def main():
"""
Test the KGTK compact processor.
TODO: Support the list output file.
"""
parser: ArgumentParser = ArgumentParser()
parser.add_argument(
dest="input_file_path",
help="The KGTK file with the input data (default=%(default)s)",
type=Path,
nargs="?",
default="-")
parser.add_argument(
"--columns",
dest="key_column_names",
help="The key columns to identify records for compaction. " +
"(default=id for node files, (node1, label, node2, id) for edge files).",
nargs='+',
default=[])
parser.add_argument(
"--keep-first",
dest="keep_first_names",
help=
"If compaction results in a list of values for any column on this list, keep only the first value after sorting. "
+ "(default=none).",
nargs='+',
default=[])
parser.add_argument(
"--compact-id",
dest="compact_id",
help=
"Indicate that the ID column in KGTK edge files should be compacted. "
+ "Normally, if the ID column exists, it is not compacted, " +
"as there are use cases that need to maintain distinct lists of secondary edges for each ID value. (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=False)
parser.add_argument(
"--deduplicate",
dest="deduplicate",
help=
"Treat all columns as key columns, overriding --columns and --compact-id. "
+
"This will remove completely duplicate records without compacting any new lists. "
+ "(default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=False,
metavar="True|False")
parser.add_argument(
"--presorted",
dest="sorted_input",
help=
"Indicate that the input has been presorted (or at least pregrouped). (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=False)
parser.add_argument(
"--verify-sort",
dest="verify_sort",
help=
"If the input has been presorted, verify its consistency (disable if only pregrouped). (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=True)
parser.add_argument(
"--lists-in-input",
dest="lists_in_input",
help=
"Assume that the input file may contain lists (disable when certain it does not). (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=True)
parser.add_argument(
"--report-lists",
dest="report_lists",
help=
"When True, report records with lists to the error output. (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=False)
parser.add_argument(
"--exclude-lists",
dest="exclude_lists",
help=
"When True, exclude records with lists from the output. (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=False)
parser.add_argument(
"--output-only-lists",
dest="output_only_lists",
help=
"When True, output only records containing lists. (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=False)
parser.add_argument("-o",
"--output-file",
dest="output_file_path",
help="The KGTK file to write (default=%(default)s).",
type=Path,
default="-")
parser.add_argument(
"--build-id",
dest="build_id",
help="Build id values in an id column. (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=False)
KgtkIdBuilderOptions.add_arguments(parser)
KgtkReader.add_debug_arguments(parser)
KgtkReaderOptions.add_arguments(parser, mode_options=True)
KgtkValueOptions.add_arguments(parser)
args: Namespace = parser.parse_args()
error_file: typing.TextIO = sys.stdout if args.errors_to_stdout else sys.stderr
# Build the option structures.
idbuilder_options: KgtkIdBuilderOptions = KgtkIdBuilderOptions.from_args(
args)
reader_options: KgtkReaderOptions = KgtkReaderOptions.from_args(args)
value_options: KgtkValueOptions = KgtkValueOptions.from_args(args)
# Show the final option structures for debugging and documentation.
if args.show_options:
print("input: %s" % str(args.input_file_path),
file=error_file,
flush=True)
print("--output-file=%s" % str(args.output_file_path),
file=error_file,
flush=True)
print("--columns %s" % " ".join(args.key_column_names),
file=error_file,
flush=True)
print("--keep-first %s" % " ".join(args.keep_first_names),
file=error_file,
flush=True)
print("--compact-id=%s" % str(args.compact_id),
file=error_file,
flush=True)
print("--deduplicate=%s" % str(args.deduplicate),
file=error_file,
flush=True)
print("--presorted=%s" % str(args.sorted_input),
file=error_file,
flush=True)
print("--verify-sort=%s" % str(args.verify_sort),
file=error_file,
flush=True)
print("--lists-in-input=%s" % str(args.lists_in_input),
file=error_file,
flush=True)
print("--report-lists=%s" % str(args.report_lists),
file=error_file,
flush=True)
print("--exclude-lists=%s" % str(args.exclude_lists),
file=error_file,
flush=True)
print("--output-only-lists=%s" % str(args.output_only_lists),
file=error_file,
flush=True)
print("--build-id=%s" % str(args.build_id),
file=error_file,
flush=True)
idbuilder_options.show(out=error_file)
reader_options.show(out=error_file)
value_options.show(out=error_file)
kc: KgtkCompact = KgtkCompact(input_file_path=args.input_file_path,
key_column_names=args.key_column_names,
keep_first_names=args.keep_first_names,
compact_id=args.compact_id,
deduplicate=args.deduplicate,
sorted_input=args.sorted_input,
verify_sort=args.verify_sort,
lists_in_input=args.lists_in_input,
report_lists=args.report_lists,
exclude_lists=args.exclude_lists,
output_only_lists=args.output_only_lists,
output_file_path=args.output_file_path,
build_id=args.build_id,
idbuilder_options=idbuilder_options,
reader_options=reader_options,
value_options=value_options,
error_file=error_file,
verbose=args.verbose,
very_verbose=args.very_verbose)
kc.process()
def main():
"""
Test the KGTK implode processor.
