cstrittmatter@0: import pickle
cstrittmatter@0: import traceback
cstrittmatter@0: import shlex
cstrittmatter@0: import subprocess
cstrittmatter@0: from threading import Timer
cstrittmatter@0: import shutil
cstrittmatter@0: import time
cstrittmatter@0: import functools
cstrittmatter@0: import os.path
cstrittmatter@0: import sys
cstrittmatter@0: import argparse
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def start_logger(workdir):
cstrittmatter@0:     time_str = time.strftime("%Y%m%d-%H%M%S")
cstrittmatter@0:     sys.stdout = Logger(workdir, time_str)
cstrittmatter@0:     logfile = sys.stdout.getLogFile()
cstrittmatter@0:     return logfile, time_str
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: class Logger(object):
cstrittmatter@0:     def __init__(self, out_directory, time_str):
cstrittmatter@0:         self.logfile = os.path.join(out_directory, str('run.' + time_str + '.log'))
cstrittmatter@0:         self.terminal = sys.stdout
cstrittmatter@0:         self.log = open(self.logfile, "w")
cstrittmatter@0: 
cstrittmatter@0:     def write(self, message):
cstrittmatter@0:         self.terminal.write(message)
cstrittmatter@0:         self.log.write(message)
cstrittmatter@0:         self.log.flush()
cstrittmatter@0: 
cstrittmatter@0:     def flush(self):
cstrittmatter@0:         pass
cstrittmatter@0: 
cstrittmatter@0:     def getLogFile(self):
cstrittmatter@0:         return self.logfile
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def checkPrograms(programs_version_dictionary):
cstrittmatter@0:     print('\n' + 'Checking dependencies...')
cstrittmatter@0:     programs = programs_version_dictionary
cstrittmatter@0:     which_program = ['which', '']
cstrittmatter@0:     listMissings = []
cstrittmatter@0:     for program in programs:
cstrittmatter@0:         which_program[1] = program
cstrittmatter@0:         run_successfully, stdout, stderr = runCommandPopenCommunicate(which_program, False, None, False)
cstrittmatter@0:         if not run_successfully:
cstrittmatter@0:             listMissings.append(program + ' not found in PATH.')
cstrittmatter@0:         else:
cstrittmatter@0:             print(stdout.splitlines()[0])
cstrittmatter@0:             if programs[program][0] is None:
cstrittmatter@0:                 print(program + ' (impossible to determine programme version) found at: ' + stdout.splitlines()[0])
cstrittmatter@0:             else:
cstrittmatter@0:                 if program.endswith('.jar'):
cstrittmatter@0:                     check_version = ['java', '-jar', stdout.splitlines()[0], programs[program][0]]
cstrittmatter@0:                     programs[program].append(stdout.splitlines()[0])
cstrittmatter@0:                 else:
cstrittmatter@0:                     check_version = [stdout.splitlines()[0], programs[program][0]]
cstrittmatter@0:                 run_successfully, stdout, stderr = runCommandPopenCommunicate(check_version, False, None, False)
cstrittmatter@0:                 if stdout == '':
cstrittmatter@0:                     stdout = stderr
cstrittmatter@0:                 if program in ['wget', 'awk']:
cstrittmatter@0:                     version_line = stdout.splitlines()[0].split(' ', 3)[2]
cstrittmatter@0:                 elif program in ['prefetch', 'fastq-dump']:
cstrittmatter@0:                     version_line = stdout.splitlines()[1].split(' ')[-1]
cstrittmatter@0:                 else:
cstrittmatter@0:                     version_line = stdout.splitlines()[0].split(' ')[-1]
cstrittmatter@0:                 replace_characters = ['"', 'v', 'V', '+', ',']
cstrittmatter@0:                 for i in replace_characters:
cstrittmatter@0:                     version_line = version_line.replace(i, '')
cstrittmatter@0:                 print(program + ' (' + version_line + ') found')
cstrittmatter@0:                 if programs[program][1] == '>=':
cstrittmatter@0:                     program_found_version = version_line.split('.')
cstrittmatter@0:                     program_version_required = programs[program][2].split('.')
