|
| 1 | +from fuzzeddataprovider import FuzzedDataProvider |
| 2 | +import pickle |
| 3 | +import io |
| 4 | + |
| 5 | +MAX_CONTAINER_SIZE = 200 # cap on generated container/string sizes to avoid OOM |
| 6 | + |
| 7 | +# Top-level operation constants for FuzzerRunOne dispatch |
| 8 | +OP_DUMPS = 0 |
| 9 | +OP_LOADS = 1 |
| 10 | +OP_PICKLER = 2 |
| 11 | +OP_ROUNDTRIP = 3 |
| 12 | + |
| 13 | +# Container type constants for build_container |
| 14 | +CTYPE_BYTES = 0 |
| 15 | +CTYPE_STRING = 1 |
| 16 | +CTYPE_INT_LIST = 2 |
| 17 | +CTYPE_TUPLE = 3 |
| 18 | +CTYPE_SET = 4 |
| 19 | +CTYPE_FROZENSET = 5 |
| 20 | +CTYPE_BYTEARRAY = 6 |
| 21 | +CTYPE_DICT = 7 |
| 22 | + |
| 23 | +# Unpickler variant constants for op_loads |
| 24 | +VARIANT_RESTRICTED = 0 |
| 25 | +VARIANT_PERSISTENT = 1 |
| 26 | +VARIANT_RESTRICTED_FIX_IMPORTS = 2 |
| 27 | + |
| 28 | + |
| 29 | +class RestrictedUnpickler(pickle.Unpickler): |
| 30 | + def find_class(self, module, name): |
| 31 | + raise pickle.UnpicklingError("restricted") |
| 32 | + |
| 33 | + |
| 34 | +class PersistentUnpickler(pickle.Unpickler): |
| 35 | + def persistent_load(self, pid): |
| 36 | + return pid |
| 37 | + |
| 38 | + def find_class(self, module, name): |
| 39 | + raise pickle.UnpicklingError("restricted") |
| 40 | + |
| 41 | + |
| 42 | +def build_container(fdp, ctype): |
| 43 | + n = fdp.ConsumeIntInRange(0, min(fdp.remaining_bytes(), MAX_CONTAINER_SIZE)) |
| 44 | + if ctype == CTYPE_BYTES: |
| 45 | + return fdp.ConsumeBytes(n) |
| 46 | + elif ctype == CTYPE_STRING: |
| 47 | + return fdp.ConsumeUnicode(n) |
| 48 | + elif ctype == CTYPE_INT_LIST: |
| 49 | + return fdp.ConsumeIntList(n, 1) |
| 50 | + elif ctype == CTYPE_TUPLE: |
| 51 | + return tuple(fdp.ConsumeIntList(n, 1)) |
| 52 | + elif ctype == CTYPE_SET: |
| 53 | + return set(fdp.ConsumeIntList(n, 1)) |
| 54 | + elif ctype == CTYPE_FROZENSET: |
| 55 | + return frozenset(fdp.ConsumeIntList(n, 1)) |
| 56 | + elif ctype == CTYPE_BYTEARRAY: |
| 57 | + return bytearray(fdp.ConsumeBytes(n)) |
| 58 | + elif ctype == CTYPE_DICT: |
| 59 | + d = {} |
| 60 | + entries = fdp.ConsumeIntInRange(0, min(n, 64)) |
| 61 | + for _ in range(entries): |
| 62 | + if fdp.remaining_bytes() == 0: |
| 63 | + break |
| 64 | + kn = fdp.ConsumeIntInRange(1, 20) |
| 65 | + key = fdp.ConsumeUnicode(kn) |
| 66 | + val = fdp.ConsumeRandomValue() |
| 67 | + d[key] = val |
| 68 | + return d |
| 69 | + return fdp.ConsumeBytes(n) |
| 70 | + |
| 71 | + |
| 72 | +def op_dumps(fdp): |
| 73 | + ctype = fdp.ConsumeIntInRange(CTYPE_BYTES, CTYPE_DICT) |
