Accurate count for CSV tasks (#991)

README.md

@@ -198,8 +198,18 @@ My Title,Hello World!
 ```
 
 The files uploaded to your Active Storage service provider will be renamed to
-include an ISO 8601 timestamp and the Task name in snake case format. The CSV is
-expected to have a trailing newline at the end of the file.
+include an ISO 8601 timestamp and the Task name in snake case format.
+
+The implicit `#count` method loads and parses the entire file to determine the
+accurate number of rows. With files with millions of rows, it takes several
+seconds to process. Consider skipping the count (defining a `count` that returns
+`nil`) or use an approximation, eg: count the number of new lines:
+
+```ruby
+def count(task)
+  task.csv_content.count("\n") - 1
+end
+```
 
 #### Batch CSV Tasks
 

app/models/maintenance_tasks/csv_collection_builder.rb

@@ -15,14 +15,13 @@ def collection(task)
       CSV.new(task.csv_content, headers: true)
     end
 
-    # The number of rows to be processed. Excludes the header row from the
-    # count and assumes a trailing newline is at the end of the CSV file.
-    # Note that this number is an approximation based on the number of
-    # newlines.
+    # The number of rows to be processed.
+    # It uses the CSV library for an accurate row count.
+    # Note that the entire file is loaded. It will take several seconds with files with millions of rows.
     #
     # @return [Integer] the approximate number of rows to process.
     def count(task)
-      task.csv_content.count("\n") - 1
+      CSV.new(task.csv_content, headers: true).count
     end
 
     # Return that the Task processes CSV content.