aboutsummaryrefslogtreecommitdiffstats
path: root/spec/models
diff options
context:
space:
mode:
authorMark Longair <mhl@pobox.com>2013-12-03 17:30:16 +0000
committerMark Longair <mhl@pobox.com>2013-12-03 19:01:01 +0000
commitf920268650b6bb906be1828ed26a3732f4d10cc1 (patch)
tree75f088d397c92db8c94c66d45cd66abcdb2b0bc4 /spec/models
parent2eb478a4bf9d6f4cf22bbf2fb63a0e16716ec20b (diff)
Fix the command-line CSV importer under Ruby 1.9
Under Ruby 1.8.7, you can parse a CSV file with the following code (Example A): require 'csv' CSV.parse('foo.csv') do |row| puts "got row: #{row.inspect}" end Rather confusingly, under Ruby 1.8.7, CSV.parse can also take a string representation of the contents of the file as its parameter, so this also works (Example B): require 'csv' CSV.parse("1,hello,red\n2,goodbye,green") do |row| puts "got row: #{row.inspect}" end However under Ruby 1.9.3, CSV.parse only expects a string representation of the contents of the CSV file, so only Example B works; Example B fails silently (interpreting the filename as a single cell CSV file, typically). The import:import_csv rake task unfortunately relied on both A and B working. This commit fixes this by adding PublicBody.import_csv_from_file, and refactoring PublicBody.import_csv to use the newly added class method, and adds a test to check for any regression in this behaviour. (This means that the usage of import_csv in the admin public body controller's import_csv action could now be changed to use PublicBody.import_csv_from_file directly from the uploaded file, which would be more efficient and cope with larger files without using lots of memory.) Fixes #1229
Diffstat (limited to 'spec/models')
-rw-r--r--spec/models/public_body_spec.rb14
1 files changed, 14 insertions, 0 deletions
diff --git a/spec/models/public_body_spec.rb b/spec/models/public_body_spec.rb
index 23842ccff..d1e2e233d 100644
--- a/spec/models/public_body_spec.rb
+++ b/spec/models/public_body_spec.rb
@@ -473,6 +473,20 @@ describe PublicBody, " when loading CSV files" do
PublicBody.count.should == original_count
end
+
+ it "should be able to load CSV from a file as well as a string" do
+ # Essentially the same code is used for import_csv_from_file
+ # as import_csv, so this is just a basic check that
+ # import_csv_from_file can load from a file at all. (It would
+ # be easy to introduce a regression that broke this, because
+ # of the confusing change in behaviour of CSV.parse between
+ # Ruby 1.8 and 1.9.)
+ original_count = PublicBody.count
+ filename = file_fixture_name('fake-authority-type-with-field-names.csv')
+ PublicBody.import_csv_from_file(filename, '', 'replace', false, 'someadmin')
+ PublicBody.count.should == original_count + 3
+ end
+
end
describe PublicBody do