gbrowseCapture

#! /usr/bin/ruby

# TheParkerLab
# Vivek Rai
# GPLv3
#
# August ??, 2018


require 'optparse'
require 'ostruct'


options = OpenStruct.new

options.png = false


parseargs = OptionParser.new do |opts|
  opts.version = 1.0

  opts.banner = <<~BANNER

  USAGE: gbrowse_capture [URL] [options] (requires cURL)

  Download PDF from UCSC Genome Browser for the corresponding URL. Simply copy
  and paste the URL corresponding to current view of your Genome Browser
  session.

  Optionally convert the output to `PNG` (requires convert).

  See -h or --help for more options.

  BANNER

  opts.on('-o', '--output', "=FILE", :REQUIRED,  "Output File") do |a|
    options.out = a
    options.prefix = File.basename(a, File.extname(a))
  end

  opts.on('--png', "Convert to PNG? (--output must be specified)") do |v|
    options.png = v
  end

  opts.on_tail("-h", "--help", "Show this message") do
    puts opts
    exit
  end
end
  
pargs = parseargs.parse!

url = pargs.first || nil

unless url
  parseargs.parse('-h')
end

STDERR.syswrite "Rendering Genome Browser PDF.."

hgsid = url.split('=').last

# Retrieve the URI of rendered PDF from the UCSC Genome Browser website and
# downloades silently using cURL.
# 
# Each Genome Browser Session has an unique `hgsid`.
COMMAND = <<-CMD
 curl -s 'https://genome.ucsc.edu/cgi-bin/hgTracks?hgsid=#{hgsid}&hgt.psOutput=on' \
  | grep -Eo 'hgt_genome_.*.pdf'                                                   \
  | xargs -I{} curl -s -o #{options.out} 'https://genome.ucsc.edu/trash/hgt/{}'
CMD

STDOUT.syswrite `#{COMMAND}`

# Convert to high quality PNG;
# TODO: Is further high quality possible?
if options.out and options.png
  STDERR.syswrite "Converting to PNG.."
  `which convert && convert -density 600 -quality 100 -flatten -trim #{options.out} #{options.prefix}.png`
end