Full revision of the base tables derived from multiple sarif input files

The new base tables produced by `sarif-extract-multi` are
    artifacts
    codeflows
    kind_pathproblem
    kind_problem
    project
    relatedLocations
    rules

The revised table overview is in the jupyter notebook
scripts/multi-table-overview.ipynb

The file notes/typegraph-multi-with-tables.pdf illustrates what original (sarif)
tables are used to form the base (derived) tables.
This commit is contained in:
Michael Hohn
2022-03-23 16:37:41 -07:00
committed by =Michael Hohn
parent db00f17137
commit d5390bb87e
4 changed files with 5440 additions and 3900 deletions

View File

@@ -59,33 +59,35 @@ typegraph.destructure(tgraph, signature_multi.start_node_2022_03_08, meta_struct
# Form output tables # Form output tables
# #
typegraph.attach_tables(tgraph) typegraph.attach_tables(tgraph)
#
# Form dataframes originally introduced by sarif-extract-tables #
# Dataframe / table collection
# #
@dataclass @dataclass
class BaseTables: class BaseTables:
kind_problem : pd.DataFrame
kind_pathproblem : pd.DataFrame
codeflows : pd.DataFrame
relatedLocations : pd.DataFrame
project : pd.DataFrame
rules : pd.DataFrame
artifacts : pd.DataFrame artifacts : pd.DataFrame
codeflows : pd.DataFrame
kind_pathproblem : pd.DataFrame
kind_problem : pd.DataFrame
project : pd.DataFrame
relatedLocations : pd.DataFrame
rules : pd.DataFrame
def __init__(self): pass def __init__(self): pass
bt = BaseTables() bt = BaseTables()
#
# Add dataframes
#
sf_2683 = tj.joins_for_sf_2683(tgraph) sf_2683 = tj.joins_for_sf_2683(tgraph)
bt.kind_problem = tj.joins_for_problem(tgraph, sf_2683) af_0350_location = tj.joins_for_af_0350_location(tgraph)
bt.kind_pathproblem = tj.joins_for_path_problem(tgraph, sf_2683)
bt.codeflows = tj.joins_for_codeflows(tgraph, sf_2683)
bt.relatedLocations = tj.joins_for_relatedLocations(tgraph, sf_2683)
#
# Form the new dataframes
#
bt.project = tj.joins_for_project(tgraph)
bt.rules = tj.joins_for_rules(tgraph)
bt.artifacts = tj.joins_for_artifacts(tgraph) bt.artifacts = tj.joins_for_artifacts(tgraph)
bt.codeflows = tj.joins_for_codeflows(tgraph, sf_2683)
bt.kind_pathproblem = tj.joins_for_path_problem(tgraph, af_0350_location)
bt.kind_problem = tj.joins_for_problem(tgraph, af_0350_location)
bt.project = tj.joins_for_project(tgraph) # multi-sarif only
bt.relatedLocations = tj.joins_for_relatedLocations(tgraph, sf_2683)
bt.rules = tj.joins_for_rules(tgraph)
# #
# Write output # Write output
# #
@@ -93,12 +95,12 @@ p = pathlib.Path(args.outdir)
p.mkdir(exist_ok=True) p.mkdir(exist_ok=True)
def write(path, frame): def write(path, frame):
with p.joinpath(path + ".csv").open(mode='wb') as fh: with p.joinpath(path + ".csv").open(mode='wb') as fh:
frame.to_csv(fh, index_label='index') frame.to_csv(fh, index=False)
write('kind_problem', bt.kind_problem)
write('kind_pathproblem', bt.kind_pathproblem)
write('codeflows', bt.codeflows)
write('relatedLocations', bt.relatedLocations)
write('project', bt.project)
write('rules', bt.rules)
write('artifacts', bt.artifacts) write('artifacts', bt.artifacts)
write('codeflows', bt.codeflows)
write('kind_pathproblem', bt.kind_pathproblem)
write('kind_problem', bt.kind_problem)
write('project', bt.project)
write('relatedLocations', bt.relatedLocations)
write('rules', bt.rules)

Binary file not shown.

