ed/gencpp
1
0
Fork 0
mirror of https://github.com/Ed94/gencpp.git synced 2025-04-23 09:20:16 -07:00
Code Issues Packages Projects Releases Wiki Activity

progress on parser and lexer revamp

Browse Source
This commit is contained in:
Edward R. Gonzalez 2025-03-19 11:48:40 -04:00
parent 1c7dd4ab32
commit ad5cb6597b
6 changed files with 141 additions and 125 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
base/components/interface.hpp
View File

@ -78,9 +78,6 @@ struct Context
StringTable StrCache;
// TODO(Ed): This needs to be just handled by a parser context
Array(Token) Lexer_Tokens;
// TODO(Ed): Active parse context vs a parse result need to be separated conceptually
ParseContext parser;

185
base/components/interface.parsing.cpp
View File

@ -22,6 +22,7 @@ ParseInfo wip_parse_str(LexedInfo lexed, ParseOpts* opts)
// TODO(Ed): ParseInfo should be set to the parser context.
ctx->parser = struct_zero();
ctx->parser.tokens = lexed.tokens;
ParseStackNode scope = NullScope;
@ -40,17 +41,16 @@ CodeClass parse_class( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
return InvalidCode;
ctx->parser = struct_zero();
ctx->parser.Tokens = toks;
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ParseStackNode scope = NullScope;
parser_push(& ctx->parser, & scope);
CodeClass result = (CodeClass) parse_class_struct( ctx, Tok_Decl_Class, parser_not_inplace_def );
parser_pop(& ctx->parser);
return result;
}
@ -62,10 +62,16 @@ CodeConstructor parse_constructor(Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ParseStackNode scope = NullScope;
parser_push(& ctx->parser, & scope);
// TODO(Ed): Constructors can have prefix attributes
CodeSpecifiers specifiers = NullCode;
@ -92,7 +98,7 @@ CodeConstructor parse_constructor(Str def )
break;
default :
log_failure( "Invalid specifier %s for variable\n%S", spec_to_str( spec ), parser_to_strbuilder(ctx->parser, ctx->Allocator_Temp) );
log_failure( "Invalid specifier %s for variable\n%S", spec_to_str( spec ), parser_to_strbuilder(& ctx->parser, ctx->Allocator_Temp) );
parser_pop(& ctx->parser);
return InvalidCode;
}
@ -106,14 +112,13 @@ CodeConstructor parse_constructor(Str def )
eat( currtok.Type );
}
if ( NumSpecifiers )
{
if ( NumSpecifiers ) {
specifiers = def_specifiers_arr( NumSpecifiers, specs_found );
// <specifiers> ...
}
ctx->parser.Tokens = toks;
CodeConstructor result = parser_parse_constructor(ctx, specifiers);
parser_pop(& ctx->parser);
return result;
}
@ -124,17 +129,16 @@ CodeDefine parse_define( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
return InvalidCode;
ctx->parser = struct_zero();
ctx->parser.Tokens = toks;
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ParseStackNode scope = NullScope;
parser_push(& ctx->parser, & scope);
CodeDefine result = parser_parse_define(ctx);
parser_pop(& ctx->parser);
return result;
}
@ -146,14 +150,16 @@ CodeDestructor parse_destructor( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
// TODO(Ed): Destructors can have prefix attributes
// TODO(Ed): Destructors can have virtual
ctx->parser.Tokens = toks;
CodeDestructor result = parser_parse_destructor(ctx, NullCode);
return result;
}
@ -165,17 +171,14 @@ CodeEnum parse_enum( Str def )
check_parse_args( def );
ParseStackNode scope = NullScope;
parser_push(& ctx->parser, & scope);
ctx->parser = struct_zero();
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
{
parser_pop(& ctx->parser);
