mirror of
https://github.com/Ed94/gencpp.git
synced 2024-12-22 15:54:45 -08:00
reduced macro usage in lexer.cpp
This commit is contained in:
parent
30dea2e9fd
commit
8d48da0b9e
@ -175,48 +175,6 @@ global FixedArena_256KB Lexer_defines_map_arena;
|
|||||||
global StringTable Lexer_defines;
|
global StringTable Lexer_defines;
|
||||||
global Array(Token) Lexer_Tokens;
|
global Array(Token) Lexer_Tokens;
|
||||||
|
|
||||||
#define current ( * ctx->scanner )
|
|
||||||
|
|
||||||
#define move_forward() \
|
|
||||||
{ \
|
|
||||||
if ( current == '\n' ) \
|
|
||||||
{ \
|
|
||||||
ctx->line++; \
|
|
||||||
ctx->column = 1; \
|
|
||||||
} \
|
|
||||||
else \
|
|
||||||
{ \
|
|
||||||
ctx->column++; \
|
|
||||||
} \
|
|
||||||
ctx->left--; \
|
|
||||||
ctx->scanner++; \
|
|
||||||
}
|
|
||||||
|
|
||||||
#define skip_whitespace() \
|
|
||||||
while ( ctx->left && char_is_space( current ) ) \
|
|
||||||
{ \
|
|
||||||
move_forward(); \
|
|
||||||
}
|
|
||||||
|
|
||||||
#define end_line() \
|
|
||||||
do \
|
|
||||||
{ \
|
|
||||||
while ( ctx->left && current == ' ' ) \
|
|
||||||
{ \
|
|
||||||
move_forward(); \
|
|
||||||
} \
|
|
||||||
if ( ctx->left && current == '\r' ) \
|
|
||||||
{ \
|
|
||||||
move_forward(); \
|
|
||||||
move_forward(); \
|
|
||||||
} \
|
|
||||||
else if ( ctx->left && current == '\n' ) \
|
|
||||||
{ \
|
|
||||||
move_forward(); \
|
|
||||||
} \
|
|
||||||
} \
|
|
||||||
while (0)
|
|
||||||
|
|
||||||
enum
|
enum
|
||||||
{
|
{
|
||||||
Lex_Continue,
|
Lex_Continue,
|
||||||
@ -234,6 +192,44 @@ struct LexContext
|
|||||||
Token token;
|
Token token;
|
||||||
};
|
};
|
||||||
|
|
||||||
|
forceinline
|
||||||
|
void lexer_move_forward( LexContext* ctx )
|
||||||
|
{
|
||||||
|
if ( * ctx->scanner == '\n' ) {
|
||||||
|
ctx->line += 1;
|
||||||
|
ctx->column = 1;
|
||||||
|
}
|
||||||
|
else {
|
||||||
|
++ ctx->column;
|
||||||
|
}
|
||||||
|
-- ctx->left;
|
||||||
|
++ ctx->scanner;
|
||||||
|
}
|
||||||
|
#define move_forward() lexer_move_forward(ctx)
|
||||||
|
|
||||||
|
forceinline
|
||||||
|
void lexer_skip_whitespace( LexContext* ctx )
|
||||||
|
{
|
||||||
|
while ( ctx->left && char_is_space( * ctx->scanner ) )
|
||||||
|
move_forward();
|
||||||
|
}
|
||||||
|
#define skip_whitespace() lexer_skip_whitespace(ctx)
|
||||||
|
|
||||||
|
forceinline
|
||||||
|
void lexer_end_line( LexContext* ctx )
|
||||||
|
{
|
||||||
|
while ( ctx->left && (* ctx->scanner) == ' ' )
|
||||||
|
move_forward();
|
||||||
|
|
||||||
|
if ( ctx->left && (* ctx->scanner) == '\r' ) {
|
||||||
|
move_forward();
|
||||||
|
move_forward();
|
||||||
|
}
|
||||||
|
else if ( ctx->left && (* ctx->scanner) == '\n' )
|
||||||
|
move_forward();
|
||||||
|
}
|
||||||
|
#define end_line() lexer_end_line(ctx)
|
||||||
|
|
||||||
forceinline
|
forceinline
|
||||||
s32 lex_preprocessor_directive( LexContext* ctx )
|
s32 lex_preprocessor_directive( LexContext* ctx )
|
||||||
{
|
{
|
||||||
@ -245,7 +241,7 @@ s32 lex_preprocessor_directive( LexContext* ctx )
|
|||||||
skip_whitespace();
|
skip_whitespace();
|
||||||
|
|
||||||
ctx->token.Text = ctx->scanner;
|
ctx->token.Text = ctx->scanner;
|
||||||
while (ctx->left && ! char_is_space(current) )
|
while (ctx->left && ! char_is_space((* ctx->scanner)) )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
ctx->token.Length++;
|
ctx->token.Length++;
|
||||||
@ -263,24 +259,24 @@ s32 lex_preprocessor_directive( LexContext* ctx )
|
|||||||
s32 within_char = false;
|
s32 within_char = false;
|
||||||
while ( ctx->left )
|
while ( ctx->left )
|
||||||
{
|
{
|
||||||
if ( current == '"' && ! within_char )
|
