Fail fast when trying to skip staging or enrich in stream enrich mode (…

…close snowplow/snowplow#3726)

lib/snowplow-emr-etl-runner/cli.rb

@@ -315,9 +315,20 @@ def self.validate_and_coalesce(args, config)
           raise ConfigError, 'resume-from and skip are mutually exclusive'
         end
 
-        if args[:resume_from] == "staging_stream_enrich" && config[:aws][:s3][:buckets][:enriched][:stream].nil?
+        if args[:resume_from] == "staging_stream_enrich" && config.dig(:aws, :s3, :buckets, :enriched, :stream).nil?
           raise ConfigError, 'staging_stream_enrich is invalid step to resume from without aws.s3.buckets.enriched.stream settings'
         end
+        unless config.dig(:aws, :s3, :buckets, :enriched, :stream).nil?
+          if args[:resume_from] == "enrich"
+            raise ConfigError, 'cannot resume from enrich in stream enrich mode'
+          end
+          if args[:skip].include?('staging') || args[:skip].include?('enrich') 
+            raise ConfigError, 'cannot skip staging nor enrich in stream enrich mode. Either skip staging_stream_enrich or resume from shred'
+          end
+          if args[:skip].include?('archive_raw') || args[:resume_from] == "archive_raw"
+              raise ConfigError, 'cannot skip nor resume from archive_raw in stream enrich mode'
+          end
+        end
 
         args[:include].each { |opt|
           unless INCLUDES.include?(opt)

spec/snowplow-emr-etl-runner/cli_spec.rb

@@ -239,9 +239,29 @@ def resource(name)
       Cli.validate_and_coalesce(a, c)
     end
 
-    it 'should accept stream-mode config' do
+    it 'should accept stream enrich mode config' do
       Cli.validate_and_coalesce(a, s)
     end
-  end
 
+    it 'should reject --skip staging in stream enrich mode' do
+      a[:skip] = [ 'staging' ]
+      expect {
+        Cli.validate_and_coalesce(a, s)
+      }.to raise_exception(ConfigError, "cannot skip staging nor enrich in stream enrich mode. Either skip staging_stream_enrich or resume from shred")
+    end
+
+    it 'should reject --resume-from enrich in stream enrich mode' do
+      a[:resume_from] = 'enrich'
+      expect {
+        Cli.validate_and_coalesce(a, s)
+      }.to raise_exception(ConfigError, "cannot resume from enrich in stream enrich mode")
+    end
+
+    it 'should reject --skip archive_raw in stream enrich mode' do
+      a[:skip] = [ 'archive_raw' ]
+      expect {
+        Cli.validate_and_coalesce(a, s)
+      }.to raise_exception(ConfigError, "cannot skip nor resume from archive_raw in stream enrich mode")
+    end
+  end
 end

spec/snowplow-emr-etl-runner/resources/stream_config.yml

@@ -12,6 +12,7 @@ aws:
       enriched:
         good: eg       # e.g. s3://my-out-bucket/enriched/good
         archive: ea    # Where to archive enriched events to, e.g. s3://my-out-bucket/enriched/archive
+        stream: es    # Where to archive enriched events to, e.g. s3://my-out-bucket/enriched/archive
       shredded:
         good: sg       # e.g. s3://my-out-bucket/shredded/good
         bad: sb       # e.g. s3://my-out-bucket/shredded/bad