Loop based (extensible) EOF1 validator

eof1_validator_loop.py

FORMAT = 0xEF
MAGIC = 0xFF # To be defined
VERSION = 0x01
S_TERMINATOR = 0x00
S_CODE = 0x01
S_DATA = 0x02

def validate_code(code: bytes) -> bool:
    # Old-style contracts are still allowed
    if len(code) == 0:
        return True
    if code[0] != FORMAT:
        return True
    if len(code) < 2:
        return True
    if code[0] == FORMAT and code[1] != MAGIC:
        return True

    # Format and magic is validated
    if len(code) < 3:
        return False
    if code[2] != VERSION:
        return False
    # No sections
    if len(code) < 4:
        return False

    # Process section headers
    total_content_size = 0
    pos = 3
    while True:
        if (pos + 1) > len(code):
            return False
        section_id = code[pos]
        pos += 1
        if section_id == S_TERMINATOR:
            break
        if section_id != S_CODE and section_id != S_DATA:
            return False

        if (pos + 2) > len(code):
            return False
        section_size = (code[pos] << 8) | code[pos + 1]
        pos += 2
        if section_size == 0:
            return False

        total_content_size += section_size
    # First section is not the code section
    if code[3] != S_CODE:
        return False
    
    return len(code) == (pos + total_content_size)

# Legacy contracts
assert validate_code(b'') == True
assert validate_code(b'\x00') == True
assert validate_code(b'\xef') == True
# Any value outside the magic
for magic in range(254):
    assert validate_code(['\xef', magic]) == True
# EOF1 contracts
assert validate_code(b'\xef\xff') == False # Only magic
assert validate_code(b'\xef\xff\x01') == False # Only version
assert validate_code(b'\xef\xff\x00') == False # Wrong version
assert validate_code(b'\xef\xff\x01\x00') == False # Only terminator
assert validate_code(b'\xef\xff\x01\x00\x0ff') == False # Trailing bytes
assert validate_code(b'\xef\xff\x01\x01') == False # Truncated section header
assert validate_code(b'\xef\xff\x01\x02') == False # Truncated section header
assert validate_code(b'\xef\xff\x01\x03') == False # Invalid section id
assert validate_code(b'\xef\xff\x01\x01\x00') == False # Truncated section size
assert validate_code(b'\xef\xff\x01\x01\x00\x00') == False # Empty section size
assert validate_code(b'\xef\xff\x01\x01\x00\x01') == False # No terminator after section
assert validate_code(b'\xef\xff\x01\x01\x00\x01\x00') == False # Missing section contents
assert validate_code(b'\xef\xff\x01\x01\x00\x01\x00\xfe') == True # Valid format with 1-byte of code
assert validate_code(b'\xef\xff\x01\x02\x00\x01\x00\xaa') == False # Only data section
assert validate_code(b'\xef\xff\x01\x01\x00\x01\x02\x00\x01\x00\xfe\xaa') == True # Code and data section
# TODO: add multiple code sections, multiple dat asections, multiple code+data sections