"""
parser: ArgumentParser = ArgumentParser()
parser.add_argument(
dest="input_file_path",
help="The KGTK file with the input data. (default=%(default)s)",
type=Path,
nargs="?",
default="-")
parser.add_argument(
"--column",
dest="column_name",
help="The name of the column to explode. (default=%(default)s).",
default="node2")
parser.add_argument(
"--types",
dest="type_names",
nargs='*',
help=
"The KGTK data types for which fields should be imploded. (default=%(default)s).",
choices=KgtkFormat.DataType.choices(),
default=KgtkFormat.DataType.choices())
parser.add_argument(
"--without",
dest="without_fields",
nargs='*',
help="The KGTK fields to do without. (default=%(default)s).",
choices=KgtkValueFields.OPTIONAL_DEFAULT_FIELD_NAMES,
default=None)
parser.add_argument("-o",
"--output-file",
dest="output_file_path",
help="The KGTK file to write (default=%(default)s).",
type=Path,
default="-")
parser.add_argument(
"--prefix",
dest="prefix",
help="The prefix for exploded column names. (default=%(default)s).",
default="node2;kgtk:")
parser.add_argument(
"--overwrite",
dest="overwrite_column",
help=
"Indicate that it is OK to overwrite an existing imploded column. (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=True)
parser.add_argument(
"--validate",
dest="validate",
help="Validate imploded values. (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=True)
parser.add_argument(
"--escape-pipes",
dest="escape_pipes",
help=
"When true, pipe characters (|) need to be escaped (\\|) per KGTK file format. (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=False)
parser.add_argument(
"--quantities-include-numbers",
dest="quantities_include_numbers",
help=
"When true, numbers are acceptable quantities. (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=True)
parser.add_argument(
"--general-strings",
dest="general_strings",
help=
"When true, strings may include language qualified strings. (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=True)
parser.add_argument(
"--remove-prefixed-columns",
dest="remove_prefixed_columns",
help=
"When true, remove all columns beginning with the prefix from the output file. (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=False)
parser.add_argument(
"--ignore-unselected-types",
dest="ignore_unselected_types",
help=
"When true, input records with valid but unselected data types will be passed through to output. (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=True)
parser.add_argument(
"--retain-unselected-types",
dest="retain_unselected_types",
help=
"When true, input records with valid but unselected data types will be retain existing data on output. (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=True)
parser.add_argument(
"--build-id",
dest="build_id",
help="Build id values in an id column. (default=%(default)s).",
type=optional_bool,
nargs='?',
const=True,
default=False)
parser.add_argument(
"--reject-file",
dest="reject_file_path",
help=
"The KGTK file into which to write rejected records (default=%(default)s).",
type=Path,
default=None)
KgtkIdBuilderOptions.add_arguments(parser)
KgtkReader.add_debug_arguments(parser)
KgtkReaderOptions.add_arguments(parser, mode_options=True)
KgtkValueOptions.add_arguments(parser)
args: Namespace = parser.parse_args()
