instruction init

classfile.c

@@ -1,4 +1,5 @@
 #include "classfile.h"
+#include "instruction.h"
 #include <assert.h>
 #include <memory.h>
 #include <stddef.h>
@@ -63,11 +64,28 @@ stack_frame_type stack_frame_type_enum(u2 frame_type) {
   }
 }
 
+ATTRIBUTE_TAGS attribute_tag(char* name) {
+  for (size_t i = 1; i < ATTRIBUTE_TAGS_END; i++) {
+    if (strcmp(ATTRIBUTE_TAGS_STRINGS[i - 1], name) == 0)
+      return (ATTRIBUTE_TAGS)i;
+  }
+  return UNKNOWN;
+}
+
+char* constant_to_string(ClassFile* cf, cp_info info) {}
+
 void parse_attribute(ClassFile* cf, attribute_info* base, FILE* file) {
   char* name = (char*)cf->constant_pool[base->attribute_name_index - 1].info.utf8_info.bytes;
-  if (strcmp("ConstantValue", name) == 0) {
+  base->_tag = attribute_tag(name);
+  switch (base->_tag) {
+    case UNKNOWN:
+      printf("unhandled attribute type: %s\n", name);
+      fseek(file, base->attribute_length, SEEK_CUR);
+      break;
+    case CONSTANT_VALUE:
       read_u2(&(base->info.constant_value_attribute.constantvalue_index), 1, file);
-  } else if (strcmp("Code", name) == 0) {
+      break;
+    case CODE:
       read_u2(&(base->info.code_attribute.max_stack), 2, file);
       read_u4(&(base->info.code_attribute.code_length), 1, file);
       base->info.code_attribute.code = (u1*)malloc(base->info.code_attribute.code_length);
@@ -85,7 +103,8 @@ void parse_attribute(ClassFile* cf, attribute_info* base, FILE* file) {
         read_u4(&(attr->attribute_length), 1, file);
         parse_attribute(cf, attr, file);
       }
-  } else if (strcmp("StackMapTable", name) == 0) {
+      break;
+    case STACK_MAP_TABLE:
       read_u2(&(base->info.stack_map_table_attribute.number_of_entries), 1, file);
       base->info.stack_map_table_attribute.entries = malloc(sizeof(stack_map_frame) * base->info.stack_map_table_attribute.number_of_entries);
       for (u2 i = 0; i < base->info.stack_map_table_attribute.number_of_entries; i++) {
@@ -156,19 +175,36 @@ void parse_attribute(ClassFile* cf, attribute_info* base, FILE* file) {
             break;
         }
       }
-  } else if (strcmp("NestHost", name) == 0) {
+      break;
+    case BOOTSTRAP_METHODS: {
+      BootstrapMethods_attribute* attr = &base->info.bootstrap_methods_attribute;
+
+      read_u2(&(attr->num_bootstrap_methods), 1, file);
+      attr->bootstrap_methods = malloc(sizeof(BootstrapMethod) * attr->num_bootstrap_methods);
+
+      for (u2 i = 0; i < attr->num_bootstrap_methods; i++) {
+        BootstrapMethod* method = &attr->bootstrap_methods[i];
+        read_u2(&(method->bootstrap_method_ref), 2, file);
+        method->bootstrap_arguments = malloc(sizeof(u2) * method->num_bootstrap_arguments);
+        read_u2(method->bootstrap_arguments, method->num_bootstrap_arguments, file);
+      }
+      break;
+    }
+    case NEST_HOST:
       read_u2(&(base->info.nest_host_attribute.host_class_index), 1, file);
-  } else if (strcmp("NestMembers", name) == 0) {
+      break;
+    case NEST_MEMBERS:
       read_u2(&(base->info.nest_members_attribute.number_of_classes), 1, file);
       base->info.nest_members_attribute.classes = malloc(sizeof(u2) * base->info.nest_members_attribute.number_of_classes);
       read_u2(base->info.nest_members_attribute.classes, base->info.nest_members_attribute.number_of_classes, file);
-  } else if (strcmp("PermittedSubclasses", name) == 0) {
+      break;
+    case PERMITTED_SUBCLASSES:
       read_u2(&(base->info.permitted_subclasses_attribute.number_of_classes), 1, file);
       base->info.permitted_subclasses_attribute.classes = malloc(sizeof(u2) * base->info.permitted_subclasses_attribute.number_of_classes);
       read_u2(base->info.permitted_subclasses_attribute.classes, base->info.permitted_subclasses_attribute.number_of_classes, file);
-  } else {
-    printf("unhandled attribute type: %s\n", name);
-    fseek(file, base->attribute_length, SEEK_CUR);
+      break;
+    case ATTRIBUTE_TAGS_END:
+      assert(0);
   }
 }
 
