From 51c7db64d6c969938a126cc6963405538c6ad074 Mon Sep 17 00:00:00 2001
From: James McKinney <james@slashpoundbang.com>
Date: Wed, 7 Oct 2015 14:01:04 -0400
Subject: [PATCH 1/2] Memoize the result of CSV#encode_re

---
 lib/csvlint/validate.rb | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

diff --git a/lib/csvlint/validate.rb b/lib/csvlint/validate.rb
index e87be357..0428ffd5 100644
--- a/lib/csvlint/validate.rb
+++ b/lib/csvlint/validate.rb
@@ -1,6 +1,16 @@
 module Csvlint
 
   class Validator
+    class LineCSV < CSV
+      ENCODE_RE = Hash.new do |h,str|
+        h[str] = Regexp.new(str)
+      end
+
+      # @see https://github.com/ruby/ruby/blob/v2_2_3/lib/csv.rb#L2273
+      def encode_re(*chunks)
+        ENCODE_RE[encode_str(*chunks)]
+      end
+    end
 
     include Csvlint::ErrorCollector
 
@@ -132,12 +142,12 @@ def parse_contents(stream, line = nil)
       @csv_options[:encoding] = @encoding
 
       begin
-        row = CSV.parse_line(stream, @csv_options)
+        row = LineCSV.parse_line(stream, @csv_options)
           # this is a one line substitute for CSV.new followed by row = CSV.shift. a CSV Row class is required
           # CSV.parse will return an array of arrays which breaks subsequent each_with_index invocations
           # TODO investigate if above would be a drag on memory
 
-      rescue CSV::MalformedCSVError => e
+      rescue LineCSV::MalformedCSVError => e
         build_exception_messages(e, stream, current_line)
       end
 
@@ -228,7 +238,7 @@ def header?
 
     def report_line_breaks(line_no=nil)
       return if @input !~ /[\r|\n]/ # Return straight away if there's no newline character - i.e. we're on the last line
-      line_break = CSV.new(@input).row_sep
+      line_break = LineCSV.new(@input).row_sep
       @line_breaks << line_break
       unless line_breaks_reported?
         if line_break != "\r\n"

From 77e54a46cb26639f4d535480597e8f49c8bf0468 Mon Sep 17 00:00:00 2001
From: James McKinney <james@slashpoundbang.com>
Date: Wed, 7 Oct 2015 15:15:24 -0400
Subject: [PATCH 2/2] Memoize the result of CSV#encode_str, CSV#escape_re, and
 disable CSV's converters feature

---
 lib/csvlint/validate.rb | 36 ++++++++++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)

diff --git a/lib/csvlint/validate.rb b/lib/csvlint/validate.rb
index 0428ffd5..2ca707bc 100644
--- a/lib/csvlint/validate.rb
+++ b/lib/csvlint/validate.rb
@@ -6,10 +6,46 @@ class LineCSV < CSV
         h[str] = Regexp.new(str)
       end
 
+      ENCODE_STR = Hash.new do |h,encoding_name|
+        h[encoding_name] = Hash.new do |h,chunks|
+          h[chunks] = chunks.map { |chunk| chunk.encode(encoding_name) }.join('')
+        end
+      end
+
+      ESCAPE_RE = Hash.new do |h,re_chars|
+        h[re_chars] = Hash.new do |h,re_esc|
+          h[re_esc] = Hash.new do |h,str|
+            h[str] = str.gsub(re_chars) {|c| re_esc + c}
+          end
+        end
+      end
+
+      # Optimization: Memoize `encode_re`.
       # @see https://github.com/ruby/ruby/blob/v2_2_3/lib/csv.rb#L2273
       def encode_re(*chunks)
         ENCODE_RE[encode_str(*chunks)]
       end
+
+      # Optimization: Memoize `encode_str`.
+      # @see https://github.com/ruby/ruby/blob/v2_2_3/lib/csv.rb#L2281
+      def encode_str(*chunks)
+        ENCODE_STR[@encoding.name][chunks]
+      end
+
+      # Optimization: Memoize `escape_re`.
+      # @see https://github.com/ruby/ruby/blob/v2_2_3/lib/csv.rb#L2265
+      def escape_re(str)
+        ESCAPE_RE[@re_chars][@re_esc][str]
+      end
+
+      # Optimization: Disable the CSV library's converters feature.
+      # @see https://github.com/ruby/ruby/blob/v2_2_3/lib/csv.rb#L2100
+      def init_converters(options, field_name = :converters)
+        @converters = []
+        @header_converters = []
+        options.delete(:unconverted_fields)
+        options.delete(field_name)
+      end
     end
 
     include Csvlint::ErrorCollector