Sha256: 8c341a53d422b074ca3d30a6ea3483b232913f919fed9815f187840204329439
Contents?: true
Size: 1.59 KB
Versions: 217
Compression:
Stored size: 1.59 KB
Contents
#!/usr/bin/env ruby
require 'rbbt-util'
require 'rbbt/util/simpleopt'
options = SOPT.setup <<EOF
Read a TSV file and dump it
$ rbbt tsv read [options] <filename.tsv|->
Use - to read from STDIN
-k--key_field* Key field
-f--fields* Fields
-t--type* Type
-m--merge Merge from multiple rows
-h--help Print this help
-tch--tokyocabinet File is a tokyocabinet hash database
-tcb--tokyocabinet_bd File is a tokyocabinet B database
EOF
rbbt_usage and exit 0 if options[:help]
file = ARGV.shift
file = STDIN if file == '-'
case
when options[:tokyocabinet]
tsv = Persist.open_tokyocabinet(file, false)
puts tsv.to_s
exit
when options[:tokyocabinet_bd]
tsv = Persist.open_tokyocabinet(file, false, nil, TokyoCabinet::BDB)
puts tsv.to_s
exit
end
parser = TSV::Parser.new TSV.get_stream(file), options
dumper = TSV::Dumper.new parser.options.merge(:sep => "\t", :sep2 => '|')
dumper.init
options[:merge] = false if options[:merge] == "false"
Thread.new do
line = parser.first_line
bar = Log::ProgressBar.new
while line
bar.tick
line = Misc.fixutf8(line)
line = parser.process line
raise SKIP_LINE if line.empty?
parts = parser.chop_line line
key, values = parser.get_values parts
values = parser.cast_values values if parser.cast?
if Array === key
key.each do |k|
dumper.add k, values
end
line = parser.stream.gets
else
dumper.add key, values
line = parser.stream.gets
end
end
dumper.close
end
stream = dumper.stream
stream = TSV.collapse_stream(dumper.stream).stream if options[:merge]
while line = stream.gets
puts line
end
Version data entries
217 entries across 217 versions & 1 rubygems