Skip to content

Commit 8932e2b

Browse files
tmatostmatos
authored
Merge pull request #15 from tmatos/multichar-operator-scanning
Multi-char operator scanning
2 parents 40925e6 + 25248e7 commit 8932e2b

6 files changed

Lines changed: 482 additions & 30 deletions

File tree

‎inout.c‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -55,7 +55,7 @@ Sinais* load_input_signals(FILE* file)
5555
// loop to read the set of all signal in the file
5656
while (1)
5757
{
58-
if ( is_symbol(it->valor[0]) )
58+
if ( is_single_char_symbol(it->valor[0]) )
5959
return show_fatal_error_corrupt_file();
6060

6161
// TODO: check signal name duplication errors

‎lex.c‎

Lines changed: 227 additions & 18 deletions
Original file line numberDiff line numberDiff line change
@@ -18,7 +18,7 @@
1818

1919
#define NUM_RESERV_KEYWORDS 108
2020

21-
char arrayPalavrasReservadas[][NUM_RESERV_KEYWORDS] = {
21+
char array_reserverd_keywords[][NUM_RESERV_KEYWORDS] = {
2222
"always",
2323
"and",
2424
"assign",
@@ -177,15 +177,15 @@ TokenClass get_token_class(const char* s_tok)
177177
int i;
178178

179179
// IMPORTANT: keep track of the count here!
180-
#define _QTD_CLASSES 48
180+
#define _QTD_CLASSES 71
181181
#define _MAX_STRLEN_IN_ARRAY 13
182182

183183
char from_str[_QTD_CLASSES][_MAX_STRLEN_IN_ARRAY] = {
184-
"and", // 1
185-
"or", // 2
186-
"not", // 3
187-
"buf",
188-
"nand",
184+
"and", // 1
185+
"or", // 2
186+
"not", // 3
187+
"buf", // 4
188+
"nand", // 5
189189
"nor",
190190
"xor",
191191
"xnor",
@@ -228,6 +228,29 @@ TokenClass get_token_class(const char* s_tok)
228228
"|", // = 45
229229
"^",
230230
"$",
231+
"**",
232+
">",
233+
">=", // = 50
234+
"<",
235+
"<=",
236+
"&&",
237+
"||",
238+
"==", // = 55
239+
"===",
240+
"!",
241+
"!=",
242+
"!==",
243+
"^~", // = 60
244+
"~^",
245+
"~&",
246+
"~|",
247+
"<<",
248+
"<<<", // = 65
249+
">>",
250+
">>>",
251+
"?:",
252+
"{{",
253+
"}}",
231254
"\x60" // = _QTD_CLASSES
232255
};
233256

@@ -271,14 +294,38 @@ TokenClass get_token_class(const char* s_tok)
271294
SYM_HASHTAG,
272295
SYM_PLUS,
273296
SYM_MINUS,
274-
SYM_ASTERISK, // = 40
297+
SYM_ASTERISK, // = 40
275298
SYM_SLASH,
276299
SYM_PERCENT,
277300
SYM_TILDE,
278301
SYM_AMPERSAND,
279-
SYM_PIPE, // = 45
302+
SYM_PIPE, // = 45
280303
SYM_CIRCUMFLEX,
281304
SYM_DOLLAR,
305+
SYM_DOUBLE_ASTERISK,
306+
SYM_GREATER_THAN,
307+
SYM_GREATER_OR_EQUAL, // = 50
308+
SYM_LESS_THAN,
309+
SYM_LESS_OR_EQUAL,
310+
SYM_DOUBLE_AMPERSAND,
311+
SYM_DOUBLE_PIPE,
312+
SYM_DOUBLE_EQ, // = 55
313+
SYM_TRIPLE_EQ,
314+
SYM_EXCLAMATION,
315+
SYM_EXCLAMATION_EQ,
316+
SYM_EXCLAMATION_DOUBLE_EQ,
317+
SYM_CIRCUMFLEX_TILDE, // = 60
318+
SYM_TILDE_CIRCUMFLEX,
319+
SYM_TILDE_AMPERSAND,
320+
SYM_TILDE_PIPE,
321+
SYM_DOUBLE_LESS_THAN,
322+
SYM_TRIPLE_LESS_THAN, // = 65
323+
SYM_DOUBLE_GREATER_THAN,
324+
SYM_TRIPLE_GREATER_THAN,
325+
SYM_QUESTION_COLON,
326+
SYM_DOUBLE_OPEN_BRACE,
327+
SYM_DOUBLE_CLOSE_BRACE, // = 70
328+
282329
SYM_GRAVE_ACCENT // = _QTD_CLASSES
283330
};
284331

@@ -429,7 +476,7 @@ int remove_tokens_by_value(ListToken* lst, const char* tok)
429476
return 1;
430477
}
431478

432-
int is_symbol(char c)
479+
int is_single_char_symbol(char c)
433480
{
434481
return (c == '(' ||
435482
c == ')' ||
@@ -456,8 +503,10 @@ int is_symbol(char c)
456503
c == '@' ||
457504
c == '$' ||
458505
c == '`' || // grave accent
459-
c == '"' ||
460-
c == '\'');
506+
c == '"' || // double quote
507+
c == '\'' || // single quote
508+
c == '\%' || // percent sign
509+
c == '^'); // circumflex
461510
}
462511

