diff options
| -rw-r--r-- | boot/stage12.elna | 234 | ||||
| -rw-r--r-- | boot/stage13.elna | 2037 |
2 files changed, 2238 insertions, 33 deletions
diff --git a/boot/stage12.elna b/boot/stage12.elna index 98751bb..68f5592 100644 --- a/boot/stage12.elna +++ b/boot/stage12.elna @@ -4,6 +4,9 @@ (* Stage 12 compiler. *) +(* - Local variables and parameters are saved in a local symbol table. *) +(* - Local variables and parameters can be referenced by their name in the symbol table. *) + const symbol_builtin_name_int := "Int"; symbol_builtin_name_word := "Word"; @@ -26,6 +29,7 @@ const (* INFO_TYPE = 1 *) (* INFO_PARAMETER = 2 *) + (* INFO_TEMPORARY = 3 *) (* Type info has the type it belongs to. *) symbol_type_info_int := S(1, @symbol_builtin_type_int); @@ -722,11 +726,25 @@ end; proc _compile_designator(); begin + v0 := _read_token(); + v4 := _symbol_table_lookup(@symbol_table_local, source_code_position, v0); + + if v4 <> 0 then + _write_z("\taddi t0, sp, \0"); + v8 := _parameter_info_get_offset(v4); + _write_i(v8); + _write_c('\n'); + _advance_token(v0); + + goto .compile_designator_end; + end; if _load_byte(source_code_position) = 'v' then _compile_local_designator(); - else - _compile_global_designator(); + goto .compile_designator_end; end; + _compile_global_designator(); + +.compile_designator_end: end; proc _compile_assignment(); @@ -834,15 +852,15 @@ begin _compile_call(); goto .compile_statement_semicolon; end; - if v0 = 'g' then + if _memcmp(source_code_position, "goto ", 5) = 0 then _compile_goto(); goto .compile_statement_semicolon; end; - if v0 = 'i' then + if _memcmp(source_code_position, "if ", 3) = 0 then _compile_if(); goto .compile_statement_semicolon; end; - if v0 = 'r' then + if _memcmp(source_code_position, "return ", 7) = 0 then _compile_return_statement(); _write_c('\n'); @@ -891,6 +909,15 @@ begin _write_c(v84); end; +proc _skip_spaces(); +begin + v0 := _load_byte(source_code_position); + if v0 = '\t' then + _advance_token(1); + _skip_spaces(); + end; +end; + proc _read_type_expression(); begin v0 := _read_token(); @@ -900,13 +927,58 @@ end; (* Parameters: *) (* a0 - Parameter index. *) -proc _create_parameter_info(v88: Word); +proc _parameter_info_create(v88: Word); begin + v8 := memory_free_pointer; + v4 := v8; + (* 2 is INFO_PARAMETER *) + _store_word(2, v4); + + v4 := v4 + 4; + (* Calculate the stack offset: 88 - (4 * parameter_counter) *) v0 := v88 * 4; v0 := 88 + -v0; + _store_word(v0, v4); - return v0 + v4 := v4 + 4; + _store_word(v4, @memory_free_pointer); + + return v8 +end; + +proc _parameter_info_get_offset(v88: Word); +begin + v88 := v88 + 4; + return _load_word(v88) +end; + +(* Parameters: *) + +(* a0 - Parameter index. *) +proc _temporary_info_create(v88: Word); +begin + v8 := memory_free_pointer; + v4 := v8; + (* 3 is INFO_TEMPORARY *) + _store_word(3, v4); + + v4 := v4 + 4; + + (* Calculate the stack offset: 4 * variable_counter. *) + v0 := v88 * 4; + _store_word(v0, v4); + + v4 := v4 + 4; + _store_word(v4, @memory_free_pointer); + + return v8 +end; + +proc _temporary_info_get_offset(v88: Word); +begin + v88 := v88 + 4; + return _load_word(v88) end; (* Parameters: *) @@ -915,6 +987,7 @@ end; proc _read_procedure_parameter(v88: Word); begin (* Read the parameter name. *) + v8 := source_code_position; v0 := _read_token(); _advance_token(v0); @@ -927,8 +1000,10 @@ begin _write_i(v88); _write_z(", \0"); - (* Calculate the stack offset: 88 - (4 * parameter_counter) *) - v4 := _create_parameter_info(v88); + v4 := _parameter_info_create(v88); + _symbol_table_enter(@symbol_table_local, v8, v0, v4); + + v4 := _parameter_info_get_offset(v4); _write_i(v4); _write_z("(sp)\n\0"); @@ -954,10 +1029,52 @@ begin _advance_token(1); end; +(* Parameters: *) +(* a0 - Variable index. *) +proc _read_procedure_temporary(v88: Word); +begin + _skip_spaces(); + v8 := source_code_position; + + (* Read and skip variable name, colon and the space *) + v0 := _read_token(); + _advance_token(v0 + 2); + + _read_type_expression(); + + v4 := _temporary_info_create(v88); + _symbol_table_enter(@symbol_table_local, v8, v0, v4); + + (* Skip semicolon and newline after the variable declaration *) + _advance_token(2); +end; + +proc _read_procedure_temporaries(); +begin + if _memcmp(source_code_position, "var", 3) <> 0 then + goto .read_local_variables_end; + end; + _advance_token(4); + v0 := 0; + +.read_local_variables_loop: + if _memcmp(source_code_position, "begin", 5) = 0 then + goto .read_local_variables_end; + end; + _read_procedure_temporary(v0); + + v0 := v0 + 1; + goto .read_local_variables_loop; + +.read_local_variables_end: +end; + proc _compile_procedure(); begin (* Skip "proc ". *) _advance_token(5); + (* Clear local symbol table. *) + _store_word(0, @symbol_table_local); (* Save the procedure name length. *) v0 := _read_token(); @@ -977,8 +1094,12 @@ begin _write_z("\taddi sp, sp, -128\n\tsw ra, 124(sp)\n\tsw s0, 120(sp)\n\taddi s0, sp, 128\n\0"); _read_procedure_parameters(); + (* Skip semicolon and newline. *) + _advance_token(2); + _read_procedure_temporaries(); + (* Skip semicolon, "begin" and newline. *) - _advance_token(8); + _advance_token(6); _compile_procedure_body(); @@ -989,15 +1110,6 @@ begin _advance_token(5); end; -proc _skip_spaces(); -begin - v0 := _load_byte(source_code_position); - if v0 = '\t' then - _advance_token(1); - _skip_spaces(); - end; -end; - (* Prints and skips a line. *) proc _skip_comment(); begin @@ -1277,37 +1389,93 @@ begin _syscall(0, 0, 0, 0, 0, 0, 93); end; +(* Looks for a symbol in the given symbol table. *) + +(* Parameters: *) +(* a0 - Symbol table. *) +(* a1 - Symbol name pointer. *) +(* a2 - Symbol name length. *) + +(* Returns the symbol pointer or 0 in a0. *) +proc _symbol_table_lookup(v88: Word, v84: Word, v80: Word); +begin + v0 := 0; + + (* The first word in the symbol table is its length, get it. *) + v4 := _load_word(v88); + + (* Go to the first symbol position. *) + v88 := v88 + 4; + + .symbol_table_lookup_loop; + if v4 = 0 then + goto .symbol_table_lookup_end; + end; + + (* Symbol name pointer and length. *) + v8 := _load_word(v88); + v12 := _load_word(v88 + 4); + + (* If lengths don't match, exit and return nil. *) + if v80 <> v12 then + goto .symbol_table_lookup_repeat; + end; + (* If names don't match, exit and return nil. *) + if _memcmp(v84, v8, v80) <> 0 then + goto .symbol_table_lookup_repeat; + end; + (* Otherwise, the symbol is found. *) + v0 := _load_word(v88 + 8); + goto .symbol_table_lookup_end; + + .symbol_table_lookup_repeat; + v88 := v88 + 12; + v4 := v4 + -1; + goto .symbol_table_lookup_loop; + + .symbol_table_lookup_end; + return v0 +end; + (* Inserts a symbol into the table. *) (* Parameters: *) -(* a0 - Symbol pointer. *) -(* a1 - Symbol name length. *) -(* a2 - Symbol name pointer. *) -(* a3 - Symbol table. *) +(* a0 - Symbol table. *) +(* a1 - Symbol name pointer. *) +(* a2 - Symbol name length. *) +(* a3 - Symbol pointer. *) proc _symbol_table_enter(v88: Word, v84: Word, v80: Word, v76: Word); begin (* The first word in the symbol table is its length, get it. *) - v0 := _load_word(v76); + v0 := _load_word(v88); (* Calculate the offset for the new symbol. *) - v4 := v0 * 4; + v4 := v0 * 12; v4 := v4 + 4; - v4 := v76 + 4; + v4 := v88 + v4; - _memcpy(v4, @v80, 12); + _store_word(v84, v4); + v4 := v4 + 4; + _store_word(v80, v4); + v4 := v4 + 4; + _store_word(v76, v4); (* Increment the symbol table length. *) v0 := v0 + 1; - _store_word(v0, v76); + _store_word(v0, v88); end; proc _symbol_table_build(); begin - _symbol_table_enter(@symbol_type_info_int, 3, symbol_builtin_name_int, @symbol_table_global); - _symbol_table_enter(@symbol_type_info_word, 4, symbol_builtin_name_word, @symbol_table_global); - _symbol_table_enter(@symbol_type_info_pointer, 7, symbol_builtin_name_pointer, @symbol_table_global); - _symbol_table_enter(@symbol_type_info_char, 4, symbol_builtin_name_char, @symbol_table_global); - _symbol_table_enter(@symbol_type_info_bool, 4, symbol_builtin_name_bool, @symbol_table_global); + (* Set the table length to 0. *) + _store_word(0, @symbol_table_global); + + (* Enter built-in symbols. *) + _symbol_table_enter(@symbol_table_global, symbol_builtin_name_int, 3, @symbol_type_info_int); + _symbol_table_enter(@symbol_table_global, symbol_builtin_name_word, 4, @symbol_type_info_word); + _symbol_table_enter(@symbol_table_global, symbol_builtin_name_pointer, 7, @symbol_type_info_pointer); + _symbol_table_enter(@symbol_table_global, symbol_builtin_name_char, 4, @symbol_type_info_char); + _symbol_table_enter(@symbol_table_global, symbol_builtin_name_bool, 4, @symbol_type_info_bool); end; diff --git a/boot/stage13.elna b/boot/stage13.elna new file mode 100644 index 0000000..6bdb58d --- /dev/null +++ b/boot/stage13.elna @@ -0,0 +1,2037 @@ +(* This Source Code Form is subject to the terms of the Mozilla Public License, *) +(* v. 2.0. If a copy of the MPL was not distributed with this file, You can *) +(* obtain one at https://mozilla.org/MPL/2.0/. *) + +(* Stage 13 compiler. *) + +const + symbol_builtin_name_int := "Int"; + symbol_builtin_name_word := "Word"; + symbol_builtin_name_pointer := "Pointer"; + symbol_builtin_name_char := "Char"; + symbol_builtin_name_bool := "Bool"; + + (* Every type info starts with a word describing what type it is. *) + + (* PRIMITIVE_TYPE = 1 *) + + (* Primitive types have only type size. *) + symbol_builtin_type_int := S(1, 4); + symbol_builtin_type_word := S(1, 4); + symbol_builtin_type_pointer := S(1, 4); + symbol_builtin_type_char := S(1, 1); + symbol_builtin_type_bool := S(1, 1); + + (* Info objects start with a word describing its type. *) + + (* INFO_TYPE = 1 *) + (* INFO_PARAMETER = 2 *) + (* INFO_TEMPORARY = 3 *) + + (* Type info has the type it belongs to. *) + symbol_type_info_int := S(1, @symbol_builtin_type_int); + symbol_type_info_word := S(1, @symbol_builtin_type_word); + symbol_type_info_pointer := S(1, @symbol_builtin_type_pointer); + symbol_type_info_char := S(1, @symbol_builtin_type_char); + symbol_type_info_bool := S(1, @symbol_builtin_type_bool); + +var + source_code: Array; + compiler_strings: Array; + symbol_table_global: Array; + symbol_table_local: Array; + classification: Array; + memory: Array; + + compiler_strings_position: Pointer := @compiler_strings; + compiler_strings_length: Word := 0; + label_counter: Word := 0; + source_code_position: Pointer := @source_code; + memory_free_pointer: Word := @memory; + +(* Calculates and returns the string token length between quotes, including the *) +(* escaping slash characters. *) + +(* Parameters: *) +(* string - String token pointer. *) + +(* Returns the length in a0. *) +proc _string_length(string: Word); +var + counter: Word; +begin + (* Reset the counter. *) + counter := 0; + + .string_length_loop; + string := string + 1; + + if _load_byte(string) <> '"' then + counter := counter + 1; + goto .string_length_loop; + end; + + return counter +end; + +(* Adds a string to the global, read-only string storage. *) + +(* Parameters: *) +(* string - String token. *) + +(* Returns the offset from the beginning of the storage to the new string in a0. *) +proc _add_string(string: Word); +var + contents: Word; + result: Word; + current_byte: Word; +begin + contents := string + 1; + result := compiler_strings_length; + + .add_string_loop; + if _load_byte(contents) <> '"' then + current_byte := _load_byte(contents); + _store_byte(current_byte, compiler_strings_position); + compiler_strings_position := compiler_strings_position + 1; + contents := contents + 1; + + if current_byte <> '\\' then + compiler_strings_length := compiler_strings_length + 1; + end; + goto .add_string_loop; + end; + + return result +end; + +(* Reads standard input into a buffer. *) +(* buffer - Buffer pointer. *) +(* size - Buffer size. *) + +(* Returns the amount of bytes written in a0. *) +proc _read_file(buffer: Word, size: Word); +begin + _syscall(0, buffer, size, 0, 0, 0, 63); +end; + +(* Writes to the standard output. *) + +(* Parameters: *) +(* buffer - Buffer. *) +(* size - Buffer length. *) +proc _write_s(buffer: Word, size: Word); +begin + _syscall(1, buffer, size, 0, 0, 0, 64); +end; + +(* Writes a number to a string buffer. *) + +(* Parameters: *) +(* number - Whole number. *) +(* output_buffer - Buffer pointer. *) + +(* Sets a0 to the length of the written number. *) +proc _print_i(number: Word, output_buffer: Word); +var + local_buffer: Word; + is_negative: Word; + current_character: Word; + result: Word; +begin + local_buffer := @result + 11; + + if number >= 0 then + is_negative := 0; + else + number = -number; + is_negative := 1; + end; + + .print_i_digit10; + current_character := number % 10; + _store_byte(current_character + '0', local_buffer); + + number := number / 10; + local_buffer := local_buffer + -1; + + if number <> 0 then + goto .print_i_digit10; + end; + if is_negative = 1 then + _store_byte('-', local_buffer); + local_buffer := local_buffer + -1; + end; + result := @result + 11; + result := result + -local_buffer; + _memcpy(output_buffer, local_buffer + 1, result); + + return result +end; + +(* Writes a number to the standard output. *) + +(* Parameters: *) +(* number - Whole number. *) +proc _write_i(number: Word); +var + local_buffer: Word; + length: Word; +begin + length := _print_i(number, @local_buffer); + _write_s(@local_buffer, length); +end; + +(* Writes a character from a0 into the standard output. *) + +(* Parameters: *) +(* character - Character to write. *) +proc _write_c(character: Word); +begin + _write_s(@character, 1); +end; + +(* Write null terminated string. *) + +(* Parameters: *) +(* string - String. *) +proc _write_z(string: Word); +var + next_byte: Word; +begin + (* Check for 0 character. *) + next_byte := _load_byte(string); + + if next_byte <> 0 then + (* Print a character. *) + _write_c(next_byte); + + (* Advance the input string by one byte. *) + _write_z(string + 1); + end; +end; + +(* Detects if a0 is an uppercase character. Sets a0 to 1 if so, otherwise to 0. *) +proc _is_upper(character: Word); +var + lhs: Word; + rhs: Word; +begin + lhs := character >= 'A'; + rhs := character <= 'Z'; + + return lhs & rhs + +end; + +(* Detects if a0 is an lowercase character. Sets a0 to 1 if so, otherwise to 0. *) +proc _is_lower(character: Word); +var + lhs: Word; + rhs: Word; +begin + lhs := character >= 'a'; + rhs := character <= 'z'; + + return lhs & rhs +end; + +(* Detects if the passed character is a 7-bit alpha character or an underscore. *) + +(* Paramters: *) +(* character - Tested character. *) + +(* Sets a0 to 1 if the character is an alpha character or underscore, sets it to 0 otherwise. *) +proc _is_alpha(character: Word); +var + is_upper_result: Word; + is_lower_result: Word; + is_alpha_result: Word; + is_underscore: Word; +begin + is_upper_result := _is_upper(character); + is_lower_result := _is_lower(character); + is_underscore := character = '_'; + + is_alpha_result := is_lower_result or is_upper_result; + return is_alpha_result or is_underscore +end; + +(* Detects whether the passed character