From eb9eabad8ad718bb031ff492254125ab90d73ad9 Mon Sep 17 00:00:00 2001
From: Andrew Cholakian <andrew@andrewvc.com>
Date: Mon, 1 Aug 2016 12:14:28 -0500
Subject: [PATCH] Add new match_direct method for optimized use by outside libs

Bump benchmark pattern test to use 10m matches to even out inconsistencies
---
 CHANGELIST                          |  3 +++
 grok.gemspec                        |  2 +-
 lib/grok-pure.rb                    | 23 ++++++++++++++++++-----
 lib/grok/pure/match.rb              |  2 +-
 test/pure-ruby/benchmark_pattern.rb |  8 ++++----
 5 files changed, 27 insertions(+), 11 deletions(-)

diff --git a/CHANGELIST b/CHANGELIST
index e07929b..f4941db 100644
--- a/CHANGELIST
+++ b/CHANGELIST
@@ -1,3 +1,6 @@
+* 0.11.3
+  - Add optimized match_direct method
+
 * 0.10.7
   - add Grok::PatternError exception class
   - raise PatternError when a pattern contains an unknown %{expansion}
diff --git a/grok.gemspec b/grok.gemspec
index 99c25e3..0fdb11d 100644
--- a/grok.gemspec
+++ b/grok.gemspec
@@ -5,7 +5,7 @@ Gem::Specification.new do |spec|
 
   #svnrev = %x{svn info}.split("\n").grep(/Revision:/).first.split(" ").last.to_i
   spec.name = "jls-grok"
-  spec.version = "0.11.2"
+  spec.version = "0.11.3"
 
   spec.summary = "grok bindings for ruby"
   spec.description = "Grok ruby bindings - pattern match/extraction tool"
diff --git a/lib/grok-pure.rb b/lib/grok-pure.rb
index 116eaff..8edfccd 100644
--- a/lib/grok-pure.rb
+++ b/lib/grok-pure.rb
@@ -1,4 +1,4 @@
-require "rubygems"
+ require "rubygems"
 require "logger"
 require "cabin"
 require "grok/pure/discovery"
@@ -172,21 +172,34 @@ def match(text)
     end
   end # def match
 
+  # Returns the matched regexp object directly for performance at the
+  # cost of usability.
+  #
+  # Returns MatchData on success, nil on failure.
+  #
+  # Can be used with #capture
+  def execute(text)
+    @regexp.match(text)
+  end
+
   # Optimized match and capture instead of calling them separately
+  # This could be DRYed up by using #match and #capture directly
+  # but there's a bit of a worry that that may lower perf.
+  # This should be benchmarked!
   def match_and_capture(text)
-    match = @regexp.match(text)
+    match = execute(text)
     if match
       @logger.debug? and @logger.debug("Regexp match object", :names => match.names,
                                        :captures => match.captures)
-      @captures_func.call(match) { |k,v| yield k,v }
+      capture(match) {|k,v| yield k,v}
       return true
     else
       return false
     end
   end # def match_and_capture
 
-  def capture(match, block)
-    @captures_func.call(match) { |k,v| block.call k,v }
+  def capture(match, &block)    
+    @captures_func.call(match,&block)
   end # def capture
 
   public
diff --git a/lib/grok/pure/match.rb b/lib/grok/pure/match.rb
index c85a027..14c663d 100644
--- a/lib/grok/pure/match.rb
+++ b/lib/grok/pure/match.rb
@@ -12,7 +12,7 @@ def initialize
 
   public
   def each_capture(&block)
-    @grok.capture(@match, block)
+    @grok.capture(@match, &block)
   end # def each_capture
 
   public
diff --git a/test/pure-ruby/benchmark_pattern.rb b/test/pure-ruby/benchmark_pattern.rb
index 47115fe..5a2ae9d 100644
--- a/test/pure-ruby/benchmark_pattern.rb
+++ b/test/pure-ruby/benchmark_pattern.rb
@@ -14,16 +14,16 @@ def init_grok(named_captures_only)
 end
 
 Benchmark.bmbm(10) do |bm|
-  bm.report("100k Named Captures On") do
+  bm.report("10m Named Captures On") do
     grok = init_grok(true)
-    (1..100000).each do
+    (1..10_000_000).each do
       match = grok.match(@log_line)
       match.each_capture { |name, val| }
     end
   end
-  bm.report("100k Named Captures Off") do
+  bm.report("10m Named Captures Off") do
     grok = init_grok(false)
-    (1..100000).each do
+    (1..10_000_000).each do
       match = grok.match(@log_line)
       match.each_capture { |name, val| }
     end