Unfortunately, I currently don't have the time to maintain or update this project. While there is a chance that I might revive it at some point, you really shoudldn't count on that. Please feel free to fork or just take anything that looks useful!
Python module capable of decoding and disassembling WebAssembly modules and bytecode, according to the MVP specification of the WASM binary format.
As there is no official text format defined yet, the text format
implemented doesn't correspond to any existing definition and is a
simple mnemonic op1, op2, ...
format. Functions are formatted in a
way similar to how Google Chrome does in the debug console.
# From PyPi
pip install wasm
# From GitHub
pip install git+https://github.com/athre0z/wasm.git
Parsing a WASM module, printing the types of sections found.
from wasm import decode_module
with open('input-samples/hello/hello.wasm', 'rb') as raw:
raw = raw.read()
mod_iter = iter(decode_module(raw))
header, header_data = next(mod_iter)
for cur_sec, cur_sec_data in mod_iter:
print(cur_sec_data.get_decoder_meta()['types']['payload'])
Possible output:
<wasm.modtypes.TypeSection object at 0x10dec52e8>
<wasm.modtypes.ImportSection object at 0x10dec5320>
<wasm.modtypes.FunctionSection object at 0x10dec5358>
<wasm.modtypes.GlobalSection object at 0x10dec5400>
<wasm.modtypes.ExportSection object at 0x10dec5438>
<wasm.modtypes.ElementSection object at 0x10dec54a8>
<wasm.modtypes.CodeSection object at 0x10dec54e0>
<wasm.modtypes.DataSection object at 0x10dec5518>
Parsing specific sections (eg. GlobalSection, ElementSection, DataSection) in WASM module, printing each section's content:
from wasm import (
decode_module,
format_instruction,
format_lang_type,
format_mutability,
SEC_DATA,
SEC_ELEMENT,
SEC_GLOBAL,
)
with open('input-samples/hello/hello.wasm', 'rb') as raw:
raw = raw.read()
mod_iter = iter(decode_module(raw))
header, header_data = next(mod_iter)
for cur_sec, cur_sec_data in mod_iter:
if cur_sec_data.id == SEC_GLOBAL:
print("GlobalSection:")
for idx, entry in enumerate(cur_sec_data.payload.globals):
print(
format_mutability(entry.type.mutability),
format_lang_type(entry.type.content_type),
)
for cur_insn in entry.init:
print(format_instruction(cur_insn))
if cur_sec_data.id == SEC_ELEMENT:
print("ElementSection:")
for idx, entry in enumerate(cur_sec_data.payload.entries):
print(entry.index, entry.num_elem, entry.elems)
for cur_insn in entry.offset:
print(format_instruction(cur_insn))
if cur_sec_data.id == SEC_DATA:
print("DataSection:")
for idx, entry in enumerate(cur_sec_data.payload.entries):
print(entry.index, entry.size, entry.data.tobytes())
for cur_insn in entry.offset:
print(format_instruction(cur_insn))
Output:
GlobalSection:
mut i32
get_global 0
end
mut i32
get_global 1
end
...
mut f32
f32.const 0x0
end
mut f32
f32.const 0x0
end
ElementSection:
0 12576 [856, 856, 856, ..., 888]
i32.const 0
end
DataSection:
0 16256 b'\x98&\x00\x00\xfe4\x00\x00\x10\x04\x00\x00\x00...\x00N10__cxxabiv121__vmi_class_type_infoE'
get_global 8
end
Manually disassemble WASM bytecode, printing each instruction.
from wasm import (
decode_bytecode,
format_instruction,
INSN_ENTER_BLOCK,
INSN_LEAVE_BLOCK,
)
raw = bytearray([2, 127, 65, 24, 16, 28, 65, 0, 15, 11])
indent = 0
for cur_insn in decode_bytecode(raw):
if cur_insn.op.flags & INSN_LEAVE_BLOCK:
indent -= 1
print(' ' * indent + format_instruction(cur_insn))
if cur_insn.op.flags & INSN_ENTER_BLOCK:
indent += 1
Output:
block -1
i32.const 24
call 28
i32.const 0
return
end
The module also comes with a simple command-line tool called wasmdump
,
dumping all module struct in sexy tree format. Optionally, it also
disassembles all functions found when invoked with --disas
(slow).
The library was successfully tested on Python 2.7, Python 3.7 and PyPy 5.4.