Upgrade \X to match an extended grapheme cluster

git-svn-id: svn://vcs.exim.org/pcre/code/trunk@1011 2f5784b3-3f2a-0410-8824-cb99058d5e15

6 files changed, 492 insertions, 33 deletions

diff --git a/testdata/testinput10 b/testdata/testinput10
index 7856518..4dfb39b 100644
--- a/testdata/testinput10
+++ b/testdata/testinput10
@@ -1026,4 +1026,68 @@
     AA\P
     AA\P\P 
 
+/-- These are tests for extended grapheme clusters --/ 
+
+/^\X/8+
+    G\x{34e}\x{34e}X
+    \x{34e}\x{34e}X
+    \x04X
+    \x{1100}X
+    \x{1100}\x{34e}X
+    \x{1b04}\x{1b04}X 
+    *These match up to the roman letters
+    \x{1111}\x{1111}L,L
+    \x{1111}\x{1111}\x{1169}L,L,V
+    \x{1111}\x{ae4c}L, LV
+    \x{1111}\x{ad89}L, LVT
+    \x{1111}\x{ae4c}\x{1169}L, LV, V
+    \x{1111}\x{ae4c}\x{1169}\x{1169}L, LV, V, V
+    \x{1111}\x{ae4c}\x{1169}\x{11fe}L, LV, V, T
+    \x{1111}\x{ad89}\x{11fe}L, LVT, T
+    \x{1111}\x{ad89}\x{11fe}\x{11fe}L, LVT, T, T
+    \x{ad89}\x{11fe}\x{11fe}LVT, T, T
+    *These match just the first codepoint (invalid sequence)
+    \x{1111}\x{11fe}L, T
+    \x{ae4c}\x{1111}LV, L
+    \x{ae4c}\x{ae4c}LV, LV
+    \x{ae4c}\x{ad89}LV, LVT
+    \x{1169}\x{1111}V, L
+    \x{1169}\x{ae4c}V, LV
+    \x{1169}\x{ad89}V, LVT
+    \x{ad89}\x{1111}LVT, L
+    \x{ad89}\x{1169}LVT, V
+    \x{ad89}\x{ae4c}LVT, LV
+    \x{ad89}\x{ad89}LVT, LVT
+    \x{11fe}\x{1111}T, L
+    \x{11fe}\x{1169}T, V
+    \x{11fe}\x{ae4c}T, LV
+    \x{11fe}\x{ad89}T, LVT
+    *Test extend and spacing mark
+    \x{1111}\x{ae4c}\x{0711}L, LV, extend
+    \x{1111}\x{ae4c}\x{1b04}L, LV, spacing mark
+    \x{1111}\x{ae4c}\x{1b04}\x{0711}\x{1b04}L, LV, spacing mark, extend, spacing mark
+    *Test CR, LF, and control
+    \x0d\x{0711}CR, extend
+    \x0d\x{1b04}CR, spacingmark
+    \x0a\x{0711}LF, extend
+    \x0a\x{1b04}LF, spacingmark
+    \x0b\x{0711}Control, extend
+    \x09\x{1b04}Control, spacingmark
+    *There are no Prepend characters, so we can't test Prepend, CR
+    
+/^(?>\X{2})X/8+
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+    
+/^\X{2,4}X/8+
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+
+/^\X{2,4}?X/8+
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+
+/-- --/
+
 /-- End of testinput10 --/ 
diff --git a/testdata/testinput6 b/testdata/testinput6
index 7010a0d..d0cad59 100644
--- a/testdata/testinput6
+++ b/testdata/testinput6
@@ -406,7 +406,13 @@
     A\x{300}\x{301}B\x{300}C\x{300}\x{301}
     A\x{300}\x{301}B\x{300}C\x{300}\x{301}X
     A\x{300}\x{301}B\x{300}C\x{300}\x{301}DA\x{300}X
-    
+
+/^\X/8
+    A
+    A\x{300}BC 
+    A\x{300}\x{301}\x{302}BC 
+    \x{300}  
+
 /^\p{Han}+/8
     \x{2e81}\x{3007}\x{2f804}\x{31a0}
     ** Failers
@@ -814,5 +820,69 @@
 /Ⱥ/8i
     Ⱥ
     ⱥ
+    
+/-- These are tests for extended grapheme clusters --/ 
+
+/^\X/8+
+    G\x{34e}\x{34e}X
+    \x{34e}\x{34e}X
+    \x04X
+    \x{1100}X
+    \x{1100}\x{34e}X
+    \x{1b04}\x{1b04}X 
+    *These match up to the roman letters
+    \x{1111}\x{1111}L,L
+    \x{1111}\x{1111}\x{1169}L,L,V
+    \x{1111}\x{ae4c}L, LV
+    \x{1111}\x{ad89}L, LVT
+    \x{1111}\x{ae4c}\x{1169}L, LV, V
+    \x{1111}\x{ae4c}\x{1169}\x{1169}L, LV, V, V
+    \x{1111}\x{ae4c}\x{1169}\x{11fe}L, LV, V, T
+    \x{1111}\x{ad89}\x{11fe}L, LVT, T
+    \x{1111}\x{ad89}\x{11fe}\x{11fe}L, LVT, T, T
+    \x{ad89}\x{11fe}\x{11fe}LVT, T, T
+    *These match just the first codepoint (invalid sequence)
+    \x{1111}\x{11fe}L, T
+    \x{ae4c}\x{1111}LV, L
+    \x{ae4c}\x{ae4c}LV, LV
+    \x{ae4c}\x{ad89}LV, LVT
+    \x{1169}\x{1111}V, L
+    \x{1169}\x{ae4c}V, LV
+    \x{1169}\x{ad89}V, LVT
+    \x{ad89}\x{1111}LVT, L
+    \x{ad89}\x{1169}LVT, V
+    \x{ad89}\x{ae4c}LVT, LV
+    \x{ad89}\x{ad89}LVT, LVT
+    \x{11fe}\x{1111}T, L
+    \x{11fe}\x{1169}T, V
+    \x{11fe}\x{ae4c}T, LV
+    \x{11fe}\x{ad89}T, LVT
+    *Test extend and spacing mark
+    \x{1111}\x{ae4c}\x{0711}L, LV, extend
+    \x{1111}\x{ae4c}\x{1b04}L, LV, spacing mark
+    \x{1111}\x{ae4c}\x{1b04}\x{0711}\x{1b04}L, LV, spacing mark, extend, spacing mark
+    *Test CR, LF, and control
+    \x0d\x{0711}CR, extend
+    \x0d\x{1b04}CR, spacingmark
+    \x0a\x{0711}LF, extend
+    \x0a\x{1b04}LF, spacingmark
+    \x0b\x{0711}Control, extend
+    \x09\x{1b04}Control, spacingmark
+    *There are no Prepend characters, so we can't test Prepend, CR
+    
+/^(?>\X{2})X/8+
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+    
+/^\X{2,4}X/8+
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+
+/^\X{2,4}?X/8+
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+
+/-- --/
 
