OS/2 Shareware BBS: 10 Tools/10-Tools.zip/numana01.zip/SRC/CONVERSI.MOD

Text File | 1996-07-31 | 31KB | 828 lines

IMPLEMENTATION MODULE Conversions;

        (********************************************************)
        (*                                                      *)
        (*              Miscellaneous type conversions          *)
        (*                                                      *)
        (*  Programmer:         P. Moylan                       *)
        (*  Last edited:        30 July 1996                    *)
        (*  Status:             Working                         *)
        (*                                                      *)
        (*      Have not fully tested the cases where the field *)
        (*       size is too small.                             *)
        (*      Seems to be a loss of accuracy when converting  *)
        (*       E-format numbers; for example 123.456E7 is     *)
        (*       converted to 1234559999 when passed through    *)
        (*       StringToReal and then RealToString.  I'm not   *)
        (*       yet sure where this is happening, but suspect  *)
        (*       that it requires some deep error analysis.     *)
        (*                                                      *)
        (********************************************************)

FROM SYSTEM IMPORT
    (* type *)  CARD8;

FROM LowLevel IMPORT
    (* proc *)  IAND, RS;

CONST tab = CHR(9);

TYPE CharSet = SET OF CHAR;

(************************************************************************)
(*                         BUFFER MANIPULATION                          *)
(************************************************************************)

PROCEDURE ShiftRight (VAR (*INOUT*) buffer: ARRAY OF CHAR;
                                first, last, amount: CARDINAL);

    (* Moves the contents of buffer[first..last] right by the specified *)
    (* number of characters, space filling at the left and discarding   *)
    (* characters shifted out at the right.                             *)

    VAR j: CARDINAL;

    BEGIN
        IF amount > 0 THEN
            FOR j := last TO first+amount BY -1 DO
                buffer[j] := buffer[j-amount];
            END (*FOR*);
            FOR j := first TO first+amount-1 DO
                buffer[j] := " ";
            END (*FOR*);
        END (*IF*);
    END ShiftRight;

(************************************************************************)
(*                      REAL NUMBER TO CARDINAL POWER                   *)
(************************************************************************)

PROCEDURE atoi (a: LONGREAL;  i: CARDINAL): LONGREAL;

    (* Calculates a**i.  This procedure does not really belong in this  *)
    (* module, but for now it doesn't seem to have any other suitable   *)
    (* home.                                                            *)

    VAR result: LONGREAL;

    BEGIN
        result := 1.0;

        (* Desired answer is result*(a)**i.  The loop below keeps this  *)
        (* quantity invariant while reducing i down to zero.            *)

        LOOP
            IF ODD(i) THEN
                DEC(i);  result := a*result;
            END (*IF*);
            IF i=0 THEN EXIT(*LOOP*) END(*IF*);
            i := i DIV 2;  a := a*a;
        END (*LOOP*);
        RETURN result;
    END atoi;

(************************************************************************)

PROCEDURE TenToPower (N: CARDINAL): LONGREAL;

    (* Calculates 10**N.        *)

    BEGIN
        RETURN atoi (10.0, N);
    END TenToPower;

(************************************************************************)
(*                      CARDINAL-TO-STRING CONVERSIONS                  *)
(************************************************************************)

PROCEDURE HexToChar (number: HexDigit): CHAR;

    (* Converts a one-digit hexadecimal number to its readable form.    *)

    BEGIN
        IF number < 10 THEN
            RETURN CHR(ORD("0")+number)
        ELSE
            RETURN CHR(ORD("A")+number-10)
        END (*IF*);
    END HexToChar;

(************************************************************************)

PROCEDURE HexByteToString (value: CARD8;
                        VAR (*OUT*) buffer: ARRAY OF CHAR;  pos: CARDINAL);

    (* Converts a byte value to 2-character hexadecimal, with the       *)
    (* result stored at buffer[pos] and buffer[pos+1].                  *)

