elna/source.elna
2025-04-12 12:05:32 +02:00

999 lines
24 KiB
Plaintext

const
SEEK_SET* := 0;
SEEK_CUR* := 1;
SEEK_END* := 2;
type
TokenKind* = (
unknown,
identifier,
_if,
_then,
_else,
_elsif,
_while,
_do,
_proc,
_begin,
_end,
_extern,
_const,
_var,
array,
_of,
_type,
_record,
_union,
pointer,
to,
boolean,
_nil,
and,
_or,
not,
_return,
_cast,
shift_left,
shift_right,
left_paren,
right_paren,
left_square,
right_square,
greater_equal,
less_equal,
greater_than,
less_than,
not_equal,
equal,
semicolon,
dot,
comma,
plus,
minus,
multiplication,
division,
remainder,
assignment,
colon,
hat,
at,
comment,
integer,
word,
character,
string,
_defer,
exclamation,
arrow
);
Position* = record
line: Word;
column: Word
end;
Location* = record
first: Position;
last: Position
end;
SourceFile* = record
buffer: [1024]Char;
handle: ^FILE;
size: Word;
index: Word
end;
FILE* = record end;
StringBuffer* = record
data: ^Byte;
size: Word;
capacity: Word
end;
SourceCode = record
position: Position;
input: ^Byte;
empty: proc(^Byte) -> Bool;
advance: proc(^Byte);
head: proc(^Byte) -> Char
end;
Token* = record
kind: TokenKind;
value: union
int_value: Int;
string: String;
boolean_value: Bool;
char_value: Char
end;
location: Location
end;
CommandLine* = record
input: ^Char;
tokenize: Bool;
syntax_tree: Bool
end;
(*
External procedures.
*)
proc fopen(pathname: ^Char, mode: ^Char) -> ^FILE; extern
proc fclose(stream: ^FILE) -> Int; extern
proc fseek(stream: ^FILE, off: Int, whence: Int) -> Int; extern
proc rewind(stream: ^FILE); extern
proc ftell(stream: ^FILE) -> Int; extern
proc fread(ptr: ^Byte, size: Word, nmemb: Word, stream: ^FILE) -> Word; extern
proc write(fd: Int, buf: ^Byte, Word: Int) -> Int; extern
proc malloc(size: Word) -> ^Byte; extern
proc free(ptr: ^Byte); extern
proc calloc(nmemb: Word, size: Word) -> ^Byte; extern
proc realloc(ptr: ^Byte, size: Word) -> ^Byte; extern
proc memset(ptr: ^Char, c: Int, n: Int) -> ^Char; extern
proc strcmp(s1: ^Char, s2: ^Char) -> Int; extern
proc strncmp(s1: ^Char, s2: ^Char, n: Word) -> Int; extern
proc strncpy(dst: ^Char, src: ^Char, dsize: Word) -> ^Char; extern
proc strcpy(dst: ^Char, src: ^Char) -> ^Char; extern
proc strlen(ptr: ^Char) -> Word; extern
proc perror(s: ^Char); extern
proc exit(code: Int) -> !; extern
(*
Standard procedures.
