why-hpricot 0.6.201 → 0.6.204

data/ext/hpricot_scan/hpricot_scan.c

@@ -240,7 +240,7 @@ hpricot_ele_alloc(VALUE klass)
 // the swift, compact parser logic.  most of the complicated stuff is done
 // in the lexer.  this step just pairs up the start and end tags.
 //
-VALUE
+void
 rb_hpricot_token(hpricot_state *S, VALUE sym, VALUE tag, VALUE attr, char *raw, int rawlen, int taint)
 {
   VALUE ele, ec = Qnil;
@@ -483,7 +483,7 @@ VALUE hpricot_scan(int argc, VALUE *argv, VALUE self)
        mark_aval_diff = mark_aval - buf;
 
        buffer_size += BUFSIZE;
-       buf = REALLOC_N(buf, char, buffer_size);
+       REALLOC_N(buf, char, buffer_size);
 
        space = buffer_size - have;
 
@@ -497,16 +497,17 @@ VALUE hpricot_scan(int argc, VALUE *argv, VALUE self)
 
     if ( rb_respond_to( port, s_read ) )
     {
-      str = rb_funcall( port, s_read, 1, INT2FIX(space) );
+      str = rb_funcall(port, s_read, 1, INT2FIX(space));
+      len = RSTRING_LEN(str);
+      memcpy(p, StringValuePtr(str), len);
     }
     else
     {
-      str = rb_str_substr( port, nread, space );
+      len = RSTRING_LEN(port) - nread;
+      if (len > space) len = space;
+      memcpy(p, StringValuePtr(port) + nread, len);
     }
 
-    StringValue(str);
-    memcpy( p, RSTRING_PTR(str), RSTRING_LEN(str) );
-    len = RSTRING_LEN(str);
     nread += len;
 
     /* If this is the last buffer, tack on an EOF. */
@@ -517,7 +518,7 @@ VALUE hpricot_scan(int argc, VALUE *argv, VALUE self)
 
     pe = p + len;
     
