Demangle function names recovered from DWARF

3 files changed, 50 insertions, 1 deletions

diff --git a/rust/Cargo.lock b/rust/Cargo.lock
index 7e03d311..8e18b38a 100644
--- a/rust/Cargo.lock
+++ b/rust/Cargo.lock
@@ -234,6 +234,15 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "acbf1af155f9b9ef647e42cdc158db4b64a1b61f743629225fde6f3e0be2a7c7"
 
 [[package]]
+name = "cpp_demangle"
+version = "0.4.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7e8227005286ec39567949b33df9896bcadfa6051bccca2488129f108ca23119"
+dependencies = [
+ "cfg-if",
+]
+
+[[package]]
 name = "crc32fast"
 version = "1.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
@@ -319,10 +328,12 @@ name = "dwarf_import"
 version = "0.1.0"
 dependencies = [
  "binaryninja",
+ "cpp_demangle",
  "dwarfreader",
  "gimli",
  "iset",
  "log",
+ "regex",
 ]
 
 [[package]]
diff --git a/rust/examples/dwarf/dwarf_import/Cargo.toml b/rust/examples/dwarf/dwarf_import/Cargo.toml
index 715838a7..1dfae181 100644
--- a/rust/examples/dwarf/dwarf_import/Cargo.toml
+++ b/rust/examples/dwarf/dwarf_import/Cargo.toml
@@ -13,3 +13,5 @@ binaryninja = { path = "../../../" }
 gimli = "0.28"
 log = "0.4.20"
 iset = "0.2.2"
+cpp_demangle = "0.4.3"
+regex = "1"
diff --git a/rust/examples/dwarf/dwarf_import/src/functions.rs b/rust/examples/dwarf/dwarf_import/src/functions.rs
index fabcc613..fa1755fa 100644
--- a/rust/examples/dwarf/dwarf_import/src/functions.rs
+++ b/rust/examples/dwarf/dwarf_import/src/functions.rs
@@ -12,11 +12,16 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use std::sync::OnceLock;
+
 use crate::dwarfdebuginfo::{DebugInfoBuilder, DebugInfoBuilderContext, TypeUID};
 use crate::helpers::*;
 use crate::types::get_type;
 
+use binaryninja::templatesimplifier::simplify_str_to_str;
+use cpp_demangle::DemangleOptions;
 use gimli::{constants, DebuggingInformationEntry, Reader, Unit};
+use regex::Regex;
 
 fn get_parameters<R: Reader<Offset = usize>>(
     unit: &Unit<R>,
@@ -69,11 +74,42 @@ pub(crate) fn parse_function_entry<R: Reader<Offset = usize>>(
     debug_info_builder: &mut DebugInfoBuilder,
 ) -> Option<usize> {
     // Collect function properties (if they exist in this DIE)
-    let full_name = debug_info_builder_context.get_name(unit, entry);
     let raw_name = get_raw_name(unit, entry, debug_info_builder_context);
     let return_type = get_type(unit, entry, debug_info_builder_context, debug_info_builder);
     let address = get_start_address(unit, entry, debug_info_builder_context);
     let (parameters, variable_arguments) = get_parameters(unit, entry, debug_info_builder_context, debug_info_builder);
 
+    // If we have a raw name, it might be mangled, see if we can demangle it into full_name
+    //  raw_name should contain a superset of the info we have in full_name
+    let mut full_name = None;
+    if let Some(possibly_mangled_name) = &raw_name {
+        if possibly_mangled_name.starts_with('_') {
+            static OPTIONS_MEM: OnceLock<DemangleOptions> = OnceLock::new();
+            let demangle_options = OPTIONS_MEM.get_or_init(|| {
+                DemangleOptions::new()
+                    .no_return_type()
+                    .hide_expression_literal_types()
+                    .no_params()
+            });
+
+            static ABI_REGEX_MEM: OnceLock<Regex> = OnceLock::new();
+            let abi_regex = ABI_REGEX_MEM.get_or_init(|| {
+                Regex::new(r"\[abi:v\d+\]").unwrap()
+            });
+            if let Ok(sym) = cpp_demangle::Symbol::new(possibly_mangled_name) {
+                if let Ok(demangled) = sym.demangle(demangle_options) {
+                    let cleaned = abi_regex.replace_all(&demangled, "");
+                    let simplified = simplify_str_to_str(&cleaned);
+                    full_name = Some(simplified.to_string());
+                }
+            }
+        }
+    }
+
+    // If we didn't demangle the raw name, fetch the name given
+    if full_name.is_none() {
+        full_name = debug_info_builder_context.get_name(unit, entry)
+    }
+
     debug_info_builder.insert_function(full_name, raw_name, return_type, address, &parameters, variable_arguments)
 }