Pg: Add "$hex" pseudo-type

FU.xs

@@ -236,6 +236,7 @@ void q(fupg_conn *c, SV *sv, ...)
 void _set_type(fupg_conn *c, SV *name, SV *sendsv, SV *recvsv)
   CODE:
     fupg_set_type(c, name, sendsv, recvsv);
+    XSRETURN(1);
 
 
 MODULE = FU   PACKAGE = FU::Pg::txn

FU/Pg.pm

@@ -619,7 +619,21 @@ supported. C<undef> always converts to SQL C<NULL>.
 =item bytea
 
 The C<bytea> type represents arbitrary binary data and this module will pass
-that along as raw binary strings.
+that along as raw binary strings. If you prefer to work with hex strings
+instead, use:
+
+  $conn->set_type(bytea => '$hex');
+
+The I<bytea> and the I<$hex> (pseudo-)types can be applied to any other type to
+convert between the PostgreSQL binary wire format and Perl strings. For
+example, if you prefer to receive integers as big-endian hex strings, you can
+do that:
+
+  $conn->set_type(int4 => recv => '$hex');
+
+Or to treat UUIDs as 16-byte strings:
+
+  $conn->set_type(uuid => 'bytea');
 
 =item timestamp / timestamptz
 
@@ -724,8 +738,6 @@ C<set_type()> to configure appropriate conversions for these types.
 
 =back
 
-I<TODO:> Some handy special types for overriding common conversions.
-
 I<TODO:> Methods to convert between the various formats.
 
 I<TODO:> Methods to query type info.

c/common.c

@@ -178,6 +178,11 @@ static double fu_timediff(const struct timespec *a, const struct timespec *b) {
 }
 
 
+static int fu_hexdig(char x) {
+    return x >= '0' && x <= '9' ? x-'0' : x >= 'A' && x <= 'F' ? x-'A'+10 : x >= 'a' && x <= 'f' ? x-'a'+10 : 0x10000;
+}
+
+
 
 /* -1 if arg is not a bool, 0 on false, 1 on true */
 static int fu_2bool(pTHX_ SV *val) {

c/jsonparse.c

@@ -30,26 +30,22 @@ static inline int fujson_parse_string_escape(pTHX_ fujson_parse_ctx *ctx, fustr
         case 'r': *(r->cur++) = 0x0d; break;
         case 'u':
             /* (awful code adapted from ncdu) */
-#define INV (1<<16)
+#define h4(b) (fu_hexdig((b)[0])<<12) + (fu_hexdig((b)[1])<<8) + (fu_hexdig((b)[2])<<4) + fu_hexdig((b)[3])
-#define hn(n) (n >= '0' && n <= '9' ? n-'0' : n >= 'A' && n <= 'F' ? n-'A'+10 : n >= 'a' && n <= 'f' ? n-'a'+10 : INV)
-#define h4(b) (hn((b)[0])<<12) + (hn((b)[1])<<8) + (hn((b)[2])<<4) + hn((b)[3])
             if (ctx->end - ctx->buf < 4) return 1;
             n = h4(ctx->buf);
-            if (n >= INV || (n & 0xfc00) == 0xdc00) return 1;
+            if (n >= 0x10000 || (n & 0xfc00) == 0xdc00) return 1;
             ctx->buf += 4;
             if ((n & 0xfc00) == 0xd800) { /* high surrogate */
                 if (ctx->end - ctx->buf < 6) return 1;
                 if (ctx->buf[0] != '\\' || ctx->buf[1] != 'u') return 1;
                 s = h4(ctx->buf+2);
-                if (s >= INV || (s & 0xfc00) != 0xdc00) return 1;
+                if (s >= 0x10000 || (s & 0xfc00) != 0xdc00) return 1;
                 n = 0x10000 + (((n & 0x03ff) << 10) | (s & 0x03ff));
                 ctx->buf += 6;
             }
             r->cur = (char *)uvchr_to_utf8((U8 *)r->cur, n);
             if (n >= 0x80) r->setutf8 = 1;
             break;
-#undef INV
-#undef hn
 #undef h4
         default:
             return 1;

c/pgtypes.c

@@ -156,6 +156,33 @@ SENDFN(bytea) {
     fustr_write(out, buf, len);
 }
 
+RECVFN(hex) {
+    SV *r = newSV(len ? len * 2 : 1);
+    SvPOK_only(r);
+    char *out = SvPVX(r);
+    const unsigned char *in = (const unsigned char *)buf;
+    int i;
+    for (i=0; i<len; i++) {
+        *out++ = PL_hexdigit[(in[i] >> 4) & 0x0f];
+        *out++ = PL_hexdigit[in[i] & 0x0f];
+    }
+    SvCUR_set(r, len * 2);
+    return r;
+}
+
+SENDFN(hex) {
+    STRLEN len;
+    const char *in = SvPV(val, len);
+    const char *end = in + len;
+    if (len % 2) SERR("Invalid hex string");
+    while (in < end) {
+        int v = (fu_hexdig(*in)<<4) + fu_hexdig(in[1]);
+        if (v > 0xff) SERR("Invalid hex string");
+        fustr_write_ch(out, v);
+        in += 2;
+    }
+}
+
 RECVFN(char) {
     RLEN(1);
     return newSVpvn(buf, len);
@@ -515,7 +542,7 @@ RECVFN(uuid) {
     RLEN(16);
     char tmp[64];
     char *out = tmp;
-    unsigned char *in = (unsigned char *)buf;
+    const unsigned char *in = (const unsigned char *)buf;
     int i;
     for (i=0; i<16; i++) {
         if (i == 4 || i == 6 || i == 8 || i == 10) *out++ = '-';
@@ -534,9 +561,8 @@ SENDFN(uuid) {
     for (; *in; in++) {
         if (*in == '}') break;
         if (dig == 0x10 && *in == '-') continue;
-        unsigned char x = *in;
+        int x = fu_hexdig(*in);
-        x = x >= '0' && x <= '9' ? x-'0' : x >= 'A' && x <= 'F' ? x-'A'+10 : x >= 'a' && x <= 'f' ? x-'a'+10 : 0x10;
+        if (x > 0x10) SERR("invalid UUID");
-        if (x == 0x10) SERR("invalid UUID");
         if (bytes >= 16) SERR("invalid UUID");
         if (dig == 0x10) dig = x;
         else {
@@ -794,7 +820,8 @@ static const fupg_type fupg_builtin[] = {
 
 /* List of special types for use with set_type() */
 #define SPECIALS\
-    T("$date_str", date_str)
+    T("$date_str",   date_str)\
+    T("$hex",        hex     )
 
 static const fupg_type fupg_specials[] = {
 #define T(name, fun) { 0, 0, {name"\0"}, fupg_send_##fun, fupg_recv_##fun },

t/pgtypes.t

@@ -67,6 +67,13 @@ v bytea => 'hello', undef, '\x68656c6c6f';
 v bytea => "\xaf\x90", undef, '\xaf90';
 f bytea => "\x{1234}";
 
+$conn->set_type(bytea => '$hex');
+v bytea => '', undef, '\x';
+v bytea => '68656c6c6f', undef, '\x68656c6c6f';
+v bytea => "af90", undef, '\xaf90';
+f bytea => 'a';
+f bytea => 'a f';
+
 v '"char"' => $_ for (1, '1', 'a', 'A', '-');
 v '"char"' => "\x84", undef, '\204';
 f '"char"' => $_ for ('', 'ab', "\x{1234}");