X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=code_ruby%2Flib%2Fmaasha%2Fbiopieces.rb;h=5fd6edf4a505154bb929976a86c58246ac5d3f85;hb=0153863ce54bf865f397a1c4b6a45b0180d71e31;hp=ca235318c815ea3021c65c74c1ad8ffd58155e55;hpb=2944deebd4724b87f68388f4178ce127aec7ce56;p=biopieces.git

diff --git a/code_ruby/lib/maasha/biopieces.rb b/code_ruby/lib/maasha/biopieces.rb
index ca23531..5fd6edf 100644
--- a/code_ruby/lib/maasha/biopieces.rb
+++ b/code_ruby/lib/maasha/biopieces.rb
@@ -1,6 +1,6 @@
 raise "Ruby 1.9 or later required" if RUBY_VERSION < "1.9"
 
-# Copyright (C) 2007-2010 Martin A. Hansen.
+# Copyright (C) 2007-2011 Martin A. Hansen.
 
 # This program is free software; you can redistribute it and/or
 # modify it under the terms of the GNU General Public License
@@ -24,11 +24,34 @@ raise "Ruby 1.9 or later required" if RUBY_VERSION < "1.9"
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
-require 'fileutils'
 require 'date'
+require 'fileutils'
 require 'optparse'
-require 'open3'
 require 'pp'
+require 'stringio'
+require 'zlib'
+
+TEST = false
+
+# Monkey patch (evil) changing the to_s method of the Hash class to
+# return a Hash in Biopieces format; keys and value pairs on one line
+# each seperated with ': ' and terminated by a line of '---'.
+class Hash
+  def to_s
+    string = ""
+
+    self.each do |key, value|
+      string << "#{key.to_s}: #{value}\n"
+    end
+
+    string << "---\n"
+
+    string
+  end
+end
+
+# Error class for all exceptions to do with the Biopieces class.
+class BiopiecesError < StandardError; end
 
 # Biopieces are command line scripts and uses OptionParser to parse command line
 # options according to a list of casts. Each cast prescribes the long and short
@@ -42,37 +65,67 @@ require 'pp'
 class Biopieces
   include Enumerable
 
-  attr_accessor :out # accessor for out stream _ios_
+  # Class method to check the integrity of a list of casts, followed by parsing
+  # options from argv and finally checking the options according to the casts.
+  def self.options_parse(argv, cast_list=[], script_path=$0)
+    casts          = Casts.new(cast_list)
+    option_handler = OptionHandler.new(argv, casts, script_path)
+    options        = option_handler.options_parse
 
-  # Initialize a Biopiece and write the status to file.
-  # Options are for testing purposes only.
-  def initialize(test=nil, input=STDIN, output=STDOUT)
-    @test   = test
-    @input  = input
-    @output = output
+    options
   end
 
-  # Check the integrity of a list of casts, followed by parsion options from argv
-  # and finally checking the options according to the casts. Returns nil if argv
-  # is empty, otherwise an options hash.
-  def parse(argv, cast_list=[], script_path=$0)
-    casts          = Casts.new(cast_list)
-    option_handler = OptionHandler.new(argv, casts, script_path, @test)
-    @options       = option_handler.options_parse
+  # Class method for opening data streams for reading and writing Biopiece
+  # records. Records are read from STDIN (default) or file (possibly gzipped)
+  # and written to STDOUT (default) or file.
+  def self.open(input = STDIN, output = STDOUT)
+    input  = self.open_input(input)
+    output = self.open_output(output)
 
-    @in = Stream.open(@options, mode="r", @input)
+    if block_given?
+      begin
+        yield input, output
+      ensure
+        input.close
+        output.close
+      end
+    else
+      return input, output
+    end
+  end
 
-    @options
+  # Class method to create a temporary directory inside the ENV["BP_TMP"] directory.
+  def self.mktmpdir
+    time = Time.now.to_i
+    user = ENV["USER"]
+    pid  = $$
+    path = File.join(ENV["BP_TMP"], [user, time + pid, pid, "bp_tmp"].join("_"))
+    Dir.mkdir(path)
+    Status.new.set_tmpdir(path)
+    path
+  end
+
+  # Initialize a Biopiece object for either reading or writing from _ios_.
+  def initialize(ios, stdio = nil)
+    @ios   = ios
+    @stdio = stdio
+  end
+
+  # Method to write a Biopiece record to _ios_.
+  def puts(foo)
+    @ios << foo.to_s
+  end
+
+  # Method to close _ios_.
+  def close
+    @ios.close unless @stdio
   end
 
