argument('--genome_name', help='UCSC Genome Browser assembly ID')

    ucsc_tools_path = ''

    toolDirectory = '.'
    extra_files_path = '.'

    # Get the args passed in parameter
    args = parser.parse_args()

    array_inputs_reference_genome = json.loads(args.fasta)

    # TODO: Replace these with the object Fasta
    input_fasta_file = array_inputs_reference_genome["false_path"]

    array_inputs_gtf = args.gtf
    array_inputs_bam = args.bam
    array_inputs_bigwig = args.bigwig

    outputFile = args.output
    json_inputs_data = args.data_json

    json_inputs_data = args.data_json

    inputs_data = json.loads(json_inputs_data)
    # We remove the spaces in ["name"] of inputs_data
    sanitize_name_inputs(inputs_data)

    if args.directory:
        toolDirectory = args.directory
    if args.extra_files_path:
        extra_files_path = args.extra_files_path

    # TODO: Check here all the binaries / tools we need. Exception if missing

    # Create the Track Hub folder
    trackHub = TrackHub(reference_genome, user_email, outputFile, extra_files_path, toolDirectory)

    #trackHub.createZip()

    # We terminate le process and so create a HTML file summarizing all the files
    trackHub.terminate()

    print "\t"
    print "--------------"
    print "Well done guys! Your data are ready to be displayed in UCSC Track Hub."

    sys.exit(0)


def sanitize_name_input(string_to_sanitize):

            if key == input_false_path:
                extensionObject = ExtensionClass(input_false_path, data_value)
                datatype_dictionary.update({data_value["order_index"]: extensionObject})
    return datatype_dictionary

if __name__ == "__main__":
    main(sys.argv)

13:25809f699cb3

"""

import argparse
import collections
import json
import logging
import os
import sys

# Internal dependencies
from Bam import Bam
from BedSimpleRepeats import BedSimpleRepeats
from Bed import Bed
from BigWig import BigWig
from util.Fasta import Fasta
from util.Filters import TraceBackFormatter
from Gff3 import Gff3
from Gtf import Gtf
from TrackHub import TrackHub

# TODO: Verify each subprocessed dependency is accessible [gff3ToGenePred, genePredToBed, twoBitInfo, faToTwoBit, bedToBigBed, sort


def main(argv):

    parser.add_argument('--user_email', help='Email of the user who launched the Hub Archive Creation')

    parser.add_argument('--genome_name', help='UCSC Genome Browser assembly ID')

    parser.add_argument('--debug_mode', action='store_true', help='Allow more details about the errors')

    # Begin init variables

    toolDirectory = '.'
    extra_files_path = '.'

    # Get the args passed in parameter
    args = parser.parse_args()

    extra_files_path = args.extra_files_path
    toolDirectory = args.directory

    #### Logging management ####
    # If we are in Debug mode, also print in stdout the debug dump

    configure_logger(extra_files_path=extra_files_path, debug=args.debug_mode)

    #### END Logging management ####

    array_inputs_reference_genome = json.loads(args.fasta)

    # TODO: Replace these with the object Fasta
    input_fasta_file = array_inputs_reference_genome["false_path"]

    array_inputs_gtf = args.gtf
    array_inputs_bam = args.bam
    array_inputs_bigwig = args.bigwig

    outputFile = args.output

    json_inputs_data = args.data_json

    inputs_data = json.loads(json_inputs_data)
    # We remove the spaces in ["name"] of inputs_data
    sanitize_name_inputs(inputs_data)


    # TODO: Check here all the binaries / tools we need. Exception if missing

    # Create the Track Hub folder
    trackHub = TrackHub(reference_genome, user_email, outputFile, extra_files_path, toolDirectory)

    #trackHub.createZip()

    # We terminate le process and so create a HTML file summarizing all the files
    trackHub.terminate()

    logging.debug('#### End of HubArchiveCreator Debug Mode: Bye! ####')

    sys.exit(0)


def sanitize_name_input(string_to_sanitize):

            if key == input_false_path:
                extensionObject = ExtensionClass(input_false_path, data_value)
                datatype_dictionary.update({data_value["order_index"]: extensionObject})
    return datatype_dictionary

def configure_logger(extra_files_path=None, debug=False):
    if not extra_files_path:
        raise Exception("Extra files path is not set. Stopping the application")


    # All case log: log everything in a .log file
    logger_file_name = ''.join([__name__, '.log'])
    logging_file_path = os.path.join(extra_files_path, logger_file_name)

    logging.basicConfig(filename=logging_file_path, level=logging.DEBUG)

    log_stdout = logging.StreamHandler(sys.stdout)
    if not debug:
        configure_logger_user(log_stdout)
    else:
        configure_logger_dev(log_stdout)

    # stderr configuration
    configure_logger_stderr()

    logging.debug('#### Welcome in HubArchiveCreator Debug Mode ####\n')

def configure_logger_user(log_stdout=None):
    """
    User Logger is defined as following:
        - User needs to have WARN, ERROR and CRITICAL but well formatted / without traceback
            in STDOUT
        - Still access to full, brute and traceback for errors
            in STDERR
        - And further access to debug if needed
            in .log
    :return:
    """
    if not log_stdout:
        raise Exception("No log_stdout given. Stopping the application")

    # stdout for INFO / WARN / ERROR / CRITICAL
    log_stdout.setLevel(logging.INFO)

    formatter = TraceBackFormatter('%(message)s')

    log_stdout.setFormatter(formatter)

    logging.getLogger().addHandler(log_stdout)

def configure_logger_dev(log_stdout=None):
    """
    Dev Logger is defined as following:
        - Dev needs to have WARN, ERROR and CRITICAL but well formatted / without traceback, in stdout
        - Still access to full, brute and traceback in stderr for errors
        - And further access to debug if needed
    :return:
    """
    if not log_stdout:
        raise Exception("No log_stdout given. Stopping the application")
    log_format = '%(message)s'

    # stdout and stderr and both identical for INFO / WARN / ERROR / CRITICAL
    log_stdout.setLevel(logging.DEBUG)

    formatter = logging.Formatter(log_format)

    log_stdout.setFormatter(formatter)

    logging.getLogger().addHandler(log_stdout)

def configure_logger_stderr():
    """
    Configure what should be logged in stderr
    :return:
    """
    log_error = logging.StreamHandler(sys.stderr)
    log_error.setLevel(logging.ERROR)
    log_error_format = '%(message)s'

    formatter_error = logging.Formatter(log_error_format)

    log_error.setFormatter(formatter_error)

    logging.getLogger().addHandler(log_error)

if __name__ == "__main__":
    logging.getLogger(__name__)
    main(sys.argv)