if ( * ctx->scanner == '"' && ! within_char )
|
||||||
within_string ^= true;
|
within_string ^= true;
|
||||||
|
|
||||||
if ( current == '\'' && ! within_string )
|
if ( * ctx->scanner == '\'' && ! within_string )
|
||||||
within_char ^= true;
|
within_char ^= true;
|
||||||
|
|
||||||
if ( current == '\\' && ! within_string && ! within_char )
|
if ( * ctx->scanner == '\\' && ! within_string && ! within_char )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
ctx->token.Length++;
|
ctx->token.Length++;
|
||||||
|
|
||||||
if ( current == '\r' )
|
if ( (* ctx->scanner) == '\r' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
ctx->token.Length++;
|
ctx->token.Length++;
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( current == '\n' )
|
if ( (* ctx->scanner) == '\n' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
ctx->token.Length++;
|
ctx->token.Length++;
|
||||||
@ -290,19 +286,19 @@ s32 lex_preprocessor_directive( LexContext* ctx )
|
|||||||
{
|
{
|
||||||
log_failure( "gen::Parser::lex: Invalid escape sequence '\\%c' (%d, %d)"
|
log_failure( "gen::Parser::lex: Invalid escape sequence '\\%c' (%d, %d)"
|
||||||
" in preprocessor directive (%d, %d)\n%.100s"
|
" in preprocessor directive (%d, %d)\n%.100s"
|
||||||
, current, ctx->line, ctx->column
|
, (* ctx->scanner), ctx->line, ctx->column
|
||||||
, ctx->token.Line, ctx->token.Column, ctx->token.Text );
|
, ctx->token.Line, ctx->token.Column, ctx->token.Text );
|
||||||
break;
|
break;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( current == '\r' )
|
if ( (* ctx->scanner) == '\r' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
ctx->token.Length++;
|
ctx->token.Length++;
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( current == '\n' )
|
if ( (* ctx->scanner) == '\n' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
ctx->token.Length++;
|
ctx->token.Length++;
|
||||||
@ -343,13 +339,13 @@ s32 lex_preprocessor_directive( LexContext* ctx )
|
|||||||
name.Length = 1;
|
name.Length = 1;
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
while ( ctx->left && ( char_is_alphanumeric(current) || current == '_' ) )
|
while ( ctx->left && ( char_is_alphanumeric((* ctx->scanner)) || (* ctx->scanner) == '_' ) )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
name.Length++;
|
name.Length++;
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( ctx->left && current == '(' )
|
if ( ctx->left && (* ctx->scanner) == '(' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
name.Length++;
|
name.Length++;
|
||||||
@ -367,12 +363,12 @@ s32 lex_preprocessor_directive( LexContext* ctx )
|
|||||||
{
|
{
|
||||||
preprocess_content.Type = Tok_String;
|
preprocess_content.Type = Tok_String;
|
||||||
|
|
||||||
if ( current != '"' && current != '<' )
|
if ( (* ctx->scanner) != '"' && (* ctx->scanner) != '<' )
|
||||||
{
|
{
|
||||||
String directive_str = string_fmt_buf( GlobalAllocator, "%.*s", min( 80, ctx->left + preprocess_content.Length ), ctx->token.Text );
|
String directive_str = string_fmt_buf( GlobalAllocator, "%.*s", min( 80, ctx->left + preprocess_content.Length ), ctx->token.Text );
|
||||||
|
|
||||||
log_failure( "gen::Parser::lex: Expected '\"' or '<' after #include, not '%c' (%d, %d)\n%s"
|
log_failure( "gen::Parser::lex: Expected '\"' or '<' after #include, not '%c' (%d, %d)\n%s"
|
||||||
, current
|
, (* ctx->scanner)
|
||||||
, preprocess_content.Line
|
, preprocess_content.Line
|
||||||
, preprocess_content.Column
|
, preprocess_content.Column
|
||||||
, (char*) directive_str
|
, (char*) directive_str
|
||||||
@ -382,7 +378,7 @@ s32 lex_preprocessor_directive( LexContext* ctx )