error_file: typing.TextIO = sys.stdout if args.errors_to_stdout else sys.stderr
# Build the option structures.
idbuilder_options: KgtkIdBuilderOptions = KgtkIdBuilderOptions.from_args(
args)
reader_options: KgtkReaderOptions = KgtkReaderOptions.from_args(args)
value_options: KgtkValueOptions = KgtkValueOptions.from_args(args)
# Show the final option structures for debugging and documentation.
if args.show_options:
# TODO: show ifempty-specific options.
print("input: %s" % str(args.input_file_path),
file=error_file,
flush=True)
print("--column %s" % args.column_name, file=error_file, flush=True)
print("--prefix %s" % args.prefix, file=error_file, flush=True)
print("--overwrite %s" % str(args.overwrite_column),
file=error_file,
flush=True)
print("--validate %s" % str(args.validate),
file=error_file,
flush=True)
print("--escape-pipes %s" % str(args.escape_pipes),
file=error_file,
flush=True)
print("--quantities-include-numbers %s" %
str(args.quantities_include_numbers),
file=error_file,
flush=True)
print("--general-strings %s" % str(args.general_strings),
file=error_file,
flush=True)
print("--remove-prefixed-columns %s" %
str(args.remove_prefixed_columns),
file=error_file,
flush=True)
print("--ignore-unselected-types %s" %
str(args.ignore_unselected_types),
file=error_file,
flush=True)
print("--retain-unselected-types %s" %
str(args.retain_unselected_types),
file=error_file,
flush=True)
if args.type_names is not None:
print("--types %s" % " ".join(args.type_names),
file=error_file,
flush=True)
if args.without_fields is not None:
print("--without %s" % " ".join(args.without_fields),
file=error_file,
flush=True)
print("--output-file=%s" % str(args.output_file_path),
file=error_file,
flush=True)
if args.reject_file_path is not None:
print("--reject-file=%s" % str(args.reject_file_path),
file=error_file,
flush=True)
print("--build-id=%s" % str(args.build_id),
file=error_file,
flush=True)
idbuilder_options.show(out=error_file)
reader_options.show(out=error_file)
value_options.show(out=error_file)
without_fields: typing.List[
str] = args.without_fields if args.without_fields is not None else list(
)
ex: KgtkImplode = KgtkImplode(
input_file_path=args.input_file_path,
column_name=args.column_name,
prefix=args.prefix,
type_names=args.type_names,
without_fields=without_fields,
overwrite_column=args.overwrite_column,
validate=args.validate,
escape_pipes=args.escape_pipes,
quantities_include_numbers=args.quantities_include_numbers,
general_strings=args.general_strings,
remove_prefixed_columns=args.remove_prefixed_columns,
ignore_unselected_types=args.ignore_unselected_types,
retain_unselected_types=args.retain_unselected_types,
output_file_path=args.output_file_path,
reject_file_path=args.reject_file_path,
build_id=args.build_id,
idbuilder_options=idbuilder_options,
reader_options=reader_options,
value_options=value_options,
error_file=error_file,
verbose=args.verbose,
very_verbose=args.very_verbose)
ex.process()
def main():
"""
Test the KGTK ntriples importer.
"""
parser: ArgumentParser = ArgumentParser()
parser.add_argument(
"-i",
"--input-files",
dest="input_file_paths",
nargs='*',
help="The file(s) with the input ntriples data. (default=%(default)s)",
type=Path,
default="-")
parser.add_argument("-o",
"--output-file",
dest="output_file_path",
help="The KGTK file to write (default=%(default)s).",
type=Path,
default="-")
parser.add_argument(
"--reject-file",
dest="reject_file_path",
help=
"The KGTK file into which to write rejected records. (default=%(default)s).",
type=Path,
default=None)
parser.add_argument(
"--namespace-file",
dest="namespace_file_path",
help="The KGTK file with known namespaces. (default=%(default)s).",
type=Path,
default=None)
parser.add_argument(
"--updated-namespace-file",
dest="updated_namespace_file_path",
help=
"An updated KGTK file with known namespaces. (default=%(default)s).",
type=Path,
default=None)
KgtkNtriples.add_arguments(parser)
KgtkIdBuilderOptions.add_arguments(parser)
KgtkReader.add_debug_arguments(parser)
KgtkReaderOptions.add_arguments(parser, mode_options=True, expert=True)
KgtkValueOptions.add_arguments(parser)
args: Namespace = parser.parse_args()
