escape_utils 0.1.0 → 0.1.1

data/.gitignore

@@ -1,3 +1,5 @@
 Makefile
 *.o
-*.bundle
+*.bundle
+pkg/*
+doc/*

data/CHANGELOG.md

@@ -1,4 +1,7 @@
 # Changelog
 
+## 0.1.1 (June 8th, 2010)
+* added javascript escaping
+
 ## 0.1.0 (June 8th, 2010)
 * initial release

data/README.rdoc

@@ -2,7 +2,7 @@
 
 Being as though we're all html escaping everything these days, why not make it faster?
 
-At the moment escape_utils supports escaping and unescaping of HTML, but I wanna add URL encoding soon
+At the moment escape_utils supports escaping and unescaping of HTML, and Javascript but I wanna add URL encoding soon
 
 It has monkey-patches for Rack::Utils, CGI, ERB::Util and Haml
 
@@ -23,6 +23,11 @@ It has monkey-patches for Rack::Utils, CGI, ERB::Util and Haml
   escaped_html = EscapeUtils.escape_html(html)
   html = EscapeUtils.unescape_html(escaped_html)
 
+=== Escaping Javascript
+
+  javascript = `curl -s http://code.jquery.com/jquery-1.4.2.js`
+  escaped_javascript = EscapeUtils.escape_javascript(javascript)
+
 === Monkey Patches
 
   require 'escape_utils/rack' # to patch Rack::Utils
@@ -33,11 +38,11 @@ It has monkey-patches for Rack::Utils, CGI, ERB::Util and Haml
 == Benchmarks
 
 In my testing, escaping is around 10-20x faster than the pure ruby implementations in wide use today.
-While unescaping is around 24x faster than CGI.unescapeHTML - also pure ruby.
+While unescaping is around 20-40x faster than CGI.unescapeHTML - also pure ruby.
 
 This output is from my laptop using the benchmark scripts in the benchmarks folder.
 
-=== Escaping
+=== HTML Escaping
 
  Rack::Utils.escape_html
   0.560000   0.040000   0.600000 (  0.589475)
@@ -50,9 +55,16 @@ This output is from my laptop using the benchmark scripts in the benchmarks fold
  EscapeUtils.escape_html
   0.050000   0.010000   0.060000 (  0.054799)
 
-=== Unescaping
+=== HTML Unescaping
 
  CGI.unescapeHTML
   1.140000   0.010000   1.150000 (  1.148470)
  EscapeUtils.unescape_html
-  0.040000   0.000000   0.040000 (  0.046166)
+  0.040000   0.000000   0.040000 (  0.046166)
+
+=== Javascript Escaping
+
+ ActionView::Helpers::JavaScriptHelper#escape_javascript
+  2.000000   0.020000   2.020000 (  2.023047)
+ EscapeUtils.escape_javascript
+  0.110000   0.010000   0.120000 (  0.121761)

data/VERSION

@@ -1 +1 @@
-0.1.0
+0.1.1

data/benchmark/javascript_escape.rb

@@ -0,0 +1,36 @@
+# encoding: utf-8
+$LOAD_PATH.unshift File.expand_path(File.dirname(__FILE__) + '/..')
+$LOAD_PATH.unshift File.expand_path(File.dirname(__FILE__) + '/../lib')
+
+require 'rubygems'
+require 'benchmark'
+
+require 'action_view'
+require 'escape_utils'
+
+class ActionPackBench
+  extend ActionView::Helpers::JavaScriptHelper
+end
+
+times = 100
+url = "http://code.jquery.com/jquery-1.4.2.js"
+javascript = `curl -s #{url}`
+puts "Escaping #{javascript.bytesize} bytes of javascript from #{url}"
+
+puts ActionPackBench.escape_javascript(javascript).eql?(EscapeUtils.escape_javascript(javascript))
+
+Benchmark.bmbm do |x|
+  x.report do
+    puts "ActionView::Helpers::JavaScriptHelper#escape_javascript"
+    times.times do
+      ActionPackBench.escape_javascript(javascript)
+    end
+  end
+
+  x.report do
+    puts "EscapeUtils.escape_javascript"
+    times.times do
+      EscapeUtils.escape_javascript(javascript)
+    end
+  end
+end

