123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301 |
- #ifndef GDSCRIPT_TOKENIZER_H
- #define GDSCRIPT_TOKENIZER_H
- #include "core/pair.h"
- #include "core/string_db.h"
- #include "core/ustring.h"
- #include "core/variant.h"
- #include "core/vmap.h"
- #include "gdscript_functions.h"
- class GDScriptTokenizer {
- public:
- enum Token {
- TK_EMPTY,
- TK_IDENTIFIER,
- TK_CONSTANT,
- TK_SELF,
- TK_BUILT_IN_TYPE,
- TK_BUILT_IN_FUNC,
- TK_OP_IN,
- TK_OP_EQUAL,
- TK_OP_NOT_EQUAL,
- TK_OP_LESS,
- TK_OP_LESS_EQUAL,
- TK_OP_GREATER,
- TK_OP_GREATER_EQUAL,
- TK_OP_AND,
- TK_OP_OR,
- TK_OP_NOT,
- TK_OP_ADD,
- TK_OP_SUB,
- TK_OP_MUL,
- TK_OP_DIV,
- TK_OP_MOD,
- TK_OP_SHIFT_LEFT,
- TK_OP_SHIFT_RIGHT,
- TK_OP_ASSIGN,
- TK_OP_ASSIGN_ADD,
- TK_OP_ASSIGN_SUB,
- TK_OP_ASSIGN_MUL,
- TK_OP_ASSIGN_DIV,
- TK_OP_ASSIGN_MOD,
- TK_OP_ASSIGN_SHIFT_LEFT,
- TK_OP_ASSIGN_SHIFT_RIGHT,
- TK_OP_ASSIGN_BIT_AND,
- TK_OP_ASSIGN_BIT_OR,
- TK_OP_ASSIGN_BIT_XOR,
- TK_OP_BIT_AND,
- TK_OP_BIT_OR,
- TK_OP_BIT_XOR,
- TK_OP_BIT_INVERT,
-
-
- TK_CF_IF,
- TK_CF_ELIF,
- TK_CF_ELSE,
- TK_CF_FOR,
- TK_CF_DO,
- TK_CF_WHILE,
- TK_CF_SWITCH,
- TK_CF_CASE,
- TK_CF_BREAK,
- TK_CF_CONTINUE,
- TK_CF_PASS,
- TK_CF_RETURN,
- TK_CF_MATCH,
- TK_PR_FUNCTION,
- TK_PR_CLASS,
- TK_PR_CLASS_NAME,
- TK_PR_EXTENDS,
- TK_PR_IS,
- TK_PR_ONREADY,
- TK_PR_TOOL,
- TK_PR_STATIC,
- TK_PR_EXPORT,
- TK_PR_SETGET,
- TK_PR_CONST,
- TK_PR_VAR,
- TK_PR_AS,
- TK_PR_VOID,
- TK_PR_ENUM,
- TK_PR_PRELOAD,
- TK_PR_ASSERT,
- TK_PR_YIELD,
- TK_PR_SIGNAL,
- TK_PR_BREAKPOINT,
- TK_PR_REMOTE,
- TK_PR_SYNC,
- TK_PR_MASTER,
- TK_PR_SLAVE,
- TK_PR_PUPPET,
- TK_PR_REMOTESYNC,
- TK_PR_MASTERSYNC,
- TK_PR_PUPPETSYNC,
- TK_BRACKET_OPEN,
- TK_BRACKET_CLOSE,
- TK_CURLY_BRACKET_OPEN,
- TK_CURLY_BRACKET_CLOSE,
- TK_PARENTHESIS_OPEN,
- TK_PARENTHESIS_CLOSE,
- TK_COMMA,
- TK_SEMICOLON,
- TK_PERIOD,
- TK_QUESTION_MARK,
- TK_COLON,
- TK_DOLLAR,
- TK_FORWARD_ARROW,
- TK_NEWLINE,
- TK_CONST_PI,
- TK_CONST_TAU,
- TK_WILDCARD,
- TK_CONST_INF,
- TK_CONST_NAN,
- TK_ERROR,
- TK_EOF,
- TK_CURSOR,
- TK_MAX
- };
- protected:
- enum StringMode {
- STRING_SINGLE_QUOTE,
- STRING_DOUBLE_QUOTE,
- STRING_MULTILINE
- };
- static const char *token_names[TK_MAX];
- public:
- static const char *get_token_name(Token p_token);
- bool is_token_literal(int p_offset = 0, bool variable_safe = false) const;
- StringName get_token_literal(int p_offset = 0) const;
- virtual const Variant &get_token_constant(int p_offset = 0) const = 0;
- virtual Token get_token(int p_offset = 0) const = 0;
- virtual StringName get_token_identifier(int p_offset = 0) const = 0;
- virtual GDScriptFunctions::Function get_token_built_in_func(int p_offset = 0) const = 0;
- virtual Variant::Type get_token_type(int p_offset = 0) const = 0;
- virtual int get_token_line(int p_offset = 0) const = 0;
- virtual int get_token_column(int p_offset = 0) const = 0;
- virtual int get_token_line_indent(int p_offset = 0) const = 0;
- virtual String get_token_error(int p_offset = 0) const = 0;
- virtual void advance(int p_amount = 1) = 0;
- #ifdef DEBUG_ENABLED
- virtual const Vector<Pair<int, String> > &get_warning_skips() const = 0;
- virtual const Set<String> &get_warning_global_skips() const = 0;
- virtual const bool is_ignoring_warnings() const = 0;
- #endif
- virtual ~GDScriptTokenizer(){};
- };
- class GDScriptTokenizerText : public GDScriptTokenizer {
- enum {
- MAX_LOOKAHEAD = 4,
- TK_RB_SIZE = MAX_LOOKAHEAD * 2 + 1
