Fix #354: yaml incorrectly escapes special characters in a string

test/app/yaml.result

@@ -271,7 +271,7 @@ ok - ucdata
     # compact: end
 ok - compact
     # output
-    1..8
+    1..9
     ok - encode for true
     ok - decode for 'yes'
     ok - encode for false
@@ -279,6 +279,7 @@ ok - compact
     ok - encode for nil
     ok - decode for ~
     ok - encode for binary
+    ok - encode for binary (2) - gh-354
     ok - tutorial string
     # output: end
 ok - output

test/app/yaml.test.lua

@@ -42,7 +42,7 @@ local function test_compact(test, s)
 end
 
 local function test_output(test, s)
-    test:plan(8)
+    test:plan(9)
     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")
@@ -51,6 +51,8 @@ local function test_output(test, s)
     test:is(s.decode("---\n~\n..."), s.NULL, "decode for ~")
     test:is(s.encode("\x80\x92\xe8s\x16"), '--- !!binary gJLocxY=\n...\n',
         "encode for binary")
+    test:is(s.encode("\x08\x5c\xc2\x80\x12\x2f"), '--- !!binary CFzCgBIv\n...\n',
+        "encode for binary (2) - gh-354")
     test:is(s.encode("Tutorial -- Header\n====\n\nText"),
         "--- |-\n  Tutorial -- Header\n  ====\n\n  Text\n...\n", "tutorial string");
 end

third_party/lua-yaml/lyaml.cc

@@ -44,6 +44,7 @@ extern "C" {
 #include <lj_state.h>
 
 #include "yaml.h"
+#include "yaml_private.h"
 #include "b64.h"
 } /* extern "C" */
 #include "lua/utils.h"
@@ -515,8 +516,24 @@ static yaml_scalar_style_t analyze_string(struct lua_yaml_dumper *dumper,
             return YAML_PLAIN_SCALAR_STYLE;
          }
 
+         /* Truncated UTF-8 sequence? */
+         if (p + continuation_bytes >= e) {
+            *is_binary = 1;
+            return YAML_PLAIN_SCALAR_STYLE;
+         }
+
+         /*
+          * Valid non-printable UTF-8? Encode as binary since otherwise
+          * the conversion may be lossy, ex: '\xc2\x80' -> '\x80'.
+          */
+         yaml_string_t ys; ys.pointer = (yaml_char_t *)p;
+         if (!IS_PRINTABLE(ys)) {
+            *is_binary = 1;
+            return YAML_PLAIN_SCALAR_STYLE;
+         }
+
          ++p;
-         while (p < e  && continuation_bytes > 0 && *p >= 0x80 && *p <= 0xBF) {
+         while (continuation_bytes > 0 && *p >= 0x80 && *p <= 0xBF) {
             ++p;
             continuation_bytes -= 1;
          }