def load_interface(spec):
"""Load and return the class for `spec`.
Parameters
----------
spec : tuple
For a standard interface, the first item is the datalad source module
and the second object name for the interface. For a plugin, the second
item should be a dictionary that maps 'file' to the path the of module.
Returns
-------
The interface class or, if importing the module fails, None.
"""
if isinstance(spec[1], dict):
intf = _load_plugin(spec[1]['file'], fail=False)
else:
lgr.log(5, "Importing module %s " % spec[0])
try:
mod = import_module(spec[0], package='datalad')
except Exception as e:
lgr.error("Internal error, cannot import interface '%s': %s",
spec[0], exc_str(e))
intf = None
else:
intf = getattr(mod, spec[1])
return intf
def load_interface(spec):
"""Load and return the class for `spec`.
Parameters
----------
spec : tuple
For a standard interface, the first item is the datalad source module
and the second object name for the interface. For a plugin, the second
item should be a dictionary that maps 'file' to the path the of module.
Returns
-------
The interface class or, if importing the module fails, None.
"""
if isinstance(spec[1], dict):
intf = _load_plugin(spec[1]['file'], fail=False)
else:
lgr.log(5, "Importing module %s " % spec[0])
try:
mod = import_module(spec[0], package='datalad')
except Exception as e:
lgr.error("Internal error, cannot import interface '%s': %s",
spec[0], exc_str(e))
intf = None
else:
intf = getattr(mod, spec[1])
return intf
def _load_plugins():
from datalad.plugin import _get_plugins
from datalad.plugin import _load_plugin
import re
camel = re.compile(r'([a-z])([A-Z])')
for pname, props in _get_plugins():
pi = _load_plugin(props['file'], fail=False)
if pi is None:
continue
globals()[camel.sub('\\1_\\2', pi.__name__).lower()] = pi.__call__
def _load_plugins():
from datalad.plugin import _get_plugins
from datalad.plugin import _load_plugin
import re
camel = re.compile(r'([a-z])([A-Z])')
for pname, props in _get_plugins():
pi = _load_plugin(props['file'], fail=False)
if pi is None:
continue
globals()[camel.sub('\\1_\\2', pi.__name__).lower()] = pi.__call__
def setup_parser(
cmdlineargs,
formatter_class=argparse.RawDescriptionHelpFormatter,
return_subparsers=False):
lgr.log(5, "Starting to setup_parser")
# delay since it can be a heavy import
from ..interface.base import dedent_docstring, get_interface_groups, \
get_cmdline_command_name, alter_interface_docs_for_cmdline
# setup cmdline args parser
parts = {}
# main parser
parser = ArgumentParserDisableAbbrev(
# cannot use '@' because we need to input JSON-LD properties (which might come wit @ prefix)
# MH: question, do we need this at all?
fromfile_prefix_chars=':',
# usage="%(prog)s ...",
description=dedent_docstring("""\
DataLad provides a unified data distribution with the convenience of git-annex
repositories as a backend. DataLad command line tools allow to manipulate
(obtain, create, update, publish, etc.) datasets and their collections."""),
epilog='"Control Your Data"',
formatter_class=formatter_class,
add_help=False)
# common options
helpers.parser_add_common_opt(parser, 'log_level')
helpers.parser_add_common_opt(parser, 'pbs_runner')
helpers.parser_add_common_opt(parser, 'change_path')
helpers.parser_add_common_opt(
parser,
'version',
version='datalad %s\n\n%s' % (datalad.__version__, _license_info()))
if __debug__:
parser.add_argument(
'--dbg', action='store_true', dest='common_debug',
help="enter Python debugger when uncaught exception happens")
parser.add_argument(
'--idbg', action='store_true', dest='common_idebug',
help="enter IPython debugger when uncaught exception happens")
parser.add_argument(
'-c', action='append', dest='cfg_overrides', metavar='KEY=VALUE',
help="""configuration variable setting. Overrides any configuration
read from a file, but is potentially overridden itself by configuration
variables in the process environment.""")
parser.add_argument(
'-f', '--output-format', dest='common_output_format',
default='default',
type=assure_unicode,
metavar="{default,json,json_pp,tailored,'<template>'}",
help="""select format for returned command results. 'default' give one line
per result reporting action, status, path and an optional message;
'json' renders a JSON object with all properties for each result (one per
line); 'json_pp' pretty-prints JSON spanning multiple lines; 'tailored'
enables a command-specific rendering style that is typically
tailored to human consumption (no result output otherwise),
'<template>' reports any value(s) of any result properties in any format
indicated by the template (e.g. '{path}'; compare with JSON
output for all key-value choices). The template syntax follows the Python
"format() language". It is possible to report individual
dictionary values, e.g. '{metadata[name]}'. If a 2nd-level key contains
a colon, e.g. 'music:Genre', ':' must be substituted by '#' in the template,
like so: '{metadata[music#Genre]}'.""")