 /-- End of testinput6 --/
diff --git a/testdata/testinput7 b/testdata/testinput7
index 6ef9230..ddb5796 100644
--- a/testdata/testinput7
+++ b/testdata/testinput7
@@ -195,15 +195,6 @@ of case for anything other than the ASCII letters. --/
     \x{c0}
     \x{e0} 
 
-/-- This should be Perl-compatible but Perl 5.11 gets \x{300} wrong. --/8
-    
-/^\X/8
-    A
-    A\x{300}BC 
-    A\x{300}\x{301}\x{302}BC 
-    *** Failers
-    \x{300}  
-    
 /-- These are PCRE's extra properties to help with Unicodizing \d etc. --/
 
 /^\p{Xan}/8
diff --git a/testdata/testoutput10 b/testdata/testoutput10
index 5009d5d..da64233 100644
--- a/testdata/testoutput10
+++ b/testdata/testoutput10
@@ -90,7 +90,7 @@ No match
  9: **
 10: *
     \x{300}\x{301}\x{302}
-No match
+ 0: \x{300}\x{301}\x{302}
 
 /\X?abc/8
     abc
@@ -100,7 +100,7 @@ No match
     A\x{300}\x{301}\x{302}A\x{300}A\x{300}A\x{300}abcxyz
  0: A\x{300}abc
     \x{300}abc  
- 0: abc
+ 0: \x{300}abc
     *** Failers
 No match
 