    BEGIN
        buffer[pos] := HexToChar (VAL(HexDigit, value DIV 16));
        buffer[pos+1] := HexToChar (VAL(HexDigit, value MOD 16));
    END HexByteToString;

(************************************************************************)

PROCEDURE HexToString (value: CARDINAL;  VAR (*OUT*) buffer: ARRAY OF CHAR);

    VAR j: CARDINAL;

    BEGIN
        FOR j := HIGH(buffer) TO 0 BY -1 DO
            buffer[j] := HexToChar (IAND(value,0FH));
            value := RS (value, 4);
        END (*FOR*);
    END HexToString;

(************************************************************************)

(*
PROCEDURE LongHexToString (value: LONGCARD;  VAR (*OUT*) buffer: EightChar);

    VAR j: [0..3];  highpart: ARRAY [0..3] OF CHAR;

    BEGIN
        HexToString (LowWord(value), buffer);
        HexToString (HighWord(value), highpart);
        FOR j := 0 TO 3 DO
            buffer[j] := highpart[j];
        END (*FOR*);
    END LongHexToString;
*)

(************************************************************************)

(*
PROCEDURE LongCardToString (number: LONGCARD;
                                        VAR (*OUT*) buffer: ARRAY OF CHAR;
                                        fieldsize: CARDINAL);

    (* Converts the number to a decimal character string in array       *)
    (* "buffer", right-justified in a field of fieldsize characters.    *)

    VAR j, remainder: CARDINAL;

    BEGIN
        IF number < 10 THEN
            IF fieldsize > 1 THEN
                FOR j := 0 TO fieldsize-2 DO
                    buffer[j] := " ";
                END (*FOR*);
            END (*IF*);
            buffer[fieldsize-1] := CHR(ORD(number) + ORD("0"));
        ELSIF fieldsize = 1 THEN
            buffer[0] := "*";
        ELSE
            LongCardToString (number DIV 10, buffer, fieldsize-1);
            remainder := CARDINAL (number MOD 10);
            buffer[fieldsize-1] := CHR(remainder + ORD("0"));
        END (*IF*);
    END LongCardToString;
*)

(*********************************************************************)

PROCEDURE CardinalToString (number: CARDINAL;
                                        VAR (*OUT*) buffer: ARRAY OF CHAR;
                                        fieldsize: CARDINAL);

    (* Converts the number to a decimal character string in array       *)
    (* "buffer", right-justified in a field of fieldsize characters.    *)

    VAR j, remainder: CARDINAL;

    BEGIN
        IF number < 10 THEN
            IF fieldsize > 1 THEN
                FOR j := 0 TO fieldsize-2 DO
                    buffer[j] := " ";
                END (*FOR*);
            END (*IF*);
            buffer[fieldsize-1] := CHR(ORD(number) + ORD("0"));
        ELSIF fieldsize = 1 THEN
            buffer[0] := "*";
        ELSE
            CardinalToString (number DIV 10, buffer, fieldsize-1);
            remainder := number MOD 10;
            buffer[fieldsize-1] := CHR(remainder + ORD("0"));
        END (*IF*);
    END CardinalToString;

(*********************************************************************)

PROCEDURE ShortCardToString (number: CARD8;
                                        VAR (*OUT*) buffer: ARRAY OF CHAR;
                                        fieldsize: CARDINAL);

    (* Converts the number to a decimal character string in array       *)
    (* "buffer", right-justified in a field of fieldsize characters.    *)

    BEGIN
        CardinalToString (VAL(CARDINAL,number), buffer, fieldsize);
    END ShortCardToString;

(*********************************************************************)

PROCEDURE AssembleCardinal (number: CARDINAL;
                                VAR (*OUT*) buffer: ARRAY OF CHAR;
                                VAR (*INOUT*) place: CARDINAL;
                                VAR (*OUT*) error: BOOLEAN);

    (* Converts number to decimal, putting it in buffer starting at     *)
    (* buffer[place].  On return, place has been updated to be just     *)
    (* beyond the last digit put in the buffer.                         *)