error_file: typing.TextIO = sys.stdout if args.errors_to_stdout else sys.stderr
# Build the option structures.
idbuilder_options: KgtkIdBuilderOptions = KgtkIdBuilderOptions.from_args(
args)
reader_options: KgtkReaderOptions = KgtkReaderOptions.from_args(args)
value_options: KgtkValueOptions = KgtkValueOptions.from_args(args)
# Show the final option structures for debugging and documentation.
if args.show_options:
print("--input-files %s" %
" ".join([str(path) for path in input_file_paths]),
file=error_file,
flush=True)
print("--output-file=%s" % str(args.output_file_path),
file=error_file,
flush=True)
# TODO: show ifempty-specific options.
if args.reject_file_path is not None:
print("--reject-file=%s" % str(args.reject_file_path),
file=error_file,
flush=True)
if args.namespace_file_path is not None:
print("--namespace-file=%s" % str(args.namespace_file_path),
file=error_file,
flush=True)
if args.updated_namespace_file_path is not None:
print("--updated-namespace-file=%s" %
str(args.updated_namespace_file_path),
file=error_file,
flush=True)
print("--namespace-id-prefix %s" % args.namespace_id_prefix,
file=error_file,
flush=True)
print("--namespace-id-use-uuid %s" % str(args.namespace_id_use_uuid),
file=error_file,
flush=True)
print("--namespace-id-counter %s" % str(args.namespace_id_counter),
file=error_file,
flush=True)
print("--namespace-id-zfill %s" % str(args.namespace_id_zfill),
file=error_file,
flush=True)
print("--output-only-used-namespaces %s" %
str(args.output_only_used_namespaces),
file=error_file,
flush=True)
print("--allow-lax-uri %s" % str(args.allow_lax_uri),
file=error_file,
flush=True)
print("--local-namespace-prefix %s" % args.local_namespace_prefix,
file=error_file,
flush=True)
print("--local-namespace-use-uuid %s" %
str(args.local_namespace_use_uuid),
file=error_file,
flush=True)
print("--prefix-expansion-label %s" % args.prefix_expansion_label,
file=error_file,
flush=True)
print("--structured-value-label %s" % args.structured_value_label,
file=error_file,
flush=True)
print("--structured-uri-label %s" % args.structured_uri_label,
file=error_file,
flush=True)
print("--newnode-prefix %s" % args.newnode_prefix,
file=error_file,
flush=True)
print("--newnode-use-uuid %s" % str(args.newnode_use_uuid),
file=error_file,
flush=True)
print("--newnode-counter %s" % str(args.newnode_counter),
file=error_file,
flush=True)
print("--newnode-zfill %s" % str(args.newnode_zfill),
file=error_file,
flush=True)
print("--build-id=%s" % str(args.build_id),
file=error_file,
flush=True)
print("--escape-pipes=%s" % str(args.escape_pipes),
file=error_file,
flush=True)
print("--validate=%s" % str(args.validate),
file=error_file,
flush=True)
if args.override_uuid is not None:
print("--override_uuid=%s" % str(args.override_uuid),
file=error_file,
flush=True)
idbuilder_options.show(out=error_file)
reader_options.show(out=error_file)
value_options.show(out=error_file)
kn: KgtkNtriples = KgtkNtriples(
input_file_paths=args.input_file_paths,
output_file_path=args.output_file_path,
reject_file_path=args.reject_file_path,
namespace_file_path=args.namespace_file_path,
updated_namespace_file_path=args.updated_namespace_file_path,
namespace_id_prefix=args.namespace_id_prefix,
namespace_id_use_uuid=args.namespace_id_use_uuid,
namespace_id_counter=args.namespace_id_counter,
namespace_id_zfill=args.namespace_id_zfill,
output_only_used_namespaces=args.output_only_used_namespaces,
newnode_prefix=args.newnode_prefix,
newnode_use_uuid=args.newnode_use_uuid,
newnode_counter=args.newnode_counter,
newnode_zfill=args.newnode_zfill,
allow_lax_uri=args.allow_lax_uri,
local_namespace_prefix=args.local_namespace_prefix,
local_namespace_use_uuid=args.local_namespace_use_uuid,
prefix_expansion_label=args.prefix_expansion_label,
structured_value_label=args.structured_value_label,
structured_uri_label=args.structured_uri_label,
build_id=args.build_id,
escape_pipes=args.escape_pipes,
idbuilder_options=idbuilder_options,
validate=args.validate,
override_uuid=args.override_uuid,
reader_options=reader_options,
value_options=value_options,
error_file=error_file,
verbose=args.verbose,
very_verbose=args.very_verbose)
kn.process()
def main():
"""
Test the KGTK compact processor.