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr ) {
return InvalidCode;
}
ctx->parser.Tokens = toks;
return parser_parse_enum(ctx, parser_not_inplace_def);
}
@ -186,11 +189,13 @@ CodeBody parse_export_body( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ctx->parser.Tokens = toks;
return parser_parse_export_body(ctx);
}
@ -201,11 +206,13 @@ CodeExtern parse_extern_link( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ctx->parser.Tokens = toks;
return parser_parse_extern_link(ctx);
}
@ -216,11 +223,13 @@ CodeFriend parse_friend( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ctx->parser.Tokens = toks;
return parser_parse_friend(ctx);
}
@ -231,11 +240,13 @@ CodeFn parse_function( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ctx->parser.Tokens = toks;
return (CodeFn) parser_parse_function(ctx);
}
@ -246,17 +257,16 @@ CodeBody parse_global_body( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
return InvalidCode;
ctx->parser = struct_zero();
ctx->parser.Tokens = toks;
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ParseStackNode scope = NullScope;
parser_push(& ctx->parser, & scope);
CodeBody result = parse_global_nspace(ctx, CT_Global_Body );
parser_pop(& ctx->parser);
return result;
}
@ -268,11 +278,13 @@ CodeNS parse_namespace( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ctx->parser.Tokens = toks;
return parser_parse_namespace(ctx);
}
@ -283,11 +295,13 @@ CodeOperator parse_operator( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ctx->parser.Tokens = toks;
return (CodeOperator) parser_parse_operator(ctx);
}
@ -298,11 +312,13 @@ CodeOpCast parse_operator_cast( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ctx->parser.Tokens = toks;
return parser_parse_operator_cast(ctx, NullCode);
}
@ -313,17 +329,16 @@ CodeStruct parse_struct( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
return InvalidCode;
ctx->parser = struct_zero();
ctx->parser.Tokens = toks;
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ParseStackNode scope = NullScope;
parser_push(& ctx->parser, & scope);
CodeStruct result = (CodeStruct) parse_class_struct( ctx, Tok_Decl_Struct, parser_not_inplace_def );
parser_pop(& ctx->parser);
return result;
}
@ -335,11 +350,13 @@ CodeTemplate parse_template( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ctx->parser.Tokens = toks;
return parser_parse_template(ctx);
}
@ -350,11 +367,13 @@ CodeTypename parse_type( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ctx->parser.Tokens = toks;
return parser_parse_type( ctx, parser_not_from_template, nullptr);
}
@ -365,11 +384,13 @@ CodeTypedef parse_typedef( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ctx->parser.Tokens = toks;
return parser_parse_typedef(ctx);
}
@ -380,11 +401,13 @@ CodeUnion parse_union( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ctx->parser.Tokens = toks;
return parser_parse_union(ctx, parser_not_inplace_def);
}
@ -395,11 +418,13 @@ CodeUsing parse_using( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ctx->parser.Tokens = toks;
return parser_parse_using(ctx);
}
@ -410,11 +435,13 @@ CodeVar parse_variable( Str def )
check_parse_args( def );
TokArray toks = lex( def );
if ( toks.Arr == nullptr )
ctx->parser = struct_zero();
LexedInfo lexed = lex(ctx, def);
ctx->parser.tokens = lexed.tokens;
if ( ctx->parser.tokens.ptr == nullptr )
return InvalidCode;
ctx->parser.Tokens = toks;
return parser_parse_variable(ctx);
}

