summaryrefslogtreecommitdiff
path: root/boot/stage12/cl.elna
diff options
context:
space:
mode:
Diffstat (limited to 'boot/stage12/cl.elna')
-rw-r--r--boot/stage12/cl.elna1944
1 files changed, 1944 insertions, 0 deletions
diff --git a/boot/stage12/cl.elna b/boot/stage12/cl.elna
new file mode 100644
index 0000000..1cf4969
--- /dev/null
+++ b/boot/stage12/cl.elna
@@ -0,0 +1,1944 @@
+(* 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 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";
+ 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: *)
+(* a0 - String token pointer. *)
+
+(* Returns the length in a0. *)
+proc _string_length(v88: Word);
+begin
+ (* Reset the counter. *)
+ v0 := 0;
+
+ .string_length_loop;
+ v88 := v88 + 1;
+
+ if _load_byte(v88) <> '"' then
+ v0 := v0 + 1;
+ goto .string_length_loop;
+ end;
+
+ return v0
+end;
+
+(* Adds a string to the global, read-only string storage. *)
+
+(* Parameters: *)
+(* a0 - String token. *)
+
+(* Returns the offset from the beginning of the storage to the new string in a0. *)
+proc _add_string(v88: Word);
+begin
+ v0 := v88 + 1;
+ v4 := compiler_strings_length;
+
+ .add_string_loop;
+ if _load_byte(v0) <> '"' then
+ v8 := _load_byte(v0);
+ _store_byte(v8, compiler_strings_position);
+ compiler_strings_position := compiler_strings_position + 1;
+ v0 := v0 + 1;
+
+ if v8 <> '\\' then
+ compiler_strings_length := compiler_strings_length + 1;
+ end;
+ goto .add_string_loop;
+ end;
+
+ return v4
+end;
+
+(* Reads standard input into a buffer. *)
+(* a0 - Buffer pointer. *)
+(* a1 - Buffer size. *)
+
+(* Returns the amount of bytes written in a0. *)
+proc _read_file(v88: Word, v84: Word);
+begin
+ return _syscall(0, v88, v84, 0, 0, 0, 63)
+end;
+
+(* Writes to the standard output. *)
+
+(* Parameters: *)
+(* a0 - Buffer. *)
+(* a1 - Buffer length. *)
+proc _write_s(v88: Word, v84: Word);
+begin
+ _syscall(1, v88, v84, 0, 0, 0, 64);
+end;
+
+(* Writes a number to a string buffer. *)
+
+(* t0 - Local buffer. *)
+(* t1 - Constant 10. *)
+(* t2 - Current character. *)
+(* t3 - Whether the number is negative. *)
+
+(* Parameters: *)
+(* a0 - Whole number. *)
+(* a1 - Buffer pointer. *)
+
+(* Sets a0 to the length of the written number. *)
+proc _print_i(v88: Word, v84: Word);
+begin
+ v0 := @v23;
+
+ if v88 >= 0 then
+ v4 := 0;
+ else
+ v88 = -v88;
+ v4 := 1;
+ end;
+
+ .print_i_digit10;
+ v8 := v88 % 10;
+ _store_byte(v8 + '0', v0);
+
+ v88 := v88 / 10;
+ v0 := v0 + -1;
+
+ if v88 <> 0 then
+ goto .print_i_digit10;
+ end;
+ if v4 = 1 then
+ _store_byte('-', v0);
+ v0 := v0 + -1;
+ end;
+ v4 := @v23 + -v0;
+ _memcpy(v84, v0 + 1, v4);
+
+ return v4
+end;
+
+(* Writes a number to the standard output. *)
+
+(* Parameters: *)
+(* a0 - Whole number. *)
+proc _write_i(v88: Word);
+begin
+ v4 := _print_i(v88, @v0);
+ _write_s(@v0, v4);
+end;
+
+(* Writes a character from a0 into the standard output. *)
+proc _write_c(v88: Word);
+begin
+ _write_s(@v88, 1);
+end;
+
+(* Write null terminated string. *)
+
+(* Parameters: *)
+(* a0 - String. *)
+proc _write_z(v88: Word);
+begin
+ (* Check for 0 character. *)
+ v0 := _load_byte(v88);
+
+ if v0 <> 0 then
+ (* Print a character. *)
+ _write_c(v0);
+
+ (* Advance the input string by one byte. *)
+ _write_z(v88 + 1);
+ end;
+end;
+
+(* Detects if a0 is an uppercase character. Sets a0 to 1 if so, otherwise to 0. *)
+proc _is_upper(v88: Word);
+begin
+ v0 := v88 >= 'A';
+ v4 := v88 <= 'Z';
+
+ return v0 & v4
+
+end;
+
