X-Git-Url: http://nsz.repo.hu/git/?a=blobdiff_plain;f=parser.c;h=0cd07167c02f8ba8331dc5ea97a22344a17bab8c;hb=3486fabdf20fb345ac0b0098071c0cfc56005568;hp=7719371f20c8ca47829af389402199f3ff970d57;hpb=05dbb10d24382cce46842d47c7eab8471a8c8834;p=cparser diff --git a/parser.c b/parser.c index 7719371..0cd0716 100644 --- a/parser.c +++ b/parser.c @@ -4,7 +4,7 @@ #include #include "parser.h" -#include "lexer_t.h" +#include "lexer.h" #include "token_t.h" #include "type_t.h" #include "type_hash.h" @@ -13,20 +13,24 @@ #include "adt/error.h" #include "adt/array.h" -#define PRINT_TOKENS +//#define PRINT_TOKENS +//#define ABORT_ON_ERROR +#define MAX_LOOKAHEAD 2 struct environment_entry_t { - symbol_t *symbol; - environment_entry_t *old_entry; - declaration_t *declaration; - unsigned short old_symbol_ID; + symbol_t *symbol; + declaration_t *old_declaration; + const void *old_context; }; static token_t token; +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 struct obstack temp_obst; static statement_t *parse_compound_statement(void); @@ -54,78 +58,11 @@ void *allocate_type_zero(size_t size) return res; } -/** - * pushs an environment_entry on the environment stack and links the - * corresponding symbol to the new entry - */ -static inline -environment_entry_t *environment_push(symbol_t *symbol) -{ - environment_entry_t *entry - = obstack_alloc(&environment_obstack, sizeof(entry[0])); - memset(entry, 0, sizeof(entry[0])); - - int top = ARR_LEN(environment_stack); - ARR_RESIZE(environment_stack, top + 1); - environment_stack[top] = entry; - - entry->old_entry = symbol->thing; - entry->symbol = symbol; - symbol->thing = entry; - - return entry; -} - -/** - * pops symbols from the environment stack until @p new_top is the top element - */ -static inline -void environment_pop_to(size_t new_top) -{ - environment_entry_t *entry = NULL; - size_t top = ARR_LEN(environment_stack); - size_t i; - - if(new_top == top) - return; - - assert(new_top < top); - i = top; - do { - entry = environment_stack[i - 1]; - - symbol_t *symbol = entry->symbol; - -#if 0 - if(entry->type == ENTRY_LOCAL_VARIABLE - && entry->e.variable->refs == 0) { - variable_declaration_statement_t *variable = entry->e.variable; - print_warning_prefix(env, variable->statement.source_position); - fprintf(stderr, "variable '%s' was declared but never read\n", - symbol->string); - } -#endif - - if(entry->declaration->storage_class == STORAGE_CLASS_TYPEDEF) { - fprintf(stderr, "pop typename '%s'\n", entry->symbol->string); - symbol->ID = entry->old_symbol_ID; - } - - assert(symbol->thing == entry); - symbol->thing = entry->old_entry; - - --i; - } while(i != new_top); - obstack_free(&environment_obstack, entry); - - ARR_SHRINKLEN(environment_stack, (int) new_top); -} - /** * returns the top element of the environment stack */ static inline -size_t environment_top() +size_t environment_top(void) { return ARR_LEN(environment_stack); } @@ -135,7 +72,11 @@ size_t environment_top() static inline void next_token(void) { - lexer_next_token(&token); + token = lookahead_buffer[lookahead_bufpos]; + lookahead_buffer[lookahead_bufpos] = lexer_token; + lexer_next_token(); + + lookahead_bufpos = (lookahead_bufpos+1) % MAX_LOOKAHEAD; #ifdef PRINT_TOKENS print_token(stderr, &token); @@ -143,6 +84,14 @@ void next_token(void) #endif } +static inline +const token_t *look_ahead(int num) +{ + assert(num > 0 && num <= MAX_LOOKAHEAD); + int pos = (lookahead_bufpos+num-1) % MAX_LOOKAHEAD; + return & lookahead_buffer[pos]; +} + static inline void eat(token_type_t type) { @@ -150,12 +99,20 @@ void eat(token_type_t type) next_token(); } -void parser_print_error_prefix(void) +void parser_print_error_prefix_pos(const source_position_t source_position) { fputs(source_position.input_name, stderr); fputc(':', stderr); fprintf(stderr, "%d", source_position.linenr); fputs(": error: ", stderr); +#ifdef ABORT_ON_ERROR + abort(); +#endif +} + +void parser_print_error_prefix(void) +{ + parser_print_error_prefix_pos(token.source_position); } static @@ -195,21 +152,60 @@ void parse_error_expected(const char *message, ...) fprintf(stderr, "\n"); } -static -void eat_until(int token_type) +static void eat_block(void) { - while(token.type != token_type) { + if(token.type == '{') + next_token(); + + while(token.type != '}') { if(token.type == T_EOF) return; + if(token.type == '{') { + eat_block(); + continue; + } next_token(); } - next_token(); + eat('}'); +} + +static void eat_statement(void) +{ + while(token.type != ';') { + if(token.type == T_EOF) + return; + if(token.type == '}') + return; + if(token.type == '{') { + eat_block(); + continue; + } + next_token(); + } + eat(';'); +} + +static void eat_brace(void) +{ + if(token.type == '(') + next_token(); + + while(token.type != ')') { + if(token.type == T_EOF) + return; + if(token.type == '{') { + eat_block(); + continue; + } + next_token(); + } + eat(')'); } #define expect(expected) \ if(UNLIKELY(token.type != (expected))) { \ parse_error_expected(NULL, (expected), 0); \ - eat_until(';'); \ + eat_statement(); \ return NULL; \ } \ next_token(); @@ -217,26 +213,136 @@ void eat_until(int token_type) #define expect_void(expected) \ if(UNLIKELY(token.type != (expected))) { \ parse_error_expected(NULL, (expected), 0); \ - eat_until(';'); \ + eat_statement(); \ return; \ } \ next_token(); -static expression_t *parse_constant_expression(void) +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; +} + +/** + * called when we find a 2nd declarator for an identifier we already have a + * declarator for + */ +static int is_compatible_declaration (declaration_t *declaration, + declaration_t *previous) { /* TODO: not correct yet */ - return parse_expression(); + return declaration->type == previous->type; +} + +/** + * pushs an environment_entry on the environment stack and links the + * corresponding symbol to the new entry + */ +static inline +declaration_t *environment_push(declaration_t *declaration, const void *context) +{ + environment_entry_t *entry + = obstack_alloc(&environment_obstack, sizeof(entry[0])); + memset(entry, 0, sizeof(entry[0])); + + int top = ARR_LEN(environment_stack); + 