improve ast printing
[cparser] / parser.c
index 914f3a6..d5d2c5e 100644 (file)
--- a/parser.c
+++ b/parser.c
@@ -13,7 +13,7 @@
 #include "adt/error.h"
 #include "adt/array.h"
 
-#define PRINT_TOKENS
+//#define PRINT_TOKENS
 #define MAX_LOOKAHEAD 2
 
 struct environment_entry_t {
@@ -27,8 +27,8 @@ static token_t               lookahead_buffer[MAX_LOOKAHEAD];
 static int                   lookahead_bufpos;
 static struct obstack        environment_obstack;
 static environment_entry_t **environment_stack = NULL;
-static translation_unit_t   *translation_unit  = NULL;
-static block_statement_t    *context           = NULL;
+static context_t            *context           = NULL;
+static declaration_t        *last_declaration  = NULL;
 
 static
 statement_t *parse_compound_statement(void);
@@ -174,6 +174,21 @@ void eat_until(int token_type)
     }                                              \
     next_token();
 
+static void set_context(context_t *new_context)
+{
+       context = new_context;
+
+       declaration_t *declaration = new_context->declarations;
+       if(declaration != NULL) {
+               while(1) {
+                       if(declaration->next == NULL)
+                               break;
+                       declaration = declaration->next;
+               }
+       }
+
+       last_declaration = declaration;
+}
 
 /**
  * pushs an environment_entry on the environment stack and links the
@@ -190,11 +205,13 @@ void environment_push(declaration_t *declaration, const void *context)
        ARR_RESIZE(environment_stack, top + 1);
        environment_stack[top] = entry;
 
+       assert(declaration->source_position.input_name != NULL);
+
        symbol_t *symbol = declaration->symbol;
        assert(declaration != symbol->declaration);
 
        if(symbol->context == context) {
-               if(context != NULL) {
+               if(symbol->declaration != NULL) {
                        assert(symbol->declaration != NULL);
                        parser_print_error_prefix_pos(declaration->source_position);
                        fprintf(stderr, "multiple definitions for symbol '%s'.\n",
@@ -255,9 +272,12 @@ static expression_t *parse_assignment_expression(void)
        return parse_expression();
 }
 
-static compound_entry_t *parse_compound_type_entries(void);
+static void parse_compound_type_entries(void);
 static void parse_declarator(declaration_t *declaration,
-                             storage_class_t storage_class, type_t *type);
+                             storage_class_t storage_class, type_t *type,
+                             int may_omit_identifier);
+static void maybe_push_declaration(declaration_t *declaration);
+static void record_declaration(declaration_t *declaration);
 
 typedef struct declaration_specifiers_t  declaration_specifiers_t;
 struct declaration_specifiers_t {
@@ -273,8 +293,11 @@ static type_t *parse_struct_specifier(void)
        struct_type->type.type       = TYPE_COMPOUND_STRUCT;
        struct_type->source_position = token.source_position;
 
+       int         top          = environment_top();
+       context_t  *last_context = context;
+       set_context(&struct_type->context);
+
        if(token.type == T_IDENTIFIER) {
-               /* TODO */
                next_token();
                if(token.type == '{') {
                        parse_compound_type_entries();
@@ -284,9 +307,13 @@ static type_t *parse_struct_specifier(void)
        } else {
                parse_error_expected("problem while parsing struct type specifiers",
                                     T_IDENTIFIER, '{', 0);
-               return NULL;
+               struct_type = NULL;
        }
 
+       assert(context == &struct_type->context);
+       set_context(last_context);
+       environment_pop_to(top);
+
        return (type_t*) struct_type;
 }
 
@@ -298,6 +325,10 @@ static type_t *parse_union_specifier(void)
        union_type->type.type       = TYPE_COMPOUND_UNION;
        union_type->source_position = token.source_position;
 
+       int         top          = environment_top();
+       context_t  *last_context = context;
+       set_context(&union_type->context);
+
        if(token.type == T_IDENTIFIER) {
                union_type->symbol = token.v.symbol;
                next_token();
@@ -309,8 +340,13 @@ static type_t *parse_union_specifier(void)
        } else {
                parse_error_expected("problem while parsing union type specifiers",
                                     T_IDENTIFIER, '{');
+               union_type = NULL;
        }
 