+(* Detects if a0 is an lowercase character. Sets a0 to 1 if so, otherwise to 0. *)
+proc _is_lower(v88: Word);
+begin
+ v0 := v88 >= 'a';
+ v4 := v88 <= 'z';
+
+ return v0 & v4
+
+end;
+
+(* Detects if the passed character is a 7-bit alpha character or an underscore. *)
+
+(* Paramters: *)
+(* a0 - Tested character. *)
+
+(* Sets a0 to 1 if the character is an alpha character or underscore, sets it to 0 otherwise. *)
+proc _is_alpha(v88: Word);
+begin
+ v0 := _is_upper(v88);
+ v4 := _is_lower(v88);
+ v8 := v88 = '_';
+
+ v12 := v0 or v4;
+ return v12 or v8
+end;
+
+(* Detects whether the passed character is a digit *)
+(* (a value between 0 and 9). *)
+
+(* Parameters: *)
+(* a0 - Exemined value. *)
+
+(* Sets a0 to 1 if it is a digit, to 0 otherwise. *)
+proc _is_digit(v88: Word);
+begin
+ v0 := v88 >= '0';
+ v4 := v88 <= '9';
+
+ return v0 & v4
+end;
+
+proc _is_alnum(v88: Word);
+begin
+ v0 := _is_alpha(v88);
+ v4 := _is_digit(v88);
+
+ return v0 or v4
+end;
+
+(* Reads the next token. *)
+
+(* Returns token length in a0. *)
+proc _read_token();
+begin
+ (* Current token position. *)
+ v0 := source_code_position;
+ (* Token length. *)
+ v4 := 0;
+
+ .read_token_loop;
+ (* Current character. *)
+ v8 := _load_byte(v0);
+
+ (* First we try to read a derictive. *)
+ (* A derictive can contain a dot and characters. *)
+ v12 := v8 = '.';
+ v16 := _is_alnum(v8);
+
+ if v12 or v16 then
+ (* Advance the source code position and token length. *)
+ v4 := v4 + 1;
+ v0 := v0 + 1;
+
+ goto .read_token_loop;
+ end;
+
+ return v4
+end;
+
+(* a0 - First pointer. *)
+(* a1 - Second pointer. *)
+(* a2 - The length to compare. *)
+
+(* Returns 0 in a0 if memory regions are equal. *)
+proc _memcmp(v88: Word, v84: Word, v80: Word);
+begin
+ v8 := 0;
+
+ .memcmp_loop;
+ if v80 <> 0 then
+ v0 := _load_byte(v88);
+ v4 := _load_byte(v84);
+ v8 := v0 + -v4;
+
+ v88 := v88 + 1;
+ v84 := v84 + 1;
+ v80 := v80 + -1;
+
+ if v8 = 0 then
+ goto .memcmp_loop;
+ end;
+ end;
+
+ return v8
+end;
+
+(* Copies memory. *)
+
+(* Parameters: *)
+(* a0 - Destination. *)
+(* a1 - Source. *)
+(* a2 - Size. *)
+
+(* Preserves a0. *)
+proc _memcpy(v88: Word, v84: Word, v80: Word);
+begin
+ .memcpy_loop;
+ if v80 <> 0 then
+ v0 := _load_byte(v84);
+ _store_byte(v0, v88);
+
+ v88 := v88 + 1;
+ v84 := v84 + 1;
+ v80 := v80 + -1;
+ goto .memcpy_loop;
+ end;
+
+ return v88
+end;
+
+(* Advances the token stream by a0 bytes. *)
+proc _advance_token(v88: Word);
+begin
+ source_code_position := source_code_position + v88;
+end;
+
+(* Prints the current token. *)
+
+(* Parameters: *)
+(* a0 - Token length. *)
+
+(* Returns a0 unchanged. *)
+proc _write_token(v88: Word);
+begin
+ _write_s(source_code_position, v88);
+ return v88
+end;
+
+proc _compile_integer_literal();
+begin
+ _write_z("\tli t0, \0");
+
+ v0 := _read_token();
+ _write_token(v0);
+ _advance_token(v0);
+
+ _write_c('\n');
+end;
+
+proc _compile_character_literal();
+begin
+ _write_z("\tli t0, \0");
+
+ _write_c('\'');
+ _advance_token(1);
+
+ v0 := _load_byte(source_code_position);
+ if v0 = '\\' then
+ _write_c('\\');
+ _advance_token(1);
+ end;
+
+ v0 := _load_byte(source_code_position);
+ _write_c(v0);
+
+ _write_c('\'');
+ _write_c('\n');
+
+ _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();
+begin
+ v0 := _string_length(source_code_position);
+ v4 := _add_string(source_code_position);
+
+ _advance_token(v0 + 2);
+ _write_z("\tla t0, strings\n\0");
+
+ _write_z("\tli t1, \0");
+ _write_i(v4);
