def add_arguments(parser: KGTKArgumentParser):
"""
Parse arguments
Args:
parser (kgtk.cli_argparse.KGTKArgumentParser)
"""
parser.add_argument(action="store", type=str, metavar="name", dest="name")
parser.add_argument("-i", "--info", action="store", type=str, dest="info")
parser.add_argument("-e",
"--error",
action="store_true",
help="raise an error")
parser.accept_shared_argument('_debug')
def add_arguments_extended(parser: KGTKArgumentParser, parsed_shared_args):
parser.accept_shared_argument('_debug')
parser.accept_shared_argument('_expert')
parser.add_input_file(options=["-i", "--input-files"], dest='input_files', allow_list=True, default_stdin=False, allow_stdin=True,
who="One or more input files to query (maybe compressed).")
parser.add_argument('--as', metavar='NAME', default={}, action=InputOptionAction, dest='alias',
help="alias name to be used for preceding input")
# future extension:
#parser.add_argument('--in-memory', default=False, type=bool, nargs=0, action=InputOptionAction, dest='in_memory',
# help="load the preceding input into a temporary in-memory table only")
parser.add_argument('--query', default=None, action='store', dest='query',
help="complete Kypher query combining all clauses," +
" if supplied, all other specialized clause arguments will be ignored")
parser.add_argument('--match', metavar='PATTERN', default='()', action='store', dest='match',
help="MATCH pattern of a Kypher query, defaults to universal node pattern `()'")
parser.add_argument('--where', metavar='CLAUSE', default=None, action='store', dest='where',
help="WHERE clause of a Kypher query")
parser.add_argument('--return', metavar='CLAUSE', default='*', action='store', dest='return_',
help="RETURN clause of a Kypher query (defaults to *)")
parser.add_argument('--order-by', metavar='CLAUSE', default=None, action='store', dest='order',
help="ORDER BY clause of a Kypher query")
parser.add_argument('--skip', metavar='CLAUSE', default=None, action='store', dest='skip',
help="SKIP clause of a Kypher query")
parser.add_argument('--limit', metavar='CLAUSE', default=None, action='store', dest='limit',
help="LIMIT clause of a Kypher query")
parser.add_argument('--para', metavar='NAME=VAL', action='append', dest='regular_paras',
help="zero or more named value parameters to be passed to the query")
parser.add_argument('--spara', metavar='NAME=VAL', action='append', dest='string_paras',
help="zero or more named string parameters to be passed to the query")
parser.add_argument('--lqpara', metavar='NAME=VAL', action='append', dest='lqstring_paras',
help="zero or more named LQ-string parameters to be passed to the query")
parser.add_argument('--no-header', action='store_true', dest='no_header',
help="do not generate a header row with column names")
parser.add_argument('--index', metavar='MODE', nargs='?', action='store', dest='index',
choices=INDEX_MODES, const=INDEX_MODES[0], default=INDEX_MODES[0],
help="control column index creation according to MODE"
+ " (%(choices)s, default: %(const)s)")
parser.add_argument('--explain', metavar='MODE', nargs='?', action='store', dest='explain',
choices=EXPLAIN_MODES, const=EXPLAIN_MODES[0],
help="explain the query execution and indexing plan according to MODE"
+ " (%(choices)s, default: %(const)s)."
+ " This will not actually run or create anything.")
parser.add_argument('--graph-cache', default=DEFAULT_GRAPH_CACHE_FILE, action='store', dest='graph_cache_file',
help="database cache where graphs will be imported before they are queried"
+ " (defaults to per-user temporary file)")
parser.add_argument('-o', '--out', default='-', action='store', dest='output',
help="output file to write to, if `-' (the default) output goes to stdout."
+ " Files with extensions .gz, .bz2 or .xz will be appropriately compressed.")
def add_arguments_extended(parser: KGTKArgumentParser, parsed_shared_args):
parser.accept_shared_argument('_debug')
parser.accept_shared_argument('_expert')
parser.add_input_file(options=["-i", "--input-files"], dest='input_files',
# default_stdin is what makes it not required which is what we need for --show-cache:
allow_list=True, default_stdin=True, allow_stdin=True,
who="One or more input files to query, maybe compressed")
parser.add_argument('--as', metavar='NAME', default={}, action=InputOptionAction, dest='alias',
help="alias name to be used for preceding input")
parser.add_argument('--comment', default=None, action=InputOptionAction, dest='comment',
help="comment string to store for the preceding input (displayed by --show-cache)")
# future extension:
#parser.add_argument('--in-memory', default=False, type=bool, nargs=0, action=InputOptionAction, dest='in_memory',
# help="load the preceding input into a temporary in-memory table only")
parser.add_argument('--query', default=None, action='store', dest='query',
help="complete Kypher query combining all clauses," +
" if supplied, all other specialized clause arguments will be ignored")
parser.add_argument('--match', metavar='PATTERN', default='()', action=MatchOptionAction, dest='match',
help="MATCH pattern of a Kypher query, defaults to universal node pattern `()'")
parser.add_argument('--where', metavar='CLAUSE', default=None, action=MatchOptionAction, dest='where',
help="WHERE clause to a preceding --match, --opt or --with clause")
parser.add_argument('--opt', '--optional', metavar='PATTERN', default=None, action=MatchOptionAction,
help="OPTIONAL MATCH pattern(s) of a Kypher query (zero or more)")
parser.add_argument('--with', metavar='CLAUSE', default='*', action=MatchOptionAction, dest='with',
help="WITH clause of a Kypher query (only 'WITH * ...' is currently supported)")
parser.add_argument('--where:', metavar='CLAUSE', default=None, action='store', dest='with_where',
help="final global WHERE clause, shorthand for 'WITH * WHERE ...'")
