X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=code_ruby%2Flib%2Fmaasha%2Ffastq.rb;h=8789899e50d796406657b10727252c32f5f6479f;hb=eb99c8692714ae906908545a7ff8f59f0d9575e9;hp=e2a96ea86c69bf2f97712c2c0b6fa7c73995b4cd;hpb=178225ead000ff26d6e76a654fceb10000bfc87d;p=biopieces.git

diff --git a/code_ruby/lib/maasha/fastq.rb b/code_ruby/lib/maasha/fastq.rb
index e2a96ea..8789899 100644
--- a/code_ruby/lib/maasha/fastq.rb
+++ b/code_ruby/lib/maasha/fastq.rb
@@ -28,25 +28,81 @@ require 'maasha/filesys'
 # Error class for all exceptions to do with FASTQ.
 class FastqError < StandardError; end
 
+# Class for parsing FASTQ entries from an ios and return as Seq objects.
 class Fastq < Filesys
-  # Method to get the next FASTQ entry form an ios and return this
+  # Method to get the next FASTQ entry from an ios and return this
   # as a Seq object. If no entry is found or eof then nil is returned.
   def get_entry
-    begin
-      seq_name       = @io.gets.chomp!
-      seq            = @io.gets.chomp!
-      qual_name      = @io.gets.chomp!
-      qual           = @io.gets.chomp!
-
-      entry          = Seq.new
-      entry.seq      = seq
-      entry.seq_name = seq_name[1 .. seq_name.length]
-      entry.qual     = qual
-      entry.type     = nil
-
-      entry
-    rescue
-      nil
+    seq_name       = @io.gets.chomp!
+    seq            = @io.gets.chomp!
+    @io.gets
+    qual           = @io.gets.chomp!
+
+    entry          = Seq.new
+    entry.seq      = seq
+    entry.seq_name = seq_name[1 .. seq_name.length]
+    entry.qual     = qual
+    entry.type     = nil
+
+    entry
+  rescue
+    nil
+  end
+end
+
+# Class for indesing FASTQ entries. The index will be
+# a hash with the FASTQ sequence name as key and a 
+# FastqElem as value. The latter contains info on
+# byte offset and length for each entry.
+class FastqIndex
+  HEADCHAR = 1
+  NEWLINE  = 1
+
+  attr_accessor :ios
+
+  # Method to initialize a FastqIndex object. For reading
+  # entries from file an _ios_ object must be supplied.
+  def initialize(ios = nil)
+    @ios    = ios
+    @index  = {}
+    @offset = 0
+  end
+
+  # Method to add a Fastq entry to a FastqIndex.
+  def add(entry)
+    offset_seq  = @offset + HEADCHAR + entry.seq_name.length + NEWLINE
+    offset_qual = @offset + HEADCHAR + entry.seq_name.length + NEWLINE + entry.length + NEWLINE + HEADCHAR + NEWLINE
+
+    @index[entry.seq_name] = FastqElem.new(offset_seq, offset_qual, entry.length)
+    
+    @offset += HEADCHAR + entry.seq_name.length + NEWLINE + entry.length + NEWLINE + HEADCHAR + NEWLINE + entry.length + NEWLINE
+  end
+
+  # Method to read from file a Fastq entry from an indexed position,
+  # and return the entry as a Seq object.
+  def get(seq_name)
+    raise FastqError, "Sequence name: #{seq_name} not found in index." unless @index[seq_name]
+
+    elem = @index[seq_name]
+    @ios.sysseek(elem.offset_seq)
+    seq = @ios.sysread(elem.length)
+    @ios.sysseek(elem.offset_qual)
+    qual = @ios.sysread(elem.length)
+
+    Seq.new(seq_name, seq, nil, qual)
+  end
+
+  private
+
+  # Class for storing index information to be used
+  # with disk based index.
+  class FastqElem
+    attr_reader :offset_seq, :offset_qual, :length
+
+    def initialize(offset_seq, offset_qual, length)
+      @offset_seq  = offset_seq
+      @offset_qual = offset_qual
+      @length      = length
     end
   end
 end