Adding a symbol resolver to the Python bindings.

Author: MIvanchev

src/rust/iced-x86-py/src/formatter.rs

@@ -54,19 +54,49 @@ pub(crate) struct Formatter {
 unsafe impl Send for Formatter {}
 unsafe impl Sync for Formatter {}
 
+struct SymbolResolverAdapter {
+	resolver: PyObject,
+}
+
+impl SymbolResolverAdapter {
+	fn new(resolver: PyObject) -> SymbolResolverAdapter {
+		SymbolResolverAdapter { resolver }
+	}
+}
+
+impl iced_x86::SymbolResolver for SymbolResolverAdapter {
+	fn symbol(
+		&mut self, _instruction: &iced_x86::Instruction, _operand: u32, _instruction_operand: Option<u32>, address: u64, _address_size: u32,
+	) -> Option<iced_x86::SymbolResult<'_>> {
+		let instr = Instruction { instr: *_instruction };
+		Python::with_gil(|py| {
+			let sym: Option<String> =
+				self.resolver.call1(py, (instr, _operand, _instruction_operand, address, _address_size)).unwrap().extract(py).unwrap();
+
+			if let Some(val) = sym {
+				Some(iced_x86::SymbolResult::with_string(address, val))
+			} else {
+				None
+			}
+		})
+	}
+}
+
 #[pymethods]
 impl Formatter {
 	#[new]
-	#[pyo3(text_signature = "(syntax)")]
-	fn new(syntax: u32) -> PyResult<Self> {
+	#[pyo3(signature = (syntax, symbol_resolver = None))]
+	fn new(syntax: u32, symbol_resolver: Option<PyObject>) -> PyResult<Self> {
+		let resolver = symbol_resolver.map(|val| -> Box<dyn iced_x86::SymbolResolver> { Box::new(SymbolResolverAdapter::new(val)) });
+
 		let formatter: Box<dyn iced_x86::Formatter> = if syntax == FormatterSyntax::Gas as u32 {
-			Box::new(iced_x86::GasFormatter::new())
+			Box::new(iced_x86::GasFormatter::with_options(resolver, None))
 		} else if syntax == FormatterSyntax::Intel as u32 {
-			Box::new(iced_x86::IntelFormatter::new())
+			Box::new(iced_x86::IntelFormatter::with_options(resolver, None))
 		} else if syntax == FormatterSyntax::Masm as u32 {
-			Box::new(iced_x86::MasmFormatter::new())
+			Box::new(iced_x86::MasmFormatter::with_options(resolver, None))
 		} else if syntax == FormatterSyntax::Nasm as u32 {
-			Box::new(iced_x86::NasmFormatter::new())
+			Box::new(iced_x86::NasmFormatter::with_options(resolver, None))
 		} else {
 			return Err(PyValueError::new_err("Invalid formatter syntax"));
 		};

src/rust/iced-x86-py/tests/formatter_gas_test.py

@@ -115,3 +115,19 @@ def test_format() -> None:
 	assert formatter.format_u16(0x89AB) == "0x89AB"
 	assert formatter.format_u32(0x89ABCDEF) == "0x89ABCDEF"
 	assert formatter.format_u64(0xFEDCBA9876543210) == "0xFEDCBA9876543210"
+
+def test_format_with_symbol_resolver():
+	def symbol_resolver(instr, op, instr_op, addr, addr_size):
+		if addr == 0x00:
+			return "foo"
+		elif addr == 0x0B:
+			return "boo"
+		else:
+			return None
+
+	instr1, _, _, instr4, _, _, instr7 = list(Decoder(32, b"\xE8\x06\x00\x00\x00\x31\xDB\xF7\xE2\xEB\xF5\x83\xC0\x0E\xC3\x8D\x81\x0B\x00\x00\x00"))
+	formatter = Formatter(FORMATTER_SYNTAX, symbol_resolver)
+
+	assert formatter.format(instr1) == "call boo"
+	assert formatter.format(instr4) == "jmp foo"
+	assert formatter.format(instr7) == "lea boo(%ecx),%eax"

src/rust/iced-x86-py/tests/formatter_intel_test.py

@@ -115,3 +115,19 @@ def test_format() -> None:
 	assert formatter.format_u16(0x89AB) == "89ABh"
 	assert formatter.format_u32(0x89ABCDEF) == "89ABCDEFh"
 	assert formatter.format_u64(0xFEDCBA9876543210) == "0FEDCBA9876543210h"
+
+def test_format_with_symbol_resolver():
+	def symbol_resolver(instr, op, instr_op, addr, addr_size):
+		if addr == 0x00:
+			return "foo"
+		elif addr == 0x0B:
+			return "boo"
+		else:
+			return None
+
+	instr1, _, _, instr4, _, _, instr7 = list(Decoder(32, b"\xE8\x06\x00\x00\x00\x31\xDB\xF7\xE2\xEB\xF5\x83\xC0\x0E\xC3\x8D\x81\x0B\x00\x00\x00"))
+	formatter = Formatter(FORMATTER_SYNTAX, symbol_resolver)
+
+	assert formatter.format(instr1) == "call boo"
+	assert formatter.format(instr4) == "jmp short foo"
+	assert formatter.format(instr7) == "lea eax,[ecx+boo]"

src/rust/iced-x86-py/tests/formatter_masm_test.py

@@ -115,3 +115,19 @@ def test_format() -> None:
 	assert formatter.format_u16(0x89AB) == "89ABh"
 	assert formatter.format_u32(0x89ABCDEF) == "89ABCDEFh"
 	assert formatter.format_u64(0xFEDCBA9876543210) == "0FEDCBA9876543210h"
+
+def test_format_with_symbol_resolver():
+	def symbol_resolver(instr, op, instr_op, addr, addr_size):
+		if addr == 0x00:
+			return "foo"
+		elif addr == 0x0B:
+			return "boo"
+		else:
+			return None
+
+	instr1, _, _, instr4, _, _, instr7 = list(Decoder(32, b"\xE8\x06\x00\x00\x00\x31\xDB\xF7\xE2\xEB\xF5\x83\xC0\x0E\xC3\x8D\x81\x0B\x00\x00\x00"))
+	formatter = Formatter(FORMATTER_SYNTAX, symbol_resolver)
+
+	assert formatter.format(instr1) == "call boo"
+	assert formatter.format(instr4) == "jmp short foo"
+	assert formatter.format(instr7) == "lea eax,[ecx+boo]"

src/rust/iced-x86-py/tests/formatter_nasm_test.py

@@ -115,3 +115,19 @@ def test_format() -> None:
 	assert formatter.format_u16(0x89AB) == "89ABh"
 	assert formatter.format_u32(0x89ABCDEF) == "89ABCDEFh"
 	assert formatter.format_u64(0xFEDCBA9876543210) == "0FEDCBA9876543210h"
+
+def test_format_with_symbol_resolver():
+	def symbol_resolver(instr, op, instr_op, addr, addr_size):
+		if addr == 0x00:
+			return "foo"
+		elif addr == 0x0B:
+			return "boo"
+		else:
+			return None
+
+	instr1, _, _, instr4, _, _, instr7 = list(Decoder(32, b"\xE8\x06\x00\x00\x00\x31\xDB\xF7\xE2\xEB\xF5\x83\xC0\x0E\xC3\x8D\x81\x0B\x00\x00\x00"))
+	formatter = Formatter(FORMATTER_SYNTAX, symbol_resolver)
+
+	assert formatter.format(instr1) == "call boo"
+	assert formatter.format(instr4) == "jmp short foo"
+	assert formatter.format(instr7) == "lea eax,[ecx+boo]"