@@ -114,7 +114,7 @@ No match
     A\x{300}\x{301}\x{302}A\x{300}A\x{300}A\x{300}abcxyz
 No match
     \x{300}abc  
-No match
+ 0: \x{300}abc
 
 /\X*abc/8
     abc
@@ -124,7 +124,7 @@ No match
     A\x{300}\x{301}\x{302}A\x{300}A\x{300}A\x{300}abcxyz
  0: A\x{300}\x{301}\x{302}A\x{300}A\x{300}A\x{300}abc
     \x{300}abc  
- 0: abc
+ 0: \x{300}abc
     *** Failers
 No match
 
@@ -138,7 +138,7 @@ No match
     *** Failers
 No match
     \x{300}abc  
-No match
+ 0: \x{300}abc
 
 /^\pL?=./8
     A=b
@@ -1133,7 +1133,7 @@ No match
     *** Failers
  0: *
     \x{300}  
-No match
+ 0: \x{300}
 
 /^[\X]/8
     X123
@@ -2100,4 +2100,172 @@ Partial match: AA
     AA\P\P 
 Partial match: AA
 
+/-- These are tests for extended grapheme clusters --/ 
+
+/^\X/8+
+    G\x{34e}\x{34e}X
+ 0: G\x{34e}\x{34e}
+ 0+ X
+    \x{34e}\x{34e}X
+ 0: \x{34e}\x{34e}
+ 0+ X
+    \x04X
+ 0: \x{04}
+ 0+ X
+    \x{1100}X
+ 0: \x{1100}
+ 0+ X
+    \x{1100}\x{34e}X
+ 0: \x{1100}\x{34e}
+ 0+ X
+    \x{1b04}\x{1b04}X 
+ 0: \x{1b04}\x{1b04}
+ 0+ X
+    *These match up to the roman letters
+ 0: *
+ 0+ These match up to the roman letters
+    \x{1111}\x{1111}L,L
+ 0: \x{1111}\x{1111}
+ 0+ L,L
+    \x{1111}\x{1111}\x{1169}L,L,V
+ 0: \x{1111}\x{1111}\x{1169}
+ 0+ L,L,V
+    \x{1111}\x{ae4c}L, LV
+ 0: \x{1111}\x{ae4c}
+ 0+ L, LV
+    \x{1111}\x{ad89}L, LVT
+ 0: \x{1111}\x{ad89}
+ 0+ L, LVT
+    \x{1111}\x{ae4c}\x{1169}L, LV, V
+ 0: \x{1111}\x{ae4c}\x{1169}
+ 0+ L, LV, V
+    \x{1111}\x{ae4c}\x{1169}\x{1169}L, LV, V, V
+ 0: \x{1111}\x{ae4c}\x{1169}\x{1169}
+ 0+ L, LV, V, V
+    \x{1111}\x{ae4c}\x{1169}\x{11fe}L, LV, V, T
+ 0: \x{1111}\x{ae4c}\x{1169}\x{11fe}
+ 0+ L, LV, V, T
+    \x{1111}\x{ad89}\x{11fe}L, LVT, T
+ 0: \x{1111}\x{ad89}\x{11fe}
+ 0+ L, LVT, T
+    \x{1111}\x{ad89}\x{11fe}\x{11fe}L, LVT, T, T
+ 0: \x{1111}\x{ad89}\x{11fe}\x{11fe}
+ 0+ L, LVT, T, T
+    \x{ad89}\x{11fe}\x{11fe}LVT, T, T
+ 0: \x{ad89}\x{11fe}\x{11fe}
+ 0+ LVT, T, T
+    *These match just the first codepoint (invalid sequence)
+ 0: *
+ 0+ These match just the first codepoint (invalid sequence)
+    \x{1111}\x{11fe}L, T
+ 0: \x{1111}
+ 0+ \x{11fe}L, T
+    \x{ae4c}\x{1111}LV, L
+ 0: \x{ae4c}
+ 0+ \x{1111}LV, L
+    \x{ae4c}\x{ae4c}LV, LV
+ 0: \x{ae4c}
+ 0+ \x{ae4c}LV, LV
+    \x{ae4c}\x{ad89}LV, LVT
+ 0: \x{ae4c}
+ 0+ \x{ad89}LV, LVT
+    \x{1169}\x{1111}V, L
+ 0: \x{1169}
+ 0+ \x{1111}V, L
+    \x{1169}\x{ae4c}V, LV
+ 0: \x{1169}
+ 0+ \x{ae4c}V, LV
+    \x{1169}\x{ad89}V, LVT
+ 0: \x{1169}
+ 0+ \x{ad89}V, LVT
+    \x{ad89}\x{1111}LVT, L
+ 0: \x{ad89}
+ 0+ \x{1111}LVT, L
+    \x{ad89}\x{1169}LVT, V
+ 0: \x{ad89}
+ 0+ \x{1169}LVT, V
+    \x{ad89}\x{ae4c}LVT, LV
+ 0: \x{ad89}