| 74 | + protocol = fdp.ConsumeIntInRange(0, 5) |
| 75 | + fix_imports = fdp.ConsumeBool() |
| 76 | + obj = build_container(fdp, ctype) |
| 77 | + pickle.dumps(obj, protocol=protocol, fix_imports=fix_imports) |
| 78 | + |
| 79 | + |
| 80 | +def op_loads(fdp): |
| 81 | + variant = fdp.ConsumeIntInRange(VARIANT_RESTRICTED, VARIANT_RESTRICTED_FIX_IMPORTS) |
| 82 | + data = fdp.ConsumeBytes(fdp.remaining_bytes()) |
| 83 | + bio = io.BytesIO(data) |
| 84 | + if variant == VARIANT_RESTRICTED: |
| 85 | + unpickler = RestrictedUnpickler(bio) |
| 86 | + elif variant == VARIANT_PERSISTENT: |
| 87 | + unpickler = PersistentUnpickler(bio) |
| 88 | + else: |
| 89 | + unpickler = RestrictedUnpickler(bio, fix_imports=True, encoding="bytes") |
| 90 | + unpickler.load() |
| 91 | + |
| 92 | + |
| 93 | +def op_pickler(fdp): |
| 94 | + protocol = fdp.ConsumeIntInRange(0, 5) |
| 95 | + n = ( |
| 96 | + fdp.ConsumeIntInRange(1, min(fdp.remaining_bytes(), MAX_CONTAINER_SIZE)) |
| 97 | + if fdp.remaining_bytes() > 0 |
| 98 | + else 0 |
| 99 | + ) |
| 100 | + if n == 0: |
| 101 | + return |
| 102 | + obj1 = fdp.ConsumeIntList(n, 1) |
| 103 | + s = fdp.ConsumeUnicode(fdp.ConsumeIntInRange(0, MAX_CONTAINER_SIZE)) |
| 104 | + bio = io.BytesIO() |
| 105 | + p = pickle.Pickler(bio, protocol) |
| 106 | + p.dump(obj1) |
| 107 | + p.clear_memo() |
| 108 | + p.dump(s) |
| 109 | + bio.getvalue() |
| 110 | + |
| 111 | + |
| 112 | +def op_roundtrip(fdp): |
| 113 | + ctype = fdp.ConsumeIntInRange(CTYPE_BYTES, CTYPE_DICT) |
| 114 | + obj = build_container(fdp, ctype) |
| 115 | + dumped = pickle.dumps(obj) |
| 116 | + pickle.loads(dumped) |
| 117 | + |
| 118 | + |
| 119 | +# Fuzzes the _pickle C module (Modules/_pickle.c). Exercises pickle.dumps() |
| 120 | +# with protocols 0-5 on various container types (bytes, strings, int lists, |
| 121 | +# tuples, sets, frozensets, bytearrays, dicts), pickle.loads() with |
| 122 | +# restricted and persistent-load unpickler variants, Pickler.dump() with |
| 123 | +# memo clearing, and dumps/loads roundtrips. |
| 124 | +def FuzzerRunOne(FuzzerInput): |
| 125 | + if len(FuzzerInput) < 1 or len(FuzzerInput) > 0x100000: |
| 126 | + return |
| 127 | + fdp = FuzzedDataProvider(FuzzerInput) |
| 128 | + op = fdp.ConsumeIntInRange(OP_DUMPS, OP_ROUNDTRIP) |
| 129 | + try: |
| 130 | + if op == OP_DUMPS: |
| 131 | + op_dumps(fdp) |
| 132 | + elif op == OP_LOADS: |
| 133 | + op_loads(fdp) |
| 134 | + elif op == OP_PICKLER: |
| 135 | + op_pickler(fdp) |
| 136 | + else: |
| 137 | + op_roundtrip(fdp) |
| 138 | + except Exception: |
| 139 | + pass |
0 commit comments