dwarffi 0.0.4

Basic Python project scaffold for dwarffi.

dwarffi

An authentic, DWARF-powered type interface for Python.

dwarffi allows you to interact with memory layouts using Intermediate Symbol Files (ISF) generated from DWARF debug information. It provides a CFFI-like experience without requiring header files---instead, it uses the actual compiled structures from your binaries.

Intermediate Symbol Files (ISF) are JSON representations of DWARF type information. They capture the exact memory layout, including padding and alignment, as defined in the original C code and compiled into the binary. You can generate ISF files using the dwarf2json tool, which extracts DWARF info from ELF binaries.

Read more about dwarf2json and ISF in the dwarf2json README.

---

🚀 Features

• DWARF-Native: No need to rewrite C headers. Just point to a .json or .json.xz ISF file (generated by dwarf2json).
• Architecture Agnostic: Handles Big Endian (PowerPC/MIPS) and Little Endian (ARM/x86) transparently.
• Dynamic cdef: Compile C code on the fly to generate types, with automatic debug-type retention to prevent compilers from stripping unused definitions.
• Recursive Typedef Handling: Automatic resolution and decay of typedef chains.
• C-Style Magic:
  • Pointer arithmetic (ptr + 5)
  • Pointer subtraction (ptr2 - ptr1)
  • Array slicing (arr[1:5])
  • Deep struct initialization via nested dictionaries
• Safety Semantics:
  • Automatic bit-masking
  • Sign extension
  • C-style integer overflow/underflow behavior
• Anonymous Struct/Union Flattening: Access anonymous union members directly (ideal for register maps).
• ISF Export Support: Save dynamically generated ISFs to .json or .json.xz.

---

📦 Installation

pip install dwarffi

Requirements for cdef()

To use dynamic compilation:

• A C compiler (gcc, clang, or cross-compiler)
• dwarf2json available in your PATH

NOTE: For some compilers you may need to disable options for optimizing away unused debug types. For example, with gcc you can use -fno-eliminate-unused-debug-types.

---

🛠️ Quick Start

1️⃣ CFFI-style cdef

from dwarffi import DFFI

ffi = DFFI()
ffi.cdef("""
    struct sensor_data {
        uint32_t timestamp;
        int16_t  readings[3];
        uint8_t  status;
    };
""")

Create an instance:

sensor = ffi.new("struct sensor_data", {
    "timestamp": 1234567,
    "readings": [10, -5, 20],
    "status": 0x01
})

print(f"Bytes: {sensor.to_bytes().hex()}")
print(f"Reading[1]: {sensor.readings[1]}")  # -5

---

2️⃣ Embedded Architecture Analysis

ffi = DFFI("mips_be_target.json")

raw_data = bytearray(b'\x00\x00\x12\x34\xff\xff\xff\xfb')
cpu_ctx = ffi.from_buffer("struct cpu_context", raw_data)

print(hex(cpu_ctx.pc))      # 0x1234
print(cpu_ctx.status)       # -5 (Sign-extended)

---

🧩 Advanced Usage

Anonymous Unions

ffi.cdef("""
struct reg_map {
    union {
        uint32_t ALL;
        struct {
            uint16_t LOW;
            uint16_t HIGH;
        };
    };
};
""")

reg = ffi.new("struct reg_map")
reg.ALL = 0x12345678
print(hex(reg.HIGH))  # 0x1234

---

Pointer Arithmetic

ptr = ffi.cast("int *", 0x4000)
next_ptr = ptr + 1
print(hex(next_ptr.address))

---

⚙️ How It Works

dwarffi operates in three phases:

1. Parsing

Reads ISF JSON files that represent the DWARF type tree.

2. Type Synthesis

Builds Python representations of:

• Base types
• Structs
• Unions
• Enums
• Typedef chains
• Arrays
• Pointers

3. Memory Mapping

Uses Python's struct.pack and struct.unpack under the hood to:

• Convert Python integers into architecture-accurate byte layouts
• Apply endianness rules
• Respect alignment and padding

Instances are bound views into bytearrays. Field access directly reads/writes into the underlying buffer.

---

📚 Core API Reference

DFFI()

Create an empty instance or load an ISF file.

cdef(source, compiler="gcc", save_isf_to=None)

Compile C source → DWARF → ISF → load into current FFI.

new(ctype, init=None)

Allocate a new instance of a C type.

from_buffer(ctype, buffer)

Bind a type to existing memory.

sizeof(ctype)

Return size in bytes.

offsetof(ctype, field)

Return byte offset of a field.

cast(ctype, value)

Reinterpret memory or create pointer instances.

addressof(instance)

Return pointer to instance.

---

🤝 Contributing

Contributions are welcome!

python -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"
pytest