# Requires python3
import re
import sqlite3
import subprocess
import shutil
import os
import codecs
import datetime
import sys
class TskDbDiff(object):
"""Compares two TSK/Autospy SQLite databases.
Attributes:
gold_artifacts:
autopsy_artifacts:
gold_attributes:
autopsy_attributes:
gold_objects:
autopsy_objects:
artifact_comparison:
attribute_comparision:
report_errors: a listof_listof_String, the error messages that will be
printed to screen in the run_diff method
passed: a boolean, did the diff pass?
autopsy_db_file:
gold_db_file:
"""
def __init__(self, output_db, gold_db, output_dir=None, gold_bb_dump=None, gold_dump=None, verbose=False):
"""Constructor for TskDbDiff.
Args:
output_db_path: path to output database (non-gold standard)
gold_db_path: path to gold database
output_dir: (optional) Path to folder where generated files will be put.
gold_bb_dump: (optional) path to file where the gold blackboard dump is located
gold_dump: (optional) path to file where the gold non-blackboard dump is located
verbose: (optional) a boolean, if true, diff results are sent to stdout.
"""
self.output_db_file = output_db
self.gold_db_file = gold_db
self.output_dir = output_dir
self.gold_bb_dump = gold_bb_dump
self.gold_dump = gold_dump
self._generate_gold_dump = False
self._generate_gold_bb_dump = False
self._bb_dump_diff = ""
self._dump_diff = ""
self._bb_dump = ""
self._dump = ""
self.verbose = verbose
if self.gold_bb_dump is None:
self._generate_gold_bb_dump = True
if self.gold_dump is None:
self._generate_gold_dump = True
def run_diff(self):
"""Compare the databases.
Raises:
TskDbDiffException: if an error occurs while diffing or dumping the database
"""
self._init_diff()
# generate the gold database dumps if necessary
if self._generate_gold_dump:
TskDbDiff._dump_output_db_nonbb(self.gold_db_file, self.gold_dump)
if self._generate_gold_bb_dump:
TskDbDiff._dump_output_db_bb(self.gold_db_file, self.gold_bb_dump)
# generate the output database dumps (both DB and BB)
TskDbDiff._dump_output_db_nonbb(self.output_db_file, self._dump)
TskDbDiff._dump_output_db_bb(self.output_db_file, self._bb_dump)
# Compare non-BB
dump_diff_pass = self._diff(self._dump, self.gold_dump, self._dump_diff)
# Compare BB
bb_dump_diff_pass = self._diff(self._bb_dump, self.gold_bb_dump, self._bb_dump_diff)
self._cleanup_diff()
return dump_diff_pass, bb_dump_diff_pass
def _init_diff(self):
"""Set up the necessary files based on the arguments given at construction"""
if self.output_dir is None:
# No stored files
self._bb_dump = TskDbDiff._get_tmp_file("BlackboardDump", ".txt")
self._bb_dump_diff = TskDbDiff._get_tmp_file("BlackboardDump-Diff", ".txt")
self._dump = TskDbDiff._get_tmp_file("DBDump", ".txt")
self._dump_diff = TskDbDiff._get_tmp_file("DBDump-Diff", ".txt")
else:
self._bb_dump = os.path.join(self.output_dir, "BlackboardDump.txt")
self._bb_dump_diff = os.path.join(self.output_dir, "BlackboardDump-Diff.txt")
self._dump = os.path.join(self.output_dir, "DBDump.txt")
self._dump_diff = os.path.join(self.output_dir, "DBDump-Diff.txt")
# Sorting gold before comparing (sort behaves differently in different environments)
new_bb = TskDbDiff._get_tmp_file("GoldBlackboardDump", ".txt")
new_db = TskDbDiff._get_tmp_file("GoldDBDump", ".txt")
if self.gold_bb_dump is not None:
srtcmdlst = ["sort", self.gold_bb_dump, "-o", new_bb]
subprocess.call(srtcmdlst)
srtcmdlst = ["sort", self.gold_dump, "-o", new_db]
subprocess.call(srtcmdlst)
self.gold_bb_dump = new_bb
self.gold_dump = new_db
def _cleanup_diff(self):
if self.output_dir is None:
#cleanup temp files
os.remove(self._dump)
os.remove(self._bb_dump)
if os.path.isfile(self._dump_diff):
os.remove(self._dump_diff)
if os.path.isfile(self._bb_dump_diff):
os.remove(self._bb_dump_diff)
if self.gold_bb_dump is None:
os.remove(self.gold_bb_dump)
os.remove(self.gold_dump)
def _diff(self, output_file, gold_file, diff_path):
"""Compare two text files.
