diff options
Diffstat (limited to 'sys/fmtio/ctod.x')
| -rw-r--r-- | sys/fmtio/ctod.x | 154 |
1 files changed, 154 insertions, 0 deletions
diff --git a/sys/fmtio/ctod.x b/sys/fmtio/ctod.x new file mode 100644 index 00000000..06b1d351 --- /dev/null +++ b/sys/fmtio/ctod.x @@ -0,0 +1,154 @@ +# Copyright(c) 1986 Association of Universities for Research in Astronomy Inc. + +include <mach.h> +include <ctype.h> + +define DECIMAL 10 + +.help +.nf _________________________________________________________________________ +Attempt to convert a string to a number: nchar = ctod (str, ip, dval) +The index IP must be set to the first character to be scanned upon entry +to CTOD, and will be left pointing at the first untranslated character. + +If the string is successfully converted, the number of characters used +is returned as the function argument. If the string (or the first few +characters of the string) cannot be interpreted as a number, zero will be +returned. Note that even if no numeric characters are encountered, the +index IP may be incremented, if leading whitespace is encountered (but the +return value N will still be zero). + +The upper case string "INDEF" is a legal real number, as is "." (. == 0.0). +Sexagesimal numbers are permitted. Excess digits of precision are ignored. +Out of range exponents are detected, and result in the value INDEF being +returned (this is not considered an ERROR condition). Any number with an +exponent greater than or equal to MAX_EXPONENT is interpreted as INDEF, +regardless of the mantissa. The number need not contain a decimal point. + +Lexical form of a sexagesimal number: + + D :== [0-9] numeric digit + E :== [eEdD] exponent symbol + + ({D}*:)+{D}*(".")?{D}*({E}("+"|"-")?{D}+)? + +The format for sexagesimal numbers is fairly permissive. Any number of +colon fields are permitted, with any number of digits (including zero) in +each field. An exponent may occur at the end of a sexagesimal number. +Leading zeros may be omitted in the fields. +.endhelp ____________________________________________________________________ + + +# CTOD -- Convert a string to double precision real. + +int procedure ctod (str, ip, dval) + +char str[ARB] # string to be converted +int ip # pointer into str +double dval # receives binary value + +bool neg +char dig[MAX_DIGITS] +int j, e, vexp, ip_start +long expon +double value, scalar +int strncmp(), gctol(), stridx() + +begin + while (IS_WHITE (str[ip])) # skip whitespace + ip = ip + 1 + ip_start = ip + dval = INDEFD + + if (strncmp (str[ip], "INDEF", 5) == 0) { # check for "INDEF" + for (ip=ip+5; IS_ALPHA (str[ip]) || str[ip] == '_'; ip=ip+1) + ; + return (ip - ip_start) + } + + neg = (str[ip] == '-') # check for sign + if (neg || str[ip] == '+') + ip = ip + 1 + + while (str[ip] == '0') # ignore leading zeros + ip = ip + 1 + + dval = 0.0 + scalar = 60.0 + + repeat { # accumulate digits + for (j=1; j <= MAX_DIGITS && IS_DIGIT(str[ip]); j=j+1) { + dig[j] = str[ip] + ip = ip + 1 + } + + for (e=0; IS_DIGIT(str[ip]); e=e+1) # ignore the rest + ip = ip + 1 + + scalar = scalar / 60.0 + if (ip > 1 && stridx(str[ip], "'\":dDhHmMsS")>0) { # sexagesimal? + ip = ip + 1 + dig[j] = EOS + value = 0.0 # convert digits + for (j=1; dig[j] != EOS; j=j+1) + value = value * 10.0D0 + TO_INTEG (dig[j]) + dval = dval + value * scalar * (10.0 ** e) + + while (str[ip] != EOS && # multiple spaces etc + stridx(str[ip]," '\":dDhHmMsS")>0) + ip = ip + 1 + } else + break + } + + if (str[ip] == '.') { # check for a fraction + ip = ip + 1 + if (j == 1) # skip leading zeros + while (str[ip] == '0') { # if str = "0.00ddd" + ip = ip + 1 + e = e - 1 + } + for (; j <= MAX_DIGITS && IS_DIGIT(str[ip]); j=j+1) { + dig[j] = str[ip] + e = e - 1 # adjust scale factor + ip = ip + 1 + } # discard insignificant + while (IS_DIGIT (str[ip])) # fractional digits + ip = ip + 1 + } + + dig[j] = EOS # no more digits + vexp = e + j - 1 # save for ovfl check + if (ip == ip_start) # not a number? + return (0) + + value = 0.0 # convert the mantissa + for (j=1; dig[j] != EOS; j=j+1) + value = value * 10.0D0 + TO_INTEG (dig[j]) + if (e != 0) + value = value * (10.0D0 ** e) # scale by e + + + # Check for exponent. + + j = ip + expon = 0 + if (stridx (str[ip], "eEdD") > 0) { # exponent? + ip = ip + 1 + if (gctol (str, ip, expon, DECIMAL) <= 0) { + ip = j # return chars + expon = 0 + } + } + + if (abs(vexp+expon) > MAX_EXPONENTD) # check for overflow + return (ip - ip_start) + + dval = dval + value * scalar + if (expon != 0) + dval = dval * (10.0D0 ** expon) # apply exponent + + if (neg) + dval = -dval + return (ip - ip_start) +end |