is a digit *) +(* (a value between 0 and 9). *) + +(* Parameters: *) +(* character - Exemined value. *) + +(* Sets a0 to 1 if it is a digit, to 0 otherwise. *) +proc _is_digit(character: Word); +var + lhs: Word; + rhs: Word; +begin + lhs := character >= '0'; + rhs := character <= '9'; + + return lhs & rhs +end; + +proc _is_alnum(character: Word); +var + lhs: Word; + rhs: Word; +begin + lhs := _is_alpha(character); + rhs := _is_digit(character); + + return lhs or rhs +end; + +(* Reads the next token. *) + +(* Returns token length in a0. *) +proc _read_token(); +var + current_token_position: Word; + token_length: Word; + current_character: Word; + is_dot: Word; + is_alnum_result: Word; +begin + current_token_position := source_code_position; + token_length := 0; + + .read_token_loop; + (* Current character. *) + current_character := _load_byte(current_token_position); + + (* First we try to read a derictive. *) + (* A derictive can contain a dot and characters. *) + is_dot := current_character = '.'; + is_alnum_result := _is_alnum(current_character); + + if is_dot or is_alnum_result then + (* Advance the source code position and token length. *) + token_length := token_length + 1; + current_token_position := current_token_position + 1; + + goto .read_token_loop; + end; + + return token_length +end; + +(* Parameters: *) +(* lhs - First pointer. *) +(* rhs - Second pointer. *) +(* count - The length to compare. *) + +(* Returns 0 if memory regions are equal. *) +proc _memcmp(lhs: Word, rhs: Word, count: Word); +var + lhs_byte: Word; + rhs_byte: Word; + result: Word; +begin + result := 0; + + .memcmp_loop; + if count <> 0 then + lhs_byte := _load_byte(lhs); + rhs_byte := _load_byte(rhs); + result := lhs_byte + -rhs_byte; + + lhs := lhs + 1; + rhs := rhs + 1; + count := count + -1; + + if result = 0 then + goto .memcmp_loop; + end; + end; + + return result +end; + +(* Copies memory. *) + +(* Parameters: *) +(* destination - Destination. *) +(* source - Source. *) +(* count - Size. *) + +(* Returns the destination. *) +proc _memcpy(destination: Word, source: Word, count: Word); +var + current_byte: Word; +begin + .memcpy_loop; + if count <> 0 then + current_byte := _load_byte(source); + _store_byte(current_byte, destination); + + destination := destination + 1; + source := source + 1; + count := count + -1; + goto .memcpy_loop; + end; + + return destination +end; + +(* Advances the token stream by a0 bytes. *) +proc _advance_token(count: Word); +begin + source_code_position := source_code_position + count; +end; + +(* Prints the current token. *) + +(* Parameters: *) +(* length - Token length. *) + +(* Returns a0 unchanged. *) +proc _write_token(length: Word); +begin + _write_s(source_code_position, length); + return length +end; + +proc _compile_integer_literal(); +var + integer_token: Word; +begin + _write_z("\tli t0, \0"); + + integer_token := _read_token(); + _write_token(integer_token); + _advance_token(integer_token); + + _write_c('\n'); +end; + +proc _compile_character_literal(); +var + character: Word; +begin + _write_z("\tli t0, '\0"); + _advance_token(1); + + character := _load_byte(source_code_position); + if character = '\\' then + _write_c('\\'); + _advance_token(1); + end; + _write_s(source_code_position, 1); + _write_s("'\n", 2); + _advance_token(2); +end; + +proc _compile_variable_expression(); +begin + _compile_designator(); + _write_z("\tlw t0, (t0)\n\0"); +end; + +proc _compile_address_expression(); +begin + (* Skip the "@" sign. *) + _advance_token(1); + _compile_designator(); +end; + +proc _compile_negate_expression(); +begin + (* Skip the "-" sign. *) + _advance_token(1); + _compile_term(); + + _write_z("\tneg t0, t0\n\0"); +end; + +proc _compile_not_expression(); +begin + (* Skip the "~" sign. *) + _advance_token(1); + _compile_term(); + + _write_z("\tnot t0, t0\n\0"); +end; + +proc _compile_string_literal(); +var + length: Word; + offset: Word; +begin + length := _string_length(source_code_position); + offset := _add_string(source_code_position); + + _advance_token(length + 2); + _write_z("\tla t0, strings\n\0"); + + _write_z("\tli t1, \0"); + _write_i(offset); + _write_c('\n'); + + _write_z("\tadd t0, t0, t1\n\0"); +end; + +proc _compile_term(); +var + current_character: Word; +begin + current_character := _load_byte(source_code_position); + + if current_character = '\'' then + _compile_character_literal(); + end; + if current_character = '@' then + _compile_address_expression(); + end; + if current_character = '-' then + _compile_negate_expression(); + end; + if current_character = '~' then + _compile_not_expression(); + end; + if current_character = '"' then + _compile_string_literal(); + end; + if current_character = '_' then + _compile_call(); + _write_z("\nmv t0, a0\n\0"); + end; + if _is_digit(current_character) = 1 then + _compile_integer_literal(); + end; + if _is_lower(current_character) = 1 then + _compile_variable_expression(); + end; +end; + +proc _compile_binary_rhs(); +begin + (* Skip the whitespace after the binary operator. *) + _advance_token(1); + _compile_term(); + + (* Load the left expression from the stack; *) + _write_z("\tlw t1, 24(sp)\n\0"); +end; + +proc _compile_expression(); +var + current_character: Word; +begin + _compile_term(); + current_character := _load_byte(source_code_position); + + if current_character <> ' ' then + goto .compile_expression_end; + end; + (* It is a binary expression. *) + + (* Save the value of the left expression on the stack. *) + _write_z("sw t0, 24(sp)\n\0"); + + (* Skip surrounding whitespace in front of the operator. *) + _advance_token(1); + current_character := _load_byte(source_code_position); + + if current_character = '+' then + _advance_token(1); + _compile_binary_rhs(); + + (* Execute the operation. *) + _write_z("add t0, t0, t1\n\0"); + + goto .compile_expression_end; + end; + if current_character = '*' then + _advance_token(1); + _compile_binary_rhs(); + + (* Execute the operation. *) + _write_z("\tmul t0, t0, t1\n\0"); + + goto .compile_expression_end; + end; + if current_character = '&' then + _advance_token(1); + _compile_binary_rhs(); + + (* Execute the operation. *) + _write_z("\tand t0, t0, t1\n\0"); + + goto .compile_expression_end; + end; + if current_character = 'o' then + _advance_token(2); + _compile_binary_rhs(); + + (* Execute the operation. *) + _write_z("or t0, t0, t1\n\0"); + + goto .compile_expression_end; + end; + if current_character = 'x' then + _advance_token(3); + _compile_binary_rhs(); + + (* Execute the operation. *) + _write_z("xor t0, t0, t1\n\0"); + + goto .compile_expression_end; + end; + if current_character = '=' then + _advance_token(1); + _compile_binary_rhs(); + + (* Execute the operation. *) + _write_z("xor t0, t0, t1\nseqz t0, t0\n\0"); + + goto .compile_expression_end; + end; + if current_character = '%' then + _advance_token(1); + _compile_binary_rhs(); + + (* Execute the operation. *) + _write_z("rem t0, t1, t0\n\0"); + + goto .compile_expression_end; + end; + if current_character = '/' then + _advance_token(1); + _compile_binary_rhs(); + + (* Execute the operation. *) + _write_z("div t0, t1, t0\n\0"); + + goto .compile_expression_end; + end; + if current_character = '<' then + _advance_token(1); + current_character := _load_byte(source_code_position); + + if current_character = '>' then + _advance_token(1); + _compile_binary_rhs(); + + (* Execute the operation. *) + _write_z("\txor t0, t0, t1\nsnez t0, t0\n\0"); + + goto .compile_expression_end; + end; + if current_character = '=' then + _advance_token(1); + _compile_binary_rhs(); + + (* Execute the operation. *) + _write_z("\tslt t0, t0, t1\nxori t0, t0, 1\n\0"); + + goto .compile_expression_end; + end; + _compile_binary_rhs(); + + (* Execute the operation. *) + _write_z("slt t0, t1, t0\n\0"); + + goto .compile_expression_end; + end; + if current_character = '>' then + _advance_token(1); + current_character := _load_byte(source_code_position); + if current_character = '=' then + _advance_token(1); + _compile_binary_rhs(); + + (* Execute the operation. *) + _write_z("\tslt t0, t1, t0\nxori t0, t0, 1\n\0"); + + goto .compile_expression_end; + end; + _compile_binary_rhs(); + + (* Execute the operation. *) + _write_z("\tslt t0, t1, t0\n\0"); + + goto .compile_expression_end; + end; + + .compile_expression_end; +end; + +proc _compile_call(); +var + name_length: Word; + name: Word; + argument_count: Word; + stack_offset: Word; +begin + name_length := _read_token(); + name := source_code_position; + argument_count := 0; + + (* Skip the identifier and left paren. *) + _advance_token(name_length + 1); + + if _load_byte(source_code_position) = ')' then + goto .compile_call_finalize + end; + .compile_call_loop; + _compile_expression(); + + (* Save the argument on the stack. *) + _write_z("\tsw t0, \0"); + + (* Calculate the stack offset: 116 - (4 * argument_counter) *) + stack_offset := argument_count * 4; + _write_i(116 + -stack_offset); + + _write_z("(sp)\n\0"); + + (* Add one to the argument counter. *) + argument_count := argument_count + 1; + + if _load_byte(source_code_position) <> ',' then + goto .compile_call_finalize; + end; + _advance_token(2); + goto .compile_call_loop; + + .compile_call_finalize; + (* Load the argument from the stack. *) + if argument_count <> 0 then + (* Decrement the argument counter. *) + argument_count := argument_count + -1; + + _write_z("\tlw a\0"); + _write_i(argument_count); + + _write_z(", \0"); + + (* Calculate the stack offset: 116 - (4 * argument_counter) *) + stack_offset := argument_count * 4; + _write_i(116 + -stack_offset); + + _write_z("(sp)\n\0"); + + goto .compile_call_finalize; + end; + + .compile_call_end; + _write_z("\tcall \0"); + _write_s(name, name_length); + + (* Skip the right paren. *) + _advance_token(1); +end; + +proc _compile_goto(); +var + next_token: Word; +begin + _advance_token(5); + + next_token := _read_token(); + _write_z("\tj \0"); + + _write_token(next_token); + _advance_token(next_token); +end; + +proc _compile_local_designator(symbol: Word, name_length: Word); +var + variable_offset: Word; +begin + + _write_z("\taddi t0, sp, \0"); + variable_offset := _parameter_info_get_offset(symbol); + _write_i(variable_offset); + _write_c('\n'); + _advance_token(name_length); +end; + +proc _compile_global_designator(); +var + name: Word; +begin + _write_z("\tla t0, \0"); + + name := _read_token(); + _write_token(name); + _advance_token(name); + + _write_c('\n'); +end; + +proc _compile_designator(); +var + name_token: Word; + lookup_result: Word; +begin + name_token := _read_token(); + lookup_result := _symbol_table_lookup(@symbol_table_local, source_code_position, name_token); + + if lookup_result <> 0 then + _compile_local_designator(lookup_result, name_token); + goto .compile_designator_end; + end; + _compile_global_designator(); + + .compile_designator_end; +end; + +proc _compile_assignment(); +begin + _compile_designator(); + + (* Save the assignee address on the stack. *) + _write_z("\tsw t0, 60(sp)\n\0"); + + (* Skip the assignment sign (:=) with surrounding whitespaces. *) + _advance_token(4); + + (* Compile the assignment. *) + _compile_expression(); + + _write_z("\tlw t1, 60(sp)\nsw t0, (t1)\n\0"); +end; + +proc _compile_return_statement(); +begin + (* Skip "return" keyword and whitespace after it. *) + _advance_token(7); + _compile_expression(); + + _write_z("mv a0, t0\n\0"); +end; + +(* Writes a label, .Ln, where n is a unique number. *) + +(* Parameters: *) +(* counter - Label counter. *) +proc _write_label(counter: Word); +begin + _write_z(".L\0"); + _write_i(counter); +end; + +proc _compile_if(); +var + after_end_label: Word; + condition_label: Word; +begin + (* Skip "if ". *) + _advance_token(3); + (* Compile condition. *) + _compile_expression(); + (* Skip " then" with newline. *) + _advance_token(6); + + after_end_label := label_counter; + label_counter := label_counter + 1; + + (* condition_label is the label in front of the next elsif condition or end. *) + condition_label := label_counter; + label_counter := label_counter + 1; + + _write_z("\tbeqz t0, \0"); + _write_label(condition_label); + _write_c('\n'); + + _compile_procedure_body(); + + _write_z("\tj \0"); + _write_label(after_end_label); + _write_c('\n'); + + _write_label(condition_label); + _write_z(":\n\0"); + + if _memcmp(source_code_position, "end", 3) = 0 then + goto .compile_if_end; + end; + if _memcmp(source_code_position, "else", 3) = 0 then + goto .compile_if_else + end; + .compile_if_else; + (* Skip "else" and newline. *) + _advance_token(5); + _compile_procedure_body(); + + .compile_if_end; + (* Skip "end". *) + _advance_token(3); + + _write_label(after_end_label); + _write_z(":\n\0"); +end; + +proc _compile_label_declaration(); +var + label_token: Word; +begin + (* Skip the dot. *) + _advance_token(1); + label_token := _read_token(); + _write_c('.'); + _write_s(source_code_position, label_token); + _write_z(":\n\0"); + _advance_token(label_token); +end; + +proc _compile_statement(); +var + current_byte: Word; +begin + _skip_spaces(); + current_byte := _load_byte(source_code_position); + + (* This is a call if the statement starts with an underscore. *) + if current_byte = '_' then + _compile_call(); + goto .compile_statement_semicolon; + end; + if _memcmp(source_code_position, "goto ", 5) = 0 then + _compile_goto(); + goto .compile_statement_semicolon; + end; + if _memcmp(source_code_position, "if ", 3) = 0 then + _compile_if(); + goto .compile_statement_semicolon; + end; + if _memcmp(source_code_position, "return ", 7) = 0 then + _compile_return_statement(); + _write_c('\n'); + + goto .compile_statement_end; + end; + if current_byte = '.' then + _compile_label_declaration(); + + goto .compile_statement_semicolon; + end; + _compile_assignment(); + goto .compile_statement_semicolon; + + .compile_statement_semicolon; + _advance_token(2); + _write_c('\n'); + + .compile_statement_end; +end; + +proc _compile_procedure_body(); +var + lhs: Word; + rhs: Word; +begin + .compile_procedure_body_loop; + _skip_empty_lines(); + _skip_spaces(); + + lhs := _memcmp(source_code_position, "end", 3) = 0; + rhs := _memcmp(source_code_position, "else", 4) = 0; + lhs := lhs or rhs; + + if lhs = 0 then + _compile_statement(); + goto .compile_procedure_body_loop; + end; +end; + +(* Writes a regster name to the standard output. *) + +(* Parameters: *) +(* register_character - Register character. *) +(* register_number - Register number. *) +proc _write_register(register_character: Word, register_number: Word); +begin + _write_c(register_character); + register_number := register_number + '0'; + _write_c(register_number); +end; + +proc _skip_spaces(); +var + current_byte: Word; +begin + current_byte := _load_byte(source_code_position); + if current_byte = '\t' then + _advance_token(1); + _skip_spaces(); + end; +end; + +proc _read_type_expression(); +var + type_name: Word; +begin + type_name := _read_token(); + _advance_token(type_name); +end; + +(* Parameters: *) + +(* parameter_index - Parameter