parser.add_argument(
'--report-status', dest='common_report_status',
choices=['success', 'failure', 'ok', 'notneeded', 'impossible', 'error'],
help="""constrain command result report to records matching the given
status. 'success' is a synonym for 'ok' OR 'notneeded', 'failure' stands
for 'impossible' OR 'error'.""")
parser.add_argument(
'--report-type', dest='common_report_type',
choices=['dataset', 'file'],
action='append',
help="""constrain command result report to records matching the given
type. Can be given more than once to match multiple types.""")
parser.add_argument(
'--on-failure', dest='common_on_failure',
choices=['ignore', 'continue', 'stop'],
# no default: better be configure per-command
help="""when an operation fails: 'ignore' and continue with remaining
operations, the error is logged but does not lead to a non-zero exit code
of the command; 'continue' works like 'ignore', but an error causes a
non-zero exit code; 'stop' halts on first failure and yields non-zero exit
code. A failure is any result with status 'impossible' or 'error'.""")
parser.add_argument(
'--run-before', dest='common_run_before',
nargs='+',
action='append',
metavar=('<PROCEDURE NAME>', 'ARGS'),
help="""Dataset procedure to run before the main command (see run-procedure
command for details). This option can be given more than once to run
multiple procedures in the order in which they were given.
It is important to specify the target dataset via the --dataset argument
of the main command."""),
parser.add_argument(
'--run-after', dest='common_run_after',
nargs='+',
action='append',
metavar=('<PROCEDURE NAME>', 'ARGS'),
help="""Like --run-before, but procedures are executed after the main command
has finished."""),
parser.add_argument(
'--cmd', dest='_', action='store_true',
help="""syntactical helper that can be used to end the list of global
command line options before the subcommand label. Options like
--run-before can take an arbitrary number of arguments and may require
to be followed by a single --cmd in order to enable identification
of the subcommand.""")
# yoh: atm we only dump to console. Might adopt the same separation later on
# and for consistency will call it --verbose-level as well for now
# log-level is set via common_opts ATM
# parser.add_argument('--log-level',
# choices=('critical', 'error', 'warning', 'info', 'debug'),
# dest='common_log_level',
# help="""level of verbosity in log files. By default
# everything, including debug messages is logged.""")
#parser.add_argument('-l', '--verbose-level',
# choices=('critical', 'error', 'warning', 'info', 'debug'),
# dest='common_verbose_level',
# help="""level of verbosity of console output. By default
# only warnings and errors are printed.""")
# Before doing anything additional and possibly expensive see may be that
# we have got the command already
need_single_subparser = False if return_subparsers else None
fail_handler = (lambda *a, **kw: True) \
if return_subparsers else fail_with_short_help
try:
parsed_args, unparsed_args = parser._parse_known_args(
cmdlineargs[1:], argparse.Namespace())
if not unparsed_args:
fail_handler(parser, msg="too few arguments", exit_code=2)
lgr.debug("Command line args 1st pass. Parsed: %s Unparsed: %s",
parsed_args, unparsed_args)
except Exception as exc:
lgr.debug("Early parsing failed with %s", exc_str(exc))
need_single_subparser = False
unparsed_args = cmdlineargs[1:] # referenced before assignment otherwise
interface_groups = get_interface_groups()
# TODO: see if we could retain "generator" for plugins
# ATM we need to make it explicit so we could check the command(s) below
# It could at least follow the same destiny as extensions so we would
# just do more iterative "load ups"
interface_groups.append(('plugins', 'Plugins', list(_get_plugins())))
# First unparsed could be either unknown option to top level "datalad"
# or a command. Among unknown could be --help/--help-np which would
# need to be dealt with
unparsed_arg = unparsed_args[0] if unparsed_args else None
if need_single_subparser is not None \
or unparsed_arg in ('--help', '--help-np', '-h'):
