feat: add header file export

Author: FASTSHIFT

ghidra_common.py

@@ -10,6 +10,7 @@
 Note: This module runs in Ghidra's Jython environment and uses Ghidra's API.
 """
 
+import os
 import re
 
 # Ghidra undefined type to standard C type mapping
@@ -443,3 +444,175 @@ def write_file_header(f, module_name, func_count, program_name=None):
     f.write("#include <stdint.h>\n")
     f.write("#include <stdbool.h>\n")
     f.write("#include <stddef.h>\n\n")
+
+
+# ============================================================
+# Header File Generation Functions
+# ============================================================
+
+
+def extract_function_signature(decompiled_code):
+    """
+    Extract function signature from decompiled code.
+    Returns the normalized function signature string or None if extraction fails.
+    """
+    if not decompiled_code:
+        return None
+
+    # Find the first function definition (before the opening brace)
+    lines = decompiled_code.strip().split("\n")
+    signature_lines = []
+
+    for line in lines:
+        if "{" in line:
+            # Include part before the brace
+            signature_lines.append(line.split("{")[0].strip())
+            break
+        signature_lines.append(line.strip())
+
+    if not signature_lines:
+        return None
+
+    signature = " ".join(signature_lines).strip()
+
+    # Clean up the signature
+    signature = re.sub(r"\s+", " ", signature)
+
+    # Skip if it looks like a variable declaration or empty
+    if not signature or signature.endswith(";"):
+        return None
+
+    # Normalize types in the signature
+    signature = normalize_code_types(signature)
+
+    return signature
+
+
+def generate_header_file(
+    output_dir, module_name, func_signatures, source_type="decompilation"
+):
+    """
+    Generate a header file for a module with function declarations.
+
+    Args:
+        output_dir: Directory to write the header file
+        module_name: Name of the module
+        func_signatures: List of (func_name, signature) tuples
+        source_type: Description of the source (e.g., "ELF decompilation", "library decompilation")
+
+    Returns:
+        Path to the generated header file
+    """
+    safe_name = sanitize_filename(module_name)
+    header_file = os.path.join(output_dir, "{}.h".format(safe_name))
+
+    guard_name = "_{}_H_".format(safe_name.upper())
+
+    with open(header_file, "w") as f:
+        f.write("/**\n")
+        f.write(" * Header: {}.h\n".format(safe_name))
+        f.write(" * Module: {}\n".format(module_name))
+        f.write(" * Functions: {}\n".format(len(func_signatures)))
+        f.write(" * \n")
+        f.write(" * Auto-generated by LibSurgeon from {}\n".format(source_type))
+        f.write(" */\n\n")
+
+        f.write("#ifndef {}\n".format(guard_name))
+        f.write("#define {}\n\n".format(guard_name))
+
+        f.write("#include <stdint.h>\n")
+        f.write("#include <stdbool.h>\n")
+        f.write("#include <stddef.h>\n")
+        f.write('#include "_types.h"\n\n')
+
+        f.write("#ifdef __cplusplus\n")
+        f.write('extern "C" {\n')
+        f.write("#endif\n\n")
+
+        # Write function declarations
+        f.write("/* Function Declarations */\n\n")
+        for func_name, signature in sorted(func_signatures, key=lambda x: x[0]):
+            if signature:
+                f.write("/* {} */\n".format(func_name))
+                f.write("{};\n\n".format(signature))
+
+        f.write("#ifdef __cplusplus\n")
+        f.write("}\n")
+        f.write("#endif\n\n")
+
+        f.write("#endif /* {} */\n".format(guard_name))
+
+    return header_file
+
+
+def generate_master_header(output_dir, module_names, program_name):
+    """
+    Generate a master header file that includes all module headers.
+
+    Args:
+        output_dir: Directory to write the header file
+        module_names: Iterable of module names
+        program_name: Name of the source program
+
+    Returns:
+        Path to the generated master header file
+    """
+    header_file = os.path.join(output_dir, "_all_headers.h")
+
+    with open(header_file, "w") as f:
+        f.write("/**\n")
+        f.write(" * Master Header File\n")
+        f.write(" * Source: {}\n".format(program_name))
+        f.write(" * Modules: {}\n".format(len(list(module_names))))
+        f.write(" * \n")
+        f.write(" * Auto-generated by LibSurgeon\n")
+        f.write(" * Include this file to get all function declarations.\n")
+        f.write(" */\n\n")
+
+        f.write("#ifndef _ALL_HEADERS_H_\n")
+        f.write("#define _ALL_HEADERS_H_\n\n")
+
+        f.write('#include "_types.h"\n\n')
+
+        for module_name in sorted(module_names):
+            safe_name = sanitize_filename(module_name)
+            f.write('#include "{}.h"\n'.format(safe_name))
+
+        f.write("\n#endif /* _ALL_HEADERS_H_ */\n")
+
+    return header_file
+
+
+def generate_types_header(output_dir):
+    """
+    Generate _types.h header with type definitions.
+
+    Args:
+        output_dir: Directory to write the header file
+
+    Returns:
+        Path to the generated types header file
+    """
+    types_file = os.path.join(output_dir, "_types.h")
+
+    with open(types_file, "w") as f:
+        f.write("/**\n")
+        f.write(" * Type Definitions for Decompiled Code\n")
+        f.write(" * \n")
+        f.write(" * This file contains typedef mappings for Ghidra-generated types.\n")
+        f.write(" * Auto-generated by LibSurgeon\n")
+        f.write(" */\n\n")
+
+        f.write("#ifndef _LIBSURGEON_TYPES_H_\n")
+        f.write("#define _LIBSURGEON_TYPES_H_\n\n")
+
+        f.write("#include <stdint.h>\n")
+        f.write("#include <stdbool.h>\n\n")
+
+        f.write("/* Unknown type definitions (signedness uncertain) */\n")
+        f.write(UNKNOWN_TYPE_DEFS)
+        f.write("\n")
+
+        f.write("#endif /* _LIBSURGEON_TYPES_H_ */\n")
+
+    return types_file

