def add_engine_args(my_parser, default_packages=None):
add_spark_args(my_parser, default_packages=default_packages)
my_parser.add_argument("--bblfsh",
default=EngineDefault.BBLFSH,
help="Babelfish server's address.")
my_parser.add_argument("--engine",
default=EngineDefault.VERSION,
help="source{d} engine version.")
my_parser.add_argument("--repository-format",
default=EngineDefault.REPOSITORY_FORMAT,
help="Repository storage input format.")
def get_parser() -> argparse.ArgumentParser:
"""
Create main parser.
:return: Parser
"""
parser = argparse.ArgumentParser()
parser.add_argument("--log-level",
default="INFO",
choices=logging._nameToLevel,
help="Logging verbosity.")
def add_feature_weight_arg(my_parser):
help_desc = "%s's weight - all features from this extractor will be multiplied by this " \
"factor"
for ex in extractors.__extractors__.values():
my_parser.add_argument("--%s-weight" % ex.NAME,
default=1,
type=float,
help=help_desc % ex.__name__)
def add_cassandra_args(my_parser):
my_parser.add_argument("--cassandra",
default="0.0.0.0:9042",
help="Cassandra's host:port.")
my_parser.add_argument("--keyspace",
default="apollo",
help="Cassandra's key space.")
my_parser.add_argument(
"--tables",
help=
"Table name mapping (JSON): bags, hashes, hashtables, hashtables2."
)
def add_wmh_args(my_parser, params_help: str, add_hash_size: bool,
required: bool):
if add_hash_size:
my_parser.add_argument("--size",
type=int,
default=128,
help="Hash size.")
my_parser.add_argument("-p",
"--params",
required=required,
help=params_help)
my_parser.add_argument("-t",
"--threshold",
required=required,
type=float,
help="Jaccard similarity threshold.")
my_parser.add_argument(
"--false-positive-weight",
type=float,
default=0.5,
help="Used to adjust the relative importance of "
"minimizing false positives count when optimizing "
"for the Jaccard similarity threshold.")
my_parser.add_argument(
"--false-negative-weight",
type=float,
default=0.5,
help="Used to adjust the relative importance of "
"minimizing false negatives count when optimizing "
"for the Jaccard similarity threshold.")
def add_template_args(my_parser, default_template):
my_parser.add_argument("--batch",
type=int,
default=100,
help="Number of hashes to query at a time.")
my_parser.add_argument("--template",
default=default_template,
help="Jinja2 template to render.")
add_cassandra_args(my_parser)
def add_dzhigurda_arg(my_parser):
my_parser.add_argument(
"--dzhigurda",
default=0,
type=int,
help="Index of the examined commit in the history.")
subparsers = parser.add_subparsers(help="Commands", dest="command")
preprocessing_parser = subparsers.add_parser(
"preprocess",
help="Generate the dataset with good candidates for duplication.")
preprocessing_parser.set_defaults(handler=preprocess_source)
preprocessing_parser.add_argument(
"-r",
"--repositories",
required=True,
help="The path to the siva files with repositories.")
preprocessing_parser.add_argument(
"-o",
"--output",
required=True,
help="[OUT] The path to the Parquet files with bag batches.")
add_dzhigurda_arg(preprocessing_parser)
preprocessing_parser.add_argument(
"-l",
"--language",
choices=("Java", "Python"),
help="The programming language to analyse.")
preprocessing_parser.add_argument("--pause",
action="store_true",
help="Do not terminate in the end.")
default_fields = [
"blob_id", "repository_id", "content", "path", "commit_hash", "uast"
]
preprocessing_parser.add_argument("-f",
"--fields",
action="append",
default=default_fields,
help="Fields to select from DF to save.")
add_engine_args(preprocessing_parser)
source2bags_parser = subparsers.add_parser(
"bags", help="Convert source code to weighted sets.")
source2bags_parser.set_defaults(handler=source2bags)
add_bow_args(source2bags_parser)
add_dzhigurda_arg(source2bags_parser)
add_repo2_args(source2bags_parser)
add_feature_args(source2bags_parser)
add_cassandra_args(source2bags_parser)
add_engine_args(source2bags_parser, default_packages=[CASSANDRA_PACKAGE])
warmup_parser = subparsers.add_parser("warmup",
help="Initialize source{d} engine.")
warmup_parser.set_defaults(handler=warmup)
add_engine_args(warmup_parser, default_packages=[CASSANDRA_PACKAGE])
hash_parser = subparsers.add_parser(
"hash", help="Run MinHashCUDA on the bag batches.")
hash_parser.set_defaults(handler=hash_batches)
hash_parser.add_argument("input",
help="Path to the directory with Parquet files.")
hash_parser.add_argument("--seed",
type=int,
default=int(time()),
help="Random generator's seed.")
hash_parser.add_argument("--mhc-verbosity",
type=int,
default=1,
help="MinHashCUDA logs verbosity level.")
hash_parser.add_argument(
"--devices",
type=int,
default=0,
help="Or-red indices of NVIDIA devices to use. 0 means all.")
hash_parser.add_argument(
"--docfreq",
default=None,
help="Path to OrderedDocumentFrequencies (file mode).")
add_wmh_args(hash_parser, "Path to the output file with WMH parameters.",
True, True)
add_cassandra_args(hash_parser)
add_spark_args(hash_parser, default_packages=[CASSANDRA_PACKAGE])
add_feature_weight_arg(hash_parser)
query_parser = subparsers.add_parser("query",
help="Query for similar files.")
query_parser.set_defaults(handler=query)
mode_group = query_parser.add_mutually_exclusive_group(required=True)
mode_group.add_argument("-i", "--id", help="Query for this id (id mode).")
mode_group.add_argument("-c",
"--file",
help="Query for this file (file mode).")
