CSV with Ruby

All about reading and writing CSV with Ruby, including large file handling.

Notes

CSV support is implemented in the CSV standard library. There are some gems that provide additional CVS processing features, but in most cases the stadnard library is just fine.

Writing Large Files

The large_file_write.rb example demonstrates 2 ways of writing large files:

• using CSV.generate to build a CSV data structure that is then written to file
• using CSV.generate_line to generate a file line-by-line

Using CSV.generate, results show a relatively large peak memory use:

$ ./large_file_write.rb generate
writing 2000000 rows using generate...
Max memory usage: 91.15625Mb

Using CSV.generate_line, results show minimal peak memory use:

$ ./large_file_write.rb generate_line
writing 2000000 rows using generate_line...
Max memory usage: 32.765625Mb

Reading Large Files

The large_file_read.rb example demonstrates 2 ways of reading large files.

• using CSV.read to load the CSV file into memory for processing
• using CSV.foreach to process the CSV as an I/O stream, line-by-line.

Note: run large_file_write.rb first to generate the csv file to be read.

Using CSV.read, results show excessive memory usage:

$ ./large_file_read.rb read
reading using read...
#<CSV::Row "col1":"row 1999999" "col2":"1999999">
Max memory usage: 1493.1875Mb

Using CSV.foreach, results show minimal peak memory use:

$ ./large_file_read.rb foreach
reading using foreach...
#<CSV::Row "col1":"row 1999999" "col2":"1999999">
Max memory usage: 37.703125Mb

Credits and References

• stdlibs: CSV
• How To Read & Parse CSV Files With Ruby
• Parsing a CSV File in Ruby
• get_process_mem