Jim Barry's Rexx Tutor part 3

The PARSE Editor

PARSE VAR InVar Out1Var Out2Var Out3Var where: InVar = String to be split Out1Var = 1st word of InVar Out2Var = 2nd word of InVar Out3Var = All the Rest of InVar PARSE VAR InVar Out1Var . Out2Var where: InVar = String to be split Out1Var = 1st word of InVar . = Loose 2nd Word Out3Var = All the Rest of InVar
How PARSING works!
If you ask REXX programmer to explain PARSING they normally tell you to come back when you have 3 weeks to spare, as a result people get frightened of it all. I have to admit it's more involved than other areas, but it's not really more complex. In these examples you can see we are splitting up the contents of one variable 'InVar' and putting the bits in other variables. Note that the contents of this first variable ('InVar') is not changes. In the first template 'Out1Var' and 'Out2Var' take the first and second words of 'InVar', The last variable gets the remainder of 'InVar'. If we want to 'loose' a word just put in a full stop instead of a variable name. This is what we have done in the second template.

PARSE VAR InVar Out1Var "," Out2Var where: InVar = The Input Variable Out1Var = Gets everything before first comma "," = the pattern marker Out2Var = Gets everything after first comma PARSE VAR InVar Out1Var (In2Var) Out2Var where: InVar = Contains the string to be split In2Var = contains the pattern marker Out1Var = gets everything before pattern marker Out2Var = gets everything after pattern marker
Pattern Markers
In the last bit a string was split on spaces. this is the default pattern marker, here in the first template we have used a comma instead. So if we code a string between two variables then we split at the point where we find a match for the string. This string can be of any length. In the second tenplate the string we want to split at is coded in a vartiable, so somehow we have to tell the PARSE editor that the variable contains a pattern marker and is not an output variable. As you can see this is achieved by putting the variable in brackets.

 If  A = '   PARSE can split up words.'
     After: 'PARSE VAR A B C D'
     B = 'PARSE'   C = 'can'
     D = ' split up words.'

 If  A = ' Use a fullstop as a dummy variable.'
     After: 'PARSE VAR A B . C '
     B = 'Use'
     C = ' fullstop as a dummy variable.'

 If  A = 'can split at pattern markers.'
     After: 'PARSE VAR A B 'at' C'
     B = 'can split '  C = ' pattern markers.'

If   A = 'The "pattern marker" now in a variable.' and PatMark= '"'
     After: 'PARSE VAR A B (PatMark) C (PatMark) D'
     B = 'The '   C = 'pattern marker'
     D = ' now in a variable.'

PARSE Examples
Here are some PARSE examples. Note that leading and trailing spaces are being stripped off. It also might be worth saying here that there may be very slight differences on stripping on different platforms, do a few tests to find out.

   PARSE VAR InVar Out1Var 23 Out2Var

   where: InVar = Input variable
          Out1Var = gets columns 1 -22 of InVar
          23      = Split Position
          Out2Var = gets 23 to end of InVar


   PAR VAR InVar . ":" Out1Var +8 .

   where: InVar  = input variable
              .  = loose everything to the first colon
             ":" = pattern marker
         Out1Var = starts ":" for 8 characters
             +8  = length of previous varible (Out1Var)
              .  = dump the rest

Numerical Pattern Markers
Now the pattern markers are numbers and not strings. In the first template we can see that coding a number indicates a split position. The previous variable (Out1Var) gets everything up to this position and the next variable (Out2Var) starts at this point. The second template shows that when a number has a plus sign it means that the number is the length of the previous variable. In this 2nd template the result would be the same if the second full stop was not coded.

 PARSE VAR InVar Out1Var =(In2Var) Out2Var 20 .

 where:    InVar = Variable containing a string
         Out1Var = Gets the first bit of InVar
       =(In2Var) = The contents of In2Var is the split point
        Out2Var  = Get 20 characters from passed split point
             20  = The length of Out2Var
             .   = Dump the rest

 PARSE VAR InVar Out1Var +(InVar2) .

 where:   InVar  = variable containing a string
         Out1Var = firstr bit of InVar
        +(InVar) = the contents of In2Var is the length of Out1Var
               . = dump the rest

Numerical Position Pattern Markers in Variables.
In both these templates numerical values are not defined as constants but in variables. In the first the equal sign and the brackets mean the contents of 'In2' determining the split point. Its value must be a zero or a positive integer number . In the second template the plus sign means that the value of the variable is the length of the previous variable (here 'Out2' ).

