def main(args=None, launch=True):
download_and_extract()
pwd = os.getcwd()
init_parser = ArgumentParser()
init.populate_parser(init_parser)
args = init_parser.parse_args(
['ap', '--name', '"Associated Press 88-90 sample"', '--rebuild', '-q'])
init.main(args)
prep_parser = ArgumentParser()
prep.populate_parser(prep_parser)
args = prep_parser.parse_args(
'ap.ini --lang en --high 2000 --low 5 -q'.split())
prep.main(args)
train_parser = ArgumentParser()
train.populate_parser(train_parser)
args = train_parser.parse_args(
"ap.ini -k 20 40 60 --context-type article --iter 20".split())
train.main(args)
from configparser import RawConfigParser as ConfigParser
config = ConfigParser()
config.read('ap.ini')
config.set("main", "label_module", "topicexplorer.extensions.ap")
config.set("main", "corpus_desc", "ap.md")
config.set("www", "icons", "ap,fingerprint,link")
config.set("www", "fulltext", "True")
shutil.copyfile(get_static_resource_path('demo/ap.md'), 'ap.md')
with open("ap.ini", "w") as configfh:
config.write(configfh)
if launch:
launch_parser = ArgumentParser()
server.populate_parser(launch_parser)
args = launch_parser.parse_args(['ap.ini'])
server.main(args)
def main(args=None, launch=True):
download_and_extract()
pwd = os.getcwd()
init_parser = ArgumentParser()
init.populate_parser(init_parser)
args = init_parser.parse_args(
['ap', '--name', '"Associated Press 88-90 sample"', '--rebuild', '-q'])
init.main(args)
prep_parser = ArgumentParser()
prep.populate_parser(prep_parser)
args = prep_parser.parse_args('ap.ini --lang en --high 2000 --low 5 -q'.split())
prep.main(args)
train_parser = ArgumentParser()
train.populate_parser(train_parser)
args = train_parser.parse_args("ap.ini -k 20 40 60 --context-type article --iter 20".split())
train.main(args)
import topicexplorer.config
config = topicexplorer.config.read('ap.ini')
config.set("main", "label_module", "topicexplorer.extensions.ap")
config.set("main", "corpus_desc", "ap.md")
config.set("www", "icons", "ap,fingerprint,link")
config.set("www", "fulltext", "True")
shutil.copyfile(get_static_resource_path('demo/ap.md'), 'ap.md')
with open("ap.ini", "w") as configfh:
config.write(configfh)
if launch:
launch_parser = ArgumentParser()
server.populate_parser(launch_parser)
args = launch_parser.parse_args(['ap.ini'])
server.main(args)
def main():
"""
The primary CLI function for the Topic Explorer.
"""
# Create the master argparse object.
parser = ThrowingArgumentParser()
# Adding the benchmarks flags.
benchmark_group = parser.add_mutually_exclusive_group()
benchmark_group.add_argument('-t',
'--time',
help="Print execution time",
action='store_true')
benchmark_group.add_argument('-p',
'--profile',
help="""Profile the command.
Optional filename saves results for use with snakeviz, pstats, or
cprofilev. Automatically launches snakeviz, if installed.""",
nargs='?',
metavar='STATS_FILE')
