Revert "Avoid generating malformed UTF-8 and replacement characters b…

…y interpolating the variable as-is if it is not a valid UTF-8 sequence (nc), Github issue xslate#88"

This reverts commit 723f10dfc38202c863c1851a73228b03b8604aba.

MANIFEST

@@ -311,7 +311,6 @@ t/900_bugs/042_perl59_issue.t
 t/900_bugs/043_issue107.t
 t/900_bugs/044_empty_result.t
 t/900_bugs/045_issue130.t
-t/900_bugs/046_issue88.t
 t/900_bugs/issue79/tmpl/contentA.tt
 t/900_bugs/issue79/tmpl/contentB.tt
 t/900_bugs/issue79/tmpl/wrapperA.tt

lib/Text/Xslate/PP/State.pm

@@ -177,7 +177,7 @@ sub print {
         if(defined ${$sv}) {
             $st->{output} .=
                 (utf8::is_utf8($st->{output}) && !utf8::is_utf8(${$sv}))
-                 ? eval {$st->encoding->decode(${$sv}, Encode::FB_CROAK())} || ${$sv}
+                 ? $st->encoding->decode(${$sv})
                  : ${$sv};
         }
         else {
@@ -188,7 +188,7 @@ sub print {
         $sv =~ s/($Text::Xslate::PP::html_metachars)/$Text::Xslate::PP::html_escape{$1}/xmsgeo;
         $st->{output} .=
             (utf8::is_utf8($st->{output}) && !utf8::is_utf8($sv))
-             ? eval {$st->encoding->decode($sv, Encode::FB_CROAK())} || $sv
+             ? $st->encoding->decode($sv)
              : $sv;
     }
     else {

src/Text-Xslate.xs

@@ -541,39 +541,19 @@ tx_unmark_raw(pTHX_ SV* const str) {
 /* does sv_catsv_nomg(dest, src), but significantly faster */
 STATIC_INLINE void
 tx_sv_cat(pTHX_ SV* const dest, SV* const src) {
-    STRLEN len;
-    const char* pv = SvPV_const(src, len);
-
     if(!SvUTF8(dest) && SvUTF8(src)) {
         sv_utf8_upgrade(dest);
     }
 
-    if(SvUTF8(dest) == SvUTF8(src)
-       || is_utf8_string((const U8 *)pv, len)) {
+    {
+        STRLEN len;
+        const char* const pv  = SvPV_const(src, len);
         STRLEN const dest_cur = SvCUR(dest);
         char* const d         = SvGROW(dest, dest_cur + len + 1 /* count '\0' */);
 
         SvCUR_set(dest, dest_cur + len);
         Copy(pv, d + dest_cur, len + 1 /* copy '\0' */, char);
     }
-    else {
-        STRLEN const dest_cur = SvCUR(dest);
-        /* Longest UTF-8 representation of each char is 2 octets. */
-        char* const d_start   = SvGROW(dest, dest_cur + 2 * len + 1 /* count '\0' */);
-        char* d = d_start + dest_cur;
-
-        while(len--) {
-            const U8 c = *pv++;
-            if (UTF8_IS_INVARIANT(c)) {
-                *(d++) = c;
-            } else {
-                *(d++) = UTF8_EIGHT_BIT_HI(c);
-                *(d++) = UTF8_EIGHT_BIT_LO(c);
-            }
-        }
-        *d = '\0';
-        SvCUR_set(dest, d - d_start);
-    }
 }
 
 static void /* doesn't care about raw-ness */
@@ -583,8 +563,6 @@ tx_sv_cat_with_html_escape_force(pTHX_ SV* const dest, SV* const src) {
     const char* const end = cur + len;
     STRLEN const dest_cur = SvCUR(dest);
     char* d;
-    const U32 upgrade_on_copy = SvUTF8(dest) && !SvUTF8(src)
-        && !is_utf8_string((const U8 *)cur, len);
 
     (void)SvGROW(dest, dest_cur + ( len * ( sizeof(""") - 1) ) + 1);
     if(!SvUTF8(dest) && SvUTF8(src)) {
@@ -617,10 +595,6 @@ tx_sv_cat_with_html_escape_force(pTHX_ SV* const dest, SV* const src) {
             // CopyToken("'", d);
             CopyToken("'", d);
         }
-        else if (upgrade_on_copy && !UTF8_IS_INVARIANT(c)) {
-            *(d++) = UTF8_EIGHT_BIT_HI(c);
-            *(d++) = UTF8_EIGHT_BIT_LO(c);
-        }
         else {
             *(d++) = c;
         }