data/escape_utils.gemspec

@@ -5,7 +5,7 @@
 
 Gem::Specification.new do |s|
   s.name = %q{escape_utils}
-  s.version = "0.1.0"
+  s.version = "0.1.1"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Brian Lopez"]
@@ -22,8 +22,9 @@ Gem::Specification.new do |s|
      "README.rdoc",
      "Rakefile",
      "VERSION",
-     "benchmark/escape.rb",
-     "benchmark/unescape.rb",
+     "benchmark/html_escape.rb",
+     "benchmark/html_unescape.rb",
+     "benchmark/javascript_escape.rb",
      "escape_utils.gemspec",
      "ext/escape_utils.c",
      "ext/extconf.rb",
@@ -34,6 +35,7 @@ Gem::Specification.new do |s|
      "lib/escape_utils/rack.rb",
      "spec/html/escape_spec.rb",
      "spec/html/unescape_spec.rb",
+     "spec/javascript/escape_spec.rb",
      "spec/rcov.opts",
      "spec/spec.opts",
      "spec/spec_helper.rb"
@@ -46,6 +48,7 @@ Gem::Specification.new do |s|
   s.test_files = [
     "spec/html/escape_spec.rb",
      "spec/html/unescape_spec.rb",
+     "spec/javascript/escape_spec.rb",
      "spec/spec_helper.rb"
   ]
 

data/ext/escape_utils.c

@@ -72,7 +72,39 @@ static size_t unescape_html(unsigned char *out, const unsigned char *in, size_t
   return total + (i-offset);
 }
 
+static size_t escape_javascript(unsigned char *out, const unsigned char *in, size_t in_len) {
+  size_t i = 0, offset = 0, total = 0;
+
+  for(;i<in_len;i++) {
+    switch(in[i]) {
+      case '\\':  APPEND_BUFFER("\\\\", 2, 1);
+      case '<':
+        if (i+1 <= in_len && in[i+1] == '/') {
+          APPEND_BUFFER("<\\/", 3, 2);
+        }
+        break;
+      case '\r':
+        if (i+1 <= in_len && in[i+1] == '\n') {
+          APPEND_BUFFER("\\n", 2, 1);
+        } else {
+          APPEND_BUFFER("\\n", 2, 1);
+        }
+        break;
+      case '\n': APPEND_BUFFER("\\n", 2, 1);
+      case '\"': APPEND_BUFFER("\\\"", 2, 1);
+      case '\'': APPEND_BUFFER("\\'", 2, 1);
+    }
+  }
+
+  // append the rest of the buffer
+  memcpy(&out[total], &in[offset], i-offset);
+
+  return total + (i-offset);
+}
+
 static VALUE rb_escape_html(VALUE self, VALUE str) {
+  Check_Type(str, T_STRING);
+
   VALUE rb_output_buf;
   unsigned char *inBuf = (unsigned char*)RSTRING_PTR(str);
   size_t len = RSTRING_LEN(str), new_len = 0;
@@ -94,6 +126,8 @@ static VALUE rb_escape_html(VALUE self, VALUE str) {
 }
 
 static VALUE rb_unescape_html(VALUE self, VALUE str) {
+  Check_Type(str, T_STRING);
+
   VALUE rb_output_buf;
   unsigned char *inBuf = (unsigned char*)RSTRING_PTR(str);
   size_t len = RSTRING_LEN(str), new_len = 0;
@@ -114,6 +148,33 @@ static VALUE rb_unescape_html(VALUE self, VALUE str) {
   return rb_output_buf;
 }
 
