How to reduce the size of scope of this batch class

Question

Due to a platform limitation that trigger on attachment won't be invoked when an attachment is sent via email I wrote a batch class to clone and parent email attachments from EmailMessage to a custom object Comm__c

global class BatchforOutboundEmailAttachments implements Database.Batchable<sObject>, Schedulable {    

    global Database.QueryLocator start(Database.BatchableContext bc){

        return Database.getQueryLocator([SELECT Name, Body, ParentID from Attachment Where Parent.Type = 'EmailMessage']);    
    }    

    global void execute(Database.BatchableContext bc, Attachment[] scope){

        set<id> parents = new set<id>();
        List<Attachment> newFiles = new List<Attachment>();
        Map<Id,EmailMessage> emailMap = new Map<Id,EmailMessage>();
        List<Attachment> oldFiles = new List<Attachment>();    

            for(Attachment att : scope){
                parents.add(att.ParentID);
            }

            if(!parents.isEmpty()){

       emailMap = new Map<ID, EmailMessage>([SELECT RelatedToId FROM EmailMessage WHERE Id in :Parents]); 

            for (Attachment att : scope) {
              if(emailMap.get(att.ParentId) != Null && emailMap.get(att.ParentId).RelatedToId != NULL){
                if (emailMap.get(att.ParentId).RelatedToId.getSObjectType() == Comm__c.SObjectType) {

                    Attachment newFile = att.clone();
                    newFile.ParentId = emailMap.get(att.ParentId).RelatedToId;
                    newFiles.add(newFile);
                    oldFiles.add(att);

                }
             }
            }     


       try{
            insert newFiles;
            delete oldFiles;
          }

           catch(DMLexception de){}    
        }

    }

    global void finish(Database.BatchableContext bc){
     }

    global void execute(SchedulableContext sc){

         ID jobId = Database.executeBatch(this,100); 

        }
}

In the start method I'm receiving 12k attachments which are EmailMessage records but not related to Comm__c i.e.( RelatedToID is not of SobjectType Comm__c)

How can I reduce the scope size so that less number of batches are processed? Someone from the client team doesn't want the unrelated attachments to be returned in the scope

Answer 1

The type of the parent object can be included as a filter as described in How to filter query for specific types of parent lookup? and you are already doing that.

But you are also asking for the SOQL query engine to accept your coded logic that the parent object type is of EmailMessage so that a RelatedToId can also be assumed. In code you might be able to inform the compiler using a cast, but AFAIK no such mechanism exists in SOQL.

So I don't think you can filter further in the SOQL and so will need to stick to the Apex filtering.