Merge pull request #14 from hankluo6/string

Support java string and concatenation

Author: jserv

Makefile

@@ -50,7 +50,8 @@ TESTS = \
 	Static \
 	Invokevirtual \
 	Inherit \
-	Initializer
+	Initializer \
+	Strings
 
 check: $(addprefix tests/,$(TESTS:=-result.out))
 

class-heap.c

@@ -109,6 +109,15 @@ void free_class_heap()
         }
         free(class_heap.class_info[i]->clazz->methods);
 
+        bootmethods_attr_t *bootstrap =
+            class_heap.class_info[i]->clazz->bootstrap;
+        if (bootstrap) {
+            for (u2 j = 0; j < bootstrap->num_bootstrap_methods; j++)
+                free(bootstrap->bootstrap_methods[j].bootstrap_arguments);
+            free(bootstrap->bootstrap_methods);
+            free(bootstrap);
+        }
+
         free(class_heap.class_info[i]->clazz);
         free(class_heap.class_info[i]->name);
         free(class_heap.class_info[i]);

classfile.c

@@ -22,7 +22,7 @@ class_info_t *get_class_info(FILE *class_file)
 }
 
 /**
- * Get the number of integer parameters that a method takes.
+ * Get the number of parameters that a method takes.
  * Use the descriptor string of the method to determine its signature.
  */
 uint16_t get_number_of_parameters(method_t *method)
@@ -175,6 +175,15 @@ char *find_class_name_from_index(uint16_t idx, class_file_t *clazz)
     return (char *) name->info;
 }
 
+bootmethods_t *find_bootstrap_method(uint16_t idx, class_file_t *clazz)
+{
+    const_pool_info *info = get_constant(&clazz->constant_pool, idx);
+    assert(info->tag == CONSTANT_InvokeDynamic && "Expected a InvokeDynanmic");
+    return &clazz->bootstrap
+                ->bootstrap_methods[((CONSTANT_InvokeDynamic_info *) info->info)
+                                        ->bootstrap_method_attr_index];
+}
+
 void read_field_attributes(FILE *class_file, field_info *info)
 {
     for (u2 i = 0; i < info->attributes_count; i++) {
@@ -223,6 +232,54 @@ void read_method_attributes(FILE *class_file,
     assert(found_code && "Missing method code");
 }
 
+bootmethods_attr_t *read_bootstrap_attribute(FILE *class_file,
+                                             constant_pool_t *cp)
+{
+    u2 attributes_count = read_u2(class_file);
+    for (u2 i = 0; i < attributes_count; i++) {
+        attribute_info ainfo = {
+            .attribute_name_index = read_u2(class_file),
+            .attribute_length = read_u4(class_file),
+        };
+        long attribute_end = ftell(class_file) + ainfo.attribute_length;
+        const_pool_info *type_constant =
+            get_constant(cp, ainfo.attribute_name_index);
+        assert(type_constant->tag == CONSTANT_Utf8 && "Expected a UTF8");
+        if (!strcmp((char *) type_constant->info, "BootstrapMethods")) {
+            bootmethods_attr_t *bootstrap = malloc(sizeof(*bootstrap));
+
+            bootstrap->num_bootstrap_methods = read_u2(class_file);
+            bootstrap->bootstrap_methods = malloc(
+                sizeof(bootmethods_t) * bootstrap->num_bootstrap_methods);
+
+            assert(bootstrap->bootstrap_methods &&
+                   "Failed to allocate bootstrap method");
+            for (int j = 0; j < bootstrap->num_bootstrap_methods; ++j) {
+                bootstrap->bootstrap_methods[j].bootstrap_method_ref =
+                    read_u2(class_file);
+                bootstrap->bootstrap_methods[j].num_bootstrap_arguments =
+                    read_u2(class_file);
+                bootstrap->bootstrap_methods[j].bootstrap_arguments = malloc(
+                    sizeof(u2) *
+                    bootstrap->bootstrap_methods[j].num_bootstrap_arguments);
+                assert(bootstrap->bootstrap_methods[j].bootstrap_arguments &&
+                       "Failed to allocate bootstrap argument");
+                for (int k = 0;
+                     k <
+                     bootstrap->bootstrap_methods[j].num_bootstrap_arguments;
+                     ++k) {
+                    bootstrap->bootstrap_methods[j].bootstrap_arguments[k] =
+                        read_u2(class_file);
+                }
+            }
+            return bootstrap;
+        }
+        /* Skip the rest of the attribute */
+        fseek(class_file, attribute_end, SEEK_SET);
+    }
+    return NULL;
+}
+
 #define IS_STATIC 0x0008
 
 field_t *get_fields(FILE *class_file, constant_pool_t *cp, class_file_t *clazz)
@@ -311,6 +368,10 @@ class_file_t get_class(FILE *class_file)
     /* Read the list of static methods */
     clazz.methods = get_methods(class_file, &clazz.constant_pool);
 
+    /* Read the list of attributes */
+    clazz.bootstrap =
+        read_bootstrap_attribute(class_file, &clazz.constant_pool);
+
     clazz.initialized = false;
 
     return clazz;

classfile.h

@@ -54,12 +54,26 @@ typedef struct {
     variable_t *static_var; /* store static fields in the class */
 } field_t;
 