# Using add_subparsers(metavar) until argparse.SUPPRESS support is fixed.
# See issue http://bugs.python.org/issue22848
parsers = parser.add_subparsers(
help="select a command",
parser_class=ArgumentParser,
metavar='{version,demo,update,init,prep,train,launch,notebook,metadata}'
)
version_parser = parsers.add_parser('version',
help="Print the version and exit")
version_parser.set_defaults(func='version')
# Init Parser
parser_init = parsers.add_parser('init',
help="Initialize the topic explorer")
init.populate_parser(parser_init)
parser_init.set_defaults(func="init")
# Prep Parser
parser_prep = parsers.add_parser(
'prep',
help="Prep the corpus",
formatter_class=argparse.RawDescriptionHelpFormatter)
prep.populate_parser(parser_prep)
parser_prep.set_defaults(func="prep")
# Train Parser
parser_train = parsers.add_parser('train', help="Train the LDA models")
train.populate_parser(parser_train)
parser_train.set_defaults(func="train")
# Launch Parser
parser_launch = parsers.add_parser('launch',
help="Serve the trained LDA models")
server.populate_parser(parser_launch)
parser_launch.set_defaults(func="launch")
# Serve Parser
parser_serve = parsers.add_parser(
'serve',
help="Serve a single LDA model, helper for `topicexplorer launch`," +
"rarely called directly")
server.populate_parser(parser_serve)
parser_serve.set_defaults(func="serve")
# Notebook Parser
parser_nb = parsers.add_parser('notebook',
help="Create a set of IPython Notebooks")
notebook.populate_parser(parser_nb)
parser_nb.set_defaults(func="notebook")
# Demo Parser
parser_demo = parsers.add_parser('demo',
help="Download and run the AP demo")
parser_demo.set_defaults(func="demo")
# Update Parser
parser_update = parsers.add_parser('update',
help="Update the Topic Explorer")
parser_update.set_defaults(func="update")
# Metadata Parser
parser_metadata = parsers.add_parser(
'metadata', help="Add spaces before unicode chars")
metadata.populate_parser(parser_metadata)
parser_metadata.set_defaults(func="metadata")
# fancy arg validation for manually injecting tempfile to profile arg
try:
try:
args = parser.parse_args()
except ArgumentParserError as e:
import sys
new_args = sys.argv[1:]
try:
# If the error was thrown by the '-p' argument not having a
# valid file, fix by manually injecting a nargs break
profile = new_args.index('-p')
if (len(new_args) > (profile + 1)
and new_args[profile + 1] in parsers.choices.keys()):
new_args.insert(profile + 1, '-')
args = parser.parse_args(new_args)
else:
raise e
except ValueError:
raise e
except ArgumentParserError as e:
import sys
# Check to see if error occurs with a subparser and cause the exception
# to arise from the subparser instead
for p in parsers.choices.keys():
if p in sys.argv[1:]:
subargs_idx = sys.argv.index(p) + 1
subargs = sys.argv[subargs_idx:]
subparser = locals()['parser_' + p]
# this might cause an error in the subparser, in which case
# we actually want to show that error first
args = subparser.parse_args(subargs)
# Use the default error mechanism for the master parser.
# If the code gets here, it means the error was not in a subparser
ArgumentParser.error(parser, e.message)
if args.profile:
if args.profile == '-':
import tempfile
temphandle, args.profile = tempfile.mkstemp(suffix='.prof',
prefix='vsm.')
print("Saving benchmark data to", args.profile)
from profilehooks import profile
def benchmark(fn):
return profile(fn,
immediate=True,
filename=args.profile,
stdout=None)
elif args.time:
from profilehooks import timecall
def benchmark(fn):
return timecall(fn, immediate=False)
else:
def benchmark(fn):
return fn
if args.func == 'version':
from topicexplorer.version import __pretty_version__
print(__pretty_version__, end='')
elif args.func == 'init':
args.config_file = benchmark(init.main)(args)
print("\nTIP: Only initalizing corpus object and config file.")
