Sha256: 09261befd6c586ad51ab66ab36ade46a6d91e97669d2c393819e9fa8798ced17
Contents?: true
Size: 1.36 KB
Versions: 337
Compression:
Stored size: 1.36 KB
Contents
#!/usr/bin/env ruby
require 'rbbt-util'
require 'rbbt/util/simpleopt'
$0 = "rbbt #{$previous_commands*""} #{ File.basename(__FILE__) }" if $previous_commands
options = SOPT.setup <<EOF
Subset entries in a tsv
$ rbbt tsv subset [options] file.tsv --subset key1,key2,key3
Subsets entries from a TSV file from a given list. Works with Tokyocabinet HDB and BDB as well.
-tch--tokyocabinet File is a TC HDB
-tcb--tokyocabinet_bd File is a TC BDB
-hh--header_hash* Change the character used to mark the header line (defaults to #)
-s--subset* Subset of keys (Comma-separated or file)
-m--merge Merge TSV rows
-h--help Help
EOF
SOPT.usage if options[:help]
file = ARGV.shift
file = STDIN if file == '-'
raise ParameterException, "Please specify the tsv file as argument" if file.nil?
subset = options[:subset]
raise ParameterException, "Please specify a subset of keys" if subset.nil?
if File.exist?(subset)
subset = Open.read(subset).split("\n")
else
subset = subset.split(',')
end
options[:fields] = options[:fields].split(/,\|/) if options[:fields]
options[:header_hash] = options["header_hash"]
case
when options[:tokyocabinet]
tsv = Persist.open_tokyocabinet(file, false)
puts tsv.summary
when options[:tokyocabinet_bd]
tsv = Persist.open_tokyocabinet(file, false, nil, TokyoCabinet::BDB)
puts tsv.summary
else
tsv = TSV.open(file, options)
end
puts tsv.select(subset)
Version data entries
337 entries across 337 versions & 1 rubygems