+static VALUE rb_escape_javascript(VALUE self, VALUE str) {
+  if (str == Qnil) {
+    return rb_str_new2("");
+  }
+
+  Check_Type(str, T_STRING);
+
+  VALUE rb_output_buf;
+  unsigned char *inBuf = (unsigned char*)RSTRING_PTR(str);
+  size_t len = RSTRING_LEN(str), new_len = 0;
+
+  // this is the max size the string could be
+  // TODO: we should try to be more intelligent about this
+  unsigned char *outBuf = (unsigned char *)malloc(sizeof(unsigned char *)*(len*2));
+
+  // perform our escape, returning the new string's length
+  new_len = escape_javascript(outBuf, inBuf, len);
+
+  // create our new ruby string
+  rb_output_buf = rb_str_new((char *)outBuf, new_len);
+
+  // free the temporary C string
+  free(outBuf);
+
+  return rb_output_buf;
+}
+
 /* Ruby Extension initializer */
 void Init_escape_utils_ext() {
   VALUE mEscape = rb_define_module("EscapeUtils");
@@ -121,4 +182,6 @@ void Init_escape_utils_ext() {
   rb_define_module_function(mEscape,  "escape_html",    rb_escape_html,   1);
   rb_define_method(mEscape,           "unescape_html",  rb_unescape_html, 1);
   rb_define_module_function(mEscape,  "unescape_html",  rb_unescape_html, 1);
+  rb_define_method(mEscape,           "escape_javascript",  rb_escape_javascript, 1);
+  rb_define_module_function(mEscape,  "escape_javascript",  rb_escape_javascript, 1);
 }

data/lib/escape_utils.rb

@@ -4,5 +4,5 @@ require 'escape_utils_ext'
 
 EscapeUtils.send(:extend, EscapeUtils)
 module EscapeUtils
-  VERSION = "0.0.1"
+  VERSION = "0.1.1"
 end

data/spec/html/escape_spec.rb

@@ -1,26 +1,24 @@
 # encoding: UTF-8
 require File.expand_path(File.dirname(__FILE__) + '/../spec_helper.rb')
 
-describe EscapeUtils do
+describe EscapeUtils, "escape_html" do
   it "should respond to escape_html" do
     EscapeUtils.should respond_to(:escape_html)
   end
 
-  context "escape_html" do
-    it "should escape a basic html tag" do
-      EscapeUtils.escape_html("<some_tag/>").should eql("&lt;some_tag/&gt;")
-    end
+  it "should escape a basic html tag" do
+    EscapeUtils.escape_html("<some_tag/>").should eql("&lt;some_tag/&gt;")
+  end
 
-    it "should escape double-quotes" do
-      EscapeUtils.escape_html("<some_tag some_attr=\"some value\"/>").should eql("&lt;some_tag some_attr=&quot;some value&quot;/&gt;")
-    end
+  it "should escape double-quotes" do
+    EscapeUtils.escape_html("<some_tag some_attr=\"some value\"/>").should eql("&lt;some_tag some_attr=&quot;some value&quot;/&gt;")
+  end
 
-    it "should escape single-quotes" do
-      EscapeUtils.escape_html("<some_tag some_attr='some value'/>").should eql("&lt;some_tag some_attr=&#39;some value&#39;/&gt;")
-    end
+  it "should escape single-quotes" do
+    EscapeUtils.escape_html("<some_tag some_attr='some value'/>").should eql("&lt;some_tag some_attr=&#39;some value&#39;/&gt;")
+  end
 
-    it "should escape the & character" do
-      EscapeUtils.escape_html("<b>Bourbon & Branch</b>").should eql("&lt;b&gt;Bourbon &amp; Branch&lt;/b&gt;")
-    end
+  it "should escape the & character" do
+    EscapeUtils.escape_html("<b>Bourbon & Branch</b>").should eql("&lt;b&gt;Bourbon &amp; Branch&lt;/b&gt;")
   end
 end

data/spec/html/unescape_spec.rb

@@ -1,26 +1,24 @@
 # encoding: UTF-8
 require File.expand_path(File.dirname(__FILE__) + '/../spec_helper.rb')
 
-describe EscapeUtils do
+describe EscapeUtils, "unescape_html" do
   it "should respond to unescape_html" do
     EscapeUtils.should respond_to(:unescape_html)
   end
 