+ 0+ \x{ae4c}LVT, LV
+    \x{ad89}\x{ad89}LVT, LVT
+ 0: \x{ad89}
+ 0+ \x{ad89}LVT, LVT
+    \x{11fe}\x{1111}T, L
+ 0: \x{11fe}
+ 0+ \x{1111}T, L
+    \x{11fe}\x{1169}T, V
+ 0: \x{11fe}
+ 0+ \x{1169}T, V
+    \x{11fe}\x{ae4c}T, LV
+ 0: \x{11fe}
+ 0+ \x{ae4c}T, LV
+    \x{11fe}\x{ad89}T, LVT
+ 0: \x{11fe}
+ 0+ \x{ad89}T, LVT
+    *Test extend and spacing mark
+ 0: *
+ 0+ Test extend and spacing mark
+    \x{1111}\x{ae4c}\x{0711}L, LV, extend
+ 0: \x{1111}\x{ae4c}\x{711}
+ 0+ L, LV, extend
+    \x{1111}\x{ae4c}\x{1b04}L, LV, spacing mark
+ 0: \x{1111}\x{ae4c}\x{1b04}
+ 0+ L, LV, spacing mark
+    \x{1111}\x{ae4c}\x{1b04}\x{0711}\x{1b04}L, LV, spacing mark, extend, spacing mark
+ 0: \x{1111}\x{ae4c}\x{1b04}\x{711}\x{1b04}
+ 0+ L, LV, spacing mark, extend, spacing mark
+    *Test CR, LF, and control
+ 0: *
+ 0+ Test CR, LF, and control
+    \x0d\x{0711}CR, extend
+ 0: \x{0d}
+ 0+ \x{711}CR, extend
+    \x0d\x{1b04}CR, spacingmark
+ 0: \x{0d}
+ 0+ \x{1b04}CR, spacingmark
+    \x0a\x{0711}LF, extend
+ 0: \x{0a}
+ 0+ \x{711}LF, extend
+    \x0a\x{1b04}LF, spacingmark
+ 0: \x{0a}
+ 0+ \x{1b04}LF, spacingmark
+    \x0b\x{0711}Control, extend
+ 0: \x{0b}
+ 0+ \x{711}Control, extend
+    \x09\x{1b04}Control, spacingmark
+ 0: \x{09}
+ 0+ \x{1b04}Control, spacingmark
+    *There are no Prepend characters, so we can't test Prepend, CR
+ 0: *
+ 0+ There are no Prepend characters, so we can't test Prepend, CR
+    
+/^(?>\X{2})X/8+
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0: \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0+ 
+    
+/^\X{2,4}X/8+
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0: \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0+ 
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0: \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0+ 
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0: \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0+ 
+
+/^\X{2,4}?X/8+
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0: \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0+ 
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0: \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0+ 
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0: \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0+ 
+
+/-- --/
+
 /-- End of testinput10 --/ 
diff --git a/testdata/testoutput6 b/testdata/testoutput6
index e88fc09..c9b5808 100644
--- a/testdata/testoutput6
+++ b/testdata/testoutput6
@@ -696,7 +696,17 @@ No match
     A\x{300}\x{301}B\x{300}C\x{300}\x{301}DA\x{300}X
  0: A\x{300}\x{301}B\x{300}C
  1: C
-    
+
+/^\X/8
+    A
+ 0: A
+    A\x{300}BC 
+ 0: A\x{300}