+ _write_c('\n');
+
+ _write_z("\tadd t0, t0, t1\n\0");
+end;
+
+proc _compile_term();
+begin
+ v0 := _load_byte(source_code_position);
+
+ if v0 = '\'' then
+ _compile_character_literal();
+ end;
+ if v0 = '@' then
+ _compile_address_expression();
+ end;
+ if v0 = '-' then
+ _compile_negate_expression();
+ end;
+ if v0 = '~' then
+ _compile_not_expression();
+ end;
+ if v0 = '"' then
+ _compile_string_literal();
+ end;
+ if v0 = '_' then
+ _compile_call();
+ _write_z("\nmv t0, a0\n\0");
+ end;
+ if _is_digit(v0) = 1 then
+ _compile_integer_literal();
+ end;
+ if _is_lower(v0) = 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();
+begin
+ _compile_term();
+ v0 := _load_byte(source_code_position);
+
+ if v0 <> ' ' 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);
+ v0 := _load_byte(source_code_position);
+
+ if v0 = '+' then
+ _advance_token(1);
+ _compile_binary_rhs();
+
+ (* Execute the operation. *)
+ _write_z("add t0, t0, t1\n\0");
+
+ goto .compile_expression_end;
+ end;
+ if v0 = '*' then
+ _advance_token(1);
+ _compile_binary_rhs();
+
+ (* Execute the operation. *)
+ _write_z("\tmul t0, t0, t1\n\0");
+
+ goto .compile_expression_end;
+ end;
+ if v0 = '&' then
+ _advance_token(1);
+ _compile_binary_rhs();
+
+ (* Execute the operation. *)
+ _write_z("\tand t0, t0, t1\n\0");
+
+ goto .compile_expression_end;
+ end;
+ if v0 = '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 v0 = '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 v0 = '=' 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 v0 = '%' then
+ _advance_token(1);
+ _compile_binary_rhs();
+
+ (* Execute the operation. *)
+ _write_z("rem t0, t1, t0\n\0");
+
+ goto .compile_expression_end;
+ end;
+ if v0 = '/' then
+ _advance_token(1);
+ _compile_binary_rhs();
+
+ (* Execute the operation. *)
+ _write_z("div t0, t1, t0\n\0");
+
+ goto .compile_expression_end;
+ end;
+ if v0 = '<' then
+ _advance_token(1);
+ v0 := _load_byte(source_code_position);
+
+ if v0 = '>' 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 v0 = '=' 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 v0 = '>' then
+ _advance_token(1);
+ v0 := _load_byte(source_code_position);
+ if v0 = '=' 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, t0, t1\n\0");
+
+ goto .compile_expression_end;
+ end;
+
+ .compile_expression_end;
+end;
+
+proc _compile_call();
+begin
+ (* Stack variables: *)
+ (* v0 - Procedure name length. *)
+ (* v4 - Procedure name pointer. *)
+ (* v8 - Argument count. *)
+
+ v0 := _read_token();
+ v4 := source_code_position;
+ v8 := 0;
+
+ (* Skip the identifier and left paren. *)
+ _advance_token(v0 + 1);
+ v12 := _load_byte(source_code_position);
+
+ if v12 = ')' 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) *)
+ v12 := v8 * 4;
+ v12 := 116 + -v12;
+ _write_i(v12);
+
+ _write_z("(sp)\n\0");
+
+ (* Add one to the argument counter. *)
+ v8 := v8 + 1;
+
+ v12 := _load_byte(source_code_position);
+
+ if v12 <> ',' then
+ goto .compile_call_finalize;
+ end;
+ _advance_token(2);
+ goto .compile_call_loop;
+
+ .compile_call_finalize;
+ (* Load the argument from the stack. *)
+ if v8 <> 0 then
+ (* Decrement the argument counter. *)
+ v8 := v8 + -1;
+
+ _write_z("\tlw a\0");
+ _write_i(v8);
+
+ _write_z(", \0");
+
+ (* Calculate the stack offset: 116 - (4 * argument_counter) *)
+ v12 := v8 * 4;
+ v12 := 116 + -v12;
+ _write_i(v12);
+
+ _write_z("(sp)\n\0");
+
+ goto .compile_call_finalize;
+ end;
+
+ .compile_call_end;
+ _write_z("\tcall \0");
