From b3c7bf872f7fbe304cdcb449a2d501933805aa48 Mon Sep 17 00:00:00 2001
From: Shane Howley <howleysv@gmail.com>
Date: Wed, 15 Jul 2015 18:46:21 +0100
Subject: stdlib: Fix bug with unicode detection in re

Fix bug with unrecognised 'unicode' option in re:split/2,3 &
re:replace/3,4 when using pre-compiled regex.
---
 lib/stdlib/src/re.erl | 101 ++++++++++++++++++++++++--------------------------
 1 file changed, 48 insertions(+), 53 deletions(-)

(limited to 'lib/stdlib/src')

diff --git a/lib/stdlib/src/re.erl b/lib/stdlib/src/re.erl
index e5d9fc51d2..80bfe38970 100644
--- a/lib/stdlib/src/re.erl
+++ b/lib/stdlib/src/re.erl
@@ -132,8 +132,9 @@ split(Subject,RE) ->
 
 split(Subject,RE,Options) ->
     try
-    {NewOpt,Convert,Unicode,Limit,Strip,Group} =
-	process_split_params(Options,iodata,false,-1,false,false),
+    {NewOpt,Convert,Limit,Strip,Group} =
+	process_split_params(Options,iodata,-1,false,false),
+    Unicode = check_for_unicode(RE, Options),
     FlatSubject = to_binary(Subject, Unicode),
     case compile_split(RE,NewOpt) of
 	{error,_Err} ->
@@ -324,8 +325,8 @@ replace(Subject,RE,Replacement) ->
 
 replace(Subject,RE,Replacement,Options) ->
     try
-    {NewOpt,Convert,Unicode} =
-	process_repl_params(Options,iodata,false),
+    {NewOpt,Convert} = process_repl_params(Options,iodata),
+    Unicode = check_for_unicode(RE, Options),
     FlatSubject = to_binary(Subject, Unicode),
     FlatReplacement = to_binary(Replacement, Unicode),
     IoList = do_replace(FlatSubject,Subject,RE,FlatReplacement,NewOpt),
@@ -367,65 +368,59 @@ do_replace(FlatSubject,Subject,RE,Replacement,Options) ->
 	    apply_mlist(FlatSubject,Replacement,[Slist])
     end.
 
