Would you like to query your database from CL?
Last month, in "Introducing the New Run SQL Command," we reviewed how you could insert, update, and delete rows/records in a database using the RUNSQL CL command. Today, we'll look at how the RUNSQL command can be used to query, or select, data from the SAMPLE database we created in the earlier article.
We previously inserted one record into the SAMPLE database using the RUNSQL command:
RUNSQL SQL('Insert into SAMPLE Values(''FIRST'', 1, ''2012-04-13'')')
COMMIT(*NONE)
Using the RUNSQL command, now insert three additional records (leaving us with four records in SAMPLE). For purposes of this article, the values that you should use are shown below.
CLASS STATUS EFFDATE
A 1 2012-05-01
ABC 2 2012-04-15
3rd 2 2012-06-01
Let's now assume that you want to query this SAMPLE data from a CL program, extracting only those records that meet certain criteria. As I mentioned last month, the SQL SELECT statement support isn't as direct as the INSERT, UPDATE, and DELETE SQL statements. The reason for this is that the RUNSQL command doesn't have any output parameters (like those found with a Retrieve CL command, for instance) where we could directly access the results of the SELECT statement. What we need to do is write the results of the SELECT to an intermediate file/table and then process the resulting data using the Receive File (RCVF) command.
The following program, named QRYSAMPLE, runs two queries over the SAMPLE file using the RUNSQL command. The first query selects only those SAMPLE records in which the CLASS field value starts with the capital letter A. The second query selects only those SAMPLE records in which the effective date (field EFFDATE) value is less than May 1 of 2012. The results of both queries are sent to the user as messages. The program also sends a message, indicating end of file, at the end of each query. Note that both of the values being selected on (the letter A and the date May 1 2012) could be implemented as parameters passed to the QRYSAMPLE program. I'm using literal values for now simply to simplify the construction of the SQL parameter of the RUNSQL command. An example of using CL variables within a SQL statement can be found in last month's article.
Pgm
DclF File(Sample) OpnID(MyResults)
Dcl Var(&EOF) Type(*Lgl)
CrtDupObj Obj(Sample) FromLib(*Libl) +
ObjType(*File) +
ToLib(QTemp) NewObj(MyResults)
RunSQL SQL('Insert into QTemp/MyResults +
(Select * +
from Sample +
where Class like(''A%''))') +
Commit(*None)
CallSubr Subr(ReadFile)
RunSQL SQL('Delete from QTemp/MyResults') +
Commit(*None)
RunSQL SQL('Insert into QTemp/MyResults +
(Select * +
from Sample +
where EffDate < ''2012-05-01'')') +
Commit(*None)
CallSubr Subr(ReadFile)
Return
Subr Subr(ReadFile)
OvrDBF File(Sample) ToFile(QTemp/MyResults)
ChgVar Var(&EOF) Value('0')
DoUntil Cond(&EOF = '1')
RcvF OpnID(MyResults)
MonMsg MsgID(CPF0864) Exec( +
ChgVar Var(&EOF) Value('1'))
If Cond(&EOF *EQ '0') Then( +
SndPgmMsg Msg( +
'Class' *BCat +
&MyResults_Class *BCat +
'is effective on' *BCat +
&MyResults_EffDate) +
ToPgmQ(*Ext))
Else Cmd( +
SndPgmMsg Msg('End of file') +
ToPgmQ(*Ext))
EndDo
Close OpnID(MyResults)
DltOvr File(Sample)
EndSubr
EndPgm
Assuming that you stored the source shown above in member QRYSAMPLE of source file QCLSRC, you can create the program using the command CRTBNDCL QRYSAMPLE. Running the program, with CALL QRYSAMPLE, should result in messages like those shown below.
Class A is effective on 2012-05-01
Class ABC is effective on 2012-04-15
End of file
Class FIRST is effective on 2012-04-13
Class ABC is effective on 2012-04-15
End of file
Now let's look at what the QRYSAMPLE program is doing.