+ _write_s(v4, v0);
+
+ (* Skip the right paren. *)
+ _advance_token(1);
+end;
+
+proc _compile_goto();
+begin
+ _advance_token(5);
+
+ v0 := _read_token();
+ _write_z("\tj \0");
+
+ _write_token(v0);
+ _advance_token();
+end;
+
+proc _compile_local_designator();
+begin
+ (* Skip "v" in the local variable name. *)
+ _advance_token(1);
+ _write_z("\t addi t0, sp, \0");
+
+ (* Read local variable stack offset and save it. *)
+ v0 := _read_token();
+ _write_token(v0);
+ _advance_token(v0);
+ _write_c('\n');
+end;
+
+proc _compile_global_designator();
+begin
+ _write_z("\tla t0, \0");
+
+ v0 := _read_token();
+ _write_token(v0);
+ _advance_token(v0);
+
+ _write_c('\n');
+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();
+ 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: *)
+(* a0 - Label counter. *)
+proc _write_label(v88: Word);
+begin
+ _write_z(".L\0");
+ _write_i(v88);
+end;
+
+proc _compile_if();
+begin
+ (* Skip "if ". *)
+ _advance_token(3);
+ (* Compile condition. *)
+ _compile_expression();
+ (* Skip " then" with newline. *)
+ _advance_token(6);
+
+ (* v0 is the label after the if statement. *)
+ v0 := label_counter;
+ label_counter := label_counter + 1;
+
+ (* v4 is the label in front of the next elsif condition or end. *)
+ v4 := label_counter;
+ label_counter := label_counter + 1;
+
+ _write_z("\tbeqz t0, \0");
+ _write_label(v4);
+ _write_c('\n');
+
+ _compile_procedure_body();
+
+ _write_z("\tj \0");
+ _write_label(v0);
+ _write_c('\n');
+
+ _write_label(v4);
+ _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(v0);
+ _write_z(":\n\0");
+end;
+
+proc _compile_label_declaration();
+begin
+ (* Skip the dot. *)
+ _advance_token(1);
+ v0 := _read_token();
+ _write_c('.');
+ _write_s(source_code_position, v0);
+ _write_z(":\n\0");
+ _advance_token(v0);
+end;
+
+proc _compile_statement();
+begin
+ _skip_spaces();
+ v0 := _load_byte(source_code_position);
+
+ (* This is a call if the statement starts with an underscore. *)
+ if v0 = '_' 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 v0 = '.' 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();
+begin
+ .compile_procedure_body_loop;
+ _skip_empty_lines();
+ _skip_spaces();
+
+ v0 := _memcmp(source_code_position, "end", 3) = 0;
+ v4 := _memcmp(source_code_position, "else", 4) = 0;
+ v4 := v0 or v4;
+
+ if v4 = 0 then
+ _compile_statement();
+ goto .compile_procedure_body_loop;
+ end;
+end;
+
+(* Writes a regster name to the standard output. *)
+
+(* Parameters: *)
+(* a0 - Register character. *)
+(* a1 - Register number. *)
+proc _write_register(v88: Word, v84: Word);
+begin
+ _write_c(v88);
+ v84 := v84 + '0';
+ _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();
+ _advance_token(v0);
+end;
+
+(* Parameters: *)
+
+(* a0 - Parameter index. *)
+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);
+
+ 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: *)
+
+(* a0 - Parameter index. *)
+proc _read_procedure_parameter(v88: Word);
+begin
+ (* Read the parameter name. *)
+ v8 := source_code_position;
+ v0 := _read_token();
+ _advance_token(v0);
+
+ (* Skip colon and space in front of the type expression. *)
+ _advance_token(2);
+
+ _read_type_expression();
+
+ _write_z("\tsw a\0");
+ _write_i(v88);
+ _write_z(", \0");
+
+ 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");
+end;
+
+proc _read_procedure_parameters();
+begin
+ (* Skip open paren. *)
+ _advance_token(1);
+ v0 := 0;
+
+ .compile_procedure_prologue_skip;