View File

@@ -6,8 +6,57 @@
provides those for the other tables. provides those for the other tables.
""" """
import pandas as pd import pandas as pd
import re
from .typegraph import tagged_array_columns, tagged_struct_columns from .typegraph import tagged_array_columns, tagged_struct_columns
def joins_for_af_0350_location(tgraph):
"""
Join all the tables used by 0350's right side into one.
"""
# Access convenience functions
sf = lambda num: tgraph.dataframes['Struct' + str(num)]
af = lambda num: tgraph.dataframes['Array' + str(num)]
sft = lambda id: sf(id).rename(columns = tagged_struct_columns(tgraph, id))
aft = lambda id: af(id).rename(columns = tagged_array_columns(tgraph, id))
af_0350_location = (
aft('0350')
#
.merge(sft(2683), how="left", left_on='t0350_id_or_value_at_index', right_on='t2683_struct_id',
validate="1:m")
.drop(columns=['t0350_id_or_value_at_index', 't2683_struct_id', 't0350_type_at_index'])
#
.merge(sft(4963), how="left", left_on='t2683_physicalLocation', right_on='t4963_struct_id',
validate="1:m")
.drop(columns=['t2683_physicalLocation', 't4963_struct_id'])
#
.merge(sft(6299), how="left", left_on='t4963_region', right_on='t6299_struct_id',
validate="1:m")
.drop(columns=['t4963_region', 't6299_struct_id'])
#
.merge(sft(2685), how="left", left_on='t4963_artifactLocation', right_on='t2685_struct_id',
validate="1:m")
.drop(columns=['t4963_artifactLocation', 't2685_struct_id'])
#
.merge(sft(2774), how="left", left_on='t2683_message', right_on='t2774_struct_id',
validate="1:m")
.drop(columns=['t2683_message', 't2774_struct_id'])
#
.rename(columns={'t0350_array_id' : 'm0350_location_array_id',
't0350_value_index' : 'm0350_location_array_index',
't2683_id' : 'm0350_location_id',
't6299_endColumn' : 'm0350_location_endColumn',
't6299_endLine' : 'm0350_location_endLine',
't6299_startColumn' : 'm0350_location_startColumn',
't6299_startLine' : 'm0350_location_startLine',
't2685_index' : 'm0350_location_index',
't2685_uri' : 'm0350_location_uri',
't2685_uriBaseId' : 'm0350_location_uriBaseId',
't2774_text' : 'm0350_location_message',
})
)
return af_0350_location
def joins_for_sf_2683(tgraph): def joins_for_sf_2683(tgraph):
""" """
Join all the tables used by 2683's right side into one. Join all the tables used by 2683's right side into one.
@@ -39,45 +88,71 @@ def joins_for_sf_2683(tgraph):
return sf_2683 return sf_2683
def joins_for_problem(tgraph, sf_2683): def joins_for_problem(tgraph, af_0350_location):
""" """
Return table providing the `problem` information. Return table providing the `problem` information.
""" """
# Access convenience functions # Access convenience functions
sf = lambda num: tgraph.dataframes['Struct' + str(num)] sf = lambda num: tgraph.dataframes['Struct' + str(num)]
af = lambda num: tgraph.dataframes['Array' + str(num)] af = lambda num: tgraph.dataframes['Array' + str(num)]
sft = lambda id: sf(id).rename(columns = tagged_struct_columns(tgraph, id))
aft = lambda id: af(id).rename(columns = tagged_array_columns(tgraph, id))
# #
# Form the message dataframe (@kind problem) via joins # Form the message dataframe (@kind problem) via joins
# #
kind_problem_1 = ( kind_problem_1 = (
af(6343) aft(6343)
.rename(columns={"value_index": "results_idx_6343", "array_id": "result_id_6343"}) .merge(sft(4055), how="inner",
.merge(sf(4055), how="inner", left_on='id_or_value_at_index', right_on='struct_id', left_on='t6343_id_or_value_at_index', right_on='t4055_struct_id',
validate="1:m") validate="1:m")
.drop(columns=['type_at_index', 'id_or_value_at_index', 'struct_id']) .drop(columns=['t6343_type_at_index', 't6343_id_or_value_at_index',
