Procedure

     NPARSD ( Double Precision parsing of a string )

     SUBROUTINE NPARSD ( STRING, X, ERROR, PTR )

Abstract

     Parse a character string that represents a number and return
     a double precision value.

Required_Reading

     None.

Keywords

     ALPHANUMERIC
     CONVERSION
     PARSING

Declarations

     IMPLICIT NONE

     CHARACTER*(*)    STRING
     DOUBLE PRECISION X
     CHARACTER*(*)    ERROR
     INTEGER          PTR

Brief_I/O

     VARIABLE  I/O  DESCRIPTION
     --------  ---  ---------------------------------------------------
     STRING     I   Character string representing a numeric value.
     X          O   Double precision value parsed from STRING.
     ERROR      O   Message indicating whether errors have occurred.
     PTR        O   Position in string where an error occurred.

Detailed_Input

     STRING   is a character string that represents a numeric value.
              Commas and spaces may be used in this string for
              ease of reading and writing the number. They
              are treated as insignificant but non-error-producing
              characters.

              For exponential representation the characters
              'E','D','e','d' may be used.

              The following are legitimate numeric expressions

               +12.2 e-1
               -3. 1415 9276
               1e12
               E10

              The program also recognizes the following  mnemonics
              'PI', 'pi', 'Pi', 'pI'
              '+PI', '+pi', '+Pi', '+pI'
              '-PI', '-pi', '-Pi', '-pI'
              and returns the value
              ( + OR - ) 3.1415 9265 3589 7932 3846 2600 D0 as
              appropriate.

Detailed_Output

     X        double precision parsed value of input string. If an
              error is encountered, X is not changed.

     ERROR    is a message indicating that the string could
              not be parsed due to use of an unexpected or misplaced
              character or due to a string representing a number
              too large for double precision. If the number was
              successfully parsed, ERROR will be returned as a blank.

              In particular, blank strings, or strings that do not
              contain either a digit or exponent character will
              be regarded as errors.

     PTR      this indicates which character was being used when
              the error occurred. If no error occurs, PTR is
              returned as 0.

Parameters

     None.

Exceptions

     Error free.

     1)  If the string is non-numeric, PTR indicates the location in
         the string where the error occurred, and ERROR contains a
         descriptive error message.

Files

     None.

Particulars

     This routine parses an input character string that represents a
     number, checks for overflow, unexpected or misplaced
     characters. It returns the double precision number or an error
     message.

Examples

     Let   LINE = 'DELTA_T_A       =   32.184'

     The following code fragment parses the line and obtains the
     double precision value.


        CALL NEXTWD ( LINE,  FIRST,  REST )
        CALL NEXTWD ( REST, SECOND,  REST )
        CALL NEXTWD ( REST,  THIRD,  REST )

        CALL NPARSD (  THIRD,  VALUE, ERROR, PTR    )

Restrictions

     1)  Due to rounding errors this routine may not be able to parse
         the decimal character string representation of the largest
         and smallest double precision numbers.

Literature_References

     None.

Author_and_Institution

     N.J. Bachman       (JPL)
     J. Diaz del Rio    (ODC Space)
     K.R. Gehringer     (JPL)
     H.A. Neilan        (JPL)
     W.L. Taber         (JPL)

Version

    SPICELIB Version 3.6.0, 12-AUG-2021 (JDR)

        Added IMPLICIT NONE statement.

        Edited the header to comply with NAIF standard. Removed
        unnecessary $Revisions section.

    SPICELIB Version 3.5.0, 15-AUG-2002 (WLT)

        Replaced the call to INSSUB with a call to ZZINSSUB so
        that this routine can legitimately call itself Error Free

    SPICELIB Version 3.4.0, 03-DEC-2001 (NJB)

        Added an extra check to make sure that ICHAR of any character
        of the input string is positive.

    SPICELIB Version 3.3.0, 29-FEB-1996 (KRG)

        The declaration for the SPICELIB function PI is now
        preceded by an EXTERNAL statement declaring PI to be an
        external function. This removes a conflict with any
        compilers that have a PI intrinsic function.

        Removed the error message and storage for the unexpected
        comma error message. This variable was set but never used,
        and according to the spec for this routine a comma is a valid
        delimiter, treated like a space, within numbers.

    SPICELIB Version 3.2.0, 10-JAN-1995 (WLT)

        Changed error strings from parameters to assignments to
        compensate for shortcomings of the Absoft FORTRAN compiler
        on the NeXT.

    SPICELIB Version 3.1.0, 12-JUL-1994 (WLT)

        The previous version of the routine assumed that the range
        of values of ICHAR was 0 to 128. That turns out not to be
        true on some machines. If a character whose ICHAR value is
        outside this range is detected, it is now handled properly
        as an unexpected character.

    SPICELIB Version 3.0.0, 24-FEB-1993 (WLT)

        The previous version of the algorithm interpreted P or p as 1.
        This was not the intent of the routine and was corrected.

    SPICELIB Version 2.0.0, 28-AUG-1992 (WLT) (KRG)

        The basic algorithm was completely re-written. As a result
        the routine now runs an order of magnitude faster than
        it did before. In addition, strings that do not contain
        enough information to assign a value to the string are now
        regarded as errors. These include blank strings or strings
        that contain only a sign characters, blanks and commas.

        In addition the error diagnosis and checking for overflow
        was greatly enhanced.

        Note: strings may now parse with slightly different values
        from the previous version of NPARSD. The current
        implementation is more accurate in converting strings to
        double precision numbers.

    SPICELIB Version 1.1.1, 10-MAR-1992 (WLT)

        Comment section for permuted index source lines was added
        following the header.

    SPICELIB Version 1.1.0, 17-APR-1990 (WLT)

        Bug fix. The subscript used to reference individual characters
        of the input string could sometimes step out of bounds. This
        went unnoticed until NAIF began compiling with the CHECK=BOUNDS
        option of the DEC Fortran compiler.


    SPICELIB Version 1.0.0, 31-JAN-1990 (WLT) (HAN) (NJB)