2
base/components/interface.untyped.cpp
View File

@ -179,7 +179,7 @@ Code untyped_token_fmt( s32 num_tokens, char const* fmt, ... )
Code untyped_toks( TokenSlice tokens )
{
if ( tokens.Num == 0 ) {
if ( tokens.num == 0 ) {
log_failure( "untyped_toks: empty token slice" );
return InvalidCode;
}

69
base/components/lexer.cpp
View File

@ -137,7 +137,7 @@ s32 lex_preprocessor_define( LexContext* ctx )
);
// GEN_DEBUG_TRAP();
}
array_append( _ctx->Lexer_Tokens, name );
array_append( ctx->tokens, name );
if ( ctx->left && (* ctx->scanner) == '(' )
{
@ -152,7 +152,7 @@ s32 lex_preprocessor_define( LexContext* ctx )
}
Token opening_paren = { { ctx->scanner, 1 }, Tok_Paren_Open, ctx->line, ctx->column, TF_Preprocess };
array_append( _ctx->Lexer_Tokens, opening_paren );
array_append( ctx->tokens, opening_paren );
move_forward();
Token last_parameter = {};
@ -168,7 +168,7 @@ s32 lex_preprocessor_define( LexContext* ctx )
move_forward();
move_forward();
array_append(_ctx->Lexer_Tokens, parameter);
array_append(ctx->tokens, parameter);
skip_whitespace();
last_parameter = parameter;
@ -202,7 +202,7 @@ s32 lex_preprocessor_define( LexContext* ctx )
move_forward();
parameter.Text.Len++;
}
array_append(_ctx->Lexer_Tokens, parameter);
array_append(ctx->tokens, parameter);
skip_whitespace();
last_parameter = parameter;
}
@ -229,7 +229,7 @@ s32 lex_preprocessor_define( LexContext* ctx )
return Lex_ReturnNull;
}
Token comma = { { ctx->scanner, 1 }, Tok_Comma, ctx->line, ctx->column, TF_Preprocess };
array_append(_ctx->Lexer_Tokens, comma);
array_append(ctx->tokens, comma);
move_forward();
}
@ -243,7 +243,7 @@ s32 lex_preprocessor_define( LexContext* ctx )
return Lex_ReturnNull;
}
Token closing_paren = { { ctx->scanner, 1 }, Tok_Paren_Close, ctx->line, ctx->column, TF_Preprocess };
array_append(_ctx->Lexer_Tokens, closing_paren);
array_append(ctx->tokens, closing_paren);
move_forward();
}
else if ( registered_macro && macro_is_functional( * registered_macro) ) {
@ -268,7 +268,7 @@ s32 lex_preprocessor_directive( LexContext* ctx )
{
char const* hash = ctx->scanner;
Token hash_tok = { { hash, 1 }, Tok_Preprocess_Hash, ctx->line, ctx->column, TF_Preprocess };
array_append( _ctx->Lexer_Tokens, hash_tok );
array_append(ctx->tokens, hash_tok);
move_forward();
skip_whitespace();
@ -344,14 +344,14 @@ s32 lex_preprocessor_directive( LexContext* ctx )
ctx->token.Text.Len = ctx->token.Text.Len + ctx->token.Text.Ptr - hash;
ctx->token.Text.Ptr = hash;
array_append( _ctx->Lexer_Tokens, ctx->token );
array_append(ctx->tokens, ctx->token);
return Lex_Continue; // Skip found token, its all handled here.
}
if ( ctx->token.Type == Tok_Preprocess_Else || ctx->token.Type == Tok_Preprocess_EndIf )
{
ctx->token.Flags |= TF_Preprocess_Cond;
array_append( _ctx->Lexer_Tokens, ctx->token );
array_append(ctx->tokens, ctx->token);
end_line();
return Lex_Continue;
}
@ -360,7 +360,7 @@ s32 lex_preprocessor_directive( LexContext* ctx )
ctx->token.Flags |= TF_Preprocess_Cond;
}
array_append( _ctx->Lexer_Tokens, ctx->token );
array_append(ctx->tokens, ctx->token);
skip_whitespace();
@ -411,7 +411,7 @@ s32 lex_preprocessor_directive( LexContext* ctx )
move_forward();
}
array_append( _ctx->Lexer_Tokens, preprocess_content );
array_append(ctx->tokens, preprocess_content);
return Lex_Continue; // Skip found token, its all handled here.
}
@ -475,14 +475,14 @@ s32 lex_preprocessor_directive( LexContext* ctx )
preprocess_content.Text.Len++;
}
array_append( _ctx->Lexer_Tokens, preprocess_content );
array_append(ctx->tokens, preprocess_content);
return Lex_Continue; // Skip found token, its all handled here.
}
void lex_found_token( LexContext* ctx )
{
if ( ctx->token.Type != Tok_Invalid ) {
array_append( _ctx->Lexer_Tokens, ctx->token );
array_append(ctx->tokens, ctx->token);
return;
}
@ -508,7 +508,7 @@ void lex_found_token( LexContext* ctx )
}
ctx->token.Type = type;
array_append( _ctx->Lexer_Tokens, ctx->token );
array_append(ctx->tokens, ctx->token);
return;
}
if ( ( type <= Tok_Star && type >= Tok_Spec_Alignas)
@ -517,13 +517,13 @@ void lex_found_token( LexContext* ctx )
{
ctx->token.Type = type;
ctx->token.Flags |= TF_Specifier;
array_append( _ctx->Lexer_Tokens, ctx->token );
array_append(ctx->tokens, ctx->token);
return;
}
if ( type != Tok_Invalid )
{
ctx->token.Type = type;
