WIP

perlpunk · Dec 31, 2023 · 48f4dde · 48f4dde
1 parent 79511dd
commit 48f4dde
Show file tree

Hide file tree

Showing 3 changed files with 29 additions and 10 deletions.
diff --git a/LibYAML/XS.xs b/LibYAML/XS.xs
@@ -124,6 +124,7 @@ emit_string_events(AV *perl_events, HV *options)
         yaml_emitter_t emitter;
         SV **val;
         SV *yaml = newSVpvn("", 0);
+        int unicode = 1;
 
         XCPT_TRY_START
         {
@@ -140,7 +141,14 @@ emit_string_events(AV *perl_events, HV *options)
             }
             yaml_emitter_set_output(&emitter, &append_output, (void *) yaml);
             yaml_emitter_set_canonical(&emitter, 0);
-            yaml_emitter_set_unicode(&emitter, 0);
+            yaml_emitter_set_unicode(&emitter, 1);
+            val = hv_fetch(options, "unicode", 7, TRUE);
+            if (val && SvOK(*val) && SvTRUE(*val)) {
+                unicode = 1;
+            }
+            else if (val && SvOK(*val) && ! SvTRUE(*val)) {
+                unicode = 0;
+            }
 
             emit_events(&emitter, perl_events);
 
@@ -155,7 +163,13 @@ emit_string_events(AV *perl_events, HV *options)
         }
 
         if (yaml) {
-            SvUTF8_off(yaml);
+            if (unicode) {
+                SvUTF8_off(yaml);
+            }
+            else {
+                (void)sv_utf8_decode(yaml);
+                SvUTF8_on(yaml);
+            }
         }
         RETVAL = yaml;
 

diff --git a/LibYAML/etc/perl_libyaml.c b/LibYAML/etc/perl_libyaml.c
@@ -324,7 +324,7 @@ perl_to_libyaml_event(yaml_emitter_t *emitter, HV *perl_event)
         }
 
         if (strEQ(type, "stream_start_event")) {
-            ok = yaml_stream_start_event_initialize(&event, 0);
+            ok = yaml_stream_start_event_initialize(&event, YAML_UTF8_ENCODING);
         }
         else if (strEQ(type, "stream_end_event")) {
             ok = yaml_stream_end_event_initialize(&event);
@@ -385,9 +385,10 @@ perl_to_libyaml_event(yaml_emitter_t *emitter, HV *perl_event)
             else {
                 croak("%s\n", "scalar value not defined");
             }
+
             ok = yaml_scalar_event_initialize(
                 &event, anchor_name, tag_name,
-                (unsigned char *) scalar_value, strlen(scalar_value), plain_implicit, quoted_implicit, style);
+                (unsigned char *) scalar_value, len, plain_implicit, quoted_implicit, style);
         }
         else if (strEQ(type, "alias_event")) {
             val = hv_fetch(perl_event, "value", 5, TRUE);

diff --git a/t/10.basic.t b/t/10.basic.t
@@ -269,16 +269,20 @@ subtest unicode => sub {
     my $ev = [];
     $yaml = "- ö";
     YAML::LibYAML::API::parse_string_events($yaml, $ev);
+    use Devel::Peek;
     my $value = encode_utf8 $ev->[3]->{value};
     cmp_ok($value, 'eq', "ö", "utf8 parse");
 
     $ev->[3]->{value} = decode_utf8 "ä";
-    my $dump = YAML::LibYAML::API::emit_string_events($ev);
-    cmp_ok($dump, '=~', qr{- "\\xE4"}i, "utf8 emit");
-
-    $ev->[3]->{value} = "\303\274 \303\300";
-    $dump = YAML::LibYAML::API::emit_string_events($ev);
-    cmp_ok($dump, '=~', qr{- "\\xC3\\xBC \\xC3\\xC0"}i, "binary emit");
+    Dump $ev->[3]->{value};
+    my $dump = YAML::LibYAML::API::emit_string_events($ev, { unicode => 0 });
+    Dump $dump;
+    cmp_ok($dump, '=~', qr{- \xe4}i, "utf8 decoded emit");
+
+#    $ev->[3]->{value} = "\303\274 \303\300";
+    $dump = YAML::LibYAML::API::emit_string_events($ev, { unicode => 1 });
+    cmp_ok($dump, '=~', qr{- ä}i, "utf8 emit");
+#    cmp_ok($dump, '=~', qr{- "\\xC3\\xBC \\xC3\\xC0"}i, "binary emit");
 };
 
 subtest indent => sub {