Rake task for removing exif from uploads

Adds a rake task which can be used for removing EXIF
data from existing uploads.

1 files changed, 120 insertions, 0 deletions

diff --git a/spec/lib/gitlab/sanitizers/exif_spec.rb b/spec/lib/gitlab/sanitizers/exif_spec.rb
new file mode 100644
index 00000000000..bd5f330c7a1
--- /dev/null
+++ b/spec/lib/gitlab/sanitizers/exif_spec.rb
@@ -0,0 +1,120 @@
+require 'spec_helper'
+
+describe Gitlab::Sanitizers::Exif do
+  let(:sanitizer) { described_class.new }
+
+  describe '#batch_clean' do
+    context 'with image uploads' do
+      let!(:uploads) { create_list(:upload, 3, :with_file, :issuable_upload) }
+
+      it 'processes all uploads if range ID is not set' do
+        expect(sanitizer).to receive(:clean).exactly(3).times
+
+        sanitizer.batch_clean
+      end
+
+      it 'processes only uploads in the selected range' do
+        expect(sanitizer).to receive(:clean).once
+
+        sanitizer.batch_clean(start_id: uploads[1].id, stop_id: uploads[1].id)
+      end
+
+      it 'pauses if sleep_time is set' do
+        expect(sanitizer).to receive(:sleep).exactly(3).times.with(1.second)
+        expect(sanitizer).to receive(:clean).exactly(3).times
+
+        sanitizer.batch_clean(sleep_time: 1)
+      end
+    end
+
+    it 'filters only jpg/tiff images' do
+      create(:upload, path: 'filename.jpg')
+      create(:upload, path: 'filename.jpeg')
+      create(:upload, path: 'filename.JPG')
+      create(:upload, path: 'filename.tiff')
+      create(:upload, path: 'filename.TIFF')
+      create(:upload, path: 'filename.png')
+      create(:upload, path: 'filename.txt')
+
+      expect(sanitizer).to receive(:clean).exactly(5).times
+      sanitizer.batch_clean
+    end
+  end
+
+  describe '#clean' do
+    let(:uploader) { create(:upload, :with_file, :issuable_upload).build_uploader }
+
+    context "no dry run" do
+      it "removes exif from the image" do
+        uploader.store!(fixture_file_upload('spec/fixtures/rails_sample.jpg'))
+
+        original_upload = uploader.upload
+        expected_args = ["exiftool", "-all=", "-tagsFromFile", "@", *Gitlab::Sanitizers::Exif::EXCLUDE_PARAMS, "--IPTC:all", "--XMP-iptcExt:all", kind_of(String)]
+
+        expect(sanitizer).to receive(:extra_tags).and_return(["", 0])
+        expect(sanitizer).to receive(:exec_remove_exif!).once.and_call_original
+        expect(uploader).to receive(:store!).and_call_original
+        expect(Gitlab::Popen).to receive(:popen).with(expected_args) do |args|
+          File.write("#{args.last}_original", "foo") if args.last.start_with?(Dir.tmpdir)
+
+          [expected_args, 0]
+        end
+
+        sanitizer.clean(uploader, dry_run: false)
+
+        expect(uploader.upload.id).not_to eq(original_upload.id)
+        expect(uploader.upload.path).to eq(original_upload.path)
+      end
+
+      it "ignores image without exif" do
+        expected_args = ["exiftool", "-all", "-j", "-sort", "--IPTC:all", "--XMP-iptcExt:all", kind_of(String)]
+
+        expect(Gitlab::Popen).to receive(:popen).with(expected_args).and_return(["[{}]", 0])
+        expect(sanitizer).not_to receive(:exec_remove_exif!)
+        expect(uploader).not_to receive(:store!)
+
+        sanitizer.clean(uploader, dry_run: false)
+      end
+
+      it "raises an error if the exiftool fails with an error" do
+        expect(Gitlab::Popen).to receive(:popen).and_return(["error", 1])
+
+        expect { sanitizer.clean(uploader, dry_run: false) }.to raise_exception(RuntimeError, "failed to get exif tags: error")
+      end
+    end
+
+    context "dry run" do
+      it "doesn't change the image" do
+        expect(sanitizer).to receive(:extra_tags).and_return({ 'foo' => 'bar' })
+        expect(sanitizer).not_to receive(:exec_remove_exif!)
+        expect(uploader).not_to receive(:store!)
+
+        sanitizer.clean(uploader, dry_run: true)
+      end
+    end
+  end
+
+  describe "#extra_tags" do
+    it "returns a list of keys for exif file" do
+      tags = '[{
+                "DigitalSourceType": "some source",
+                "ImageHeight": 654
+              }]'
+
+      expect(Gitlab::Popen).to receive(:popen).and_return([tags, 0])
+
+      expect(sanitizer.extra_tags('filename')).not_to be_empty
+    end
+
+    it "returns an empty list for file with only whitelisted and ignored tags" do
+      tags = '[{
+                "ImageHeight": 654,
+                "Megapixels": 0.641
+              }]'
+
+      expect(Gitlab::Popen).to receive(:popen).and_return([tags, 0])
+
+      expect(sanitizer.extra_tags('some file')).to be_empty
+    end
+  end
+end