parser.add_argument('--return', metavar='CLAUSE', default='*', action='store', dest='return',
help="RETURN clause of a Kypher query (defaults to *)")
parser.add_argument('--order-by', metavar='CLAUSE', default=None, action='store', dest='order',
help="ORDER BY clause of a Kypher query")
parser.add_argument('--skip', metavar='CLAUSE', default=None, action='store', dest='skip',
help="SKIP clause of a Kypher query")
parser.add_argument('--limit', metavar='CLAUSE', default=None, action='store', dest='limit',
help="LIMIT clause of a Kypher query")
parser.add_argument('--para', metavar='NAME=VAL', action='append', dest='regular_paras',
help="zero or more named value parameters to be passed to the query")
parser.add_argument('--spara', metavar='NAME=VAL', action='append', dest='string_paras',
help="zero or more named string parameters to be passed to the query")
parser.add_argument('--lqpara', metavar='NAME=VAL', action='append', dest='lqstring_paras',
help="zero or more named LQ-string parameters to be passed to the query")
parser.add_argument('--no-header', action='store_true', dest='no_header',
help="do not generate a header row with column names")
parser.add_argument('--force', action='store_true', dest='force',
help="force problematic queries to run against advice")
parser.add_argument('--index', '--index-mode', metavar='MODE', nargs='+', action='store',
dest='index_mode', default=[INDEX_MODES[0]],
help="default index creation MODE for all inputs"
+ f" (default: {INDEX_MODES[0]});"
+ " can be overridden with --idx for specific inputs")
parser.add_argument('--idx', '--input-index', metavar='SPEC', nargs='+', default=None,
action=InputOptionAction, dest='index_specs',
help="create index(es) according to SPEC for the preceding input only")
parser.add_argument('--explain', metavar='MODE', nargs='?', action='store', dest='explain',
choices=EXPLAIN_MODES, const=EXPLAIN_MODES[0],
help="explain the query execution and indexing plan according to MODE"
+ " (%(choices)s, default: %(const)s)."
+ " This will not actually run or create anything.")
parser.add_argument('--graph-cache', '--gc', action='store', dest='graph_cache_file',
help="database cache where graphs will be imported before they are queried"
+ " (defaults to per-user temporary file)")
parser.add_argument('--show-cache', '--sc', action='store_true', dest='show_cache',
help="describe the current content of the graph cache and exit"
+ " (does not actually run a query or import data)")
parser.add_argument('--read-only', '--ro', action='store_true', dest='readonly',
help="do not create or update the graph cache in any way"
+ ", only run queries against already imported and indexed data")
parser.add_argument('--import', metavar='MODULE_LIST', default=None, action='store', dest='import',
help="Python modules needed to define user extensions to built-in functions")
parser.add_argument('-o', '--out', default='-', action='store', dest='output',
help="output file to write to, if `-' (the default) output goes to stdout."
+ " Files with extensions .gz, .bz2 or .xz will be appropriately compressed.")
def add_arguments(parser: KGTKArgumentParser):
from kgtk.utils.argparsehelpers import optional_bool
parser.accept_shared_argument('_debug')
# input file
# parser.add_argument('input_file', nargs='?', type=argparse.FileType('r'), default=sys.stdin)
parser.add_input_file(positional=True)
# model name
all_models_names = ALL_EMBEDDING_MODELS_NAMES
parser.add_argument('-m',
'--model',
action='store',
nargs='+',
dest='all_models_names',
default="bert-base-nli-cls-token",
choices=all_models_names,
help="the model to used for embedding")
# parser.add_argument('-i', '--input', action='store', nargs='+', dest='input_uris',
# help="input path", )
parser.add_argument(
'-f',
'--input-data-format',
action='store',
dest='data_format',
choices=("test_format", "kgtk_format"),
default="kgtk_format",
help=
"the input file format, could either be `test_format` or `kgtk_format`, default is `kgtk_format`",
)
parser.add_argument(
'-p',
'--property-labels-file',
action='store',
nargs='+',
dest='property_labels_file_uri',
help="the path to the property labels file.",
)
# This should probably default to "--label-properties" if not specified.
parser.add_argument(
'--property-labels-filter',
action='store',
nargs='+',
dest='property_labels_filter',
default=["label"],
help=
"The label columns value(s) of the edges to process in the property labels file. Default is [\"label\"]."