query_parser.add_argument(
"--docfreq", help="Path to OrderedDocumentFrequencies (file mode).")
query_parser.add_argument("--bblfsh",
default="localhost:9432",
help="Babelfish server's endpoint.")
add_feature_args(query_parser, required=False)
query_parser.add_argument("-x",
"--precise",
action="store_true",
help="Calculate the precise set.")
add_wmh_args(query_parser, "Path to the Weighted MinHash parameters.",
False, False)
add_template_args(query_parser, "query.md.jinja2")
db_parser = subparsers.add_parser(
"resetdb", help="Destructively initialize the database.")
db_parser.set_defaults(handler=reset_db)
add_cassandra_args(db_parser)
db_parser.add_argument(
"--hashes-only",
action="store_true",
help=
"Only clear the tables: hashes, hashtables, hashtables2. Do not touch the rest."
)
cc_parser = subparsers.add_parser(
"cc",
help=
"Load the similar pairs of files and run connected components analysis."
)
cc_parser.set_defaults(handler=find_connected_components)
add_cassandra_args(cc_parser)
cc_parser.add_argument(
"-o",
"--output",
help="Path to save the asdf file with connected components.")
dumpcc_parser = subparsers.add_parser(
"dumpcc", help="Output the connected components to stdout.")
dumpcc_parser.set_defaults(handler=dumpcc)
dumpcc_parser.add_argument("input", help="Path to the asdf file with CCs.")
community_parser = subparsers.add_parser(
"cmd",
help=
"Run Community Detection analysis on the connected components from \"cc\"."
)
community_parser.set_defaults(handler=detect_communities)
community_parser.add_argument(
"-i",
"--input",
required=True,
help="The path to connected components ASDF model.")
community_parser.add_argument(
"-o",
"--output",
required=True,
help="Output path to the communities ASDF model.")
community_parser.add_argument(
"--edges",
choices=("linear", "quadratic", "1", "2"),
default="linear",
help="The method to generate the graph's edges: bipartite - "
"linear and fast, but may not fit some the CD algorithms, "
"or all to all within a bucket - quadratic and slow, but "
"surely fits all the algorithms.")
cmd_choices = [k[10:] for k in dir(Graph) if k.startswith("community_")]
community_parser.add_argument(
"-a",
"--algorithm",
choices=cmd_choices,
default="walktrap",
help="The community detection algorithm to apply.")
community_parser.add_argument(
"-p",
"--params",
type=json.loads,
default={},
help="Parameters for the algorithm (**kwargs, JSON format).")
community_parser.add_argument("--no-spark",
action="store_true",
help="Do not use Spark.")
add_spark_args(community_parser)
dumpcmd_parser = subparsers.add_parser(
"dumpcmd", help="Output the detected communities to stdout.")
dumpcmd_parser.set_defaults(handler=dumpcmd)
dumpcmd_parser.add_argument("input",
help="Path to the asdf file with communities.")
add_template_args(dumpcmd_parser, "report.md.jinja2")
evalcc_parser = subparsers.add_parser(
"evalcc",
help=
"Evaluate the communities: calculate the precise similarity and the "
"fitness metric.")
evalcc_parser.set_defaults(handler=evaluate_communities)
evalcc_parser.add_argument("-t",
"--threshold",
required=True,
type=float,
help="Jaccard similarity threshold.")
evalcc_parser.add_argument("-i",
"--input",
required=True,
help="Path to the communities model.")
add_spark_args(evalcc_parser, default_packages=[CASSANDRA_PACKAGE])
add_cassandra_args(evalcc_parser)
# TODO: retable [.....] -> [.] [.] [.] [.] [.]
return parser
def get_parser() -> argparse.ArgumentParser:
"""
Creates the cmdline argument parser.
"""
parser = argparse.ArgumentParser(
formatter_class=args.ArgumentDefaultsHelpFormatterNoNone)
parser.add_argument("--log-level",
default="INFO",
choices=logging._nameToLevel,
help="Logging verbosity.")
# Create and construct subparsers
subparsers = parser.add_subparsers(help="Commands", dest="command")
def add_parser(name, help_message):
return subparsers.add_parser(
name,
help=help_message,
formatter_class=args.ArgumentDefaultsHelpFormatterNoNone)
# ------------------------------------------------------------------------
preprocessing_parser = subparsers.add_parser(
"preprocrepos",
help="Convert siva to parquet files with extracted information.")
preprocessing_parser.set_defaults(handler=cmd.preprocess_repos)
preprocessing_parser.add_argument(
"-x",
"--mode",
choices=Moder.Options.__all__,
default="file",
help="What to extract from repositories.")
args.add_repo2_args(preprocessing_parser)
args.add_dzhigurda_arg(preprocessing_parser)
preprocessing_parser.add_argument(
"-o",
"--output",
required=True,
help="[OUT] Path to the parquet files with bag batches.")
default_fields = ("blob_id", "repository_id", "content", "path",
"commit_hash", "uast")
preprocessing_parser.add_argument("-f",
"--fields",
nargs="+",
default=default_fields,
help="Fields to save.")
# ------------------------------------------------------------------------
repos2bow_parser = add_parser(
"repos2bow", "Convert source code to the bag-of-words model.")
repos2bow_parser.set_defaults(handler=cmd.repos2bow)
args.add_df_args(repos2bow_parser)
args.add_repo2_args(repos2bow_parser)
args.add_feature_args(repos2bow_parser)
args.add_bow_args(repos2bow_parser)
args.add_repartitioner_arg(repos2bow_parser)
# ------------------------------------------------------------------------
repos2df_parser = add_parser(
"repos2df",
"Calculate document frequencies of features extracted from source code."