@@ -324,18 +360,18 @@ ClassFile* ClassFile_load(const char* path) {
 }
 
 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",
+  printf("Metadata:\n"
+      "\tmagic=%X\n"
+      "\tminor_version=%hu\n"
+      "\tmajor_version=%hu\n"
+      "\tconstant_pool_count=%hu\n"
+      "\taccess_flags=%hx\n"
+      "\tthis_class=%hu\n"
+      "\tsuper_class=%hu\n"
+      "\tinterfaces_count=%hu\n"
+      "\tfields_count=%hu\n"
+      "\tmethods_count=%hu\n"
+      "\tattributes_count=%hu\n\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);
 
@@ -362,10 +398,12 @@ void ClassFile_info(const ClassFile* cf) {
       case CONSTANT_Long:
         merged = merge_bytes(cf->constant_pool[i].info.long_info.high_bytes, cf->constant_pool[i].info.long_info.low_bytes);
         printf("\t%ld\n", *(long*)&merged);
+        i++;
         break;
       case CONSTANT_Double:
         merged = merge_bytes(cf->constant_pool[i].info.double_info.high_bytes, cf->constant_pool[i].info.double_info.low_bytes);
         printf("\t%lf\n", *(double*)&merged);
+        i++;
         break;
       case CONSTANT_Class:
         printf("\t#%hu\t\t\t// %s\n", cf->constant_pool[i].info.class_info.name_index,
@@ -389,15 +427,69 @@ void ClassFile_info(const ClassFile* cf) {
                buffer);
         free(buffer);
       default:
-        putchar('\n');
+        printf("\n");
     }
   }
 
+  // class
+  char* this_name = (char*)cf->constant_pool[cf->constant_pool[cf->this_class - 1].info.class_info.name_index - 1].info.utf8_info.bytes;
+  char* super_name =
+      cf->super_class != 0 ? (char*)cf->constant_pool[cf->constant_pool[cf->super_class - 1].info.class_info.name_index - 1].info.utf8_info.bytes : NULL;
+  char* type = (cf->access_flags & ACC_INTERFACE ? "interface" : (cf->access_flags & ACC_ENUM ? "enum" : (cf->access_flags & ACC_MODULE ? "module" : "class")));
+  printf("\n%s%s%s%s %s%s%s ", cf->access_flags & ACC_PUBLIC ? "public " : "", cf->access_flags & ACC_FINAL ? "final " : "",
+         cf->access_flags & ACC_ABSTRACT ? "abstract " : "", type, this_name, super_name ? " extends " : "", super_name ? super_name : "");
+
   // interfaces
+  if (cf->interfaces_count > 0)
+    printf("implements ");
+  for (u2 i = 0; i < cf->interfaces_count; i++) {
+    printf("%s ", cf->constant_pool[cf->interfaces[i] - 1].info.utf8_info.bytes);
+  }
+
+  printf("{\n");
 
   // fields
+  for (u2 i = 0; i < cf->fields_count; i++) {
+    u2 flags = cf->fields[i].access_flags;
+    char* access = flags & ACC_PUBLIC ? "public" : flags & ACC_PRIVATE ? "private" : flags & ACC_PROTECTED ? "protected" : "";
+    char* descriptor = (char*)cf->constant_pool[cf->fields[i].descriptor_index - 1].info.utf8_info.bytes;
+    char* name = (char*)cf->constant_pool[cf->fields[i].name_index - 1].info.utf8_info.bytes;
+    printf("\t%s %s%s%s%s %s", access, flags & ACC_STATIC ? "static " : "", flags & ACC_FINAL ? "final " : "", flags & ACC_VOLATLE ? "volatile " : "",
+           descriptor, name);
+
+    // TODO: print field attributes
+    for (u2 j = 0; j < cf->fields[i].attributes_count; j++) {
+    }
+
+    printf(";\n");
+  }
 
   // methods
+  for (u2 i = 0; i < cf->methods_count; i++) {
+    u2 flags = cf->methods[i].access_flags;
+    char* access = flags & ACC_PUBLIC ? "public" : flags & ACC_PRIVATE ? "private" : flags & ACC_PROTECTED ? "protected" : "";
+    char* descriptor = (char*)cf->constant_pool[cf->methods[i].descriptor_index - 1].info.utf8_info.bytes;
+    char* name = (char*)cf->constant_pool[cf->methods[i].name_index - 1].info.utf8_info.bytes;
+    printf("\t%s %s%s%s%s %s", access, flags & ACC_STATIC ? "static " : "", flags & ACC_FINAL ? "final " : "", flags & ACC_VOLATLE ? "volatile " : "",
+           descriptor, name);
+
+    // TODO: print field attributes
+    for (u2 j = 0; j < cf->methods[i].attributes_count; j++) {
+    }
+
+    printf("{\n");
+
+    // code
+    for (u2 j = 0; j < cf->methods[i].attributes_count; j++) {
+      if (cf->methods[i].attributes[j]._tag == CODE) {
+        Instruction* code = Instruction_parse(cf->methods[i].attributes[j].info.code_attribute.code_length, cf->methods[i].attributes[j].info.code_attribute.code);
+        Instruction_print_code(code);
+        // TODO: free code
+      }
+    }
+
+    printf("\t}\n\n");
+  }
 }
 