+       assert(context == &union_type->context);
+       set_context(last_context);
+       environment_pop_to(top);
+
        return (type_t*) union_type;
 }
 
@@ -433,7 +469,6 @@ typedef enum {
        case T_enum:            \
        COMPLEX_SPECIFIERS      \
        IMAGINARY_SPECIFIERS
-/* TODO: T_IDENTIFIER && typename */
 
 #define DECLARATION_START   \
        STORAGE_CLASSES         \
@@ -462,6 +497,7 @@ void parse_declaration_specifiers(declaration_specifiers_t *specifiers)
        type_t        *type            = NULL;
        unsigned       type_qualifiers = 0;
        unsigned       type_specifiers = 0;
+       int            newtype         = 0;
 
        while(1) {
                switch(token.type) {
@@ -682,6 +718,7 @@ finish_specifiers:
                atomic_type_t *atype = allocate_type_zero(sizeof(atype[0]));
                atype->type.type     = TYPE_ATOMIC;
                atype->atype         = atomic_type;
+               newtype              = 1;
 
                type = (type_t*) atype;
        } else {
@@ -693,15 +730,11 @@ finish_specifiers:
        type->qualifiers = type_qualifiers;
 
        type_t *result = typehash_insert(type);
-       if(result != (type_t*) type) {
+       if(newtype && result != (type_t*) type) {
                obstack_free(type_obst, type);
        }
 
        specifiers->type = result;
-
-       fprintf(stderr, "Specifiers type: ");
-       print_type(stderr, result);
-       fprintf(stderr, "\n");
 }
 
 static
@@ -724,12 +757,22 @@ unsigned parse_type_qualifiers()
 }
 
 static
-type_t *parse_pointer(type_t *type)
+int parse_pointers(void)
 {
+       int result = 0;
+
        while(token.type == '*') {
-               /* pointer */
                next_token();
+               result++;
+       }
 
+       return result;
+}
+
+static
+type_t *make_pointers(type_t *type, int count)
+{
+       for(int i = 0; i < count; ++i) {
                pointer_type_t *pointer_type
                        = allocate_type_zero(sizeof(pointer_type[0]));
                pointer_type->type.type = TYPE_POINTER;
@@ -749,7 +792,7 @@ type_t *parse_pointer(type_t *type)
 }
 
 static
-void parse_identifier_list()
+void parse_identifier_list(void)
 {
        while(1) {
                if(token.type != T_IDENTIFIER) {
@@ -765,41 +808,76 @@ void parse_identifier_list()
 }
 
 static
-void parse_parameter()
+declaration_t *parse_parameter(void)
 {
-       if(token.type == T_DOTDOTDOT) {
-               next_token();
-               return;
-       }
-
        declaration_specifiers_t specifiers;
        memset(&specifiers, 0, sizeof(specifiers));
 
        parse_declaration_specifiers(&specifiers);
-       specifiers.type = parse_pointer(specifiers.type);
 
-       if(token.type == '(' || token.type == T_IDENTIFIER) {
-               declaration_t declaration;
-               memset(&declaration, 0, sizeof(declaration));
-               parse_declarator(&declaration, specifiers.storage_class,
-                                specifiers.type);
+       declaration_t *declaration = allocate_ast_zero(sizeof(declaration[0]));
+       parse_declarator(declaration, specifiers.storage_class,
+                        specifiers.type, 1);
+
+#if 0
+       if(declaration->symbol != NULL) {
+               maybe_push_declaration(declaration);
+               record_declaration(declaration);
        }
+#endif
+
+       return declaration;
 }
 
 static
-void parse_parameters()
+void parse_parameters(method_type_t *type)
 {
        if(token.type == T_IDENTIFIER) {
-               parse_identifier_list();
+               symbol_t      *symbol      = token.v.symbol;
+               declaration_t *declaration = symbol->declaration;
+               if(declaration == NULL
+                               || declaration->storage_class != STORAGE_CLASS_TYPEDEF) {
+                       parse_identifier_list();
+                       return;
+               }
+       }
+
+       if(token.type == ')') {
+               type->unspecified_parameters = 1;
+               return;
+       }
+       if(token.type == T_void && la(1)->type == ')') {
+               next_token();
                return;
        }
 