cstrittmatter@0:                     if len(program_version_required) == 3:
cstrittmatter@0:                         if len(program_found_version) == 2:
cstrittmatter@0:                             program_found_version.append(0)
cstrittmatter@0:                         else:
cstrittmatter@0:                             program_found_version[2] = program_found_version[2].split('_')[0]
cstrittmatter@0:                     for i in range(0, len(program_version_required)):
cstrittmatter@0:                         if int(program_found_version[i]) > int(program_version_required[i]):
cstrittmatter@0:                             break
cstrittmatter@0:                         elif int(program_found_version[i]) == int(program_version_required[i]):
cstrittmatter@0:                             continue
cstrittmatter@0:                         else:
cstrittmatter@0:                             listMissings.append('It is required ' + program + ' with version ' +
cstrittmatter@0:                                                 programs[program][1] + ' ' + programs[program][2])
cstrittmatter@0:                 else:
cstrittmatter@0:                     if version_line != programs[program][2]:
cstrittmatter@0:                         listMissings.append('It is required ' + program + ' with version ' + programs[program][1] +
cstrittmatter@0:                                             ' ' + programs[program][2])
cstrittmatter@0:     return listMissings
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def requiredPrograms():
cstrittmatter@0:     programs_version_dictionary = {}
cstrittmatter@0:     programs_version_dictionary['rematch.py'] = ['--version', '>=', '4.0']
cstrittmatter@0:     missingPrograms = checkPrograms(programs_version_dictionary)
cstrittmatter@0:     if len(missingPrograms) > 0:
cstrittmatter@0:         sys.exit('\n' + 'Errors:' + '\n' + '\n'.join(missingPrograms))
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def general_information(logfile, version, outdir, time_str):
cstrittmatter@0:     # Check if output directory exists
cstrittmatter@0: 
cstrittmatter@0:     print('\n' + '==========> patho_typing <==========')
cstrittmatter@0:     print('\n' + 'Program start: ' + time.ctime())
cstrittmatter@0: 
cstrittmatter@0:     # Tells where the logfile will be stored
cstrittmatter@0:     print('\n' + 'LOGFILE:')
cstrittmatter@0:     print(logfile)
cstrittmatter@0: 
cstrittmatter@0:     # Print command
cstrittmatter@0:     print('\n' + 'COMMAND:')
cstrittmatter@0:     script_path = os.path.join(os.path.dirname(os.path.dirname(os.path.realpath(__file__))), 'patho_typing.py')
cstrittmatter@0:     print(sys.executable + ' ' + ' '.join(sys.argv))
cstrittmatter@0: 
cstrittmatter@0:     # Print directory where programme was lunch
cstrittmatter@0:     print('\n' + 'PRESENT DIRECTORY:')
cstrittmatter@0:     present_directory = os.path.abspath(os.getcwd())
cstrittmatter@0:     print(present_directory)
cstrittmatter@0: 
cstrittmatter@0:     # Print program version
cstrittmatter@0:     print('\n' + 'VERSION:')
cstrittmatter@0:     script_version_git(version, present_directory, script_path)
cstrittmatter@0: 
cstrittmatter@0:     # Check programms
cstrittmatter@0:     requiredPrograms()
cstrittmatter@0: 
cstrittmatter@0:     return script_path
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def setPATHvariable(doNotUseProvidedSoftware, script_path):
cstrittmatter@0:     path_variable = os.environ['PATH']
cstrittmatter@0:     script_folder = os.path.dirname(script_path)
cstrittmatter@0:     # Set path to use provided softwares
cstrittmatter@0:     if not doNotUseProvidedSoftware:
cstrittmatter@0:         bowtie2 = os.path.join(script_folder, 'src', 'bowtie2-2.2.9')
cstrittmatter@0:         samtools = os.path.join(script_folder, 'src', 'samtools-1.3.1', 'bin')
cstrittmatter@0:         bcftools = os.path.join(script_folder, 'src', 'bcftools-1.3.1', 'bin')
cstrittmatter@0: 
cstrittmatter@0:         os.environ['PATH'] = str(':'.join([bowtie2, samtools, bcftools, path_variable]))
cstrittmatter@0: 
