[tarantool-patches] [PATCH] Extend range of printable unicode characters

[Ivan Koptelov]

Before the patch unicode characters encoded with 4 bytes
were always treated as non-printable and displayed as byte
sequences (with 'binary' tag).
With the patch, range of printable characters is extended and
include characters encoded with 4 bytes.
Currently it is: (old printable range) U (icu printable range).
Corresponding changes are also made in tarantool/libyaml.

Closes: #4090
---
Branch https://github.com/tarantool/tarantool/tree/sudobobo/gh-4090-valid-chars-displ-bin
Issue https://github.com/tarantool/tarantool/issues/4090

 src/lib/core/CMakeLists.txt |  2 +-
 src/lib/core/util.c         |  5 ++++-
 test/app-tap/yaml.test.lua  | 10 +++++++++-
 3 files changed, 14 insertions(+), 3 deletions(-)

diff --git a/src/lib/core/CMakeLists.txt b/src/lib/core/CMakeLists.txt
index 66e430a25..4b96d1105 100644
--- a/src/lib/core/CMakeLists.txt
+++ b/src/lib/core/CMakeLists.txt
@@ -40,7 +40,7 @@ add_library(core STATIC ${core_sources})
 
 target_link_libraries(core salad small uri decNumber ${LIBEV_LIBRARIES}
                       ${LIBEIO_LIBRARIES} ${LIBCORO_LIBRARIES}
-                      ${MSGPUCK_LIBRARIES})
+                      ${MSGPUCK_LIBRARIES} ${ICU_LIBRARIES})
 
 if (ENABLE_BACKTRACE AND NOT TARGET_OS_DARWIN)
     target_link_libraries(core gcc_s ${UNWIND_LIBRARIES})
diff --git a/src/lib/core/util.c b/src/lib/core/util.c
index 9458695b9..4ca99e177 100644
--- a/src/lib/core/util.c
+++ b/src/lib/core/util.c
@@ -41,6 +41,8 @@
 #include <unistd.h>
 #include <limits.h>
 
+#include <unicode/utf8.h>
+#include <unicode/uchar.h>
 #include <msgpuck/msgpuck.h> /* mp_char2escape[] table */
 
 #include "say.h"
@@ -271,7 +273,8 @@ utf8_check_printable(const char *start, size_t length)
 		      (pointer[0] == 0xEF &&
 		       !(pointer[1] == 0xBB && pointer[2] == 0xBF) &&
 		       !(pointer[1] == 0xBF &&
-			 (pointer[2] == 0xBE || pointer[2] == 0xBF)))
+			 (pointer[2] == 0xBE || pointer[2] == 0xBF))) ||
+		      (u_isprint(value))
 		      )
 		    ) {
 			return 0;
diff --git a/test/app-tap/yaml.test.lua b/test/app-tap/yaml.test.lua
index c0eecea03..4669b6102 100755
--- a/test/app-tap/yaml.test.lua
+++ b/test/app-tap/yaml.test.lua
@@ -42,7 +42,7 @@ local function test_compact(test, s)
 end
 
 local function test_output(test, s)
-    test:plan(12)
+    test:plan(17)
     test:is(s.encode({true}), '---\n- true\n...\n', "encode for true")
     test:is(s.decode("---\nyes\n..."), true, "decode for 'yes'")
     test:is(s.encode({false}), '---\n- false\n...\n', "encode for false")
@@ -55,6 +55,14 @@ local function test_output(test, s)
         "encode for binary (2) - gh-354")
     test:is(s.encode("\xe0\x82\x85\x00"), '--- !!binary 4IKFAA==\n...\n',
         "encode for binary (3) - gh-1302")
+    -- gh-4090: some printable unicode characters displayed as byte sequences.
+    -- The following tests ensures that various 4-byte encoded unicode characters
+    -- displayed as expected.
+    test:is(s.encode("\xF0\x9F\x86\x98"), '--- 🆘\n...\n', "encode - gh-4090 (1)")
+    test:is(s.encode("\xF0\x9F\x84\xBD"), '--- 🄽\n...\n', "encode - gh-4090 (2)")
+    test:is(s.encode("\xF0\x9F\x85\xA9"), '--- 🅩\n...\n', "encode - gh-4090 (3)")
+    test:is(s.encode("\xF0\x9F\x87\xA6"), '--- 🇦\n...\n', "encode - gh-4090 (4)")
+    test:is(s.encode("\xF0\x9F\x88\xB2"), '--- 🈲\n...\n', "encode - gh-4090 (5)")
     -- gh-883: console can hang tarantool process
     local t = {}
     for i=0x8000,0xffff,1 do
-- 
2.20.1