Accepted Solutions
Your %sendmail() macro shouldn't be doing anything else than sending emails. All the other stuff where you create attachments (or don't) and then decide what email text with/without attachments to send does NOT belong into the %sendmail() macro but into an upstream process (another node, eventually user written).
I would create a generic reusable %sendmail() macro implemented in DIS via a custom transformation with prompts for the parameters of the %sendmail() macro.
@ScottBass published such %sendmail() macro code here which you could use.
If using Scott's macro then your upstream process could simply populate as SAS table with the required parameter values and you then pass in the name of this table to macro parameter mail.
It's in this node where you populate this parameter table that you build the logic if you need to send an attachment or not - but that's just simple datastep logic very likely implemented as user written code.
OK - You'll get now in the following quite a bit more than what I normally would share in the forums here. But I acknowledge that Q&A with DIS jobs is a bit harder than talking SAS code only.
I've attached for you a fully executable self contained DIS job (version 4.903). If you select "My Folder" for import of the .spk then it will create the following sub-folder and objects.
You can use this job as template for what you need to implement. Below how the sample job looks like:
For running/testing the sample job the only thing which requires change is the email address (hard coded in node 8).
Just replace in the code of node 8 "uw - send email" the dummy address with your own email address. Then you can execute the code and should receive an email with 2 attachments.
To test the case where one of the source tables has no data: Go into node 1 and comment/uncomment code in order to execute the test case for "no data".
Once you're happy with what the template job does, replicate the relevant bits into your real job.
Important: For the current code to work the input tables to Node 3 "uw check" must be SAS tables (not views, not database). If that's not the case then you need to amend the code.
I'd recommend that you don't hard code email addresses into the code. Either use a macro variable instead which is defined somewhere in a config area or if hard coding use a group email address. Your goal must be that adding/removing recipients doesn't require any change to the DIS job.
SAS DIS best practice is to use out-of-the-box transformations before user written code. Even following this best practice I still had to use quite a bit of user written code. BUT: Even with user written code try to implement as metadata driven as you can. You'll see in the job posted that also the user written code uses metadata like DIS generated and metadata driven macro variable &_output for a target table instead of a hard coded table name.
On a side note:
I'd still prefer to have a reusable custom transformation for sending emails. You then would just have a step before this custom transformation which prepares the parameters (like sending an attachment or not) and then pass this to the reusable sendmail custom transformation (=replacing node 8 with a user written transformation preparing the parameters and a reusable custom transformation then sending the email).
Doing this here was just too much work so I went for a fully user written approach.
I've had another look into the %sendmail macro code which I've proposed earlier for you to use as a starting point for implementation of a site specific %sendmail macro. I need to revoke this recommendation. This code is too involved and uses a whole framework of other utility macros which you also would need to implement. But: There is quite a bit of SAS code for sending emails published so I'm sure if implementing a generic %sendmail macro then a bit of Googling would allow you to not have to start from scratch.
I'm happy to answer more questions. My expectation though is that you now dive into the sample I've provided here and try to fully understand what it does and how it works before you ask follow-up questions.
And last but not least:
I'm writing the .csv to the SAS Work folder which only exists during a SAS session (path in external file metadata like: %sysfunc(pathname(work))/class.csv). I've done it this way to keep the whole DIS job self contained with a working path but not interfering with anything else you have in your environment. SAS will delete WORK as soon as you end your SAS session so there won't be any "left-overs". For your actual implementation: Consider writing the files to another disk location. Reason: Email delivery is not always guaranteed (i.e. email server down). So if this happens then you have still the .csv's available without re-running the job.
Also here: You need to implement in a way that no change to the DIS job is required when moving from one environment to the next. Make sure that the path you use will work for all environments (i.e. for different Lev's). I normally have a config where I define root paths in macro variables and then use in my code/DIS jobs paths like: &pathToProject/output
And about sending email attachments: If these attachments can get big then consider to zip them first and attach the zip archive (filename zip or ods package lets you create such zip archives using SAS).
Need to connect to databases in SAS Viya? SAS’ David Ghan shows you two methods – via SAS/ACCESS LIBNAME and SAS Data Connector SASLIBS – in this video.
Find more tutorials on the SAS Users YouTube channel.
-
7290 views
-
3 likes
-
-