|
|||||||
move_forward();
|
move_forward();
|
||||||
preprocess_content.Length++;
|
preprocess_content.Length++;
|
||||||
|
|
||||||
while ( ctx->left && current != '"' && current != '>' )
|
while ( ctx->left && (* ctx->scanner) != '"' && (* ctx->scanner) != '>' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
preprocess_content.Length++;
|
preprocess_content.Length++;
|
||||||
@ -391,12 +387,12 @@ s32 lex_preprocessor_directive( LexContext* ctx )
|
|||||||
move_forward();
|
move_forward();
|
||||||
preprocess_content.Length++;
|
preprocess_content.Length++;
|
||||||
|
|
||||||
if ( current == '\r' && ctx->scanner[1] == '\n' )
|
if ( (* ctx->scanner) == '\r' && ctx->scanner[1] == '\n' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
move_forward();
|
move_forward();
|
||||||
}
|
}
|
||||||
else if ( current == '\n' )
|
else if ( (* ctx->scanner) == '\n' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
}
|
}
|
||||||
@ -411,24 +407,24 @@ s32 lex_preprocessor_directive( LexContext* ctx )
|
|||||||
// SkipWhitespace();
|
// SkipWhitespace();
|
||||||
while ( ctx->left )
|
while ( ctx->left )
|
||||||
{
|
{
|
||||||
if ( current == '"' && ! within_char )
|
if ( (* ctx->scanner) == '"' && ! within_char )
|
||||||
within_string ^= true;
|
within_string ^= true;
|
||||||
|
|
||||||
if ( current == '\'' && ! within_string )
|
if ( (* ctx->scanner) == '\'' && ! within_string )
|
||||||
within_char ^= true;
|
within_char ^= true;
|
||||||
|
|
||||||
if ( current == '\\' && ! within_string && ! within_char )
|
if ( (* ctx->scanner) == '\\' && ! within_string && ! within_char )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
preprocess_content.Length++;
|
preprocess_content.Length++;
|
||||||
|
|
||||||
if ( current == '\r' )
|
if ( (* ctx->scanner) == '\r' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
preprocess_content.Length++;
|
preprocess_content.Length++;
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( current == '\n' )
|
if ( (* ctx->scanner) == '\n' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
preprocess_content.Length++;
|
preprocess_content.Length++;
|
||||||
@ -441,20 +437,20 @@ s32 lex_preprocessor_directive( LexContext* ctx )
|
|||||||
|
|
||||||
log_failure( "gen::Parser::lex: Invalid escape sequence '\\%c' (%d, %d)"
|
log_failure( "gen::Parser::lex: Invalid escape sequence '\\%c' (%d, %d)"
|
||||||
" in preprocessor directive '%s' (%d, %d)\n%s"
|
" in preprocessor directive '%s' (%d, %d)\n%s"
|
||||||
, current, ctx->line, ctx->column
|
, (* ctx->scanner), ctx->line, ctx->column
|
||||||
, directive_str, preprocess_content.Line, preprocess_content.Column
|
, directive_str, preprocess_content.Line, preprocess_content.Column
|
||||||
, content_str );
|
, content_str );
|
||||||
break;
|
break;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( current == '\r' )
|
if ( (* ctx->scanner) == '\r' )
|
||||||
{
|
{
|
||||||
break;
|
break;
|
||||||
//move_forward();
|
//move_forward();
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( current == '\n' )
|
if ( (* ctx->scanner) == '\n' )
|
||||||
{
|
{
|
||||||
//move_forward();
|
//move_forward();
|
||||||
break;
|
break;
|
||||||
@ -493,7 +489,7 @@ void lex_found_token( LexContext* ctx )
|
|||||||
{
|
{
|
||||||
skip_whitespace();
|
skip_whitespace();
|
||||||
|
|
||||||
if ( current != '"' )
|
if ( (* ctx->scanner) != '"' )
|
||||||
{
|
{
|
||||||
type = Tok_Spec_Extern;
|
type = Tok_Spec_Extern;
|
||||||
ctx->token.Flags |= TF_Specifier;
|
ctx->token.Flags |= TF_Specifier;
|
||||||
@ -523,7 +519,7 @@ void lex_found_token( LexContext* ctx )