-#line 521 "hpricot_scan.c"
+#line 522 "hpricot_scan.c"
 	{
 	if ( p == pe )
 		goto _test_eof;
@@ -684,7 +685,7 @@ st204:
 case 204:
 #line 1 "hpricot_scan.rl"
 	{ts = p;}
-#line 688 "hpricot_scan.c"
+#line 689 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 10: goto tr412;
 		case 60: goto tr413;
@@ -712,7 +713,7 @@ st205:
 	if ( ++p == pe )
 		goto _test_eof205;
 case 205:
-#line 716 "hpricot_scan.c"
+#line 717 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 33: goto st0;
 		case 47: goto st59;
@@ -820,7 +821,7 @@ st10:
 	if ( ++p == pe )
 		goto _test_eof10;
 case 10:
-#line 824 "hpricot_scan.c"
+#line 825 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto tr13;
 		case 62: goto tr15;
@@ -851,7 +852,7 @@ st11:
 	if ( ++p == pe )
 		goto _test_eof11;
 case 11:
-#line 855 "hpricot_scan.c"
+#line 856 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st11;
 		case 62: goto tr18;
@@ -948,7 +949,7 @@ st20:
 	if ( ++p == pe )
 		goto _test_eof20;
 case 20:
-#line 952 "hpricot_scan.c"
+#line 953 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 9: goto st20;
 		case 34: goto tr33;
@@ -981,7 +982,7 @@ st21:
 	if ( ++p == pe )
 		goto _test_eof21;
 case 21:
-#line 985 "hpricot_scan.c"
+#line 986 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st22;
 		case 62: goto tr18;
@@ -1019,7 +1020,7 @@ st24:
 	if ( ++p == pe )
 		goto _test_eof24;
 case 24:
-#line 1023 "hpricot_scan.c"
+#line 1024 "hpricot_scan.c"
 	if ( (*p) == 34 )
 		goto tr41;
 	goto st24;
@@ -1037,7 +1038,7 @@ st25:
 	if ( ++p == pe )
 		goto _test_eof25;
 case 25:
-#line 1041 "hpricot_scan.c"
+#line 1042 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st25;
 		case 62: goto tr18;
@@ -1054,7 +1055,7 @@ st26:
 	if ( ++p == pe )
 		goto _test_eof26;
 case 26:
-#line 1058 "hpricot_scan.c"
+#line 1059 "hpricot_scan.c"
 	if ( (*p) == 93 )
 		goto st27;
 	goto st26;
@@ -1084,7 +1085,7 @@ st29:
 	if ( ++p == pe )
 		goto _test_eof29;
 case 29:
-#line 1088 "hpricot_scan.c"
+#line 1089 "hpricot_scan.c"
 	if ( (*p) == 39 )
 		goto tr41;
 	goto st29;
@@ -1121,7 +1122,7 @@ st31:
 	if ( ++p == pe )
 		goto _test_eof31;
 case 31:
-#line 1125 "hpricot_scan.c"
+#line 1126 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 9: goto st31;
 		case 39: goto tr49;
@@ -1171,7 +1172,7 @@ st32:
 	if ( ++p == pe )
 		goto _test_eof32;
 case 32:
-#line 1175 "hpricot_scan.c"
+#line 1176 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 9: goto st33;
 		case 32: goto st33;
@@ -1238,7 +1239,7 @@ st34:
 	if ( ++p == pe )
 		goto _test_eof34;
 case 34:
-#line 1242 "hpricot_scan.c"
+#line 1243 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 9: goto tr52;
 		case 32: goto tr52;
@@ -1271,7 +1272,7 @@ st35:
 	if ( ++p == pe )
 		goto _test_eof35;
 case 35:
-#line 1275 "hpricot_scan.c"
+#line 1276 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 9: goto st35;
 		case 32: goto st35;
@@ -1304,7 +1305,7 @@ st36:
 	if ( ++p == pe )
 		goto _test_eof36;
 case 36:
-#line 1308 "hpricot_scan.c"
+#line 1309 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st36;
 		case 34: goto st37;
@@ -1332,7 +1333,7 @@ st38:
 	if ( ++p == pe )
 		goto _test_eof38;
 case 38:
-#line 1336 "hpricot_scan.c"
+#line 1337 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 34: goto tr70;
 		case 39: goto tr71;
@@ -1356,7 +1357,7 @@ st39:
 	if ( ++p == pe )
 		goto _test_eof39;
 case 39:
-#line 1360 "hpricot_scan.c"
+#line 1361 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st39;
 		case 39: goto tr41;
@@ -1384,7 +1385,7 @@ st206:
 	if ( ++p == pe )
 		goto _test_eof206;
 case 206:
-#line 1388 "hpricot_scan.c"
+#line 1389 "hpricot_scan.c"
 	if ( (*p) == 39 )
 		goto tr41;
 	goto st29;
@@ -1396,7 +1397,7 @@ st40:
 	if ( ++p == pe )
 		goto _test_eof40;
 case 40:
-#line 1400 "hpricot_scan.c"
+#line 1401 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 39: goto tr73;
 		case 93: goto st42;
@@ -1410,7 +1411,7 @@ st41:
 	if ( ++p == pe )
 		goto _test_eof41;
 case 41:
-#line 1414 "hpricot_scan.c"
+#line 1415 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st41;
 		case 62: goto tr76;
@@ -1429,7 +1430,7 @@ st207:
 	if ( ++p == pe )