index. *) +proc _parameter_info_create(parameter_index: Word); +var + offset: Word; + current_word: Word; + result: Word; +begin + result := memory_free_pointer; + current_word := result; + (* 2 is INFO_PARAMETER *) + _store_word(2, current_word); + + current_word := current_word + 4; + + (* Calculate the stack offset: 88 - (4 * parameter_counter) *) + offset := parameter_index * 4; + _store_word(88 + -offset, current_word); + + memory_free_pointer := current_word + 4; + + return result +end; + +proc _parameter_info_get_offset(info: Word); +begin + info := info + 4; + return _load_word(info) +end; + +(* Parameters: *) + +(* temporary_index - Parameter index. *) +proc _temporary_info_create(temporary_index: Word); +var + offset: Word; + current_word: Word; + result: Word; +begin + result := memory_free_pointer; + current_word := result; + (* 3 is INFO_TEMPORARY *) + _store_word(3, current_word); + + current_word := current_word + 4; + + (* Calculate the stack offset: 4 * variable_counter. *) + _store_word(temporary_index * 4, current_word); + + memory_free_pointer := current_word + 4; + + return result +end; + +proc _temporary_info_get_offset(info: Word); +begin + info := info + 4; + return _load_word(info) +end; + +(* Parameters: *) + +(* parameter_index - Parameter index. *) +proc _read_procedure_parameter(parameter_index: Word); +var + name_length: Word; + info: Word; + name_position: Word; +begin + (* Read the parameter name. *) + name_position := source_code_position; + name_length := _read_token(); + _advance_token(name_length); + + (* Skip colon and space in front of the type expression. *) + _advance_token(2); + + _read_type_expression(); + + _write_z("\tsw a\0"); + _write_i(parameter_index); + _write_z(", \0"); + + info := _parameter_info_create(parameter_index); + _symbol_table_enter(@symbol_table_local, name_position, name_length, info); + + info := _parameter_info_get_offset(info); + _write_i(info); + + _write_z("(sp)\n\0"); +end; + +proc _read_procedure_parameters(); +var + parameter_counter: Word; +begin + (* Skip open paren. *) + _advance_token(1); + parameter_counter := 0; + + .compile_procedure_prologue_skip; + if _load_byte(source_code_position) <> ')' then + _read_procedure_parameter(parameter_counter); + parameter_counter := parameter_counter + 1; + + if _load_byte(source_code_position) = ',' then + _advance_token(2); + goto .compile_procedure_prologue_skip; + end; + end; + (* Skip close paren. *) + _advance_token(1); +end; + +(* Parameters: *) +(* variable_index - Variable index. *) +proc _read_procedure_temporary(variable_index: Word); +var + name_length: Word; + info: Word; + name_position: Word; +begin + _skip_spaces(); + name_position := source_code_position; + + (* Read and skip variable name, colon and the space *) + name_length := _read_token(); + _advance_token(name_length + 2); + + _read_type_expression(); + + info := _temporary_info_create(variable_index); + _symbol_table_enter(@symbol_table_local, name_position, name_length, info); + + (* Skip semicolon and newline after the variable declaration *) + _advance_token(2); +end; + +proc _read_procedure_temporaries(); +var + temporary_counter: Word; +begin + if _memcmp(source_code_position, "var", 3) <> 0 then + goto .read_local_variables_end; + end; + _advance_token(4); + temporary_counter := 0; + + .read_local_variables_loop; + if _memcmp(source_code_position, "begin", 5) = 0 then + goto .read_local_variables_end; + end; + _read_procedure_temporary(temporary_counter); + + temporary_counter := temporary_counter + 1; + goto .read_local_variables_loop; + + .read_local_variables_end; +end; + +proc _compile_procedure(); +var + name_length: Word; +begin + (* Skip "proc ". *) + _advance_token(5); + (* Clear local symbol table. *) + _store_word(0, @symbol_table_local); + + name_length := _read_token(); + + (* Write .type _procedure_name, @function. *) + _write_z(".type \0"); + + _write_token(name_length); + _write_z(", @function\n\0"); + + (* Write procedure label, _procedure_name: *) + _write_token(name_length); + _write_z(":\n\0"); + + (* Skip procedure name. *) + _advance_token(name_length); + _write_z("\taddi sp, sp, -128\n\tsw ra, 124(sp)\n\tsw s0, 120(sp)\n\taddi s0, sp, 128\n\0"); + _read_procedure_parameters(); + + (* Skip semicolon and newline. *) + _advance_token(2); + _read_procedure_temporaries(); + + (* Skip semicolon, "begin" and newline. *) + _advance_token(6); + + _compile_procedure_body(); + + (* Write the epilogue. *) + _write_z("\tlw ra, 124(sp)\n\tlw s0, 120(sp)\n\taddi sp, sp, 128\n\tret\n\0"); + + (* Skip the "end" keyword, semicolon and newline. *) + _advance_token(5); +end; + +(* Prints and skips a line. *) +proc _skip_comment(); +begin + .skip_comment_loop; + + (* Check for newline character. *) + if _load_byte(source_code_position) <> '\n' then + (* Advance the input string by one byte. *) + _advance_token(1); + + goto .skip_comment_loop; + end; + (* Skip the newline. *) + _advance_token(1); +end; + +(* Skip newlines and comments. *) +proc _skip_empty_lines(); +var + current_position: Word; + current_byte: Word; +begin + .skip_empty_lines_rerun; + current_position := source_code_position; + + .skip_empty_lines_loop; + current_byte := _load_byte(current_position); + + if current_byte = '\n' then + goto .skip_empty_lines_newline; + end; + if current_byte = '\t' then + goto .skip_empty_lines_tab; + end; + if current_byte <> '(' then + goto .skip_empty_lines_end; + end; + current_byte := _load_byte(current_position + 1); + + if current_byte = '*' then + goto .skip_empty_lines_comment + end; + goto .skip_empty_lines_end; + + .skip_empty_lines_comment; + source_code_position := current_position; + _skip_comment(); + goto .skip_empty_lines_rerun; + + .skip_empty_lines_newline; + source_code_position := current_position + 1; + goto .skip_empty_lines_rerun; + + .skip_empty_lines_tab; + current_position := current_position + 1; + goto .skip_empty_lines_loop + + .skip_empty_lines_end; +end; + +proc _compile_global_initializer(); +var + current_byte: Word; + length: Word; +begin + current_byte := _load_byte(source_code_position); + + if current_byte = '"' then + _write_z("\n\t.word strings + \0"); + length := _string_length(source_code_position); + + _add_string(source_code_position); + _write_i(); + + (* Skip the quoted string. *) + _advance_token(length + 2); + + goto .compile_global_initializer_end; + end; + if current_byte = 'S' then + (* Skip "S(". *) + _advance_token(2); + + if _load_byte(source_code_position) = ')' then + goto .compile_global_initializer_closing; + end; + goto .compile_global_initializer_loop; + end; + if current_byte = '@' then + (* Skip @. *) + _advance_token(1); + _write_z("\n\t.word \0"); + current_byte := _read_token(); + _write_token(current_byte); + _advance_token(current_byte); + + goto .compile_global_initializer_end; + end; + if _is_digit(current_byte) = 1 then + _write_z("\n\t.word \0"); + current_byte := _read_token(); + _write_token(current_byte); + _advance_token(1); + + goto .compile_global_initializer_end; + end; + + .compile_global_initializer_loop; + _compile_global_initializer(); + + if _load_byte(source_code_position) <> ')' then + (* Skip comma and whitespace after it. *) + _advance_token(2); + + goto .compile_global_initializer_loop; + end; + + .compile_global_initializer_closing; + (* Skip ")" *) + _advance_token(1); + + goto .compile_global_initializer_end; + + .compile_global_initializer_end; +end; + +proc _compile_constant_declaration(); +var + name_length: Word; +begin + name_length := _read_token(); + + _write_z(".type \0"); + _write_token(name_length); + _write_z(", @object\n\0"); + + _write_token(name_length); + _write_c(':'); + + (* Skip the constant name with assignment sign and surrounding whitespaces. *) + _advance_token(name_length + 4); + _compile_global_initializer(); + (* Skip semicolon and newline. *) + _advance_token(2); + _write_c('\n'); +end; + +proc _compile_const_part(); +begin + _skip_empty_lines(); + + if _memcmp(source_code_position, "const\0", 5) <> 0 then + goto .compile_const_part_end; + end; + (* Skip "const" with the newline after it. *) + _advance_token(6); + _write_z(".section .rodata # Compiled from const section.