need_single_subparser = False
add_entrypoints_to_interface_groups(interface_groups)
elif unparsed_arg.startswith('-'): # unknown option
fail_with_short_help(parser,
msg="unrecognized argument %s" % unparsed_arg,
exit_code=2)
# if we could get a list of options known to parser,
# we could suggest them
# known=get_all_options(parser), provided=unparsed_arg)
else: # the command to handle
known_commands = get_commands_from_groups(interface_groups)
if unparsed_arg not in known_commands:
# need to load all the extensions and try again
add_entrypoints_to_interface_groups(interface_groups)
known_commands = get_commands_from_groups(interface_groups)
if unparsed_arg not in known_commands:
# check if might be coming from known extensions
from ..interface import _known_extension_commands
extension_commands = {
c: e
for e, commands in _known_extension_commands.items()
for c in commands
}
hint = None
if unparsed_arg in extension_commands:
hint = "Command %s is provided by (not installed) extension %s." \
% (unparsed_arg, extension_commands[unparsed_arg])
fail_with_short_help(
parser,
hint=hint,
provided=unparsed_arg,
known=list(known_commands.keys()) + list(extension_commands.keys())
)
if need_single_subparser is None:
need_single_subparser = unparsed_arg
# --help specification was delayed since it causes immediate printout of
# --help output before we setup --help for each command
helpers.parser_add_common_opt(parser, 'help')
grp_short_descriptions = []
# create subparser, use module suffix as cmd name
subparsers = parser.add_subparsers()
for _, _, _interfaces \
in sorted(interface_groups, key=lambda x: x[1]):
# for all subcommand modules it can find
cmd_short_descriptions = []
for _intfspec in _interfaces:
cmd_name = get_cmdline_command_name(_intfspec)
if need_single_subparser and cmd_name != need_single_subparser:
continue
if isinstance(_intfspec[1], dict):
# plugin
_intf = _load_plugin(_intfspec[1]['file'], fail=False)
if _intf is None:
# TODO: add doc why we could skip this one... makes this
# loop harder to extract into a dedicated function
continue
else:
# turn the interface spec into an instance
lgr.log(5, "Importing module %s " % _intfspec[0])
try:
_mod = import_module(_intfspec[0], package='datalad')
except Exception as e:
lgr.error("Internal error, cannot import interface '%s': %s",
_intfspec[0], exc_str(e))
continue
_intf = getattr(_mod, _intfspec[1])
# deal with optional parser args
if hasattr(_intf, 'parser_args'):
parser_args = _intf.parser_args
else:
parser_args = dict(formatter_class=formatter_class)
# use class description, if no explicit description is available
intf_doc = '' if _intf.__doc__ is None else _intf.__doc__.strip()
if hasattr(_intf, '_docs_'):
# expand docs
intf_doc = intf_doc.format(**_intf._docs_)
parser_args['description'] = alter_interface_docs_for_cmdline(
intf_doc)
subparser = subparsers.add_parser(cmd_name, add_help=False, **parser_args)
# our own custom help for all commands
helpers.parser_add_common_opt(subparser, 'help')
# let module configure the parser
_intf.setup_parser(subparser)
# logger for command
# configure 'run' function for this command
plumbing_args = dict(
func=_intf.call_from_parser,
logger=logging.getLogger(_intf.__module__),
subparser=subparser)
if hasattr(_intf, 'result_renderer_cmdline'):
plumbing_args['result_renderer'] = _intf.result_renderer_cmdline
subparser.set_defaults(**plumbing_args)
parts[cmd_name] = subparser
# store short description for later
sdescr = getattr(_intf, 'short_description',
parser_args['description'].split('\n')[0])
cmd_short_descriptions.append((cmd_name, sdescr))
grp_short_descriptions.append(cmd_short_descriptions)
# create command summary
if '--help' in cmdlineargs or '--help-np' in cmdlineargs:
parser.description = get_description_with_cmd_summary(
grp_short_descriptions,
interface_groups,
parser.description)
parts['datalad'] = parser
lgr.log(5, "Finished setup_parser")
if return_subparsers:
return parts
else:
return parser
def setup_parser(
cmdlineargs,
formatter_class=argparse.RawDescriptionHelpFormatter,
return_subparsers=False):
lgr.log(5, "Starting to setup_parser")