+typedef struct {
+    u2 bootstrap_method_ref;
+    u2 num_bootstrap_arguments;
+    u2 *bootstrap_arguments;
+} bootmethods_t;
+
+typedef struct {
+    u2 attribute_name_index;
+    u4 attribute_length;
+    u2 num_bootstrap_methods;
+    bootmethods_t *bootstrap_methods;
+} bootmethods_attr_t;
+
 typedef struct class_file {
     constant_pool_t constant_pool;
     class_info_t *info;
     method_t *methods;
     field_t *fields;
     u2 fields_count;
+    bootmethods_attr_t *bootstrap;
     bool initialized;
     struct class_file *next;
     struct class_file *prev;
@@ -89,4 +103,7 @@ char *find_field_info_from_index(uint16_t idx,
                                  char **name_info,
                                  char **descriptor_info);
 void read_field_attributes(FILE *class_file, field_info *info);
+bootmethods_t *find_bootstrap_method(uint16_t idx, class_file_t *clazz);
+bootmethods_attr_t *read_bootstrap_attribute(FILE *class_file,
+                                             constant_pool_t *cp);
 field_t *get_fields(FILE *class_file, constant_pool_t *cp, class_file_t *clazz);

constant-pool.c

@@ -48,6 +48,23 @@ CONSTANT_Class_info *get_class_name(constant_pool_t *cp, u2 idx)
     return (CONSTANT_Class_info *) class->info;
 }
 
+CONSTANT_MethodHandle_info *get_method_handle(constant_pool_t *cp, u2 idx)
+{
+    const_pool_info *handle = get_constant(cp, idx);
+    assert(handle->tag == CONSTANT_MethodHandle && "Expected a MethodHandle");
+    return (CONSTANT_MethodHandle_info *) handle->info;
+}
+
+char *get_string_utf(constant_pool_t *cp, u2 idx)
+{
+    const_pool_info *str = get_constant(cp, idx);
+    assert(str->tag == CONSTANT_String && "Expected a String");
+    const_pool_info *utf8 =
+        get_constant(cp, ((CONSTANT_String_info *) str->info)->string_index);
+    assert(utf8->tag == CONSTANT_Utf8 && "Expected a UTF8");
+    return (char *) utf8->info;
+}
+
 constant_pool_t get_constant_pool(FILE *class_file)
 {
     constant_pool_t cp = {
@@ -120,6 +137,32 @@ constant_pool_t get_constant_pool(FILE *class_file)
             break;
         }
 
+        case CONSTANT_String: {
+            CONSTANT_String_info *value = malloc(sizeof(*value));
+            assert(value && "Failed to allocate String constant");
+            value->string_index = read_u2(class_file);
+            constant->info = (u1 *) value;
+            break;
+        }
+
+        case CONSTANT_InvokeDynamic: {
+            CONSTANT_InvokeDynamic_info *value = malloc(sizeof(*value));
+            assert(value && "Failed to allocate InvokeDynamic constant");
+            value->bootstrap_method_attr_index = read_u2(class_file);
+            value->name_and_type_index = read_u2(class_file);
+            constant->info = (u1 *) value;
+            break;
+        }
+
+        case CONSTANT_MethodHandle: {
+            CONSTANT_MethodHandle_info *value = malloc(sizeof(*value));
+            assert(value && "Failed to allocate MethodHandle constant");
+            value->reference_kind = read_u1(class_file);
+            value->reference_index = read_u2(class_file);
+            constant->info = (u1 *) value;
+            break;
+        }
+
         default:
             fprintf(stderr, "Unknown constant type %d\n", constant->tag);
             exit(1);

constant-pool.h

@@ -13,9 +13,12 @@ typedef enum {
     CONSTANT_Integer = 3,
     CONSTANT_Long = 5,
     CONSTANT_Class = 7,
+    CONSTANT_String = 8,
     CONSTANT_FieldRef = 9,
     CONSTANT_MethodRef = 10,
     CONSTANT_NameAndType = 12,
+    CONSTANT_MethodHandle = 15,
+    CONSTANT_InvokeDynamic = 18,
 } const_pool_tag_t;
 
 typedef struct {
@@ -41,6 +44,20 @@ typedef struct {
     u2 descriptor_index;
 } CONSTANT_NameAndType_info;
 
+typedef struct {
+    u2 string_index;
+} CONSTANT_String_info;
+
+typedef struct {
+    u2 bootstrap_method_attr_index;
+    u2 name_and_type_index;
+} CONSTANT_InvokeDynamic_info;
+
+typedef struct {
+    u1 reference_kind;
+    u2 reference_index;
+} CONSTANT_MethodHandle_info;
+
 typedef struct {
     const_pool_tag_t tag;
     u1 *info;
@@ -58,3 +75,5 @@ const_pool_info *get_constant(constant_pool_t *constant_pool, u2 index);
 constant_pool_t get_constant_pool(FILE *class_file);
 CONSTANT_FieldOrMethodRef_info *get_methodref(constant_pool_t *cp, u2 idx);
 CONSTANT_Class_info *get_class_name(constant_pool_t *cp, u2 idx);
+CONSTANT_MethodHandle_info *get_method_handle(constant_pool_t *cp, u2 idx);
+char *get_string_utf(constant_pool_t *cp, u2 idx);