"""
parser: ArgumentParser = ArgumentParser()
parser.add_argument(dest="input_file_path", help="The KGTK file with the input data (default=%(default)s)", type=Path, nargs="?", default="-")
parser.add_argument( "--columns", dest="key_column_names",
help="The key columns to identify records for compaction. " +
"(default=id for node files, (node1, label, node2, id) for edge files).", nargs='+', default=[ ])
parser.add_argument( "--compact-id", dest="compact_id",
help="Indicate that the ID column in KGTK edge files should be compacted. " +
"Normally, if the ID column exists, it is not compacted, " +
"as there are use cases that need to maintain distinct lists of secondary edges for each ID value. (default=%(default)s).",
type=optional_bool, nargs='?', const=True, default=False)
parser.add_argument( "--presorted", dest="sorted_input",
help="Indicate that the input has been presorted (or at least pregrouped). (default=%(default)s).",
type=optional_bool, nargs='?', const=True, default=False)
parser.add_argument( "--verify-sort", dest="verify_sort",
help="If the input has been presorted, verify its consistency (disable if only pregrouped). (default=%(default)s).",
type=optional_bool, nargs='?', const=True, default=True)
parser.add_argument("-o", "--output-file", dest="output_file_path", help="The KGTK file to write (default=%(default)s).", type=Path, default="-")
parser.add_argument( "--build-id", dest="build_id",
help="Build id values in an id column. (default=%(default)s).",
type=optional_bool, nargs='?', const=True, default=False)
KgtkIdBuilderOptions.add_arguments(parser)
KgtkReader.add_debug_arguments(parser)
KgtkReaderOptions.add_arguments(parser, mode_options=True)
KgtkValueOptions.add_arguments(parser)
args: Namespace = parser.parse_args()
error_file: typing.TextIO = sys.stdout if args.errors_to_stdout else sys.stderr
# Build the option structures.
idbuilder_options: KgtkIdBuilderOptions = KgtkIdBuilderOptions.from_args(args)
reader_options: KgtkReaderOptions = KgtkReaderOptions.from_args(args)
value_options: KgtkValueOptions = KgtkValueOptions.from_args(args)
# Show the final option structures for debugging and documentation.
if args.show_options:
print("input: %s" % str(args.input_file_path), file=error_file, flush=True)
print("--columns %s" % " ".join(args.key_column_names), file=error_file, flush=True)
print("--compact-id=%s" % str(args.compact_id), file=error_file, flush=True)
print("--presorted=%s" % str(args.sorted_input), file=error_file, flush=True)
print("--verify-sort=%s" % str(args.verify_sort), file=error_file, flush=True)
print("--output-file=%s" % str(args.output_file_path), file=error_file, flush=True)
print("--build-id=%s" % str(args.build_id), file=error_file, flush=True)
idbuilder_options.show(out=error_file)
reader_options.show(out=error_file)
value_options.show(out=error_file)
kc: KgtkCompact = KgtkCompact(
input_file_path=args.input_file_path,
key_column_names=args.key_column_names,
compact_id=args.compact_id,
sorted_input=args.sorted_input,
verify_sort=args.verify_sort,
output_file_path=args.output_file_path,
build_id=args.build_id,
idbuilder_options=idbuilder_options,
reader_options=reader_options,
value_options=value_options,
error_file=error_file,
verbose=args.verbose,
very_verbose=args.very_verbose)
kc.process()