def main():
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument(
"-f",
"--target-file",
type=str,
dest="target_files",
default=[],
action="append",
help=
"Path to file to be tokenized; option may be specified multiple times for multiple files."
)
parser.add_argument(
"-t",
"--target-type",
type=str,
dest="target_types",
default=[],
choices=["trees", "chars", "all"],
action="append",
help=
"Input data file types (default='all' if '-f'/'--file' argument not given); option may be specified multiple times."
)
parser.add_argument(
"-r",
"--repeat",
type=int,
default=10,
help=
"Repeat each tokenization this number of times (default=%(default)s).")
parser.add_argument(
"--delimited-output",
action="store_true",
default=False,
help="Output in tab-delimited instead of aligned format")
args = parser.parse_args()
messenger = messaging.ConsoleMessenger(name="-benchmark")
src_descs = []
src_paths = []
results = []
if args.target_files:
for f in args.target_files:
ff = os.path.expanduser(os.path.expandvars(f))
src_paths.append(ff)
src_descs.append(("User", f))
if not args.target_types and not args.target_files:
messenger.info(
"No sources specified: adding default benchmark target set")
args.target_types = ["all"]
if "all" in args.target_types or "trees" in args.target_types:
for f in TREE_FILENAMES:
ff = pathmap.tree_source_path(f)
src_paths.append(ff)
src_descs.append(("Trees", f))
if "all" in args.target_types or "chars" in args.target_types:
for f in CHAR_FILENAMES:
ff = pathmap.char_source_path(f)
src_paths.append(ff)
src_descs.append(("Alignment", f))
for src_path, src_desc in zip(src_paths, src_descs):
messenger.info("Processing: '{}'".format(src_desc[1]))
t = timeit.Timer(tokenizing_fn_factory([src_path]))
result = min(t.repeat(args.repeat, 1))
messenger.info("Best time (of {} repetions): {:.10f} seconds".format(
args.repeat, result))
results.append(result)
messenger.info("Benchmarking complete: all files processed")
if args.delimited_output:
result_template = "{}\t{}\t{:.10f}\n"
header_template = "{}\t{}\t{}\n"
else:
max_len1 = max(len(r[0]) for r in src_descs)
max_len2 = max(len(r[1]) for r in src_descs)
col1 = "{{:{}}}".format(max_len1)
col2 = "{{:{}}}".format(max_len2)
result_template = "[" + col1 + "] " + col2 + " {:.10f}\n"
header_template = col1 + " " + col2 + " {}\n"
sys.stdout.write(header_template.format("Type", "File", "Seconds"))
for result, src_desc in zip(results, src_descs):
sys.stdout.write(
result_template.format(src_desc[0], src_desc[1], result))
#! /usr/bin/env python
import dendropy
from dendropy.calculate import treescore
from dendropy.utility import messaging
_messenger = messaging.ConsoleMessenger()
trees_src_path = "data/cetaceans.mcmc.nex"
chars_src_path = "data/cetaceans.chars.nex"
burnin = 100
tns = dendropy.TaxonNamespace()
_messenger.info("Reading trees from: {}".format(trees_src_path))
trees = dendropy.TreeList.get(
path=trees_src_path,
schema="nexus",
taxon_namespace=tns,
tree_offset=burnin,
)
_messenger.info("{} trees ({} taxa) read from: {}".format(
len(trees), len(tns), trees_src_path))
_messenger.info("Reading character data from: {}".format(chars_src_path))
chars = dendropy.DnaCharacterMatrix.get(
path=chars_src_path,
schema="nexus",
taxon_namespace=tns,
)
def main():
# description = "%s %s %s" % (_program_name, _program_version, _program_subtitle)
# usage = "%prog [options] TREES-FILE [TREES-FILE [TREES-FILE [...]]"
parser = optparse.OptionParser()
parser.add_option("-s",
"--schema",
action="store",
dest="schema",
default=DEFAULT_SCHEMA_PATH,
help="path to schema (default='%default')")
parser.add_option("-x",
"--xmllint",
action="store_true",
dest="run_xmllint",
default=False,
help="run xmllint as parser")
parser.add_option("-p",
"--xmllint-path",
action="store",
dest="xmllint_path",
default=DEFAULT_XMLLINT_PATH,
help="path to xmllint program (default='%default')")
parser.add_option("-v",
"--verbosity",
action="store",
dest="verbosity",
type="int",
default=3,
help="control noise level")
(opts, args) = parser.parse_args()
if opts.verbosity >= 3:
messaging_level = messaging.ConsoleMessenger.INFO_MESSAGING_LEVEL
elif opts.verbosity >= 2:
messaging_level = messaging.ConsoleMessenger.WARNING_MESSAGING_LEVEL
else:
messaging_level = messaging.ConsoleMessenger.ERROR_MESSAGING_LEVEL
messenger = messaging.ConsoleMessenger(name="nexmlvalidator",
messaging_level=messaging_level)
if len(args) == 0:
sys.exit("No NeXML files to validate specified")
for arg in args:
if opts.run_xmllint:
validator = XmllintValidator(schema_path=opts.schema,
xmllint_path=opts.xmllint_path)
parser = "xmllint"
else:
validator = StandardNexmlValidator(schema_path=opts.schema)
parser = "nexmlvalidator.jar"
validator.validate_nexml(arg)
if validator.is_invalid():
messenger.send_error("[%s]: %s failed to validate with %d errors" %
(parser, arg, len(validator.error_log)),
wrap=False)
if opts.verbosity >= 3:
for fpath, line_num, error in validator.error_log:
messenger.send_info(" % 6d: %s" % (line_num, error),
wrap=False)
else:
messenger.send_info("[%s]: %s validates with %d errors" %
(parser, arg, len(validator.error_log)),
wrap=False)