    BEGIN
        IF number > 9 THEN
            AssembleCardinal (number DIV 10, buffer, place, error);
            IF error THEN RETURN END(*IF*);
        END (*IF*);
        error := place > HIGH(buffer);
        IF NOT error THEN
            buffer[place] := CHR (number MOD 10 + ORD("0"));
            INC (place);
        END (*IF*);
    END AssembleCardinal;

(************************************************************************)
(*                      REAL-TO-STRING CONVERSIONS                      *)
(************************************************************************)

PROCEDURE AssembleExponent (number: INTEGER;
                        VAR (*OUT*) buffer: ARRAY OF CHAR;
                        VAR (*INOUT*) position: CARDINAL;
                        VAR (*OUT*) error: BOOLEAN);

    (* Puts a field of the format Ennn or E-nnn into the buffer,        *)
    (* starting at buffer[position].  On return, position has been      *)
    (* updated so that buffer[position] is the first character not      *)
    (* altered by this procedure.                                       *)

    BEGIN
        error := FALSE;
        IF number <> 0 THEN
            error := position > HIGH(buffer);
            IF NOT error THEN
                buffer[position] := "E";  INC(position);
                IF number < 0 THEN
                    error := position > HIGH(buffer);
                    IF NOT error THEN
                        buffer[position] := "-";  INC(position);
                        number := -number;
                    END (*IF*);
                END (*IF*);
            END (*IF*);
            IF NOT error THEN
                AssembleCardinal (number, buffer, position, error);
            END (*IF*);
        END (*IF*);
    END AssembleExponent;

(************************************************************************)

PROCEDURE Roundup (VAR (*INOUT*) buffer: ARRAY OF CHAR;
                                        first, last: CARDINAL);

    (* Takes the decimal number in buffer[first..last] and increments   *)
    (* its least significant digit, propagating the carry upwards as    *)
    (* far as necessary.                                                *)

    VAR position, pointposition: CARDINAL;
        code: CHAR;

    BEGIN
        position := last+1;  pointposition := position;
        REPEAT
            DEC (position);
            code := buffer[position];
            IF code = "9" THEN buffer[position] := "0"
            ELSIF code = "." THEN
                pointposition := position;  code := "9";
            ELSE
                INC (buffer[position]);
            END (*IF*);
        UNTIL (code <> "9") OR (position = first);

        (* The job is now done, except for one special case.  If we     *)
        (* have left the above loop after incrementing a "9", the carry *)
        (* has propagated off the left end of the number.  In that case *)
        (* every digit must have been a "9", so the result is 10000...  *)
        (* with a decimal point inserted at the appropriate place.      *)

        IF code = "9" THEN
            IF pointposition <= last THEN
                buffer[pointposition] := "0";
                IF pointposition < last THEN
                    INC (pointposition);  buffer[pointposition] := ".";
                END (*IF*);
            END (*IF*);
            buffer[first] := "1";
        END (*IF*);

    END Roundup;

(************************************************************************)

PROCEDURE Fformat (number: LONGREAL;  VAR (*OUT*) buffer: ARRAY OF CHAR;
                        start: CARDINAL;  VAR (*INOUT*) finish: CARDINAL;
                        LeftJustified: BOOLEAN;  VAR (*OUT*) error: BOOLEAN);

    (* Formats the second argument as a decimal number, left or right   *)
    (* justified depending on the value of LeftJustified, in            *)
    (* buffer[start..finish].  This procedure is known to be called     *)
    (* only with start=0 or start=1 with a sign in buffer[0]; so we     *)
    (* perform the justification on all of buffer[0..finish] if right   *)
    (* justification is specified.  In the case of left justification,  *)
    (* finish is updated to show the last buffer position actually      *)
    (* used; and this character position is followed by one or more NUL *)
    (* characters, except in the case where we have used the entire     *)
    (* field to hold the result.                                        *)

    VAR position: CARDINAL;
        integerpart: CARDINAL;  nextdigit: [0..9];