-  # Open Biopiece input stream if not open and iterate over all Biopiece
-  # records in the stream.
+  # Method to parse and yield a Biopiece record from _ios_.
   def each_record
-    return if @in.nil?
-    
     record = {}
 
-    @in.each_line do |line|
+    @ios.each_line do |line|
       case line
       when /^([^:]+): (.*)$/
         record[$1.to_sym] = $2
@@ -80,7 +133,7 @@ class Biopieces
         yield record unless record.empty?
         record = {}
       else
-        raise "Bad record format: #{line}"
+        raise BiopiecesError, "Bad record format: #{line}"
       end
     end
 
@@ -91,29 +144,42 @@ class Biopieces
 
   alias :each :each_record
 
-  # Open Biopiece output stream if not open and puts record to the stream.
-  def puts(record)
-    @out = Stream.open(@options, mode="w", @output) unless @out.is_a? IO
+  private
+
+  # Class method for opening data stream for reading Biopiece records.
+  # Records are read from STDIN (default) or file (possibly gzipped).
+  def self.open_input(input)
+    if input.nil?
+      if STDIN.tty?
+        input = self.new(StringIO.new)
+      else
+        input = self.new(STDIN, stdio = true)
+      end
+    elsif File.exists? input
+      ios = File.open(input, mode='r')
+
+      begin
+          ios = Zlib::GzipReader.new(ios)
+      rescue
+          ios.rewind
+      end
 
-    record.each do |key,value|
-      @out.print "#{key.to_s}: #{value}\n"
+      input = self.new(ios)
     end
 
-    @out.print "---\n"
+    input
   end
 
-  def to_s
-  end
+  # Class method for opening data stream for writing Biopiece records.
+  # Records are written to STDOU (default) or file.
+  def self.open_output(output)
+    if output.nil?
+      output = self.new(STDOUT, stdio = true)
+    elsif not output.is_a? IO
+      output = self.new(File.open(output, mode='w'))
+    end
 
-  # Create a temporary directory inside the ENV["BP_TMP"] dir.
-  def mktmpdir
-    time = Time.now.to_i
-    user = ENV["USER"]
-    pid  = $$
-    path = ENV["BP_TMP"] + "/" + [user, time + pid, pid, "bp_tmp"].join("_")
-    Dir.mkdir(path)
-    Status.new.set_tmpdir(path)
-    path
+    output
   end
 end
 
@@ -121,7 +187,6 @@ end
 # Error class for all exceptions to do with option casts.
 class CastError < StandardError; end
 
-
 # Class to handle casts of command line options. Each cast prescribes the long and
 # short name of the option, the type, if it is mandatory, the default value, and
 # allowed and disallowed values. An optional list of extra casts can be supplied,
@@ -145,7 +210,7 @@ class Casts < Array
   # Add ubiquitous options casts.
   def ubiquitous
     @cast_list << {:long=>'help',       :short=>'?', :type=>'flag',   :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
-    @cast_list << {:long=>'stream_in',  :short=>'I', :type=>'files!', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
+    @cast_list << {:long=>'stream_in',  :short=>'I', :type=>'file!',  :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
     @cast_list << {:long=>'stream_out', :short=>'O', :type=>'file',   :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
     @cast_list << {:long=>'verbose',    :short=>'v', :type=>'flag',   :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
   end
@@ -253,7 +318,6 @@ class Casts < Array
       cast[:long] = cast[:long].to_sym
     end
   end
-
 end
 
 
@@ -266,11 +330,10 @@ class OptionHandler
   REGEX_INT    = /^(int|uint)$/
   REGEX_STRING = /^(file|file!|dir|dir!|genome)$/
 
-  def initialize(argv, casts, script_path, test=nil)
+  def initialize(argv, casts, script_path)
     @argv        = argv
     @casts       = casts
     @script_path = script_path
-    @test        = test
   end
 
