kse-02/instrument.py

from typing import Optional
import os.path

import tqdm
from frozendict import frozendict

import ast
import astunparse
import sys
import random
from deap import creator, base, tools, algorithms
from operators import compute_distances

# hyperparameters
NPOP = 300
NGEN = 200
INDMUPROB = 0.05
MUPROB = 0.1
CXPROB = 0.5
TOURNSIZE = 3
LOW = -1000
UP = 1000
REPS = 10
MAX_STRING_LENGTH = 10

ROOT_DIR: str = os.path.dirname(__file__)
IN_DIR: str = os.path.join(ROOT_DIR, 'benchmark')
OUT_DIR: str = os.path.join(ROOT_DIR, 'instrumented')
SUFFIX: str = "_instrumented"

distances_true: dict[int, int] = {}
distances_false: dict[int, int] = {}
branches: list[int] = [1, 2, 3, 4, 5]
archive_true_branches: dict[int, str] = {}
archive_false_branches: dict[int, str] = {}


class BranchTransformer(ast.NodeTransformer):
    branch_num: int
    instrumented_name: Optional[str]

    def __init__(self):
        self.branch_num = 0
        self.instrumented_name = None

    @staticmethod
    def to_instrumented_name(name: str):
        return name + SUFFIX

    @staticmethod
    def to_original_name(name: str):
        assert name.endswith(SUFFIX)
        return name[:len(name) - len(SUFFIX)]

    def visit_Assert(self, ast_node):
        # Disable recursion in asserts, i.e. do not instrument assert conditions
        # TODO: may fail if assertion calls method (which must be renamed)
        return ast_node

    def visit_Return(self, ast_node):
        # Same thing for return statements
        # TODO: may fail if return statement calls method (which must be renamed)
        return ast_node

    def visit_FunctionDef(self, ast_node):
        self.instrumented_name = ast_node.name
        ast_node.name = BranchTransformer.to_instrumented_name(ast_node.name)
        inner_node = self.generic_visit(ast_node)
        self.instrumented_name = None
        return inner_node

    def visit_Call(self, ast_node):
        if isinstance(ast_node.func, ast.Name) and ast_node.func.id == self.instrumented_name:
            ast_node.func.id = BranchTransformer.to_instrumented_name(ast_node.func.id)
        return ast_node

    def visit_Compare(self, ast_node):
        if ast_node.ops[0] in [ast.Is, ast.IsNot, ast.In, ast.NotIn]:
            return ast_node

        self.branch_num += 1
        return ast.Call(func=ast.Name("evaluate_condition", ast.Load()),
                        args=[ast.Num(self.branch_num),
                              ast.Str(ast_node.ops[0].__class__.__name__),
                              ast_node.left,
                              ast_node.comparators[0]],
                        keywords=[],
                        starargs=None,
                        kwargs=None)


def update_maps(condition_num, d_true, d_false):
    global distances_true, distances_false

    if condition_num in distances_true.keys():
        distances_true[condition_num] = min(distances_true[condition_num], d_true)
    else:
        distances_true[condition_num] = d_true

    if condition_num in distances_false.keys():
        distances_false[condition_num] = min(distances_false[condition_num], d_false)
    else:
        distances_false[condition_num] = d_false


def evaluate_condition(num, op, lhs, rhs):  # type: ignore
    if op == "In":
        if isinstance(lhs, str):
            lhs = ord(lhs)

        minimum = sys.maxsize
        for elem in rhs.keys():
            distance = abs(lhs - ord(elem))
            if distance < minimum:
                minimum = distance

        distance_true, distance_false = minimum, 1 if minimum == 0 else 0
    else:
        distance_true, distance_false = compute_distances(op, lhs, rhs)

    update_maps(num, distance_true, distance_false)

    # distance == 0 equivalent to actual test by construction
    return distance_true == 0


def normalize(x):
    return x / (1.0 + x)


def get_fitness_cgi(individual):
    x = individual[0]
    # Reset any distance values from previous executions
    global distances_true, distances_false
    global branches, archive_true_branches, archive_false_branches
    distances_true = {}
    distances_false = {}

    # TODO: fix this
    # Run the function under test
    # try:
    #    cgi_decode_instrumented(x)
    # except BaseException:
    #    pass