    BEGIN
        position := start;
        integerpart := VAL (CARDINAL, number);
        AssembleCardinal (integerpart, buffer, position, error);
        IF error THEN RETURN END(*IF*);

        IF position <= finish THEN
            buffer[position] := ".";
            INC (position);
            number := number - VAL (LONGREAL, integerpart);

            WHILE (position <= finish) DO
                number := 10.0*number;
                nextdigit := VAL (CARDINAL, number);
                buffer[position] := CHR(ORD("0") + nextdigit);
                INC (position);
                number := number - VAL (LONGREAL, nextdigit);
            END (*WHILE*);

            (* If the remainder is 0.5 or more, adjust the result by    *)
            (* rounding up.                                             *)

            IF number >= 0.5 THEN
                Roundup (buffer, start, finish);
            END (*IF*);

            (* Strip off the trailing zeros.    *)

            DEC (position);
            WHILE buffer[position] = '0' DO
                buffer[position] := CHR(0);
                DEC (position);
            END (*WHILE*);

            (* If we are left with a whole number, strip off the        *)
            (* decimal point.                                           *)

            IF buffer[position] = '.' THEN
                buffer[position] := CHR(0);
                DEC (position);
            END (*IF*);

            (* Right justify the result or modify finish, as specified. *)

            IF LeftJustified THEN
                finish := position;
            ELSE
                ShiftRight (buffer, 0, finish, finish-position);
            END (*IF*);

        END (*IF*);

    END Fformat;

(************************************************************************)

PROCEDURE Scale (VAR (*INOUT*) mantissa: LONGREAL;
                        VAR (*INOUT*) exponent: INTEGER;
                        power: CARDINAL;  lower, upper: LONGREAL);

    (* Adjusts mantissa so that lower <= mantissa < upper, while        *)
    (* keeping the quantity  (mantissa * 10^exponent) invariant.  To    *)
    (* save us some calculation, the caller must ensure that            *)
    (* upper = 10^power and lower = 10^(-power).                        *)

    BEGIN
        WHILE mantissa >= upper DO
            INC (exponent, power);  mantissa := lower*mantissa;
        END (*WHILE*);

        WHILE mantissa < lower DO
            DEC (exponent, power);  mantissa := upper*mantissa;
        END (*WHILE*);
    END Scale;

(************************************************************************)

PROCEDURE Separate (number: LONGREAL;  VAR (*OUT*) mantissa: LONGREAL;
                                        VAR (*OUT*) exponent: INTEGER);

    (* Separates the first argument into a mantissa and exponent part,  *)
    (* so that  number = mantissa * 10^exponent.                        *)

    BEGIN
        mantissa := number;  exponent := 0;
        Scale (mantissa, exponent, 256, 1.0E-256, 1.0E256);
        Scale (mantissa, exponent, 64, 1.0E-64, 1.0E64);
        Scale (mantissa, exponent, 16, 1.0E-16, 1.0E16);
        Scale (mantissa, exponent, 4, 1.0E-4, 1.0E4);
        Scale (mantissa, exponent, 1, 1.0E-1, 1.0E1);
    END Separate;

(************************************************************************)

PROCEDURE Eformat (number: LONGREAL;  VAR (*OUT*) buffer: ARRAY OF CHAR;
                                        start, finish: CARDINAL;
                                        VAR (*OUT*) error: BOOLEAN);

    (* Puts number into buffer[start..finish] in E format, with the     *)
    (* whole of buffer[0..finish] right justified.                      *)

    VAR mantissa: LONGREAL;  exponent: INTEGER;
        position: CARDINAL;

    BEGIN
        Separate (number, mantissa, exponent);

        (* Put the exponent into the buffer first, in order to find out *)
        (* how much space will be left for the mantissa.                *)

        position := start;
        AssembleExponent (exponent, buffer, position, error);
        error := error OR (position > finish);

        IF error THEN
            IF finish < HIGH(buffer) THEN
                buffer[finish+1] := CHR(0);
            END (*IF*);
        ELSE
            ShiftRight (buffer, start, finish, finish-position+1);

