[ID 20000626.007] h2xs man page contains trailing garbage

[p5sagit/p5-mst-13.2.git] / utf8.c

diff --git a/utf8.c b/utf8.c
index 212c555..b77cfdc 100644 (file)
--- a/utf8.c
+++ b/utf8.c
@@ -101,6 +101,39 @@ Perl_uv_to_utf8(pTHX_ U8 *d, UV uv)
 #endif
 }
 
+/* Tests if some arbitrary number of bytes begins in a valid UTF-8 character.
+ * The actual number of bytes in the UTF-8 character will be returned if it
+ * is valid, otherwise 0. */
+int
+Perl_is_utf8_char(pTHX_ U8 *s)
+{
+    U8 u = *s;
+    int slen, len;
+
+    if (!(u & 0x80))
+       return 1;
+
+    if (!(u & 0x40))
+       return 0;
+
+    if      (!(u & 0x20))      { len = 2; }
+    else if (!(u & 0x10))      { len = 3; }
+    else if (!(u & 0x08))      { len = 4; }
+    else if (!(u & 0x04))      { len = 5; }
+    else if (!(u & 0x02))      { len = 6; }
+    else if (!(u & 0x01))      { len = 7; }
+    else                       { len = 13; } /* whoa! */
+
+    slen = len - 1;
+    s++;
+    while (slen--) {
+       if ((*s & 0xc0) != 0x80)
+           return 0;
+       s++;
+    }
+    return len;
+}
+
 UV
 Perl_utf8_to_uv(pTHX_ U8* s, I32* retlen)
 {
@@ -189,6 +222,73 @@ Perl_utf8_hop(pTHX_ U8 *s, I32 off)
     return s;
 }
 