|
|||||||
}
|
}
|
||||||
|
|
||||||
u64 key = 0;
|
u64 key = 0;
|
||||||
if ( current == '(')
|
if ( (* ctx->scanner) == '(')
|
||||||
key = crc32( ctx->token.Text, ctx->token.Length + 1 );
|
key = crc32( ctx->token.Text, ctx->token.Length + 1 );
|
||||||
else
|
else
|
||||||
key = crc32( ctx->token.Text, ctx->token.Length );
|
key = crc32( ctx->token.Text, ctx->token.Length );
|
||||||
@ -534,18 +530,18 @@ void lex_found_token( LexContext* ctx )
|
|||||||
ctx->token.Type = Tok_Preprocess_Macro;
|
ctx->token.Type = Tok_Preprocess_Macro;
|
||||||
|
|
||||||
// Want to ignore any arguments the define may have as they can be execution expressions.
|
// Want to ignore any arguments the define may have as they can be execution expressions.
|
||||||
if ( ctx->left && current == '(' )
|
if ( ctx->left && (* ctx->scanner) == '(' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
ctx->token.Length++;
|
ctx->token.Length++;
|
||||||
|
|
||||||
s32 level = 0;
|
s32 level = 0;
|
||||||
while ( ctx->left && (current != ')' || level > 0) )
|
while ( ctx->left && ((* ctx->scanner) != ')' || level > 0) )
|
||||||
{
|
{
|
||||||
if ( current == '(' )
|
if ( (* ctx->scanner) == '(' )
|
||||||
level++;
|
level++;
|
||||||
|
|
||||||
else if ( current == ')' && level > 0 )
|
else if ( (* ctx->scanner) == ')' && level > 0 )
|
||||||
level--;
|
level--;
|
||||||
|
|
||||||
move_forward();
|
move_forward();
|
||||||
@ -556,12 +552,12 @@ void lex_found_token( LexContext* ctx )
|
|||||||
ctx->token.Length++;
|
ctx->token.Length++;
|
||||||
}
|
}
|
||||||
|
|
||||||
//if ( current == '\r' && ctx->scanner[1] == '\n' )
|
//if ( (* ctx->scanner) == '\r' && ctx->scanner[1] == '\n' )
|
||||||
//{
|
//{
|
||||||
// move_forward();
|
// move_forward();
|
||||||
// ctx->token.Length++;
|
// ctx->token.Length++;
|
||||||
//}
|
//}
|
||||||
//else if ( current == '\n' )
|
//else if ( (* ctx->scanner) == '\n' )
|
||||||
//{
|
//{
|
||||||
// move_forward();
|
// move_forward();
|
||||||
// ctx->token.Length++;
|
// ctx->token.Length++;
|
||||||
@ -637,13 +633,13 @@ TokArray lex( StrC content )
|
|||||||
|
|
||||||
if ( c.column == 1 )
|
if ( c.column == 1 )
|
||||||
{
|
{
|
||||||
if ( current == '\r')
|
if ( (* ctx->scanner) == '\r')
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length = 1;
|
c.token.Length = 1;
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( current == '\n' )
|
if ( (* ctx->scanner) == '\n' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
@ -661,12 +657,8 @@ TokArray lex( StrC content )
|
|||||||
if ( c.left <= 0 )
|
if ( c.left <= 0 )
|
||||||
break;
|
break;
|
||||||
|
|
||||||
switch ( current )
|
switch ( (* ctx->scanner) )
|
||||||
{
|
{
|
||||||
if (array_back(Lexer_Tokens)->Length > 100 ) {
|
|
||||||
__debugbreak();
|
|
||||||
}
|
|
||||||
|
|
||||||
case '#':
|
case '#':
|
||||||
{
|
{
|
||||||
s32 result = lex_preprocessor_directive( ctx );
|
s32 result = lex_preprocessor_directive( ctx );
|
||||||
@ -681,13 +673,13 @@ TokArray lex( StrC content )
|
|||||||
Token thanks_c = { c.scanner, 0, Tok_Invalid, c.line, c.column, TF_Null };
|
Token thanks_c = { c.scanner, 0, Tok_Invalid, c.line, c.column, TF_Null };
|
||||||
c.token = thanks_c;
|
c.token = thanks_c;
|
||||||
}
|
}
|
||||||
if ( current == '\r')
|
if ( (* ctx->scanner) == '\r')
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length = 1;
|
c.token.Length = 1;
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( current == '\n' )
|
if ( (* ctx->scanner) == '\n' )
|
||||||
{
|
{
|
||||||
c.token.Type = Tok_NewLine;
|
c.token.Type = Tok_NewLine;
