Proper Field Validation Methods and other annoyances...

[Rob Cozens]

Hi Len,

>1) Why does  put "4" is a date return true?  If this function 
>doesn't actually work, why is it still available?  I'd like to be 
>able to have users enter a date without any kind of punctuation 
>(i.e., 08142007) and be able to validate it as a valid date.  The 
>same applies to entering 1315 as a time (no colon).

Others have explained why "is a date" returns true for any integer.

Here is a function that validates user input as valid according to 
the current system date format and returns the date in dateItems format.

  function validDate theString -- 5 April 04:RCC
   put stripBlanks(theString,false) into theString -- strips leading 
or trailing blanks from a string
   get systemDateFormat() -- see Called Routines below
   set the itemDelimiter to char -1 of it -- the Separater
   if the number of items of theString <> 3 then return false
   delete the last char of it
   delete char offset("mm",it) of it -- remove double characters, if any...
                                                          -- which 
there will be until RR system date
                                                          -- 
correctly recognizes suppression of leading
                                                          -- zeroes
   delete char offset("dd",it) of it
   delete char offset("yy",it) of it
   repeat with x = 1 to 3
     switch char x of it
     case "m"
       put item x of theString into theMonth
       if not validDigits(theMonth) then return false  -- each 
char >= "0" and <= "9"
       if theMonth < 1 or theMonth > 12 then return false
       break
     case "d"
       put item x of theString into theDay
       if not validDigits(theDay) then return false
       if theDay < 1 or theDay > 31 then return false
       break
     case "y"
       put item x of theString into theYear
       if not validDigits(theYear) then return false
       put length(theYear) into charCount
       if charCount <> 2 and charCount <> 4 then return false
       if charCount is 2 then add 2000 to theYear
       if theYear < 1 then return false
       break
     end switch
   end repeat
   if theDay < 29 then return 
true&return&theYear&comma&theMonth&comma&theDay&",0,0,0,0"
   if theMonth is 2 then
     if theDay > 29 or not leapYear(theYear) then return false -- 
leapYear returns boolean
     else return true&return&theYear&comma&theMonth&comma&theDay&",0,0,0,0"
   else
     if offset("0",theMonth) = 1 then delete char 1 of theMonth
     set the itemDelimiter to comma
     if theMonth is among the items of "1,3,5,7,8,10,12" then return 
true&return&theYear&comma&theMonth&comma&theDay&",0,0,0,0"
     if theDay is 31 then return false
     else return true&return&theYear&comma&theMonth&comma&theDay&",0,0,0,0"
   end if
end validDate

Cal;led routines:

function systemDateFormat
   put "1999,1,9,0,0,0,0" into testDate
   convert testDate to short system date -- RR pads day/month 
regardless of settings
                                                                        -- 
which means function will not recognize the
                                                                       -- 
absence of leading zeroes until this is fixed
   put 2 into characterNumber
   repeat while char characterNumber of testDate is in "1,9"
     add 1 to characterNumber
   end repeat
   put char characterNumber of testDate into dateSeparater
   put empty into returnDateFormat
   set the itemDelimiter to dateSeparater
   repeat for each item datePart in testDate
     if datePart > 9 then
       if datePart > 99 then put "y" after returnDateFormat
       put "y" after returnDateFormat
     else if datePart > 1 then
       if length(datePart) > 1 then put "d" after returnDateFormat
       put "d" after returnDateFormat
     else
       if length(datePart) > 1 then put "m" after returnDateFormat
       put "m" after returnDateFormat
     end if
   end repeat
   return returnDateFormat&dateSeparater
end systemDateFormat


>2) This is the "biggie."  What is the proper way to validate the 
>contents of a field.  I tried writing a "closeField" script that 
>would do the validation, put up a message box if the format/contents 
>are wrong, then set the focus back to the field for correction by 
>the user. I end up getting two insertion point cursors (one in the 
>"bad" field and one in the next field.  The keyboard does go to the 
>bad field but having more than one insertion point is confusing to 
>me let alone my end users.  By the way, I have tried not passing the 
>closeField message but I'm not sure if the "tab to next field" takes 
>place anyway.

Among my back burner projects is a Data Dictionary-driven RAD, which 
will validate user input in a field as compliant with the Data 
Dictionary specifications for the data that field captures (eg: 
required?, numeric?, # of decimal positions, show thousands 
separator?, show currency symbol?, lookup table, autoentry?, max 
length, max lines, etc).  Only some of the edits have been tested at 
this stage; but all that our working are triggered by closeField, 
enterInField (and returnInField for single-line fields),or internal 
logic if auto-entry fields, and I have not experienced the problems 
you describe.

However, I would note that my user interface keeps ALL fields without 
focus locked while accepting user input in a field.  When the user 
signals "I'm done", the field contents are validated to DD specs.  If 
invalid, an error message is displayed; if valid, the field is locked 
and the next (either by ordinal or based on the user input) field is 
unlocked and given focus.

>3) Related to #2 above, I frequently get vertical lines (insertion 
>point cursors) in a field as I move the mouse across it.  Once they 
>start showing up, (and they stay so I end up with vertical bars 
>between every letter in the field!), I can get them to appear by 
>moving the mouse ABOVE the field (not in it) like I have a vertical 
>line that runs the height of the entire screen.  Is this something 
>I'm doing or is it a known problem?

Do you get the same effect from a standalone compiled from your stack?

Rob Cozens CCW
Serendipity Software Company

"And I, which was two fooles, do so grow three;
  Who are a little wise, the best fooles bee."

  from "The Triple Foole" by John Donne (1572-1631)