Merge pull request #6430 from CartoDB/6421-check_in2csv_input

Raise exception when CSV file with XLS extension is detected

Makefile

@@ -96,6 +96,7 @@ WORKING_SPECS_2 = \
   services/importer/spec/unit/csv_normalizer_spec.rb \
 	services/importer/spec/unit/shp_normalizer_spec.rb \
 	services/importer/spec/unit/shp_helper_spec.rb \
+  services/importer/spec/unit/excel2csv_spec.rb \
   services/importer/spec/unit/downloader_spec.rb \
   services/importer/spec/unit/georeferencer_spec.rb \
   services/importer/spec/unit/json2csv_spec.rb \

services/importer/lib/importer/excel2csv.rb

@@ -32,13 +32,12 @@ module CartoDB
         job.log "Converting #{@format.upcase} to CSV"
 
         Open3.popen3("file -b --mime-type #{filepath}") do |stdin, stdout, stderr, process|
-          @file_mime_type = stdout.read.delete("\n")
+          file_mime_type = stdout.read.delete("\n")
           job.log "Can't get the mime type of the file" unless process.value.to_s =~ /exit 0/
+          # CSV files with XLS extensions are considered malformed files
+          raise CartoDB::Importer2::MalformedXLSException.new if file_mime_type == "text/plain"
         end
-
-        # Take into account that here should come or csv files with xls extensions or xls documents
-        file_format = (@file_mime_type == "text/plain") ? "-f csv" : ""
-        in2csv_command_line = %Q[in2csv #{filepath} #{file_format} | #{in2csv_warning_filter} | #{newline_remover_path} > #{converted_filepath}]
+        in2csv_command_line = %Q[in2csv #{filepath} | #{in2csv_warning_filter} | #{newline_remover_path} > #{converted_filepath}]
         job.log "About to execute in2csv: " + in2csv_command_line
         Open3.popen3(in2csv_command_line) do |stdin, stdout, stderr, process|
           raise CartoDB::Importer2::MalformedXLSException.new if stderr.read =~ /Unsupported format, or corrupt file:/

services/importer/spec/unit/excel2csv_spec.rb

@@ -31,17 +31,17 @@ describe CartoDB::Importer2::Excel2Csv do
         @excel2csv.stubs(:converted_filepath).returns("/tmp")
         @excel2csv.run
       end
-      it "if a csv file is passed as xls should parse it" do
+      it "raise if a csv file is passed as xls" do
         filepath = path_to('csv_as_xls.xls')
         @excel2csv      = CartoDB::Importer2::Excel2Csv.new("xls", filepath, @job, @csv_normalizer)
         @excel2csv.stubs(:converted_filepath).returns("/tmp")
-        @excel2csv.run
+        expect { @excel2csv.run }.to raise_error CartoDB::Importer2::MalformedXLSException
       end
-      it "if a csv file is passed as xlsx should parse it" do
-        filepath = path_to('csv_as_xls.xlsx')
+      it "raise if a csv file is passed as xlsx" do
+        filepath = path_to('csv_as_xlsx.xlsx')
         @excel2csv      = CartoDB::Importer2::Excel2Csv.new("xlsx", filepath, @job, @csv_normalizer)
         @excel2csv.stubs(:converted_filepath).returns("/tmp")
-        @excel2csv.run
+        expect { @excel2csv.run }.to raise_error CartoDB::Importer2::MalformedXLSException
       end
     end
   end