From 475219957807032ab27f0f1bffbce971b89a293d Mon Sep 17 00:00:00 2001
From: Matthew Flatt <mflatt@racket-lang.org>
Date: Sun, 20 Jan 2019 16:32:47 -0700
Subject: [PATCH] fix `string-titlecase` on special-casing characters

original commit: 0ae58f75bcf76fb41e571579a81ab2080379455c
---
 LOG         | 3 +++
 mats/5_4.ms | 4 ++++
 s/5_4.ss    | 8 +++++---
 3 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/LOG b/LOG
index e5aae978a0..1e0276354a 100644
--- a/LOG
+++ b/LOG
@@ -1027,3 +1027,6 @@
     cpnanopass.ss, x86_64.ss, x86.ss, foreign2.c, foreign.ms
 - added initialization of seginfo sorted and trigger_ephemerons fields.
     segment.c
+- fix `string-titlecase` on special-casing characters like #\xDF and
+   #\xFB00
+    5_4.ss, 5_4.ms
diff --git a/mats/5_4.ms b/mats/5_4.ms
index e9535566a2..74592dd1f7 100644
--- a/mats/5_4.ms
+++ b/mats/5_4.ms
@@ -1162,6 +1162,10 @@
   (test (string-titlecase "r6rs") "R6rs") ; this example appears to be wrong in R6RS (Sept 2007 version)
   (test (string-titlecase "R6RS") "R6rs") ; this one, too
 
+  (test (string-titlecase "stra \xDF;e") "Stra Sse")
+  (test (string-titlecase "one \xFB00;e") "One Ffe")
+  (test (string-titlecase "one\xFB00;e") "One\xFB00;e")
+
   (test (string-downcase "A\x3A3;'x") "a\x3C3;'x") ; ' is a MidLetter
 
   (test (string-ci<? "a" "Z") #t)
diff --git a/s/5_4.ss b/s/5_4.ss
index b92e339a02..3a2d1988f6 100644
--- a/s/5_4.ss
+++ b/s/5_4.ss
@@ -403,9 +403,11 @@
           (define (trans s i c seen-cased? ac)
             (if seen-cased?
                 (trans1 s i ($str-downcase c) #t ac)
-                (if ($char-cased? c)
-                    (trans1 s i ($str-titlecase c) #t ac)
-                    (trans1 s i c #f ac))))
+                (let ([tc ($str-titlecase c)])
+                  (if (or ($char-cased? c)
+                          (not (eqv? c tc)))
+                      (trans1 s i tc #t ac)
+                      (trans1 s i c #f ac)))))
           ; NB: if used as a pattern for word breaking, take care not to break between CR & LF (WB3)
           ; NB: and between regional-indicators (WB13c).  also take care not to let handling of WB6 and
           ; NB: WB7 here prevent breaks in, e.g., "a." when not followed by, e.g., another letter.