- };
- struct TokenData {
- Token type;
- StringName identifier;
- Variant constant;
- union {
- Variant::Type vtype;
- GDScriptFunctions::Function func;
- int warning_code;
- };
- int line, col;
- TokenData() {
- type = TK_EMPTY;
- line = col = 0;
- vtype = Variant::NIL;
- }
- };
- void _make_token(Token p_type);
- void _make_newline(int p_spaces = 0);
- void _make_identifier(const StringName &p_identifier);
- void _make_built_in_func(GDScriptFunctions::Function p_func);
- void _make_constant(const Variant &p_constant);
- void _make_type(const Variant::Type &p_type);
- void _make_error(const String &p_error);
- String code;
- int len;
- int code_pos;
- const CharType *_code;
- int line;
- int column;
- TokenData tk_rb[TK_RB_SIZE * 2 + 1];
- int tk_rb_pos;
- String last_error;
- bool error_flag;
- #ifdef DEBUG_ENABLED
- Vector<Pair<int, String> > warning_skips;
- Set<String> warning_global_skips;
- bool ignore_warnings;
- #endif
- void _advance();
- public:
- void set_code(const String &p_code);
- virtual Token get_token(int p_offset = 0) const;
- virtual StringName get_token_identifier(int p_offset = 0) const;
- virtual GDScriptFunctions::Function get_token_built_in_func(int p_offset = 0) const;
- virtual Variant::Type get_token_type(int p_offset = 0) const;
- virtual int get_token_line(int p_offset = 0) const;
- virtual int get_token_column(int p_offset = 0) const;
- virtual int get_token_line_indent(int p_offset = 0) const;
- virtual const Variant &get_token_constant(int p_offset = 0) const;
- virtual String get_token_error(int p_offset = 0) const;
- virtual void advance(int p_amount = 1);
- #ifdef DEBUG_ENABLED
- virtual const Vector<Pair<int, String> > &get_warning_skips() const { return warning_skips; }
- virtual const Set<String> &get_warning_global_skips() const { return warning_global_skips; }
- virtual const bool is_ignoring_warnings() const { return ignore_warnings; }
- #endif
- };
- class GDScriptTokenizerBuffer : public GDScriptTokenizer {
- enum {
- TOKEN_BYTE_MASK = 0x80,
- TOKEN_BITS = 8,
- TOKEN_MASK = (1 << TOKEN_BITS) - 1,
- TOKEN_LINE_BITS = 24,
- TOKEN_LINE_MASK = (1 << TOKEN_LINE_BITS) - 1,
- };
- Vector<StringName> identifiers;
- Vector<Variant> constants;
- VMap<uint32_t, uint32_t> lines;
- Vector<uint32_t> tokens;
- Variant nil;
- int token;
- public:
- Error set_code_buffer(const Vector<uint8_t> &p_buffer);
- static Vector<uint8_t> parse_code_string(const String &p_code);
- virtual Token get_token(int p_offset = 0) const;
- virtual StringName get_token_identifier(int p_offset = 0) const;
- virtual GDScriptFunctions::Function get_token_built_in_func(int p_offset = 0) const;
- virtual Variant::Type get_token_type(int p_offset = 0) const;
- virtual int get_token_line(int p_offset = 0) const;
- virtual int get_token_column(int p_offset = 0) const;
- virtual int get_token_line_indent(int p_offset = 0) const;
- virtual const Variant &get_token_constant(int p_offset = 0) const;
- virtual String get_token_error(int p_offset = 0) const;
- virtual void advance(int p_amount = 1);
- #ifdef DEBUG_ENABLED
- virtual const Vector<Pair<int, String> > &get_warning_skips() const {
- static Vector<Pair<int, String> > v;
- return v;
- }
- virtual const Set<String> &get_warning_global_skips() const {
- static Set<String> s;
- return s;
- }
- virtual const bool is_ignoring_warnings() const { return true; }
- #endif
- GDScriptTokenizerBuffer();
- };
- #endif
|