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) { + declaration_t *previous_declaration = symbol->declaration; + if(symbol->declaration != NULL) { + if(!is_compatible_declaration(declaration, previous_declaration)) { + parser_print_error_prefix_pos(declaration->source_position); + fprintf(stderr, "definition of symbol '%s' with type ", + declaration->symbol->string); + print_type(declaration->type); + fputc('\n', stderr); + parser_print_error_prefix_pos( + previous_declaration->source_position); + fprintf(stderr, "is incompatible with previous declaration " + "of type "); + print_type(previous_declaration->type); + fputc('\n', stderr); + } + return previous_declaration; + } + } + + entry->old_declaration = symbol->declaration; + entry->old_context = symbol->context; + entry->symbol = symbol; + symbol->declaration = declaration; + symbol->context = context; + + return declaration; +} + +/** + * pops symbols from the environment stack until @p new_top is the top element + */ +static inline +void environment_pop_to(size_t new_top) +{ + environment_entry_t *entry = NULL; + size_t top = ARR_LEN(environment_stack); + size_t i; + + if(new_top == top) + return; + + assert(new_top < top); + i = top; + do { + entry = environment_stack[i - 1]; + + symbol_t *symbol = entry->symbol; + + symbol->declaration = entry->old_declaration; + symbol->context = entry->old_context; + + --i; + } while(i != new_top); + obstack_free(&environment_obstack, entry); + + ARR_SHRINKLEN(environment_stack, (int) new_top); +} + + + +static expression_t *parse_constant_expression(void) +{ + /* start parsing at precedence 7 (conditional expression) */ + return parse_sub_expression(7); } static expression_t *parse_assignment_expression(void) { - /* TODO: not correct yet */ - return parse_expression(); + /* start parsing at precedence 2 (assignment expression) */ + return parse_sub_expression(2); } -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_be_abstract); +static declaration_t *record_declaration(declaration_t *declaration); typedef struct declaration_specifiers_t declaration_specifiers_t; struct declaration_specifiers_t { @@ -244,56 +350,88 @@ struct declaration_specifiers_t { type_t *type; }; -static type_t *parse_struct_specifier(void) +static compound_type_t *find_compound_type(compound_type_t *types, + const symbol_t *symbol) { - eat(T_struct); - - compound_type_t *struct_type = allocate_type_zero(sizeof(struct_type[0])); - struct_type->type.type = TYPE_COMPOUND_STRUCT; - struct_type->source_position = source_position; - - if(token.type == T_IDENTIFIER || token.type == T_TYPENAME) { - /* TODO */ - next_token(); - if(token.type == '{') { - parse_compound_type_entries(); - } - } else if(token.type == '{') { - parse_compound_type_entries(); - } else { - parse_error_expected("problem while parsing struct type specifiers", - T_IDENTIFIER, '{', 0); - return NULL; + compound_type_t *type = types; + for( ; type != NULL; type = type->next) { + if(type->symbol == symbol) + return type; } - return (type_t*) struct_type; + return NULL; } -static type_t *parse_union_specifier(void) +static type_t *parse_compound_type_specifier(int is_struct) { - eat(T_union); + if(is_struct) { + eat(T_struct); + } else { + eat(T_union); + } - compound_type_t *union_type = allocate_type_zero(sizeof(union_type[0])); - union_type->type.type = TYPE_COMPOUND_UNION; - union_type->source_position = source_position; + symbol_t *symbol = NULL; + compound_type_t *compound_type = NULL; if(token.type == T_IDENTIFIER) { - union_type->symbol = token.v.symbol; + symbol = token.v.symbol; next_token(); - if(token.type == '{') { - parse_compound_type_entries(); + + if(context != NULL) { + if(is_struct) { + compound_type = find_compound_type(context->structs, symbol); + } else { + compound_type = find_compound_type(context->unions, symbol); + } + } + } else if(token.type != '{') { + if(is_struct) { + parse_error_expected("problem while parsing struct type specifier", + T_IDENTIFIER, '{', 0); + } else { + parse_error_expected("problem while parsing union type specifier", + T_IDENTIFIER, '{', 0); + } + + return NULL; + } + + if(compound_type == NULL) { + compound_type = allocate_type_zero(sizeof(compound_type[0])); + + if(is_struct) { + compound_type->type.type = TYPE_COMPOUND_STRUCT; + } else { + compound_type->type.type = TYPE_COMPOUND_UNION; + } + compound_type->source_position = token.source_position; + compound_type->symbol = symbol; + } + + if(token.type == '{') { + if(compound_type->defined) { + parser_print_error_prefix(); + fprintf(stderr, "multiple definition of %s %s\n", + is_struct ? "struct" : "union", symbol->string); + compound_type->context.declarations = NULL; } - } else if(token.type == '{') { + compound_type->defined = 1; + + int top = environment_top(); + context_t *last_context = context; + set_context(&compound_type->context); + parse_compound_type_entries(); - } else { - parse_error_expected("problem while parsing union type specifiers", - T_IDENTIFIER, '{'); + + assert(context == &compound_type->context); + set_context(last_context); + environment_pop_to(top); } - return (type_t*) union_type; + return (type_t*) compound_type; } -static void parse_enum_type_entries() +static void parse_enum_entries(void) { eat('{'); @@ -304,18 +442,26 @@ static void parse_enum_type_entries() } do { + declaration_t *entry = allocate_ast_zero(sizeof(entry[0])); + if(token.type != T_IDENTIFIER) { parse_error_expected("problem while parsing enum entry", T_IDENTIFIER, 0); - eat_until('}'); + eat_block(); return; } + entry->storage_class = STORAGE_CLASS_ENUM_ENTRY; + entry->symbol = token.v.symbol; + entry->source_position = token.source_position; next_token(); if(token.type == '=') { - parse_constant_expression(); + next_token(); + entry->initializer = parse_constant_expression(); } + record_declaration(entry); + if(token.type != ',') break; next_token(); @@ -324,30 +470,135 @@ static void parse_enum_type_entries() expect_void('}'); } +static enum_type_t *find_enum_type(enum_type_t *types, const symbol_t *symbol) +{ + enum_type_t *type = types; + for( ; type != NULL; type = type->next) { + if(type->symbol == symbol) + return type; + } + + return NULL; +} + static