def token_mutate_series_any_fix(prog, max_num_mutations, num_mutated_progs, include_kind=False):
assert max_num_mutations > 0 and num_mutated_progs > 0, "Invalid argument(s) supplied to the function token_mutate"
global mutator_obj
corrupt_fix_pair = set()
for _ in range(num_mutated_progs):
num_mutations = random.choice(range(max_num_mutations)) + 1
this_corrupted = prog
lines = set()
mutation_count = 0
loop_counter = 0
loop_count_threshold = 50
if include_kind:
fix_kinds = {}
while(mutation_count < num_mutations):
loop_counter += 1
if loop_counter == loop_count_threshold:
print "mutation_count", mutation_count
raise LoopCountThresholdExceededException
line = None
if include_kind:
this_corrupted, fix, line, kind = mutator_obj.easy_mutate2(prog, this_corrupted, include_kind=True)
else:
this_corrupted, fix, line = mutator_obj.easy_mutate2(prog, this_corrupted)
if line is not None:
fix = fetch_line(prog, line)
corrupt_line = fetch_line(this_corrupted, line)
if fix != corrupt_line:
lines.add(line)
mutation_count += 1
if include_kind:
if str(line) not in fix_kinds:
fix_kinds[str(line)] = [kind]
else:
fix_kinds[str(line)].append(kind)
assert len(lines) > 0, "Could not mutate!"
flag_empty_line_in_corrupted = False
for _line_ in get_lines(this_corrupted):
if _line_.strip() == '':
flag_empty_line_in_corrupted = True
break
if flag_empty_line_in_corrupted:
continue
lines = sorted(lines)
ranked_lines = map(lambda x:(x,lines.index(x)+1), lines)
random.shuffle(ranked_lines)
random.shuffle(lines)
for line, fix_number in ranked_lines:
fix = fetch_line(prog, line)
corrupt_line = fetch_line(this_corrupted, line)
assert len(fetch_line(prog, line, include_line_number=False).strip()) != 0, "empty fix"
assert len(fetch_line(this_corrupted, line, include_line_number=False).strip()) != 0, "empty corrupted line"
if fix != corrupt_line:
if include_kind:
if len(fix_kinds[str(line)]) == 1: # remove later
for kind in fix_kinds[str(line)]:
corrupt_fix_pair.add((this_corrupted, fix, fix_number, kind))
else:
corrupt_fix_pair.add((this_corrupted, fix, fix_number))
try:
this_corrupted = do_fix_at_line(this_corrupted, line, fetch_line(prog, line, include_line_number=False))
except IndexError:
raise
if include_kind:
return map( lambda (w,x,y,z):(add_fix_number(w, y), x, z), list(corrupt_fix_pair))
else:
return map( lambda (w,x,y):(add_fix_number(w, y), x), list(corrupt_fix_pair))
def token_mutate_for_tsne_with_specific_errors(prog, num_mutations, action, include_kind=False):
assert num_mutations > 0, "Invalid argument(s) supplied to the function token_mutate"
global mutator_obj
specific_mutate = mutator_obj.specific_mutate
corrupt_fix_pair = set()
for _ in range(1):
this_corrupted = prog
lines = set()
mutation_count = 0
loop_counter = 0
loop_count_threshold = 50
if include_kind:
fix_kinds = {}
while(mutation_count < num_mutations):
loop_counter += 1
if loop_counter == loop_count_threshold:
print "mutation_count", mutation_count
raise LoopCountThresholdExceededException
line = None
if include_kind:
this_corrupted, fix, line, kind = specific_mutate(prog, this_corrupted, action, include_kind=True)
else:
this_corrupted, fix, line = specific_mutate(prog, this_corrupted, action)
if line is not None:
fix = fetch_line(prog, line)
corrupt_line = fetch_line(this_corrupted, line)
if fix != corrupt_line:
lines.add(line)
mutation_count += 1
if include_kind:
if str(line) not in fix_kinds:
fix_kinds[str(line)] = [kind]
else:
fix_kinds[str(line)].append(kind)
assert len(lines) > 0, "Could not mutate!"
empty_line_in_corrupted = False
for _line_ in get_lines(this_corrupted):
if _line_.strip() == '':
empty_line_in_corrupted = True
break
if empty_line_in_corrupted:
continue
sorted_lines = sorted(lines)
for line in sorted_lines:
fix = fetch_line(prog, line)
corrupt_line = fetch_line(this_corrupted, line)
assert len(fetch_line(prog, line, include_line_number=False).strip()) != 0, "empty fix"
assert len(fetch_line(this_corrupted, line, include_line_number=False).strip()) != 0, "empty corrupted line"
if fix != corrupt_line:
corrupt_fix_pair.add((this_corrupted, fix))
break
return list(corrupt_fix_pair)
def typo_mutate(mutator_obj,
prog,
max_num_mutations,
num_mutated_progs,
just_one=False):
assert len(
prog
) > 10 and max_num_mutations > 0 and num_mutated_progs > 0, "Invalid argument(s) supplied to the function token_mutate_series_network2"
corrupt_fix_pair = set()
for _ in range(num_mutated_progs):
num_mutations = mutator_obj.rng.choice(
range(max_num_mutations)) + 1 if max_num_mutations > 1 else 1
this_corrupted = prog
lines = set()
mutation_count = 0
loop_counter = 0
loop_count_threshold = 50
mutations = {}
while mutation_count < num_mutations:
loop_counter += 1
if loop_counter == loop_count_threshold:
print("mutation_count", mutation_count)
raise LoopCountThresholdExceededException
line = None
this_corrupted, line, mutation_name = mutator_obj.easy_mutate(
this_corrupted) # line is line_number here!
if line is not None:
fix = fetch_line(prog, line)
corrupt_line = fetch_line(this_corrupted, line)
if fix != corrupt_line:
lines.add(line)
mutation_count += 1
if line not in mutations:
mutations[line] = [mutation_name]
else:
mutations[line].append(mutation_name)
assert len(lines) > 0, "Could not mutate!"
flag_empty_line_in_corrupted = False
for _line_ in get_lines(this_corrupted):
if _line_.strip() == '':
flag_empty_line_in_corrupted = True
break
if flag_empty_line_in_corrupted:
continue
sorted_lines = sorted(lines)
for line in sorted_lines:
fix = fetch_line(prog, line)
corrupt_line = fetch_line(this_corrupted, line)
assert len(
fetch_line(
prog, line,
include_line_number=False).strip()) != 0, "empty fix"
assert len(
fetch_line(this_corrupted, line, include_line_number=False).
strip()) != 0, "empty corrupted line"
if fix != corrupt_line:
corrupt_fix_pair.add((this_corrupted, fix))
mutator_obj.update_mutation_distribution(mutations[line])
if just_one:
break
try:
this_corrupted = do_fix_at_line(
this_corrupted, line,
fetch_line(prog, line, include_line_number=False))
except IndexError:
raise
if len(corrupt_fix_pair) > 0:
mutator_obj.update_pmf()
return list(corrupt_fix_pair)