+    A\x{300}\x{301}\x{302}BC 
+ 0: A\x{300}\x{301}\x{302}
+    \x{300}  
+ 0: \x{300}
+
 /^\p{Han}+/8
     \x{2e81}\x{3007}\x{2f804}\x{31a0}
  0: \x{2e81}\x{3007}\x{2f804}
@@ -1372,5 +1382,173 @@ No match
  0: \x{23a}
     ⱥ
  0: \x{2c65}
+    
+/-- These are tests for extended grapheme clusters --/ 
+
+/^\X/8+
+    G\x{34e}\x{34e}X
+ 0: G\x{34e}\x{34e}
+ 0+ X
+    \x{34e}\x{34e}X
+ 0: \x{34e}\x{34e}
+ 0+ X
+    \x04X
+ 0: \x{04}
+ 0+ X
+    \x{1100}X
+ 0: \x{1100}
+ 0+ X
+    \x{1100}\x{34e}X
+ 0: \x{1100}\x{34e}
+ 0+ X
+    \x{1b04}\x{1b04}X 
+ 0: \x{1b04}\x{1b04}
+ 0+ X
+    *These match up to the roman letters
+ 0: *
+ 0+ These match up to the roman letters
+    \x{1111}\x{1111}L,L
+ 0: \x{1111}\x{1111}
+ 0+ L,L
+    \x{1111}\x{1111}\x{1169}L,L,V
+ 0: \x{1111}\x{1111}\x{1169}
+ 0+ L,L,V
+    \x{1111}\x{ae4c}L, LV
+ 0: \x{1111}\x{ae4c}
+ 0+ L, LV
+    \x{1111}\x{ad89}L, LVT
+ 0: \x{1111}\x{ad89}
+ 0+ L, LVT
+    \x{1111}\x{ae4c}\x{1169}L, LV, V
+ 0: \x{1111}\x{ae4c}\x{1169}
+ 0+ L, LV, V
+    \x{1111}\x{ae4c}\x{1169}\x{1169}L, LV, V, V
+ 0: \x{1111}\x{ae4c}\x{1169}\x{1169}
+ 0+ L, LV, V, V
+    \x{1111}\x{ae4c}\x{1169}\x{11fe}L, LV, V, T
+ 0: \x{1111}\x{ae4c}\x{1169}\x{11fe}
+ 0+ L, LV, V, T
+    \x{1111}\x{ad89}\x{11fe}L, LVT, T
+ 0: \x{1111}\x{ad89}\x{11fe}
+ 0+ L, LVT, T
+    \x{1111}\x{ad89}\x{11fe}\x{11fe}L, LVT, T, T
+ 0: \x{1111}\x{ad89}\x{11fe}\x{11fe}
+ 0+ L, LVT, T, T
+    \x{ad89}\x{11fe}\x{11fe}LVT, T, T
+ 0: \x{ad89}\x{11fe}\x{11fe}
+ 0+ LVT, T, T
+    *These match just the first codepoint (invalid sequence)
+ 0: *
+ 0+ These match just the first codepoint (invalid sequence)
+    \x{1111}\x{11fe}L, T
+ 0: \x{1111}
+ 0+ \x{11fe}L, T
+    \x{ae4c}\x{1111}LV, L
+ 0: \x{ae4c}
+ 0+ \x{1111}LV, L
+    \x{ae4c}\x{ae4c}LV, LV
+ 0: \x{ae4c}
+ 0+ \x{ae4c}LV, LV
+    \x{ae4c}\x{ad89}LV, LVT
+ 0: \x{ae4c}
+ 0+ \x{ad89}LV, LVT
+    \x{1169}\x{1111}V, L
+ 0: \x{1169}
+ 0+ \x{1111}V, L
+    \x{1169}\x{ae4c}V, LV
+ 0: \x{1169}
+ 0+ \x{ae4c}V, LV
+    \x{1169}\x{ad89}V, LVT
+ 0: \x{1169}
+ 0+ \x{ad89}V, LVT
+    \x{ad89}\x{1111}LVT, L
+ 0: \x{ad89}
+ 0+ \x{1111}LVT, L
+    \x{ad89}\x{1169}LVT, V
+ 0: \x{ad89}
+ 0+ \x{1169}LVT, V
+    \x{ad89}\x{ae4c}LVT, LV
+ 0: \x{ad89}
+ 0+ \x{ae4c}LVT, LV
+    \x{ad89}\x{ad89}LVT, LVT
+ 0: \x{ad89}
+ 0+ \x{ad89}LVT, LVT
+    \x{11fe}\x{1111}T, L
+ 0: \x{11fe}
+ 0+ \x{1111}T, L
+    \x{11fe}\x{1169}T, V
+ 0: \x{11fe}
+ 0+ \x{1169}T, V
+    \x{11fe}\x{ae4c}T, LV
+ 0: \x{11fe}
+ 0+ \x{ae4c}T, LV
+    \x{11fe}\x{ad89}T, LVT
+ 0: \x{11fe}
+ 0+ \x{ad89}T, LVT
+    *Test extend and spacing mark
+ 0: *
+ 0+ Test extend and spacing mark
+    \x{1111}\x{ae4c}\x{0711}L, LV, extend
+ 0: \x{1111}\x{ae4c}\x{711}
