java-llvm/classfile.c

#include "classfile.h"
#include <assert.h>
#include <memory.h>
#include <stddef.h>
#include <stdio.h>
#include <stdlib.h>

#define SWAP_ENDIANNESS

void read_u2(void *ptr, size_t n, FILE *file) {
  u2 buffer[n];
  fread(&buffer, sizeof(u2), n, file);
#ifdef SWAP_ENDIANNESS
  for (size_t i = 0; i < n; i++) {
    buffer[i] = (buffer[i] >> 8) | (buffer[i] << 8);
  }
#endif
  memcpy(ptr, &buffer, n * sizeof(u2));
}

void read_u4(void *ptr, size_t n, FILE *file) {
  u4 buffer[n];
  fread(&buffer, sizeof(u4), n, file);
#ifdef SWAP_ENDIANNESS
  for (size_t i = 0; i < n; i++) {
    buffer[i] = ((buffer[i] >> 24) & 0xff) |    // move byte 3 to byte 0
                ((buffer[i] << 8) & 0xff0000) | // move byte 1 to byte 2
                ((buffer[i] >> 8) & 0xff00) |   // move byte 2 to byte 1
                ((buffer[i] << 24) & 0xff000000);
  }
#endif
  memcpy(ptr, &buffer, n * sizeof(u4));
}

ClassFile *ClassFile_load(const char *path) {
  FILE *file = fopen(path, "rb");
  if (file == NULL)
    return NULL;

  ClassFile *cf = (ClassFile *)malloc(sizeof(ClassFile));

  // magic 4bytes, minor_version 2bytes, major_version 2bytes,
  // constant_pool_count 2bytes
  read_u4(&(cf->magic), 1, file);
  read_u2(&(cf->minor_version), 3, file);

  // constant_pool
  cf->constant_pool =
      (cp_info *)malloc(sizeof(cp_info) * (cf->constant_pool_count - 1));
  for (u2 i = 0; i < (cf->constant_pool_count - 1); i++) {
    // cp_info.tag 2bytes
    fread(&(cf->constant_pool[i].tag), 1, 1, file);
    size_t size;
    switch (cf->constant_pool[i].tag) {
    case CONSTANT_Class:
      read_u2(&(cf->constant_pool[i].info.class_info.name_index), 1, file);
      break;
    case CONSTANT_Fieldref:
      read_u2(&(cf->constant_pool[i].info.fieldref_info.class_index), 1, file);
      read_u2(&(cf->constant_pool[i].info.fieldref_info.name_and_type_index), 1,
              file);
      break;
    case CONSTANT_Methodref:
      read_u2(&(cf->constant_pool[i].info.methodref_info.class_index), 1, file);
      read_u2(&(cf->constant_pool[i].info.methodref_info.name_and_type_index),
              1, file);
      break;
    case CONSTANT_InterfaceMethodref:
      read_u2(&(cf->constant_pool[i].info.interface_methodred_info.class_index),
              1, file);
      read_u2(&(cf->constant_pool[i]
                    .info.interface_methodred_info.name_and_type_index),
              1, file);
      break;
    case CONSTANT_String:
      read_u2(&(cf->constant_pool[i].info.string_info.string_index), 1, file);
      break;
    case CONSTANT_Integer:
      read_u4(&(cf->constant_pool[i].info.integer_info.bytes), 1, file);
      break;
    case CONSTANT_Float:
      read_u4(&(cf->constant_pool[i].info.float_info.bytes), 1, file);
      break;
    case CONSTANT_Long:
      read_u4(&(cf->constant_pool[i].info.long_info.high_bytes), 1, file);
      read_u4(&(cf->constant_pool[i].info.long_info.low_bytes), 1, file);
      break;
    case CONSTANT_Double:
      read_u4(&(cf->constant_pool[i].info.double_info.high_bytes), 1, file);
      read_u4(&(cf->constant_pool[i].info.double_info.low_bytes), 1, file);
      break;
    case CONSTANT_NameAndType:
      read_u2(&(cf->constant_pool[i].info.name_and_type_info.name_index), 1,
              file);
      read_u2(&(cf->constant_pool[i].info.name_and_type_info.descriptor_index),
              1, file);
      break;
    case CONSTANT_Utf8:
      read_u2(&(cf->constant_pool[i].info.utf8_info.length), 1, file);
      fread(&(cf->constant_pool[i].info.utf8_info.bytes),
            cf->constant_pool[i].info.utf8_info.length, 1, file);
      break;
    case CONSTANT_MethodHandle:
      fread(&(cf->constant_pool[i].info.method_handle_info.reference_kind), 1,
            1, file);
      read_u2(&(cf->constant_pool[i].info.method_handle_info.reference_index),
              1, file);
      break;
    case CONSTANT_MethodType:
      read_u2(&(cf->constant_pool[i].info.method_type_info.descriptor_index), 1,
              file);
      break;
    case CONSTANT_Dynamic:
      read_u2(
          &(cf->constant_pool[i].info.dynamic_info.bootstrap_method_attr_index),
          1, file);
      read_u2(&(cf->constant_pool[i].info.dynamic_info.name_and_type_index), 1,
              file);
      break;
    case CONSTANT_InvokeDynamic:
      read_u2(&(cf->constant_pool[i]
                    .info.invoke_dynamic_info.bootstrap_method_attr_index),
              1, file);
      read_u2(
          &(cf->constant_pool[i].info.invoke_dynamic_info.name_and_type_index),
          1, file);
      break;
    case CONSTANT_Module:
      read_u2(&(cf->constant_pool[i].info.module_info.name_index), 1, file);
      break;
    case CONSTANT_Package:
      read_u2(&(cf->constant_pool[i].info.package_info.name_index), 1, file);
      break;
    default:
      printf("unknown constant tag: %hhu\n", cf->constant_pool[i].tag);
      // assert(0);
    }
  }

