From 49cbf5aa10ecf06501e6a3094489c7c198fa491e Mon Sep 17 00:00:00 2001
From: Klaus Crusius <KlausC@users.noreply.github.com>
Date: Fri, 29 Sep 2017 17:40:04 +0200
Subject: [PATCH] allow non-ascii characters in printf format strings (#23899)

fixes issue #23880
---
 base/printf.jl | 15 ++++++++-------
 test/printf.jl | 15 +++++++++++++++
 2 files changed, 23 insertions(+), 7 deletions(-)

diff --git a/base/printf.jl b/base/printf.jl
index 2f8347f6f8a95..c6e121405f02f 100644
--- a/base/printf.jl
+++ b/base/printf.jl
@@ -36,23 +36,24 @@ end
 ### printf format string parsing ###
 
 function parse(s::AbstractString)
-    # parse format string in to stings and format tuples
+    # parse format string into strings and format tuples
     list = []
     i = j = start(s)
+    j1 = 0 # invariant: j1 == prevind(s, j)
     while !done(s,j)
         c, k = next(s,j)
         if c == '%'
-            isempty(s[i:j-1]) || push!(list, s[i:j-1])
+            i > j1 || push!(list, s[i:j1])
             flags, width, precision, conversion, k = parse1(s,k)
             '\'' in flags && error("printf format flag ' not yet supported")
             conversion == 'n'    && error("printf feature %n not supported")
             push!(list, conversion == '%' ? "%" : (flags,width,precision,conversion))
-            i = j = k
-        else
-            j = k
+            i = k
         end
+        j1 = j
+        j = k
     end
-    isempty(s[i:end]) || push!(list, s[i:end])
+    i > endof(s) || push!(list, s[i:end])
     # coalesce adjacent strings
     i = 1
     while i < length(list)
@@ -98,7 +99,7 @@ function parse1(s::AbstractString, k::Integer)
     while c in "#0- + '"
         c, k = next_or_die(s,k)
     end
-    flags = String(s[j:k-2])
+    flags = String(s[j:prevind(s,k)-1]) # exploiting that all flags are one-byte.
     # parse width
     while '0' <= c <= '9'
         width = 10*width + c-'0'
diff --git a/test/printf.jl b/test/printf.jl
index 90c0e10769621..0674d172d66d1 100644
--- a/test/printf.jl
+++ b/test/printf.jl
@@ -272,3 +272,18 @@ end
 
 # Check bug with trailing nul printing BigFloat
 @test (@sprintf("%.330f", BigFloat(1)))[end] != '\0'
+
+# Check utf8 strings #23880
+@test (@sprintf("X%d", 2)) == "X2"
+@test (@sprintf("\u00d0%d", 2)) == "\u00d02"
+@test (@sprintf("\u0f00%d", 2)) == "\u0f002"
+@test (@sprintf("\U0001ffff%d", 2)) == "\U0001ffff2"
+@test (@sprintf("%dX%d", 1, 2)) == "1X2"
+@test (@sprintf("%d\u00d0%d", 1, 2)) == "1\u00d02"
+@test (@sprintf("%d\u0f00%d", 1, 2)) == "1\u0f002"
+@test (@sprintf("%d\U0001ffff%d", 1, 2)) == "1\U0001ffff2"
+@test (@sprintf("%d\u2203%d\u0203", 1, 2)) == "1\u22032\u0203"
+@test_throws ArgumentError @macroexpand(@sprintf("%y%d", 1, 2))
+@test_throws ArgumentError @macroexpand(@sprintf("%\u00d0%d", 1, 2))
+@test_throws ArgumentError @macroexpand(@sprintf("%\u0f00%d", 1, 2))
+@test_throws ArgumentError @macroexpand(@sprintf("%\U0001ffff%d", 1, 2))