+       declaration_t *declaration;
+       method_parameter_type_t *parameter_type;
+       method_parameter_type_t *last_parameter_type = NULL;
+
        while(1) {
                switch(token.type) {
                case T_DOTDOTDOT:
+                       next_token();
+                       type->variadic = 1;
+                       return;
+
+               case T_IDENTIFIER:
                DECLARATION_START
-                       parse_parameter();
+                       declaration = parse_parameter();
+
+                       parameter_type = allocate_type_zero(sizeof(parameter_type[0]));
+                       parameter_type->type   = declaration->type;
+                       parameter_type->symbol = declaration->symbol;
+
+                       if(last_parameter_type != NULL) {
+                               last_parameter_type->next = parameter_type;
+                       } else {
+                               type->parameter_types = parameter_type;
+                       }
+                       last_parameter_type = parameter_type;
                        break;
+
                default:
                        return;
                }
@@ -814,7 +892,6 @@ void parse_attributes(void)
 {
        while(token.type == T___attribute__) {
                next_token();
-               fprintf(stderr, "TODO: __attribute__ not handled yet\n");
 
                expect_void('(');
                int depth = 1;
@@ -838,25 +915,156 @@ void parse_attributes(void)
        }
 }
 
+typedef struct declarator_part declarator_part;
+struct declarator_part {
+       int              pointers;
+       method_type_t   *method_type;
+       declarator_part *inner;
+};
+
+static struct obstack  temp_obst;
+
+static
+declarator_part *parse_inner_declarator(declaration_t *declaration)
+{
+       declarator_part *part = obstack_alloc(&temp_obst, sizeof(part[0]));
+       memset(part, 0, sizeof(part[0]));
+
+       part->pointers = parse_pointers();
+
+       switch(token.type) {
+       case T_IDENTIFIER:
+               declaration->symbol          = token.v.symbol;
+               declaration->source_position = token.source_position;
+               next_token();
+               break;
+       case '(':
+               next_token();
+               part->inner = parse_inner_declarator(declaration);
+               expect(')');
+               break;
+       default:
+               parse_error_expected("problem while parsing declarator", T_IDENTIFIER,
+                                    '(', 0);
+       }
+
+       while(1) {
+               switch(token.type) {
+               case '(':
+                       next_token();
+
+#if 0
+                       int         top          = environment_top();
+                       context_t  *last_context = context;
+                       set_context(&declaration->context);
+#endif
+
+                       method_type_t *method_type
+                               = allocate_type_zero(sizeof(method_type[0]));
+                       method_type->type.type   = TYPE_METHOD;
+
+                       parse_parameters(method_type);
+
+#if 0
+                       assert(context == &declaration->context);
+                       set_context(last_context);
+                       environment_pop_to(top);
+#endif
+
+                       part->method_type = method_type;
+
+                       expect(')');
+                       break;
+               case '[':
+                       next_token();
+
+                       if(token.type == T_static) {
+                               next_token();
+                       }
+
+                       unsigned type_qualifiers = parse_type_qualifiers();
+                       if(type_qualifiers != 0) {
+                               if(token.type == T_static) {
+                                       next_token();
+                               }
+                       }
+
+                       /* TODO */
+
+                       if(token.type == '*' && la(1)->type == ']') {
+                               next_token();
+                       } else if(token.type != ']') {
+                               parse_assignment_expression();
+                       }
+
+                       expect(']');
+                       break;
+               default:
+                       goto declarator_finished;
+               }
+       }
+
+declarator_finished:
+       parse_attributes();
+
+       return part;
+}
+
+static
+type_t *construct_declarator_type(declarator_part *part, type_t *type)
+{
+       do {
+               type = make_pointers(type, part->pointers);
+
+               method_type_t *method_type = part->method_type;
+               if(method_type != NULL) {
+                       method_type->result_type = type;
+
+                       type = (type_t*) method_type;
+               }
+
+               part = part->inner;
+       } while(part != NULL);
+
+       return type;
+}
+
+static
+void parse_declarator(declaration_t *declaration, storage_class_t storage_class,
+                      type_t *type, int may_omit_identifier)
+{
+       (void) may_omit_identifier;
+       declarator_part *part = parse_inner_declarator(declaration);
+
+       declaration->type          = construct_declarator_type(part, type);
+       declaration->storage_class = storage_class;
+       obstack_free(&temp_obst, part);
+}
+
+#if 0
 static
 void parse_declarator(declaration_t *declaration, storage_class_t storage_class,
-                      type_t *type)
+                      type_t *type, int may_omit_identifier)
 {
-       type = parse_pointer(type);
+       ir_type *outer_type = parse_pointers(type);
+
        declaration->storage_class = storage_class;
-       declaration->type          = type;
+       declaration->type          = outer_type;
 