            (* Now assemble the mantissa into the buffer.       *)

            DEC (finish, position-start);
            Fformat (mantissa, buffer, start, finish, FALSE, error);
        END (*IF*);

    END Eformat;

(************************************************************************)
(*              CONVERSION OF REAL NUMBER TO CHARACTER STRING           *)
(************************************************************************)

PROCEDURE LongRealToString (number: LONGREAL;
                                        VAR (*OUT*) buffer: ARRAY OF CHAR;
                                        fieldsize: CARDINAL);

    (* Converts the number to a decimal character string in array       *)
    (* "buffer", right-justified in a field of "places" characters.     *)

    VAR start, finish, j: CARDINAL;  small: LONGREAL;  error: BOOLEAN;

    BEGIN
        IF fieldsize = 0 THEN RETURN END(*IF*);

        start := 0;  finish := fieldsize-1;  error := FALSE;

        (* Make sure that the string will fit into the buffer, and that *)
        (* it will be properly terminated.                              *)

        IF finish > HIGH(buffer) THEN
            DEC (fieldsize, finish-HIGH(buffer));
            finish := HIGH(buffer);
        ELSIF finish < HIGH(buffer) THEN
            buffer[finish+1] := CHR(0);
        END (*IF*);

        (* For a negative number, insert a minus sign.  *)

        IF number < 0.0 THEN
            IF fieldsize <= 1 THEN
                error := TRUE;
            ELSE
                buffer[0] := "-";  start := 1;  DEC(fieldsize);
                number := -number;
            END (*IF*);
        END (*IF*);

        IF NOT error THEN

            (* Now decide on whether to use E format, based on the      *)
            (* value to be converted.                                   *)

            small := 100.0 / TenToPower(fieldsize);
            IF number = 0.0 THEN
                Fformat (number, buffer, start, finish, FALSE, error);
            ELSIF (number >= TenToPower(fieldsize))
                    OR (number > VAL(LONGREAL, MAX(CARDINAL)))
                        OR (number < small) THEN
                Eformat (number, buffer, start, finish, error);
            ELSE
                Fformat (number, buffer, start, finish, FALSE, error);
            END (*IF*);

        END (*IF*);

        IF error THEN
            FOR j := 0 TO finish DO
                buffer[j] := '*';
            END (*FOR*);
        END (*IF*);

    END LongRealToString;

(************************************************************************)

PROCEDURE RealToString (number: REAL;  VAR (*OUT*) buffer: ARRAY OF CHAR;
                                        fieldsize: CARDINAL);

    (* Like LongRealToString, except for argument type. *)

    BEGIN
        LongRealToString (VAL(LONGREAL,number), buffer, fieldsize);
    END RealToString;

(************************************************************************)

PROCEDURE LongRealToF (number: LONGREAL;  VAR (*INOUT*) fieldsize: CARDINAL;
                        decimalplaces: CARDINAL;  LeftJustified: BOOLEAN;
                        VAR (*OUT*) buffer: ARRAY OF CHAR);

    (* Converts the number to an F-format string, of up to fieldsize    *)
    (* characters with decimalplaces digits after the decimal point.    *)
    (* The result is left justified if LeftJustified = TRUE is          *)
    (* specified by the caller, and right justified with space fill     *)
    (* otherwise.  On return fieldsize gives the number of character    *)
    (* positions actually used.  The result string is terminated with   *)
    (* at least one CHR(0) (which is not counted in fieldsize), except  *)
    (* where the result fills the entire buffer.                        *)

    VAR start, finish, j: CARDINAL;  scalefactor: LONGREAL;  error: BOOLEAN;

    BEGIN
        IF fieldsize = 0 THEN RETURN END(*IF*);

        start := 0;  finish := fieldsize-1;  error := FALSE;

        (* Make sure that the string will fit into the buffer, and that *)
        (* it will be properly terminated.                              *)