 char* ClassFile_resolve_NameAnyType(const ClassFile* cf, CONSTANT_NameAndType_info info) {

classfile.h

@@ -8,6 +8,7 @@
 typedef uint8_t u1;
 typedef uint16_t u2;
 typedef uint32_t u4;
+typedef uint64_t u8;
 
 typedef enum {
   ACC_PUBLIC = 0x0001,
@@ -261,12 +262,14 @@ typedef struct {
 } StackMapTable_attribute;
 
 typedef struct {
-  u2 num_bootstrap_methods;
-  struct {
   u2 bootstrap_method_ref;
   u2 num_bootstrap_arguments;
   u2* bootstrap_arguments;
-  } * bootstrap_methods;
+} BootstrapMethod;
+
+typedef struct {
+  u2 num_bootstrap_methods;
+  BootstrapMethod* bootstrap_methods;
 } BootstrapMethods_attribute;
 
 typedef struct {
@@ -283,9 +286,26 @@ typedef struct {
   u2* classes;
 } PermittedSubclasses_attribute;
 
+typedef enum {
+  UNKNOWN,
+  CONSTANT_VALUE,
+  CODE,
+  STACK_MAP_TABLE,
+  BOOTSTRAP_METHODS,
+  NEST_HOST,
+  NEST_MEMBERS,
+  PERMITTED_SUBCLASSES,
+  ATTRIBUTE_TAGS_END
+} ATTRIBUTE_TAGS;
+
+static const char* ATTRIBUTE_TAGS_STRINGS[] = {
+    "ConstantValue", "Code", "StackMapTable", "BootstrapMethods", "NestHost", "NestMembers", "PermittedSubclasses",
+};
+
 typedef struct {
   u2 attribute_name_index;
   u4 attribute_length;
+  ATTRIBUTE_TAGS _tag;
   union {
     ConstantValue_attribute constant_value_attribute;
     Code_attribute code_attribute;

instruction.c

@@ -0,0 +1,65 @@
+#include "instruction.h"
+#include <stdbool.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include "classfile.h"
+
+Instruction* Instruction_parse(uint32_t bytes_length, uint8_t* bytes) {
+  Instruction* start = (Instruction*)malloc(sizeof(Instruction));
+  Instruction* current = start;
+
+  u4 i = 0, n = 0;
+
+  while (i < bytes_length) {
+    current->opcode = bytes[i++];
+
+    switch (current->opcode) {
+      case ALOAD_0:
+      case RETURN:
+        current->operands = NULL;
+        break;
+      case LDC:
+        *(u1*)&current->operands = bytes[i++];
+        break;
+      case GETSTATIC:
+      case INVOKESPECIAL:
+      case INVOKEVIRTUAL:
+        // current->operands = malloc(sizeof(u2));
+        *(u2*)&current->operands = (*(u2*)&bytes[i] << 8) | bytes[i + 1];
+        i += 2;
+        break;
+    }
+
+    if (i < bytes_length) {
+      current->next = (Instruction*)malloc(sizeof(Instruction));
+      current = current->next;
+    } else {
+      current->next = NULL;
+    }
+  }
+
+  return start;
+}
+
+void Instruction_print_code(Instruction* code) {
+  u4 i = 0;
+  while (code) {
+    printf("\t\t%hx ", (u1)code[i].opcode);
+    switch (code[i].opcode) {
+      case ALOAD_0:
+      case RETURN:
+        putchar('\n');
+        break;
+      case LDC:
+        printf("#%hhu\n", *(u1*)&code[i].operands);
+        break;
+      case GETSTATIC:
+      case INVOKESPECIAL:
+      case INVOKEVIRTUAL:
+        printf("#%hu\n", *(u2*)&code[i].operands);
+        break;
+    }
+    code = code->next;
+  }
+}

instruction.h

@@ -0,0 +1,24 @@
+#include <stddef.h>
+#include <stdint.h>
+
+#pragma pack(1)
+
+typedef enum {
+  ALOAD_0 = 0x2a,
+  INVOKESPECIAL = 0xb7,
+  RETURN = 0xb1,
+  GETSTATIC = 0xb2,
+  LDC = 0x12,
+  INVOKEVIRTUAL = 0xb6,
+} OPCODE;
+
+typedef struct Instruction {
+  OPCODE opcode;
+  uint8_t* operands;
+  struct Instruction* next;
+} Instruction;
+
+Instruction* Instruction_parse(uint32_t bytes_length, uint8_t* bytes);
+void Instruction_print_code(Instruction* code);
+
+#pragma pack()

main.c

@@ -1,6 +1,6 @@
 #include "classfile.h"
 
 int main(int argc, char **argv) {
-  ClassFile *cf = ClassFile_load("test/Main.class");
+  ClassFile *cf = ClassFile_load("Main.class");
   ClassFile_info(cf);
 }