463512
void show_token_list(ListToken* tokens)
@@ -533,7 +582,7 @@ int is_reserverd_word(Token* tk)
533582

534583
for (i = 0; i < NUM_RESERV_KEYWORDS; ++i)
535584
{
536-
if ( iguais(tk->valor, arrayPalavrasReservadas[i]) )
585+
if ( iguais(tk->valor, array_reserverd_keywords[i]) )
537586
return 1;
538587
}
539588

@@ -710,12 +759,172 @@ ListToken* tokeniza(FILE* arquivo)
710759

711760
// B: a parte B do automato (na folha de papel)
712761

713-
if (is_symbol(c)) {
714-
// TODO: capture symbols larger than 1 char
762+
if ( is_single_char_symbol(c) ) {
715763
symbols_capture_state:
716764
coluna++;
717-
insert_token_of_char(tokens, c, linha, coluna);
718-
goto start_state;
765+
char previous_c = c;
766+
767+
switch (c)
768+
{
769+
case '*':
770+
c = fgetc(arquivo);
771+
if (c == '*') {
772+
insert_token_of_string(tokens, "**", linha, coluna);
773+
coluna += 2;
774+
goto start_state;
775+
}
776+
break;
777+
case '<':
778+
c = fgetc(arquivo);
779+
if (c == '=') {
780+
insert_token_of_string(tokens, "<=", linha, coluna);
781+
coluna += 2;
782+
goto start_state;
783+
}
784+
else if (c == '<') {
785+
c = fgetc(arquivo);
786+
if (c == '<') {
787+
insert_token_of_string(tokens, "<<<", linha, coluna);
788+
coluna += 3;
789+
goto start_state;
790+
}
791+
else {
792+
ungetc(c, arquivo);
793+
}
794+
insert_token_of_string(tokens, "<<", linha, coluna);
795+
coluna += 2;
796+
goto start_state;
797+
}
798+
break;
799+
case '>':
800+
c = fgetc(arquivo);
801+
if (c == '=') {
802+
insert_token_of_string(tokens, ">=", linha, coluna);
803+
coluna += 2;
804+
goto start_state;
805+
}
806+
else if (c == '>') {
807+
c = fgetc(arquivo);
808+
if (c == '>') {
809+
insert_token_of_string(tokens, ">>>", linha, coluna);
810+
coluna += 3;
811+
goto start_state;
812+
}
813+
else {
814+
ungetc(c, arquivo);
815+
}
816+
insert_token_of_string(tokens, ">>", linha, coluna);
817+
coluna += 2;
818+
goto start_state;
819+
}
820+
break;
821+
case '&':
822+
c = fgetc(arquivo);
823+
if (c == '&') {
824+
insert_token_of_string(tokens, "&&", linha, coluna);
825+
coluna += 2;
826+
goto start_state;
827+
}
828+
break;
829+
case '|':
830+
c = fgetc(arquivo);
831+
if (c == '|') {
832+
insert_token_of_string(tokens, "||", linha, coluna);
833+
coluna += 2;
834+
goto start_state;
835+
}
836+
break;
837+
case '=':
838+
c = fgetc(arquivo);
839+
if (c == '=') {
840+
c = fgetc(arquivo);
841+
if (c == '=') {
842+
insert_token_of_string(tokens, "===", linha, coluna);
843+
coluna += 3;
844+
goto start_state;
845+
}
846+
else {
847+
ungetc(c, arquivo);
848+
}
849+
insert_token_of_string(tokens, "==", linha, coluna);
850+
coluna += 2;
851+
goto start_state;
852+
}
853+
break;
854+
case '!':
855+
c = fgetc(arquivo);
856+
if (c == '=') {
857+
c = fgetc(arquivo);
858+
if (c == '=') {
859+
insert_token_of_string(tokens, "!==", linha, coluna);
860+
coluna += 3;
861+
goto start_state;
862+
}
863+
else {
864+
ungetc(c, arquivo);
865+
}
866+
insert_token_of_string(tokens, "!=", linha, coluna);
867+
coluna += 2;
868+
goto start_state;
869+
}
870+
break;
871+
case '^':
872+
c = fgetc(arquivo);
873+
if (c == '~') {
874+
insert_token_of_string(tokens, "^~", linha, coluna);
875+
coluna += 2;
876+
goto start_state;
877+
}
878+
break;
879+
case '~':
880+
c = fgetc(arquivo);
881+
if (c == '^') {
882+
insert_token_of_string(tokens, "~^", linha, coluna);
883+
coluna += 2;
884+
goto start_state;
885+
}
886+
else if (c == '&') {
887+
insert_token_of_string(tokens, "~&", linha, coluna);
888+
coluna += 2;
889+
goto start_state;
890+
}
891+
else if (c == '|') {
892+
insert_token_of_string(tokens, "~|", linha, coluna);
893+
coluna += 2;
894+
goto start_state;
895+
}
896+
break;
897+
case '?':
898+
c = fgetc(arquivo);
899+
if (c == ':') {
900+
insert_token_of_string(tokens, "?:", linha, coluna);
901+
coluna += 2;
902+
goto start_state;
903+
}
904+
break;
905+
case '{':
906+
c = fgetc(arquivo);
907+
if (c == '{') {
908+
insert_token_of_string(tokens, "{{", linha, coluna);
909+
coluna += 2;
910+
goto start_state;
911+
}
912+
break;
913+
case '}':
914+
c = fgetc(arquivo);
915+
if (c == '}') {
916+
insert_token_of_string(tokens, "}}", linha, coluna);
917+
coluna += 2;
918+
goto start_state;
919+
}
920+
break;
921+
default:
922+
insert_token_of_char(tokens, c, linha, coluna);
923+
goto start_state;
924+
}
925+
926+
insert_token_of_char(tokens, previous_c, linha, coluna);
927+
goto start_after_getc_state;
719928
}
720929