type_t *parse_enum_specifier(void) { eat(T_enum); - enum_type_t *enum_type = allocate_type_zero(sizeof(enum_type[0])); - enum_type->type.type = TYPE_ENUM; - enum_type->source_position = source_position; + symbol_t *symbol = NULL; + enum_type_t *enum_type = NULL; if(token.type == T_IDENTIFIER) { - enum_type->symbol = token.v.symbol; + symbol = token.v.symbol; next_token(); - if(token.type == '{') { - parse_enum_type_entries(); + + if(context != NULL) { + enum_type = find_enum_type(context->enums, symbol); } - } else if(token.type == '{') { - parse_enum_type_entries(); - } else { - parse_error_expected("problem while parsing enum type specifiers", - T_IDENTIFIER, '{'); + } else if(token.type != '{') { + parse_error_expected("problem while parsing enum type specifier", + T_IDENTIFIER, '{', 0); + return NULL; + } + + if(enum_type == NULL) { + enum_type = allocate_type_zero(sizeof(enum_type[0])); + enum_type->type.type = TYPE_ENUM; + enum_type->source_position = token.source_position; + enum_type->symbol = symbol; + } + + if(token.type == '{') { + if(enum_type->defined) { + parser_print_error_prefix(); + fprintf(stderr, "multiple definitions of enum %s\n", + symbol->string); + enum_type->entries_begin = NULL; + enum_type->entries_end = NULL; + } + enum_type->defined = 1; + + declaration_t *before = last_declaration; + + parse_enum_entries(); + + if(before == NULL) { + enum_type->entries_begin = context->declarations; + } else { + enum_type->entries_begin = before->next; + } + enum_type->entries_end = last_declaration; } return (type_t*) enum_type; } +static +const char *parse_string_literals(void) +{ + assert(token.type == T_STRING_LITERAL); + const char *result = token.v.string; + + next_token(); + + while(token.type == T_STRING_LITERAL) { + result = concat_strings(result, token.v.string); + next_token(); + } + + return result; +} + +static +void parse_attributes(void) +{ + while(1) { + switch(token.type) { + case T___attribute__: + next_token(); + + expect_void('('); + int depth = 1; + while(depth > 0) { + switch(token.type) { + case T_EOF: + parse_error("EOF while parsing attribute"); + break; + case '(': + next_token(); + depth++; + break; + case ')': + next_token(); + depth--; + break; + default: + next_token(); + } + } + break; + case T_asm: + next_token(); + expect_void('('); + if(token.type != T_STRING_LITERAL) { + parse_error_expected("while parsing assembler attribute", + T_STRING_LITERAL); + eat_brace(); + break; + } else { + parse_string_literals(); + } + expect_void(')'); + break; + default: + goto attributes_finished; + } + } + +attributes_finished: + ; +} + typedef enum { SPECIFIER_SIGNED = 1 << 0, SPECIFIER_UNSIGNED = 1 << 1, @@ -379,8 +630,7 @@ typedef enum { case T_const: \ case T_restrict: \ case T_volatile: \ - case T_inline: \ - case T___extension__: + case T_inline: #ifdef PROVIDE_COMPLEX #define COMPLEX_SPECIFIERS \ @@ -397,7 +647,6 @@ typedef enum { #endif #define TYPE_SPECIFIERS \ - case T_TYPENAME: \ case T_void: \ case T_char: \ case T_short: \ @@ -437,9 +686,11 @@ type_t *create_builtin_type(symbol_t *symbol) static void parse_declaration_specifiers(declaration_specifiers_t *specifiers) { - type_t *type = NULL; - unsigned type_qualifiers = 0; - unsigned type_specifiers = 0; + declaration_t *declaration; + type_t *type = NULL; + unsigned type_qualifiers = 0; + unsigned type_specifiers = 0; + int newtype = 0; while(1) { switch(token.type) { @@ -515,11 +766,12 @@ void parse_declaration_specifiers(declaration_specifiers_t *specifiers) } break; + /* TODO: if type != NULL for the following rules issue an error */ case T_struct: - type = parse_struct_specifier(); + type = parse_compound_type_specifier(1); break; case T_union: - type = parse_union_specifier(); + type = parse_compound_type_specifier(0); break; case T_enum: type = parse_enum_specifier(); @@ -529,12 +781,19 @@ void parse_declaration_specifiers(declaration_specifiers_t *specifiers) next_token(); break; - case T_TYPENAME: - if(type != NULL || type_specifiers != 0) { + case T___attribute__: + /* TODO */ + parse_attributes(); + break; + + case T_IDENTIFIER: + declaration = token.v.symbol->declaration; + if(declaration == NULL || + declaration->storage_class != STORAGE_CLASS_TYPEDEF) { goto finish_specifiers; } - type = token.v.symbol->thing->declaration->type; + type = declaration->type; assert(type != NULL); next_token(); break; @@ -657,6 +916,7 @@ 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 { @@ -668,19 +928,15 @@ 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 -unsigned parse_type_qualifiers() +unsigned parse_type_qualifiers(void) { unsigned type_qualifiers = 0; @@ -698,19 +954,44 @@ unsigned parse_type_qualifiers() } } +typedef struct parsed_pointer_t parsed_pointer_t; +struct parsed_pointer_t { + unsigned type_qualifiers; + parsed_pointer_t *next; +}; + static -type_t *parse_pointer(type_t *type) +parsed_pointer_t *parse_pointers(void) { + parsed_pointer_t *result = NULL; + parsed_pointer_t *last_pointer = NULL; + while(token.type == '*') { - /* pointer */ next_token(); + parsed_pointer_t *pointer + = obstack_alloc(&temp_obst, sizeof(pointer[0])); + pointer->type_qualifiers = parse_type_qualifiers(); + if(last_pointer != NULL) { + last_pointer->next = pointer; + } else { + result = pointer; + } + last_pointer = pointer; + } + + return result; +} + +static +type_t *make_pointers(type_t *type, parsed_pointer_t *pointer) +{ + for( ; pointer != NULL; pointer = pointer->next) { pointer_type_t *pointer_type = allocate_type_zero(sizeof(pointer_type[0])); - pointer_type->type.type = TYPE_POINTER; - pointer_type->points_to = type; - - pointer_type->type.qualifiers = parse_type_qualifiers(); + pointer_type->type.type = TYPE_POINTER; + pointer_type->points_to = type; + pointer_type->type.qualifiers = pointer->type_qualifiers; type_t *result = typehash_insert((type_t*) pointer_type); if(result != (type_t*) pointer_type) { @@ -724,7 +1005,7 @@ type_t *parse_pointer(type_t *type) } static -void parse_identifier_list() +void parse_identifier_list(void) { while(1) { if(token.type != T_IDENTIFIER) { @@ -740,102 +1021,124 @@ 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 - || token.type == T_TYPENAME) { - 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); + + return declaration; } static -void parse_parameters() +declaration_t *parse_parameters(method_type_t *type) { if(token.type == T_IDENTIFIER) { - parse_identifier_list(); - return; + symbol_t *symbol = token.v.symbol; + declaration_t *declaration = symbol->declaration; + if(declaration == NULL + || declaration->storage_class != STORAGE_CLASS_TYPEDEF) { + /* TODO */ + parse_identifier_list(); + return NULL; + } + } + + if(token.type == ')') { + type->unspecified_parameters = 1; + return NULL; } + if(token.type == T_void && look_ahead(1)->type == ')') { + next_token(); + return NULL; + } + + declaration_t *declarations = NULL; + declaration_t *declaration; + declaration_t *last_declaration = NULL; + method_parameter_t *parameter; + method_parameter_t *last_parameter = NULL; while(1) { switch(token.type) { case T_DOTDOTDOT: + next_token(); + type->variadic = 1; + return declarations; + + case T_IDENTIFIER: + case T___extension__: DECLARATION_START - parse_parameter(); + declaration = parse_parameter(); + + parameter = allocate_type_zero(sizeof(parameter[0])); + parameter->type = declaration->type; + + if(last_parameter != NULL) { + last_declaration->next = declaration; + last_parameter->next = parameter; + } else { + type->parameters = parameter; + declarations = declaration; + } + last_parameter = parameter; + last_declaration = declaration; break; + default: - return; + return declarations; } if(token.type != ',') - return; + return declarations; next_token(); } } -static -void parse_attributes(void) -{ - while(token.type == T___attribute__) { - next_token(); - fprintf(stderr, "TODO: __attribute__ not handled yet\n"); - - expect_void('('); - int depth = 1; - while(depth > 0) { - switch(token.type) { - case T_EOF: - parse_error("EOF while parsing attribute"); - break; - case '(': - next_token(); - depth++; - break; - case ')': - next_token(); - depth--; - break; - default: - next_token(); - } - } - } -} +typedef struct declarator_part declarator_part; +struct declarator_part { + parsed_pointer_t *pointers; + method_type_t *method_type; + declarator_part *inner; +}; + static -void parse_declarator(declaration_t *declaration, storage_class_t storage_class, - type_t *type) +declarator_part *parse_inner_declarator(declaration_t *declaration, + int may_be_abstract) { - type = parse_pointer(type); - declaration->storage_class = storage_class; - declaration->type = type; + declarator_part *part = obstack_alloc(&temp_obst, sizeof(part[0])); + memset(part, 0, sizeof(part[0])); + + part->pointers = parse_pointers(); + + /* TODO: find out if this is correct */ + parse_attributes(); switch(token.type) { - case T_TYPENAME: case T_IDENTIFIER: - declaration->symbol = token.v.symbol; + if(declaration == NULL) { + parse_error("no identifier expected in typename"); + } else { + declaration->symbol = token.v.symbol; + declaration->source_position = token.source_position; + } next_token(); break; case '(': next_token(); - parse_declarator(declaration, storage_class, type); - expect_void(')'); + part->inner = parse_inner_declarator(declaration, may_be_abstract); + expect(')'); break; default: - parse_error_expected("problem while parsing declarator", T_TYPENAME, - T_IDENTIFIER, '(', 0); + if(may_be_abstract) + break; + parse_error_expected("problem while parsing declarator", T_IDENTIFIER, + '(', 0); } while(1) { @@ -843,9 +1146,18 @@ void parse_declarator(declaration_t *declaration, storage_class_t storage_class, case '(': next_token(); - parse_parameters(); + method_type_t *method_type + = allocate_type_zero(sizeof(method_type[0])); + method_type->type.type = TYPE_METHOD; - expect_void(')'); + declaration_t *parameters = parse_parameters(method_type); + if(declaration != NULL) { + declaration->context.declarations = parameters; + } + + part->method_type = method_type; + + expect(')'); break; case '[': next_token(); @@ -861,13 +1173,15 @@ void parse_declarator(declaration_t *declaration, storage_class_t storage_class, } } - if(token.type == '*' /* TODO: && lookahead == ']' */) { + /* TODO */ + + if(token.type == '*' && look_ahead(1)->type == ']') { next_token(); } else if(token.type != ']') { parse_assignment_expression(); } - expect_void(']'); + expect(']'); break; default: goto declarator_finished; @@ -877,44 +1191,152 @@ void parse_declarator(declaration_t *declaration, storage_class_t storage_class, declarator_finished: parse_attributes(); - fprintf(stderr, "Declarator type: "); - print_type(stderr, type); - fprintf(stderr, "\n"); + return part; +} - symbol_t *symbol = declaration->symbol; +static +type_t *construct_declarator_type(declarator_part *part, type_t *type) +{ + do { + type = make_pointers(type, part->pointers); - if(symbol != NULL) { - environment_entry_t *entry = environment_push(symbol); - entry->declaration = declaration; - entry->old_symbol_ID = symbol->ID; + method_type_t *method_type = part->method_type; + if(method_type != NULL) { + method_type->result_type = type; - if(storage_class == STORAGE_CLASS_TYPEDEF) { - symbol->ID = T_TYPENAME; - fprintf(stderr, "typedef '%s'\n", symbol->string); - } else { - symbol->ID = T_IDENTIFIER; + type_t *result = typehash_insert((type_t*) method_type); + if(result != (type_t*) method_type) { + obstack_free(type_obst, method_type); + } + type = result; } + + 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_be_abstract) +{ + declarator_part *part + = parse_inner_declarator(declaration, may_be_abstract); + + if(part != NULL) { + declaration->type = construct_declarator_type(part, type); + declaration->storage_class = storage_class; + obstack_free(&temp_obst, part); + } +} + +static +type_t *parse_abstract_declarator(type_t *base_type) +{ + declarator_part *part = parse_inner_declarator(NULL, 1); + + if(part == NULL) + return NULL; + + type_t *result = construct_declarator_type(part, base_type); + obstack_free(&temp_obst, part); + + return result; +} + +static declaration_t *record_declaration(declaration_t *declaration) +{ + if(context == NULL) + return declaration; + + symbol_t *symbol = declaration->symbol; + if(symbol != NULL) { + declaration_t *alias = environment_push(declaration, context); + if(alias != declaration) + return alias; + } + + if(last_declaration != NULL) { + last_declaration->next = declaration; + } else { + context->declarations = declaration; } + last_declaration = declaration; + + return declaration; +} + +static +void parser_error_multiple_definition(declaration_t *previous, + declaration_t *declaration) +{ + parser_print_error_prefix_pos(declaration->source_position); + fprintf(stderr, "multiple definition of symbol '%s'\n", + declaration->symbol->string); + parser_print_error_prefix_pos(previous->source_position); + fprintf(stderr, "this is the location of the previous " + "definition.