print(" Next prepare the corpus using:")
print(" topicexplorer prep", args.config_file)
print(" Or skip directly to training LDA models using:")
print(" topicexplorer train", args.config_file)
elif args.func == 'prep':
benchmark(prep.main)(args)
print("\nTIP: Train the LDA models with:")
print(" topicexplorer train", args.config_file)
elif args.func == 'train':
benchmark(train.main)(args)
if not args.dry_run:
print("\nTIP: launch the topic explorer with:")
print(" topicexplorer launch", args.config_file)
print(" or the notebook server with:")
print(" topicexplorer notebook", args.config_file)
elif args.func == 'launch' or args.func == 'serve':
# Note that we are only benchmarking the creation process - obviously
# benches of the serve process will take longer
app = benchmark(server.create_app)(args)
server.main(args, app)
elif args.func == 'notebook':
benchmark(notebook.main)(args)
elif args.func == 'demo':
benchmark(demo.main)(args)
elif args.func == 'update':
benchmark(update.main)(args)
elif args.func == 'metadata':
benchmark(metadata.main)(args)
if args.profile:
try:
import snakeviz.cli
print("\n\n")
snakeviz.cli.main([args.profile])
except ImportError:
print(
"""\nSnakeviz is not installed. Install with `pip install snakeviz`,
then run `snakeviz {}`.""".format(args.profile))
def main():
parser = ArgumentParser()
parsers = parser.add_subparsers(help="select a command")
# Init Parser
parser_init = parsers.add_parser('init',
help="Initialize the topic explorer")
init.populate_parser(parser_init)
parser_init.set_defaults(func="init")
# Prep Parser
parser_prep = parsers.add_parser(
'prep',
help="Prep the corpus",
formatter_class=argparse.RawDescriptionHelpFormatter)
prep.populate_parser(parser_prep)
parser_prep.set_defaults(func="prep")
# Train Parser
parser_train = parsers.add_parser('train', help="Train the LDA models")
train.populate_parser(parser_train)
parser_train.set_defaults(func="train")
# Launch Parser
parser_launch = parsers.add_parser('launch',
help="Serve the trained LDA models")
launch.populate_parser(parser_launch)
parser_launch.set_defaults(func="launch")
# Serve Parser
parser_serve = parsers.add_parser(
'serve',
help="Serve a single LDA model, helper for `vsm launch`," +
"rarely called directly")
server.populate_parser(parser_serve)
parser_serve.set_defaults(func="serve")
# Notebook Parser
parser_nb = parsers.add_parser('notebook',
help="Create a set of IPython Notebooks")
notebook.populate_parser(parser_nb)
parser_nb.set_defaults(func="notebook")
# Demo Parser
parser_demo = parsers.add_parser('demo',
help="Download and run the AP demo")
parser_demo.set_defaults(func="demo")
args = parser.parse_args()
if args.func == 'init':
args.config_file = init.main(args)
print "\nTIP: Only initalizing corpus object and config file."
print " Next prepare the corpus using:"
print " vsm prep", args.config_file
print " Or skip directly to training LDA models using:"
print " vsm train", args.config_file
elif args.func == 'prep':
prep.main(args)
print "\nTIP: Train the LDA models with:"
print " vsm train", args.config_file
elif args.func == 'train':
train.main(args)
print "\nTIP: launch the topic explorer with:"
print " vsm launch", args.config_file
print " or the notebook server with:"
print " vsm notebook", args.config_file
elif args.func == 'launch':
launch.main(args)
elif args.func == 'serve':
server.main(args)
elif args.func == 'notebook':
notebook.main(args)
elif args.func == 'demo':
demo.main()
def main():
parser = ArgumentParser()
parsers = parser.add_subparsers(help="select a command")
# Init Parser
parser_init = parsers.add_parser('init', help="Initialize the topic explorer")
init.populate_parser(parser_init)
parser_init.set_defaults(func="init")
# Prep Parser
parser_prep = parsers.add_parser('prep', help="Prep the corpus",
formatter_class=argparse.RawDescriptionHelpFormatter)
prep.populate_parser(parser_prep)
parser_prep.set_defaults(func="prep")
# Train Parser
parser_train = parsers.add_parser('train', help="Train the LDA models")
train.populate_parser(parser_train)
parser_train.set_defaults(func="train")
# Launch Parser
parser_launch = parsers.add_parser('launch', help="Serve the trained LDA models")
launch.populate_parser(parser_launch)
parser_launch.set_defaults(func="launch")
# Serve Parser
parser_serve = parsers.add_parser('serve',
help="Serve a single LDA model, helper for `vsm launch`,"+
"rarely called directly")
server.populate_parser(parser_serve)
parser_serve.set_defaults(func="serve")
# Notebook Parser
parser_nb = parsers.add_parser('notebook',
help="Create a set of IPython Notebooks")
notebook.populate_parser(parser_nb)
parser_nb.set_defaults(func="notebook")
# Demo Parser
parser_demo = parsers.add_parser('demo',
help="Download and run the AP demo")
parser_demo.set_defaults(func="demo")
args = parser.parse_args()
if args.func == 'init':
args.config_file = init.main(args)
print "\nTIP: Only initalizing corpus object and config file."