-process_repl_params([],Convert,Unicode) ->
-    {[],Convert,Unicode};
-process_repl_params([unicode|T],C,_U) ->
-    {NT,NC,NU} = process_repl_params(T,C,true), 
-    {[unicode|NT],NC,NU};
-process_repl_params([report_errors|_],_,_) ->
+process_repl_params([],Convert) ->
+    {[],Convert};
+process_repl_params([report_errors|_],_) ->
     throw(badopt);
-process_repl_params([{capture,_,_}|_],_,_) ->
+process_repl_params([{capture,_,_}|_],_) ->
     throw(badopt);
-process_repl_params([{capture,_}|_],_,_) ->
+process_repl_params([{capture,_}|_],_) ->
     throw(badopt);
-process_repl_params([{return,iodata}|T],_C,U) ->
-    process_repl_params(T,iodata,U);
-process_repl_params([{return,list}|T],_C,U) ->
-    process_repl_params(T,list,U);
-process_repl_params([{return,binary}|T],_C,U) ->
-    process_repl_params(T,binary,U);
-process_repl_params([{return,_}|_],_,_) ->
+process_repl_params([{return,iodata}|T],_C) ->
+    process_repl_params(T,iodata);
+process_repl_params([{return,list}|T],_C) ->
+    process_repl_params(T,list);
+process_repl_params([{return,binary}|T],_C) ->
+    process_repl_params(T,binary);
+process_repl_params([{return,_}|_],_) ->
     throw(badopt);
-process_repl_params([H|T],C,U) ->
-    {NT,NC,NU} = process_repl_params(T,C,U),
-    {[H|NT],NC,NU}.
-
-process_split_params([],Convert,Unicode,Limit,Strip,Group) ->
-    {[],Convert,Unicode,Limit,Strip,Group};
-process_split_params([unicode|T],C,_U,L,S,G) ->
-    {NT,NC,NU,NL,NS,NG} = process_split_params(T,C,true,L,S,G), 
-    {[unicode|NT],NC,NU,NL,NS,NG};
-process_split_params([trim|T],C,U,_L,_S,G) ->
-    process_split_params(T,C,U,-1,true,G); 
-process_split_params([{parts,0}|T],C,U,_L,_S,G) ->
-    process_split_params(T,C,U,-1,true,G); 
-process_split_params([{parts,N}|T],C,U,_L,_S,G) when is_integer(N), N >= 1 ->
-    process_split_params(T,C,U,N-1,false,G); 
-process_split_params([{parts,infinity}|T],C,U,_L,_S,G) ->
-    process_split_params(T,C,U,-1,false,G); 
-process_split_params([{parts,_}|_],_,_,_,_,_) ->
+process_repl_params([H|T],C) ->
+    {NT,NC} = process_repl_params(T,C),
+    {[H|NT],NC}.
+
+process_split_params([],Convert,Limit,Strip,Group) ->
+    {[],Convert,Limit,Strip,Group};
+process_split_params([trim|T],C,_L,_S,G) ->
+    process_split_params(T,C,-1,true,G); 
+process_split_params([{parts,0}|T],C,_L,_S,G) ->
+    process_split_params(T,C,-1,true,G); 
+process_split_params([{parts,N}|T],C,_L,_S,G) when is_integer(N), N >= 1 ->
+    process_split_params(T,C,N-1,false,G); 
+process_split_params([{parts,infinity}|T],C,_L,_S,G) ->
+    process_split_params(T,C,-1,false,G); 
+process_split_params([{parts,_}|_],_,_,_,_) ->
     throw(badopt); 
-process_split_params([group|T],C,U,L,S,_G) ->
-    process_split_params(T,C,U,L,S,true); 
-process_split_params([global|_],_,_,_,_,_) ->
+process_split_params([group|T],C,L,S,_G) ->
+    process_split_params(T,C,L,S,true); 
+process_split_params([global|_],_,_,_,_) ->
     throw(badopt);
-process_split_params([report_errors|_],_,_,_,_,_) ->
+process_split_params([report_errors|_],_,_,_,_) ->
     throw(badopt);
-process_split_params([{capture,_,_}|_],_,_,_,_,_) ->
+process_split_params([{capture,_,_}|_],_,_,_,_) ->
     throw(badopt);
-process_split_params([{capture,_}|_],_,_,_,_,_) ->
+process_split_params([{capture,_}|_],_,_,_,_) ->
     throw(badopt);
-process_split_params([{return,iodata}|T],_C,U,L,S,G) ->
-    process_split_params(T,iodata,U,L,S,G);
-process_split_params([{return,list}|T],_C,U,L,S,G) ->
-    process_split_params(T,list,U,L,S,G);
-process_split_params([{return,binary}|T],_C,U,L,S,G) ->
-    process_split_params(T,binary,U,L,S,G);
-process_split_params([{return,_}|_],_,_,_,_,_) ->
+process_split_params([{return,iodata}|T],_C,L,S,G) ->
+    process_split_params(T,iodata,L,S,G);
+process_split_params([{return,list}|T],_C,L,S,G) ->
+    process_split_params(T,list,L,S,G);
+process_split_params([{return,binary}|T],_C,L,S,G) ->
+    process_split_params(T,binary,L,S,G);
+process_split_params([{return,_}|_],_,_,_,_) ->
     throw(badopt);
-process_split_params([H|T],C,U,L,S,G) ->
-    {NT,NC,NU,NL,NS,NG} = process_split_params(T,C,U,L,S,G),
-    {[H|NT],NC,NU,NL,NS,NG}.
+process_split_params([H|T],C,L,S,G) ->
+    {NT,NC,NL,NS,NG} = process_split_params(T,C,L,S,G),
+    {[H|NT],NC,NL,NS,NG}.
 
 apply_mlist(Subject,Replacement,Mlist) ->
     do_mlist(Subject,Subject,0,precomp_repl(Replacement), Mlist).
-- 
cgit v1.2.3