# delay since it can be a heavy import
from ..interface.base import dedent_docstring, get_interface_groups, \
get_cmdline_command_name, alter_interface_docs_for_cmdline
# setup cmdline args parser
parts = {}
# main parser
parser = argparse.ArgumentParser(
# cannot use '@' because we need to input JSON-LD properties (which might come wit @ prefix)
# MH: question, do we need this at all?
fromfile_prefix_chars=':',
# usage="%(prog)s ...",
description=dedent_docstring("""\
DataLad provides a unified data distribution with the convenience of git-annex
repositories as a backend. DataLad command line tools allow to manipulate
(obtain, create, update, publish, etc.) datasets and their collections."""),
epilog='"Control Your Data"',
formatter_class=formatter_class,
add_help=False)
# common options
helpers.parser_add_common_opt(parser, 'help')
helpers.parser_add_common_opt(parser, 'log_level')
helpers.parser_add_common_opt(parser, 'pbs_runner')
helpers.parser_add_common_opt(parser, 'change_path')
helpers.parser_add_common_opt(
parser,
'version',
version='datalad %s\n\n%s' % (datalad.__version__, _license_info()))
if __debug__:
parser.add_argument(
'--dbg', action='store_true', dest='common_debug',
help="enter Python debugger when uncaught exception happens")
parser.add_argument(
'--idbg', action='store_true', dest='common_idebug',
help="enter IPython debugger when uncaught exception happens")
parser.add_argument(
'-c', action='append', dest='cfg_overrides', metavar='KEY=VALUE',
help="""configuration variable setting. Overrides any configuration
read from a file, but is potentially overridden itself by configuration
variables in the process environment.""")
parser.add_argument(
'-f', '--output-format', dest='common_output_format',
default='default',
type=assure_unicode,
metavar="{default,json,json_pp,tailored,'<template>'}",
help="""select format for returned command results. 'default' give one line
per result reporting action, status, path and an optional message;
'json' renders a JSON object with all properties for each result (one per
line); 'json_pp' pretty-prints JSON spanning multiple lines; 'tailored'
enables a command-specific rendering style that is typically
tailored to human consumption (no result output otherwise),
'<template>' reports any value(s) of any result properties in any format
indicated by the template (e.g. '{path}'; compare with JSON
output for all key-value choices). The template syntax follows the Python
"format() language". It is possible to report individual
dictionary values, e.g. '{metadata[name]}'. If a 2nd-level key contains
a colon, e.g. 'music:Genre', ':' must be substituted by '#' in the template,
like so: '{metadata[music#Genre]}'.""")
parser.add_argument(
'--report-status', dest='common_report_status',
choices=['success', 'failure', 'ok', 'notneeded', 'impossible', 'error'],
help="""constrain command result report to records matching the given
status. 'success' is a synonym for 'ok' OR 'notneeded', 'failure' stands
for 'impossible' OR 'error'.""")
parser.add_argument(
'--report-type', dest='common_report_type',
choices=['dataset', 'file'],
action='append',
help="""constrain command result report to records matching the given
type. Can be given more than once to match multiple types.""")
parser.add_argument(
'--on-failure', dest='common_on_failure',
choices=['ignore', 'continue', 'stop'],
# no default: better be configure per-command
help="""when an operation fails: 'ignore' and continue with remaining
operations, the error is logged but does not lead to a non-zero exit code
of the command; 'continue' works like 'ignore', but an error causes a
non-zero exit code; 'stop' halts on first failure and yields non-zero exit
code. A failure is any result with status 'impossible' or 'error'.""")
#parser.add_argument(
# '--run-before', dest='common_run_before',
# nargs='+',
# action='append',
# metavar='PLUGINSPEC',
# help="""DataLad plugin to run after the command. PLUGINSPEC is a list
# comprised of a plugin name plus optional `key=value` pairs with arguments
# for the plugin call (see `plugin` command documentation for details).
# This option can be given more than once to run multiple plugins
# in the order in which they were given.
# For running plugins that require a --dataset argument it is important
# to provide the respective dataset as the --dataset argument of the main
# command, if it is not in the list of plugin arguments."""),
#parser.add_argument(
# '--run-after', dest='common_run_after',
# nargs='+',
# action='append',
# metavar='PLUGINSPEC',
# help="""Like --run-before, but plugins are executed after the main command
# has finished."""),
parser.add_argument(
'--cmd', dest='_', action='store_true',
help="""syntactical helper that can be used to end the list of global
command line options before the subcommand label. Options like
--run-before can take an arbitray number of arguments and may require
to be followed by a single --cmd in order to enable identification
of the subcommand.""")