print " Next prepare the corpus using:"
print " vsm prep", args.config_file
print " Or skip directly to training LDA models using:"
print " vsm train", args.config_file
elif args.func == 'prep':
prep.main(args)
print "\nTIP: Train the LDA models with:"
print " vsm train", args.config_file
elif args.func == 'train':
train.main(args)
print "\nTIP: launch the topic explorer with:"
print " vsm launch", args.config_file
print " or the notebook server with:"
print " vsm notebook", args.config_file
elif args.func == 'launch':
launch.main(args)
elif args.func == 'serve':
server.main(args)
elif args.func == 'notebook':
notebook.main(args)
elif args.func == 'demo':
demo.main()
def main():
"""
The primary CLI function for the Topic Explorer.
"""
# Create the master argparse object.
parser = ThrowingArgumentParser()
# Adding the benchmarks flags.
benchmark_group = parser.add_mutually_exclusive_group()
benchmark_group.add_argument('-t', '--time', help="Print execution time",
action='store_true')
benchmark_group.add_argument('-p', '--profile', help="""Profile the command.
Optional filename saves results for use with snakeviz, pstats, or
cprofilev. Automatically launches snakeviz, if installed.""",
nargs='?', metavar='STATS_FILE')
# Using add_subparsers(metavar) until argparse.SUPPRESS support is fixed.
# See issue http://bugs.python.org/issue22848
parsers = parser.add_subparsers(help="select a command",
parser_class=ArgumentParser,
metavar='{version,demo,update,init,prep,train,launch,notebook,metadata}')
version_parser = parsers.add_parser('version', help="Print the version and exit")
version_parser.set_defaults(func='version')
# Init Parser
parser_init = parsers.add_parser('init', help="Initialize the topic explorer")
init.populate_parser(parser_init)
parser_init.set_defaults(func="init")
# Prep Parser
parser_prep = parsers.add_parser('prep', help="Prep the corpus",
formatter_class=argparse.RawDescriptionHelpFormatter)
prep.populate_parser(parser_prep)
parser_prep.set_defaults(func="prep")
# Train Parser
parser_train = parsers.add_parser('train', help="Train the LDA models")
train.populate_parser(parser_train)
parser_train.set_defaults(func="train")
# Launch Parser
parser_launch = parsers.add_parser('launch', help="Serve the trained LDA models")
server.populate_parser(parser_launch)
parser_launch.set_defaults(func="launch")
# Serve Parser
parser_serve = parsers.add_parser('serve',
help="Serve a single LDA model, helper for `topicexplorer launch`," +
"rarely called directly")
server.populate_parser(parser_serve)
parser_serve.set_defaults(func="serve")
# Notebook Parser
parser_nb = parsers.add_parser('notebook',
help="Create a set of IPython Notebooks")
notebook.populate_parser(parser_nb)
parser_nb.set_defaults(func="notebook")
# Demo Parser
parser_demo = parsers.add_parser('demo',
help="Download and run the AP demo")
parser_demo.set_defaults(func="demo")
# Update Parser
parser_update = parsers.add_parser('update',
help="Update the Topic Explorer")
parser_update.set_defaults(func="update")
# Metadata Parser
parser_metadata = parsers.add_parser('metadata',
help="Add spaces before unicode chars")
metadata.populate_parser(parser_metadata)
parser_metadata.set_defaults(func="metadata")
# Export Parser
parser_export = parsers.add_parser('export', help="Export a tez archive")
export.populate_parser(parser_export)
parser_export.set_defaults(func="export")
