Updated the scanner to parse registers, assembly keywords and identifiers.

This commit is contained in:
2022-05-04 20:07:47 +00:00
parent 2c9478d311
commit 859035da2a
5 changed files with 83 additions and 89 deletions
+75 -27
View File
@@ -1,9 +1,11 @@
#include "../includes/scanner.h"
#include <ctype.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
const char* SourceCode;
int Line = 0;
int Position = 0;
int SourceLength = 0;
int ScannerAtEnd(void);
@@ -12,6 +14,8 @@ char AdvanceScanner(void);
Token* ParseString(void);
Token* ParseDirective(void);
Token* ParseIdentifier(void);
Token* ParseNumber(void);
void IgnoreLine(void);
List* GenerateTokenList(const char* source) {
List* tokens = CreateList();
@@ -23,15 +27,20 @@ List* GenerateTokenList(const char* source) {
SourceLength = strlen(source);
for(int i = 0; i < SourceLength; i++) {
char c = AdvanceScanner();
char c = PeekScanner();
switch(c) {
case ' ':
case '\r':
case '\t':
AdvanceScanner();
break; //Ignore whitespace
case '\n':
AdvanceScanner();
Line++;
break;
case ';':
while(!ScannerAtEnd() && PeekScanner() != '\n') AdvanceScanner();
IgnoreLine();
break;
case '.': //directive like ".org" or ".db"
token = ParseDirective();
@@ -42,15 +51,15 @@ List* GenerateTokenList(const char* source) {
if (token) AddListItem(token, sizeof(Token), tokens);
break;
case ',':
AdvanceScanner();
AddListItem(CreateToken(COMMA, ","), sizeof(Token), tokens);
break;
default:
// if (c >= '0' && c <= '9') {
// //Number
// break;
// }
if (c >= '0' && c <= '9') {
ParseNumber();
break;
}
token = ParseIdentifier();
if (token) AddListItem(token, sizeof(Token), tokens);
@@ -64,31 +73,66 @@ List* GenerateTokenList(const char* source) {
return tokens;
}
Token* ParseDirective(void) {
char directive[10] = { 0 };
int length = 0;
Token* ParseNumber(void) {
int start = Position;
for(; length < sizeof directive - 1; length++) {
if (PeekScanner() == ' ') break;
while(!ScannerAtEnd() && PeekScanner() >= '0' && PeekScanner() <= '9')
AdvanceScanner();
directive[length] = AdvanceScanner();
}
char* lexeme = calloc(Position - start + 1, sizeof(char));
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");
if (!lexeme) {
fprintf(stderr, "Failed to calloc space for number.\n");
return NULL;
}
fprintf(stderr, "[Error] Unknown assembler directive '%s'\n", directive);
//Ignore the line
while(PeekScanner() != '\n') AdvanceScanner();
memcpy(lexeme, &SourceCode[start], Position - start);
return CreateToken(NUMBER, lexeme);
}
Token* ParseDirective(void) {
int start = Position;
while(!ScannerAtEnd() && PeekScanner() != ' ' && PeekScanner() != '\n') AdvanceScanner();
if (start == Position) return NULL;
char* directive = calloc(Position - start + 1, sizeof(char));
if (!directive) {
fprintf(stderr, "Failed to calloc for the assmebler directive.\n");
return NULL;
}
memcpy(directive, &SourceCode[start], Position - start);
if (strcmp(directive, ".db") == 0) {
return CreateToken(DB, directive);
}
// else if (strcmp(directive, ".org") == 0) {
// fprintf(stderr, "[Warning] Org is not a supported directive.\n");
// IgnoreLine();
// return NULL;
// }
fprintf(stderr, "[Error] Unknown assembler directive '%s'\n", directive);
free(directive);
//Ignore the line
IgnoreLine();
return NULL;
}
void IgnoreLine(void) {
while(!ScannerAtEnd() && PeekScanner() != '\n') AdvanceScanner();
}
Token* ParseString(void) {
AdvanceScanner(); //Consume the first double quote.
int start = Position;
while(!ScannerAtEnd() && PeekScanner() != '"') {
@@ -96,9 +140,9 @@ Token* ParseString(void) {
}
char* lexeme = calloc(Position - start, sizeof(char));
memccpy(lexeme, &SourceCode[start], sizeof(char), Position - start);
memcpy(lexeme, &SourceCode[start], Position - start);
AdvanceScanner();
AdvanceScanner(); //Consume the trailing double quote.
Token* token = CreateToken(STRING, lexeme);
@@ -108,20 +152,24 @@ Token* ParseString(void) {
Token* ParseIdentifier(void) {
int start = Position;
while(!ScannerAtEnd() && PeekScanner() != ' ' && PeekScanner() != '\n') {
while(!ScannerAtEnd() && PeekScanner() != ',' && 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);
TokenType type;
memcpy(lexeme, &SourceCode[start], Position - start);
if (IsOpcode(lexeme, &type)) return CreateToken(type, lexeme);
if (IsRegister(lexeme, &type)) return CreateToken(type, lexeme);
return CreateToken(IDENTIFIER, lexeme);
}
char PeekScanner(void) {
if (Position > SourceLength) return '\0';
if (Position >= SourceLength) return '\0';
return SourceCode[Position];
}