flex&bison_homework_1

1.1 Will the calculator accept a line that contains only a comment? Why not? Would it be easier to fix this in the scanner or in the parser?

The program before will not accept a line that contains only a comment. Add “calclist EOL” to “calclist”, (parser), is easier to fix this

// fbcal.y
calclist:
    | calclist exp EOL { printf("= %d\n", $2);}
    | calclist EOL { printf("= 0\n"); }
    ;

1.2 Make the calculator into a hex calculator that accepts both hex and decimal num- bers. In the scanner add a pattern such as 0x[a-f0-9]+ to match a hex number, and in the action code use strtol to convert the string to a number that you store in yylval; then return a NUMBER token. Adjust the output printf to print the result in both decimal and hex.

Using sscanf() to convert Hexadecimal number to decimal. and using %x to print Hexadecimal number

// fbcal.l
0x[A-Fa-f0-9]+ { sscanf(yytext,"%i",&yylval); return NUMBER; }
// fbcal.y
calclist exp EOL { printf("= %d or 0x%x\n", $2, $2);}

1.3 (extra credit) Add bit operators such as AND and OR to the calculator. The obvious operator to use for OR is a vertical bar, but that’s already the unary absolute value operator. What happens if you also use it as a binary OR operator, for example, exp ABS factor?

// fbcal.l
"&" { return AND; }
"|" { return OR; }
// fbcal.y
factor: ter { $$ = $1; }
    | factor MUL ter { $$ = $1 * $3; }
    | factor DIV ter { $$ = $1 / $3; }
    ;
ter:  term
    | ter AND term {$$ = $1 & $3; }
    | ter OR term { $$ = $1 | $3; }
    ;
term: NUMBER { $$ = $1; }
    | OR term { $$ = $2 >= 0 ? $2 : -$2; }
    | OP exp CP { $$ = $2; }
    ;

1.4 Does the handwritten version of the scanner from Example 1-4 recognize exactly the same tokens as the flex version?

This part is not exactly same, yylval might overflow

if (isdigit(c)) {
    int i = c - '0';
    while( isdigit(c = getc(yyin)))
        i = (10*i) + c-'0';
    yylval = i;
}

1.5 Can you think of languages for which flex wouldn’t be a good tool to write a scanner?

I searched on the internet. Flex is not a good tool to scan these languages that ambiguous, such as c++ and javascript

1.6 Rewrite the word count program in C. Run some large files through both versions. Is the C version noticeably faster? How much harder was it to debug?

#include <stdio.h>

int main(int argc, char **argv) {
    int chars = 0;
    int words = 0;
    int lines = 0;
    int inword = 0;
    char s[] = "The boy stood on the burning deck\n shelling peanuts by the peck\n";
    for( int i = 0; s[i] != '\0'; ++i) {
        chars++;
        if (('a'<=s[i] && s[i]<='z') || ('A'<=s[i] && s[i]<='Z')) {
            if (!inword) {
                inword = 1;
                words++;
            }
        } else {
            inword = 0;
        }
        if (s[i] == '\n') lines++;
    }
    printf("%8d%8d%8d\n", lines, words, chars);
    return 0;
}