)
repos2df_parser.set_defaults(handler=cmd.repos2df)
args.add_df_args(repos2df_parser)
args.add_repo2_args(repos2df_parser)
args.add_feature_args(repos2df_parser)
# ------------------------------------------------------------------------
repos2ids_parser = subparsers.add_parser(
"repos2ids", help="Convert source code to a bag of identifiers.")
repos2ids_parser.set_defaults(handler=cmd.repos2ids)
args.add_repo2_args(repos2ids_parser)
args.add_split_stem_arg(repos2ids_parser)
args.add_repartitioner_arg(repos2ids_parser)
repos2ids_parser.add_argument(
"-o",
"--output",
required=True,
help="[OUT] output path to the CSV file with identifiers.")
repos2ids_parser.add_argument(
"--idfreq",
action="store_true",
help="Adds identifier frequencies to the output CSV file."
"num_repos is the number of repositories where the identifier appears in."
"num_files is the number of files where the identifier appears in."
"num_occ is the total number of occurences of the identifier.")
# ------------------------------------------------------------------------
repos2coocc_parser = add_parser(
"repos2coocc",
"Convert source code to the sparse co-occurrence matrix of identifiers."
)
repos2coocc_parser.set_defaults(handler=cmd.repos2coocc)
args.add_df_args(repos2coocc_parser)
args.add_repo2_args(repos2coocc_parser)
args.add_split_stem_arg(repos2coocc_parser)
args.add_repartitioner_arg(repos2coocc_parser)
repos2coocc_parser.add_argument(
"-o",
"--output",
required=True,
help="[OUT] Path to the Cooccurrences model.")
# ------------------------------------------------------------------------
repos2roles_and_ids = add_parser(
"repos2roleids",
"Converts a UAST to a list of pairs, where pair is a role and "
"identifier. Role is merged generic roles where identifier was found.")
repos2roles_and_ids.set_defaults(handler=cmd.repos2roles_and_ids)
args.add_repo2_args(repos2roles_and_ids)
args.add_split_stem_arg(repos2roles_and_ids)
repos2roles_and_ids.add_argument(
"-o",
"--output",
required=True,
help="[OUT] Path to the directory where spark should store the result. "
"Inside the direcory you find result is csv format, status file and sumcheck files."
)
# ------------------------------------------------------------------------
repos2identifier_distance = add_parser(
"repos2id_distance", "Converts a UAST to a list of identifier pairs "
"and distance between them.")
repos2identifier_distance.set_defaults(handler=cmd.repos2id_distance)
args.add_repo2_args(repos2identifier_distance)
args.add_split_stem_arg(repos2identifier_distance)
repos2identifier_distance.add_argument(
"-t",
"--type",
required=True,
choices=IdentifierDistance.DistanceType.All,
help="Distance type.")
repos2identifier_distance.add_argument(
"--max-distance",
default=IdentifierDistance.DEFAULT_MAX_DISTANCE,
type=int,
help="Maximum distance to save.")
repos2identifier_distance.add_argument(
"-o",
"--output",
required=True,
help="[OUT] Path to the directory where spark should store the result. "
"Inside the direcory you find result is csv format, status file and sumcheck files."
)
# ------------------------------------------------------------------------
repos2id_sequence = add_parser(
"repos2idseq",
"Converts a UAST to sequence of identifiers sorted by order of appearance."
)
repos2id_sequence.set_defaults(handler=cmd.repos2id_sequence)
args.add_repo2_args(repos2id_sequence)
args.add_split_stem_arg(repos2id_sequence)
repos2id_sequence.add_argument(
"--skip-docname",
default=False,
action="store_true",
help="Do not save document name in CSV file, only identifier sequence."
)
repos2id_sequence.add_argument(
"-o",
"--output",
required=True,
help="[OUT] Path to the directory where spark should store the result. "
"Inside the direcory you find result is csv format, status file and sumcheck files."
)
# ------------------------------------------------------------------------
preproc_parser = add_parser(
"id2vec-preproc",
"Convert a sparse co-occurrence matrix to the Swivel shards.")
preproc_parser.set_defaults(handler=cmd.id2vec_preprocess)
args.add_df_args(preproc_parser)
preproc_parser.add_argument("-s",
"--shard-size",
default=4096,
type=int,
help="The shard (submatrix) size.")
preproc_parser.add_argument(
"-i", "--input", help="Concurrence model produced by repos2coocc.")
preproc_parser.add_argument("-o",
"--output",
required=True,
help="Output directory.")
# ------------------------------------------------------------------------
train_parser = add_parser("id2vec-train",
"Train identifier embeddings using Swivel.")
train_parser.set_defaults(handler=cmd.run_swivel)
mirror_tf_args(train_parser)
# ------------------------------------------------------------------------
id2vec_postproc_parser = add_parser(
"id2vec-postproc",
"Combine row and column embeddings produced by Swivel and write them to an .asdf."
)
id2vec_postproc_parser.set_defaults(handler=cmd.id2vec_postprocess)
id2vec_postproc_parser.add_argument(
"-i",
"--swivel-data",
required=True,
help="Folder with swivel row and column embeddings data. "
"You can get it using id2vec_train subcommand.")
id2vec_postproc_parser.add_argument(
"-o",
"--output",
required=True,
help="Output directory for Id2Vec model.")
# ------------------------------------------------------------------------
id2vec_project_parser = add_parser(
"id2vec-project", "Present id2vec model in Tensorflow Projector.")
id2vec_project_parser.set_defaults(handler=cmd.id2vec_project)
args.add_df_args(id2vec_project_parser, required=False)
id2vec_project_parser.add_argument("-i",
"--input",
required=True,
help="id2vec model to present.")
id2vec_project_parser.add_argument("-o",
"--output",
required=True,
help="Projector output directory.")
id2vec_project_parser.add_argument("--no-browser",
action="store_true",
help="Do not open the browser.")