\n"); } static void parse_init_declarators(const declaration_specifiers_t *specifiers) { while(1) { - declaration_t *declaration = allocate_ast_zero(sizeof(declaration[0])); + declaration_t *ndeclaration + = allocate_ast_zero(sizeof(ndeclaration[0])); - parse_declarator(declaration, specifiers->storage_class, - specifiers->type); + parse_declarator(ndeclaration, specifiers->storage_class, + specifiers->type, 0); + declaration_t *declaration = record_declaration(ndeclaration); if(token.type == '=') { next_token(); + + /* TODO: check that this is an allowed type (esp. no method type) */ + + if(declaration->initializer != NULL) { + parser_error_multiple_definition(declaration, ndeclaration); + } + if(token.type == '{') { // TODO expect_void('}'); } else { - parse_assignment_expression(); + declaration->initializer = parse_assignment_expression(); } } else if(token.type == '{') { - parse_compound_statement(); + if(declaration->type->type != TYPE_METHOD) { + parser_print_error_prefix(); + fprintf(stderr, "Declarator "); + print_type_ext(declaration->type, declaration->symbol, NULL); + fprintf(stderr, " is not a method type.\n"); + } + + if(declaration->initializer != NULL) { + parser_error_multiple_definition(declaration, ndeclaration); + } + if(ndeclaration != declaration) { + memcpy(&declaration->context, &ndeclaration->context, + sizeof(declaration->context)); + } + + int top = environment_top(); + context_t *last_context = context; + set_context(&declaration->context); + + /* push function parameters */ + declaration_t *parameter = declaration->context.declarations; + for( ; parameter != NULL; parameter = parameter->next) { + environment_push(parameter, context); + } + + statement_t *statement = parse_compound_statement(); + + assert(context == &declaration->context); + set_context(last_context); + environment_pop_to(top); + + declaration->statement = statement; return; } @@ -929,21 +1351,23 @@ 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, 1); + + /* TODO: check for doubled fields */ + record_declaration(declaration); if(token.type == ':') { next_token(); parse_constant_expression(); - /* TODO */ + /* TODO (bitfields) */ } } @@ -954,12 +1378,10 @@ 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)); @@ -969,9 +1391,10 @@ 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) @@ -991,12 +1414,17 @@ type_t *parse_typename(void) { declaration_specifiers_t specifiers; memset(&specifiers, 0, sizeof(specifiers)); - /* TODO not correct storage class elements are not allowed here */ parse_declaration_specifiers(&specifiers); + if(specifiers.storage_class != STORAGE_CLASS_NONE) { + /* TODO: improve error message, user does probably not know what a + * storage class is... + */ + parse_error("typename may not have a storage class"); + } - specifiers.type = parse_pointer(specifiers.type); + type_t *result = parse_abstract_declarator(specifiers.type); - return specifiers.type; + return result; } @@ -1037,9 +1465,7 @@ expression_t *parse_string_const(void) string_literal_t *cnst = allocate_ast_zero(sizeof(cnst[0])); cnst->expression.type = EXPR_STRING_LITERAL; - cnst->value = token.v.string; - - next_token(); + cnst->value = parse_string_literals(); return (expression_t*) cnst; } @@ -1062,25 +1488,107 @@ expression_t *parse_reference(void) { reference_expression_t *ref = allocate_ast_zero(sizeof(ref[0])); - ref->expression.type = EXPR_REFERENCE; - ref->symbol = token.v.symbol; + ref->expression.type = EXPR_REFERENCE; + ref->symbol = token.v.symbol; + + if(ref->symbol->declaration == NULL) { + parser_print_error_prefix(); + fprintf(stderr, "unknown symbol '%s' found.\n", ref->symbol->string); + } + ref->declaration = ref->symbol->declaration; next_token(); return (expression_t*) ref; } +static +expression_t *parse_cast(void) +{ + unary_expression_t *cast = allocate_ast_zero(sizeof(cast[0])); + + cast->expression.type = EXPR_UNARY; + cast->type = UNEXPR_CAST; + cast->expression.source_position = token.source_position; + + type_t *type = parse_typename(); + + expect(')'); + expression_t *value = parse_sub_expression(20); + + cast->expression.datatype = type; + cast->value = value; + + return (expression_t*) cast; +} + +static +expression_t *parse_statement_expression(void) +{ + statement_expression_t *expression + = allocate_ast_zero(sizeof(expression[0])); + expression->expression.type = EXPR_STATEMENT; + expression->statement = parse_compound_statement(); + + expect(')'); + + return (expression_t*) expression; +} + static expression_t *parse_brace_expression(void) { eat('('); + declaration_t *declaration; + switch(token.type) { + case '{': + /* gcc extension: a stement expression */ + return parse_statement_expression(); + + TYPE_QUALIFIERS + TYPE_SPECIFIERS + return parse_cast(); + case T_IDENTIFIER: + declaration = token.v.symbol->declaration; + if(declaration != NULL && + (declaration->storage_class == STORAGE_CLASS_TYPEDEF)) { + return parse_cast(); + } + } + expression_t *result = parse_expression(); expect(')'); return result; } +static +expression_t *parse_function_keyword(void) +{ + eat(T___FUNCTION__); + /* TODO */ + + string_literal_t *expression = allocate_ast_zero(sizeof(expression[0])); + expression->expression.type = EXPR_FUNCTION; + expression->value = "TODO: FUNCTION"; + + return (expression_t*) expression; +} + +static +expression_t *parse_pretty_function_keyword(void) +{ + eat(T___PRETTY_FUNCTION__); + /* TODO */ + + string_literal_t *expression = allocate_ast_zero(sizeof(expression[0])); + expression->expression.type = EXPR_PRETTY_FUNCTION; + expression->value = "TODO: PRETTY FUNCTION"; + + return (expression_t*) expression; +} + static expression_t *parse_primary_expression(void) { @@ -1091,11 +1599,19 @@ expression_t *parse_primary_expression(void) return parse_string_const(); case T_IDENTIFIER: return parse_reference(); + case T___FUNCTION__: + return parse_function_keyword(); + case T___PRETTY_FUNCTION__: + return parse_pretty_function_keyword(); case '(': return parse_brace_expression(); } - /* TODO: error message */ + parser_print_error_prefix(); + fprintf(stderr, "unexpected token "); + print_token(stderr, &token); + fprintf(stderr, "\n"); + eat_statement(); return NULL; } @@ -1131,6 +1647,32 @@ type_t *get_expression_type(const expression_t *expression) return NULL; } +static +int is_declaration_specifier(const token_t *token, int only_type_specifiers) +{ + declaration_t *declaration; + + switch(token->type) { + TYPE_SPECIFIERS + return 1; + case T_IDENTIFIER: + declaration = token->v.symbol->declaration; + if(declaration == NULL) + return 0; + if(declaration->storage_class != STORAGE_CLASS_TYPEDEF) + return 0; + return 1; + STORAGE_CLASSES + TYPE_QUALIFIERS + if(only_type_specifiers) + return 0; + return 1; + + default: + return 0; + } +} + static expression_t *parse_sizeof(unsigned precedence) { @@ -1140,13 +1682,14 @@ expression_t *parse_sizeof(unsigned precedence) = allocate_ast_zero(sizeof(sizeof_expression[0])); sizeof_expression->expression.type = EXPR_SIZEOF; - if(token.type == '(' /* && LA1 is type_specifier */) { + if(token.type == '(' && is_declaration_specifier(look_ahead(1), 1)) { next_token(); sizeof_expression->type = parse_typename(); expect(')'); } else { - expression_t *expression = parse_sub_expression(precedence); - sizeof_expression->type = get_expression_type(expression); + expression_t *expression = parse_sub_expression(precedence); + sizeof_expression->type = get_expression_type(expression); + sizeof_expression->size_expression = expression; } return (expression_t*) sizeof_expression; @@ -1158,7 +1701,7 @@ expression_t *parse_select_expression(unsigned precedence, { (void) precedence; - assert(token.type == '.' || token.type == T_SELECT); + assert(token.type == '.' || token.type == T_MINUSGREATER); next_token(); select_expression_t *select = allocate_ast_zero(sizeof(select[0])); @@ -1196,7 +1739,9 @@ expression_t *parse_call_expression(unsigned precedence, while(1) { call_argument_t *argument = allocate_ast_zero(sizeof(argument[0])); - argument->expression = parse_expression(); + /* we start parsing at precedence 2 so we don't get comma operators + * parsed */ + argument->expression = parse_sub_expression(2); if(last_argument == NULL) { call->arguments = argument; } else { @@ -1214,6 +1759,32 @@ expression_t *parse_call_expression(unsigned precedence, return (expression_t*) call; } +static +expression_t *parse_conditional_expression(unsigned precedence, + expression_t *expression) +{ + eat('?'); + + conditional_expression_t *conditional + = allocate_ast_zero(sizeof(conditional[0])); + conditional->condition = expression; + + conditional->true_expression = parse_expression(); + expect(':'); + conditional->false_expression = parse_sub_expression(precedence); + + return (expression_t*) conditional; +} + +static expression_t *parse_extension(unsigned precedence) +{ + eat(T___extension__); + + /* TODO enable extensions */ + + return parse_sub_expression(precedence); +} + #define CREATE_UNARY_EXPRESSION_PARSER(token_type, unexpression_type) \ static \ expression_t *parse_##unexpression_type(unsigned precedence) \ @@ -1269,14 +1840,15 @@ expression_t *parse_##binexpression_type(unsigned precedence, \ \ binary_expression_t *binexpr \ = allocate_ast_zero(sizeof(binexpr[0])); \ - binexpr->expression.type = EXPR_BINARY; \ - binexpr->type = binexpression_type; \ - binexpr->left = left; \ - binexpr->right = right; \ + binexpr->expression.type = EXPR_BINARY; \ + binexpr->type = binexpression_type; \ + binexpr->left = left; \ + binexpr->right = right; \ \ return (expression_t*) binexpr; \ } +CREATE_BINEXPR_PARSER(',', BINEXPR_COMMA) CREATE_BINEXPR_PARSER('*', BINEXPR_MUL) CREATE_BINEXPR_PARSER('/', BINEXPR_DIV) CREATE_BINEXPR_PARSER('+', BINEXPR_ADD) @@ -1285,14 +1857,26 @@ CREATE_BINEXPR_PARSER('<', BINEXPR_LESS) CREATE_BINEXPR_PARSER('>', BINEXPR_GREATER) CREATE_BINEXPR_PARSER('=', BINEXPR_ASSIGN) CREATE_BINEXPR_PARSER(T_EQUALEQUAL, BINEXPR_EQUAL) -CREATE_BINEXPR_PARSER(T_SLASHEQUAL, BINEXPR_NOTEQUAL) +CREATE_BINEXPR_PARSER(T_EXCLAMATIONMARKEQUAL, BINEXPR_NOTEQUAL) CREATE_BINEXPR_PARSER(T_LESSEQUAL, BINEXPR_LESSEQUAL) CREATE_BINEXPR_PARSER(T_GREATEREQUAL, BINEXPR_GREATEREQUAL) CREATE_BINEXPR_PARSER('&', BINEXPR_BITWISE_AND) CREATE_BINEXPR_PARSER('|', BINEXPR_BITWISE_OR) CREATE_BINEXPR_PARSER('^', BINEXPR_BITWISE_XOR) +CREATE_BINEXPR_PARSER(T_ANDAND, BINEXPR_LOGICAL_AND) +CREATE_BINEXPR_PARSER(T_PIPEPIPE, BINEXPR_LOGICAL_OR) CREATE_BINEXPR_PARSER(T_LESSLESS, BINEXPR_SHIFTLEFT) CREATE_BINEXPR_PARSER(T_GREATERGREATER, BINEXPR_SHIFTRIGHT) +CREATE_BINEXPR_PARSER(T_PLUSEQUAL, BINEXPR_ADD_ASSIGN) +CREATE_BINEXPR_PARSER(T_MINUSEQUAL, BINEXPR_SUB_ASSIGN) +CREATE_BINEXPR_PARSER(T_ASTERISKEQUAL, BINEXPR_MUL_ASSIGN) +CREATE_BINEXPR_PARSER(T_SLASHEQUAL, BINEXPR_DIV_ASSIGN) +CREATE_BINEXPR_PARSER(T_PERCENTEQUAL, BINEXPR_MOD_ASSIGN) +CREATE_BINEXPR_PARSER(T_LESSLESSEQUAL, BINEXPR_SHIFTLEFT_ASSIGN) +CREATE_BINEXPR_PARSER(T_GREATERGREATEREQUAL, BINEXPR_SHIFTRIGHT_ASSIGN) +CREATE_BINEXPR_PARSER(T_ANDEQUAL, BINEXPR_BITWISE_AND_ASSIGN) +CREATE_BINEXPR_PARSER(T_PIPEEQUAL, BINEXPR_BITWISE_OR_ASSIGN) +CREATE_BINEXPR_PARSER(T_CARETEQUAL, BINEXPR_BITWISE_XOR_ASSIGN) static expression_t *parse_sub_expression(unsigned precedence) @@ -1303,7 +1887,7 @@ expression_t *parse_sub_expression(unsigned precedence) expression_parser_function_t *parser = &expression_parsers[token.type]; - source_position_t source_position = source_position; + source_position_t source_position = token.source_position; expression_t *left; if(parser->parser != NULL) { @@ -1383,25 +1967,48 @@ void init_expression_parsers(void) T_LESSLESS, 