/* dtchk.sas Daniel Brockman 070721 Check dates */


%macro dtchk (iset,yr,mo,dy) ;

/* 
* dtchk prints an error message in the log if it detects an invalid 
* date in iset.
*
* iset   = name of data set to consider.
* yr     = name of year variable stored in iset.
*           The variable contains an integer between 1 and 9999 inclusive.
*             1 or 2 digits represents a year not more than 50 years from
*                  this year. 
*             3 or 4 digits represents the year number unambiguously.
* mo     = name of month variable stored in iset.
*           The variable contains an integer between 1 and 12 inclusive.
* dy     = name of day variable stored in iset.
*           The variable contains an integer between 1 and 31.
*
* Notes: 
* 1. The above limits on the numbers in yr, mo and dy apply for valid
* dates. The numbers for invalid dates may not comply with these limits.
* 2. You may have to create the 3 variables yr, mo and dy in the data 
* set you want to check. To maintain generality, dtchk makes no assumptions
* about the data set except that it contains variables representing year, 
* month and day as described above.
* 3. dtchk executes a data step.
* 4. dtchk writes nothing in the log when it encounters a valid date.
* 5. a date 50 years from this year is an ambiguous case for a 2-digit year.
*
* Example:
* 
* 
* data mydataset ;
*   
*   input somekindadate ;
*   
*   yr = int(somekindadate/10000) ;
*   mo = int((somekindadate - 10000*yr)/100) ;
*   dy = somekindadate - 10000*yr -100*mo ;
* 
* cards;
* 19000229
* 19000228
* 20000229
* 20000228
* 000229
* 000228
* 14921012
* 920431
* August 15, 1914
* 920430
* 929230
* 920229
* 920228
* 570712
* 580712
* 560712
* 5
* 25
* 225
* 1225
* 01225
* 101225
* 101225.5678
* 0101225
* 20101225
* 220101225
* .
* 0
* notanumber
* -88
* 3.1416
* 11:15
* 20070132
* 20070229
* 20070332
* 20070431
* 20070532
* 20070631
* 20070732
* 20070832
* 20070931
* 20071032
* 20071131
* 20071232
* 20070131
* 20070228
* 20070331
* 20070430
* 20070531
* 20070630
* 20070731
* 20070831
* 20070930
* 20071031
* 20071130
* 20071231

* 
* run;
*
* %dtchk(mydataset,yr,mo,dy);
*   
*/ 

data _NULL_ ;
  set &iset ;
  
  * number of days in each month of non-leap year ;
  
  * days per month ;
  a1=31;
  a2=28 ;
  a3=31 ;
  a4=30 ;
  a5=31 ;
  a6=30 ;
  a7=31 ;
  a8=31 ;
  a9=30 ;
  a10=31 ;
  a11=30 ;
  a12=31 ;
  array dpm(12) a1-a12;

  good=1;                        * starting assumption: date is good ;
    
       if (&dy ne int(&dy)) then good=0;      * yr not an integer? ;
  else if (&mo ne int(&mo)) then good=0;      * yr not an integer? ;
  else if (&yr ne int(&yr)) then good=0;      * yr not an integer? ;
  else do;                                    * all integers ? ;

    if (&yr<100) then do;          * 2 digit year? ;
      tyyyy = year(today());       * this year ;
      tcc = int(tyyyy/100) ;       * this century ;
      
      tcc&yr = tcc*100 + &yr ;     * this year if yr is in this century ;
      if (tcc&yr - tyyyy > 50) then tcc=tcc-1 ;   * yr must be prev century ;
      if (tyyyy - tcc&yr >=50) then tcc=tcc+1 ;   * yr must be next century ;
      &yr = tcc*100 + &yr ;        * we will work with this year number ;
          
    end ;                          * 2 digit year ;
  
  
         if (&yr<1 or &yr>9999) then good=0;    * yr out of range?;
    else if (&mo<1 or &mo>12)   then good=0;    * mo out of range?;
    else if (&dy<1 or &dy>31)   then good=0;    * dy out of range?;
    else if (&mo = 2) then do;                  * February?? ;

      * discover whether leap year ;
  
      leap=0 ;                                  * assume not a leap year;
      if (mod(&yr,400)=0) then leap=1;          * yr divides by 400? ;
      else if (mod(&yr,100)=0) then  leap = 0 ; * century yr? ;
      else if (mod(&yr,4)=0) then leap=1 ;      * yr divides by 4? ;
      else leap = 0 ;                           * regular year ;
  
      if (leap and &dy>29) then good=0;       * >29 days in leap yr? ;
      else if ((not leap) and &dy>28) then good=0;   * >28 days in regular yr? ;
   
    end;                                        * February ;
  
    else do ;                                   * some other month? ;
  
      * too many days? ;
      if (&dy>dpm(&mo)) then good=0;     
  
      * too many days? ;
      * if (&dy>30 and (&mo=9 or &mo=4 or &mo=6 or &mo=11)) then good=0; 
  
    end;                                        * some other month ;

  end;                                          * all integers ;

  if (not good) then put 'ERROR BAD DATE ' _N_= ' ' &yr.= ' ' &mo= ' ' &dy= ;
  
  
%mend dtchk ;

* --------------------------------------------------------- ;