cstrittmatter@0:     # Print PATH variable
cstrittmatter@0:     print('\n' + 'PATH variable:')
cstrittmatter@0:     print(os.environ['PATH'])
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def script_version_git(version, current_directory, script_path, no_git_info=False):
cstrittmatter@0:     """
cstrittmatter@0:     Print script version and get GitHub commit information
cstrittmatter@0: 
cstrittmatter@0:     Parameters
cstrittmatter@0:     ----------
cstrittmatter@0:     version : str
cstrittmatter@0:         Version of the script, e.g. "4.0"
cstrittmatter@0:     current_directory : str
cstrittmatter@0:         Path to the directory where the script was start to run
cstrittmatter@0:     script_path : str
cstrittmatter@0:         Path to the script running
cstrittmatter@0:     no_git_info : bool, default False
cstrittmatter@0:         True if it is not necessary to retreive the GitHub commit information
cstrittmatter@0: 
cstrittmatter@0:     Returns
cstrittmatter@0:     -------
cstrittmatter@0: 
cstrittmatter@0:     """
cstrittmatter@0:     print('Version {}'.format(version))
cstrittmatter@0: 
cstrittmatter@0:     if not no_git_info:
cstrittmatter@0:         try:
cstrittmatter@0:             os.chdir(os.path.dirname(os.path.dirname(script_path)))
cstrittmatter@0:             command = ['git', 'log', '-1', '--date=local', '--pretty=format:"%h (%H) - Commit by %cn, %cd) : %s"']
cstrittmatter@0:             run_successfully, stdout, stderr = runCommandPopenCommunicate(command, False, 15, False)
cstrittmatter@0:             print(stdout)
cstrittmatter@0:             command = ['git', 'remote', 'show', 'origin']
cstrittmatter@0:             run_successfully, stdout, stderr = runCommandPopenCommunicate(command, False, 15, False)
cstrittmatter@0:             print(stdout)
cstrittmatter@0:         except:
cstrittmatter@0:             print('HARMLESS WARNING: git command possibly not found. The GitHub repository information will not be'
cstrittmatter@0:                   ' obtained.')
cstrittmatter@0:         finally:
cstrittmatter@0:             os.chdir(current_directory)
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def runTime(start_time):
cstrittmatter@0:     end_time = time.time()
cstrittmatter@0:     time_taken = end_time - start_time
cstrittmatter@0:     hours, rest = divmod(time_taken, 3600)
cstrittmatter@0:     minutes, seconds = divmod(rest, 60)
cstrittmatter@0:     print('Runtime :' + str(hours) + 'h:' + str(minutes) + 'm:' + str(round(seconds, 2)) + 's')
cstrittmatter@0:     return round(time_taken, 2)
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def timer(function, name):
cstrittmatter@0:     @functools.wraps(function)
cstrittmatter@0:     def wrapper(*args, **kwargs):
cstrittmatter@0:         print('\n' + 'RUNNING {0}\n'.format(name))
cstrittmatter@0:         start_time = time.time()
cstrittmatter@0: 
cstrittmatter@0:         results = list(function(*args, **kwargs))  # guarantees return is a list to allow .insert()
cstrittmatter@0: 
cstrittmatter@0:         time_taken = runTime(start_time)
cstrittmatter@0:         print('END {0}'.format(name))
cstrittmatter@0: 
cstrittmatter@0:         results.insert(0, time_taken)
cstrittmatter@0:         return results
cstrittmatter@0:     return wrapper
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def removeDirectory(directory):
cstrittmatter@0:     if os.path.isdir(directory):
cstrittmatter@0:         shutil.rmtree(directory)
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def saveVariableToPickle(variableToStore, pickleFile):
cstrittmatter@0:     with open(pickleFile, 'wb') as writer:
cstrittmatter@0:         pickle.dump(variableToStore, writer)
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def extractVariableFromPickle(pickleFile):
cstrittmatter@0:     with open(pickleFile, 'rb') as reader:
cstrittmatter@0:         variable = pickle.load(reader)
cstrittmatter@0:     return variable
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def trace_unhandled_exceptions(func):
cstrittmatter@0:     @functools.wraps(func)
cstrittmatter@0:     def wrapped_func(*args, **kwargs):