*)
proc reallocarray(ptr: ^Byte, n: Word, size: Word) -> ^Byte;
begin
return realloc(ptr, n * size)
end
proc write_s(value: String);
begin
write(0, cast(value.ptr: ^Byte), cast(value.length: Int))
end
proc write_z(value: ^Char);
begin
write(0, cast(value: ^Byte), cast(strlen(value): Int))
end
proc write_b(value: Bool);
begin
if value then
write_s("true")
else
write_s("false")
end
end
proc write_c(value: Char);
begin
write(0, cast(@value: ^Byte), 1)
end
proc write_i(value: Int);
var
digit: Int;
n: Word;
buffer: [10]Char;
begin
n := 10u;
if value = 0 then
write_c('0')
end;
while value <> 0 do
digit := value % 10;
value := value / 10;
buffer[n] := cast(cast('0': Int) + digit: Char);
n := n - 1u
end;
while n < 10u do
n := n + 1u;
write_c(buffer[n])
end
end
proc write_u(value: Word);
begin
write_i(cast(value: Int))
end
proc is_digit(c: Char) -> Bool;
begin
return cast(c: Int) >= cast('0': Int) & cast(c: Int) <= cast('9': Int)
end
proc is_alpha(c: Char) -> Bool;
begin
return cast(c: Int) >= cast('A': Int) & cast(c: Int) <= cast('z': Int)
end
proc is_alnum(c: Char) -> Bool;
begin
return is_digit(c) or is_alpha(c)
end
proc is_space(c: Char) -> Bool;
begin
return c = ' ' or c = '\n' or c = '\t'
end
proc substring(string: String, start: Word, count: Word) -> String;
begin
return String(string.ptr + start, count)
end
proc open_substring(string: String, start: Word) -> String;
begin
return substring(string, start, string.length - start)
end
proc string_dup(origin: String) -> String;
var
copy: ^Char;
begin
copy := cast(malloc(origin.length): ^Char);
strncpy(copy, origin.ptr, origin.length);
return String(copy, origin.length)
end
proc string_buffer_new() -> StringBuffer;
var
result: StringBuffer;
begin
result.capacity := 64u;
result.data := malloc(result.capacity);
result.size := 0u;
return result
end
proc string_buffer_push(buffer: ^StringBuffer, char: Char);
begin
if buffer^.size >= buffer^.capacity then
buffer^.capacity := buffer^.capacity + 1024u;
buffer^.data := realloc(buffer^.data, buffer^.capacity)
end;
(buffer^.data + buffer^.size)^ := cast(char: Byte);
buffer^.size := buffer^.size + 1u
end
proc string_buffer_pop(buffer: ^StringBuffer, count: Word);
begin
buffer^.size := buffer^.size - count
end
proc string_buffer_clear(buffer: ^StringBuffer) -> String;
var
result: String;
begin
result := String(cast(buffer^.data: ^Char), buffer^.size);
buffer^.size := 0u;
return result
end
(*
End of standard procedures.
*)
proc make_position() -> Position;
begin
return Position(1u, 1u)
end
proc read_source(filename: ^Char) -> ^SourceFile;
var
result: ^SourceFile;
file_handle: ^FILE;
begin
file_handle := fopen(filename, "rb\0".ptr);
if file_handle <> nil then
result := cast(malloc(#size(SourceFile)): ^SourceFile);
result^.handle := file_handle;
result^.size := 0u;
result^.index := 1u
end;
return result
end
proc escape_char(escape: Char, result: ^Char) -> Bool;
var
successful: Bool;
begin
if escape = 'n' then
result^ := '\n';
successful := true;
elsif escape = 'a' then
result^ := '\a';
successful := true
elsif escape = 'b' then
result^ := '\b';
successful := true
elsif escape = 't' then
result^ := '\t';
successful := true
elsif escape = 'f' then
result^ := '\f';
successful := true
elsif escape = 'r' then
result^ := '\r';
successful := true
elsif escape = 'v' then
result^ := '\v';
successful := true
elsif escape = '\\' then
result^ := '\\';
successful := true
elsif escape = '\'' then
result^ := '\'';
successful := true
elsif escape = '"' then
result^ := '"';
successful := true
elsif escape = '?' then
result^ := '\?';
successful := true
elsif escape = '0' then
result^ := '\0';
successful := true
else
successful := false
end;
return successful
end
proc source_file_empty(source_input: ^Byte) -> Bool;
var
source_file: ^SourceFile;
begin
source_file := cast(source_input: ^SourceFile);
if source_file^.index > source_file^.size then
source_file^.size := fread(cast(@source_file^.buffer: ^Byte), 1u, 1024u, source_file^.handle);