QRYSAMPLE starts by declaring the file SAMPLE. The QRYSAMPLE program, as we'll see later, extracts all fields of the SAMPLE file (though a subset of the records), so we can use SAMPLE as a model file for the results of the various SQL SELECT statements. The DCLF of SAMPLE simply provides an expedient way for the program to declare the variables &CLASS, &STATUS, and &EFFDATE (as &MyResults_Class, &MyResults_Status, and &MyResults_EffDate, respectively, due to the OPNID keyword of the DCLF command) that will be later retrieved. If QRYSAMPLE was using the SQL SELECT to derive additional fields (for instance, a count of the records with a given status), to join to fields in another file (for instance, to get a text description of the status value), or the like, then we would want to create a model file that defines the results of the SELECT. To avoid possible confusion later, between the SAMPLE file and the results of subsequent RUNSQL commands run in the program, the declared file uses an Open file identifier of MyResults.
The program also declares a logical variable &EOF, which is used to track whether or not end of file has been encountered. QRYSAMPLE could be written without using this variable, but I prefer to explicitly track state information (such as the status of a file) so it's declared.
With the program declares out of the way, the first command run by QRYSAMPLE (CRTDUPOBJ) creates an empty instance of the SAMPLE file named MYRESULTS in library QTEMP. There are many, many possible ways of creating the MYRESULTS file, with the CRTDUPOBJ being shown just one (very simplistic) approach. I say this is a simplistic approach as there is no MONMSG associated with the CRTDUPOBJ command to handle the situation of MYRESULTS already existing in QTEMP (as it would if you ran QRYSAMPLE twice in the same job). More robust solutions might include these:
- Checking whether QTEMP/MYRESULTS already exists and, if so, clearing the file rather than creating it
- Monitoring failure of the CRTDUPOBJ due to MYRESULTS already existing and then clearing the file
- After each successful run of QRYSAMPLE, deleting QTEMP/MYRESULTS prior to returning
- Always attempting to delete QTEMP/MYRESULTS (and monitoring for object not found) prior to running the CRTDUPOBJ
To avoid getting into the pros and cons of these various approaches (which is decidedly off topic to our introduction to the RUNSQL command), the program simply tries to create MYRESULTS into QTEMP, leaving the knotty details of a more complete implementation to you.
Having created MYRESULTS, QRYSAMPLE then uses the RUNSQL command to insert (write) all records of SAMPLE into MYRESULTS where the CLASS field starts with the letter A. QRYSAMPLE then calls subroutine ReadFile to process the results of the RUNSQL command.
The ReadFile subroutine does the following:
- Overrides the declared file of SAMPLE to file MYRESULTS in QTEMP (where our results are stored)
- Changes the variable &EOF to "off," indicating that end of file has not been reached for MYRESULTS
- Enters into a DOUNTIL loop to process each record found in MYRESULTS. The DOUNTIL QRYSAMPLE reads (receives) the next record of MYRESULTS. If a record is successfully read, the class and effective date are shown by way of a message. If end of file was encountered, the message "End of file" is shown and the DOUNTIL is exited.
- Closes the file MYRESULTS
- Deletes the override of SAMPLE to QTEMP/MYRESULTS
- Returns to the main logic of QRYSAMPLE
Having displayed the results of the first RUNSQL SELECT command, QRYSAMPLE now performs these actions:
- Uses the RUNSQL command to delete the results of the previous RUNSQL SELECT command from MYRESULTS
- Runs a second SELECT to insert (write) all records of SAMPLE into MYRESULTS where the effective date is prior to May 1 2012
- Calls subroutine ReadFile to now process the results of the second SELECT. Due to the use of the CLOSE command (a very handy command introduced in 6.1) in the first call to the ReadFile subroutine, the initial RCVF command run in our second call to ReadFile will cause MYRESULTS to be reopened and positioned to the first record returned by the second RUNSQL SELECT command. The ReadFile subroutine will then process this second set of results in the same manner as described previously.
Having displayed the results of the second RUNSQL SELECT command, QRYSAMPLE ends.
QRYSAMPLE demonstrates how you might access the results of SQL SELECT statements within a CL program. In future articles, we'll look at other possible usages of RUNSQL within CL.
More CL Questions?
Wondering how to accomplish a function in CL? Send your CL-related questions to me at
LATEST COMMENTS
MC Press Online