For all of these 'A' = '123456789abcdefghij' After: 'PARSE VAR A 3 B 8 C' A = '12' B = '34567' C = '89abcdefghij' After: 'PARSE VAR A 3 B +8 C' A = '12' B = '3456789a' C = 'bcdefghij' If Pos= '4' and Len= '6' After: 'PARSE VAR A =(Pos) B +(Len) C' A = '123' B = '456789' C = 'abcdefghij'
PARSE Examples - 2
Here are some more PARSE examples showing positional pattern markers. No stripped is done on these of course. You can mix positional and literal pattern markers.

  PARSE VAR InVar Out1Var Out2Var =(Out1Var) .

  where: InVar = variable containing a string
       Out1Var = the first word of Invar
      Out2Var  = the next part og InVar
    =(Out1Var) = Out1Var holds the length of Out2Var
             . = dump the rest

  PARSE VAR InVar Out1Var 15 Out2Var 1 Out3Var
  where: InVar = variable containing a string
       Out1Var = gets the first 14 chars. of InVar
            15 = split position
      Out2Var  = gets 15 to end of InVar
             1 = split position
      Out3Var  = whole of inVar

Complex PARSE instructions
Life's getting slightly more complex here. In the first template the value of the positional pattern variable is extracted from the input string just before its used to determine the length of the next variable! The second template details what happens when a numeric positional pattern has a value less than a previous positional pattern - it just goes back to that position. In the example here we go back to the 1st character so 'Out3Var' would be the same as 'InVar'.


  PARSE VALUE "a long string" WITH Out1Var Out2Var etc

  where: "a long string" = the input to PARSE
            Out1Var etc  = output variables and pattern markers

  PARSE VALUE funct() WITH Out1Var etc

  where:    funct()  = function result
         Out1Var etc = output variables and pattern markers

PARSE using VALUE & WITH
In the last templates the input to the editor could only be a variable. Using VALUE and WITH the input can be just about anything. In the first template here we are splitting up a literal string constant that has been coded. To be honest you are unlikely to want to do this in real life, however you will want to do what is shown in the second template. Here we are splitting up the return from a function. Anything after the word 'WITH' are either the output variables or pattern markers. This produces a very powerful feature.

/* REXX. PARSE Editor continued   */
SAY "If  TIME()  = '" || TIME() || "'"
SAY ' After: ''PARSE VALUE TIME() WITH hh ":" mm ":" ss'''
               PARSE VALUE TIME() WITH hh ":" mm ":" ss
SAY "    hh = '" ||  hh || "'"
SAY "    mm = '" ||  mm || "'"
SAY "    ss = '" ||  ss || "'"
A = "one two three four five six seven"
SAY "A = ""one two three four five six seven"""
SAY "If you keep executing: 'PARSE VAR A . A'"
SAY "THEN..."
DO WHILE  A \= ""
   SAY "     A= '" || A"'"
   PARSE VAR A . A
   END
-------------------------------------------------------
  If  TIME()  = '13:39:39'
   After: 'PARSE VALUE TIME() WITH hh ":" mm ":" ss'
      hh = '13'
      mm = '39'
      ss = '39'
  A = "one two three four five six seven"
  If you keep executing: 'PARSE VAR A . A'
  THEN...
     A= 'one two three four five six seven'
     A= ' two three four five six seven'
     A= ' three four five six seven'
     A= ' four five six seven'
     A= ' five six seven'
     A= ' six seven'
     A= ' seven'

PARSE Examples
Here are a couple of other examples of PARSE. In the first we are execution the TIME function and splitting the output, all in one instruction. In the second example we are breaking down a string word by word. We do this be coding the input string as one of the outputs. We are just take off the first word on each execution as you can see.

     PARSE UPPER VAR InVar OutVar

     where:  UPPER = converts all chars to upper case.


     PARSE UPPER PULL
     where:  UPPER = converts all chars to upper case.
              PULL = get line from input stream


     PARSE UPPER ARG
     where:  UPPER = converts all chars to upper case.
               ARG = gets arguments going into program or function

UPPER, PULL & ARG
We haven't seem these before. UPPER can be coded after the word PARSE to ensure that we also do a case conversion. Simple and useful. In the second template here we have dropped the word 'OutVar', now we are getting our input from a different place. In the first line it is coming from the input stream. Exactly what that is will depend on where you are running your REXX. In second line ARG has been coded, here the input is coming from arguments, either passed to a function if it's coded in a function, or arguments passed to the program if its not coded in a function.
Now here something interesting, coding PULL actually gives you a PARSE UPPER PULL and coding ARG you get an PARSE UPPER ARG!

TOP . . BACK