    # Sum up branch distances
    fitness = 0.0
    for branch in branches:
        if branch in distances_true:
            if distances_true[branch] == 0 and branch not in archive_true_branches:
                archive_true_branches[branch] = x
            if branch not in archive_true_branches:
                fitness += normalize(distances_true[branch])
    for branch in branches:
        if branch in distances_false:
            if distances_false[branch] == 0 and branch not in archive_false_branches:
                archive_false_branches[branch] = x
            if branch not in archive_false_branches:
                fitness += normalize(distances_false[branch])

    return fitness,


def random_string():
    l = random.randint(0, MAX_STRING_LENGTH)
    s = ""
    for i in range(l):
        random_character = chr(random.randrange(32, 127))
        s = s + random_character
    return s


def crossover(individual1, individual2):
    parent1 = individual1[0]
    parent2 = individual2[0]
    if len(parent1) > 1 and len(parent2) > 1:
        pos = random.randint(1, len(parent1))
        offspring1 = parent1[:pos] + parent2[pos:]
        offspring2 = parent2[:pos] + parent1[pos:]
        individual1[0] = offspring1
        individual2[0] = offspring2

    return individual1, individual2


def mutate(individual):
    chromosome = individual[0]
    mutated = chromosome[:]
    if len(mutated) > 0:
        prob = 1.0 / len(mutated)
        for pos in range(len(mutated)):
            if random.random() < prob:
                new_c = chr(random.randrange(32, 127))
                mutated = mutated[:pos] + new_c + mutated[pos + 1:]
        individual[0] = mutated
    return individual,


def generate():
    global archive_true_branches, archive_false_branches

    creator.create("Fitness", base.Fitness, weights=(-1.0,))
    creator.create("Individual", list, fitness=creator.Fitness)

    toolbox = base.Toolbox()
    toolbox.register("attr_str", random_string)
    toolbox.register("individual", tools.initRepeat, creator.Individual, toolbox.attr_str, n=1)
    toolbox.register("population", tools.initRepeat, list, toolbox.individual)
    toolbox.register("evaluate", get_fitness_cgi)
    toolbox.register("mate", crossover)
    toolbox.register("mutate", mutate)
    toolbox.register("select", tools.selTournament, tournsize=TOURNSIZE)

    coverage = []
    for i in range(REPS):
        archive_true_branches = {}
        archive_false_branches = {}
        population = toolbox.population(n=NPOP)
        algorithms.eaSimple(population, toolbox, CXPROB, MUPROB, NGEN, verbose=False)
        cov = len(archive_true_branches) + len(archive_false_branches)
        print(cov, archive_true_branches, archive_false_branches)
        coverage.append(cov)


ArgType = str
Arg = tuple[str, ArgType]
Params = frozendict[str, any]
SignatureDict = dict[str, list[Arg]]

functions: SignatureDict = {}
module_of: dict[str, str] = {}


def instrument(source_path: str, target_path: str, save_instrumented=True):
    global functions

    with open(source_path, "r") as f:
        source = f.read()

    node = ast.parse(source)
    # print(ast.dump(node, indent=2))
    BranchTransformer().visit(node)
    node = ast.fix_missing_locations(node)  # Make sure the line numbers are ok before printing

    if save_instrumented:
        with open(target_path, "w") as f:
            print(astunparse.unparse(node), file=f)

    current_module = sys.modules[__name__]
    code = compile(node, filename="<ast>", mode="exec")
    exec(code, current_module.__dict__)

    # Figure out the top level function definitions
    assert isinstance(node, ast.Module)
    top_level_f_ast: list[ast.FunctionDef] = [f for f in node.body if isinstance(f, ast.FunctionDef)]

    for f in top_level_f_ast:
        if f.name in functions:
            raise ValueError(f"Function '{f.name}' already loaded from another file")

        arg_types: list[Arg] = []

        for arg in f.args.args:
            # fetch annotation type if found else fetch none
            # noinspection PyUnresolvedReferences
            arg_type = None if arg.annotation is None else arg.annotation.id

            arg_types.append((arg.arg, arg_type))

        functions[f.name] = arg_types
        module_of[f.name] = os.path.normpath(os.path.relpath(source_path, ROOT_DIR)) \
            .replace(".py", "") \
            .replace("/", ".")


def invoke(f_name: str, f_args: Params) -> any:
    global functions

    current_module = sys.modules[__name__]

    if f_name not in functions:
        raise ValueError(f"Function '{f_name}' not loaded")

    f_args_signature = functions[f_name]
    for arg_name, arg_type in f_args_signature:
        if arg_name not in f_args:
            raise ValueError(f"Required argument '{arg_name}' not provided")

    return getattr(current_module, f_name)(**f_args)


def find_py_files(search_dir: str):
    for (cwd, dirs, files) in os.walk(search_dir):
        for file in files:
            if file.endswith(".py"):
                yield os.path.join(cwd, file)


def load_benchmark(save_instrumented=True):
    for file in tqdm.tqdm(find_py_files(IN_DIR), desc="Instrumenting"):
        instrument(file, os.path.join(OUT_DIR, os.path.basename(file)), save_instrumented=save_instrumented)


def call_statement(f_name: str, f_args: Params) -> str:
    arg_list: list[str] = []
    for k, v in f_args.items():
        if type(v) == str:
            arg_list.append(f"{k}='{v}'")  # quote strings
        else:
            arg_list.append(f"{k}={v}")

    return f"{f_name}({', '.join(arg_list)})"


if __name__ == '__main__':
    load_benchmark(save_instrumented=True)