+ 0+ L, LV, extend
+    \x{1111}\x{ae4c}\x{1b04}L, LV, spacing mark
+ 0: \x{1111}\x{ae4c}\x{1b04}
+ 0+ L, LV, spacing mark
+    \x{1111}\x{ae4c}\x{1b04}\x{0711}\x{1b04}L, LV, spacing mark, extend, spacing mark
+ 0: \x{1111}\x{ae4c}\x{1b04}\x{711}\x{1b04}
+ 0+ L, LV, spacing mark, extend, spacing mark
+    *Test CR, LF, and control
+ 0: *
+ 0+ Test CR, LF, and control
+    \x0d\x{0711}CR, extend
+ 0: \x{0d}
+ 0+ \x{711}CR, extend
+    \x0d\x{1b04}CR, spacingmark
+ 0: \x{0d}
+ 0+ \x{1b04}CR, spacingmark
+    \x0a\x{0711}LF, extend
+ 0: \x{0a}
+ 0+ \x{711}LF, extend
+    \x0a\x{1b04}LF, spacingmark
+ 0: \x{0a}
+ 0+ \x{1b04}LF, spacingmark
+    \x0b\x{0711}Control, extend
+ 0: \x{0b}
+ 0+ \x{711}Control, extend
+    \x09\x{1b04}Control, spacingmark
+ 0: \x{09}
+ 0+ \x{1b04}Control, spacingmark
+    *There are no Prepend characters, so we can't test Prepend, CR
+ 0: *
+ 0+ There are no Prepend characters, so we can't test Prepend, CR
+    
+/^(?>\X{2})X/8+
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0: \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0+ 
+    
+/^\X{2,4}X/8+
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0: \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0+ 
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0: \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0+ 
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0: \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0+ 
+
+/^\X{2,4}?X/8+
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0: \x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0+ 
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0: \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0+ 
+    \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0: \x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
+ 0+ 
+
+/-- --/
 
 /-- End of testinput6 --/
diff --git a/testdata/testoutput7 b/testdata/testoutput7
index ccfdad9..8c5bfb2 100644
--- a/testdata/testoutput7
+++ b/testdata/testoutput7
@@ -423,20 +423,6 @@ of case for anything other than the ASCII letters. --/
     \x{e0} 
  0: \x{e0}
 
-/-- This should be Perl-compatible but Perl 5.11 gets \x{300} wrong. --/8
-    
-/^\X/8
-    A
- 0: A
-    A\x{300}BC 
- 0: A\x{300}
-    A\x{300}\x{301}\x{302}BC 
- 0: A\x{300}\x{301}\x{302}
-    *** Failers
- 0: *
-    \x{300}  
-No match
-    
 /-- These are PCRE's extra properties to help with Unicodizing \d etc. --/
 
 /^\p{Xan}/8
@@ -1194,11 +1180,13 @@ No match
 
 /^S(\X*)e(\X*)$/8
     Stéréo
-No match
+ 0: Ste\x{301}re\x{301}o
+ 1: te\x{301}r
+ 2: \x{301}o
     
 /^\X/8 
     ́réo
-No match
+ 0: \x{301}
 
 /^a\X41z/<JS>
     aX41z