cstrittmatter@0:         try:
cstrittmatter@0:             func(*args, **kwargs)
cstrittmatter@0:         except:
cstrittmatter@0:             print('Exception in ' + func.__name__)
cstrittmatter@0:             traceback.print_exc()
cstrittmatter@0:     return wrapped_func
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def kill_subprocess_Popen(subprocess_Popen, command):
cstrittmatter@0:     print('Command run out of time: ' + str(command))
cstrittmatter@0:     subprocess_Popen.kill()
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def runCommandPopenCommunicate(command, shell_True, timeout_sec_None, print_comand_True):
cstrittmatter@0:     run_successfully = False
cstrittmatter@0:     if not isinstance(command, str):
cstrittmatter@0:         command = ' '.join(command)
cstrittmatter@0:     command = shlex.split(command)
cstrittmatter@0: 
cstrittmatter@0:     if print_comand_True:
cstrittmatter@0:         print('Running: ' + ' '.join(command))
cstrittmatter@0: 
cstrittmatter@0:     if shell_True:
cstrittmatter@0:         command = ' '.join(command)
cstrittmatter@0:         proc = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=True)
cstrittmatter@0:     else:
cstrittmatter@0:         proc = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
cstrittmatter@0: 
cstrittmatter@0:     not_killed_by_timer = True
cstrittmatter@0:     if timeout_sec_None is None:
cstrittmatter@0:         stdout, stderr = proc.communicate()
cstrittmatter@0:     else:
cstrittmatter@0:         time_counter = Timer(timeout_sec_None, kill_subprocess_Popen, args=(proc, command,))
cstrittmatter@0:         time_counter.start()
cstrittmatter@0:         stdout, stderr = proc.communicate()
cstrittmatter@0:         time_counter.cancel()
cstrittmatter@0:         not_killed_by_timer = time_counter.isAlive()
cstrittmatter@0: 
cstrittmatter@0:     stdout = stdout.decode("utf-8")
cstrittmatter@0:     stderr = stderr.decode("utf-8")
cstrittmatter@0: 
cstrittmatter@0:     if proc.returncode == 0:
cstrittmatter@0:         run_successfully = True
cstrittmatter@0:     else:
cstrittmatter@0:         if not print_comand_True and not_killed_by_timer:
cstrittmatter@0:             print('Running: ' + str(command))
cstrittmatter@0:         if len(stdout) > 0:
cstrittmatter@0:             print('STDOUT')
cstrittmatter@0:             print(stdout)
cstrittmatter@0:         if len(stderr) > 0:
cstrittmatter@0:             print('STDERR')
cstrittmatter@0:             print(stderr)
cstrittmatter@0:     return run_successfully, stdout, stderr
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def required_length(tuple_length_options, argument_name):
cstrittmatter@0:     class RequiredLength(argparse.Action):
cstrittmatter@0:         def __call__(self, parser, args, values, option_string=None):
cstrittmatter@0:             if len(values) not in tuple_length_options:
cstrittmatter@0:                 msg = 'Option {argument_name} requires one of the following number of' \
cstrittmatter@0:                       ' arguments: {tuple_length_options}'.format(argument_name=self.argument_name,
cstrittmatter@0:                                                                   tuple_length_options=tuple_length_options)
cstrittmatter@0:                 raise argparse.ArgumentTypeError(msg)
cstrittmatter@0:             setattr(args, self.dest, values)
cstrittmatter@0:     return RequiredLength
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def get_sequence_information(fasta_file, length_extra_seq):
cstrittmatter@0:     sequence_dict = {}
cstrittmatter@0:     headers = {}
cstrittmatter@0: 
cstrittmatter@0:     with open(fasta_file, 'rtU') as reader:
cstrittmatter@0:         blank_line_found = False
cstrittmatter@0:         sequence_counter = 0
cstrittmatter@0:         temp_sequence_dict = {}
cstrittmatter@0:         for line in reader:
cstrittmatter@0:             line = line.splitlines()[0]
cstrittmatter@0:             if len(line) > 0:
cstrittmatter@0:                 if not blank_line_found:
cstrittmatter@0:                     if line.startswith('>'):