.rename(columns={"message": "result_message_4055", 't4055_struct_id'])
"relatedLocations": "relatedLocations_id"})
# locations
.merge(af('0350'), how="left", left_on='locations', right_on='array_id', validate="1:m")
.drop(columns=['locations', 'array_id', 'type_at_index'])
#
.merge(sf_2683, how="left", left_on='id_or_value_at_index', right_on='struct_id_2683', validate="1:m")
.drop(columns=['id_or_value_at_index', 'struct_id_2683'])
#
.merge(sf(2774), how="left", left_on='result_message_4055', right_on='struct_id', validate="1:m")
.drop(columns=['struct_id', 'result_message_4055'])
.rename(columns={"text": "message_text_4055"})
#
.merge(sf(4199), how="left", left_on='partialFingerprints', right_on='struct_id', validate="1:m")
.drop(columns=['struct_id', 'partialFingerprints'])
# #
.merge( .merge(af_0350_location, how="left", left_on='t4055_locations',
sf(3942).rename(columns={"id": "rule_id", "index": "rule_index"}), right_on='m0350_location_array_id', validate="1:m")
how="left", left_on='rule', right_on='struct_id', validate="1:m") .drop(columns=['t4055_locations', 'm0350_location_array_id'])
.drop(columns=['struct_id', 'rule'])
# #
.merge(af_0350_location.rename(columns=lambda x: re.sub('m0350_location',
'm0350_relatedLocation',
x)),
how="left", left_on='t4055_relatedLocations',
right_on='m0350_relatedLocation_array_id', validate="1:m")
.drop(columns=['t4055_relatedLocations', 'm0350_relatedLocation_array_id'])
#
.merge(sft(2774), how="left", left_on='t4055_message', right_on='t2774_struct_id')
.drop(columns=['t4055_message', 't2774_struct_id'])
.rename(columns={"t2774_text": "t4055_message_text"})
#
.merge(sft(4199), how="left", left_on='t4055_partialFingerprints',
right_on='t4199_struct_id')
.drop(columns=['t4055_partialFingerprints', 't4199_struct_id'])
#
.merge(sft(3942), how="left", left_on='t4055_rule',
right_on='t3942_struct_id')
.drop(columns=['t4055_rule', 't3942_struct_id'])
) )
return kind_problem_1
kind_problem_2 = (
kind_problem_1
.rename({
't6343_array_id' : 'results_array_id',
't6343_value_index' : 'results_array_index',
't4055_ruleId' : 'ruleId',
't4055_ruleIndex' : 'ruleIndex',
't4055_message_text' : 'message_text',
't3942_id' : 'rule_id',
't3942_index' : 'rule_index',
}, axis='columns')
# Strip type prefix for the rest
.rename(columns = lambda x: re.sub('m0350_|t4199_', '', x))
)
# # TODO potential cleanup
# # Remove dummy locations previously injected by signature.fillsig
# kind_problem_2 = kind_problem_1[kind_problem_1.uri != 'scli-dyys dummy value']
# #
return kind_problem_2
def joins_for_codeflows(tgraph, sf_2683): def joins_for_codeflows(tgraph, sf_2683):
""" """
@@ -87,7 +162,7 @@ def joins_for_codeflows(tgraph, sf_2683):
sf = lambda num: tgraph.dataframes['Struct' + str(num)] sf = lambda num: tgraph.dataframes['Struct' + str(num)]
af = lambda num: tgraph.dataframes['Array' + str(num)] af = lambda num: tgraph.dataframes['Array' + str(num)]
# #
af_9799 = ( codeflows = (
af(9799).rename(columns={"array_id": "t9799_array_id", "value_index": "t9799_idx"}) af(9799).rename(columns={"array_id": "t9799_array_id", "value_index": "t9799_idx"})
# #
.merge(sf(7122), how="left", left_on='id_or_value_at_index', right_on='struct_id', validate="1:m") .merge(sf(7122), how="left", left_on='id_or_value_at_index', right_on='struct_id', validate="1:m")
@@ -111,66 +186,79 @@ def joins_for_codeflows(tgraph, sf_2683):
.merge(sf_2683, how="left", left_on='location', right_on='struct_id_2683', validate="1:m") .merge(sf_2683, how="left", left_on='location', right_on='struct_id_2683', validate="1:m")