# Export Parser
parser_import = parsers.add_parser('import', help="Import the tez archive")
tezimport.populate_parser(parser_import)
parser_import.set_defaults(func="import")
# fancy arg validation for manually injecting tempfile to profile arg
try:
try:
args = parser.parse_args()
except ArgumentParserError as e:
import sys
new_args = sys.argv[1:]
try:
# If the error was thrown by the '-p' argument not having a
# valid file, fix by manually injecting a nargs break
profile = new_args.index('-p')
if (len(new_args) > (profile + 1) and
new_args[profile + 1] in parsers.choices.keys()):
new_args.insert(profile + 1, '-')
args = parser.parse_args(new_args)
else:
raise e
except ValueError:
raise e
except ArgumentParserError as e:
import sys
# Check to see if error occurs with a subparser and cause the exception
# to arise from the subparser instead
for p in parsers.choices.keys():
if p in sys.argv[1:]:
subargs_idx = sys.argv.index(p) + 1
subargs = sys.argv[subargs_idx:]
subparser = locals()['parser_' + p]
# this might cause an error in the subparser, in which case
# we actually want to show that error first
args = subparser.parse_args(subargs)
# Use the default error mechanism for the master parser.
# If the code gets here, it means the error was not in a subparser
ArgumentParser.error(parser, e.message)
if args.profile:
if args.profile == '-':
import tempfile
temphandle, args.profile = tempfile.mkstemp(suffix='.prof', prefix='vsm.')
print("Saving benchmark data to", args.profile)
from profilehooks import profile
def benchmark(fn):
return profile(fn, immediate=True, filename=args.profile, stdout=None)
elif args.time:
from profilehooks import timecall
def benchmark(fn):
return timecall(fn, immediate=False)
else:
def benchmark(fn):
return fn
if args.func == 'version':
from topicexplorer.version import __pretty_version__
print(__pretty_version__, end='')
elif args.func == 'init':
args.config_file = benchmark(init.main)(args)
print("\nTIP: Only initalizing corpus object and config file.")
print(" Next prepare the corpus using:")
print(" topicexplorer prep", args.config_file)
print(" Or skip directly to training LDA models using:")
print(" topicexplorer train", args.config_file)
elif args.func == 'prep':
benchmark(prep.main)(args)
print("\nTIP: Train the LDA models with:")
print(" topicexplorer train", args.config_file)
elif args.func == 'train':
benchmark(train.main)(args)
if not args.dry_run:
print("\nTIP: launch the topic explorer with:")
print(" topicexplorer launch", args.config_file)
print(" or the notebook server with:")
print(" topicexplorer notebook", args.config_file)
elif args.func == 'launch' or args.func == 'serve':
# Note that we are only benchmarking the creation process - obviously
# benches of the serve process will take longer
app = benchmark(server.create_app)(args)
server.main(args, app)
elif args.func == 'notebook':
benchmark(notebook.main)(args)
elif args.func == 'demo':
benchmark(demo.main)(args)
elif args.func == 'update':
benchmark(update.main)(args)
elif args.func == 'metadata':
benchmark(metadata.main)(args)
elif args.func == 'export':
benchmark(export.main)(args)
elif args.func == 'import':
benchmark(tezimport.main)(args)
if args.profile:
try:
import snakeviz.cli
print("\n\n")
snakeviz.cli.main([args.profile])
except ImportError:
print("""\nSnakeviz is not installed. Install with `pip install snakeviz`,
then run `snakeviz {}`.""".format(args.profile))