# yoh: atm we only dump to console. Might adopt the same separation later on
# and for consistency will call it --verbose-level as well for now
# log-level is set via common_opts ATM
# parser.add_argument('--log-level',
# choices=('critical', 'error', 'warning', 'info', 'debug'),
# dest='common_log_level',
# help="""level of verbosity in log files. By default
# everything, including debug messages is logged.""")
#parser.add_argument('-l', '--verbose-level',
# choices=('critical', 'error', 'warning', 'info', 'debug'),
# dest='common_verbose_level',
# help="""level of verbosity of console output. By default
# only warnings and errors are printed.""")
# subparsers
subparsers = None
# auto detect all available interfaces and generate a function-based
# API from them
cmdlineargs = set(cmdlineargs) if cmdlineargs else set()
grp_short_descriptions = []
interface_groups = get_interface_groups()
interface_groups.append(('plugins', 'Plugins', _get_plugins()))
for grp_name, grp_descr, _interfaces \
in sorted(interface_groups, key=lambda x: x[1]):
# for all subcommand modules it can find
cmd_short_descriptions = []
for _intfspec in _interfaces:
cmd_name = get_cmdline_command_name(_intfspec)
# for each interface to be imported decide if it is necessary
# test conditions from frequent to infrequent occasions
# we want to import everything for help requests of any kind
# including a cluecless `datalad` without args
if not (len(cmdlineargs) == 1 or
cmd_name in cmdlineargs or
'--help' in cmdlineargs or
'-h' in cmdlineargs or
'--help-np' in cmdlineargs):
continue
if isinstance(_intfspec[1], dict):
# plugin
_intf = _load_plugin(_intfspec[1]['file'], fail=False)
if _intf is None:
continue
else:
# turn the interface spec into an instance
lgr.log(5, "Importing module %s " % _intfspec[0])
try:
_mod = import_module(_intfspec[0], package='datalad')
except Exception as e:
lgr.error("Internal error, cannot import interface '%s': %s",
_intfspec[0], exc_str(e))
continue
_intf = getattr(_mod, _intfspec[1])
# deal with optional parser args
if hasattr(_intf, 'parser_args'):
parser_args = _intf.parser_args
else:
parser_args = dict(formatter_class=formatter_class)
# use class description, if no explicit description is available
intf_doc = '' if _intf.__doc__ is None else _intf.__doc__.strip()
if hasattr(_intf, '_docs_'):
# expand docs
intf_doc = intf_doc.format(**_intf._docs_)
parser_args['description'] = alter_interface_docs_for_cmdline(
intf_doc)
# create subparser, use module suffix as cmd name
if subparsers is None:
subparsers = parser.add_subparsers()
subparser = subparsers.add_parser(cmd_name, add_help=False, **parser_args)
# our own custom help for all commands
helpers.parser_add_common_opt(subparser, 'help')
# let module configure the parser
_intf.setup_parser(subparser)
# logger for command
# configure 'run' function for this command
plumbing_args = dict(
func=_intf.call_from_parser,
logger=logging.getLogger(_intf.__module__),
subparser=subparser)
if hasattr(_intf, 'result_renderer_cmdline'):
plumbing_args['result_renderer'] = _intf.result_renderer_cmdline
subparser.set_defaults(**plumbing_args)
# store short description for later
sdescr = getattr(_intf, 'short_description',
parser_args['description'].split('\n')[0])
cmd_short_descriptions.append((cmd_name, sdescr))
parts[cmd_name] = subparser
grp_short_descriptions.append(cmd_short_descriptions)
# create command summary
cmd_summary = []
console_width = shutil.get_terminal_size()[0] \
if hasattr(shutil, 'get_terminal_size') else 80
for i, grp in enumerate(
sorted(interface_groups, key=lambda x: x[1])):
grp_descr = grp[1]
grp_cmds = grp_short_descriptions[i]
cmd_summary.append('\n*%s*\n' % (grp_descr,))
for cd in grp_cmds:
cmd_summary.append(' %s\n%s'
% ((cd[0],
textwrap.fill(
cd[1].rstrip(' .'),
console_width - 5,
initial_indent=' ' * 6,
subsequent_indent=' ' * 6))))
# we need one last formal section to not have the trailed be
# confused with the last command group
cmd_summary.append('\n*General information*\n')
parser.description = '%s\n%s\n\n%s' \
% (parser.description,
'\n'.join(cmd_summary),
textwrap.fill(dedent_docstring("""\
Detailed usage information for individual commands is
available via command-specific --help, i.e.:
datalad <command> --help"""),
console_width - 5, initial_indent='', subsequent_indent=''))
parts['datalad'] = parser
lgr.log(5, "Finished setup_parser")
if return_subparsers:
return parts
else:
return parser