bytes<->utf8 fixes

Message-ID: <slrn8ldoih.fbd.simon@justanother.perlhacker.org>

p4raw-id: //depot/cfgperl@6242

diff --git a/embed.pl b/embed.pl
index 419ae4e..5b63a35 100755 (executable)
--- a/embed.pl
+++ b/embed.pl
@@ -2063,7 +2063,7 @@ Ap        |U8*    |utf16_to_utf8_reversed|U16* p|U8 *d|I32 bytelen
 Ap     |I32    |utf8_distance  |U8 *a|U8 *b
 Ap     |U8*    |utf8_hop       |U8 *s|I32 off
 Ap     |U8*    |utf8_to_bytes  |U8 *s|STRLEN len
-Ap     |U8*    |bytes_to_utf8  |U8 *s|STRLEN len
+Ap     |U8*    |bytes_to_utf8  |U8 *s|STRLEN *len
 Ap     |UV     |utf8_to_uv     |U8 *s|I32* retlen
 Ap     |U8*    |uv_to_utf8     |U8 *d|UV uv
 p      |void   |vivify_defelem |SV* sv

diff --git a/perlapi.c b/perlapi.c
index a911be2..ccb7c8f 100755 (executable)
--- a/perlapi.c
+++ b/perlapi.c
@@ -3352,7 +3352,7 @@ Perl_utf8_to_bytes(pTHXo_ U8 *s, STRLEN len)
 
 #undef  Perl_bytes_to_utf8
 U8*
-Perl_bytes_to_utf8(pTHXo_ U8 *s, STRLEN len)
+Perl_bytes_to_utf8(pTHXo_ U8 *s, STRLEN *len)
 {
     return ((CPerlObj*)pPerl)->Perl_bytes_to_utf8(s, len);
 }

diff --git a/pod/perlapi.pod b/pod/perlapi.pod
index f274641..1e48809 100644 (file)
--- a/pod/perlapi.pod
+++ b/pod/perlapi.pod
@@ -153,9 +153,10 @@ Found in file av.c
 =item bytes_to_utf8
 
 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
-Returns a pointer to the newly-created string.
+Returns a pointer to the newly-created string, and sets C<len> to
+reflect the new length.
 
-       U8 *    bytes_to_utf8(U8 *s, STRLEN len)
+       U8 *    bytes_to_utf8(U8 *s, STRLEN *len)
 
 =for hackers
 Found in file utf8.c
@@ -2942,6 +2943,7 @@ Found in file handy.h
 
 Converts a string C<s> of length C<len> from UTF8 into ASCII encoding.
 Unlike C<bytes_to_utf8>, this over-writes the original string.
+Returns zero on failure after converting as much as possible.
 
        U8 *    utf8_to_bytes(U8 *s, STRLEN len)
 

diff --git a/proto.h b/proto.h
index 2171aa1..c1d9a66 100644 (file)
--- a/proto.h
+++ b/proto.h
@@ -810,7 +810,7 @@ PERL_CALLCONV U8*   Perl_utf16_to_utf8_reversed(pTHX_ U16* p, U8 *d, I32 bytelen);
 PERL_CALLCONV I32      Perl_utf8_distance(pTHX_ U8 *a, U8 *b);
 PERL_CALLCONV U8*      Perl_utf8_hop(pTHX_ U8 *s, I32 off);
 PERL_CALLCONV U8*      Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN len);
-PERL_CALLCONV U8*      Perl_bytes_to_utf8(pTHX_ U8 *s, STRLEN len);
+PERL_CALLCONV U8*      Perl_bytes_to_utf8(pTHX_ U8 *s, STRLEN *len);
 PERL_CALLCONV UV       Perl_utf8_to_uv(pTHX_ U8 *s, I32* retlen);
 PERL_CALLCONV U8*      Perl_uv_to_utf8(pTHX_ U8 *d, UV uv);
 PERL_CALLCONV void     Perl_vivify_defelem(pTHX_ SV* sv);

diff --git a/utf8.c b/utf8.c
index b77cfdc..9bb89a4 100644 (file)
--- a/utf8.c
+++ b/utf8.c
@@ -227,6 +227,7 @@ Perl_utf8_hop(pTHX_ U8 *s, I32 off)
 
 Converts a string C<s> of length C<len> from UTF8 into ASCII encoding.
 Unlike C<bytes_to_utf8>, this over-writes the original string.
+Returns zero on failure after converting as much as possible.
 
 =cut
 */
@@ -247,6 +248,10 @@ Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN len)
         else {
             I32 ulen;
             UV uv = utf8_to_uv(s, &ulen);
+            if (uv > 255) {
+                *d = '\0';
+                return 0;
+            }
             s += ulen;
             *d++ = (U8)uv;
         }
@@ -256,24 +261,25 @@ Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN len)
 }
 
 /*
-=for apidoc Am|U8 *|bytes_to_utf8|U8 *s|STRLEN len
+=for apidoc Am|U8 *|bytes_to_utf8|U8 *s|STRLEN *len
 
 Converts a string C<s> of length C<len> from ASCII into UTF8 encoding.
-Returns a pointer to the newly-created string.
+Returns a pointer to the newly-created string, and sets C<len> to
+reflect the new length.
 
 =cut
 */
 
 U8*
-Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN len)
+Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len)
 {
     dTHR;
     U8 *send;
     U8 *d;
     U8 *dst;
-    send = s + len;
+    send = s + (*len);
 
-    Newz(801, d, len * 2 + 1, U8);
+    Newz(801, d, (*len) * 2 + 1, U8);
     dst = d;
 
     while (s < send) {
@@ -286,6 +292,7 @@ Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN len)
         }
     }
     *d = '\0';
+    *len = d-dst;
     return dst;
 }