Args:
output_file: a pathto_File, the latest text file
gold_file: a pathto_File, the gold text file
diff_path: The file to write the differences to
Returns False if different
"""
if (not os.path.isfile(output_file)):
return False
if (not os.path.isfile(gold_file)):
return False
# It is faster to read the contents in and directly compare
output_data = codecs.open(output_file, "r", "utf_8").read()
gold_data = codecs.open(gold_file, "r", "utf_8").read()
if (gold_data == output_data):
return True
# If they are different, invoke 'diff'
diff_file = codecs.open(diff_path, "wb", "utf_8")
# Gold needs to be passed in as 1st arg and output as 2nd
dffcmdlst = ["diff", gold_file, output_file]
subprocess.call(dffcmdlst, stdout = diff_file)
# create file path for gold files inside output folder. In case of diff, both gold and current run files
# are available in the report output folder. Prefix Gold- is added to the filename.
gold_file_in_output_dir = output_file[:output_file.rfind("/")] + "/Gold-" + output_file[output_file.rfind("/")+1:]
shutil.copy(gold_file, gold_file_in_output_dir)
return False
def _dump_output_db_bb(db_file, bb_dump_file):
"""Dumps sorted text results to the given output location.
Smart method that deals with a blackboard comparison to avoid issues
with different IDs based on when artifacts were created.
Args:
db_file: a pathto_File, the output database.
bb_dump_file: a pathto_File, the sorted dump file to write to
"""
unsorted_dump = TskDbDiff._get_tmp_file("dump_data", ".txt")
conn = sqlite3.connect(db_file)
conn.text_factory = lambda x: x.decode("utf-8", "ignore")
conn.row_factory = sqlite3.Row
artifact_cursor = conn.cursor()
# Get the list of all artifacts (along with type and associated file)
# @@@ Could add a SORT by parent_path in here since that is how we are going to later sort it.
artifact_cursor.execute("SELECT tsk_files.parent_path, tsk_files.name, blackboard_artifact_types.display_name, blackboard_artifacts.artifact_id FROM blackboard_artifact_types INNER JOIN blackboard_artifacts ON blackboard_artifact_types.artifact_type_id = blackboard_artifacts.artifact_type_id INNER JOIN tsk_files ON tsk_files.obj_id = blackboard_artifacts.obj_id")
database_log = codecs.open(unsorted_dump, "wb", "utf_8")
row = artifact_cursor.fetchone()
appnd = False
counter = 0
artifact_count = 0
artifact_fail = 0
# Cycle through artifacts
try:
while (row != None):
# File Name and artifact type
if(row["parent_path"] != None):
database_log.write(row["parent_path"] + row["name"] + ' ')
else:
database_log.write(row["name"] + ' ')
# Get attributes for this artifact
attribute_cursor = conn.cursor()
looptry = True
artifact_count += 1
try:
art_id = ""
art_id = str(row["artifact_id"])
attribute_cursor.execute("SELECT blackboard_attributes.source, blackboard_attribute_types.display_name, blackboard_attributes.value_type, blackboard_attributes.value_text, blackboard_attributes.value_int32, blackboard_attributes.value_int64, blackboard_attributes.value_double, blackboard_attributes.value_byte FROM blackboard_attributes INNER JOIN blackboard_attribute_types ON blackboard_attributes.attribute_type_id = blackboard_attribute_types.attribute_type_id WHERE artifact_id =? ORDER BY blackboard_attributes.source, blackboard_attribute_types.display_name, blackboard_attributes.value_type, blackboard_attributes.value_text, blackboard_attributes.value_int32, blackboard_attributes.value_int64, blackboard_attributes.value_double", [art_id])
attributes = attribute_cursor.fetchall()
# Print attributes
if (len(attributes) == 0):
# @@@@ This should be
database_log.write(' \n')
row = artifact_cursor.fetchone()
continue
src = attributes[0][0]
for attr in attributes:
numvals = 0
for x in range(3, 6):
if(attr[x] != None):
numvals += 1
if(numvals > 1):
