More fixes to backslash and pipe escaping/unescaping

History.txt

@@ -1,7 +1,7 @@
 == In Git
 
 === New Features
-* Table cells can now contain escaped bars - \|. (#48. Gregory Hnatiuk)
+* Table cells can now contain escaped bars - \|. (#48. Gregory Hnatiuk, Aslak Hellesøy)
 * Luxemburgish (lu) added. (Christoph König)
 
 == 1.0.26 (2010-05-09)

features/escaped_pipes.feature

@@ -1,4 +1,8 @@
 Feature: Escaped pipes
   Scenario: They are the future
     Given they have arrived
-      | \|OK\| | (OK) |
+      | æ | o |
+      | a | ø |
+    Given they have arrived
+      | æ   | \|o |
+      | \|a | ø\\ |

features/step_definitions/eyeball_steps.rb

@@ -0,0 +1,3 @@
+Given /^they have arrived$/ do |table|
+  announce table.raw
+end

java/src/main/java/gherkin/formatter/PrettyFormatter.java

@@ -124,7 +124,11 @@ public void step(String keyword, String name, int line) {
 
     public void row(List<String> row, int line) {
         if (rows == null) rows = new ArrayList<List<String>>();
-        rows.add(row);
+        List<String> escapedRow = new ArrayList<String>(row.size());
+        for(String cell : row) {
+            escapedRow.add(cell.replaceAll("\\|", "\\\\|").replaceAll("\\\\(?!\\|)", "\\\\\\\\"));
+        }
+        rows.add(escapedRow);
     }
 
     public void pyString(String string, int line) {

lib/gherkin/formatter/escaping.rb

@@ -0,0 +1,15 @@
+module Gherkin
+  module Formatter
+    module Escaping
+      # Escapes a pipes and backslashes:
+      #
+      # * | becomes \|
+      # * \ becomes \\
+      #
+      # This is used in the pretty formatter.
+      def escape_cell(s)
+        s.gsub(/\|/, "\\|").gsub(/\\(?!\|)/, "\\\\\\\\")
+      end
+    end
+  end
+end

lib/gherkin/formatter/pretty_formatter.rb

@@ -3,6 +3,7 @@
 require 'gherkin/formatter/colors'
 require 'gherkin/formatter/monochrome_format'
 require 'gherkin/formatter/argument'
+require 'gherkin/formatter/escaping'
 
 module Gherkin
   module Formatter
@@ -11,6 +12,7 @@ class PrettyFormatter
       java_impl('gherkin.jar')
 
       include Colors
+      include Escaping
 
       def initialize(io, monochrome=false)
         @io = io
@@ -67,7 +69,7 @@ def step(keyword, name, line, status=nil, exception=nil, arguments=nil, location
 
       def row(row, line)
         @rows ||= []
-        @rows << row.map{|cell| cell.gsub(/\|/, "\\|")}
+        @rows << row.map{|cell| escape_cell(cell)}
       end
 
       def py_string(string, line)

ragel/lexer.c.rl.erb

@@ -181,8 +181,10 @@ static VALUE rb_eGherkinLexingError;
   action store_cell_content {
     VALUE con = ENCODED_STR_NEW(PTR_TO(content_start), LEN(content_start, p));
     rb_funcall(con, rb_intern("strip!"), 0);
-    VALUE re = rb_reg_regcomp(rb_str_new2("\\\\\\|"));
-    rb_funcall(con, rb_intern("gsub!"), 2, re, rb_str_new2("|"));
+    VALUE re_pipe      = rb_reg_regcomp(rb_str_new2("\\\\\\|"));
+    VALUE re_backslash = rb_reg_regcomp(rb_str_new2("\\\\\\\\"));
+    rb_funcall(con, rb_intern("gsub!"), 2, re_pipe,      rb_str_new2("|"));
+    rb_funcall(con, rb_intern("gsub!"), 2, re_backslash, rb_str_new2("\\"));
 
     rb_ary_push(current_row, con);
   }

ragel/lexer.java.rl.erb

@@ -115,7 +115,7 @@ public class <%= @i18n.underscored_iso_code.upcase %> implements Lexer {
 
     action store_cell_content {
       String con = substring(data, contentStart, p).trim();
-      currentRow.add(con.replaceAll("\\\\\\|", "|"));
+      currentRow.add(con.replaceAll("\\\\\\|", "|").replaceAll("\\\\\\\\", "\\\\"));
     }
 
     action store_row {

ragel/lexer.rb.rl.erb

@@ -103,7 +103,7 @@ module Gherkin
 
         action store_cell_content {
           con = utf8_pack(data[@content_start...p]).strip
-          current_row << con.gsub(/\\\|/, "|")
+          current_row << con.gsub(/\\\|/, "|").gsub(/\\\\/, "\\")
         }
 
         action store_row {

spec/gherkin/formatter/pretty_formatter_spec.rb

@@ -113,7 +113,7 @@ def assert_pretty(input, output=input)
       """
     And there is another step
       | æ   | \\|o |
-      | \\|a | ø   |
+      | \\|a | ø\\\\ |
     Then we will see steps
 })
       end
@@ -148,6 +148,15 @@ def assert_pretty(input, output=input)
       it "should preserve tabs" do
         assert_pretty(IO.read(File.dirname(__FILE__) + '/tabs.feature'), IO.read(File.dirname(__FILE__) + '/spaces.feature'))
       end
+
+      it "should escape backslashes and pipes" do
+        io = StringIO.new
+        l = PrettyFormatter.new(io, true)
+        l.row(['|', '\\'], 1)
+        l.flush_table
+        io.rewind
+        io.read.should == '      | \\| | \\\\ |' + "\n"
+      end
     end
   end
 end