X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=utf8.c;h=0153fd6cd86858228ce7269f52b10395560cc3ef;hb=1be79c1b33149964355b8298e509bb8974fe3030;hp=4bb2e9bd788fd4ef190295e545271f0eabbeeb2d;hpb=0453d815b8a74697ff1e5451c27aba2fe537b8e0;p=p5sagit%2Fp5-mst-13.2.git

diff --git a/utf8.c b/utf8.c
index 4bb2e9b..0153fd6 100644
--- a/utf8.c
+++ b/utf8.c
@@ -68,8 +68,8 @@ Perl_uv_to_utf8(pTHX_ U8 *d, UV uv)
 	*d++ = (( uv        & 0x3f) | 0x80);
 	return d;
     }
-#ifdef Quad_t
-    if (uv < 0x2000000000)
+#ifdef HAS_QUAD
+    if (uv < 0x1000000000LL)
 #endif
     {
 	*d++ =                        0xfe;	/* Can't match U+FEFF! */
@@ -81,7 +81,7 @@ Perl_uv_to_utf8(pTHX_ U8 *d, UV uv)
 	*d++ = (( uv        & 0x3f) | 0x80);
 	return d;
     }
-#ifdef Quad_t
+#ifdef HAS_QUAD
     {
 	*d++ =                        0xff;	/* Can't match U+FFFE! */
 	*d++ = (((uv >> 36) & 0x3f) | 0x80);
@@ -261,6 +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];
@@ -277,6 +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];
@@ -309,6 +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];
@@ -316,6 +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)
 {
@@ -349,6 +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 */
@@ -361,6 +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 */
@@ -385,11 +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)
 {
@@ -408,7 +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)
 {
@@ -425,6 +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);
@@ -439,6 +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)
@@ -471,6 +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)
@@ -479,6 +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)