Basic memory profiling setup.

This makes it a bit easier to profile memory usage of certain components and
plot them using Gnuplot. In the past I would write one-off scripts for this and
throw them away, only to figure out I needed them again later on.

Profiling samples are written to profile/samples and can be plotted using
corresponding Gnuplot scripts found in profile/plot. The latter requires
Gnuplot to be installed.

.gitignore

@@ -7,3 +7,6 @@ lib/oga/xml/lexer.rb
 lib/oga/xml/parser.rb
 
 benchmark/fixtures/big.xml
+
+profile/samples/*.txt
+profile/samples/lexer/*.txt

profile/lexer/big_xml.rb

@@ -0,0 +1,8 @@
+require_relative '../profile_helper'
+
+xml   = read_big_xml
+lexer = Oga::XML::Lexer.new(xml)
+
+profile_memory('lexer/big_xml')
+
+lexer.advance { |tok| }

profile/plot/lexer/big_xml.gnuplot

@@ -0,0 +1,20 @@
+#!/usr/bin/env gnuplot
+
+set title "Lexing 10MB of XML"
+
+set xlabel "Sample"
+set ylabel "Memory (MB)"
+
+set yrange [0:*]
+
+set term qt persist
+set grid
+
+set style line 1 lc rgb "#0060ad" lt 1 lw 2 pt 7 ps 1
+
+plot "profile/samples/lexer/big_xml.txt" \
+    using 0:($1 / 1024 ** 2) \
+    with linespoints ls 1 \
+    title "Memory"
+
+# vim: set ft=gnuplot:

profile/profile_helper.rb

@@ -0,0 +1,41 @@
+require_relative '../lib/oga'
+
+##
+# Returns memory usage in bytes. This relies on the /proc filesystem, it won't
+# work without it.
+#
+# @return [Fixnum]
+#
+def memory_usage
+  return File.read('/proc/self/status').match(/VmRSS:\s+(\d+)/)[1].to_i * 1024
+end
+
+##
+# Reads a big XML file and returns it as a String.
+#
+# @return [String]
+#
+def read_big_xml
+  return File.read(File.expand_path('../../benchmark/fixtures/big.xml', __FILE__))
+end
+
+##
+# Writes memory samples to a file until the thread is killed.
+#
+# @param [String] name The name of the samples file.
+# @param [Fixnum] interval The sample interval. The default is 200 ms.
+# @return [Thread]
+#
+def profile_memory(name, interval = 0.2)
+  return Thread.new do
+    path        = File.expand_path("../samples/#{name}.txt", __FILE__)
+    handle      = File.open(path, 'w')
+    handle.sync = true
+
+    loop do
+      handle.write("#{memory_usage}\n")
+
+      sleep(interval)
+    end
+  end
+end