)
# properties (only valid for kgtk format input/output data)
parser.add_argument(
'--label-properties',
action='store',
nargs='+',
dest='label_properties',
default=["label"],
help=
"""The names of the edges for label properties, Default is ["label"]. \n
This argument is only valid for input in kgtk format."""
)
parser.add_argument(
'--description-properties',
action='store',
nargs='+',
dest='description_properties',
default=["description"],
help=
"""The names of the edges for description properties, Default is ["description"].\n
This argument is only valid for input in kgtk format."""
)
parser.add_argument(
'--isa-properties',
action='store',
nargs='+',
dest='isa_properties',
default=["P31"],
help=
"""The names of the edges for `isa` properties, Default is ["P31"] (the `instance of` node in
wikidata).""")
parser.add_argument(
'--has-properties',
action='store',
nargs='+',
dest='has_properties',
default=[],
help=
"""The names of the edges for `has` properties, Default is ["all"] (will automatically append all
properties found for each node).""")
parser.add_argument(
'--property-value',
action='store',
nargs='+',
dest='property_values',
default=[],
help=
"""For those edges found in `has` properties, the nodes specified here will display with
corresponding edge(property) values. instead of edge name. """
)
parser.add_argument(
'--property-value-file',
action='store',
dest='property_values_file',
help=
"""Read the properties for --property-value option from an KGTK edge file"""
)
parser.add_argument(
'--output-property',
action='store',
dest='output_properties',
default="text_embedding",
help=
"""The output property name used to record the embedding. Default is `output_properties`. \n
This argument is only valid for output in kgtk format."""
)
# output
parser.add_argument(
'--save-embedding-sentence',
action='store_true',
dest='save_embedding_sentence',
help="if set, will also save the embedding sentences to output.")
parser.add_argument(
'-o',
'--embedding-projector-metadata-path',
action='store',
dest='output_uri',
default="",
help=
"output path for the metadata file, default will be current user's home directory"
)
parser.add_argument(
'--output-data-format',
action='store',
dest='output_data_format',
default="kgtk_format",
choices=("tsv_format", "kgtk_format"),
help=
"output format, can either be `tsv_format` or `kgtk_format`. \nIf choose `tsv_format`, the output "
"will be a tsv file, with each row contains only the vector representation of a node. Each "
"dimension is separated by a tab")
parser.add_argument(
'--embedding-projector-metadata',
action='store',
nargs='+',
dest='metadata_properties',
default=[],
help=
"""list of properties used to construct a metadata file for use in the Google Embedding Projector:
http://projector.tensorflow.org. \n Default: the label and description of each node."""
)
# black list file
parser.add_argument(
'-b',
'--black-list',
nargs='+',
action='store',
dest='black_list_files',
default=[],
help=
"the black list file, contains the Q nodes which should not consider as candidates."
)
# dimensional reduction relate
parser.add_argument(
"--dimensional-reduction",
nargs='?',
action='store',
default="none",
dest="dimensional_reduction",
choices=("pca", "tsne", "none"),
help=
'whether to run dimensional reduction algorithm or not after the embedding, default is None (not '
'run). ')
parser.add_argument(
"--dimension",
type=int,
nargs='?',
action='store',
default=2,
dest="dimension_val",
help=
'How many dimension should remained after reductions, only valid when set to run dimensional '
'reduction, default value is 2 ')
parser.add_argument(
"--parallel",
nargs='?',
action='store',
default="1",
dest="parallel_count",
help="How many processes to be run in same time, default is 1.")
# cache config
parser.add_argument(
"--use-cache",
type=optional_bool,
nargs='?',
action='store',
default=False,
dest="use_cache",
help=
"whether to use cache to get some embedding vectors quicker, default is False"
)
parser.add_argument(
"--cache-host",
nargs='?',
action='store',
default="dsbox01.isi.edu",
dest="cache_host",
help="cache host address, default is `dsbox01.isi.edu`")
parser.add_argument("--cache-port",
nargs='?',
action='store',
default="6379",
dest="cache_port",
help="cache server port, default is `6379`")
# query server
parser.add_argument(
"--query-server",
nargs='?',
action='store',
default="",
dest="query_server",
help=
"sparql query endpoint used for test_format input files, default is "
"https://query.wikidata.org/sparql ")
KgtkReader.add_debug_arguments(parser, expert=False)
KgtkReaderOptions.add_arguments(parser, mode_options=True, expert=False)
KgtkValueOptions.add_arguments(parser, expert=False)