ghidra_decompile_elf.py

@@ -52,7 +52,11 @@
     demangle_cpp_name,
     enhance_decompiled_code,
     extract_class_from_method,
+    extract_function_signature,
     extract_namespace,
+    generate_header_file,
+    generate_master_header,
+    generate_types_header,
     normalize_code_types,
     normalize_ghidra_type,
     sanitize_filename,
@@ -495,109 +499,6 @@ def write_file_header(f, module_name, func_count):
     f.write('#include "../include/_types.h"\n\n')
 
 
-def extract_function_signature(decompiled_code):
-    """
-    Extract function signature from decompiled code.
-    Returns the normalized function signature string or None if extraction fails.
-    """
-    if not decompiled_code:
-        return None
-
-    # Find the first function definition (before the opening brace)
-    lines = decompiled_code.strip().split("\n")
-    signature_lines = []
-
-    for line in lines:
-        if "{" in line:
-            # Include part before the brace
-            signature_lines.append(line.split("{")[0].strip())
-            break
-        signature_lines.append(line.strip())
-
-    if not signature_lines:
-        return None
-
-    signature = " ".join(signature_lines).strip()
-
-    # Clean up the signature
-    signature = re.sub(r"\s+", " ", signature)
-
-    # Normalize types in the signature (already done in decompiled code, but just in case)
-    signature = normalize_code_types(signature)
-
-    return signature
-
-
-def generate_header_file(output_dir, module_name, func_signatures):
-    """Generate a header file for a module with function declarations."""
-    safe_name = sanitize_filename(module_name)
-    header_file = os.path.join(output_dir, "{}.h".format(safe_name))
-
-    guard_name = "_{}_H_".format(safe_name.upper())
-
-    with open(header_file, "w") as f:
-        f.write("/**\n")
-        f.write(" * Header: {}.h\n".format(safe_name))
-        f.write(" * Module: {}\n".format(module_name))
-        f.write(" * Functions: {}\n".format(len(func_signatures)))
-        f.write(" * \n")
-        f.write(" * Auto-generated by LibSurgeon from ELF decompilation\n")
-        f.write(" */\n\n")
-
-        f.write("#ifndef {}\n".format(guard_name))
-        f.write("#define {}\n\n".format(guard_name))
-
-        f.write("#include <stdint.h>\n")
-        f.write("#include <stdbool.h>\n")
-        f.write("#include <stddef.h>\n")
-        f.write('#include "_types.h"\n\n')
-
-        f.write("#ifdef __cplusplus\n")
-        f.write('extern "C" {\n')
-        f.write("#endif\n\n")
-
-        # Write function declarations
-        f.write("/* Function Declarations */\n\n")
-        for func_name, signature in sorted(func_signatures, key=lambda x: x[0]):
-            if signature:
-                f.write("/* {} */\n".format(func_name))
-                f.write("{};\n\n".format(signature))
-
-        f.write("#ifdef __cplusplus\n")
-        f.write("}\n")
-        f.write("#endif\n\n")
-
-        f.write("#endif /* {} */\n".format(guard_name))
-
-    return header_file
-
-
-def generate_master_header(output_dir, module_names, program_name):
-    """Generate a master header file that includes all module headers."""
-    header_file = os.path.join(output_dir, "_all_headers.h")
-
-    with open(header_file, "w") as f:
-        f.write("/**\n")
-        f.write(" * Master Header File\n")
-        f.write(" * Source: {}\n".format(program_name))
-        f.write(" * Modules: {}\n".format(len(module_names)))
-        f.write(" * \n")
-        f.write(" * Auto-generated by LibSurgeon from ELF decompilation\n")
-        f.write(" * Include this file to get all function declarations.\n")
-        f.write(" */\n\n")
-
-        f.write("#ifndef _ALL_HEADERS_H_\n")
-        f.write("#define _ALL_HEADERS_H_\n\n")
-
-        for module_name in sorted(module_names):
-            safe_name = sanitize_filename(module_name)
-            f.write('#include "{}.h"\n'.format(safe_name))
-
-        f.write("\n#endif /* _ALL_HEADERS_H_ */\n")
-
-    return header_file
-
-
 def format_data_type(dt, indent=0):
     """Format a Ghidra DataType to C code string"""
     if dt is None:
@@ -1106,10 +1007,16 @@ def main():
     for module_name in sorted(module_signatures.keys()):
         signatures = module_signatures[module_name]
         if signatures:
-            generate_header_file(include_dir, module_name, signatures)
+            generate_header_file(
+                include_dir, module_name, signatures, "ELF decompilation"
+            )
             header_count += 1
             total_signatures += len(signatures)
 
+    # Generate types header
+    if header_count > 0:
+        generate_types_header(include_dir)
+
     # Generate master header
     if header_count > 0:
         generate_master_header(include_dir, module_signatures.keys(), program_name)