.drop(columns=['location', 'struct_id_2683']) .drop(columns=['location', 'struct_id_2683'])
) )
return af_9799 codeflows_1 = (
codeflows
.drop(columns=['id_2683'])
.rename({
't9799_array_id': 'codeflow_id',
't9799_idx': 'codeflow_index',
't1597_idx': 'threadflow_index',
't1075_locations_idx': 'location_index',
'location_index_2685': 'artifact_index',
'message_text_2683': 'message',
}, axis='columns')
)
return codeflows_1
def joins_for_path_problem(tgraph, sf_2683): def joins_for_path_problem(tgraph, af_0350_location):
""" """
Return table providing the `path-problem` information. Return table providing the `path-problem` information.
""" """
# Access convenience functions # Access convenience functions
sf = lambda num: tgraph.dataframes['Struct' + str(num)] sf = lambda num: tgraph.dataframes['Struct' + str(num)]
af = lambda num: tgraph.dataframes['Array' + str(num)] af = lambda num: tgraph.dataframes['Array' + str(num)]
# sft = lambda id: sf(id).rename(columns = tagged_struct_columns(tgraph, id))
aft = lambda id: af(id).rename(columns = tagged_array_columns(tgraph, id))
kind_pathproblem_1 = ( kind_pathproblem_1 = (
af(6343) aft(6343)
.rename(columns={"value_index": "t6343_result_idx", "array_id": "t6343_result_id"}) .merge(sft(9699), how="inner", left_on='t6343_id_or_value_at_index', right_on='t9699_struct_id',
.merge(sf(9699), how="inner", left_on='id_or_value_at_index', right_on='struct_id',
validate="1:m") validate="1:m")
.rename(columns={"codeFlows" : "t9699_codeFlows", .drop(columns=['t6343_id_or_value_at_index', 't9699_struct_id', 't6343_type_at_index'])
"locations" : "t9699_locations",
"message" : "t9699_message",
"partialFingerprints" : "t9699_partialFingerprints",
"relatedLocations" : "t9699_relatedLocations",
"rule" : "t9699_rule",
"ruleId" : "t9699_ruleId",
"ruleIndex" : "t9699_ruleIndex",
})
.drop(columns=['id_or_value_at_index', 'struct_id', 'type_at_index'])
# 9699.locations
.merge(af('0350').rename(columns={"value_index": "t0350_location_idx"}),
how="left", left_on='t9699_locations', right_on='array_id', validate="1:m")
.drop(columns=['t9699_locations', 'array_id', 'type_at_index'])
#
.merge(sf_2683, how="left", left_on='id_or_value_at_index', right_on='struct_id_2683', validate="1:m")
.drop(columns=['id_or_value_at_index', 'struct_id_2683'])
# #
# # TODO: merge or keep separate? .merge(af_0350_location, how="left", left_on='t9699_locations',
# # 9699.codeFlows right_on='m0350_location_array_id', validate="1:m")
# .merge(af_9799, how="left", left_on='t9699_codeFlows', right_on='t9799_array_id', validate="1:m") .drop(columns=['t9699_locations', 'm0350_location_array_id'])
# #
# 9699.message .merge(af_0350_location.rename(columns=lambda x: re.sub('m0350_location',
.merge(sf(2774), how="left", left_on='t9699_message', right_on='struct_id', validate="1:m") 'm0350_relatedLocation',
.drop(columns=['struct_id', 't9699_message']) x)),
.rename(columns={"text": "t9699_message_text"}) how="left", left_on='t9699_relatedLocations',
# right_on='m0350_relatedLocation_array_id', validate="1:m")
# 9699.partialFingerprints .drop(columns=['t9699_relatedLocations', 'm0350_relatedLocation_array_id'])
.merge(sf(4199), how="left", left_on='t9699_partialFingerprints', right_on='struct_id', validate="1:m")
.drop(columns=['struct_id', 't9699_partialFingerprints'])
# #
# 9699.relatedLocations -- keep ids .merge(sft(2774), how="left", left_on='t9699_message', right_on='t2774_struct_id')
.drop(columns=['t9699_message', 't2774_struct_id'])
.rename(columns={"t2774_text": "t9699_message_text"})
# #
# 9699.rule .merge(sft(4199), how="left", left_on='t9699_partialFingerprints',
.merge( right_on='t4199_struct_id')
