Goofing around with minitar, typhoeus, and progressbar

for purposes of actually working with PostgreSQL and weather data.

postgresql/tutorial/weather/import-data

@@ -1,25 +1,52 @@
-#!/bin/bash
+#!/usr/bin/env ruby
 
-FTP_BASE='ftp://ftp.ncdc.noaa.gov/pub/data/gsod'
+require 'fileutils'
 
-pushd $(dirname $(readlink -f $0))
+require 'bundler/setup'
+require 'archive/tar/minitar'
+require 'progressbar'
+require 'typhoeus'
 
-pushd ./data
-if [[ ! -f ish-history.csv ]]
-then
-    curl -O $FTP_BASE/ish-history.csv
-fi
+FTP_BASE = 'ftp://ftp.ncdc.noaa.gov/pub/data/gsod'
+YEARS = %w(1950 1960 1970 1980 1990 2000 2010)
 
-for year in 1950 1960 1970 1980 1990 2000 2010
-do
-    tarname=gsod_$year.tar
-    if [[ ! -f $tarname ]]
-    then
-        curl -O $FTP_BASE/$year/$tarname
-    fi
-    tar xf $tarname
-    for gzfile in *-$year.op.gz
-    do
-        :
-    done
-done
+def main
+  include Archive::Tar
+  dest_dir = File.expand_path('../data', __FILE__)
+  Dir.chdir(dest_dir)
+
+  download_here_maybe?("#{FTP_BASE}/ish-history.csv")
+
+  YEARS.each do |year|
+    tarname = "gsod_#{year}.tar"
+    url = "#{FTP_BASE}/#{year}/#{tarname}"
+    dest = "#{dest_dir}/#{tarname}"
+    download_here_maybe?(url)
+    to_unpack = Minitar.open(dest, 'r').collect(&:full_name).select do |f|
+      f =~ /\.gz$/ && !File.exist?(f)
+    end
+    if !to_unpack.empty?
+      @progress = ProgressBar.new(File.basename(dest), to_unpack.length)
+      Minitar.unpack(dest, dest_dir, to_unpack) do |action,name,stats|
+        if action == :file_done
+          @progress.inc(1)
+        end
+      end
+      puts
+    end
+  end
+end
+
+def download_here_maybe?(url)
+  outfile = File.basename(url)
+  if !File.exist?(outfile)
+    File.open(outfile, 'w') do |f|
+      puts "Writing #{url} to #{outfile}"
+      f.write(Typhoeus::Request.get(url).body)
+    end
+  end
+end
+
+if $0 == __FILE__
+  main
+end