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
@ -696,7 +688,6 @@ TokArray lex( StrC content )
|
|||||||
array_append( Lexer_Tokens, c.token );
|
array_append( Lexer_Tokens, c.token );
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
|
|
||||||
@ -718,10 +709,10 @@ TokArray lex( StrC content )
|
|||||||
move_forward();
|
move_forward();
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( current == '.' )
|
if ( (* ctx->scanner) == '.' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
if( current == '.' )
|
if( (* ctx->scanner) == '.' )
|
||||||
{
|
{
|
||||||
c.token.Length = 3;
|
c.token.Length = 3;
|
||||||
c.token.Type = Tok_Varadic_Argument;
|
c.token.Type = Tok_Varadic_Argument;
|
||||||
@ -732,7 +723,7 @@ TokArray lex( StrC content )
|
|||||||
{
|
{
|
||||||
String context_str = string_fmt_buf( GlobalAllocator, "%s", c.scanner, min( 100, c.left ) );
|
String context_str = string_fmt_buf( GlobalAllocator, "%s", c.scanner, min( 100, c.left ) );
|
||||||
|
|
||||||
log_failure( "gen::lex: invalid varadic argument, expected '...' got '..%c' (%d, %d)\n%s", current, c.line, c.column, context_str );
|
log_failure( "gen::lex: invalid varadic argument, expected '...' got '..%c' (%d, %d)\n%s", (* ctx->scanner), c.line, c.column, context_str );
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
@ -749,7 +740,7 @@ TokArray lex( StrC content )
|
|||||||
if (c.left)
|
if (c.left)
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
if ( current == '&' ) // &&
|
if ( (* ctx->scanner) == '&' ) // &&
|
||||||
{
|
{
|
||||||
c.token.Length = 2;
|
c.token.Length = 2;
|
||||||
c.token.Type = Tok_Ampersand_DBL;
|
c.token.Type = Tok_Ampersand_DBL;
|
||||||
@ -771,7 +762,7 @@ TokArray lex( StrC content )
|
|||||||
if (c.left)
|
if (c.left)
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
if ( current == ':' )
|
if ( (* ctx->scanner) == ':' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Type = Tok_Access_StaticSymbol;
|
c.token.Type = Tok_Access_StaticSymbol;
|
||||||
@ -811,7 +802,7 @@ TokArray lex( StrC content )
|
|||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
if ( current == ']' )
|
if ( (* ctx->scanner) == ']' )
|
||||||
{
|
{
|
||||||
c.token.Length = 2;
|
c.token.Length = 2;
|
||||||
c.token.Type = Tok_Operator;
|
c.token.Type = Tok_Operator;
|
||||||
@ -859,19 +850,19 @@ TokArray lex( StrC content )
|
|||||||
|
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
if ( c.left && current == '\\' )
|
if ( c.left && (* ctx->scanner) == '\\' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
|
|
||||||
if ( current == '\'' )
|
if ( (* ctx->scanner) == '\'' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
while ( c.left && current != '\'' )
|
while ( c.left && (* ctx->scanner) != '\'' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
@ -906,7 +897,7 @@ TokArray lex( StrC content )
|
|||||||
if (c.left)
|
if (c.left)
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
if ( current == '=' )
|
if ( (* ctx->scanner) == '=' )
|
||||||
{
|
{
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
c.token.Flags |= TF_Assign;
|
c.token.Flags |= TF_Assign;
|
||||||
@ -941,13 +932,13 @@ TokArray lex( StrC content )
|
|||||||
move_forward();
|
move_forward();
|
||||||
while ( c.left )
|
while ( c.left )
|
||||||
{
|
{
|
||||||
if ( current == '"' )
|
if ( (* ctx->scanner) == '"' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
break;
|
break;
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( current == '\\' )
|
if ( (* ctx->scanner) == '\\' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
