hohn/sarif-cli
1
0
Fork 0
mirror of https://github.com/hohn/sarif-cli.git synced 2025-12-16 17:23:03 +01:00
Code Issues Packages Projects Releases Wiki Activity

sarif-extract-tables: Form the codeFlows dataframe and write it out

Browse Source 
One of the shorter multi-path results from
     cd ~/local/sarif-cli/data/treeio
     ../../bin/sarif-results-summary -r results.sarif |less
follows; the dataframe formed here starts with the codeFlows-containing table 9699
and has the content of the PATH * output below.

    RESULT: static/js/jquery-ui-1.10.3/demos/dialog/modal-form.html:89:35:93:14: [DOM text](1) is reinte
    rpreted as HTML without escaping meta-characters.
    [DOM text](2) is reinterpreted as HTML without escaping meta-characters.
    [DOM text](3) is reinterpreted as HTML without escaping meta-characters.
    REFERENCE: static/js/jquery-ui-1.10.3/demos/dialog/modal-form.html:90:17:90:27: DOM text
    REFERENCE: static/js/jquery-ui-1.10.3/demos/dialog/modal-form.html:91:17:91:28: DOM text
    REFERENCE: static/js/jquery-ui-1.10.3/demos/dialog/modal-form.html:92:17:92:31: DOM text
    PATH 0
    FLOW STEP 0: static/js/jquery-ui-1.10.3/demos/dialog/modal-form.html:90:17:90:27: name.val()
    FLOW STEP 1: static/js/jquery-ui-1.10.3/demos/dialog/modal-form.html:89:35:93:14: "<tr>"  ... "</tr>
    "
    PATH 1
    FLOW STEP 0: static/js/jquery-ui-1.10.3/demos/dialog/modal-form.html:91:17:91:28: email.val()
    FLOW STEP 1: static/js/jquery-ui-1.10.3/demos/dialog/modal-form.html:89:35:93:14: "<tr>"  ... "</tr>"
    PATH 2
    FLOW STEP 0: static/js/jquery-ui-1.10.3/demos/dialog/modal-form.html:92:17:92:31: password.val()
    FLOW STEP 1: static/js/jquery-ui-1.10.3/demos/dialog/modal-form.html:89:35:93:14: "<tr>"  ... "</tr>"
This commit is contained in:
Michael Hohn
2022-02-22 16:50:44 -08:00
committed by =Michael Hohn
parent ad738abed3
commit 1dbd240b5b
1 changed files with 65 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

67
bin/sarif-extract-tables
View File

@@ -163,20 +163,83 @@ dr1 = (
.drop(columns=['struct_id', 'message_2683'])
)
# Keep columns of interest
dr2 = (dr1[['struct_id_4055', 'uri', 'startLine', 'startColumn', 'endLine', 'endColumn', 'text']]
.rename({'text': 'message'}, axis='columns'))
# Remove dummy locations previously injected by signature.fillsig
dr3 = dr2[dr2.uri != 'scli-dyys dummy value']
#
# Form the codeFlows dataframe
#
dco1 = (
sf(9699)
.merge(af(9799), how="left", left_on='codeFlows', right_on='array_id', validate="1:m")
.drop(columns=['struct_id', 'codeFlows', 'array_id', 'type_at_index'])
#
.merge(sf(7122), how="left", left_on='id_or_value_at_index', right_on='struct_id', validate="1:m")
.drop(columns=['id_or_value_at_index', 'struct_id'])
#
.merge(af(1597), how="left", left_on='threadFlows', right_on='array_id',
suffixes=("_codeFlow_9799", "_threadFlows_1597"), validate="1:m")
.drop(columns=['threadFlows', 'array_id', 'type_at_index'])
#
.merge(sf(4194), how="left", left_on='id_or_value_at_index', right_on='struct_id',
suffixes=("_9699", "_4194"), validate="1:m")
.drop(columns=['id_or_value_at_index', 'struct_id'])
#
.merge(af(1075), how="left", left_on='locations_4194', right_on='array_id', validate="1:m")
.drop(columns=['locations_4194', 'array_id', 'type_at_index'])
.rename(columns={"value_index": "value_index_locations_1075"})
#
.merge(sf('0987'), how="left", left_on='id_or_value_at_index', right_on='struct_id', validate="1:m")
.drop(columns=['id_or_value_at_index', 'struct_id'])
#
.merge(sf(2683), how="left", left_on='location', right_on='struct_id',
suffixes=("_9699", "_2683"), validate="1:m")
.drop(columns=['location', 'struct_id'])
#
# The below is similar to dr1
#
.merge(sf(4963), how="left", left_on='physicalLocation', right_on='struct_id', validate="1:m")
.drop(columns=['struct_id', 'physicalLocation'])
#
.merge(sf(6299), how="left", left_on='region', right_on='struct_id', validate="1:m")
.drop(columns=['struct_id', 'region'])
#
.merge(sf(2685), how="left", left_on='artifactLocation', right_on='struct_id', validate="1:m")
.drop(columns=['struct_id', 'artifactLocation'])
#
.merge(sf(2774), how="left", left_on='message_2683', right_on='struct_id', validate="1:m")
.drop(columns=['struct_id', 'message_2683'])
)
# Keep columns of interest
dco2 = (dco1[['uri',
'startLine', 'startColumn', 'endLine', 'endColumn',
'text',
'ruleIndex', 'value_index_codeFlow_9799',
'value_index_threadFlows_1597', 'value_index_locations_1075',
]]
.rename({'text': 'message',
'value_index_codeFlow_9799': 'idx_codeFlow',
'value_index_threadFlows_1597': 'idx_threadFlows',
'value_index_locations_1075': 'idx_locations'}, axis='columns'))
# Remove dummy locations previously injected by signature.fillsig
dco3 = dco2[dco2.uri != 'scli-dyys dummy value']
#
# Write output
#
if args.output_format == 'csv':
p = pathlib.Path(args.outdir)
p.mkdir(exist_ok=True)
with p.joinpath('messages.csv').open(mode='wb') as messages:
d2.to_csv(messages, index_label='index')
with p.joinpath('problem.csv').open(mode='wb') as problem:
d2.to_csv(problem, index_label='index')
with p.joinpath('path-problem.csv').open(mode='wb') as path_problem:
dco3.to_csv(path_problem, index_label='index')
with p.joinpath('relatedLocations.csv').open(mode='wb') as relo:
dr3.to_csv(relo, index_label='index')