# ------------------------------------------------------------------------
bow2vw_parser = add_parser(
"bow2vw",
"Convert a bag-of-words model to the dataset in Vowpal Wabbit format.")
bow2vw_parser.set_defaults(handler=cmd.bow2vw)
bow2vw_parser.add_argument("--bow",
help="URL or path to a bag-of-words model.")
bow2vw_parser.add_argument(
"--id2vec", help="URL or path to the identifier embeddings.")
bow2vw_parser.add_argument("-o",
"--output",
required=True,
help="Path to the output file.")
# ------------------------------------------------------------------------
bigartm_postproc_parser = add_parser(
"bigartm2asdf",
"Convert a human-readable BigARTM model to Modelforge format.")
bigartm_postproc_parser.set_defaults(handler=cmd.bigartm2asdf)
bigartm_postproc_parser.add_argument("input")
bigartm_postproc_parser.add_argument("output")
# ------------------------------------------------------------------------
bigartm_parser = add_parser(
"bigartm", "Install bigartm/bigartm to the current working directory.")
bigartm_parser.set_defaults(handler=install_bigartm)
bigartm_parser.add_argument(
"--tmpdir",
help="Store intermediate files in this directory instead of /tmp.")
bigartm_parser.add_argument("--output",
default=os.getcwd(),
help="Output directory.")
# ------------------------------------------------------------------------
dump_parser = add_parser("dump", "Dump a model to stdout.")
dump_parser.set_defaults(handler=cmd.dump_model)
dump_parser.add_argument("input",
help="Path to the model file, URL or UUID.")
dump_parser.add_argument("--gcs",
default=None,
dest="gcs_bucket",
help="GCS bucket to use.")
# ------------------------------------------------------------------------
merge_df = add_parser("merge-df",
"Merge DocumentFrequencies models to a singe one.")
merge_df.set_defaults(handler=cmd.merge_df)
args.add_filter_arg(merge_df)
args.add_min_docfreq(merge_df)
args.add_vocabulary_size_arg(merge_df)
merge_df.add_argument("-o",
"--output",
required=True,
help="Path to the merged DocumentFrequencies model.")
merge_df.add_argument("-i",
"--input",
required=True,
nargs="+",
help="DocumentFrequencies models input files."
"Use `-i -` to read input files from stdin.")
merge_df.add_argument(
"--ordered",
action="store_true",
default=False,
help="Save OrderedDocumentFrequencies. "
"If not specified DocumentFrequencies model will be saved")
# ------------------------------------------------------------------------
merge_coocc = add_parser("merge-coocc",
"Merge several Cooccurrences models together.")
merge_coocc.set_defaults(handler=cmd.merge_coocc)
add_spark_args(merge_coocc)
args.add_filter_arg(merge_coocc)
merge_coocc.add_argument("-o",
"--output",
required=True,
help="Path to the merged Cooccurrences model.")
merge_coocc.add_argument("-i",
"--input",
required=True,
help="Cooccurrences models input files."
"Use `-i -` to read input files from stdin.")
merge_coocc.add_argument(
"--docfreq",
required=True,
help="[IN] Specify OrderedDocumentFrequencies model. "
"Identifiers that are not present in the model will be ignored.")
merge_coocc.add_argument(
"--no-spark",
action="store_true",
default=False,
help="Use the local reduction instead of PySpark. "
"Can be faster and consume less memory if the data fits into RAM.")
return parser
def get_parser() -> argparse.ArgumentParser:
"""
Creates the cmdline argument parser.
"""
parser = argparse.ArgumentParser(formatter_class=args.ArgumentDefaultsHelpFormatterNoNone)
parser.add_argument("--log-level", default="INFO",
choices=logging._nameToLevel,
help="Logging verbosity.")
# Create and construct subparsers
subparsers = parser.add_subparsers(help="Commands", dest="command")
def add_parser(name, help_message):
return subparsers.add_parser(
name, help=help_message, formatter_class=args.ArgumentDefaultsHelpFormatterNoNone)
# ------------------------------------------------------------------------
preprocessing_parser = subparsers.add_parser(
"preprocrepos", help="Convert siva to parquet files with extracted information.")
preprocessing_parser.set_defaults(handler=cmd.preprocess_repos)
preprocessing_parser.add_argument("-x", "--mode", choices=Moder.Options.__all__,
default="file", help="What to extract from repositories.")
args.add_repo2_args(preprocessing_parser)
preprocessing_parser.add_argument(
"-o", "--output", required=True,
help="[OUT] Path to the parquet files with bag batches.")
default_fields = ("blob_id", "repository_id", "content", "path", "commit_hash", "uast", "lang")
preprocessing_parser.add_argument(
"-f", "--fields", nargs="+", default=default_fields,
help="Fields to save.")
# ------------------------------------------------------------------------
repos2bow_parser = add_parser(
"repos2bow", "Convert source code to the bag-of-words model.")
repos2bow_parser.set_defaults(handler=cmd.repos2bow)
args.add_df_args(repos2bow_parser)
args.add_repo2_args(repos2bow_parser)
args.add_feature_args(repos2bow_parser)
args.add_bow_args(repos2bow_parser)
args.add_repartitioner_arg(repos2bow_parser)
args.add_cached_index_arg(repos2bow_parser)
# ------------------------------------------------------------------------
repos2bow_index_parser = add_parser(
"repos2bow_index", "Creates the index, quant and docfreq model of the bag-of-words model.")
