Refactoring splitter

README.md

@@ -81,7 +81,7 @@ headers may need to be installed first, for example on Debian:
 ```
   blastxmlparser [options] file(s)
 
-    -p, --parser name                Use full|split parser (default full)
+    -p, --parser name                Use split|nosplit parser (default split)
         --filter filter              Filtering expression
         --threads num                Use parallel threads
     -e, --exec filter                Evaluate filter (deprecated)
@@ -229,10 +229,10 @@ Likewise, using the RDF template
 
 ## Additional options
 
-To use the low-mem (iterated slower) version of the parser use
+To use the high-mem version of the parser (slightly faster on single core) use
 
 ```sh
-  blastxmlparser --parser split -n 'hsp.evalue,hsp.qseq' --filter 'hsp.evalue<0.01 and hit.len>100' test/data/nt_example_blastn.m7
+  blastxmlparser --parser nosplit --threads 1 -n 'hsp.evalue,hsp.qseq' --filter 'hsp.evalue<0.01 and hit.len>100' test/data/nt_example_blastn.m7
 ```
 
 ## API (Ruby library)

bin/blastxmlparser

@@ -48,7 +48,7 @@ opts = OptionParser.new do |o|
 
   o.separator ""
 
-  o.on("-p name", "--parser name", "Use full|split parser (default full)") do |p|
+  o.on("-p name", "--parser name", "Use split|nosplit parser (default split)") do |p|
     options.parser = p.to_sym
   end
 
@@ -127,10 +127,11 @@ begin
 
   ARGV.each do | fn |
     logger.info("XML parsing #{fn}")
-    n = if options.parser == :split
-      Bio::BlastXMLParser::XmlSplitterIterator.new(fn).to_enum
+    n = if options.parser == :nosplit
+      NokogiriBlastXml.new(File.new(@fn)).to_enum
     else
-      Bio::BlastXMLParser::XmlIterator.new(fn).to_enum
+      # default
+      Bio::BlastXMLParser::XmlSplitterIterator.new(fn).to_enum
     end
     chunks = []
     chunks_count = 0

lib/bio/db/blast/xmliterator.rb

@@ -11,7 +11,7 @@ def initialize blastfilename
 
       def to_enum
         logger = Bio::Log::LoggerPlus['bio-blastxmlparser']
-        logger.info("parsing (full) #{@fn}")
+        logger.info("parsing (:nosplit) #{@fn}")
         NokogiriBlastXml.new(File.new(@fn)).to_enum
       end
     end

lib/bio/db/blast/xmlsplitter.rb

@@ -4,27 +4,21 @@ module Bio
   module BlastXMLParser
     # Reads a full XML result and splits it out into a buffer for each
     # Iteration (query result).
-    class XmlSplitterIterator
-      # include Enumerable
-
+    class BlastXmlSplitter
       def initialize fn
         @fn = fn
       end
-
-      def to_enum 
-        Enumerator.new do | yielder | 
-          logger = Bio::Log::LoggerPlus['bio-blastxmlparser']
-          logger.info("split file parsing #{@fn}")
-          f = File.open(@fn)
-          # Skip BLAST header
-          f.each_line do | line |
-            break if line.strip == "<Iteration>"
-          end
-          # Return each Iteration as an XML DOM
-          each_iteration(f) do | buf |
-            iteration = Nokogiri::XML.parse(buf.join) { | cfg | cfg.noblanks }
-            yielder.yield NokogiriBlastIterator.new(iteration,self,:prefix=>nil)
-          end
+      def each
+        logger = Bio::Log::LoggerPlus['bio-blastxmlparser']
+        logger.info("split file parsing #{@fn}")
+        f = File.open(@fn)
+        # Skip BLAST header
+        f.each_line do | line |
+          break if line.strip == "<Iteration>"
+        end
+        # Return each Iteration as an XML DOM
+        each_iteration(f) do | buf |
+          yield buf
         end
       end
 
@@ -43,5 +37,22 @@ def each_iteration f
         end
       end
     end
+
+    class XmlSplitterIterator
+      # include Enumerable
+
+      def initialize fn
+        @splitter = BlastXmlSplitter.new(fn)
+      end
+
+      def to_enum 
+        Enumerator.new do | yielder | 
+          @splitter.each do | buf |
+            iteration = Nokogiri::XML.parse(buf.join) { | cfg | cfg.noblanks }
+            yielder.yield NokogiriBlastIterator.new(iteration,self,:prefix=>nil)
+          end
+        end
+      end
+    end
   end
 end