   # Parse options from argv using OptionParser and casts denoting long and
@@ -315,7 +378,7 @@ class OptionHandler
     option_parser.parse!(@argv)
 
     if print_usage_full?
-      print_usage_and_exit(true)
+      print_usage_and_exit(full=true)
     elsif print_usage_short?
       print_usage_and_exit
     end
@@ -341,7 +404,7 @@ class OptionHandler
 
   # Check if short "usage info" should be printed.
   def print_usage_short?
-    if not $stdin.tty?
+    if not STDIN.tty?
       return false
     elsif @options[:stream_in]
       return false
@@ -358,7 +421,7 @@ class OptionHandler
   # using a system() call and exit. An optional 'full' flag
   # outputs the full usage info.
   def print_usage_and_exit(full=nil)
-    if @test
+    if TEST
       return
     else
       if full
@@ -547,7 +610,7 @@ class Status
 
     elap     = time_diff(time0, time1)
     command  = [script, args].join(" ") 
-    log_file = ENV["BP_LOG"] + "/biopieces.log"
+    log_file = File.join(ENV["BP_LOG"], "biopieces.log")
 
     File.open(log_file, mode = "a") { |file| file.puts [time0, time1, elap, user, exit_status, command].join("\t") }
   end
@@ -564,7 +627,7 @@ class Status
     user   = ENV["USER"]
     script = File.basename($0)
     pid    = $$
-    path   = ENV["BP_TMP"] + "/" + [user, script, pid, "status"].join(".")
+    path   = File.join(ENV["BP_TMP"], [user, script, pid, "status"].join("."))
   end
 
   # Get the elapsed time from the difference between two time stamps.
@@ -574,87 +637,6 @@ class Status
 end
 
 
-class Stream < IO
-  # Open Biopieces output data stream for reading from stdin or a file
-  # specified in options[:stream_in] OR writing to stdout or a file
-  # specified in options[:stream_out] or options[:data_out].
-  def self.open(options, mode, stdio)
-    if mode == "r"
-      if options[:data_in] and options[:data_in].first == "-"
-        self.nread(["-"])
-      else
-        $stdin.tty? ? read(options[:stream_in]) : stdio
-      end
-    elsif mode == "w"
-      options[:stream_out] ? self.write(options[:stream_out], options[:compress]) : stdio
-    else
-      raise "Bad mode #{mode}"
-    end
-  end
-
-  private
-
-  # Opens a reads stream to a list of files.
-  def self.read(files)
-    return if files.nil? #TODO case/when
-    self.zipped?(files) ? self.zread(files) : self.nread(files)
-  end
-
-  # Opens a write stream to a file and returns a _io_ object.
-  def self.write(file, zip=nil)
-    zip ? self.zwrite(file) : self.nwrite(file)
-  end
-
-  # Opens a list of gzipped files for reading and return an _io_ object.
-  def self.zread(files)
-    stdin, stdout, stderr = Open3.popen3("zcat " + files.join(' '));
-    stdin.close
-    stderr.close
-    stdout
-  end
-
-  # Opens a file for gzipped writing and return an _io_ object.
-  def self.zwrite(file)
-    stdin, stdout, stderr = Open3.popen3("gzip -f > #{file}")
-    stderr.close
-    stdout.close
-    stdin
-  end
-
-  # Opens a list of files for reading and return an _io_ object.
-  def self.nread(files)
-    stdin, stdout, stderr = Open3.popen3("cat " + files.join(' '));
-    stdin.close
-    stderr.close
-    stdout
-  end
-
-  # Opens a file for writing and return an _io_ object.
-  def self.nwrite(file)
-    File.open(file, mode="w")
-  end
-
-  # Test if a list of files are gzipped or not.
-  # Raises if files are mixed zipped and unzipped.
-  def self.zipped?(files)
-    type_hash = {}
-
-    files.each do |file|
-      type = `file #{file}`
-
-      if type =~ /gzip compressed/
-        type_hash[:gzip] = true
-      else
-        type_hash[:ascii] = true
-      end
-    end
-
-    raise "Mixture of zipped and unzipped files" if type_hash.size == 2
-
-    type_hash[:gzip]
-  end
-end
-
 # Set status when 'biopieces' is required.
 Status.new.set
 
@@ -682,3 +664,4 @@ end
 
 
 __END__
+