source_file^.index := 1u
end;
return source_file^.size = 0u
end
proc source_file_head(source_input: ^Byte) -> Char;
var
source_file: ^SourceFile;
begin
source_file := cast(source_input: ^SourceFile);
return source_file^.buffer[source_file^.index]
end
proc source_file_advance(source_input: ^Byte);
var
source_file: ^SourceFile;
begin
source_file := cast(source_input: ^SourceFile);
source_file^.index := source_file^.index + 1u
end
proc source_code_empty(source_code: ^SourceCode) -> Bool;
begin
return source_code^.empty(source_code^.input)
end
proc source_code_head(source_code: SourceCode) -> Char;
begin
return source_code.head(source_code.input)
end
proc source_code_advance(source_code: ^SourceCode);
begin
source_code^.advance(source_code^.input);
source_code^.position.column := source_code^.position.column
end
proc source_code_break(source_code: ^SourceCode);
begin
source_code^.position.line := source_code^.position.line + 1u;
source_code^.position.column := 0u
end
proc source_code_expect(source_code: ^SourceCode, expected: Char) -> Bool;
begin
return ~source_code_empty(source_code) & source_code_head(source_code^) = expected
end
proc skip_spaces(source_code: ^SourceCode);
var
current: Char;
begin
while ~source_code_empty(source_code), loop do
current := source_code_head(source_code^);
if ~is_space(current) then
break loop
elsif current = '\n' then
source_code_break(source_code)
end;
source_code_advance(source_code)
end
end
proc is_ident(char: Char) -> Bool;
begin
return is_alnum(char) or char = '_'
end
proc lex_identifier(source_code: ^SourceCode, token_content: ^StringBuffer);
var
content_length: Word;
begin
while ~source_code_empty(source_code) & is_ident(source_code_head(source_code^)) do
string_buffer_push(token_content, source_code_head(source_code^));
source_code_advance(source_code)
end
end
proc lex_comment(source_code: ^SourceCode, token_content: ^StringBuffer) -> Bool;
var
trailing: Word;
begin
trailing := 0u;
while ~source_code_empty(source_code) & trailing < 2u do
if source_code_head(source_code^) = '*' then
string_buffer_push(token_content, '*');
trailing := 1u
elsif source_code_head(source_code^) = ')' & trailing = 1u then
string_buffer_pop(token_content, 1u);
trailing := 2u
else
string_buffer_push(token_content, source_code_head(source_code^));
trailing := 0u
end;
source_code_advance(source_code)
end;
return trailing = 2u
end
proc lex_character(source_code: ^SourceCode, token_content: ^Char) -> Bool;
var
successful: Bool;
begin
successful := ~source_code_empty(source_code);
if successful then
if source_code_head(source_code^) = '\\' then
source_code_advance(source_code);
successful := ~source_code_empty(source_code) & escape_char(source_code_head(source_code^), token_content)
else
token_content^ := source_code_head(source_code^);
successful := true
end
end;
if successful then
source_code_advance(source_code)
end;
return successful
end
proc lex_string(source_code: ^SourceCode, token_content: ^StringBuffer) -> Bool;
var
token_end, constructed_string: ^Char;
token_length: Word;
is_valid: Bool;
next_char: Char;
begin
is_valid := true;
while is_valid & ~source_code_empty(source_code) & source_code_head(source_code^) <> '"' do
is_valid := lex_character(source_code, @next_char);
if is_valid then
string_buffer_push(token_content, next_char)
end
end;
if is_valid & source_code_expect(source_code, '"') then
source_code_advance(source_code)
else
is_valid := false
end;
return is_valid
end
proc lex_number(source_code: ^SourceCode, token_content: ^Int);
begin
token_content^ := 0;
while ~source_code_empty(source_code) & is_digit(source_code_head(source_code^)) do
token_content^ := token_content^ * 10 + (cast(source_code_head(source_code^): Int) - cast('0': Int));
source_code_advance(source_code)
end
end
proc print_tokens(tokens: ^Token, tokens_size: Word);
var
current_token: ^Token;
i: Word;
begin
i := 0u;
while i < tokens_size do
current_token := tokens + i;
case current_token^.kind of
TokenKind._if:
write_s("IF")
| TokenKind._then:
write_s("THEN")
| TokenKind._else:
write_s("ELSE")
| TokenKind._elsif:
write_s("ELSIF")
| TokenKind._while:
write_s("WHILE")
| TokenKind._do:
write_s("DO")
| TokenKind._proc:
write_s("PROC")
| TokenKind._begin:
write_s("BEGIN")
| TokenKind._end:
write_s("END")
| TokenKind._extern:
write_s("EXTERN")
| TokenKind._const:
write_s("CONST")
| TokenKind._var:
write_s("VAR")
| TokenKind.array:
write_s("ARRAY")
| TokenKind._of:
write_s("OF")
| TokenKind._type:
write_s("TYPE")
| TokenKind._record:
write_s("RECORD")
| TokenKind._union:
write_s("UNION")
| TokenKind.pointer:
write_s("POINTER")
| TokenKind.to:
write_s("TO")
| TokenKind.boolean:
write_s("BOOLEAN<");
write_b(current_token^.value.boolean_value);
write_c('>')
| TokenKind._nil:
write_s("NIL")
| TokenKind.and:
write_s("AND")
| TokenKind._or:
write_s("OR")
| TokenKind.not:
write_s("NOT")
| TokenKind._return:
write_s("RETURN")
| TokenKind._cast:
write_s("CAST")
| TokenKind.shift_left:
write_s("<<")
| TokenKind.shift_right:
write_s(">>")
| TokenKind.identifier:
write_c('<');
write_s(current_token^.value.string);
write_c('>')
| TokenKind.left_paren:
write_s("(")
| TokenKind.right_paren:
write_s(")")
| TokenKind.left_square:
write_s("[")
| TokenKind.right_square:
write_s("]")
| TokenKind.greater_equal:
write_s(">=")
| TokenKind.less_equal:
write_s("<=")
| TokenKind.greater_than:
write_s(">")
| TokenKind.less_than:
write_s("<")
| TokenKind.equal:
write_s("=")
| TokenKind.not_equal:
write_s("<>")
| TokenKind.semicolon:
write_c(';')
| TokenKind.dot:
write_c('.')
| TokenKind.comma:
write_c(',')
| TokenKind.plus:
write_c('+')
| TokenKind.minus:
write_c('-')
| TokenKind.multiplication:
write_c('*')
| TokenKind.division:
write_c('/')
| TokenKind.remainder:
write_c('%')
| TokenKind.assignment:
write_s(":=")
| TokenKind.colon:
write_c(':')
| TokenKind.hat:
write_c('^')
| TokenKind.at:
write_c('@')
| TokenKind.comment:
write_s("(* COMMENT *)")
| TokenKind.integer:
write_c('<');
write_i(current_token^.value.int_value);
write_c('>')
| TokenKind.word:
write_c('<');
write_i(current_token^.value.int_value);
write_s("u>")
| TokenKind.character:
write_c('<');
write_i(cast(current_token^.value.char_value: Int));
write_s("c>")
| TokenKind.string:
write_s("\"...\"")
| TokenKind._defer:
write_s("DEFER")
| TokenKind.exclamation:
write_c('!')
| TokenKind.arrow:
write_s("->")
else
write_s("UNKNOWN<");
write_i(cast(current_token^.kind: Int));
write_c('>')
end;
write_c(' ');
i := i + 1u;
end;
write_c('\n')
end
proc categorize_identifier(token_content: String) -> Token;
var
current_token: Token;
begin
if "if" = token_content then
current_token.kind := TokenKind._if
elsif "then" = token_content then
current_token.kind := TokenKind._then
elsif "else" = token_content then
current_token.kind := TokenKind._else
elsif "elsif" = token_content then
current_token.kind := TokenKind._elsif
elsif "while" = token_content then
current_token.kind := TokenKind._while
elsif "do" = token_content then
current_token.kind := TokenKind._do
elsif "proc" = token_content then
current_token.kind := TokenKind._proc
elsif "begin" = token_content then
current_token.kind := TokenKind._begin
elsif "end" = token_content then
current_token.kind := TokenKind._end
elsif "extern" = token_content then
current_token.kind := TokenKind._extern
elsif "const" = token_content then
current_token.kind := TokenKind._const
elsif "var" = token_content then
current_token.kind := TokenKind._var
elsif "array" = token_content then
current_token.kind := TokenKind.array
elsif "of" = token_content then
current_token.kind := TokenKind._of
elsif "type" = token_content then
current_token.kind := TokenKind._type
elsif "record" = token_content then
current_token.kind := TokenKind._record
elsif "union" = token_content then
current_token.kind := TokenKind._union
elsif "pointer" = token_content then
current_token.kind := TokenKind.pointer
elsif "to" = token_content then
current_token.kind := TokenKind.to
