lib/assimilate/batch.rb in assimilate-0.0.1 vs lib/assimilate/batch.rb in assimilate-0.0.2

@@ -12,10 +12,11 @@
 
     @noops = []
     @changes = []
     @adds = []
     @deletes = []
+    @resolved = false
   end
 
   def load_baseline
     stored_records = @catalog.catalog.find(@catalog.domainkey => @domain).to_a
     @baseline = stored_records.each_with_object({}) do |rec, h|
@@ -50,27 +51,32 @@
   end
 
   # compute anything needed before we can write updates to permanent store
   # * find records that have been deleted
   def resolve
-    @deleted_keys = @baseline.keys - @seen.keys
+    if !@resolved
+      @deleted_keys = @baseline.keys - @seen.keys
 
-    @updated_field_counts = @changes.each_with_object(Hash.new(0)) do |rec,h|
-      key = rec[idfield]
-      diffs = rec.diff(stripped_record_for(key))
-      diffs.keys.each do |f|
-        h[f] += 1
+      @updated_field_counts = @changes.each_with_object(Hash.new(0)) do |rec,h|
+        key = rec[idfield]
+        diffs = rec.diff(stripped_record_for(key))
+        diffs.keys.each do |f|
+          h[f] += 1
+        end
       end
+
+      @resolved = true
     end
   end
 
   def stats
     resolve
     {
       :baseline_count => @baseline.size,
       :final_count => @baseline.size + @adds.count,
       :adds_count => @adds.count,
+      :new_ids => @adds.map {|rec| rec[idfield]},
       :deletes_count => @deleted_keys.count,
       :updates_count => @changes.count,
       :unchanged_count => @noops.count,
       :updated_fields => @updated_field_counts
     }
@@ -84,11 +90,13 @@
     apply_inserts
     apply_updates
   end
 
   def record_batch
-    raise(Assimilate::DuplicateImportError, "duplicate batch") if @catalog.batches.find('domain' => @domain, 'datestamp' => @datestamp).to_a.any?
+    raise(Assimilate::DuplicateImportError, "duplicate batch for datestamp #{datestamp}") if @catalog.batches.find('domain' => @domain, 'datestamp' => @datestamp).to_a.any?
+    raise(Assimilate::DuplicateImportError, "duplicate batch for file #{@filename}") if @catalog.batches.find('domain' => @domain, 'filename' => @filename).to_a.any?
+
     @catalog.batches.insert({
       'domain' => @domain,
       'datestamp' => @datestamp,
       'filename' => @filename
       })
@@ -134,6 +142,9 @@
     end
   end
 end
 
 class Assimilate::DuplicateImportError < StandardError
+end
+
+class Assimilate::CorruptDataError < StandardError
 end