@ -990,7 +981,7 @@ TokArray lex( StrC content )
|
|||||||
if (c.left)
|
if (c.left)
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
if ( current == '=' )
|
if ( (* ctx->scanner) == '=' )
|
||||||
{
|
{
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
c.token.Flags = TF_Operator;
|
c.token.Flags = TF_Operator;
|
||||||
@ -1045,7 +1036,7 @@ TokArray lex( StrC content )
|
|||||||
if (c.left)
|
if (c.left)
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
if ( current == '=' )
|
if ( (* ctx->scanner) == '=' )
|
||||||
{
|
{
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
c.token.Flags |= TF_Assign;
|
c.token.Flags |= TF_Assign;
|
||||||
@ -1055,7 +1046,7 @@ TokArray lex( StrC content )
|
|||||||
if (c.left)
|
if (c.left)
|
||||||
move_forward();
|
move_forward();
|
||||||
}
|
}
|
||||||
else while ( c.left && current == *(c.scanner - 1) && c.token.Length < 3 )
|
else while ( c.left && (* ctx->scanner) == *(c.scanner - 1) && c.token.Length < 3 )
|
||||||
{
|
{
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
|
|
||||||
@ -1077,21 +1068,21 @@ TokArray lex( StrC content )
|
|||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
if ( current == '>' )
|
if ( (* ctx->scanner) == '>' )
|
||||||
{
|
{
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
// token.Type = Tok_Access_PointerToMemberSymbol;
|
// token.Type = Tok_Access_PointerToMemberSymbol;
|
||||||
c.token.Flags |= TF_AccessOperator;
|
c.token.Flags |= TF_AccessOperator;
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
if ( current == '*' )
|
if ( (* ctx->scanner) == '*' )
|
||||||
{
|
{
|
||||||
// token.Type = Tok_Access_PointerToMemberOfPointerSymbol;
|
// token.Type = Tok_Access_PointerToMemberOfPointerSymbol;
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
move_forward();
|
move_forward();
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
else if ( current == '=' )
|
else if ( (* ctx->scanner) == '=' )
|
||||||
{
|
{
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
// token.Type = Tok_Assign_Subtract;
|
// token.Type = Tok_Assign_Subtract;
|
||||||
@ -1100,7 +1091,7 @@ TokArray lex( StrC content )
|
|||||||
if (c.left)
|
if (c.left)
|
||||||
move_forward();
|
move_forward();
|
||||||
}
|
}
|
||||||
else while ( c.left && current == *(c.scanner - 1) && c.token.Length < 3 )
|
else while ( c.left && (* ctx->scanner) == *(c.scanner - 1) && c.token.Length < 3 )
|
||||||
{
|
{
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
|
|
||||||
@ -1121,32 +1112,32 @@ TokArray lex( StrC content )
|
|||||||
|
|
||||||
if ( c.left )
|
if ( c.left )
|
||||||
{
|
{
|
||||||
if ( current == '=' )
|
if ( (* ctx->scanner) == '=' )
|
||||||
{
|
{
|
||||||
// token.Type = TokeType::Assign_Divide;
|
// token.Type = TokeType::Assign_Divide;
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
c.token.Flags = TF_Assign;
|
c.token.Flags = TF_Assign;
|
||||||
}
|
}
|
||||||
else if ( current == '/' )
|
else if ( (* ctx->scanner) == '/' )
|
||||||
{
|
{
|
||||||
c.token.Type = Tok_Comment;
|
c.token.Type = Tok_Comment;
|
||||||
c.token.Length = 2;
|
c.token.Length = 2;
|
||||||
c.token.Flags = TF_Null;
|
c.token.Flags = TF_Null;
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
while ( c.left && current != '\n' && current != '\r' )
|
while ( c.left && (* ctx->scanner) != '\n' && (* ctx->scanner) != '\r' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( current == '\r' )
|
if ( (* ctx->scanner) == '\r' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
}
|
}
|
||||||
if ( current == '\n' )
|
if ( (* ctx->scanner) == '\n' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