721930
if ( !isalnum(c) && c != '_' ) {
@@ -746,7 +955,7 @@ ListToken* tokeniza(FILE* arquivo)
746955
insert_token_of_string(tokens, tok, linha, coluna - len(tok));
747956
goto comments_state;
748957
}
749-
else if (is_symbol(c)) {
958+
else if (is_single_char_symbol(c)) {
750959
insert_token_of_string(tokens, tok, linha, coluna - len(tok));
751960
goto symbols_capture_state;
752961
}

‎lex.h‎

Lines changed: 34 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -24,7 +24,7 @@ extern "C" {
2424
/** @brief Enumeracao para a classe do token detectado.
2525
*/
2626
typedef enum en_token_class {
27-
KW_ALWAYS,
27+
KW_ALWAYS = 0,
2828
KW_AND,
2929
KW_ASSIGN,
3030
KW_BEGIN,
@@ -70,6 +70,8 @@ typedef enum en_token_class {
7070
SYM_CLOSE_SQUAREBRACKET, // ]
7171
SYM_OPEN_BRACE, // {
7272
SYM_CLOSE_BRACE, // }
73+
SYM_DOUBLE_OPEN_BRACE, // {{
74+
SYM_DOUBLE_CLOSE_BRACE, // }}
7375
SYM_PLUS, // +
7476
SYM_MINUS, // -
7577
SYM_ASTERISK, // *
@@ -81,6 +83,27 @@ typedef enum en_token_class {
8183
SYM_CIRCUMFLEX, // ^
8284
SYM_DOLLAR, // $
8385
SYM_GRAVE_ACCENT, // `
86+
SYM_DOUBLE_ASTERISK, // **
87+
SYM_GREATER_THAN, // >
88+
SYM_GREATER_OR_EQUAL, // >=
89+
SYM_LESS_THAN, // <
90+
SYM_LESS_OR_EQUAL, // <=
91+
SYM_DOUBLE_AMPERSAND, // &&
92+
SYM_DOUBLE_PIPE, // ||
93+
SYM_DOUBLE_EQ, // ==
94+
SYM_TRIPLE_EQ, // ===
95+
SYM_EXCLAMATION, // !
96+
SYM_EXCLAMATION_EQ, // !=
97+
SYM_EXCLAMATION_DOUBLE_EQ, // !==
98+
SYM_CIRCUMFLEX_TILDE, // ^~
99+
SYM_TILDE_CIRCUMFLEX, // ~^
100+
SYM_TILDE_AMPERSAND, // ~&
101+
SYM_TILDE_PIPE, // ~|
102+
SYM_DOUBLE_LESS_THAN, // <<
103+
SYM_TRIPLE_LESS_THAN, // <<<
104+
SYM_DOUBLE_GREATER_THAN, // >>
105+
SYM_TRIPLE_GREATER_THAN, // >>>
106+
SYM_QUESTION_COLON, // ?:
84107

85108
NUM_BASE_BINARY,
86109
NUM_BASE_DECIMAL,
@@ -92,6 +115,15 @@ typedef enum en_token_class {
92115
_UNKNOWN
93116
} TokenClass;
94117

118+
/** @brief Struct for pairs that represents the mapping between
119+
* token strings and their respective TokenClass.
120+
*/
121+
typedef struct st_pair_string_tokenclass
122+
{
123+
char token_string[MAX_TOKEN_SIZE];
124+
TokenClass token_class;
125+
} PairStringTokenClass;
126+
95127
/** @brief Tipo basico para o elemento que representa um token.
96128
*/
97129
typedef struct st_token {
@@ -158,7 +190,7 @@ int remove_tokens_by_value(ListToken* lst, const char* tok);
158190
* @param c Um char qualquer.
159191
* @return True se c for simbolo, False caso contrario.
160192
*/
161-
int is_symbol(char c);
193+
int is_single_char_symbol(char c);
162194

163195
/** @brief Imprime na tela os tokens dessa lista, um por linha.
164196
* @return Void.

0 commit comments

Comments
 (0)