From 45d3ea361ec59e09dec248e81f4447d08b7c987b Mon Sep 17 00:00:00 2001
From: satyr <murky.satyr@gmail.com>
Date: Wed, 23 Oct 2013 16:17:09 +0900
Subject: [PATCH] heregexes now keep escaped whitespaces
 (jashkenas/coffee-script#3059)

---
 lib/lexer.js  | 15 +++++++++++----
 src/lexer.co  | 10 +++++++---
 test/regex.co |  9 +++++++++
 3 files changed, 27 insertions(+), 7 deletions(-)

diff --git a/lib/lexer.js b/lib/lexer.js
index 411ae9ae8..3e4bdbbe9 100644
--- a/lib/lexer.js
+++ b/lib/lexer.js
@@ -1,4 +1,4 @@
-var string, TABS, unlines, enlines, enslash, reslash, camelize, character, KEYWORDS_SHARED, KEYWORDS_UNUSED, KEYWORDS, ID, SYMBOL, SPACE, MULTIDENT, SIMPLESTR, BSTOKEN, JSTOKEN, NUMBER, NUMBER_OMIT, REGEX, HEREGEX_OMIT, LASTDENT, INLINEDENT, NONASCII, OPENERS, CLOSERS, INVERSES, CHAIN, ARG, BLOCK_USERS, slice$ = [].slice;
+var string, TABS, unlines, enlines, enslash, reslash, camelize, deheregex, character, KEYWORDS_SHARED, KEYWORDS_UNUSED, KEYWORDS, ID, SYMBOL, SPACE, MULTIDENT, SIMPLESTR, BSTOKEN, JSTOKEN, NUMBER, NUMBER_OMIT, REGEX, LASTDENT, INLINEDENT, NONASCII, OPENERS, CLOSERS, INVERSES, CHAIN, ARG, BLOCK_USERS, slice$ = [].slice;
 exports.lex = function(code, options){
   return (clone$(exports)).tokenize(code || '', options || {});
 };
@@ -424,7 +424,7 @@ exports.doHeregex = function(code, index){
       if (t[0] === 'TOKENS') {
         tokens.push.apply(tokens, t[1]);
       } else {
-        val = t[1].replace(HEREGEX_OMIT, '');
+        val = deheregex(t[1]);
         if (one && !val) {
           continue;
         }
@@ -443,7 +443,7 @@ exports.doHeregex = function(code, index){
     }
     this.token(flag === '$' ? ')' : ')CALL', '');
   } else {
-    this.regex(reslash(parts[0][1].replace(HEREGEX_OMIT, '')), flag);
+    this.regex(reslash(deheregex(parts[0][1])), flag);
   }
   return delim.length + parts.size + flag.length;
 };
@@ -1064,6 +1064,14 @@ reslash = replacer(/(\\.)|\//g, function(){
 camelize = replacer(/-[a-z]/ig, function(it){
   return it.charAt(1).toUpperCase();
 });
+deheregex = replacer(/\s+(?:#.*)?|(\\[\s\S])/g, function(arg$, bs){
+  bs || (bs = '');
+  if ('\n' === bs.charAt(1)) {
+    return '\\n';
+  } else {
+    return bs;
+  }
+});
 function lchomp(it){
   return it.slice(1 + it.lastIndexOf('\n', 0));
 }
@@ -1506,7 +1514,6 @@ JSTOKEN = /(`+)([^`][\s\S]*?)\1|/g;
 NUMBER = /0x[\dA-Fa-f][\dA-Fa-f_]*|([2-9]|[12]\d|3[0-6])r([\dA-Za-z]\w*)|((\d[\d_]*)(\.\d[\d_]*)?(?:e[+-]?\d[\d_]*)?)[$\w]*|/g;
 NUMBER_OMIT = /_+/g;
 REGEX = /\/([^[\/\n\\]*(?:(?:\\.|\[[^\]\n\\]*(?:\\.[^\]\n\\]*)*\])[^[\/\n\\]*)*)\/([gimy]{1,4}|\$?)|/g;
-HEREGEX_OMIT = /\s+(?:#.*)?/g;
 LASTDENT = /\n[^\n\S]*$/;
 INLINEDENT = /[^\n\S]*[^#\s]|/g;
 NONASCII = /[\x80-\uFFFF]/;
diff --git a/src/lexer.co b/src/lexer.co
index 24cabab96..d015a3c00 100644
--- a/src/lexer.co
+++ b/src/lexer.co
@@ -310,7 +310,7 @@ exports import
         if t.0 is \TOKENS
           tokens.push ...t.1
         else
-          val = t.1.replace HEREGEX_OMIT, ''
+          val = deheregex t.1
           continue if one and not val
           one = tokens.push t <<< [\STRNUM @string \' enslash val]
         tokens.push [\+- \+ tokens[*-1]2]
@@ -319,7 +319,7 @@ exports import
         @token \, \,
         if dynaflag then tokens.push ...dynaflag else @token \STRNUM "'#flag'"
       @token (if flag is \$ then \) else \)CALL), ''
-    else @regex reslash(parts.0.1.replace HEREGEX_OMIT, ''), flag
+    else @regex reslash(deheregex parts.0.1), flag
     delim.length + parts.size + flag.length
 
   # Matches a word literal, or ignores a sequence of whitespaces.
@@ -748,6 +748,11 @@ reslash = replacer /(\\.)|\//g -> @@1 or \\\/
 # Transforms hyphenated-words to camelCase.
 camelize = replacer /-[a-z]/ig -> it.char-at 1 .to-upper-case!
 
+# ESifies a heregex.
+deheregex = replacer do
+  /\s+(?:#.*)?|(\\[\s\S])/g
+  (, bs || '') -> if \\n is bs.charAt 1 then \\\n else bs
+
 # Deletes the first character if newline.
 function lchomp then it.slice 1 + it.lastIndexOf \\n 0
 
@@ -1069,7 +1074,6 @@ REGEX = //
      )*
   )/ ([gimy]{1,4}|\$?)
 |//g
-HEREGEX_OMIT = /\s+(?:#.*)?/g
 
 LASTDENT   = /\n[^\n\S]*$/
 INLINEDENT = /[^\n\S]*[^#\s]|/g
diff --git a/test/regex.co b/test/regex.co
index 7bce167d4..87a516e60 100644
--- a/test/regex.co
+++ b/test/regex.co
@@ -68,6 +68,7 @@ eq \01234 // 0 #{
   // 1 #{ //2//source } 3 //source
 } 4 //source
 
+# `this` interpolation
 let this = \THIS
   ok //^ \\##@#this $//test //\#THISTHIS//source
 
@@ -82,6 +83,14 @@ ok /:\/[/]goog/.test 'http://google.com'
 eq /0/['source'], //#{0}//['source']
 
 
+# [coffee#3059](https://github.com/jashkenas/coffee-script/pull/3059)
+# Keep escaped whitespaces.
+ok  //^
+  a \ b　\　c \
+  d
+$//test 'a b\u3000c\nd'
+
+
 ### $ flag
 eq \string typeof /^$/$
 eq \string typeof //^$//$