        IF finish > HIGH(buffer) THEN
            DEC (fieldsize, finish-HIGH(buffer));
            finish := HIGH(buffer);
        ELSIF finish < HIGH(buffer) THEN
            buffer[finish+1] := CHR(0);
        END (*IF*);

        (* For a negative number, insert a minus sign.  *)

        IF number < 0.0 THEN
            IF fieldsize <= 1 THEN
                error := TRUE;
            ELSE
                buffer[0] := "-";  start := 1;  DEC(fieldsize);
                number := -number;
            END (*IF*);
        END (*IF*);

        IF NOT error THEN

            (* Round the number to the desired number of decimal places. *)

            scalefactor := TenToPower (decimalplaces);
            number := scalefactor*number + 0.5;
            number := VAL(LONGREAL, VAL(CARDINAL, number)) / scalefactor;

            (* Perform the conversion.  *)

            Fformat (number, buffer, start, finish, LeftJustified, error);

        END (*IF*);

        IF error THEN
            FOR j := 0 TO finish DO
                buffer[j] := '*';
            END (*FOR*);
        END (*IF*);

        fieldsize := finish + 1;

    END LongRealToF;

(************************************************************************)

PROCEDURE RealToF (number: REAL;  VAR (*INOUT*) fieldsize: CARDINAL;
                        decimalplaces: CARDINAL;  LeftJustified: BOOLEAN;
                        VAR (*OUT*) buffer: ARRAY OF CHAR);

    (* Like LongRealToF, except for argument type.      *)

    BEGIN
        LongRealToF (VAL(LONGREAL,number), fieldsize, decimalplaces,
                                        LeftJustified, buffer);
    END RealToF;

(************************************************************************)
(*                  CONVERSION OF STRING TO CARDINAL                    *)
(************************************************************************)

PROCEDURE StringToHex (string: ARRAY OF CHAR): CARDINAL;

    (* Converts a hexadecimal character string to numeric, stopping at  *)
    (* the first non-digit character.  Leading spaces are permitted.    *)

    CONST HexChars = CharSet {"0".."9", "a".."f", "A".."F"};

    VAR position, value: CARDINAL;

    BEGIN
        position := 0;
        WHILE (position <= HIGH(string)) AND (string[position] = ' ') DO
            INC (position);
        END (*WHILE*);
        value := 0;
        WHILE (position <= HIGH(string)) AND (string[position] IN HexChars) DO
            value := 16*value;
            IF string[position] IN CharSet{"a".."f"} THEN
                value := value + 10 + ORD(string[position]) - ORD('a');
            ELSIF string[position] IN CharSet{"A".."F"} THEN
                value := value + 10 + ORD(string[position]) - ORD('A');
            ELSE
                value := value + ORD(string[position]) - ORD('0');
            END (*IF*);
            INC (position);
        END (*WHILE*);
        RETURN value;
    END StringToHex;

(************************************************************************)

(*
PROCEDURE StringToLongCard (string: ARRAY OF CHAR): LONGCARD;

    (* Converts a character string to decimal, stopping at the first    *)
    (* non-digit character.  Leading spaces are permitted.              *)

    VAR position: CARDINAL;  value: LONGCARD;

    BEGIN
        position := 0;
        WHILE (position <= HIGH(string)) AND (string[position] = ' ') DO
            INC (position);
        END (*WHILE*);
        value := 0;
        WHILE (position <= HIGH(string)) AND (string[position] >= '0')
                AND (string[position] <= '9') DO
            value := 10*value + VAL(LONGCARD, ORD(string[position]) - ORD('0'));
            INC (position);
        END (*WHILE*);
        RETURN value;
    END StringToLongCard;
*)

(************************************************************************)

PROCEDURE StringToCardinal (string: ARRAY OF CHAR): CARDINAL;

    (* Converts a character string to decimal, stopping at the first    *)
    (* non-digit character.  Leading spaces are permitted.              *)

    VAR position, value: CARDINAL;