+ if _load_byte(source_code_position) <> ')' then
+ _read_procedure_parameter(v0);
+ v0 := v0 + 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: *)
+(* 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();
+
+ (* Write .type _procedure_name, @function. *)
+ _write_z(".type \0");
+
+ _write_token(v0);
+ _write_z(", @function\n\0");
+
+ (* Write procedure label, _procedure_name: *)
+ _write_token(v0);
+ _write_z(":\n\0");
+
+ (* Skip procedure name. *)
+ _advance_token(v0);
+ _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;
+ v0 := _load_byte(source_code_position);
+
+ (* Check for newline character. *)
+ if v0 <> '\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();
+begin
+ .skip_empty_lines_rerun;
+ v0 := source_code_position;
+
+ .skip_empty_lines_loop;
+ v4 := _load_byte(v0);
+
+ if v4 = '\n' then
+ goto .skip_empty_lines_newline;
+ end;
+ if v4 = '\t' then
+ goto .skip_empty_lines_tab;
+ end;
+ if v4 <> '(' then
+ goto .skip_empty_lines_end;
+ end;
+ v4 := v0 + 1;
+
+ if _load_byte(v4) = '*' then
+ goto .skip_empty_lines_comment
+ end;
+ goto .skip_empty_lines_end;
+
+ .skip_empty_lines_comment;
+ source_code_position := v0;
+ _skip_comment();
+ goto .skip_empty_lines_rerun;
+
+ .skip_empty_lines_newline;
+ source_code_position := v0 + 1;
+ goto .skip_empty_lines_rerun;
+
+ .skip_empty_lines_tab;
+ v0 := v0 + 1;
+ goto .skip_empty_lines_loop
+
+ .skip_empty_lines_end;
+end;
+
+proc _compile_global_initializer();
+begin
+ v0 := _load_byte(source_code_position);
+
+ if v0 = '"' then
+ _write_z("\n\t.word strings + \0");
+ v4 := _string_length(source_code_position);
+
+ _add_string(source_code_position);
+ _write_i();
+
+ (* Skip the quoted string. *)
+ _advance_token(v4 + 2);
+
+ goto .compile_global_initializer_end;
+ end;
+ if v0 = '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 v0 = '@' then
+ (* Skip @. *)
+ _advance_token(1);
+ _write_z("\n\t.word \0");
+ v0 := _read_token();
+ _write_token(v0);
+ _advance_token(v0);
+
+ goto .compile_global_initializer_end;
+ end;
+ if _is_digit(v0) = 1 then
+ _write_z("\n\t.word \0");
+ v0 := _read_token();
+ _write_token(v0);
+ _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();
+begin
+ v0 := _read_token();
+
+ _write_z(".type \0");
+ _write_token(v0);
+ _write_z(", @object\n\0");
+
+ _write_token(v0);
+ _write_c(':');
+
+ (* Skip the constant name with assignment sign and surrounding whitespaces. *)
+ _advance_token(v0 + 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();
+begin
+ v0 := _read_token();
+
+ _write_z(".type \0");
+ _write_token(v0);
+ _write_z(", @object\n\0");
+
+ _write_token(v0);
+ _write_c(':');
+
+ (* Skip the variable name and colon with space before the type. *)
+ _advance_token(v0 + 2);
+
+ (* Skip the type name. *)
+ v4 := _read_token();
+ _advance_token(v4);
+
+ 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();
+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();
+ v0 := _load_byte(source_code_position);
+
+ if v0 = '\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();
+begin
+ _write_z(".globl _start\n\n\0");
+ _compile_module();
+
+ _write_z(".section .rodata\n.type strings, @object\nstrings: .ascii \0");
+ _write_c('"');
+
+ v0 := @compiler_strings;
+ v4 := compiler_strings_position;
+
+ .compile_loop;
+ if v0 < v4 then
+ v8 := _load_byte(v0);
+ v0 := v0 + 1;
+ _write_c(v8);
+
+ 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: *)