        switch(token.type) {
        case T_IDENTIFIER:
-               declaration->symbol = token.v.symbol;
+               declaration->symbol          = token.v.symbol;
+               declaration->source_position = token.source_position;
                next_token();
                break;
        case '(':
                next_token();
-               parse_declarator(declaration, storage_class, type);
+               parse_declarator(declaration, storage_class, type, 0);
                expect_void(')');
                break;
        default:
+               if(may_omit_identifier)
+                       goto declarator_finished;
                parse_error_expected("problem while parsing declarator", T_IDENTIFIER,
                                     '(', 0);
        }
@@ -866,7 +1074,23 @@ void parse_declarator(declaration_t *declaration, storage_class_t storage_class,
                case '(':
                        next_token();
 
-                       parse_parameters();
+                       int         top          = environment_top();
+                       context_t  *last_context = context;
+                       set_context(&declaration->context);
+
+                       method_type_t *method_type
+                               = allocate_type_zero(sizeof(method_type[0]));
+                       method_type->type.type   = TYPE_METHOD;
+                       method_type->result_type = outer_type;
+
+                       parse_parameters(method_type);
+
+                       assert(context == &declaration->context);
+                       set_context(last_context);
+                       environment_pop_to(top);
+
+                       replace_type(&declaration->type, outer_type,
+                                    method_type);
 
                        expect_void(')');
                        break;
@@ -884,7 +1108,7 @@ void parse_declarator(declaration_t *declaration, storage_class_t storage_class,
                                }
                        }
 
-                       if(token.type == '*' /* TODO: && lookahead == ']' */) {
+                       if(token.type == '*' && la(1)->type == ']') {
                                next_token();
                        } else if(token.type != ']') {
                                parse_assignment_expression();
@@ -899,11 +1123,23 @@ void parse_declarator(declaration_t *declaration, storage_class_t storage_class,
 
 declarator_finished:
        parse_attributes();
+}
+#endif
 
-       fprintf(stderr, "Declarator type: ");
-       print_type(stderr, type);
-       fprintf(stderr, "\n");
+static void record_declaration(declaration_t *declaration)
+{
+       if(last_declaration != NULL) {
+               last_declaration->next = declaration;
+       } else {
+               if(context != NULL)
+                       context->declarations = declaration;
+       }
+       last_declaration = declaration;
+}
 
+static
+void maybe_push_declaration(declaration_t *declaration)
+{
        symbol_t *symbol = declaration->symbol;
 
        if(symbol != NULL) {
@@ -918,7 +1154,9 @@ void parse_init_declarators(const declaration_specifiers_t *specifiers)
                declaration_t *declaration = allocate_ast_zero(sizeof(declaration[0]));
 
                parse_declarator(declaration, specifiers->storage_class,
-                                specifiers->type);
+                                specifiers->type, 0);
+               maybe_push_declaration(declaration);
+               record_declaration(declaration);
                if(token.type == '=') {
                        next_token();
                        if(token.type == '{') {
@@ -928,7 +1166,8 @@ void parse_init_declarators(const declaration_specifiers_t *specifiers)
                                parse_assignment_expression();
                        }
                } else if(token.type == '{') {
-                       parse_compound_statement();
+                       statement_t *statement = parse_compound_statement();
+                       declaration->statement = statement;
                        return;
                }
 
