X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=utf8.c;h=a4703769dea371e541da418a8f4f0e858c85f340;hb=cfcb0b09a4c7cff7274c3b329d4dc27f192b7434;hp=8c7aee2d8945b05346a00202a50f83473bee18c2;hpb=cea2e8a9dd23747fd2b66edc86c58c64e9970321;p=p5sagit%2Fp5-mst-13.2.git

diff --git a/utf8.c b/utf8.c
index 8c7aee2..a470376 100644
--- a/utf8.c
+++ b/utf8.c
@@ -107,7 +107,9 @@ Perl_utf8_to_uv(pTHX_ U8* s, I32* retlen)
 	return *s;
     }
     if (!(uv & 0x40)) {
-	Perl_warn(aTHX_ "Malformed UTF-8 character");
+        dTHR;
+	if (ckWARN_d(WARN_UTF8))     
+	    Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
 	if (retlen)
 	    *retlen = 1;
 	return *s;
@@ -127,7 +129,9 @@ Perl_utf8_to_uv(pTHX_ U8* s, I32* retlen)
     s++;
     while (len--) {
 	if ((*s & 0xc0) != 0x80) {
-	    Perl_warn(aTHX_ "Malformed UTF-8 character");
+            dTHR;
+	    if (ckWARN_d(WARN_UTF8))     
+	        Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
 	    if (retlen)
 		*retlen -= len + 1;
 	    return 0xfffd;
@@ -203,9 +207,11 @@ Perl_utf16_to_utf8(pTHX_ U16* p, U8* d, I32 bytelen)
 	    continue;
 	}
 	if (uv >= 0xd800 && uv < 0xdbff) {	/* surrogates */
+            dTHR;
 	    int low = *p++;
 	    if (low < 0xdc00 || low >= 0xdfff) {
-		Perl_warn(aTHX_ "Malformed UTF-16 surrogate");
+		if (ckWARN_d(WARN_UTF8))     
+	    	    Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-16 surrogate");
 		p--;
 		uv = 0xfffd;
 	    }
@@ -255,6 +261,14 @@ Perl_is_uni_alnum(pTHX_ U32 c)
 }
 
 bool
+Perl_is_uni_alnumc(pTHX_ U32 c)
+{
+    U8 tmpbuf[10];
+    uv_to_utf8(tmpbuf, (UV)c);
+    return is_utf8_alnumc(tmpbuf);
+}
+
+bool
 Perl_is_uni_idfirst(pTHX_ U32 c)
 {
     U8 tmpbuf[10];
@@ -271,6 +285,14 @@ Perl_is_uni_alpha(pTHX_ U32 c)
 }
 
 bool
+Perl_is_uni_ascii(pTHX_ U32 c)
+{
+    U8 tmpbuf[10];
+    uv_to_utf8(tmpbuf, (UV)c);
+    return is_utf8_ascii(tmpbuf);
+}
+
+bool
 Perl_is_uni_space(pTHX_ U32 c)
 {
     U8 tmpbuf[10];
@@ -303,6 +325,22 @@ Perl_is_uni_lower(pTHX_ U32 c)
 }
 
 bool
+Perl_is_uni_cntrl(pTHX_ U32 c)
+{
+    U8 tmpbuf[10];
+    uv_to_utf8(tmpbuf, (UV)c);
+    return is_utf8_cntrl(tmpbuf);
+}
+
+bool
+Perl_is_uni_graph(pTHX_ U32 c)
+{
+    U8 tmpbuf[10];
+    uv_to_utf8(tmpbuf, (UV)c);
+    return is_utf8_graph(tmpbuf);
+}
+
+bool
 Perl_is_uni_print(pTHX_ U32 c)
 {
     U8 tmpbuf[10];
@@ -310,6 +348,22 @@ Perl_is_uni_print(pTHX_ U32 c)
     return is_utf8_print(tmpbuf);
 }
 
+bool
+Perl_is_uni_punct(pTHX_ U32 c)
+{
+    U8 tmpbuf[10];
+    uv_to_utf8(tmpbuf, (UV)c);
+    return is_utf8_punct(tmpbuf);
+}
+
+bool
+Perl_is_uni_xdigit(pTHX_ U32 c)
+{
+    U8 tmpbuf[10];
+    uv_to_utf8(tmpbuf, (UV)c);
+    return is_utf8_xdigit(tmpbuf);
+}
+
 U32
 Perl_to_uni_upper(pTHX_ U32 c)
 {
@@ -343,6 +397,12 @@ Perl_is_uni_alnum_lc(pTHX_ U32 c)
 }
 