elsif "true" = token_content then
current_token.kind := TokenKind.boolean;
current_token.value.boolean_value := true
elsif "false" = token_content then
current_token.kind := TokenKind.boolean;
current_token.value.boolean_value := false
elsif "nil" = token_content then
current_token.kind := TokenKind._nil
elsif "and" = token_content then
current_token.kind := TokenKind.and
elsif "or" = token_content then
current_token.kind := TokenKind._or
elsif "not" = token_content then
current_token.kind := TokenKind.not
elsif "return" = token_content then
current_token.kind := TokenKind._return
elsif "cast" = token_content then
current_token.kind := TokenKind._cast
elsif "defer" = token_content then
current_token.kind := TokenKind._defer
else
current_token.kind := TokenKind.identifier;
current_token.value.string := string_dup(token_content)
end;
return current_token
end
proc tokenize(source_code: SourceCode, tokens_size: ^Word) -> ^Token;
var
tokens, current_token: ^Token;
first_char: Char;
token_buffer: StringBuffer;
begin
tokens_size^ := 0u;
tokens := nil;
token_buffer := string_buffer_new();
skip_spaces(@source_code);
while ~source_code_empty(@source_code) do
tokens := cast(reallocarray(cast(tokens: ^Byte), tokens_size^ + 1u, #size(Token)): ^Token);
current_token := tokens + tokens_size^;
first_char := source_code_head(source_code);
if is_alpha(first_char) or first_char = '_' then
lex_identifier(@source_code, @token_buffer);
current_token^ := categorize_identifier(string_buffer_clear(@token_buffer))
elsif is_digit(first_char) then
lex_number(@source_code, @current_token^.value.int_value);
if source_code_expect(@source_code, 'u') then
current_token^.kind := TokenKind.word;
source_code_advance(@source_code)
else
current_token^.kind := TokenKind.integer
end
elsif first_char = '(' then
source_code_advance(@source_code);
if source_code_empty(@source_code) then
current_token^.kind := TokenKind.left_paren
elsif source_code_head(source_code) = '*' then
source_code_advance(@source_code);
if lex_comment(@source_code, @token_buffer) then
current_token^.value.string := string_dup(string_buffer_clear(@token_buffer));
current_token^.kind := TokenKind.comment
else
current_token^.kind := TokenKind.unknown
end
else
current_token^.kind := TokenKind.left_paren
end
elsif first_char = ')' then
current_token^.kind := TokenKind.right_paren;
source_code_advance(@source_code)
elsif first_char = '\'' then
source_code_advance(@source_code);
if lex_character(@source_code, @current_token^.value.char_value) & source_code_expect(@source_code, '\'') then
current_token^.kind := TokenKind.character;
source_code_advance(@source_code)
else
current_token^.kind := TokenKind.unknown
end
elsif first_char = '"' then
source_code_advance(@source_code);
if lex_string(@source_code, @token_buffer) then
current_token^.kind := TokenKind.string;
current_token^.value.string := string_dup(string_buffer_clear(@token_buffer))
else
current_token^.kind := TokenKind.unknown
end
elsif first_char = '[' then
current_token^.kind := TokenKind.left_square;
source_code_advance(@source_code)
elsif first_char = ']' then
current_token^.kind := TokenKind.right_square;
source_code_advance(@source_code)
elsif first_char = '>' then
source_code_advance(@source_code);
if source_code_empty(@source_code) then
current_token^.kind := TokenKind.greater_than
elsif source_code_head(source_code) = '=' then
current_token^.kind := TokenKind.greater_equal;
source_code_advance(@source_code)
elsif source_code_head(source_code) = '>' then
current_token^.kind := TokenKind.shift_right;
source_code_advance(@source_code)
else
current_token^.kind := TokenKind.greater_than
end
elsif first_char = '<' then
source_code_advance(@source_code);
if source_code_empty(@source_code) then
current_token^.kind := TokenKind.less_than
elsif source_code_head(source_code) = '=' then
current_token^.kind := TokenKind.less_equal;
source_code_advance(@source_code)
elsif source_code_head(source_code) = '<' then