+(* 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 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(v88);
+
+ (* Calculate the offset for the new symbol. *)
+ v4 := v0 * 12;
+ v4 := v4 + 4;
+ v4 := v88 + v4;
+
+ _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, v88);
+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: *)
+(* a0 - Array pointer. *)
+(* a1 - Index (word offset into the array). *)
+(* a2 - Data to assign. *)
+proc _assign_at(v88: Word, v84: Word, v80: Word);
+begin
+ v0 := v84 + -1;
+ v0 := v0 * 4;
+ v0 := v88 + v0;
+
+ _store_word(v80, v0);
+end;
+
+proc _create_classification();
+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);
+
+ v0 := 129;
+
+(* Set the remaining 129 - 256 bytes to transitionClassOther. *)
+ .create_classification_loop;
+ _assign_at(@classification, v0, 22);
+ v0 := v0 + 1;
+
+ if v0 < 257 then
+ goto .create_classification_loop;
+ end;
+end;
+
+(* Parameters: *)
+(* a0 - Current state (first index into transitions table). *)
+(* a1 - Transition (second index into transitions table).. *)
+(* a2 - Action to assign. *)
+(* a3 - Next state to assign. *)
+proc _set_transition(v88: Word, v84: Word, v80: Word, v76: Word);
+begin
+ (* Transitions start at offset in classification array. Save the transitions start in v0. *)
+ v0 := @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. *)
+
+ v4 := v88 + -1;
+ v4 := v4 * 128;
+
+ v8 := v84 + -1;
+ v8 := v8 * 8;
+
+ v12 := v0 + v4;
+ v12 := v12 + v8;
+
+ _store_word(v80, v12);
+ v12 := v12 + 4;
+ _store_word(v76, v12);
+end;
+
+(* Parameters: *)
+(* a0 - Current state (Transition state enumeration). *)
+(* a1 - Default action (Callback). *)
+(* a2 - Next state (Transition state enumeration). *)
+proc _set_default_transition(v88: Word, v84: Word, v80: Word);
+begin
+ _set_transition(v88, 1, v84, v80);
+ _set_transition(v88, 2, v84, v80);
+ _set_transition(v88, 3, v84, v80);
+ _set_transition(v88, 4, v84, v80);
+ _set_transition(v88, 5, v84, v80);
+ _set_transition(v88, 6, v84, v80);
+ _set_transition(v88, 7, v84, v80);
+ _set_transition(v88, 8, v84, v80);
+ _set_transition(v88, 9, v84, v80);
+ _set_transition(v88, 10, v84, v80);
+ _set_transition(v88, 11, v84, v80);
+ _set_transition(v88, 12, v84, v80);
+ _set_transition(v88, 13, v84, v80);
+ _set_transition(v88, 14, v84, v80);
+ _set_transition(v88, 15, v84, v80);
+ _set_transition(v88, 16, v84, v80);
+ _set_transition(v88, 17, v84, v80);
+ _set_transition(v88, 18, v84, v80);
+ _set_transition(v88, 19, v84, v80);
+ _set_transition(v88, 20, v84, v80);
+ _set_transition(v88, 21, v84, v80);
+ _set_transition(v88, 22, v84, v80);
+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();
+begin
+ (* Lexer state is saved after the transition tables. The offset is 256 + 16 * 22. *)
+ v0 := @classification;
+ v4 := 16 * 22;
+ v0 := v0 + 256;
+
+ return v0 + v4
+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();
+begin
+ (* Transition start state is 1. *)
+ v0 := _lexer_get_state();
+ _store_word(1, v4);
+
+ (* Text pointer to the beginning of the currently read token. *)
+ v4 := _lexer_get_current();
+ _store_word(source_code_position, v4);
+
+ (* 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. *)
+ v4 := @source_code;
+
+ .start_read;
+ (* Second argument is buffer size. Modifying update the source_code definition. *)
+ v0 := _read_file(v4, 81920);
+ if v0 > 0 then
+ v4 := v4 + v0;
+ goto .start_read;
+ end;
+ _compile();
+
+ _exit(0);
+end;
generated by cgit v1.2.3 (git 2.26.2) at 2026-06-26 21:42:53 +0000