@@ -943,21 +1182,22 @@ static
 void parse_struct_declarators(const declaration_specifiers_t *specifiers)
 {
        while(1) {
-               declaration_t declaration;
-               compound_entry_t *entry = allocate_ast_zero(sizeof(entry[0]));
-
                if(token.type == ':') {
                        next_token();
                        parse_constant_expression();
-                       /* TODO */
+                       /* TODO (bitfields) */
                } else {
-                       parse_declarator(&declaration, specifiers->storage_class,
-                                        specifiers->type);
+                       declaration_t *declaration
+                               = allocate_ast_zero(sizeof(declaration[0]));
+                       parse_declarator(declaration, specifiers->storage_class,
+                                        specifiers->type, 0);
+                       maybe_push_declaration(declaration);
+                       record_declaration(declaration);
 
                        if(token.type == ':') {
                                next_token();
                                parse_constant_expression();
-                               /* TODO */
+                               /* TODO (bitfields) */
                        }
                }
 
@@ -968,12 +1208,10 @@ void parse_struct_declarators(const declaration_specifiers_t *specifiers)
        expect_void(';');
 }
 
-static compound_entry_t *parse_compound_type_entries(void)
+static void parse_compound_type_entries(void)
 {
        eat('{');
 
-       compound_entry_t *entries = NULL;
-
        while(token.type != '}' && token.type != T_EOF) {
                declaration_specifiers_t specifiers;
                memset(&specifiers, 0, sizeof(specifiers));
@@ -983,9 +1221,10 @@ static compound_entry_t *parse_compound_type_entries(void)
 
                parse_struct_declarators(&specifiers);
        }
+       if(token.type == T_EOF) {
+               parse_error("unexpected error while parsing struct");
+       }
        next_token();
-
-       return entries;
 }
 
 void parse_declaration(void)
@@ -1008,7 +1247,8 @@ type_t *parse_typename(void)
        /* TODO not correct storage class elements are not allowed here */
        parse_declaration_specifiers(&specifiers);
 
-       specifiers.type = parse_pointer(specifiers.type);
+       int pointers = parse_pointers();
+       specifiers.type = make_pointers(specifiers.type, pointers);
 
        return specifiers.type;
 }
@@ -1682,17 +1922,25 @@ statement_t *parse_compound_statement(void)
 {
        eat('{');
 
-       int top = environment_top();
+       compound_statement_t *compound_statement
+               = allocate_ast_zero(sizeof(compound_statement[0]));
+       compound_statement->statement.type = STATEMENT_COMPOUND;
+
+       int        top          = environment_top();
+       context_t *last_context = context;
+       set_context(&compound_statement->context);
 
        while(token.type != '}') {
                parse_statement();
        }
 
+       assert(context == &compound_statement->context);
+       set_context(last_context);
        environment_pop_to(top);
 
        next_token();
 
-       return NULL;
+       return (statement_t*) compound_statement;
 }
 
 static
@@ -1700,15 +1948,17 @@ translation_unit_t *parse_translation_unit(void)
 {
        translation_unit_t *unit = allocate_ast_zero(sizeof(unit[0]));
 
-       assert(translation_unit == NULL);
        assert(context == NULL);
-       translation_unit = unit;
+       set_context(&unit->context);
 
        while(token.type != T_EOF) {
                parse_declaration();
        }
 
-       translation_unit = NULL;
+       assert(context == &unit->context);
+       context          = NULL;
+       last_declaration = NULL;
+
        return unit;
 }
 
@@ -1732,8 +1982,10 @@ translation_unit_t *parse(void)
 void init_parser(void)
 {
        init_expression_parsers();
+       obstack_init(&temp_obst);
 }
 
 void exit_parser(void)
 {
+       obstack_free(&temp_obst, NULL);
 }