sf(3942).rename(columns={"id": "t3942_rule_id", "index": "t3942_rule_idx"}), .drop(columns=['t9699_partialFingerprints', 't4199_struct_id'])
how="left", left_on='t9699_rule', right_on='struct_id', validate="1:m") #
.drop(columns=['struct_id', 't9699_rule']) .merge(sft(3942), how="left", left_on='t9699_rule',
right_on='t3942_struct_id')
.drop(columns=['t9699_rule', 't3942_struct_id'])
) )
strip_colums = lambda x: re.sub('t9699_|m0350_|t4199_', '', x)
kind_pathproblem_2 = (kind_pathproblem_1
.rename({
't6343_array_id' : 'results_array_id',
't6343_value_index' : 'results_array_index',
't9699_codeFlows' : 'codeFlows_id',
't9699_ruleId' : 'ruleId',
't9699_ruleIndex' : 'ruleIndex',
't9699_message_text' : 'message_text',
't3942_id' : 'rule_id',
't3942_index' : 'rule_index',
}, axis='columns')
# Strip type prefix for the rest
.rename(columns = strip_colums))
# # TODO potential cleanup # # TODO potential cleanup
# # Remove dummy locations previously injected by signature.fillsig # # Remove dummy locations previously injected by signature.fillsig
# kind_pathproblem_2 = kind_pathproblem_1[kind_pathproblem_1.uri != 'scli-dyys dummy value'] # kind_pathproblem_2 = kind_pathproblem_1[kind_pathproblem_1.uri != 'scli-dyys dummy value']
# # # #
return kind_pathproblem_1 return kind_pathproblem_2
def joins_for_relatedLocations(tgraph, sf_2683): def joins_for_relatedLocations(tgraph, sf_2683):
""" """
@@ -267,7 +355,18 @@ def joins_for_project(tgraph):
.drop(columns=['id_or_value_at_index', 'struct_id']) .drop(columns=['id_or_value_at_index', 'struct_id'])
# #
) )
return project_df # Keep columns of interest
project_df_1 = (
project_df
.drop(columns=['value_index_7481', 'versionControl_value_index_5511'])
.rename({
'version_6787': 'sarif_version',
'value_index_0177': 'run_index',
'driver_name_7820': 'driver_name',
'driver_version_7820': 'driver_version',
}, axis='columns')
)
return project_df_1
def joins_for_rules(tgraph): def joins_for_rules(tgraph):
""" """
@@ -310,8 +409,19 @@ def joins_for_rules(tgraph):
.merge(aft(7069), how="left", left_on='t7849_tags', .merge(aft(7069), how="left", left_on='t7849_tags',
right_on='t7069_array_id', validate="1:m") right_on='t7069_array_id', validate="1:m")
.drop(columns=['t7849_tags', 't7069_array_id', 't7069_type_at_index']) .drop(columns=['t7849_tags', 't7069_array_id', 't7069_type_at_index'])
) )
return rules_df rules_2 = (
rules_df
.rename({
't8754_array_id' : 'rules_array_id',
't8754_value_index' : 'rules_array_index',
't7069_value_index' : 'tag_index',
't7069_id_or_value_at_index' : 'tag_text',
}, axis='columns')
# Strip type prefix for the rest
.rename(columns = lambda x: re.sub('t6818_t2774_|t6818_|t8581_|t7849_', '', x))
)
return rules_2
def joins_for_artifacts(tgraph): def joins_for_artifacts(tgraph):
""" """
@@ -330,7 +440,17 @@ def joins_for_artifacts(tgraph):
# #
.merge(sf(2685), how="left", left_on='location', right_on='struct_id', validate="1:m") .merge(sf(2685), how="left", left_on='location', right_on='struct_id', validate="1:m")
.drop(columns=['location', 'struct_id']) .drop(columns=['location', 'struct_id'])
.rename(columns={"index": "location_index_2685", "uri": "location_uri_2685",
"uriBaseId": "location_uriBaseId_2685"})
) )
return artifacts_df # Keep columns of interest and rename
df_1 = (
artifacts_df
.rename({
'array_id': 'artifacts_id',
'artifact_index_4640': 'artifacts_array_index',
}, axis='columns')
)
if (df_1['artifacts_array_index'] == df_1['index']).all():
df_1 = df_1.drop(columns=['artifacts_array_index'])
return df_1

File diff suppressed because it is too large Load Diff