msg = "There were too many values for attribute type: " + attr["display_name"] + " for artifact with id #" + str(row["artifact_id"]) + ".\n"
if(not attr["source"] == src):
msg = "There were inconsistent sources for artifact with id #" + str(row["artifact_id"]) + ".\n"
try:
if attr["value_type"] == 0:
attr_value_as_string = str(attr["value_text"])
elif attr["value_type"] == 1:
attr_value_as_string = str(attr["value_int32"])
elif attr["value_type"] == 2:
attr_value_as_string = str(attr["value_int64"])
elif attr["value_type"] == 3:
attr_value_as_string = str(attr["value_double"])
elif attr["value_type"] == 4:
attr_value_as_string = "bytes"
elif attr["value_type"] == 5:
attr_value_as_string = str(attr["value_int64"])
if attr["display_name"] == "Associated Artifact":
attr_value_as_string = getAssociatedArtifactType(db_file, attr_value_as_string)
patrn = re.compile("[\n\0\a\b\r\f]")
attr_value_as_string = re.sub(patrn, ' ', attr_value_as_string)
database_log.write('')
except IOError as e:
print("IO error")
raise TskDbDiffException("Unexpected IO error while writing to database log." + str(e))
except sqlite3.Error as e:
msg = "Attributes in artifact id (in output DB)# " + str(row["artifact_id"]) + " encountered an error: " + str(e) +" .\n"
print("Attributes in artifact id (in output DB)# ", str(row["artifact_id"]), " encountered an error: ", str(e))
print()
looptry = False
artifact_fail += 1
database_log.write('Error Extracting Attributes')
database_log.close()
raise TskDbDiffException(msg)
finally:
attribute_cursor.close()
# @@@@ This should be
database_log.write(' \n')
row = artifact_cursor.fetchone()
if(artifact_fail > 0):
msg ="There were " + str(artifact_count) + " artifacts and " + str(artifact_fail) + " threw an exception while loading.\n"
except Exception as e:
raise TskDbDiffException("Unexpected error while dumping blackboard database: " + str(e))
finally:
database_log.close()
artifact_cursor.close()
conn.close()
# Now sort the file
srtcmdlst = ["sort", unsorted_dump, "-o", bb_dump_file]
subprocess.call(srtcmdlst)
def _dump_output_db_nonbb(db_file, dump_file):
"""Dumps a database to a text file.
Does not dump the artifact and attributes.
Args:
db_file: a pathto_File, the database file to dump
dump_file: a pathto_File, the location to dump the non-blackboard database items
"""
# Make a copy that we can modify
backup_db_file = TskDbDiff._get_tmp_file("tsk_backup_db", ".db")
shutil.copy(db_file, backup_db_file)
# We sometimes get situations with messed up permissions
os.chmod (backup_db_file, 0o777)
conn = sqlite3.connect(backup_db_file)
id_path_table = build_id_table(conn.cursor())
conn.text_factory = lambda x: x.decode("utf-8", "ignore")
# Delete the blackboard tables
conn.execute("DROP TABLE blackboard_artifacts")
conn.execute("DROP TABLE blackboard_attributes")
# Write to the database dump
with codecs.open(dump_file, "wb", "utf_8") as db_log:
for line in conn.iterdump():
line = normalize_db_entry(line, id_path_table)
db_log.write('%s\n' % line)
# Now sort the file
srtcmdlst = ["sort", dump_file, "-o", dump_file]
subprocess.call(srtcmdlst)
conn.close()
# cleanup the backup
os.remove(backup_db_file)
def dump_output_db(db_file, dump_file, bb_dump_file):
"""Dumps the given database to text files for later comparison.
Args:
db_file: a pathto_File, the database file to dump
dump_file: a pathto_File, the location to dump the non-blackboard database items
bb_dump_file: a pathto_File, the location to dump the blackboard database items
"""
TskDbDiff._dump_output_db_nonbb(db_file, dump_file)
TskDbDiff._dump_output_db_bb(db_file, bb_dump_file)
def _get_tmp_file(base, ext):
time = datetime.datetime.now().time().strftime("%H%M%f")
return os.path.join(os.environ['TMP'], base + time + ext)
class TskDbDiffException(Exception):
pass
def normalize_db_entry(line, table):
""" Make testing more consistent and reasonable by doctoring certain db entries.