    BEGIN
        position := 0;
        WHILE (position <= HIGH(string)) AND (string[position] = ' ') DO
            INC (position);
        END (*WHILE*);
        value := 0;
        WHILE (position <= HIGH(string)) AND (string[position] >= '0')
                AND (string[position] <= '9') DO
            value := 10*value + ORD(string[position]) - ORD('0');
            INC (position);
        END (*WHILE*);
        RETURN value;
    END StringToCardinal;

(************************************************************************)
(*                      CONVERSION OF STRING TO REAL                    *)
(************************************************************************)

PROCEDURE StringToLongReal (string: ARRAY OF CHAR): LONGREAL;

    (* Converts a decimal text string (with optional leading minus      *)
    (* sign) to real.  Leading blanks are ignored.  The conversion      *)
    (* stops at the end of the array or at the first character which    *)
    (* cannot be part of the number, and in the latter case all         *)
    (* subsequent characters are ignored.                               *)

    VAR result, placevalue: LONGREAL;
        position: CARDINAL;
        nextchar: CHAR;
        exponent: CARDINAL;  negative, negativeexp: BOOLEAN;

    (********************************************************************)

    PROCEDURE GetNextChar;

        (* Puts the next character in the input into variable nextchar, *)
        (* or sets nextchar := CHR(0) if there is no next character.    *)
        (* The position in the array is updated.                        *)

        CONST EndMarker = CHR(0);

        BEGIN
            IF position > HIGH(string) THEN
                nextchar := EndMarker;
            ELSE
                nextchar := string[position];  INC (position);
            END (*IF*);
        END GetNextChar;

    (********************************************************************)

    BEGIN
        result := 0.0;  position := 0;  negative := FALSE;

        (* Skip leading spaces and tabs.        *)

        REPEAT
            GetNextChar;
        UNTIL (nextchar <> " ") AND (nextchar <> tab);

        (* Check for a sign.    *)

        IF (nextchar = "-") OR (nextchar = "+") THEN
            negative := (nextchar = "-");

            (* There might be some more spaces to skip. *)

            REPEAT
                GetNextChar;
            UNTIL (nextchar <> " ") AND (nextchar <> tab);
        END (*IF*);

        (* Read the part before the decimal point.      *)

        WHILE nextchar IN CharSet {"0".."9"} DO
            result := 10.0*result + VAL(LONGREAL, ORD(nextchar)-ORD("0") );
            GetNextChar;
        END (*WHILE*);

        (* Now the part after the decimal point, if any.        *)

        IF nextchar = "." THEN
            GetNextChar;  placevalue := 0.1;
            WHILE nextchar IN CharSet {"0".."9"} DO
                result := result +
                        placevalue * VAL(LONGREAL, ORD(nextchar)-ORD("0") );
                placevalue := 0.1*placevalue;
                GetNextChar;
            END (*WHILE*);
        END (*IF*);

        (* Check for Ennn part. *)

        IF (nextchar = "E") OR (nextchar = "e") THEN
            GetNextChar;
            exponent := 0;  negativeexp := FALSE;
            IF nextchar = "+" THEN
                GetNextChar;
            ELSIF nextchar = "-" THEN
                negativeexp := TRUE;  GetNextChar;
            END (*IF*);
            WHILE nextchar IN CharSet {"0".."9"} DO
                exponent := 10*exponent + ORD(nextchar) - ORD("0");
                GetNextChar;
            END (*WHILE*);
            IF negativeexp THEN
                result := result / TenToPower(exponent);
            ELSE
                result := result * TenToPower(exponent);
            END (*IF*);
        END (*IF*);

        IF negative THEN
            result := -result;
        END (*IF*);
        RETURN result;

    END StringToLongReal;

(************************************************************************)

PROCEDURE StringToReal (string: ARRAY OF CHAR): REAL;

    (* Like StringToLongReal except for the result type.        *)

    BEGIN
        RETURN VAL(REAL,StringToLongReal(string));
    END StringToReal;

(************************************************************************)

END Conversions.