repos2bow_index_parser.set_defaults(handler=cmd.repos2bow_index)
args.add_df_args(repos2bow_index_parser)
args.add_repo2_args(repos2bow_index_parser)
args.add_feature_args(repos2bow_index_parser)
args.add_repartitioner_arg(repos2bow_index_parser)
args.add_cached_index_arg(repos2bow_index_parser, create=True)
# ------------------------------------------------------------------------
repos2df_parser = add_parser(
"repos2df", "Calculate document frequencies of features extracted from source code.")
repos2df_parser.set_defaults(handler=cmd.repos2df)
args.add_df_args(repos2df_parser)
args.add_repo2_args(repos2df_parser)
args.add_feature_args(repos2df_parser)
# ------------------------------------------------------------------------
repos2ids_parser = subparsers.add_parser(
"repos2ids", help="Convert source code to a bag of identifiers.")
repos2ids_parser.set_defaults(handler=cmd.repos2ids)
args.add_repo2_args(repos2ids_parser)
args.add_split_stem_arg(repos2ids_parser)
args.add_repartitioner_arg(repos2ids_parser)
repos2ids_parser.add_argument(
"-o", "--output", required=True,
help="[OUT] output path to the CSV file with identifiers.")
repos2ids_parser.add_argument(
"--idfreq", action="store_true",
help="Adds identifier frequencies to the output CSV file."
"num_repos is the number of repositories where the identifier appears in."
"num_files is the number of files where the identifier appears in."
"num_occ is the total number of occurrences of the identifier.")
# ------------------------------------------------------------------------
repos2coocc_parser = add_parser(
"repos2coocc", "Convert source code to the sparse co-occurrence matrix of identifiers.")
repos2coocc_parser.set_defaults(handler=cmd.repos2coocc)
args.add_df_args(repos2coocc_parser)
args.add_repo2_args(repos2coocc_parser)
args.add_split_stem_arg(repos2coocc_parser)
args.add_repartitioner_arg(repos2coocc_parser)
repos2coocc_parser.add_argument(
"-o", "--output", required=True,
help="[OUT] Path to the Cooccurrences model.")
# ------------------------------------------------------------------------
repos2roles_and_ids = add_parser(
"repos2roleids", "Converts a UAST to a list of pairs, where pair is a role and "
"identifier. Role is merged generic roles where identifier was found.")
repos2roles_and_ids.set_defaults(handler=cmd.repos2roles_and_ids)
args.add_repo2_args(repos2roles_and_ids)
args.add_split_stem_arg(repos2roles_and_ids)
repos2roles_and_ids.add_argument(
"-o", "--output", required=True,
help="[OUT] Path to the directory where spark should store the result. "
"Inside the directory you find result is csv format, status file and sumcheck files.")
# ------------------------------------------------------------------------
repos2identifier_distance = add_parser(
"repos2id_distance", "Converts a UAST to a list of identifier pairs "
"and distance between them.")
repos2identifier_distance.set_defaults(handler=cmd.repos2id_distance)
args.add_repo2_args(repos2identifier_distance)
args.add_split_stem_arg(repos2identifier_distance)
repos2identifier_distance.add_argument(
"-t", "--type", required=True, choices=extractors.IdentifierDistance.DistanceType.All,
help="Distance type.")
repos2identifier_distance.add_argument(
"--max-distance", default=extractors.IdentifierDistance.DEFAULT_MAX_DISTANCE, type=int,
help="Maximum distance to save.")
repos2identifier_distance.add_argument(
"-o", "--output", required=True,
help="[OUT] Path to the directory where spark should store the result. "
"Inside the directory you find result is csv format, status file and sumcheck files.")
# ------------------------------------------------------------------------
repos2id_sequence = add_parser(
"repos2idseq", "Converts a UAST to sequence of identifiers sorted by order of appearance.")
repos2id_sequence.set_defaults(handler=cmd.repos2id_sequence)
args.add_repo2_args(repos2id_sequence)
args.add_split_stem_arg(repos2id_sequence)
repos2id_sequence.add_argument(
"--skip-docname", default=False, action="store_true",
help="Do not save document name in CSV file, only identifier sequence.")
repos2id_sequence.add_argument(
"-o", "--output", required=True,
help="[OUT] Path to the directory where spark should store the result. "
"Inside the directory you find result is csv format, status file and sumcheck files.")
# ------------------------------------------------------------------------
preproc_parser = add_parser(
"id2vec-preproc", "Convert a sparse co-occurrence matrix to the Swivel shards.")
preproc_parser.set_defaults(handler=cmd.id2vec_preprocess)
args.add_df_args(preproc_parser)
preproc_parser.add_argument("-s", "--shard-size", default=4096, type=int,
help="The shard (submatrix) size.")
preproc_parser.add_argument(
"-i", "--input",
help="Concurrence model produced by repos2coocc.")
preproc_parser.add_argument("-o", "--output", required=True, help="Output directory.")
# ------------------------------------------------------------------------
train_parser = add_parser(
"id2vec-train", "Train identifier embeddings using Swivel.")
train_parser.set_defaults(handler=cmd.run_swivel)
mirror_tf_args(train_parser)
# ------------------------------------------------------------------------
id2vec_postproc_parser = add_parser(
"id2vec-postproc",
"Combine row and column embeddings produced by Swivel and write them to an .asdf.")
id2vec_postproc_parser.set_defaults(handler=cmd.id2vec_postprocess)
id2vec_postproc_parser.add_argument(
"-i", "--swivel-data", required=True,
help="Folder with swivel row and column embeddings data. "
"You can get it using id2vec_train subcommand.")
id2vec_postproc_parser.add_argument(
"-o", "--output", required=True,
help="Output directory for Id2Vec model.")
