X-Git-Url: http://nsz.repo.hu/git/?a=blobdiff_plain;f=parser.c;h=278c524526c39eea108a77a4206d93f8dac595b6;hb=d1d999b6563a06706be2500a02217986c54ec13f;hp=4d67fc359b96fa5c106a476bd28379df811f2a82;hpb=daab1c4b9ec9146053f6d4f16ca2b92222104ddc;p=cparser diff --git a/parser.c b/parser.c index 4d67fc3..278c524 100644 --- a/parser.c +++ b/parser.c @@ -3,17 +3,44 @@ #include #include -#include "lexer_t.h" +#include "parser.h" +#include "lexer.h" #include "token_t.h" #include "type_t.h" #include "type_hash.h" #include "ast_t.h" #include "adt/bitfiddle.h" #include "adt/error.h" +#include "adt/array.h" -#define PRINT_TOKENS +//#define PRINT_TOKENS +//#define ABORT_ON_ERROR +#define MAX_LOOKAHEAD 2 -static token_t token; +struct environment_entry_t { + 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 context_t *context = NULL; +static declaration_t *last_declaration = NULL; +static struct obstack temp_obst; + +static +statement_t *parse_compound_statement(void); +static +statement_t *parse_statement(void); + +static +expression_t *parse_sub_expression(unsigned precedence); +static +expression_t *parse_expression(void); static inline void *allocate_ast_zero(size_t size) @@ -23,10 +50,33 @@ void *allocate_ast_zero(size_t size) return res; } +static inline +void *allocate_type_zero(size_t size) +{ + void *res = obstack_alloc(type_obst, size); + memset(res, 0, size); + return res; +} + +/** + * returns the top element of the environment stack + */ +static inline +size_t environment_top(void) +{ + return ARR_LEN(environment_stack); +} + + + 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); @@ -34,6 +84,14 @@ void next_token(void) #endif } +static inline +const token_t *la(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) { @@ -41,12 +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 @@ -87,12 +153,12 @@ void parse_error_expected(const char *message, ...) } static -void eat_until_semi(void) +void eat_until(int token_type) { - while(token.type != ';') { - next_token(); + while(token.type != token_type) { if(token.type == T_EOF) return; + next_token(); } next_token(); } @@ -100,7 +166,7 @@ void eat_until_semi(void) #define expect(expected) \ if(UNLIKELY(token.type != (expected))) { \ parse_error_expected(NULL, (expected), 0); \ - eat_until_semi(); \ + eat_until(';'); \ return NULL; \ } \ next_token(); @@ -108,11 +174,320 @@ void eat_until_semi(void) #define expect_void(expected) \ if(UNLIKELY(token.type != (expected))) { \ parse_error_expected(NULL, (expected), 0); \ - eat_until_semi(); \ + eat_until(';'); \ return; \ } \ 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 + * corresponding symbol to the new entry + */ +static inline +void 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) { + if(symbol->declaration != NULL) { + assert(symbol->declaration != NULL); + parser_print_error_prefix_pos(declaration->source_position); + fprintf(stderr, "multiple definitions for symbol '%s'.\n", + symbol->string); + parser_print_error_prefix_pos(symbol->declaration->source_position); + fprintf(stderr, "this is the location of the previous declaration.\n"); + } + } + + entry->old_declaration = symbol->declaration; + entry->old_context = symbol->context; + entry->symbol = symbol; + symbol->declaration = declaration; + symbol->context = context; +} + +/** + * 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) +{ + /* TODO: not correct yet */ + return parse_expression(); +} + +static expression_t *parse_assignment_expression(void) +{ + /* TODO: not correct yet */ + return parse_expression(); +} + +static void parse_compound_type_entries(void); +static void parse_declarator(declaration_t *declaration, + storage_class_t storage_class, type_t *type, + int may_be_abstract); +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 { + storage_class_t storage_class; + type_t *type; +}; + +static type_t *parse_struct_specifier(void) +{ + 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 = token.source_position; + + int top = environment_top(); + context_t *last_context = context; + set_context(&struct_type->context); + + if(token.type == T_IDENTIFIER) { + 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); + struct_type = NULL; + } + + assert(context == &struct_type->context); + set_context(last_context); + environment_pop_to(top); + + return (type_t*) struct_type; +} + +static type_t *parse_union_specifier(void) +{ + 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 = 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(); + if(token.type == '{') { + parse_compound_type_entries(); + } + } else if(token.type == '{') { + parse_compound_type_entries(); + } 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; +} + +static enum_entry_t *parse_enum_type_entries(void) +{ + eat('{'); + + if(token.type == '}') { + next_token(); + parse_error("empty enum not allowed"); + return NULL; + } + + enum_entry_t *result = NULL; + enum_entry_t *last_entry = NULL; + do { + enum_entry_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('}'); + return result; + } + entry->symbol = token.v.symbol; + next_token(); + + if(token.type == '=') { + next_token(); + entry->value = parse_constant_expression(); + } + + if(last_entry != NULL) { + last_entry->next = entry; + } else { + result = entry; + } + last_entry = entry; + + if(token.type != ',') + break; + next_token(); + } while(token.type != '}'); + + expect('}'); + return result; +} + +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 = token.source_position; + + if(token.type == T_IDENTIFIER) { + enum_type->symbol = token.v.symbol; + next_token(); + if(token.type == '{') { + enum_type->entries = parse_enum_type_entries(); + } + } else if(token.type == '{') { + enum_type->entries = parse_enum_type_entries(); + } else { + parse_error_expected("problem while parsing enum type specifiers", + T_IDENTIFIER, '{'); + } + + 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_until(')'); + break; + } else { + parse_string_literals(); + } + expect_void(')'); + break; + default: + goto attributes_finished; + } + } + +attributes_finished: + ; +} typedef enum { SPECIFIER_SIGNED = 1 << 0, @@ -134,28 +509,79 @@ typedef enum { #endif } specifiers_t; -typedef enum { - STORAGE_CLASS_NONE, - STORAGE_CLASS_TYPEDEF, - STORAGE_CLASS_EXTERN, - STORAGE_CLASS_STATIC, - STORAGE_CLASS_AUTO, - STORAGE_CLASS_REGISTER -} storage_class_t; +#define STORAGE_CLASSES \ + case T_typedef: \ + case T_extern: \ + case T_static: \ + case T_auto: \ + case T_register: -typedef struct declaration_specifiers_t declaration_specifiers_t; -struct declaration_specifiers_t { - storage_class_t storage_class; - type_t *type; -}; +#define TYPE_QUALIFIERS \ + case T_const: \ + case T_restrict: \ + case T_volatile: \ + case T_inline: \ + case T___extension__: + +#ifdef PROVIDE_COMPLEX +#define COMPLEX_SPECIFIERS \ + case T__Complex: +#else +#define COMPLEX_SPECIFIERS +#endif + +#ifdef PROVIDE_IMAGINARY +#define IMAGINARY_SPECIFIERS \ + case T__Imaginary: +#else +#define IMAGINARY_SPECIFIERS +#endif + +#define TYPE_SPECIFIERS \ + case T_void: \ + case T_char: \ + case T_short: \ + case T_int: \ + case T_long: \ + case T_float: \ + case T_double: \ + case T_signed: \ + case T_unsigned: \ + case T__Bool: \ + case T_struct: \ + case T_union: \ + case T_enum: \ + COMPLEX_SPECIFIERS \ + IMAGINARY_SPECIFIERS + +#define DECLARATION_START \ + STORAGE_CLASSES \ + TYPE_QUALIFIERS \ + TYPE_SPECIFIERS + +static +type_t *create_builtin_type(symbol_t *symbol) +{ + builtin_type_t *type = allocate_type_zero(sizeof(type[0])); + type->type.type = TYPE_BUILTIN; + type->symbol = symbol; + + type_t *result = typehash_insert((type_t*) type); + if(result != (type_t*) type) { + obstack_free(type_obst, type); + } + + return result; +} static void parse_declaration_specifiers(declaration_specifiers_t *specifiers) { - type_type_t type_type = TYPE_INVALID; - atomic_type_type_t atomic_type = ATOMIC_TYPE_INVALID; - 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) { @@ -189,6 +615,11 @@ void parse_declaration_specifiers(declaration_specifiers_t *specifiers) MATCH_TYPE_QUALIFIER(T_volatile, TYPE_QUALIFIER_VOLATILE); MATCH_TYPE_QUALIFIER(T_inline, TYPE_QUALIFIER_INLINE); + case T___extension__: + /* TODO */ + next_token(); + break; + /* type specifiers */ #define MATCH_SPECIFIER(token, specifier, name) \ case token: \ @@ -218,7 +649,7 @@ void parse_declaration_specifiers(declaration_specifiers_t *specifiers) case T_long: next_token(); if(type_specifiers & SPECIFIER_LONG_LONG) { - parse_error("too many long type specifiers given"); + parse_error("multiple type specifiers given"); } else if(type_specifiers & SPECIFIER_LONG) { type_specifiers |= SPECIFIER_LONG_LONG; } else { @@ -226,10 +657,36 @@ 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(); + break; + case T_union: + type = parse_union_specifier(); + break; case T_enum: + type = parse_enum_specifier(); + break; + case T___builtin_va_list: + type = create_builtin_type(token.v.symbol); + next_token(); + break; + + case T___attribute__: /* TODO */ - assert(0); + parse_attributes(); + break; + + case T_IDENTIFIER: + declaration = token.v.symbol->declaration; + if(declaration == NULL || + declaration->storage_class != STORAGE_CLASS_TYPEDEF) { + goto finish_specifiers; + } + + type = declaration->type; + assert(type != NULL); + next_token(); break; /* function specifier */ @@ -239,7 +696,10 @@ void parse_declaration_specifiers(declaration_specifiers_t *specifiers) } finish_specifiers: - if(type_type == TYPE_INVALID) { + + if(type == NULL) { + atomic_type_type_t atomic_type; + /* match valid basic types */ switch(type_specifiers) { case SPECIFIER_VOID: @@ -283,14 +743,16 @@ finish_specifiers: case SPECIFIER_UNSIGNED | SPECIFIER_LONG | SPECIFIER_INT: atomic_type = ATOMIC_TYPE_ULONG; break; - case SPECIFIER_LONG_LONG: - case SPECIFIER_SIGNED | SPECIFIER_LONG_LONG: - case SPECIFIER_LONG_LONG | SPECIFIER_INT: - case SPECIFIER_SIGNED | SPECIFIER_LONG_LONG | SPECIFIER_INT: + case SPECIFIER_LONG | SPECIFIER_LONG_LONG: + case SPECIFIER_SIGNED | SPECIFIER_LONG | SPECIFIER_LONG_LONG: + case SPECIFIER_LONG | SPECIFIER_LONG_LONG | SPECIFIER_INT: + case SPECIFIER_SIGNED | SPECIFIER_LONG | SPECIFIER_LONG_LONG + | SPECIFIER_INT: atomic_type = ATOMIC_TYPE_LONGLONG; break; - case SPECIFIER_UNSIGNED | SPECIFIER_LONG_LONG: - case SPECIFIER_UNSIGNED | SPECIFIER_LONG_LONG | SPECIFIER_INT: + case SPECIFIER_UNSIGNED | SPECIFIER_LONG | SPECIFIER_LONG_LONG: + case SPECIFIER_UNSIGNED | SPECIFIER_LONG | SPECIFIER_LONG_LONG + | SPECIFIER_INT: atomic_type = ATOMIC_TYPE_ULONGLONG; break; case SPECIFIER_FLOAT: @@ -342,31 +804,30 @@ finish_specifiers: atomic_type = ATOMIC_TYPE_INVALID; } - atomic_type_t *type = obstack_alloc(type_obst, sizeof(type[0])); - memset(type, 0, sizeof(type[0])); - type->type.type = TYPE_ATOMIC; - type->type.qualifiers = type_qualifiers; - type->atype = atomic_type; + atomic_type_t *atype = allocate_type_zero(sizeof(atype[0])); + atype->type.type = TYPE_ATOMIC; + atype->atype = atomic_type; + newtype = 1; - type_t *result = typehash_insert((type_t*) type); - if(result != (type_t*) type) { - obstack_free(type_obst, type); - } - - specifiers->type = result; - - fprintf(stderr, "Specifiers type: "); - print_type(stderr, result); - fprintf(stderr, "\n"); + type = (type_t*) atype; } else { if(type_specifiers != 0) { parse_error("multiple datatypes in declaration"); } } + + type->qualifiers = type_qualifiers; + + type_t *result = typehash_insert(type); + if(newtype && result != (type_t*) type) { + obstack_free(type_obst, type); + } + + specifiers->type = result; } static -unsigned parse_type_qualifiers() +unsigned parse_type_qualifiers(void) { unsigned type_qualifiers = 0; @@ -384,22 +845,44 @@ unsigned parse_type_qualifiers() } } +typedef struct parsed_pointer_t parsed_pointer_t; +struct parsed_pointer_t { + unsigned type_qualifiers; + parsed_pointer_t *next; +}; + static -void parse_declarator(const declaration_specifiers_t *specifiers) +parsed_pointer_t *parse_pointers(void) { - type_t *type = specifiers->type; + 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(); - pointer_type_t *pointer_type - = obstack_alloc(type_obst, sizeof(pointer_type[0])); - memset(pointer_type, 0, sizeof(pointer_type[0])); - pointer_type->type.type = TYPE_POINTER; - pointer_type->points_to = type; + if(last_pointer != NULL) { + last_pointer->next = pointer; + } else { + result = pointer; + } + last_pointer = pointer; + } + + return result; +} - pointer_type->type.qualifiers = parse_type_qualifiers(); +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 = pointer->type_qualifiers; type_t *result = typehash_insert((type_t*) pointer_type); if(result != (type_t*) pointer_type) { @@ -409,48 +892,333 @@ void parse_declarator(const declaration_specifiers_t *specifiers) type = result; } + return type; +} + +static +void parse_identifier_list(void) +{ + while(1) { + if(token.type != T_IDENTIFIER) { + parse_error_expected("problem while parsing parameter identifier " + "list", T_IDENTIFIER, 0); + return; + } + next_token(); + if(token.type != ',') + break; + next_token(); + } +} + +static +declaration_t *parse_parameter(void) +{ + declaration_specifiers_t specifiers; + memset(&specifiers, 0, sizeof(specifiers)); + + parse_declaration_specifiers(&specifiers); + + declaration_t *declaration = allocate_ast_zero(sizeof(declaration[0])); + parse_declarator(declaration, specifiers.storage_class, + specifiers.type, 1); + + return declaration; +} + +static +void parse_parameters(method_type_t *type) +{ + if(token.type == T_IDENTIFIER) { + 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 + 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; + } + if(token.type != ',') + return; + next_token(); + } +} + +typedef struct declarator_part declarator_part; +struct declarator_part { + parsed_pointer_t *pointers; + method_type_t *method_type; + declarator_part *inner; +}; + + +static +declarator_part *parse_inner_declarator(declaration_t *declaration, + int may_be_abstract) +{ + 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_IDENTIFIER: - (void) 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(specifiers); - expect_void(')'); - return; + part->inner = parse_inner_declarator(declaration, may_be_abstract); + expect(')'); + break; default: - parse_error("problem while parsing declarator"); + if(may_be_abstract) + break; + parse_error_expected("problem while parsing declarator", T_IDENTIFIER, + '(', 0); } - if(token.type == '(') { - next_token(); + while(1) { + switch(token.type) { + case '(': + next_token(); - /* parse parameter-type-list or identifier-list */ + method_type_t *method_type + = allocate_type_zero(sizeof(method_type[0])); + method_type->type.type = TYPE_METHOD; - expect_void(')'); - } else if(token.type == '[') { - next_token(); + parse_parameters(method_type); - /* multiple type qualifiers, and static */ + part->method_type = method_type; - /* assignment_expression or '*' or nothing */ + 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 */ - expect_void(']'); + if(token.type == '*' && la(1)->type == ']') { + next_token(); + } else if(token.type != ']') { + parse_assignment_expression(); + } + + expect(']'); + break; + default: + goto declarator_finished; + } } - fprintf(stderr, "Declarator type: "); - print_type(stderr, type); - fprintf(stderr, "\n"); +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_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); + + type_t *result = construct_declarator_type(part, base_type); + obstack_free(&temp_obst, part); + + return result; +} + +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) { + environment_push(declaration, context); + } +} + +static void parse_init_declarators(const declaration_specifiers_t *specifiers) { - parse_declarator(specifiers); - if(token.type == '=') { + while(1) { + 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(); + if(token.type == '{') { + // TODO + expect_void('}'); + } else { + parse_assignment_expression(); + } + } else if(token.type == '{') { + statement_t *statement = parse_compound_statement(); + declaration->statement = statement; + return; + } + + if(token.type != ',') + break; next_token(); - //parse_initialize(); } + expect_void(';'); +} + +static +void parse_struct_declarators(const declaration_specifiers_t *specifiers) +{ + while(1) { + if(token.type == ':') { + next_token(); + parse_constant_expression(); + /* TODO (bitfields) */ + } else { + 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 (bitfields) */ + } + } + + if(token.type != ',') + break; + next_token(); + } + expect_void(';'); +} + +static void parse_compound_type_entries(void) +{ + eat('{'); + + while(token.type != '}' && token.type != T_EOF) { + declaration_specifiers_t specifiers; + memset(&specifiers, 0, sizeof(specifiers)); + /* TODO not correct as this allows storage class stuff... but only + * specifiers and qualifiers sould be allowed here */ + parse_declaration_specifiers(&specifiers); + + parse_struct_declarators(&specifiers); + } + if(token.type == T_EOF) { + parse_error("unexpected error while parsing struct"); + } + next_token(); } void parse_declaration(void) @@ -459,15 +1227,32 @@ void parse_declaration(void) memset(&specifiers, 0, sizeof(specifiers)); parse_declaration_specifiers(&specifiers); + if(token.type == ';') { + next_token(); + return; + } parse_init_declarators(&specifiers); } +type_t *parse_typename(void) +{ + declaration_specifiers_t specifiers; + memset(&specifiers, 0, sizeof(specifiers)); + 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"); + } + + type_t *result = parse_abstract_declarator(specifiers.type); + + return result; +} + -static -expression_t *parse_sub_expression(unsigned precedence); -static -expression_t *parse_expression(void); typedef expression_t* (*parse_expression_function) (unsigned precedence); typedef expression_t* (*parse_expression_infix_function) (unsigned precedence, @@ -504,9 +1289,7 @@ 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; } @@ -537,11 +1320,44 @@ expression_t *parse_reference(void) 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_brace_expression(void) { eat('('); + declaration_t *declaration; + switch(token.type) { + 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(')'); @@ -590,14 +1406,33 @@ expression_t *parse_array_expression(unsigned precedence, return (expression_t*) array_access; } +static +type_t *get_expression_type(const expression_t *expression) +{ + (void) expression; + /* TODO */ + return NULL; +} + static expression_t *parse_sizeof(unsigned precedence) { - (void) precedence; eat(T_sizeof); - /* TODO... */ - return NULL; + sizeof_expression_t *sizeof_expression + = allocate_ast_zero(sizeof(sizeof_expression[0])); + sizeof_expression->expression.type = EXPR_SIZEOF; + + if(token.type == '(' /* && LA1 is type_specifier */) { + next_token(); + sizeof_expression->type = parse_typename(); + expect(')'); + } else { + expression_t *expression = parse_sub_expression(precedence); + sizeof_expression->type = get_expression_type(expression); + } + + return (expression_t*) sizeof_expression; } static @@ -662,6 +1497,23 @@ 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; +} + #define CREATE_UNARY_EXPRESSION_PARSER(token_type, unexpression_type) \ static \ expression_t *parse_##unexpression_type(unsigned precedence) \ @@ -677,14 +1529,14 @@ expression_t *parse_##unexpression_type(unsigned precedence) \ return (expression_t*) unary_expression; \ } -CREATE_UNARY_EXPRESSION_PARSER('-', UNEXPR_NEGATE); -CREATE_UNARY_EXPRESSION_PARSER('+', UNEXPR_PLUS); -CREATE_UNARY_EXPRESSION_PARSER('!', UNEXPR_NOT); -CREATE_UNARY_EXPRESSION_PARSER('*', UNEXPR_DEREFERENCE); -CREATE_UNARY_EXPRESSION_PARSER('&', UNEXPR_TAKE_ADDRESS); -CREATE_UNARY_EXPRESSION_PARSER('~', UNEXPR_BITWISE_NEGATE); -CREATE_UNARY_EXPRESSION_PARSER(T_PLUSPLUS, UNEXPR_PREFIX_INCREMENT); -CREATE_UNARY_EXPRESSION_PARSER(T_MINUSMINUS, UNEXPR_PREFIX_DECREMENT); +CREATE_UNARY_EXPRESSION_PARSER('-', UNEXPR_NEGATE) +CREATE_UNARY_EXPRESSION_PARSER('+', UNEXPR_PLUS) +CREATE_UNARY_EXPRESSION_PARSER('!', UNEXPR_NOT) +CREATE_UNARY_EXPRESSION_PARSER('*', UNEXPR_DEREFERENCE) +CREATE_UNARY_EXPRESSION_PARSER('&', UNEXPR_TAKE_ADDRESS) +CREATE_UNARY_EXPRESSION_PARSER('~', UNEXPR_BITWISE_NEGATE) +CREATE_UNARY_EXPRESSION_PARSER(T_PLUSPLUS, UNEXPR_PREFIX_INCREMENT) +CREATE_UNARY_EXPRESSION_PARSER(T_MINUSMINUS, UNEXPR_PREFIX_DECREMENT) #define CREATE_UNARY_POSTFIX_EXPRESSION_PARSER(token_type, unexpression_type) \ static \ @@ -703,8 +1555,8 @@ expression_t *parse_##unexpression_type(unsigned precedence, \ return (expression_t*) unary_expression; \ } -CREATE_UNARY_POSTFIX_EXPRESSION_PARSER(T_PLUSPLUS, UNEXPR_POSTFIX_INCREMENT); -CREATE_UNARY_POSTFIX_EXPRESSION_PARSER(T_MINUSMINUS, UNEXPR_POSTFIX_DECREMENT); +CREATE_UNARY_POSTFIX_EXPRESSION_PARSER(T_PLUSPLUS, UNEXPR_POSTFIX_INCREMENT) +CREATE_UNARY_POSTFIX_EXPRESSION_PARSER(T_MINUSMINUS, UNEXPR_POSTFIX_DECREMENT) #define CREATE_BINEXPR_PARSER(token_type, binexpression_type) \ static \ @@ -725,22 +1577,24 @@ expression_t *parse_##binexpression_type(unsigned precedence, \ return (expression_t*) binexpr; \ } -CREATE_BINEXPR_PARSER('*', BINEXPR_MUL); -CREATE_BINEXPR_PARSER('/', BINEXPR_DIV); -CREATE_BINEXPR_PARSER('+', BINEXPR_ADD); -CREATE_BINEXPR_PARSER('-', BINEXPR_SUB); -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_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_LESSLESS, BINEXPR_SHIFTLEFT); -CREATE_BINEXPR_PARSER(T_GREATERGREATER, BINEXPR_SHIFTRIGHT); +CREATE_BINEXPR_PARSER('*', BINEXPR_MUL) +CREATE_BINEXPR_PARSER('/', BINEXPR_DIV) +CREATE_BINEXPR_PARSER('+', BINEXPR_ADD) +CREATE_BINEXPR_PARSER('-', BINEXPR_SUB) +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_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) static expression_t *parse_sub_expression(unsigned precedence) @@ -751,7 +1605,7 @@ 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) { @@ -844,6 +1698,9 @@ void init_expression_parsers(void) 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_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, T_EQUAL, 2); register_expression_infix_parser(parse_array_expression, '[', 30); @@ -867,12 +1724,6 @@ void init_expression_parsers(void) } -static -statement_t *parse_compound_statement(void); - -static -statement_t *parse_statement(void); - static statement_t *parse_case_statement(void) { @@ -898,7 +1749,7 @@ static statement_t *parse_label_statement(void) { eat(T_IDENTIFIER); - expect(';'); + expect(':'); parse_statement(); return NULL; @@ -997,7 +1848,11 @@ 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 @@ -1013,16 +1868,36 @@ 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; + if(token.type != ';') { + statement->return_value = parse_expression(); + } expect(';'); + return (statement_t*) statement; +} + +static +statement_t *parse_declaration_statement(void) +{ + parse_declaration(); + return NULL; +} + +static +statement_t *parse_expression_statement(void) +{ + parse_expression(); return NULL; } static statement_t *parse_statement(void) { - statement_t *statement = NULL; + declaration_t *declaration; + statement_t *statement = NULL; /* declaration or statement */ switch(token.type) { @@ -1034,10 +1909,6 @@ statement_t *parse_statement(void) statement = parse_default_statement(); break; - case T_IDENTIFIER: - statement = parse_label_statement(); - break; - case '{': statement = parse_compound_statement(); break; @@ -1081,6 +1952,26 @@ statement_t *parse_statement(void) case ';': statement = NULL; break; + + case T_IDENTIFIER: + if(la(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; + + DECLARATION_START + statement = parse_declaration_statement(); + break; } return statement; @@ -1089,55 +1980,81 @@ statement_t *parse_statement(void) static statement_t *parse_compound_statement(void) { - expect('{'); + eat('{'); - while(token.type != '}') { - parse_statement(); + 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); + + statement_t *last_statement = NULL; + + while(token.type != '}' && token.type != T_EOF) { + statement_t *statement = parse_statement(); + + if(last_statement != NULL) { + last_statement->next = statement; + } else { + compound_statement->statements = statement; + } + 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 -void parse_translation_unit(void) +translation_unit_t *parse_translation_unit(void) { - /* - declaration_specifiers_t specifiers; - memset(&specifiers, 0, sizeof(specifiers)); - parse_declaration_specifiers(&specifiers); - */ + translation_unit_t *unit = allocate_ast_zero(sizeof(unit[0])); - while(token.type != T_EOF) { - if(token.type == '{') { - next_token(); - continue; - } + assert(context == NULL); + set_context(&unit->context); + while(token.type != T_EOF) { parse_declaration(); - /* multiple declarations? */ - - if(token.type == '{') { - parse_compound_statement(); - } else if(token.type == ';') { - next_token(); - } else { - parse_error_expected("while parsing declarations", '{', ';', 0); - } } + + assert(context == &unit->context); + context = NULL; + last_declaration = NULL; + + return unit; } -void parse(void) +translation_unit_t *parse(void) { - next_token(); - parse_translation_unit(); + obstack_init(&environment_obstack); + environment_stack = NEW_ARR_F(environment_entry_t*, 0); + + 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); + obstack_free(&environment_obstack, NULL); + + return unit; } void init_parser(void) { init_expression_parsers(); + obstack_init(&temp_obst); } void exit_parser(void) { + obstack_free(&temp_obst, NULL); }