-  context "unescape_html" do
-    it "should unescape a basic html tag" do
-      EscapeUtils.unescape_html("&lt;some_tag/&gt;").should eql("<some_tag/>")
-    end
+  it "should unescape a basic html tag" do
+    EscapeUtils.unescape_html("&lt;some_tag/&gt;").should eql("<some_tag/>")
+  end
 
-    it "should unescape double-quotes" do
-      EscapeUtils.unescape_html("&lt;some_tag some_attr=&quot;some value&quot;/&gt;").should eql("<some_tag some_attr=\"some value\"/>")
-    end
+  it "should unescape double-quotes" do
+    EscapeUtils.unescape_html("&lt;some_tag some_attr=&quot;some value&quot;/&gt;").should eql("<some_tag some_attr=\"some value\"/>")
+  end
 
-    it "should unescape single-quotes" do
-      EscapeUtils.unescape_html("&lt;some_tag some_attr=&#39;some value&#39;/&gt;").should eql("<some_tag some_attr='some value'/>")
-    end
+  it "should unescape single-quotes" do
+    EscapeUtils.unescape_html("&lt;some_tag some_attr=&#39;some value&#39;/&gt;").should eql("<some_tag some_attr='some value'/>")
+  end
 
-    it "should unescape the & character" do
-      EscapeUtils.unescape_html("&lt;b&gt;Bourbon &amp; Branch&lt;/b&gt;").should eql("<b>Bourbon & Branch</b>")
-    end
+  it "should unescape the & character" do
+    EscapeUtils.unescape_html("&lt;b&gt;Bourbon &amp; Branch&lt;/b&gt;").should eql("<b>Bourbon & Branch</b>")
   end
 end

data/spec/javascript/escape_spec.rb

@@ -0,0 +1,25 @@
+# encoding: UTF-8
+require File.expand_path(File.dirname(__FILE__) + '/../spec_helper.rb')
+
+describe EscapeUtils, "escape_javascript" do
+  it "should respond to escape_javascript" do
+    EscapeUtils.should respond_to(:escape_javascript)
+  end
+
+  # these are from the ActionView tests
+  it "should return an empty string if passed nil" do
+    EscapeUtils.escape_javascript(nil).should eql("")
+  end
+  
+  it "should escape quotes and newlines" do
+    EscapeUtils.escape_javascript(%(This "thing" is really\n netos')).should eql(%(This \\"thing\\" is really\\n netos\\'))
+  end
+
+  it "should escape backslashes" do
+    EscapeUtils.escape_javascript(%(backslash\\test)).should eql(%(backslash\\\\test))
+  end
+
+  it "should escape closed html tags" do
+    EscapeUtils.escape_javascript(%(dont </close> tags)).should eql(%(dont <\\/close> tags))
+  end
+end

metadata

@@ -5,8 +5,8 @@ version: !ruby/object:Gem::Version
   segments: 
   - 0
   - 1
-  - 0
-  version: 0.1.0
+  - 1
+  version: 0.1.1
 platform: ruby
 authors: 
 - Brian Lopez
@@ -33,8 +33,9 @@ files:
 - README.rdoc
 - Rakefile
 - VERSION
-- benchmark/escape.rb
-- benchmark/unescape.rb
+- benchmark/html_escape.rb
+- benchmark/html_unescape.rb
+- benchmark/javascript_escape.rb
 - escape_utils.gemspec
 - ext/escape_utils.c
 - ext/extconf.rb
@@ -45,6 +46,7 @@ files:
 - lib/escape_utils/rack.rb
 - spec/html/escape_spec.rb
 - spec/html/unescape_spec.rb
+- spec/javascript/escape_spec.rb
 - spec/rcov.opts
 - spec/spec.opts
 - spec/spec_helper.rb
@@ -82,4 +84,5 @@ summary: Faster string escaping routines for your web apps
 test_files: 
 - spec/html/escape_spec.rb
 - spec/html/unescape_spec.rb
+- spec/javascript/escape_spec.rb
 - spec/spec_helper.rb