@ -1154,14 +1145,14 @@ TokArray lex( StrC content )
|
|||||||
array_append( Lexer_Tokens, c.token );
|
array_append( Lexer_Tokens, c.token );
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
else if ( current == '*' )
|
else if ( (* ctx->scanner) == '*' )
|
||||||
{
|
{
|
||||||
c.token.Type = Tok_Comment;
|
c.token.Type = Tok_Comment;
|
||||||
c.token.Length = 2;
|
c.token.Length = 2;
|
||||||
c.token.Flags = TF_Null;
|
c.token.Flags = TF_Null;
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
bool star = current == '*';
|
bool star = (* ctx->scanner) == '*';
|
||||||
bool slash = c.scanner[1] == '/';
|
bool slash = c.scanner[1] == '/';
|
||||||
bool at_end = star && slash;
|
bool at_end = star && slash;
|
||||||
while ( c.left && ! at_end )
|
while ( c.left && ! at_end )
|
||||||
@ -1169,7 +1160,7 @@ TokArray lex( StrC content )
|
|||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
|
|
||||||
star = current == '*';
|
star = (* ctx->scanner) == '*';
|
||||||
slash = c.scanner[1] == '/';
|
slash = c.scanner[1] == '/';
|
||||||
at_end = star && slash;
|
at_end = star && slash;
|
||||||
}
|
}
|
||||||
@ -1177,12 +1168,12 @@ TokArray lex( StrC content )
|
|||||||
move_forward();
|
move_forward();
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
if ( current == '\r' )
|
if ( (* ctx->scanner) == '\r' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
}
|
}
|
||||||
if ( current == '\n' )
|
if ( (* ctx->scanner) == '\n' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
@ -1196,13 +1187,13 @@ TokArray lex( StrC content )
|
|||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( char_is_alpha( current ) || current == '_' )
|
if ( char_is_alpha( (* ctx->scanner) ) || (* ctx->scanner) == '_' )
|
||||||
{
|
{
|
||||||
c.token.Text = c.scanner;
|
c.token.Text = c.scanner;
|
||||||
c.token.Length = 1;
|
c.token.Length = 1;
|
||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
while ( c.left && ( char_is_alphanumeric(current) || current == '_' ) )
|
while ( c.left && ( char_is_alphanumeric((* ctx->scanner)) || (* ctx->scanner) == '_' ) )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
@ -1210,7 +1201,7 @@ TokArray lex( StrC content )
|
|||||||
|
|
||||||
goto FoundToken;
|
goto FoundToken;
|
||||||
}
|
}
|
||||||
else if ( char_is_digit(current) )
|
else if ( char_is_digit((* ctx->scanner)) )
|
||||||
{
|
{
|
||||||
// This is a very brute force lex, no checks are done for validity of literal.
|
// This is a very brute force lex, no checks are done for validity of literal.
|
||||||
|
|
||||||
@ -1221,15 +1212,15 @@ TokArray lex( StrC content )
|
|||||||
move_forward();
|
move_forward();
|
||||||
|
|
||||||
if (c.left
|
if (c.left
|
||||||
&& ( current == 'x' || current == 'X'
|
&& ( (* ctx->scanner) == 'x' || (* ctx->scanner) == 'X'
|
||||||
|| current == 'b' || current == 'B'
|
|| (* ctx->scanner) == 'b' || (* ctx->scanner) == 'B'
|
||||||
|| current == 'o' || current == 'O' )
|
|| (* ctx->scanner) == 'o' || (* ctx->scanner) == 'O' )
|
||||||
)
|
)
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
|
|
||||||
while ( c.left && char_is_hex_digit(current) )
|
while ( c.left && char_is_hex_digit((* ctx->scanner)) )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
@ -1238,18 +1229,18 @@ TokArray lex( StrC content )
|
|||||||
goto FoundToken;
|
goto FoundToken;
|
||||||
}
|
}
|
||||||
|
|
||||||
while ( c.left && char_is_digit(current) )
|
while ( c.left && char_is_digit((* ctx->scanner)) )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( c.left && current == '.' )
|
if ( c.left && (* ctx->scanner) == '.' )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
