opal · Apr 27, 2015 · Apr 27, 2015
Showing with 38 additions and 18 deletions.

+20 −16 opal/corelib/string.rb

+6 −0 spec/filters/bugs/string.rb

+2 −0 spec/filters/unsupported/regular_expressions.rb

+9 −1 spec/filters/unsupported/tainted.rb

+1 −1 spec/rubyspecs
diff --git a/opal/corelib/string.rb b/opal/corelib/string.rb
@@ -430,10 +430,10 @@ def gsub(pattern, replacement = undefined, &block)
       var result = '', match_data = nil, index = 0, match, _replacement;
 
       if (pattern.$$is_regexp) {
-        pattern = new RegExp(pattern.source, 'g' + (pattern.multiline ? 'm' : '') + (pattern.ignoreCase ? 'i' : ''));
+        pattern = new RegExp(pattern.source, 'gm' + (pattern.ignoreCase ? 'i' : ''));
       } else {
         pattern = #{Opal.coerce_to(`pattern`, String, :to_str)};
-        pattern = new RegExp(pattern.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'), 'g');
+        pattern = new RegExp(pattern.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'), 'gm');
       }
 
       while (true) {
@@ -821,7 +821,7 @@ def rindex(search, offset = undefined)
 
       if (search.$$is_regexp) {
         m = null;
-        r = new RegExp(search.source, 'g' + (search.multiline ? 'm' : '') + (search.ignoreCase ? 'i' : ''));
+        r = new RegExp(search.source, 'gm' + (search.ignoreCase ? 'i' : ''));
         while (true) {
           _m = r.exec(self);
           if (_m === null || _m.index > offset) {
@@ -872,7 +872,7 @@ def rpartition(sep)
 
       if (sep.$$is_regexp) {
         m = null;
-        r = new RegExp(sep.source, 'g' + (sep.multiline ? 'm' : '') + (sep.ignoreCase ? 'i' : ''));
+        r = new RegExp(sep.source, 'gm' + (sep.ignoreCase ? 'i' : ''));
 
         while (true) {
           _m = r.exec(self);
@@ -914,27 +914,31 @@ def rstrip
 
   def scan(pattern, &block)
     %x{
-      if (pattern.global) {
-        // should we clear it afterwards too?
-        pattern.lastIndex = 0;
-      }
-      else {
-        // rewrite regular expression to add the global flag to capture pre/post match
-        pattern = new RegExp(pattern.source, 'g' + (pattern.multiline ? 'm' : '') + (pattern.ignoreCase ? 'i' : ''));
-      }
+      var result = [],
+          match_data = nil,
+          match;
 
-      var result = [];
-      var match;
+      if (pattern.$$is_regexp) {
+        pattern = new RegExp(pattern.source, 'gm' + (pattern.ignoreCase ? 'i' : ''));
+      } else {
+        pattern = #{Opal.coerce_to(`pattern`, String, :to_str)};
+        pattern = new RegExp(pattern.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'), 'gm');
+      }
 
       while ((match = pattern.exec(self)) != null) {
-        var match_data = #{MatchData.new `pattern`, `match`};
+        match_data = #{MatchData.new `pattern`, `match`};
         if (block === nil) {
           match.length == 1 ? result.push(match[0]) : result.push(#{`match_data`.captures});
         } else {
-          match.length == 1 ? block(match[0]) : block.apply(self, #{`match_data`.captures});
+          match.length == 1 ? block(match[0]) : block.call(self, #{`match_data`.captures});
+        }
+        if (pattern.lastIndex === match.index) {
+          pattern.lastIndex += 1;
         }
       }
 
+      #{$~ = `match_data`}
+
       return (block !== nil ? self : result);
     }
   end

diff --git a/spec/filters/bugs/string.rb b/spec/filters/bugs/string.rb
@@ -9,6 +9,12 @@
   fails "String#lines uses $/ as the separator when none is given"
   fails "String#lines yields subclass instances for subclasses"
 
+  #The following failure is a side-effect of a symbol being used in one of the assertions:
+  fails "String#scan raises a TypeError if pattern isn't a Regexp and can't be converted to a String"
+
+  #The following failure is a side-effect of the currently broken MatchData#offset implementation:
+  fails "String#scan with pattern and block sets $~ for access from the block"
+
   fails "String#slice with Range calls to_int on range arguments"
 
   fails "String#split with String returns subclass instances based on self"

diff --git a/spec/filters/unsupported/regular_expressions.rb b/spec/filters/unsupported/regular_expressions.rb
@@ -7,4 +7,6 @@
   fails "String#gsub with pattern and replacement doesn't freak out when replacing ^" #Only fails "Text\nFoo".gsub(/^/, ' ').should == " Text\n Foo"
   fails "String#gsub with pattern and replacement supports \\G which matches at the beginning of the remaining (non-matched) string"
   fails "String#gsub with pattern and replacement returns a copy of self with all occurrences of pattern replaced with replacement" #Only fails str.gsub(/\Ah\S+\s*/, "huh? ").should == "huh? homely world. hah!"
+
+  fails "String#scan supports \\G which matches the end of the previous match / string start for first match"
 end
diff --git a/spec/filters/unsupported/tainted.rb b/spec/filters/unsupported/tainted.rb
@@ -120,9 +120,17 @@
   fails "String#rjust with length, padding taints result when self or padstr is tainted"
   fails "String#rjust with length, padding when padding is tainted and self is untainted returns a tainted string if and only if length is longer than self"
 
-
   fails "String#rstrip taints the result when self is tainted"
 
+  fails "String#scan taints the results if the String argument is tainted"
+  fails "String#scan taints the results when passed a String argument if self is tainted"
+  fails "String#scan taints the results if the Regexp argument is tainted"
+  fails "String#scan taints the results when passed a Regexp argument if self is tainted"
+  fails "String#scan with pattern and block taints the results if the String argument is tainted"
+  fails "String#scan with pattern and block taints the results when passed a String argument if self is tainted"
+  fails "String#scan with pattern and block taints the results if the Regexp argument is tainted"
+  fails "String#scan with pattern and block taints the results when passed a Regexp argument if self is tainted"
+
   fails "String#slice with index, length always taints resulting strings when self is tainted"
   fails "String#slice with Range always taints resulting strings when self is tainted"
   fails "String#slice with Regexp always taints resulting strings when self or regexp is tainted"

diff --git a/spec/rubyspecs b/spec/rubyspecs
@@ -154,7 +154,7 @@ corelib/core/string/rindex_spec
 corelib/core/string/rjust_spec
 corelib/core/string/rpartition_spec
 corelib/core/string/rstrip_spec
-# corelib/core/string/scan_spec - infinite loop on parsing
+corelib/core/string/scan_spec
 corelib/core/string/setbyte_spec
 corelib/core/string/size_spec
 corelib/core/string/slice_spec