current_token^.kind := TokenKind.shift_left;
source_code_advance(@source_code)
elsif source_code_head(source_code) = '>' then
current_token^.kind := TokenKind.not_equal;
source_code_advance(@source_code)
else
current_token^.kind := TokenKind.less_than
end
elsif first_char = '=' then
current_token^.kind := TokenKind.equal;
source_code_advance(@source_code)
elsif first_char = ';' then
current_token^.kind := TokenKind.semicolon;
source_code_advance(@source_code)
elsif first_char = '.' then
current_token^.kind := TokenKind.dot;
source_code_advance(@source_code)
elsif first_char = ',' then
current_token^.kind := TokenKind.comma;
source_code_advance(@source_code)
elsif first_char = '+' then
current_token^.kind := TokenKind.plus;
source_code_advance(@source_code)
elsif first_char = '-' then
source_code_advance(@source_code);
if source_code_empty(@source_code) then
current_token^.kind := TokenKind.minus
elsif source_code_head(source_code) = '>' then
current_token^.kind := TokenKind.arrow;
source_code_advance(@source_code)
else
current_token^.kind := TokenKind.minus
end
elsif first_char = '*' then
current_token^.kind := TokenKind.multiplication;
source_code_advance(@source_code)
elsif first_char = '/' then
current_token^.kind := TokenKind.division;
source_code_advance(@source_code)
elsif first_char = '%' then
current_token^.kind := TokenKind.remainder;
source_code_advance(@source_code)
elsif first_char = ':' then
source_code_advance(@source_code);
if source_code_empty(@source_code) then
current_token^.kind := TokenKind.colon
elsif source_code_head(source_code) = '=' then
current_token^.kind := TokenKind.assignment;
source_code_advance(@source_code)
else
current_token^.kind := TokenKind.colon
end
elsif first_char = '^' then
current_token^.kind := TokenKind.hat;
source_code_advance(@source_code)
elsif first_char = '@' then
current_token^.kind := TokenKind.at;
source_code_advance(@source_code)
elsif first_char = '!' then
current_token^.kind := TokenKind.exclamation;
source_code_advance(@source_code)
else
current_token^.kind := TokenKind.unknown;
source_code_advance(@source_code)
end;
if current_token^.kind <> TokenKind.unknown then
tokens_size^ := tokens_size^ + 1u;
skip_spaces(@source_code)
else
write_s("Lexical analysis error on \"");
write_c(first_char);
write_s("\".\n")
end
end;
return tokens
end
proc parse_command_line*(argc: Int, argv: ^^Char) -> ^CommandLine;
var
parameter: ^^Char;
i: Int;
result: ^CommandLine;
begin
i := 1;
result := cast(malloc(#size(CommandLine)): ^CommandLine);
result^.tokenize := false;
result^.syntax_tree := false;
result^.input := nil;
while i < argc do
parameter := argv + i;
if strcmp(parameter^, "--tokenize\0".ptr) = 0 then
result^.tokenize := true
elsif strcmp(parameter^, "--syntax-tree\0".ptr) = 0 then
result^.syntax_tree := true
elsif parameter^^ <> '-' then
result^.input := parameter^
else
write_s("Fatal error: Unknown command line options:");
write_c(' ');
write_z(parameter^);
write_s(".\n");
return nil
end;
i := i + 1
end;
if result^.input = nil then
write_s("Fatal error: no input files.\n");
return nil
end;
return result
end
proc process(argc: Int, argv: ^^Char) -> Int;
var
tokens: ^Token;
tokens_size: Word;
source_code: SourceCode;
command_line: ^CommandLine;
return_code: Int;
begin
return_code := 0;
command_line := parse_command_line(argc, argv);
if command_line = nil then
return_code := 2
end;
if return_code = 0 then
source_code.position := make_position();
source_code.input := cast(read_source(command_line^.input): ^Byte);
source_code.empty := source_file_empty;
source_code.head := source_file_head;
source_code.advance := source_file_advance;
if source_code.input = nil then
perror(command_line^.input);
return_code := 3
end
end;
if return_code = 0 then
tokens := tokenize(source_code, @tokens_size);
fclose(cast(source_code.input: ^SourceFile)^.handle);
if command_line^.tokenize then
print_tokens(tokens, tokens_size)
end
end;
return return_code
end
begin
exit(process(count, parameters))
end.