The C Programming Language, 2nd Edition, by Kernighan and Ritchie
Exercise 5.02 on page 97
Write getfloat , the floating-point analog of getint . What type does getfloat return as its function value?
Solutions by Chris Mears and Gregory Pietsch
Here is Chris's solution:
/* * Exercise 5-2 from The C Programming Language, 2nd edition, by Kernighan * and Ritchie. * * "Write getfloat, the floating-point analog of getint. What type does * getfloat return as its function value?" */ /* * Here's the getint function, from section 5.2: */ #include <ctype.h> #include <stdio.h> int getch(void); void ungetch(int); /* getint: get next integer from input into *pn */ int getint(int *pn) { int c, sign; while (isspace(c = getch())) /* skip white space */ ; if (!isdigit(c) && c != EOF && c != '+' && c != '-') { ungetch(c); /* it is not a number */ return 0; } sign = (c == '-') ? -1 : 1; if (c == '+' || c == '-') c = getch(); for (*pn = 0; isdigit(c); c = getch()) *pn = 10 * *pn + (c - '0'); *pn *= sign; if (c != EOF) ungetch(c); return c; } /* * The getch and ungetch functions, from section 4.3, are also required. */ #include <stdio.h> #define BUFSIZE 100 char buf[BUFSIZE]; /* buffer for ungetch */ int bufp = 0; /* next free position in buf */ int getch(void) /* get a (possibly pushed-back) character */ { return (bufp > 0) ? buf[--bufp] : getchar(); } void ungetch(int c) /* push character back on input */ { if (bufp >= BUFSIZE) printf("ungetch: too many characters\n"); else buf[bufp++] = c; } /* * The getfloat function. * * Reads the next number from input, and puts it into *fp. Returns EOF for * end of file, zero if the next input is not a number, and a positive * value of the input contains a valid number. * * Based heavily on the getint function from K&R2. */ #include <ctype.h> #include <math.h> int getfloat(float *fp) { int ch; int sign; int fraction; int digits; while (isspace(ch = getch())) /* skip white space */ ; if (!isdigit(ch) && ch != EOF && ch != '+' && ch != '-' && ch != '.') { ungetch(ch); return 0; } sign = (ch == '-') ? -1 : 1; if (ch == '+' || ch == '-') { ch = getch(); if (!isdigit(ch) && ch != '.') { if (ch == EOF) { return EOF; } else { ungetch(ch); return 0; } } } *fp = 0; fraction = 0; digits = 0; for ( ; isdigit(ch) || ch == '.' ; ch = getch()) { if (ch == '.') { fraction = 1; } else { if (!fraction) { *fp = 10 * *fp + (ch - '0'); } else { *fp = *fp + ((ch - '0') / pow(10, fraction)); fraction++; } digits++; } } *fp *= sign; if (ch == EOF) { return EOF; } else { ungetch(ch); return (digits) ? ch : 0; } } /* * Test module. */ #include <stdio.h> int main(void) { int ret; do { float f; fputs("Enter a number: ", stdout); fflush(stdout); ret = getfloat(&f); if (ret > 0) { printf("You entered: %f\n", f); } } while (ret > 0); if (ret == EOF) { puts("Stopped by EOF."); } else { puts("Stopped by bad input."); } return 0; }
...and here is Greg's solution:
/* Gregory Pietsch <gkp1@flash.net> Exercise 5-2 dated 2001-01-08 */ #include <ctype.h> #include <limits.h> /* also uses getch and ungetch from Section 4.3 */ /* number of significant digits in a double */ #define SIG_MAX 32 /* store double in d; return next character */ int getfloat(double *d) { const char point = '.'; /* localeconv->decimal_point[0]; */ int c; char buf[SIG_MAX], sign, sawsign, sawe, sawesign, esign; double x; static double fac[] = {0.0, 1.0e8, 1.0e16, 1.0e24, 1.0e32}; double dpow; int ndigit, nsig, nzero, olead, opoint, n; char *pc; long lo[SIG_MAX / 8 + 1], lexp; long *pl; /* skip white space */ while (isspace(c = getch())) ; if (sawsign = (c == '-' || c == '+')) { sign = c; c = getch(); } else sign = '+'; olead = -1; opoint = -1; ndigit = 0; nsig = 0; nzero = 0; while (c != EOF) { if (c == point) { if (0 <= opoint) break; /* already seen point */ else opoint = ndigit; } else if (c == '0') { /* saw a zero */ nzero++; ndigit++; } else if (!isdigit(c)) break; /* found nondigit */ else { /* got a nonzero digit */ if (olead < 0) olead = nzero; else { /* deliver zeros */ for ( ; 0 < nzero && nsig < SIG_MAX; --nzero) buf[nsig++] = 0; } ++ndigit; /* deliver digit */ if (nsig < SIG_MAX) buf[nsig++] = (c - '0'); } c = getch(); } if (ndigit == 0) { /* no digits? */ *d = 0.0; if (c != EOF) ungetch(c); if (0 <= opoint) { /* saw point */ ungetch(c = point); } if (sawsign) { /* saw sign */ ungetch(c = sign); } return c; } /* skip trailing digits */ for ( ; 0 < nsig && buf[nsig - 1] == 0; --nsig) ; /* compute significand */ pc = buf; pl = &(lo[nsig >> 3]); for (*pl = 0, n = nsig; 0 < n; --n) { if ((n & 7) == 0) /* start new sum */ *--pl = *pc++; else *pl = *pl * 10 + *pc++; } for (*d = (double)(lo[0]), n = 0; ++n <= (nsig >> 3); ) if (lo[n] != 0) *d += fac[n] * (double)(lo[n]); /* fold in any explicit exponent */ lexp = 0; if (c == 'e' || c == 'E') { /* we have an explicit exponent */ sawe = c; c = getch(); if (sawesign = (c == '+' || c == '-')) { esign = c; c = getch(); } else esign = '+'; if (!isdigit(c)) { /* ill-formed exponent */ if (c != EOF) ungetch(c); if (sawesign) ungetch(c = esign); c = sawe; } else { /* get exponent */ while (isdigit(c)) { /* get explicit exponent digits */ if (lexp < 100000) lexp = lexp * 10 + (c - '0'); /* else overflow */ c = getch(); } if (esign == '-') lexp = -lexp; } } if (c != EOF) ungetch(c); if (opoint < 0) lexp += ndigit - nsig; else lexp += opoint - olead - nsig; /* this is where I pray I don't lose precision */ esign = (lexp < 0) ? '-' : '+'; /* if anyone has a better way of handling overflow, tell Richard Heathfield */ if (lexp < SHRT_MIN) lexp = SHRT_MIN; if (lexp > SHRT_MAX) lexp = SHRT_MAX; if (lexp < 0) lexp = -lexp; if (lexp != 0) { dpow = (esign == '-') ? 0.1 : 10.0; while (lexp != 0) { /* form 10.0 to the lexp power */ if ((lexp & 1) != 0) /* lexp is positive */ *d *= dpow; lexp >>= 1; dpow *= dpow; } } /* if there was a minus sign in front, negate *d */ if (sign == '-') *d = -(*d); return c; }