# ------------------------------------------------------------------------
id2vec_project_parser = add_parser(
"id2vec-project", "Present id2vec model in Tensorflow Projector.")
id2vec_project_parser.set_defaults(handler=cmd.id2vec_project)
args.add_df_args(id2vec_project_parser, required=False)
id2vec_project_parser.add_argument("-i", "--input", required=True,
help="id2vec model to present.")
id2vec_project_parser.add_argument("-o", "--output", required=True,
help="Projector output directory.")
id2vec_project_parser.add_argument("--no-browser", action="store_true",
help="Do not open the browser.")
# ------------------------------------------------------------------------
train_id_split_parser = add_parser(
"train-id-split", "Train a neural network to split identifiers.")
train_id_split_parser.set_defaults(handler=cmd.train_id_split)
# common arguments for CNN/RNN models
train_id_split_parser.add_argument("-i", "--input", required=True,
help="Path to the input data in CSV format:"
"num_files,num_occ,num_repos,token,token_split")
train_id_split_parser.add_argument("-e", "--epochs", type=int, default=10,
help="Number of training epochs. The more the better"
"but the training time is proportional.")
train_id_split_parser.add_argument("-b", "--batch-size", type=int, default=500,
help="Batch size. Higher values better utilize GPUs"
"but may harm the convergence.")
train_id_split_parser.add_argument("-l", "--length", type=int, default=40,
help="RNN sequence length.")
train_id_split_parser.add_argument("-o", "--output", required=True,
help="Path to store the trained model.")
train_id_split_parser.add_argument("-t", "--test-ratio", type=float, default=0.2,
help="Fraction of the dataset to use for evaluation.")
train_id_split_parser.add_argument("-p", "--padding", default="post", choices=("pre", "post"),
help="Whether to pad before or after each sequence.")
train_id_split_parser.add_argument("--optimizer", default="Adam", choices=("RMSprop", "Adam"),
help="Algorithm to use as an optimizer for the neural net.")
train_id_split_parser.add_argument("--lr", default=0.001, type=float,
help="Initial learning rate.")
train_id_split_parser.add_argument("--final-lr", default=0.00001, type=float,
help="Final learning rate. The decrease from "
"the initial learning rate is done linearly.")
train_id_split_parser.add_argument("--samples-before-report", type=int, default=5*10**6,
help="Number of samples between each validation report"
"and training updates.")
train_id_split_parser.add_argument("--val-batch-size", type=int, default=2000,
help="Batch size for validation."
"It can be increased to speed up the pipeline but"
"it proportionally increases the memory consumption.")
train_id_split_parser.add_argument("--seed", type=int, default=1989,
help="Random seed.")
train_id_split_parser.add_argument("--devices", default="0",
help="Device(s) to use. '-1' means CPU.")
train_id_split_parser.add_argument("--csv-identifier", default=3,
help="Column name in the CSV file for the raw identifier.")
train_id_split_parser.add_argument("--csv-identifier-split", default=4,
help="Column name in the CSV file for the split"
"identifier.")
train_id_split_parser.add_argument("--include-csv-header", action="store_true",
help="Treat the first line of the input CSV as a regular"
"line.")
train_id_split_parser.add_argument("--model", type=str, choices=("RNN", "CNN"), required=True,
help="Neural Network model to use to learn the identifier"
"splitting task.")
train_id_split_parser.add_argument("-s", "--stack", default=2, type=int,
help="Number of layers stacked on each other.")
# RNN specific arguments
train_id_split_parser.add_argument("--type-cell", default="LSTM",
choices=("GRU", "LSTM", "CuDNNLSTM", "CuDNNGRU"),
help="Recurrent layer type to use.")
train_id_split_parser.add_argument("-n", "--neurons", default=256, type=int,
help="Number of neurons on each layer.")
# CNN specific arguments
train_id_split_parser.add_argument("-f", "--filters", default="64,32,16,8",
help="Number of filters for each kernel size.")
train_id_split_parser.add_argument("-k", "--kernel-sizes", default="2,4,8,16",
help="Sizes for sliding windows.")
train_id_split_parser.add_argument("--dim-reduction", default=32, type=int,
help="Number of 1-d kernels to reduce dimensionality"
"after each layer.")
# ------------------------------------------------------------------------
bow2vw_parser = add_parser(
"bow2vw", "Convert a bag-of-words model to the dataset in Vowpal Wabbit format.")
bow2vw_parser.set_defaults(handler=cmd.bow2vw)
bow2vw_parser.add_argument(
"--bow", help="URL or path to a bag-of-words model.")
bow2vw_parser.add_argument(
"--id2vec", help="URL or path to the identifier embeddings.")
bow2vw_parser.add_argument(
"-o", "--output", required=True, help="Path to the output file.")
# ------------------------------------------------------------------------
bigartm_postproc_parser = add_parser(
"bigartm2asdf", "Convert a human-readable BigARTM model to Modelforge format.")
bigartm_postproc_parser.set_defaults(handler=cmd.bigartm2asdf)
bigartm_postproc_parser.add_argument("input")
bigartm_postproc_parser.add_argument("output")
# ------------------------------------------------------------------------
bigartm_parser = add_parser(
"bigartm", "Install bigartm/bigartm to the current working directory.")
bigartm_parser.set_defaults(handler=install_bigartm)
bigartm_parser.add_argument(
"--tmpdir", help="Store intermediate files in this directory instead of /tmp.")
bigartm_parser.add_argument("--output", default=os.getcwd(), help="Output directory.")
# ------------------------------------------------------------------------
merge_df = add_parser("merge-df", "Merge DocumentFrequencies models to a single one.")
merge_df.set_defaults(handler=cmd.merge_df)
args.add_min_docfreq(merge_df)
args.add_vocabulary_size_arg(merge_df)
merge_df.add_argument(
"-o", "--output", required=True,
help="Path to the merged DocumentFrequencies model.")
merge_df.add_argument(
"-i", "--input", required=True, nargs="+",
help="DocumentFrequencies models input files."
