worked on unit tests for sam.rb

[biopieces.git] / code_ruby / lib / maasha / sam.rb

diff --git a/code_ruby/lib/maasha/sam.rb b/code_ruby/lib/maasha/sam.rb
index 67e950c082347425865b3745dc5cb8f2a4a5d95d..df518e65d281068c42dcd60394d2bb80171eab30 100644 (file)
--- a/code_ruby/lib/maasha/sam.rb
+++ b/code_ruby/lib/maasha/sam.rb
@@ -38,19 +38,33 @@ REGEX_COMMENT = Regexp.new(/^@CO\t.*/)
 
 # Class to parse and write SAM files.
 class Sam < Filesys
-  attr_accessor :io
+  attr_accessor :io, :header
 
   # Method to initialize a Sam object.
   def initialize(io = nil)
-    @io          = io
-    @header_hash = {}
+    @io     = io
+    @header = {}
+
+    parse_header
   end
 
-  # Method to parse the header of a SAM file.
+  def each
+    @io.each_line do |line|
+      unless line[0] == '@'
+        entry = parse_alignment(line.chomp)
+
+        yield entry if block_given?
+      end
+    end
+  end
+
+  private
+
+  # Method to parse the header section of a SAM file.
   # Each header line should match:
   # /^@[A-Za-z][A-Za-z](\t[A-Za-z][A-Za-z0-9]:[ -~]+)+$/ or /^@CO\t.*/.
   # Tags containing lowercase letters are reserved for end users.
-  def header
+  def parse_header
     @io.each_line do |line|
       if line =~ /^@([A-Za-z][A-Za-z])/
         line.chomp!
@@ -58,11 +72,11 @@ class Sam < Filesys
         tag = $1
 
         case tag
-        when 'HD' then parse_header(line)
-        when 'SQ' then parse_sequence(line)
-        when 'RG' then parse_read_group(line)
-        when 'PG' then parse_program(line)
-        when 'CO' then parse_comment(line)
+        when 'HD' then subparse_header(line)
+        when 'SQ' then subparse_sequence(line)
+        when 'RG' then subparse_read_group(line)
+        when 'PG' then subparse_program(line)
+        when 'CO' then subparse_comment(line)
         else
           raise SamError, "Unknown header tag: #{tag}"
         end
@@ -72,23 +86,11 @@ class Sam < Filesys
       end
     end
 
-    return @header_hash.empty? ? nil : @header_hash
-  end
-
-  def each
-    @io.each_line do |line|
-      unless line[0] == '@'
-        entry = parse_alignment(line.chomp)
-
-        yield entry if block_given?
-      end
-    end
+    return @header.empty? ? nil : @header
   end
 
-  private
-
   # Method to subparse header lines.
-  def parse_header(line)
+  def subparse_header(line)
     hash   = {}
     fields = line.split("\t")
 
@@ -106,12 +108,12 @@ class Sam < Filesys
       end
     end
 
-    @header_hash[:HD] = hash
+    @header[:HD] = hash
   end
 
   # Method to subparse sequence lines.
-  def parse_sequence(line)
-    @header_hash[:SQ] = Hash.new unless @header_hash[:SQ].is_a? Hash
+  def subparse_sequence(line)
+    @header[:SQ] = Hash.new unless @header[:SQ].is_a? Hash
     hash = {}
 
     fields = line.split("\t")
@@ -136,18 +138,18 @@ class Sam < Filesys
       end
     end
 
-    @header_hash[:SQ][:SN] = Hash.new unless @header_hash[:SQ][:SN].is_a? Hash
+    @header[:SQ][:SN] = Hash.new unless @header[:SQ][:SN].is_a? Hash
 
-    if @header_hash[:SQ][:SN].has_key? seq_name
+    if @header[:SQ][:SN].has_key? seq_name
       raise SamError, "Non-unique sequence name: #{seq_name}"
     else
-      @header_hash[:SQ][:SN][seq_name] = hash
+      @header[:SQ][:SN][seq_name] = hash
     end
   end
 
   # Method to subparse read group lines.
-  def parse_read_group(line)
-    @header_hash[:RG] = Hash.new unless @header_hash[:RG].is_a? Hash
+  def subparse_read_group(line)
+    @header[:RG] = Hash.new unless @header[:RG].is_a? Hash
     hash = {}
 
     fields = line.split("\t")
@@ -178,18 +180,18 @@ class Sam < Filesys
       end
     end
 
-    @header_hash[:RG][:ID] = Hash.new unless @header_hash[:RG][:ID].is_a? Hash
+    @header[:RG][:ID] = Hash.new unless @header[:RG][:ID].is_a? Hash
 
-    if @header_hash[:RG][:ID].has_key? id
+    if @header[:RG][:ID].has_key? id
       raise SamError, "Non-unique read group identifier: #{id}"
     else
-      @header_hash[:RG][:ID][id] = hash
+      @header[:RG][:ID][id] = hash
     end
   end
 
   # Method to subparse program lines.
-  def parse_program(line)
-    @header_hash[:PG] = Hash.new unless @header_hash[:PG].is_a? Hash
+  def subparse_program(line)
+    @header[:PG] = Hash.new unless @header[:PG].is_a? Hash
     hash = {}
 
     fields = line.split("\t")
@@ -208,21 +210,21 @@ class Sam < Filesys
       end
     end
 
-    @header_hash[:PG][:ID] = Hash.new unless @header_hash[:PG][:ID].is_a? Hash
+    @header[:PG][:ID] = Hash.new unless @header[:PG][:ID].is_a? Hash
 
-    if @header_hash[:PG][:ID].has_key? id
+    if @header[:PG][:ID].has_key? id
       raise SamError, "Non-unique program record identifier: #{id}"
     else
-      @header_hash[:PG][:ID][id] = hash
+      @header[:PG][:ID][id] = hash
     end
   end
 
   # Method to subparse comment lines.
-  def parse_comment(line)
-    @header_hash[:CO] = Array.new unless @header_hash[:CO].is_a? Array
+  def subparse_comment(line)
+    @header[:CO] = Array.new unless @header[:CO].is_a? Array
 
     if line =~ /^@CO\t(.+)/
-      @header_hash[:CO] << $1
+      @header[:CO] << $1
     else
       raise SamError, "Bad comment line: #{line}"
     end
@@ -258,6 +260,8 @@ class Sam < Filesys
     raise SamError, "Bad seq: #{seq}"     unless seq  =~ /^(\*|[A-Za-z=.]+)$/
     raise SamError, "Bad qual: #{qual}"   unless qual =~ /^[!-~]+$/
 
+    check_rname(rname)
+
     entry = {}
     entry[:QNAME] = qname
     entry[:FLAG]  = flag
@@ -273,6 +277,17 @@ class Sam < Filesys
 
     entry
   end
+
+  # Method to check if rname, when not '*' and
+  # @SQ header lines are present, is located in
+  # the header hash.
+  def check_rname(rname)
+    unless @header.empty? or rname == '*'
+      unless @header[:SQ][:SN].has_key? rname.to_sym
+        raise SamError, "rname not found in header hash: #{rname}"
+      end
+    end
+  end
 end