array_append( _ctx->Lexer_Tokens, ctx->token );
array_append(ctx->tokens, ctx->token);
return;
}
@ -561,7 +561,7 @@ void lex_found_token( LexContext* ctx )
ctx->token.Type = Tok_Identifier;
}
array_append( _ctx->Lexer_Tokens, ctx->token );
array_append(ctx->tokens, ctx->token);
}
// TODO(Ed): We should dynamically allocate the lexer's array in Allocator_DyanmicContainers.
@ -579,8 +579,7 @@ LexedInfo lex(Context* lib_ctx, Str content)
c.scanner = content.Ptr;
c.line = 1;
c.column = 1;
Array(Token) tokens = array_init_reserve(Token, lib_ctx->Allocator_DyanmicContainers, lib_ctx->InitSize_LexerTokens );
c.tokens = array_init_reserve(Token, lib_ctx->Allocator_DyanmicContainers, lib_ctx->InitSize_LexerTokens );
// TODO(Ed): Re-implement to new constraints:
// 1. Ability to continue on error
@ -592,18 +591,10 @@ LexedInfo lex(Context* lib_ctx, Str content)
return info;
}
array_clear(_ctx->Lexer_Tokens);
b32 preprocess_args = true;
while (c.left )
{
#if 0
if (Tokens.num()) {
log_fmt("\nLastTok: %SB", Tokens.back().to_strbuilder());
}
#endif
c.token = struct_init(Token) { { c.scanner, 0 }, Tok_Invalid, c.line, c.column, TF_Null };
bool is_define = false;
@ -623,7 +614,7 @@ LexedInfo lex(Context* lib_ctx, Str content)
c.token.Type = Tok_NewLine;
c.token.Text.Len++;
array_append( _ctx->Lexer_Tokens, c.token );
array_append(c.tokens, c.token);
continue;
}
}
@ -662,7 +653,7 @@ LexedInfo lex(Context* lib_ctx, Str content)
c.token.Text.Len++;
move_forward();
array_append( _ctx->Lexer_Tokens, c.token );
array_append(c.tokens, c.token);
}
}
continue;
@ -1118,7 +1109,7 @@ LexedInfo lex(Context* lib_ctx, Str content)
move_forward();
c.token.Text.Len++;
}
array_append( _ctx->Lexer_Tokens, c.token );
array_append(c.tokens, c.token);
continue;
}
else if ( (* ctx->scanner) == '*' )
@ -1154,7 +1145,7 @@ LexedInfo lex(Context* lib_ctx, Str content)
move_forward();
c.token.Text.Len++;
}
array_append( _ctx->Lexer_Tokens, c.token );
array_append(c.tokens, c.token);
// end_line();
continue;
}
@ -1242,14 +1233,14 @@ LexedInfo lex(Context* lib_ctx, Str content)
}
else
{
s32 start = max( 0, array_num(_ctx->Lexer_Tokens) - 100 );
s32 start = max( 0, array_num(c.tokens) - 100 );
log_fmt("\n%d\n", start);
for ( s32 idx = start; idx < array_num(_ctx->Lexer_Tokens); idx++ )
for ( s32 idx = start; idx < array_num(c.tokens); idx++ )
{
log_fmt( "Token %d Type: %s : %.*s\n"
, idx
, toktype_to_str( _ctx->Lexer_Tokens[ idx ].Type ).Ptr
, _ctx->Lexer_Tokens[ idx ].Text.Len, _ctx->Lexer_Tokens[ idx ].Text.Ptr
, toktype_to_str( c.tokens[ idx ].Type ).Ptr
, c.tokens[ idx ].Text.Len, c.tokens[ idx ].Text.Ptr
);
}
@ -1265,7 +1256,7 @@ LexedInfo lex(Context* lib_ctx, Str content)
FoundToken:
{
lex_found_token( ctx );
TokType last_type = array_back(_ctx->Lexer_Tokens)->Type;
TokType last_type = array_back(c.tokens)->Type;
if ( last_type == Tok_Preprocess_Macro_Stmt || last_type == Tok_Preprocess_Macro_Expr )
{
Token thanks_c = { { c.scanner, 0 }, Tok_Invalid, c.line, c.column, TF_Null };
@ -1280,21 +1271,21 @@ LexedInfo lex(Context* lib_ctx, Str content)
c.token.Text.Len++;
move_forward();
array_append( _ctx->Lexer_Tokens, c.token );
array_append(c.tokens, c.token);
continue;
}
}
}
}
if ( array_num(_ctx->Lexer_Tokens) == 0 ) {
if ( array_num(c.tokens) == 0 ) {
log_failure( "Failed to lex any tokens" );
return info;
}
info.messages = c.messages;
info.text = content;
info.tokens = struct_init(TokenSlice) { tokens, scast(s32, array_num(tokens)) };
info.tokens = struct_init(TokenSlice) { pcast(Token*, c.tokens), scast(s32, array_num(c.tokens)) };
return info;
}

6
base/components/parser.cpp
View File

@ -126,14 +126,14 @@ bool lex__eat(Context* ctx, ParseContext* parser, TokType type)
internal
void parser_init(Context* ctx)
{
ctx->Lexer_Tokens = array_init_reserve(Token, ctx->Allocator_DyanmicContainers, ctx->InitSize_LexerTokens );
// ctx->Lexer_Tokens = array_init_reserve(Token, ctx->Allocator_DyanmicContainers, ctx->InitSize_LexerTokens );
}
internal
void parser_deinit(Context* ctx)
{
Array(Token) null_array = { nullptr };
ctx->Lexer_Tokens = null_array;
// Array(Token) null_array = { nullptr };
// ctx->Lexer_Tokens = null_array;
}
#pragma region Helper Macros

1
base/components/parser_types.hpp
View File

@ -135,6 +135,7 @@ struct LexContext
s32 line;
s32 column;
Token token;
Array(Token) tokens;
};
struct LexedInfo