"Use `-i -` to read input files from stdin.")
merge_df.add_argument(
"--ordered", action="store_true", default=False,
help="Save OrderedDocumentFrequencies. "
"If not specified DocumentFrequencies model will be saved")
# ------------------------------------------------------------------------
merge_coocc = add_parser("merge-coocc", "Merge several Cooccurrences models together.")
merge_coocc.set_defaults(handler=cmd.merge_coocc)
add_spark_args(merge_coocc)
merge_coocc.add_argument(
"-o", "--output", required=True,
help="Path to the merged Cooccurrences model.")
merge_coocc.add_argument(
"-i", "--input", required=True,
help="Cooccurrences models input files."
"Use `-i -` to read input files from stdin.")
merge_coocc.add_argument(
"--docfreq", required=True,
help="[IN] Specify OrderedDocumentFrequencies model. "
"Identifiers that are not present in the model will be ignored.")
merge_coocc.add_argument(
"--no-spark", action="store_true", default=False,
help="Use the local reduction instead of PySpark. "
"Can be faster and consume less memory if the data fits into RAM.")
# ------------------------------------------------------------------------
merge_bow = add_parser("merge-bow", "Merge BOW models to a single one.")
merge_bow.set_defaults(handler=cmd.merge_bow)
merge_bow.add_argument(
"-i", "--input", required=True, nargs="+",
help="BOW models input files."
"Use `-i -` to read input files from stdin.")
merge_bow.add_argument(
"-o", "--output", required=True,
help="Path to the merged BOW model.")
merge_bow.add_argument(
"-f", "--features", nargs="+",
choices=[ex.NAME for ex in extractors.__extractors__.values()],
default=None, help="To keep only specific features, if not specified all will be kept.")
# ------------------------------------------------------------------------
id2role_eval = add_parser("id2role-eval",
"Compare the embeddings quality on role prediction problem.")
id2role_eval.set_defaults(handler=cmd.id2role_eval)
id2role_eval.add_argument(
"-m", "--models", required=True, nargs="+",
help="Id2Vec models to compare."
"Use `-i -` to read input files from stdin.")
id2role_eval.add_argument(
"-d", "--dataset", required=True,
help="Dataset directory. You can collect dataset via repos2roleids command.")
id2role_eval.add_argument(
"-s", "--seed", default=420,
help="Random seed for reproducible results.")
return parser
def get_parser() -> argparse.ArgumentParser:
"""
Create the cmdline argument parser.
"""
parser = argparse.ArgumentParser(formatter_class=ArgumentDefaultsHelpFormatterNoNone)
parser.add_argument("--log-level", default="INFO", choices=logging._nameToLevel,
help="Logging verbosity.")
def add_feature_weight_arg(my_parser):
help_desc = "%s's weight - all features from this extractor will be multiplied by this " \
"factor"
for ex in extractors.__extractors__.values():
my_parser.add_argument("--%s-weight" % ex.NAME, default=1, type=float,
help=help_desc % ex.__name__)
def add_cassandra_args(my_parser):
my_parser.add_argument(
"--cassandra", default="0.0.0.0:9042", help="Cassandra's host:port.")
my_parser.add_argument("--keyspace", default="apollo",
help="Cassandra's key space.")
my_parser.add_argument(
"--tables", help="Table name mapping (JSON): bags, hashes, hashtables, hashtables2.")
def add_wmh_args(my_parser, params_help: str, add_hash_size: bool, required: bool):
if add_hash_size:
my_parser.add_argument("--size", type=int, default=128, help="Hash size.")
my_parser.add_argument("-p", "--params", required=required, help=params_help)
my_parser.add_argument("-t", "--threshold", required=required, type=float,
help="Jaccard similarity threshold.")
my_parser.add_argument("--false-positive-weight", type=float, default=0.5,
help="Used to adjust the relative importance of "
"minimizing false positives count when optimizing "
"for the Jaccard similarity threshold.")
my_parser.add_argument("--false-negative-weight", type=float, default=0.5,
help="Used to adjust the relative importance of "
"minimizing false negatives count when optimizing "
"for the Jaccard similarity threshold.")
def add_template_args(my_parser, default_template):
my_parser.add_argument("--batch", type=int, default=100,
help="Number of hashes to query at a time.")
my_parser.add_argument("--template", default=default_template,
help="Jinja2 template to render.")
def add_dzhigurda_arg(my_parser):
my_parser.add_argument(
"--dzhigurda", default=0, type=int,
help="Index of the examined commit in the history.")
# Create and construct subparsers
subparsers = parser.add_subparsers(help="Commands", dest="command")
# ------------------------------------------------------------------------
warmup_parser = subparsers.add_parser(
"warmup", help="Initialize source{d} engine.")
warmup_parser.set_defaults(handler=warmup)
add_engine_args(warmup_parser, default_packages=[CASSANDRA_PACKAGE])
# ------------------------------------------------------------------------
db_parser = subparsers.add_parser("resetdb", help="Destructively initialize the database.")
db_parser.set_defaults(handler=reset_db)
add_cassandra_args(db_parser)
db_parser.add_argument(
"--hashes-only", action="store_true",
help="Only clear the tables: hashes, hashtables, hashtables2. Do not touch the rest.")
# ------------------------------------------------------------------------
preprocessing_parser = subparsers.add_parser(
"preprocess", help="Generate the dataset with good candidates for duplication.")
preprocessing_parser.set_defaults(handler=preprocess_source)
preprocessing_parser.add_argument("-x", "--mode", choices=Moder.Options.__all__,
default="file", help="What to select for analysis.")
add_repo2_args(preprocessing_parser)
add_dzhigurda_arg(preprocessing_parser)
preprocessing_parser.add_argument(
"-o", "--output", required=True,
help="[OUT] Path to the Parquet files with bag batches.")
default_fields = ["blob_id", "repository_id", "content", "path", "commit_hash", "uast"]
preprocessing_parser.add_argument(
"-f", "--fields", action="append", default=default_fields,
help="Fields to select from DF to save.")