  // access_flags 2bytes, this_class 2bytes, super_class 2bytes, interface_count
  // 2byte
  read_u2(&(cf->access_flags), 4, file);

  // interfaces
  cf->interfaces = (u2 *)malloc(2 * cf->interfaces_count);
  read_u2(cf->interfaces, cf->interfaces_count, file);

  // fields_count 2bytes
  read_u2(&(cf->fields_count), 1, file);

  // fields
  cf->fields = (field_info *)malloc(sizeof(field_info) * cf->fields_count);
  for (u2 i = 0; i < cf->fields_count; i++) {
    // access_flags 2bytes, name_index 2bytes, descriptor_index 2bytes,
    // attributes_count 2bytes
    read_u2(&(cf->fields[i].access_flags), 4, file);
    cf->fields[i].attributes = (attribute_info *)malloc(
        sizeof(attribute_info) * cf->fields[i].attributes_count);
    for (u2 j = 0; j < cf->fields[i].attributes_count; j++) {
      // attribute_name_index 2bytes, attribute_length 4bytes
      read_u2(&(cf->fields[i].attributes[j].attribute_name_index), 1, file);
      read_u4(&(cf->fields[i].attributes[j].attribute_length), 1, file);
      cf->fields[i].attributes[j].info =
          (u1 *)malloc(cf->fields[i].attributes[j].attribute_length);
      fread(cf->fields[i].attributes[j].info,
            cf->fields[i].attributes[j].attribute_length, 1, file);
    }
  }

  // methods_count 2bytes
  read_u2(&(cf->methods_count), 1, file);

  // methods
  cf->methods = (method_info *)malloc(sizeof(method_info) * cf->methods_count);
  for (u2 i = 0; i < cf->methods_count; i++) {
    // access_flags 2bytes, name_index 2bytes, descriptor_index 2bytes,
    // attributes_count 2bytes
    read_u2(&(cf->methods[i].access_flags), 4, file);
    cf->methods[i].attributes = (attribute_info *)malloc(
        sizeof(attribute_info) * cf->methods[i].attributes_count);
    for (u2 j = 0; j < cf->methods[i].attributes_count; j++) {
      // attribute_name_index 2bytes, attribute_length 4bytes
      read_u2(&(cf->methods[i].attributes[j].attribute_name_index), 1, file);
      read_u4(&(cf->methods[i].attributes[j].attribute_length), 1, file);
      cf->methods[i].attributes[j].info =
          (u1 *)malloc(cf->methods[i].attributes[j].attribute_length);
      fread(cf->methods[i].attributes[j].info,
            cf->methods[i].attributes[j].attribute_length, 1, file);
    }
  }

  // attributes_count 2bytes
  read_u2(&(cf->attributes_count), 1, file);

  // attributes
  cf->attributes =
      (attribute_info *)malloc(sizeof(attribute_info) * cf->attributes_count);
  for (u2 i = 0; i < cf->attributes_count; i++) {
    // attribute_name_index 2bytes, attribute_length 4bytes
    read_u2(&(cf->attributes[i].attribute_name_index), 1, file);
    read_u4(&(cf->attributes[i].attribute_length), 1, file);
    cf->attributes[i].info = (u1 *)malloc(cf->attributes[i].attribute_length);
    fread(cf->attributes[i].info, cf->attributes[i].attribute_length, 1, file);
  }

  fclose(file);

  return cf;
}

void ClassFile_info(const ClassFile *cf) {
  printf("magic=%X\n"
         "minor_version=%hu\n"
         "major_version=%hu\n"
         "constant_pool_count=%hu\n"
         "access_flags=%hx\n"
         "this_class=%hu\n"
         "super_class=%hu\n"
         "interfaces_count=%hu\n"
         "fields_count=%hu\n"
         "methods_count=%hu\n"
         "attributes_count=%hu\n",
         cf->magic, cf->minor_version, cf->major_version,
         cf->constant_pool_count, cf->access_flags, cf->this_class,
         cf->super_class, cf->interfaces_count, cf->fields_count,
         cf->methods_count, cf->attributes_count);
}