cstrittmatter@0:                         if len(temp_sequence_dict) > 0:
cstrittmatter@0:                             if list(temp_sequence_dict.values())[0]['length'] - 2 * length_extra_seq > 0:
cstrittmatter@0:                                 sequence_dict[list(temp_sequence_dict.keys())[0]] = list(temp_sequence_dict.values())[0]
cstrittmatter@0:                                 headers[list(temp_sequence_dict.values())[0]['header'].lower()] = sequence_counter
cstrittmatter@0:                             else:
cstrittmatter@0:                                 print(list(temp_sequence_dict.values())[0]['header'] + ' sequence ignored due to '
cstrittmatter@0:                                                                                        'length <= 0')
cstrittmatter@0:                             temp_sequence_dict = {}
cstrittmatter@0: 
cstrittmatter@0:                         if line[1:].lower() in headers:
cstrittmatter@0:                             sys.exit('Found duplicated sequence headers')
cstrittmatter@0: 
cstrittmatter@0:                         sequence_counter += 1
cstrittmatter@0:                         temp_sequence_dict[sequence_counter] = {'header': line[1:].lower(), 'sequence': '', 'length': 0}
cstrittmatter@0:                     else:
cstrittmatter@0:                         temp_sequence_dict[sequence_counter]['sequence'] += line.upper()
cstrittmatter@0:                         temp_sequence_dict[sequence_counter]['length'] += len(line)
cstrittmatter@0:                 else:
cstrittmatter@0:                     sys.exit('It was found a blank line between the fasta file above line ' + line)
cstrittmatter@0:             else:
cstrittmatter@0:                 blank_line_found = True
cstrittmatter@0: 
cstrittmatter@0:         if len(temp_sequence_dict) > 0:
cstrittmatter@0:             if list(temp_sequence_dict.values())[0]['length'] - 2 * length_extra_seq > 0:
cstrittmatter@0:                 sequence_dict[list(temp_sequence_dict.keys())[0]] = list(temp_sequence_dict.values())[0]
cstrittmatter@0:                 headers[list(temp_sequence_dict.values())[0]['header'].lower()] = sequence_counter
cstrittmatter@0:             else:
cstrittmatter@0:                 print(list(temp_sequence_dict.values())[0]['header'] + ' sequence ignored due to length <= 0')
cstrittmatter@0: 
cstrittmatter@0:     return sequence_dict, headers
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def simplify_sequence_dict(sequence_dict):
cstrittmatter@0:     simple_sequence_dict = {}
cstrittmatter@0:     for counter, info in list(sequence_dict.items()):
cstrittmatter@0:         simple_sequence_dict[info['header']] = info
cstrittmatter@0:         del simple_sequence_dict[info['header']]['header']
cstrittmatter@0:     return simple_sequence_dict
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def chunkstring(string, length):
cstrittmatter@0:     return (string[0 + i:length + i] for i in range(0, len(string), length))
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def clean_headers_sequences(sequence_dict):
cstrittmatter@0:     problematic_characters = ["|", " ", ",", ".", "(", ")", "'", "/", ":"]
cstrittmatter@0:     # print 'Checking if reference sequences contain ' + str(problematic_characters) + '\n'
cstrittmatter@0: 
cstrittmatter@0:     headers_changed = False
cstrittmatter@0:     new_headers = {}
cstrittmatter@0:     for i in sequence_dict:
cstrittmatter@0:         if any(x in sequence_dict[i]['header'] for x in problematic_characters):
cstrittmatter@0:             for x in problematic_characters:
cstrittmatter@0:                 sequence_dict[i]['header'] = sequence_dict[i]['header'].replace(x, '_')
cstrittmatter@0:             headers_changed = True
cstrittmatter@0:         new_headers[sequence_dict[i]['header'].lower()] = i
cstrittmatter@0: 
cstrittmatter@0:     if headers_changed:
cstrittmatter@0:         print('At least one of the those characters was found. Replacing those with _' + '\n')
cstrittmatter@0: 
cstrittmatter@0:     return sequence_dict, new_headers