|
|
||||||
while ( c.left && char_is_digit(current) )
|
while ( c.left && char_is_digit((* ctx->scanner)) )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
@ -1257,18 +1248,18 @@ TokArray lex( StrC content )
|
|||||||
|
|
||||||
// Handle number literal suffixes in a botched way
|
// Handle number literal suffixes in a botched way
|
||||||
if (c.left && (
|
if (c.left && (
|
||||||
current == 'l' || current == 'L' || // long/long long
|
(* ctx->scanner) == 'l' || (* ctx->scanner) == 'L' || // long/long long
|
||||||
current == 'u' || current == 'U' || // unsigned
|
(* ctx->scanner) == 'u' || (* ctx->scanner) == 'U' || // unsigned
|
||||||
current == 'f' || current == 'F' || // float
|
(* ctx->scanner) == 'f' || (* ctx->scanner) == 'F' || // float
|
||||||
current == 'i' || current == 'I' || // imaginary
|
(* ctx->scanner) == 'i' || (* ctx->scanner) == 'I' || // imaginary
|
||||||
current == 'z' || current == 'Z')) // complex
|
(* ctx->scanner) == 'z' || (* ctx->scanner) == 'Z')) // complex
|
||||||
{
|
{
|
||||||
char prev = current;
|
char prev = (* ctx->scanner);
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
|
|
||||||
// Handle 'll'/'LL' as a special case when we just processed an 'l'/'L'
|
// Handle 'll'/'LL' as a special case when we just processed an 'l'/'L'
|
||||||
if (c.left && (prev == 'l' || prev == 'L') && (current == 'l' || current == 'L'))
|
if (c.left && (prev == 'l' || prev == 'L') && ((* ctx->scanner) == 'l' || (* ctx->scanner) == 'L'))
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
@ -1292,10 +1283,10 @@ TokArray lex( StrC content )
|
|||||||
}
|
}
|
||||||
|
|
||||||
String context_str = string_fmt_buf( GlobalAllocator, "%.*s", min( 100, c.left ), c.scanner );
|
String context_str = string_fmt_buf( GlobalAllocator, "%.*s", min( 100, c.left ), c.scanner );
|
||||||
log_failure( "Failed to lex token '%c' (%d, %d)\n%s", current, c.line, c.column, context_str );
|
log_failure( "Failed to lex token '%c' (%d, %d)\n%s", (* ctx->scanner), c.line, c.column, context_str );
|
||||||
|
|
||||||
// Skip to next whitespace since we can't know if anything else is valid until then.
|
// Skip to next whitespace since we can't know if anything else is valid until then.
|
||||||
while ( c.left && ! char_is_space( current ) )
|
while ( c.left && ! char_is_space( (* ctx->scanner) ) )
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
}
|
}
|
||||||
@ -1309,13 +1300,13 @@ TokArray lex( StrC content )
|
|||||||
{
|
{
|
||||||
Token thanks_c = { c.scanner, 0, Tok_Invalid, c.line, c.column, TF_Null };
|
Token thanks_c = { c.scanner, 0, Tok_Invalid, c.line, c.column, TF_Null };
|
||||||
c.token = thanks_c;
|
c.token = thanks_c;
|
||||||
if ( current == '\r')
|
if ( (* ctx->scanner) == '\r')
|
||||||
{
|
{
|
||||||
move_forward();
|
move_forward();
|
||||||
c.token.Length = 1;
|
c.token.Length = 1;
|
||||||
}
|
}
|
||||||
|
|
||||||
if ( current == '\n' )
|
if ( (* ctx->scanner) == '\n' )
|
||||||
{
|
{
|
||||||
c.token.Type = Tok_NewLine;
|
c.token.Type = Tok_NewLine;
|
||||||
c.token.Length++;
|
c.token.Length++;
|
||||||
@ -1342,8 +1333,8 @@ TokArray lex( StrC content )
|
|||||||
TokArray result = { Lexer_Tokens, 0 };
|
TokArray result = { Lexer_Tokens, 0 };
|
||||||
return result;
|
return result;
|
||||||
}
|
}
|
||||||
#undef current
|
|
||||||
#undef move_forward
|
#undef move_forward
|
||||||
#undef SkipWhitespace
|
#undef skip_whitespace
|
||||||
|
#undef end_line
|
||||||
|
|
||||||
GEN_NS_PARSER_END
|
GEN_NS_PARSER_END
|
||||||
|
Loading…
Reference in New Issue
Block a user