def __init__(self, search_path='.', max_files_and_dirs_to_process=50000, run=Run(), progress=Progress()):
self.run = run
self.progress = progress
self.search_path = search_path
self.max_files_and_dirs_to_process = max_files_and_dirs_to_process
self.anvio_dbs = {}
for db_path, level in self.walk(depth=3):
db_info = DBInfo(db_path, dont_raise=True)
if db_info is not None:
if db_info.db_type not in self.anvio_dbs:
self.anvio_dbs[db_info.db_type] = []
# Add a cheeky `level` attribute to db_info for no other reason than to order dbs after the walk
db_info.level = level
self.anvio_dbs[db_info.db_type].append(db_info)
# sort by level, so we know what is closest to the search_path root directory
for db_type in self.anvio_dbs:
self.anvio_dbs[db_type] = sorted(self.anvio_dbs[db_type], key=lambda d: d.level)
self.anvio_dbs_found = True
def __init__(self, from_address='admin@localhost', server_address='localhost', server_port=25,
init_tls=False, username = None, password = None, run = Run(verbose = False),
progress = Progress(verbose=False)):
self.from_address = from_address
self.server_address = server_address
self.server_port = server_port
self.init_tls = init_tls
self.username = username
self.password = password
self.server = None
self.config_ini_path = None
self.run = run
self.progress = progress
self.config_template = {
'SMTP': {
'from_address' : {'mandatory': True, 'test': lambda x: str(x)},
'server_address' : {'mandatory': True, 'test': lambda x: str(x)},
'server_port' : {'mandatory': True, 'test': lambda x: RepresentsInt(x) and int(x) > 0, 'required': 'an integer'},
'init_tls' : {'mandatory': True, 'test': lambda x: x in ['True', 'False'], 'required': 'True or False'},
'username' : {'mandatory': True, 'test': lambda x: str(x)},
'password' : {'mandatory': True, 'test': lambda x: str(x)},
},
}
def gen_output_directory(output_directory,
progress=Progress(verbose=False),
run=Run(),
delete_if_exists=False):
if os.path.exists(output_directory) and delete_if_exists:
try:
run.warning(
'filesnpaths::gen_output_directory: the client asked the existing directory \
"%s" to be removed.. Just so you know :/ (You have 5 seconds to press\
CTRL + C).' % output_directory)
time.sleep(5)
shutil.rmtree(output_directory)
except:
progress.end()
raise FilesNPathsError, "I was instructed to remove this directory, but I failed: '%s' :/" % output_directory
if not os.path.exists(output_directory):
try:
os.makedirs(output_directory)
except:
progress.end()
raise FilesNPathsError, "Output directory does not exist (attempt to create one failed as well): '%s'" % \
(output_directory)
if not os.access(output_directory, os.W_OK):
progress.end()
raise FilesNPathsError, "You do not have write permission for the output directory: '%s'" % output_directory
return output_directory
def gen_output_directory(output_directory, progress=Progress(verbose=False), run=Run(), delete_if_exists=False, dont_warn=False):
if not output_directory:
raise FilesNPathsError("Someone called `gen_output_directory` function without an output\
directory name :( An embarrassing moment for everyone involved.")
if os.path.exists(output_directory) and delete_if_exists and not is_dir_empty(output_directory):
try:
if not dont_warn:
run.warning('The existing directory "%s" is about to be removed... (You have \
20 seconds to press CTRL + C). [filesnpaths::gen_output_directory]' % output_directory,
header = '!!! READ THIS NOW !!!')
time.sleep(20)
shutil.rmtree(output_directory)
except:
progress.end()
raise FilesNPathsError("I was instructed to remove this directory, but I failed: '%s' :/" % output_directory)
if not os.path.exists(output_directory):
try:
os.makedirs(output_directory)
except:
progress.end()
raise FilesNPathsError("Output directory does not exist (attempt to create one failed as well): '%s'" % \
(output_directory))
if not os.access(output_directory, os.W_OK):
progress.end()
raise FilesNPathsError("You do not have write permission for the output directory: '%s'" % output_directory)
return output_directory
def K(param_id, params_dict={}):
kwargs = copy.deepcopy(D[param_id][1])
for key in params_dict:
kwargs[key] = params_dict[key]
return kwargs
# The rest of this file is composed of code that responds to '-v' or '--version' calls from clients,
# and provides access to the database version numbers for all anvi'o modules.
import anvio.tables as t
from anvio.terminal import Run
run = Run()
def set_version():
try:
__version__ = pkg_resources.require("anvio")[0].version
except:
# maybe it is not installed but being run from the codebase dir?
try:
__version__ = open(
os.path.normpath(os.path.dirname(os.path.abspath(__file__))) +
'/../VERSION').read().strip()
except:
__version__ = 'unknown'
return __version__, t.contigs_db_version, t.profile_db_version, t.samples_info_db_version, t.auxiliary_hdf5_db_version
# -*- coding: utf-8
"""Classes and functions for handling, storing, and retrieving atomic data from contigs and splits"""
import anvio
from anvio.sequence import Coverage
from anvio.terminal import Run, Progress
from anvio.terminal import pretty_print as pp
from anvio.variability import VariablityTestFactory
import anvio.tables as t
run = Run()
progress = Progress()
progress.verbose = False
try:
from anvio.columnprofile import ColumnProfile
except ImportError:
run.info_single('C extension for ColumnProfile failed to load, falling back to the Python implementation...', mc = 'gray', nl_after = 1)
from anvio.variability import ColumnProfile
__author__ = "A. Murat Eren"
__copyright__ = "Copyright 2015, The anvio Project"
__credits__ = ["Faruk Uzun"]
__license__ = "GPL 3.0"
__version__ = anvio.__version__
__maintainer__ = "A. Murat Eren"
__email__ = "*****@*****.**"
__status__ = "Development"
# -*- coding: utf-8
"""Classes and functions for handling, storing, and retrieving atomic data from contigs and splits"""
import anvio
from anvio.sequence import Coverage
from anvio.terminal import Run, Progress
from anvio.terminal import pretty_print as pp
from anvio.variability import VariablityTestFactory
import anvio.tables as t
run = Run()
progress = Progress()
progress.verbose = False
try:
from anvio.columnprofile import ColumnProfile
except ImportError:
run.info_single(
'C extension for ColumnProfile failed to load, falling back to the Python implementation...',
mc='gray',
nl_after=1)
from anvio.variability import ColumnProfile
__author__ = "A. Murat Eren"
__copyright__ = "Copyright 2015, The anvio Project"
__credits__ = ["Faruk Uzun"]
__license__ = "GPL 3.0"
__version__ = anvio.__version__
__maintainer__ = "A. Murat Eren"
#!/usr/bin/env python
# -*- coding: utf-8
import time
from anvio.columnprofile import ColumnProfile as ColumnProfile_in_C
from anvio.variability import ColumnProfile as ColumnProfile_in_Python
# setup the test factory:
from anvio.variability import VariablityTestFactory
variability_test_class = VariablityTestFactory()
# pretty output
from anvio.summaryhtml import pretty
from anvio.terminal import Run
run = Run(width=55)
# available profiles dict:
ColumnProfile = {'Python': ColumnProfile_in_Python, 'C': ColumnProfile_in_C}
# test function:
def test(column,
consensus='A',
quiet=False,
test_class=variability_test_class):
coverage = len(column)
results = {}
if not quiet:
__author__ = "A. Murat Eren"
__copyright__ = "Copyright 2015, The anvio Project"
__credits__ = []
__license__ = "GPL 3.0"
__version__ = anvio.__version__
__maintainer__ = "A. Murat Eren"
__email__ = "*****@*****.**"
__status__ = "Development"
# Mock progress object that will not report anything, for general clarity.
progress = Progress()
progress.verbose = False
run = Run()
run.verbose = False
def rev_comp(seq):
return seq.translate(complements)[::-1]
class Multiprocessing:
def __init__(self, target_function, num_thread = None):
self.cpu_count = multiprocessing.cpu_count()
self.num_thread = num_thread or (self.cpu_count - (int(round(self.cpu_count / 10.0)) or 1))
self.target_function = target_function
self.processes = []
self.manager = multiprocessing.Manager()
__author__ = "A. Murat Eren"
__copyright__ = "Copyright 2015, The anvio Project"
__credits__ = []
__license__ = "GPL 3.0"
__version__ = anvio.__version__
__maintainer__ = "A. Murat Eren"
__email__ = "*****@*****.**"
__status__ = "Development"
# Mock progress object that will not report anything, for general clarity.
progress = Progress()
progress.verbose = False
run = Run()
run.verbose = False
def rev_comp(seq):
return seq.translate(complements)[::-1]
class Multiprocessing:
def __init__(self, target_function, num_thread=None):
self.cpu_count = multiprocessing.cpu_count()
self.num_thread = num_thread or (self.cpu_count - (int(round(self.cpu_count / 10.0)) or 1))
self.target_function = target_function
self.processes = []
self.manager = multiprocessing.Manager()
# -*- coding: utf-8
"""Classes for handling contigs and splits"""
import anvio
from anvio.terminal import Run
from anvio.variability import VariablityTestFactory
from anvio.sequence import Coverage
run = Run()
try:
from anvio.columnprofile import ColumnProfile
except ImportError:
run.info_single(
"C extension for ColumnProfile failed to load, falling back to the Python implementation...",
mc="gray",
nl_after=1,
)
from anvio.variability import ColumnProfile
__author__ = "A. Murat Eren"
__copyright__ = "Copyright 2015, The anvio Project"
__credits__ = ["Faruk Uzun"]
__license__ = "GPL 3.0"
__version__ = anvio.__version__
__maintainer__ = "A. Murat Eren"
__email__ = "*****@*****.**"
__status__ = "Development"