16); register_expression_infix_parser(parse_BINEXPR_SHIFTRIGHT, T_GREATERGREATER, 16); - register_expression_infix_parser(parse_BINEXPR_ADD, '+', 15); - register_expression_infix_parser(parse_BINEXPR_SUB, '-', 15); - register_expression_infix_parser(parse_BINEXPR_LESS, '<', 14); - register_expression_infix_parser(parse_BINEXPR_GREATER, '>', 14); - register_expression_infix_parser(parse_BINEXPR_LESSEQUAL, T_LESSEQUAL, 14); + register_expression_infix_parser(parse_BINEXPR_ADD, '+', 15); + register_expression_infix_parser(parse_BINEXPR_SUB, '-', 15); + register_expression_infix_parser(parse_BINEXPR_LESS, '<', 14); + register_expression_infix_parser(parse_BINEXPR_GREATER, '>', 14); + register_expression_infix_parser(parse_BINEXPR_LESSEQUAL, T_LESSEQUAL, 14); register_expression_infix_parser(parse_BINEXPR_GREATEREQUAL, - T_GREATEREQUAL, 14); + T_GREATEREQUAL, 14); register_expression_infix_parser(parse_BINEXPR_EQUAL, T_EQUALEQUAL, 13); register_expression_infix_parser(parse_BINEXPR_NOTEQUAL, - T_EXCLAMATIONMARKEQUAL, 13); + T_EXCLAMATIONMARKEQUAL, 13); register_expression_infix_parser(parse_BINEXPR_BITWISE_AND, '&', 12); register_expression_infix_parser(parse_BINEXPR_BITWISE_XOR, '^', 11); register_expression_infix_parser(parse_BINEXPR_BITWISE_OR, '|', 10); - register_expression_infix_parser(parse_BINEXPR_ASSIGN, T_EQUAL, 2); + register_expression_infix_parser(parse_BINEXPR_LOGICAL_AND, T_ANDAND, 9); + register_expression_infix_parser(parse_BINEXPR_LOGICAL_OR, T_PIPEPIPE, 8); + register_expression_infix_parser(parse_conditional_expression, '?', 7); + register_expression_infix_parser(parse_BINEXPR_ASSIGN, '=', 2); + register_expression_infix_parser(parse_BINEXPR_ADD_ASSIGN, T_PLUSEQUAL, 2); + register_expression_infix_parser(parse_BINEXPR_SUB_ASSIGN, T_MINUSEQUAL, 2); + register_expression_infix_parser(parse_BINEXPR_MUL_ASSIGN, + T_ASTERISKEQUAL, 2); + register_expression_infix_parser(parse_BINEXPR_DIV_ASSIGN, T_SLASHEQUAL, 2); + register_expression_infix_parser(parse_BINEXPR_MOD_ASSIGN, + T_PERCENTEQUAL, 2); + register_expression_infix_parser(parse_BINEXPR_SHIFTLEFT_ASSIGN, + T_LESSLESSEQUAL, 2); + register_expression_infix_parser(parse_BINEXPR_SHIFTRIGHT_ASSIGN, + T_GREATERGREATEREQUAL, 2); + register_expression_infix_parser(parse_BINEXPR_BITWISE_AND_ASSIGN, + T_ANDEQUAL, 2); + register_expression_infix_parser(parse_BINEXPR_BITWISE_OR_ASSIGN, + T_PIPEEQUAL, 2); + register_expression_infix_parser(parse_BINEXPR_BITWISE_XOR_ASSIGN, + T_CARETEQUAL, 2); + + register_expression_infix_parser(parse_BINEXPR_COMMA, ',', 1); register_expression_infix_parser(parse_array_expression, '[', 30); register_expression_infix_parser(parse_call_expression, '(', 30); register_expression_infix_parser(parse_select_expression, '.', 30); - register_expression_infix_parser(parse_select_expression, T_SELECT, 30); + register_expression_infix_parser(parse_select_expression, + T_MINUSGREATER, 30); register_expression_infix_parser(parse_UNEXPR_POSTFIX_INCREMENT, T_PLUSPLUS, 30); register_expression_infix_parser(parse_UNEXPR_POSTFIX_DECREMENT, @@ -1416,6 +2023,7 @@ void init_expression_parsers(void) register_expression_parser(parse_UNEXPR_PREFIX_INCREMENT, T_PLUSPLUS, 25); register_expression_parser(parse_UNEXPR_PREFIX_DECREMENT, T_MINUSMINUS, 25); register_expression_parser(parse_sizeof, T_sizeof, 25); + register_expression_parser(parse_extension, T___extension__, 25); } @@ -1423,28 +2031,38 @@ static statement_t *parse_case_statement(void) { eat(T_case); - parse_expression(); + case_label_statement_t *label = allocate_ast_zero(sizeof(label[0])); + label->statement.type = STATEMENT_CASE_LABEL; + label->statement.source_position = token.source_position; + + label->expression = parse_expression(); + expect(':'); - parse_statement(); + label->statement.next = parse_statement(); - return NULL; + return (statement_t*) label; } static statement_t *parse_default_statement(void) { eat(T_default); + + case_label_statement_t *label = allocate_ast_zero(sizeof(label[0])); + label->statement.type = STATEMENT_CASE_LABEL; + label->statement.source_position = token.source_position; + expect(':'); - parse_statement(); + label->statement.next = parse_statement(); - return NULL; + return (statement_t*) label; } static statement_t *parse_label_statement(void) { eat(T_IDENTIFIER); - expect(';'); + expect(':'); parse_statement(); return NULL; @@ -1454,77 +2072,118 @@ static statement_t *parse_if(void) { eat(T_if); + + if_statement_t *statement = allocate_ast_zero(sizeof(statement[0])); + statement->statement.type = STATEMENT_IF; + statement->statement.source_position = token.source_position; + expect('('); - parse_expression(); + statement->condition = parse_expression(); expect(')'); - parse_statement(); + statement->true_statement = parse_statement(); if(token.type == T_else) { next_token(); - parse_statement(); + statement->false_statement = parse_statement(); } - return NULL; + return (statement_t*) statement; } static statement_t *parse_switch(void) { eat(T_switch); + + switch_statement_t *statement = allocate_ast_zero(sizeof(statement[0])); + statement->statement.type = STATEMENT_SWITCH; + statement->statement.source_position = token.source_position; + expect('('); - parse_expression(); + statement->expression = parse_expression(); expect(')'); - parse_statement(); + statement->body = parse_statement(); - return NULL; + return (statement_t*) statement; } static statement_t *parse_while(void) { eat(T_while); + + while_statement_t *statement = allocate_ast_zero(sizeof(statement[0])); + statement->statement.type = STATEMENT_WHILE; + statement->statement.source_position = token.source_position; + expect('('); - parse_expression(); + statement->condition = parse_expression(); expect(')'); - parse_statement(); + statement->body = parse_statement(); - return NULL; + return (statement_t*) statement; } static statement_t *parse_do(void) { eat(T_do); - parse_statement(); + + do_while_statement_t *statement = allocate_ast_zero(sizeof(statement[0])); + statement->statement.type = STATEMENT_DO_WHILE; + statement->statement.source_position = token.source_position; + + statement->body = parse_statement(); expect(T_while); expect('('); - parse_expression(); + statement->condition = parse_expression(); expect(')'); + expect(';'); - return NULL; + return (statement_t*) statement; } static statement_t *parse_for(void) { eat(T_for); + + for_statement_t *statement = allocate_ast_zero(sizeof(statement[0])); + statement->statement.type = STATEMENT_FOR; + statement->statement.source_position = token.source_position; + expect('('); + + int top = environment_top(); + context_t *last_context = context; + set_context(&statement->context); + if(token.type != ';') { - /* TODO not correct... this could also be a declaration */ - parse_expression(); + if(is_declaration_specifier(&token, 0)) { + parse_declaration(); + } else { + statement->initialisation = parse_expression(); + expect(';'); + } + } else { + expect(';'); } - expect(';'); + if(token.type != ';') { - parse_expression(); + statement->condition = parse_expression(); } expect(';'); if(token.type != ')') { - parse_expression(); + statement->step = parse_expression(); } expect(')'); - parse_statement(); + statement->body = parse_statement(); - return NULL; + assert(context == &statement->context); + set_context(last_context); + environment_pop_to(top); + + return (statement_t*) statement; } static @@ -1543,7 +2202,11 @@ statement_t *parse_continue(void) eat(T_continue); expect(';'); - return NULL; + statement_t *statement = allocate_ast_zero(sizeof(statement[0])); + statement->source_position = token.source_position; + statement->type = STATEMENT_CONTINUE; + + return statement; } static @@ -1552,30 +2215,79 @@ statement_t *parse_break(void) eat(T_break); expect(';'); - return NULL; + statement_t *statement = allocate_ast_zero(sizeof(statement[0])); + statement->source_position = token.source_position; + statement->type = STATEMENT_BREAK; + + return statement; } static statement_t *parse_return(void) { eat(T_return); - parse_expression(); + + return_statement_t *statement = allocate_ast_zero(sizeof(statement[0])); + + statement->statement.type = STATEMENT_RETURN; + statement->statement.source_position = token.source_position; + if(token.type != ';') { + statement->return_value = parse_expression(); + } expect(';'); - return NULL; + return (statement_t*) statement; } static statement_t *parse_declaration_statement(void) { - parse_declaration(); - return NULL; + declaration_t *before = last_declaration; + + declaration_statement_t *statement + = allocate_ast_zero(sizeof(statement[0])); + statement->statement.type = STATEMENT_DECLARATION; + statement->statement.source_position = token.source_position; + + declaration_specifiers_t specifiers; + memset(&specifiers, 0, sizeof(specifiers)); + parse_declaration_specifiers(&specifiers); + + if(token.type == ';') { + eat(';'); + } else { + parse_init_declarators(&specifiers); + } + + if(before == NULL) { + statement->declarations_begin = context->declarations; + } else { + statement->declarations_begin = before->next; + } + statement->declarations_end = last_declaration; + + return (statement_t*) statement; +} + +static +statement_t *parse_expression_statement(void) +{ + expression_statement_t *statement = allocate_ast_zero(sizeof(statement[0])); + statement->statement.type = STATEMENT_EXPRESSION; + statement->statement.source_position = token.source_position; + + statement->expression = parse_expression(); + + expect(';'); + + return (statement_t*) statement; } static statement_t *parse_statement(void) { - statement_t *statement = NULL; + declaration_t *declaration; + statement_t *statement = NULL; /* declaration or statement */ switch(token.type) { @@ -1587,10 +2299,6 @@ statement_t *parse_statement(void) statement = parse_default_statement(); break; - case T_IDENTIFIER: - statement = parse_label_statement(); - break; - case '{': statement = parse_compound_statement(); break; @@ -1632,14 +2340,46 @@ statement_t *parse_statement(void) break; case ';': + next_token(); statement = NULL; break; + case T_IDENTIFIER: + if(look_ahead(1)->type == ':') { + statement = parse_label_statement(); + break; + } + + declaration = token.v.symbol->declaration; + if(declaration != NULL && + declaration->storage_class == STORAGE_CLASS_TYPEDEF) { + statement = parse_declaration_statement(); + break; + } + + statement = parse_expression_statement(); + break; + + case T___extension__: + /* this can be a prefix to a declaration or an expression statement */ + /* we simply eat it now and parse the rest with tail recursion */ + do { + next_token(); + } while(token.type == T___extension__); + statement = parse_statement(); + break; + DECLARATION_START statement = parse_declaration_statement(); break; + + default: + statement = parse_expression_statement(); + break; } + assert(statement == NULL || statement->source_position.input_name != NULL); + return statement; } @@ -1648,17 +2388,41 @@ 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; + compound_statement->statement.source_position = token.source_position; - while(token.type != '}') { - parse_statement(); + int top = environment_top(); + context_t *last_context = context; + set_context(&compound_statement->context); + + statement_t *last_statement = NULL; + + while(token.type != '}' && token.type != T_EOF) { + statement_t *statement = parse_statement(); + if(statement == NULL) + continue; + + if(last_statement != NULL) { + last_statement->next = statement; + } else { + compound_statement->statements = statement; + } + + while(statement->next != NULL) + statement = statement->next; + + last_statement = statement; } + assert(context == &compound_statement->context); + set_context(last_context); environment_pop_to(top); next_token(); - return NULL; + return (statement_t*) compound_statement; } static @@ -1666,15 +2430,17 @@ 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; } @@ -1683,7 +2449,12 @@ translation_unit_t *parse(void) obstack_init(&environment_obstack); environment_stack = NEW_ARR_F(environment_entry_t*, 0); - next_token(); + type_set_output(stderr); + + lookahead_bufpos = 0; + for(int i = 0; i < MAX_LOOKAHEAD + 2; ++i) { + next_token(); + } translation_unit_t *unit = parse_translation_unit(); DEL_ARR_F(environment_stack); @@ -1695,8 +2466,10 @@ 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); }