# ------------------------------------------------------------------------
source2bags_parser = subparsers.add_parser(
"bags", help="Convert source code to weighted sets.")
source2bags_parser.set_defaults(handler=source2bags)
add_bow_args(source2bags_parser)
add_dzhigurda_arg(source2bags_parser)
add_repo2_args(source2bags_parser, default_packages=[CASSANDRA_PACKAGE])
add_feature_args(source2bags_parser)
add_cassandra_args(source2bags_parser)
add_df_args(source2bags_parser)
# ------------------------------------------------------------------------
hash_parser = subparsers.add_parser(
"hash", help="Run MinHashCUDA on the bag batches.")
hash_parser.set_defaults(handler=hash_batches)
hash_parser.add_argument("-i", "--input",
help="Path to the directory with Parquet files.")
hash_parser.add_argument("--seed", type=int, default=int(time()),
help="Random generator's seed.")
hash_parser.add_argument("--mhc-verbosity", type=int, default=1,
help="MinHashCUDA logs verbosity level.")
hash_parser.add_argument("--devices", type=int, default=0,
help="Or-red indices of NVIDIA devices to use. 0 means all.")
add_wmh_args(hash_parser, "Path to the output file with WMH parameters.", True, True)
add_cassandra_args(hash_parser)
add_spark_args(hash_parser, default_packages=[CASSANDRA_PACKAGE])
add_feature_weight_arg(hash_parser)
# ------------------------------------------------------------------------
query_parser = subparsers.add_parser("query", help="Query for similar files.")
query_parser.set_defaults(handler=query)
mode_group = query_parser.add_mutually_exclusive_group(required=True)
mode_group.add_argument("-i", "--id", help="Query for this id (id mode).")
mode_group.add_argument("-c", "--file", help="Query for this file (file mode).")
query_parser.add_argument("--docfreq", help="Path to OrderedDocumentFrequencies (file mode).")
query_parser.add_argument("--min-docfreq", default=1, type=int,
help="The minimum document frequency of each feature.")
query_parser.add_argument(
"--bblfsh", default="localhost:9432", help="Babelfish server's address.")
query_parser.add_argument("--precise", action="store_true",
help="Calculate the precise set.")
add_wmh_args(query_parser, "Path to the Weighted MinHash parameters.", False, False)
add_feature_args(query_parser, required=False)
add_template_args(query_parser, "query.md.jinja2")
add_cassandra_args(query_parser)
# ------------------------------------------------------------------------
cc_parser = subparsers.add_parser(
"cc", help="Load the similar pairs of files and run connected components analysis.")
cc_parser.set_defaults(handler=find_connected_components)
add_cassandra_args(cc_parser)
cc_parser.add_argument("-o", "--output", required=True,
help="[OUT] Path to connected components ASDF model.")
# ------------------------------------------------------------------------
dumpcc_parser = subparsers.add_parser(
"dumpcc", help="Output the connected components to stdout.")
dumpcc_parser.set_defaults(handler=dumpcc)
dumpcc_parser.add_argument("-i", "--input", required=True,
help="Path to connected components ASDF model.")
# ------------------------------------------------------------------------
community_parser = subparsers.add_parser(
"cmd", help="Run Community Detection analysis on the connected components from \"cc\".")
community_parser.set_defaults(handler=detect_communities)
community_parser.add_argument("-i", "--input", required=True,
help="Path to connected components ASDF model.")
community_parser.add_argument("-o", "--output", required=True,
help="[OUT] Path to the communities ASDF model.")
community_parser.add_argument("--edges", choices=("linear", "quadratic", "1", "2"),
default="linear",
help="The method to generate the graph's edges: bipartite - "
"linear and fast, but may not fit some the CD algorithms, "
"or all to all within a bucket - quadratic and slow, but "
"surely fits all the algorithms.")
cmd_choices = [k[10:] for k in dir(Graph) if k.startswith("community_")]
community_parser.add_argument("-a", "--algorithm", choices=cmd_choices,
default="walktrap",
help="The community detection algorithm to apply.")
community_parser.add_argument("-p", "--params", type=json.loads, default={},
help="Parameters for the algorithm (**kwargs, JSON format).")
community_parser.add_argument("--no-spark", action="store_true", help="Do not use Spark.")
add_spark_args(community_parser)
# ------------------------------------------------------------------------
dumpcmd_parser = subparsers.add_parser(
"dumpcmd", help="Output the detected communities to stdout.")
dumpcmd_parser.set_defaults(handler=dumpcmd)
dumpcmd_parser.add_argument("input", help="Path to the communities ASDF model.")
add_template_args(dumpcmd_parser, "report.md.jinja2")
add_cassandra_args(dumpcmd_parser)
# ------------------------------------------------------------------------
evalcc_parser = subparsers.add_parser(
"evalcc", help="Evaluate the communities: calculate the precise similarity and the "
"fitness metric.")
evalcc_parser.set_defaults(handler=evaluate_communities)
evalcc_parser.add_argument("-t", "--threshold", required=True, type=float,
help="Jaccard similarity threshold.")
evalcc_parser.add_argument("-i", "--input", required=True,
help="Path to the communities model.")
add_spark_args(evalcc_parser, default_packages=[CASSANDRA_PACKAGE])
add_cassandra_args(evalcc_parser)
# TODO: retable [.....] -> [.] [.] [.] [.] [.]
return parser