+/*
+=for apidoc Am|U8 *|utf8_to_bytes|U8 *s|STRLEN len
+
+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.
+
+=cut
+*/
+
+U8 *
+Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN len)
+{
+    dTHR;
+    U8 *send;
+    U8 *d;
+    U8 *save;
+
+    send = s + len;
+    d = save = s;
+    while (s < send) {
+        if (*s < 0x80)
+            *d++ = *s++;
+        else {
+            I32 ulen;
+            UV uv = utf8_to_uv(s, &ulen);
+            s += ulen;
+            *d++ = (U8)uv;
+        }
+    }
+    *d = '\0';
+    return save;
+}
+
+/*
+=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.
+
+=cut
+*/
+
+U8*
+Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN len)
+{
+    dTHR;
+    U8 *send;
+    U8 *d;
+    U8 *dst;
+    send = s + len;
+
+    Newz(801, d, len * 2 + 1, U8);
+    dst = d;
+
+    while (s < send) {
+        if (*s < 0x80)
+            *d++ = *s++;
+        else {
+            UV uv = *s++;
+            *d++ = (( uv >>  6)         | 0xc0);
+            *d++ = (( uv        & 0x3f) | 0x80);
+        }
+    }
+    *d = '\0';
+    return dst;
+}
+
 /* XXX NOTHING CALLS THE FOLLOWING TWO ROUTINES YET!!! */
 /*
  * Convert native or reversed UTF-16 to UTF-8.
@@ -500,6 +600,8 @@ Perl_to_uni_lower_lc(pTHX_ U32 c)
 bool
 Perl_is_utf8_alnum(pTHX_ U8 *p)
 {
+    if (!is_utf8_char(p))
+       return FALSE;
     if (!PL_utf8_alnum)
        PL_utf8_alnum = swash_init("utf8", "IsAlnum", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_alnum, p);
@@ -515,6 +617,8 @@ Perl_is_utf8_alnum(pTHX_ U8 *p)
 bool
 Perl_is_utf8_alnumc(pTHX_ U8 *p)
 {
+    if (!is_utf8_char(p))
+       return FALSE;
     if (!PL_utf8_alnum)
        PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_alnum, p);
@@ -536,6 +640,8 @@ Perl_is_utf8_idfirst(pTHX_ U8 *p)
 bool
 Perl_is_utf8_alpha(pTHX_ U8 *p)
 {
+    if (!is_utf8_char(p))
+       return FALSE;
     if (!PL_utf8_alpha)
        PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_alpha, p);
@@ -544,6 +650,8 @@ Perl_is_utf8_alpha(pTHX_ U8 *p)
 bool
 Perl_is_utf8_ascii(pTHX_ U8 *p)
 {
+    if (!is_utf8_char(p))
+       return FALSE;
     if (!PL_utf8_ascii)
        PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_ascii, p);
@@ -552,6 +660,8 @@ Perl_is_utf8_ascii(pTHX_ U8 *p)
 bool
 Perl_is_utf8_space(pTHX_ U8 *p)
 {
+    if (!is_utf8_char(p))
+       return FALSE;
     if (!PL_utf8_space)
        PL_utf8_space = swash_init("utf8", "IsSpace", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_space, p);
@@ -560,6 +670,8 @@ Perl_is_utf8_space(pTHX_ U8 *p)
 bool
 Perl_is_utf8_digit(pTHX_ U8 *p)
 {
+    if (!is_utf8_char(p))
+       return FALSE;
     if (!PL_utf8_digit)
        PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_digit, p);
@@ -568,6 +680,8 @@ Perl_is_utf8_digit(pTHX_ U8 *p)
 bool
 Perl_is_utf8_upper(pTHX_ U8 *p)
 {
+    if (!is_utf8_char(p))
+       return FALSE;
     if (!PL_utf8_upper)
        PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_upper, p);
@@ -576,6 +690,8 @@ Perl_is_utf8_upper(pTHX_ U8 *p)
 bool
 Perl_is_utf8_lower(pTHX_ U8 *p)
 {
+    if (!is_utf8_char(p))
+       return FALSE;
     if (!PL_utf8_lower)
        PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_lower, p);
@@ -584,6 +700,8 @@ Perl_is_utf8_lower(pTHX_ U8 *p)
 bool
 Perl_is_utf8_cntrl(pTHX_ U8 *p)
 {
+    if (!is_utf8_char(p))
+       return FALSE;
     if (!PL_utf8_cntrl)
        PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_cntrl, p);
@@ -592,6 +710,8 @@ Perl_is_utf8_cntrl(pTHX_ U8 *p)
 bool
 Perl_is_utf8_graph(pTHX_ U8 *p)
 {
+    if (!is_utf8_char(p))
+       return FALSE;
     if (!PL_utf8_graph)
        PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_graph, p);
@@ -600,6 +720,8 @@ Perl_is_utf8_graph(pTHX_ U8 *p)
 bool
 Perl_is_utf8_print(pTHX_ U8 *p)
 {
+    if (!is_utf8_char(p))
+       return FALSE;
     if (!PL_utf8_print)
        PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_print, p);
@@ -608,6 +730,8 @@ Perl_is_utf8_print(pTHX_ U8 *p)
 bool
 Perl_is_utf8_punct(pTHX_ U8 *p)
 {
+    if (!is_utf8_char(p))
+       return FALSE;
     if (!PL_utf8_punct)
        PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_punct, p);
@@ -616,6 +740,8 @@ Perl_is_utf8_punct(pTHX_ U8 *p)
 bool
 Perl_is_utf8_xdigit(pTHX_ U8 *p)
 {
+    if (!is_utf8_char(p))
+       return FALSE;
     if (!PL_utf8_xdigit)
        PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_xdigit, p);
@@ -624,6 +750,8 @@ Perl_is_utf8_xdigit(pTHX_ U8 *p)
 bool
 Perl_is_utf8_mark(pTHX_ U8 *p)
 {
+    if (!is_utf8_char(p))
+       return FALSE;
     if (!PL_utf8_mark)
        PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0);
     return swash_fetch(PL_utf8_mark, p);
@@ -730,7 +858,7 @@ Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr)
 
     if (hv == PL_last_swash_hv &&
        klen == PL_last_swash_klen &&
-       (!klen || memEQ(ptr,PL_last_swash_key,klen)) )
+       (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) )
     {
        tmps = PL_last_swash_tmps;
        slen = PL_last_swash_slen;