137 lines
3.3 KiB
C
137 lines
3.3 KiB
C
#include "../includes/scanner.h"
|
|
#include <stdio.h>
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
|
|
const char* SourceCode;
|
|
int Position = 0;
|
|
int SourceLength = 0;
|
|
int ScannerAtEnd(void);
|
|
char PeekScanner(void);
|
|
char AdvanceScanner(void);
|
|
Token* ParseString(void);
|
|
Token* ParseDirective(void);
|
|
Token* ParseIdentifier(void);
|
|
|
|
List* GenerateTokenList(const char* source) {
|
|
List* tokens = CreateList();
|
|
Token* token = NULL;
|
|
|
|
if (!tokens) return NULL;
|
|
|
|
SourceCode = source;
|
|
SourceLength = strlen(source);
|
|
|
|
for(int i = 0; i < SourceLength; i++) {
|
|
char c = AdvanceScanner();
|
|
switch(c) {
|
|
case ' ':
|
|
case '\r':
|
|
case '\t':
|
|
break; //Ignore whitespace
|
|
case ';':
|
|
while(!ScannerAtEnd() && PeekScanner() != '\n') AdvanceScanner();
|
|
|
|
break;
|
|
case '.': //directive like ".org" or ".db"
|
|
token = ParseDirective();
|
|
if (token) AddListItem(token, sizeof(Token), tokens);
|
|
break;
|
|
case '"':
|
|
token = ParseString();
|
|
if (token) AddListItem(token, sizeof(Token), tokens);
|
|
break;
|
|
case ',':
|
|
AddListItem(CreateToken(COMMA, ","), sizeof(Token), tokens);
|
|
break;
|
|
default:
|
|
// if (c >= '0' && c <= '9') {
|
|
// //Number
|
|
|
|
// break;
|
|
// }
|
|
|
|
token = ParseIdentifier();
|
|
|
|
if (token) AddListItem(token, sizeof(Token), tokens);
|
|
|
|
break;
|
|
}
|
|
|
|
token = NULL;
|
|
}
|
|
|
|
return tokens;
|
|
}
|
|
|
|
Token* ParseDirective(void) {
|
|
char directive[10] = { 0 };
|
|
int length = 0;
|
|
|
|
for(; length < sizeof directive - 1; length++) {
|
|
if (PeekScanner() == ' ') break;
|
|
|
|
directive[length] = AdvanceScanner();
|
|
}
|
|
|
|
if (strcmp(directive, ".db") == 0) {
|
|
return CreateToken(DB, ".db");
|
|
} else if (strcmp(directive, ".org") == 0) {
|
|
//fprintf(stderr, "[Warning] Org is not a supported directive.\n");
|
|
return NULL;
|
|
}
|
|
|
|
fprintf(stderr, "[Error] Unknown assembler directive '%s'\n", directive);
|
|
//Ignore the line
|
|
while(PeekScanner() != '\n') AdvanceScanner();
|
|
|
|
return NULL;
|
|
}
|
|
|
|
Token* ParseString(void) {
|
|
int start = Position;
|
|
|
|
while(!ScannerAtEnd() && PeekScanner() != '"') {
|
|
AdvanceScanner();
|
|
}
|
|
|
|
char* lexeme = calloc(Position - start, sizeof(char));
|
|
memccpy(lexeme, &SourceCode[start], sizeof(char), Position - start);
|
|
|
|
AdvanceScanner();
|
|
|
|
Token* token = CreateToken(STRING, lexeme);
|
|
|
|
return token;
|
|
}
|
|
|
|
Token* ParseIdentifier(void) {
|
|
int start = Position;
|
|
|
|
while(!ScannerAtEnd() && PeekScanner() != ' ' && PeekScanner() != '\n') {
|
|
AdvanceScanner();
|
|
}
|
|
|
|
if (Position - start == 0) return NULL;
|
|
|
|
char* lexeme = calloc(Position - start, sizeof(char));
|
|
memccpy(lexeme, &SourceCode[start], sizeof(char), Position - start);
|
|
|
|
return CreateToken(IDENTIFIER, lexeme);
|
|
}
|
|
|
|
char PeekScanner(void) {
|
|
if (Position > SourceLength) return '\0';
|
|
|
|
return SourceCode[Position];
|
|
}
|
|
|
|
char AdvanceScanner(void) {
|
|
if (ScannerAtEnd()) return '\0';
|
|
|
|
return SourceCode[Position++];
|
|
}
|
|
|
|
int ScannerAtEnd(void) {
|
|
return Position >= SourceLength;
|
|
} |