Args:
line: a String, the line to remove the object id from.
table: a map from object ids to file paths.
"""
files_index = line.find('INSERT INTO "tsk_files"')
path_index = line.find('INSERT INTO "tsk_files_path"')
object_index = line.find('INSERT INTO "tsk_objects"')
report_index = line.find('INSERT INTO "reports"')
layout_index = line.find('INSERT INTO "tsk_file_layout"')
parens = line[line.find('(') + 1 : line.find(')')]
fields_list = parens.replace(" ", "").split(',')
# remove object ID
if (files_index != -1):
obj_id = fields_list[0]
path = table[int(obj_id)]
newLine = ('INSERT INTO "tsk_files" VALUES(' + ', '.join(fields_list[1:]) + ');')
return newLine
# remove object ID
elif (path_index != -1):
obj_id = fields_list[0]
path = table[int(obj_id)]
newLine = ('INSERT INTO "tsk_files_path" VALUES(' + path + ', '.join(fields_list[1:]) + ');')
return newLine
# remove object ID
elif (layout_index != -1):
obj_id = fields_list[0]
path= table[int(obj_id)]
newLine = ('INSERT INTO "tsk_file_layout" VALUES(' + path + ', '.join(fields_list[1:]) + ');')
return newLine
# remove object ID
elif (object_index != -1):
obj_id = fields_list[0]
parent_id = fields_list[1]
try:
path = table[int(obj_id)]
parent_path = table[int(parent_id)]
newLine = ('INSERT INTO "tsk_objects" VALUES(' + path + ', ' + parent_path + ', ' + ', '.join(fields_list[2:]) + ');')
return newLine
except Exception as e:
# objects table has things that aren't files. if lookup fails, don't replace anything.
return line
# remove time-based information, ie Test_6/11/14 -> Test
elif (report_index != -1):
fields_list[1] = "AutopsyTestCase"
fields_list[2] = "0"
newLine = ('INSERT INTO "reports" VALUES(' + ','.join(fields_list) + ');')
return newLine
else:
return line
def getAssociatedArtifactType(db_file, artifact_id):
# Make a copy that we can modify
backup_db_file = TskDbDiff._get_tmp_file("tsk_backup_db", ".db")
shutil.copy(db_file, backup_db_file)
# We sometimes get situations with messed up permissions
os.chmod (backup_db_file, 0o777)
conn = sqlite3.connect(backup_db_file)
cur = conn.cursor()
#artifact_cursor.execute("SELECT display_name FROM blackboard_artifact_types WHERE artifact_id=?",[artifact_id])
cur.execute("SELECT tsk_files.parent_path, blackboard_artifact_types.display_name FROM blackboard_artifact_types INNER JOIN blackboard_artifacts ON blackboard_artifact_types.artifact_type_id = blackboard_artifacts.artifact_type_id INNER JOIN tsk_files ON tsk_files.obj_id = blackboard_artifacts.obj_id WHERE artifact_id=?",[artifact_id])
info = cur.fetchone()
conn.close()
# cleanup the backup
os.remove(backup_db_file)
return "File path: " + info[0] + " Artifact Type: " + info[1]
def build_id_table(artifact_cursor):
"""Build the map of object ids to file paths.
Args:
artifact_cursor: the database cursor
"""
# for each row in the db, take the object id, parent path, and name, then create a tuple in the dictionary
# with the object id as the key and the full file path (parent + name) as the value
mapping = dict([(row[0], str(row[1]) + str(row[2])) for row in artifact_cursor.execute("SELECT obj_id, parent_path, name FROM tsk_files")])
return mapping
def main():
try:
sys.argv.pop(0)
output_db = sys.argv.pop(0)
gold_db = sys.argv.pop(0)
except:
print("usage: tskdbdiff [OUPUT DB PATH] [GOLD DB PATH]")
sys.exit(1)
db_diff = TskDbDiff(output_db, gold_db, output_dir=".")
dump_passed, bb_dump_passed = db_diff.run_diff()
if dump_passed and bb_dump_passed:
print("Database comparison passed.")
if not dump_passed:
print("Non blackboard database comparison failed.")
if not bb_dump_passed:
print("Blackboard database comparison failed.")
sys.exit(0)
if __name__ == "__main__":
if sys.hexversion < 0x03000000:
print("Python 3 required")
sys.exit(1)
main()