\n\n\0"); + + .compile_const_part_loop; + _skip_empty_lines(); + + (* If the character at the line beginning is not indentation, *) + (* it is probably the next code section. *) + if _load_byte(source_code_position) = '\t' then + _advance_token(1); + + _compile_constant_declaration(); + goto .compile_const_part_loop; + end; + + .compile_const_part_end; +end; + +proc _compile_variable_declaration(); +var + name_length: Word; +begin + name_length := _read_token(); + + _write_z(".type \0"); + _write_token(name_length); + _write_z(", @object\n\0"); + + _write_token(name_length); + _write_c(':'); + + (* Skip the variable name and colon with space before the type. *) + _advance_token(name_length + 2); + _read_type_expression(); + + if _load_byte(source_code_position) <> ' ' then + (* Else we assume this is a zeroed 81920 bytes big array. *) + _write_z(" .zero 81920\0"); + else + (* Skip the assignment sign with surrounding whitespaces. *) + _advance_token(4); + _compile_global_initializer(); + end; + + (* Skip semicolon and newline. *) + _advance_token(2); + _write_c('\n'); +end; + +proc _compile_var_part(); +var + current_character: Word; +begin + if _memcmp(source_code_position, "var\0", 3) <> 0 then + goto .compile_var_part_end; + end; + (* Skip "var" and newline. *) + _advance_token(4); + _write_z(".section .data\n\0"); + + .compile_var_part_loop; + _skip_empty_lines(); + current_character := _load_byte(source_code_position); + + if current_character = '\t' then + _advance_token(1); + _compile_variable_declaration(); + goto .compile_var_part_loop; + end; + + .compile_var_part_end; +end; + +(* Process the source code and print the generated code. *) +proc _compile_module(); +begin + _compile_const_part(); + _skip_empty_lines(); + _compile_var_part(); + + _write_z(".section .text\n\n\0"); + _write_z(".type _syscall, @function\n_syscall:\n\tmv a7, a6\n\tecall\n\tret\n\n\0"); + _write_z(".type _load_byte, @function\n_load_byte:\n\tlb a0, (a0)\nret\n\n\0"); + _write_z(".type _load_word, @function\n_load_word:\n\tlw a0, (a0)\nret\n\n\0"); + _write_z(".type _store_byte, @function\n_store_byte:\n\tsb a0, (a1)\nret\n\n\0"); + _write_z(".type _store_word, @function\n_store_word:\n\tsw a0, (a1)\nret\n\n\0"); + + .compile_module_loop; + _skip_empty_lines(); + + if _load_byte(source_code_position) <> 0 then + (* 5 is "proc " length. Space is needed to distinguish from "procedure". *) + if _memcmp(source_code_position, "proc ", 5) = 0 then + _compile_procedure(); + goto .compile_module_loop; + end; + end; + .compile_module_end; +end; + +proc _compile(); +var + compiler_strings_copy: Word; + compiler_strings_end: Word; + current_byte: Word; +begin + _write_z(".globl _start\n\n\0"); + _compile_module(); + + _write_z(".section .rodata\n.type strings, @object\nstrings: .ascii \0"); + _write_c('"'); + + compiler_strings_copy := @compiler_strings; + compiler_strings_end := compiler_strings_position; + + .compile_loop; + if compiler_strings_copy < compiler_strings_end then + current_byte := _load_byte(compiler_strings_copy); + compiler_strings_copy := compiler_strings_copy + 1; + _write_c(current_byte); + + goto .compile_loop; + end; + _write_c('"'); + _write_c('\n'); +end; + +(* Terminates the program. a0 contains the return code. *) + +(* Parameters: *) +(* a0 - Status code. *) +proc _exit(); +begin + _syscall(0, 0, 0, 0, 0, 0, 93); +end; + +(* Looks for a symbol in the given symbol table. *) + +(* Parameters: *) +(* symbol_table - Symbol table. *) +(* symbol_name - Symbol name pointer. *) +(* name_length - Symbol name length. *) + +(* Returns the symbol pointer or 0 in a0. *) +proc _symbol_table_lookup(symbol_table: Word, symbol_name: Word, name_length: Word); +var + result: Word; + symbol_table_length: Word; + current_name: Word; + current_length: Word; +begin + result := 0; + + (* The first word in the symbol table is its length, get it. *) + symbol_table_length := _load_word(symbol_table); + + (* Go to the first symbol position. *) + symbol_table := symbol_table + 4; + + .symbol_table_lookup_loop; + if symbol_table_length = 0 then + goto .symbol_table_lookup_end; + end; + + (* Symbol name pointer and length. *) + current_name := _load_word(symbol_table); + current_length := _load_word(symbol_table + 4); + + (* If lengths don't match, exit and return nil. *) + if name_length <> current_length then + goto .symbol_table_lookup_repeat; + end; + (* If names don't match, exit and return nil. *) + if _memcmp(symbol_name, current_name, name_length) <> 0 then + goto .symbol_table_lookup_repeat; + end; + (* Otherwise, the symbol is found. *) + result := _load_word(symbol_table + 8); + goto .symbol_table_lookup_end; + + .symbol_table_lookup_repeat; + symbol_table := symbol_table + 12; + symbol_table_length := symbol_table_length + -1; + goto .symbol_table_lookup_loop; + + .symbol_table_lookup_end; + return result +end; + +(* Inserts a symbol into the table. *) + +(* Parameters: *) +(* symbol_table - Symbol table. *) +(* symbol_name - Symbol name pointer. *) +(* name_length - Symbol name length. *) +(* symbol - Symbol pointer. *) +proc _symbol_table_enter(symbol_table: Word, symbol_name: Word, name_length: Word, symbol: Word); +var + table_length: Word; + symbol_pointer: Word; +begin + (* The first word in the symbol table is its length, get it. *) + table_length := _load_word(symbol_table); + + (* Calculate the offset for the new symbol. *) + symbol_pointer := table_length * 12; + symbol_pointer := symbol_pointer + 4; + symbol_pointer := symbol_table + symbol_pointer; + + _store_word(symbol_name, symbol_pointer); + symbol_pointer := symbol_pointer + 4; + _store_word(name_length, symbol_pointer); + symbol_pointer := symbol_pointer + 4; + _store_word(symbol, symbol_pointer); + + (* Increment the symbol table length. *) + table_length := table_length + 1; + _store_word(table_length, symbol_table); +end; + +proc _symbol_table_build(); +begin + (* Set the table length to 0. *) + _store_word(0, @symbol_table_global); + + (* Enter built-in symbols. *) + _symbol_table_enter(@symbol_table_global, symbol_builtin_name_int, 3, @symbol_type_info_int); + _symbol_table_enter(@symbol_table_global, symbol_builtin_name_word, 4, @symbol_type_info_word); + _symbol_table_enter(@symbol_table_global, symbol_builtin_name_pointer, 7, @symbol_type_info_pointer); + _symbol_table_enter(@symbol_table_global, symbol_builtin_name_char, 4, @symbol_type_info_char); + _symbol_table_enter(@symbol_table_global, symbol_builtin_name_bool, 4, @symbol_type_info_bool); +end; + + +(* Classification table assigns each possible character to a group (class). All *) +(* characters of the same group a handled equivalently. *) + +(* Classification: *) + +(* TransitionClass = ( *) +(* transitionClassInvalid = 1, *) +(* transitionClassDigit = 2, *) +(* transitionClassAlpha = 3, *) +(* transitionClassSpace = 4, *) +(* transitionClassColon = 5, *) +(* transitionClassEquals = 6, *) +(* transitionClassLeftParen = 7, *) +(* transitionClassRightParen = 8, *) +(* transitionClassAsterisk = 9, *) +(* transitionClassUnderscore = 10, *) +(* transitionClassSingle = 11, *) +(* transitionClassHex = 12, *) +(* transitionClassZero = 13, *) +(* transitionClassX = 14, *) +(* transitionClassEof = 15, *) +(* transitionClassDot = 16, *) +(* transitionClassMinus = 17, *) +(* transitionClassSingleQuote = 18, *) +(* transitionClassDoubleQuote = 19, *) +(* transitionClassGreater = 20, *) +(* transitionClassLess = 21, *) +(* transitionClassOther = 22 *) +(* ); *) +(* TransitionState = ( *) +(* transitionStateStart = 1, *) +(* transitionStateColon = 2, *) +(* transitionStateIdentifier = 3, *) +(* transitionStateDecimal = 4, *) +(* transitionStateGreater = 5, *) +(* transitionStateMinus = 6, *) +(* transitionStateLeftParen = 7, *) +(* transitionStateLess = 8, *) +(* transitionStateDot = 9, *) +(* transitionStateComment = 10, *) +(* transitionStateClosingComment = 11, *) +(* transitionStateCharacter = 12, *) +(* transitionStateString = 13, *) +(* transitionStateLeadingZero = 14, *) +(* transitionStateDecimalSuffix = 15, *) +(* transitionStateEnd = 16 *) +(* ); *) +(* Transition = record *) +(* action: TransitionAction; *) +(* next_state: TransitionState *) +(* end; *) +(* TransitionAction = ( *) +(* none = 1, *) +(* accumulate = 2, *) +(* skip = 3, *) +(* single = 4, *) +(* eof = 5, *) +(* finalize = 6, *) +(* composite = 7, *) +(* key_id = 8, *) +(* integer = 9, *) +(* delimited = 10 *) +(* ); *) + +(* Assigns some value to at array index. *) + +(* Parameters: *) +(* array - Array pointer. *) +(* index - Index (word offset into the array). *) +(* data - Data to assign. *) +proc _assign_at(array: Word, index: Word, data: Word); +var + target: Word; +begin + target := index + -1; + target := target * 4; + target := array + target; + + _store_word(data, target); +end; + +proc _create_classification(); +var + code: Word; +begin + _assign_at(@classification, 1, 15); + _assign_at(@classification, 2, 1); + _assign_at(@classification, 3, 1); + _assign_at(@classification, 4, 1); + _assign_at(@classification, 5, 1); + _assign_at(@classification, 6, 1); + _assign_at(@classification, 7, 1); + _assign_at(@classification, 8, 1); + _assign_at(@classification, 9, 1); + _assign_at(@classification, 10, 4); + _assign_at(@classification, 11, 4); + _assign_at(@classification, 12, 1); + _assign_at(@classification, 13, 1); + _assign_at(@classification, 14, 4); + _assign_at(@classification, 15, 1); + _assign_at(@classification, 16, 1); + _assign_at(@classification, 17, 1); + _assign_at(@classification, 18, 1); + _assign_at(@classification, 19, 1); + _assign_at(@classification, 20, 1); + _assign_at(@classification, 21, 1); + _assign_at(@classification, 22, 1); + _assign_at(@classification, 23, 1); + _assign_at(@classification, 24, 1); + _assign_at(@classification, 25, 1); + _assign_at(@classification, 26, 1); + _assign_at(@classification, 27, 1); + _assign_at(@classification, 28, 1); + _assign_at(@classification, 29, 1); + _assign_at(@classification, 30, 1); + _assign_at(@classification, 31, 1); + _assign_at(@classification, 32, 1); + _assign_at(@classification, 33, 4); + _assign_at(@classification, 34, 11); + _assign_at(@classification, 35, 19); + _assign_at(@classification, 36, 22); + _assign_at(@classification, 37, 22); + _assign_at(@classification, 38, 11); + _assign_at(@classification, 39, 11); + _assign_at(@classification, 40, 18); + _assign_at(@classification, 41, 7); + _assign_at(@classification, 42, 8); + _assign_at(@classification, 43, 9); + _assign_at(@classification, 44, 11); + _assign_at(@classification, 45, 11); + _assign_at(@classification, 46, 17); + _assign_at(@classification, 47, 16); + _assign_at(@classification, 48, 11); + _assign_at(@classification, 49, 13); + _assign_at(@classification, 50, 2); + _assign_at(@classification, 51, 2); + _assign_at(@classification, 52, 2); + _assign_at(@classification, 53, 2); + _assign_at(@classification, 54, 2); + _assign_at(@classification, 55, 2); + _assign_at(@classification, 56, 2); + _assign_at(@classification, 57, 2); + _assign_at(@classification, 58, 2); + _assign_at(@classification, 59, 5); + _assign_at(@classification, 60, 11); + _assign_at(@classification, 61, 21); + _assign_at(@classification, 62, 6); + _assign_at(@classification, 63, 20); + _assign_at(@classification, 64, 22); + _assign_at(@classification, 65, 11); + _assign_at(@classification, 66, 3); + _assign_at(@classification, 67, 3); + _assign_at(@classification, 68, 3); + _assign_at(@classification, 69, 3); + _assign_at(@classification, 70, 3); + _assign_at(@classification, 71, 3); + _assign_at(@classification, 72, 3); + _assign_at(@classification, 73, 3); + _assign_at(@classification, 74, 3); + _assign_at(@classification, 75, 3); + _assign_at(@classification, 76, 3); + _assign_at(@classification, 77, 3); + _assign_at(@classification, 78, 3); + _assign_at(@classification, 79, 3); + _assign_at(@classification, 80, 3); + _assign_at(@classification, 81, 3); + _assign_at(@classification, 82, 3); + _assign_at(@classification, 83, 3); + _assign_at(@classification, 84, 3); + _assign_at(@classification, 85, 3); + _assign_at(@classification, 86, 3); + _assign_at(@classification, 87, 3); + _assign_at(@classification, 88, 3); + _assign_at(@classification, 89, 3); + _assign_at(@classification, 90, 3); + _assign_at(@classification, 91, 3); + _assign_at(@classification, 92, 11); + _assign_at(@classification, 93, 22); + _assign_at(@classification, 94, 11); + _assign_at(@classification, 95, 11); + _assign_at(@classification, 96, 10); + _assign_at(@classification, 97, 22); + _assign_at(@classification, 98, 12); + _assign_at(@classification, 99, 12); + _assign_at(@classification, 100, 12); + _assign_at(@classification, 101, 12); + _assign_at(@classification, 102, 12); + _assign_at(@classification, 103, 12); + _assign_at(@classification, 104, 3); + _assign_at(@classification, 105, 3); + _assign_at(@classification, 106, 3); + _assign_at(@classification, 107, 3); + _assign_at(@classification, 108, 3); + _assign_at(@classification, 109, 3); + _assign_at(@classification, 110, 3); + _assign_at(@classification, 111, 3); + _assign_at(@classification, 112, 3); + _assign_at(@classification, 113, 3); + _assign_at(@classification, 114, 3); + _assign_at(@classification, 115, 3); + _assign_at(@classification, 116, 3); + _assign_at(@classification, 117, 3); + _assign_at(@classification, 118, 3); + _assign_at(@classification, 119, 3); + _assign_at(@classification, 120, 3); + _assign_at(@classification, 121, 14); + _assign_at(@classification, 122, 3); + _assign_at(@classification, 123, 3); + _assign_at(@classification, 124, 22); + _assign_at(@classification, 125, 11); + _assign_at(@classification, 126, 22); + _assign_at(@classification, 127, 11); + _assign_at(@classification, 128, 1); + + code := 129; + + (* Set the remaining 129 - 256 bytes to transitionClassOther. *) + .create_classification_loop; + _assign_at(@classification, code, 22); + code := code + 1; + + if code < 257 then + goto .create_classification_loop; + end; +end; + +(* Parameters: *) +(* current_state - Current state (first index into transitions table). *) +(* transition - Transition (second index into transitions table). *) +(* action - Action to assign. *) +(* next_state - Next state to assign. *) +proc _set_transition(current_state: Word, transition: Word, action: Word, next_state: Word); +var + transition_table: Word; + row_position: Word; + state_position: Word; + target: Word; +begin + (* Transitions start at offset in classification array. *) + transition_table := @classification + 256 + + (* Each state is 8 bytes long (2 words: action and next state). *) + (* There are 16 transition classes, so a transition 8 * 16 = 128 bytes long. *) + + row_position := current_state + -1; + row_position := row_position * 128; + + state_position := transition + -1; + state_position := state_position * 8; + + target := transition_table + row_position; + target := target + state_position; + + _store_word(action, target); + target := target + 4; + _store_word(next_state, target); +end; + +(* Parameters: *) +(* current_state - Current state (Transition state enumeration). *) +(* default_action - Default action (Callback). *) +(* next_state - Next state (Transition state enumeration). *) +proc _set_default_transition(current_state: Word, default_action: Word, next_state: Word); +begin + _set_transition(current_state, 1, default_action, next_state); + _set_transition(current_state, 2, default_action, next_state); + _set_transition(current_state, 3, default_action, next_state); + _set_transition(current_state, 4, default_action, next_state); + _set_transition(current_state, 5, default_action, next_state); + _set_transition(current_state, 6, default_action, next_state); + _set_transition(current_state, 7, default_action, next_state); + _set_transition(current_state, 8, default_action, next_state); + _set_transition(current_state, 9, default_action, next_state); + _set_transition(current_state, 10, default_action, next_state); + _set_transition(current_state, 11, default_action, next_state); + _set_transition(current_state, 12, default_action, next_state); + _set_transition(current_state, 13, default_action, next_state); + _set_transition(current_state, 14, default_action, next_state); + _set_transition(current_state, 15, default_action, next_state); + _set_transition(current_state, 16, default_action, next_state); + _set_transition(current_state, 17, default_action, next_state); + _set_transition(current_state, 18, default_action, next_state); + _set_transition(current_state, 19, default_action, next_state); + _set_transition(current_state, 20, default_action, next_state); + _set_transition(current_state, 21, default_action, next_state); + _set_transition(current_state, 22, default_action, next_state); +end; + + +(* The transition table describes transitions from one state to another, given *) +(* a symbol (character class). *) + +(* The table has m rows and n columns, where m is the amount of states and n is *) +(* the amount of classes. So given the current state and a classified character *) +(* the table can be used to look up the next state. *) + +(* Each cell is a word long. *) +(* - The least significant byte of the word is a row number (beginning with 0). *) +(* It specifies the target state. "ff" means that this is an end state and no *) +(* transition is possible. *) +(* - The next byte is the action that should be performed when transitioning. *) +(* For the meaning of actions see labels in the lex_next function, which *) +(* handles each action. *) +proc _create_transitions(); +begin + (* Start state. *) + _set_transition(1, 1, 1, 16); + _set_transition(1, 2, 2, 4); + _set_transition(1, 3, 2, 3); + _set_transition(1, 4, 3, 1); + _set_transition(1, 5, 2, 5); + _set_transition(1, 6, 4, 16); + _set_transition(1, 7, 2, 7); + _set_transition(1, 8, 4, 16); + _set_transition(1, 9, 4, 16); + _set_transition(1, 10, 2, 3); + _set_transition(1, 11, 4, 16); + _set_transition(1, 12, 2, 3); + _set_transition(1, 13, 2, 14); + _set_transition(1, 14, 2, 3); + _set_transition(1, 15, 5, 16); + _set_transition(1, 16, 2, 9); + _set_transition(1, 17, 2, 6); + _set_transition(1, 18, 2, 12); + _set_transition(1, 19, 2, 13); + _set_transition(1, 20, 2, 5); + _set_transition(1, 21, 2, 8); + _set_transition(1, 22, 1, 16); + + (* Colon state. *) + _set_default_transition(2, 6, 16); + _set_transition(2, 6, 7, 16); + + (* Identifier state. *) + _set_default_transition(3, 8, 16); + _set_transition(3, 2, 2, 3); + _set_transition(3, 3, 2, 3); + _set_transition(3, 10, 2, 3); + _set_transition(3, 12, 2, 3); + _set_transition(3, 13, 2, 3); + _set_transition(3, 14, 2, 3); + + (* Decimal state. *) + _set_default_transition(4, 9, 16); + _set_transition(4, 2, 2, 4); + _set_transition(4, 3, 2, 15); + _set_transition(4, 10, 1, 16); + _set_transition(4, 12, 2, 15); + _set_transition(4, 13, 2, 4); + _set_transition(4, 14, 2, 15); + + (* Greater state. *) + _set_default_transition(5, 6, 16); + _set_transition(5, 6, 7, 16); + + (* Minus state. *) + _set_default_transition(6, 6, 16); + _set_transition(6, 20, 7, 16); + + (* Left paren state. *) + _set_default_transition(7, 6, 16); + _set_transition(7, 9, 2, 10); + + (* Less state. *) + _set_default_transition(8, 6, 16); + _set_transition(8, 6, 7, 16); + _set_transition(8, 20, 7, 16); + + (* Hexadecimal after 0x. *) + _set_default_transition(9, 6, 16); + _set_transition(9, 16, 7, 16); + + (* Comment. *) + _set_default_transition(10, 2, 10); + _set_transition(10, 9, 2, 11); + _set_transition(10, 15, 1, 16); + + (* Closing comment. *) + _set_default_transition(11, 2, 10); + _set_transition(11, 1, 1, 16); + _set_transition(11, 8, 10, 16); + _set_transition(11, 9, 2, 11); + _set_transition(11, 15, 1, 16); + + (* Character. *) + _set_default_transition(12, 2, 12); + _set_transition(12, 1, 1, 16); + _set_transition(12, 15, 1, 16); + _set_transition(12, 18, 10, 16); + + (* String. *) + _set_default_transition(13, 2, 13); + _set_transition(13, 1, 1, 16); + _set_transition(13, 15, 1, 16); + _set_transition(13, 19, 10, 16); + + (* Leading zero. *) + _set_default_transition(14, 9, 16); + _set_transition(14, 2, 1, 16); + _set_transition(14, 3, 1, 16); + _set_transition(14, 10, 1, 16); + _set_transition(14, 12, 1, 16); + _set_transition(14, 13, 1, 16); + _set_transition(14, 14, 1, 16); + + (* Digit with a character suffix. *) + _set_default_transition(15, 9, 16); + _set_transition(15, 3, 1, 16); + _set_transition(15, 2, 1, 16); + _set_transition(15, 12, 1, 16); + _set_transition(15, 13, 1, 16); + _set_transition(15, 14, 1, 16); +end; + +proc _lexer_get_state(); +var + offset: Word; + size: Word; +begin + (* Lexer state is saved after the transition tables. The offset is 256 + 16 * 22. *) + offset := @classification; + size := 16 * 22; + offset := offset + 256; + + return offset + size +end; + +(* Gets pointer to the current source text. *) +proc _lexer_get_current(); +begin + return _lexer_get_state() + 4 +end; + +(* Resets the lexer state for reading the next token. *) +proc _lexer_reset(); +var + state: Word; + current: Word; +begin + (* Transition start state is 1. *) + state := _lexer_get_state(); + _store_word(1, state); + + (* Text pointer to the beginning of the currently read token. *) + current := _lexer_get_current(); + _store_word(source_code_position, current); + + (* Initial length of the token is 0. *) + _store_word(0, source_code_position + 4); +end; + +(* One time lexer initialization. *) +proc _lexer_initialize(); +begin + _create_classification(); + _create_transitions(); +end; + +(* Entry point. *) +proc _start(); +begin + _lexer_initialize(); + _symbol_table_build(); + + (* Read the source from the standard input. *) + (* Second argument is buffer size. Modifying update the source_code definition. *) + _read_file(@source_code, 81920); + _compile(); + + _exit(0); +end; |