 bool
+Perl_is_uni_alnumc_lc(pTHX_ U32 c)
+{
+    return is_uni_alnumc(c);	/* XXX no locale support yet */
+}
+
+bool
 Perl_is_uni_idfirst_lc(pTHX_ U32 c)
 {
     return is_uni_idfirst(c);	/* XXX no locale support yet */
@@ -355,6 +415,12 @@ Perl_is_uni_alpha_lc(pTHX_ U32 c)
 }
 
 bool
+Perl_is_uni_ascii_lc(pTHX_ U32 c)
+{
+    return is_uni_ascii(c);	/* XXX no locale support yet */
+}
+
+bool
 Perl_is_uni_space_lc(pTHX_ U32 c)
 {
     return is_uni_space(c);	/* XXX no locale support yet */
@@ -379,11 +445,35 @@ Perl_is_uni_lower_lc(pTHX_ U32 c)
 }
 
 bool
+Perl_is_uni_cntrl_lc(pTHX_ U32 c)
+{
+    return is_uni_cntrl(c);	/* XXX no locale support yet */
+}
+
+bool
+Perl_is_uni_graph_lc(pTHX_ U32 c)
+{
+    return is_uni_graph(c);	/* XXX no locale support yet */
+}
+
+bool
 Perl_is_uni_print_lc(pTHX_ U32 c)
 {
     return is_uni_print(c);	/* XXX no locale support yet */
 }
 
+bool
+Perl_is_uni_punct_lc(pTHX_ U32 c)
+{
+    return is_uni_punct(c);	/* XXX no locale support yet */
+}
+
+bool
+Perl_is_uni_xdigit_lc(pTHX_ U32 c)
+{
+    return is_uni_xdigit(c);	/* XXX no locale support yet */
+}
+
 U32
 Perl_to_uni_upper_lc(pTHX_ U32 c)
 {
@@ -402,7 +492,6 @@ Perl_to_uni_lower_lc(pTHX_ U32 c)
     return to_uni_lower(c);	/* XXX no locale support yet */
 }
 
-
 bool
 Perl_is_utf8_alnum(pTHX_ U8 *p)
 {
@@ -419,6 +508,21 @@ Perl_is_utf8_alnum(pTHX_ U8 *p)
 }
 
 bool
+Perl_is_utf8_alnumc(pTHX_ U8 *p)
+{
+    if (!PL_utf8_alnum)
+	PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
+    return swash_fetch(PL_utf8_alnum, p);
+/*    return is_utf8_alpha(p) || is_utf8_digit(p); */
+#ifdef SURPRISINGLY_SLOWER  /* probably because alpha is usually true */
+    if (!PL_utf8_alnum)
+	PL_utf8_alnum = swash_init("utf8", "",
+	    sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0);
+    return swash_fetch(PL_utf8_alnum, p);
+#endif
+}
+
+bool
 Perl_is_utf8_idfirst(pTHX_ U8 *p)
 {
     return *p == '_' || is_utf8_alpha(p);
@@ -433,6 +537,14 @@ Perl_is_utf8_alpha(pTHX_ U8 *p)
 }
 
 bool
+Perl_is_utf8_ascii(pTHX_ U8 *p)
+{
+    if (!PL_utf8_ascii)
+	PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
+    return swash_fetch(PL_utf8_ascii, p);
+}
+
+bool
 Perl_is_utf8_space(pTHX_ U8 *p)
 {
     if (!PL_utf8_space)
@@ -465,6 +577,22 @@ Perl_is_utf8_lower(pTHX_ U8 *p)
 }
 
 bool
+Perl_is_utf8_cntrl(pTHX_ U8 *p)
+{
+    if (!PL_utf8_cntrl)
+	PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
+    return swash_fetch(PL_utf8_cntrl, p);
+}
+
+bool
+Perl_is_utf8_graph(pTHX_ U8 *p)
+{
+    if (!PL_utf8_graph)
+	PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
+    return swash_fetch(PL_utf8_graph, p);
+}
+
+bool
 Perl_is_utf8_print(pTHX_ U8 *p)
 {
     if (!PL_utf8_print)
@@ -473,6 +601,22 @@ Perl_is_utf8_print(pTHX_ U8 *p)
 }
 
 bool
+Perl_is_utf8_punct(pTHX_ U8 *p)
+{
+    if (!PL_utf8_punct)
+	PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
+    return swash_fetch(PL_utf8_punct, p);
+}
+
+bool
+Perl_is_utf8_xdigit(pTHX_ U8 *p)
+{
+    if (!PL_utf8_xdigit)
+	PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
+    return swash_fetch(PL_utf8_xdigit, p);
+}
+
+bool
 Perl_is_utf8_mark(pTHX_ U8 *p)
 {
     if (!PL_utf8_mark)