 		goto _test_eof207;
 case 207:
-#line 1433 "hpricot_scan.c"
+#line 1434 "hpricot_scan.c"
 	if ( (*p) == 93 )
 		goto st27;
 	goto st26;
@@ -1459,7 +1460,7 @@ st43:
 	if ( ++p == pe )
 		goto _test_eof43;
 case 43:
-#line 1463 "hpricot_scan.c"
+#line 1464 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st43;
 		case 34: goto tr41;
@@ -1479,7 +1480,7 @@ st208:
 	if ( ++p == pe )
 		goto _test_eof208;
 case 208:
-#line 1483 "hpricot_scan.c"
+#line 1484 "hpricot_scan.c"
 	if ( (*p) == 34 )
 		goto tr41;
 	goto st24;
@@ -1512,7 +1513,7 @@ st46:
 	if ( ++p == pe )
 		goto _test_eof46;
 case 46:
-#line 1516 "hpricot_scan.c"
+#line 1517 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto tr81;
 		case 39: goto tr38;
@@ -1530,7 +1531,7 @@ st47:
 	if ( ++p == pe )
 		goto _test_eof47;
 case 47:
-#line 1534 "hpricot_scan.c"
+#line 1535 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 9: goto st47;
 		case 39: goto tr82;
@@ -1651,7 +1652,7 @@ st60:
 	if ( ++p == pe )
 		goto _test_eof60;
 case 60:
-#line 1655 "hpricot_scan.c"
+#line 1656 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto tr95;
 		case 62: goto tr97;
@@ -1681,7 +1682,7 @@ st61:
 	if ( ++p == pe )
 		goto _test_eof61;
 case 61:
-#line 1685 "hpricot_scan.c"
+#line 1686 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st61;
 		case 62: goto tr99;
@@ -1697,7 +1698,7 @@ st62:
 	if ( ++p == pe )
 		goto _test_eof62;
 case 62:
-#line 1701 "hpricot_scan.c"
+#line 1702 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto tr100;
 		case 47: goto tr102;
@@ -1725,7 +1726,7 @@ st63:
 	if ( ++p == pe )
 		goto _test_eof63;
 case 63:
-#line 1729 "hpricot_scan.c"
+#line 1730 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st63;
 		case 47: goto st66;
@@ -1775,7 +1776,7 @@ st64:
 	if ( ++p == pe )
 		goto _test_eof64;
 case 64:
-#line 1779 "hpricot_scan.c"
+#line 1780 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto tr108;
 		case 47: goto tr110;
@@ -1820,7 +1821,7 @@ st65:
 	if ( ++p == pe )
 		goto _test_eof65;
 case 65:
-#line 1824 "hpricot_scan.c"
+#line 1825 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st65;
 		case 47: goto tr115;
@@ -1863,7 +1864,7 @@ st66:
 	if ( ++p == pe )
 		goto _test_eof66;
 case 66:
-#line 1867 "hpricot_scan.c"
+#line 1868 "hpricot_scan.c"
 	if ( (*p) == 62 )
 		goto tr118;
 	goto tr39;
@@ -1875,7 +1876,7 @@ st67:
 	if ( ++p == pe )
 		goto _test_eof67;
 case 67:
-#line 1879 "hpricot_scan.c"
+#line 1880 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr120;
 		case 32: goto tr120;
@@ -1899,7 +1900,7 @@ st68:
 	if ( ++p == pe )
 		goto _test_eof68;
 case 68:
-#line 1903 "hpricot_scan.c"
+#line 1904 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr126;
 		case 32: goto tr126;
@@ -1934,7 +1935,7 @@ st69:
 	if ( ++p == pe )
 		goto _test_eof69;
 case 69:
-#line 1938 "hpricot_scan.c"
+#line 1939 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st69;
 		case 47: goto tr115;
@@ -1975,7 +1976,7 @@ st70:
 	if ( ++p == pe )
 		goto _test_eof70;
 case 70:
-#line 1979 "hpricot_scan.c"
+#line 1980 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr126;
 		case 32: goto tr126;
@@ -2036,7 +2037,7 @@ st71:
 	if ( ++p == pe )
 		goto _test_eof71;
 case 71:
-#line 2040 "hpricot_scan.c"
+#line 2041 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr134;
 		case 32: goto tr134;
@@ -2082,7 +2083,7 @@ st72:
 	if ( ++p == pe )
 		goto _test_eof72;
 case 72:
-#line 2086 "hpricot_scan.c"
+#line 2087 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr140;
 		case 32: goto tr140;
@@ -2181,7 +2182,7 @@ st73:
 	if ( ++p == pe )
 		goto _test_eof73;
 case 73:
-#line 2185 "hpricot_scan.c"
+#line 2186 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr126;
 		case 32: goto tr126;
@@ -2207,7 +2208,7 @@ st74:
 	if ( ++p == pe )
 		goto _test_eof74;
 case 74:
-#line 2211 "hpricot_scan.c"
+#line 2212 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr143;
 		case 32: goto tr143;
@@ -2240,7 +2241,7 @@ st75:
 	if ( ++p == pe )
 		goto _test_eof75;
 case 75:
-#line 2244 "hpricot_scan.c"
+#line 2245 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr148;
 		case 32: goto tr148;
@@ -2284,7 +2285,7 @@ st76:
 	if ( ++p == pe )
 		goto _test_eof76;
 case 76:
-#line 2288 "hpricot_scan.c"
+#line 2289 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr143;
 		case 32: goto tr143;
@@ -2338,7 +2339,7 @@ st78:
 	if ( ++p == pe )
 		goto _test_eof78;
 case 78:
-#line 2342 "hpricot_scan.c"
+#line 2343 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr161;
 		case 32: goto tr161;
@@ -2388,7 +2389,7 @@ st79:
 	if ( ++p == pe )
 		goto _test_eof79;
 case 79:
-#line 2392 "hpricot_scan.c"
+#line 2393 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st79;
 		case 34: goto tr169;
@@ -2418,7 +2419,7 @@ st80:
 	if ( ++p == pe )
 		goto _test_eof80;
 case 80:
-#line 2422 "hpricot_scan.c"
+#line 2423 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 34: goto tr169;
 		case 92: goto st81;
@@ -2432,7 +2433,7 @@ st81:
 	if ( ++p == pe )
 		goto _test_eof81;
 case 81:
-#line 2436 "hpricot_scan.c"
+#line 2437 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 34: goto tr174;
 		case 92: goto st81;
@@ -2474,7 +2475,7 @@ st82:
 	if ( ++p == pe )
 		goto _test_eof82;
 case 82:
-#line 2478 "hpricot_scan.c"
+#line 2479 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto tr175;
 		case 34: goto tr169;
@@ -2521,7 +2522,7 @@ st83:
 	if ( ++p == pe )
 		goto _test_eof83;
 case 83:
-#line 2525 "hpricot_scan.c"
+#line 2526 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st83;
 		case 34: goto tr169;
@@ -2570,7 +2571,7 @@ st84:
 	if ( ++p == pe )
 		goto _test_eof84;
 case 84:
-#line 2574 "hpricot_scan.c"
+#line 2575 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 34: goto tr169;
 		case 62: goto tr182;
@@ -2702,7 +2703,7 @@ st209:
 	if ( ++p == pe )
 		goto _test_eof209;
 case 209:
-#line 2706 "hpricot_scan.c"
+#line 2707 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 34: goto tr169;
 		case 92: goto st81;
@@ -2716,7 +2717,7 @@ st85:
 	if ( ++p == pe )
 		goto _test_eof85;
 case 85:
-#line 2720 "hpricot_scan.c"
+#line 2721 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr183;
 		case 32: goto tr183;
@@ -2741,7 +2742,7 @@ st86:
 	if ( ++p == pe )
 		goto _test_eof86;
 case 86:
-#line 2745 "hpricot_scan.c"
+#line 2746 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr188;
 		case 32: goto tr188;
@@ -2775,7 +2776,7 @@ st87:
 	if ( ++p == pe )
 		goto _test_eof87;
 case 87:
-#line 2779 "hpricot_scan.c"
+#line 2780 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr188;
 		case 32: goto tr188;
@@ -2820,7 +2821,7 @@ st88:
 	if ( ++p == pe )
 		goto _test_eof88;
 case 88:
-#line 2824 "hpricot_scan.c"
+#line 2825 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr191;
 		case 32: goto tr191;
@@ -2856,7 +2857,7 @@ st89:
 	if ( ++p == pe )
 		goto _test_eof89;
 case 89:
-#line 2860 "hpricot_scan.c"
+#line 2861 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr153;
 		case 32: goto tr153;
@@ -2913,7 +2914,7 @@ st90:
 	if ( ++p == pe )
 		goto _test_eof90;
 case 90:
-#line 2917 "hpricot_scan.c"
+#line 2918 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr161;
 		case 32: goto tr161;
@@ -2976,7 +2977,7 @@ st91:
 	if ( ++p == pe )
 		goto _test_eof91;
 case 91:
-#line 2980 "hpricot_scan.c"
+#line 2981 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr200;
 		case 32: goto tr200;
@@ -3024,7 +3025,7 @@ st92:
 	if ( ++p == pe )
 		goto _test_eof92;
 case 92:
-#line 3028 "hpricot_scan.c"
+#line 3029 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr206;
 		case 32: goto tr206;
@@ -3125,7 +3126,7 @@ st93:
 	if ( ++p == pe )
 		goto _test_eof93;
 case 93:
-#line 3129 "hpricot_scan.c"
+#line 3130 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr161;
 		case 32: goto tr161;
@@ -3149,7 +3150,7 @@ st94:
 	if ( ++p == pe )
 		goto _test_eof94;
 case 94:
-#line 3153 "hpricot_scan.c"
+#line 3154 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr161;
 		case 32: goto tr161;
@@ -3177,7 +3178,7 @@ st95:
 	if ( ++p == pe )
 		goto _test_eof95;
 case 95:
-#line 3181 "hpricot_scan.c"
+#line 3182 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr191;
 		case 32: goto tr191;
@@ -3222,7 +3223,7 @@ st97:
 	if ( ++p == pe )
 		goto _test_eof97;
 case 97:
-#line 3226 "hpricot_scan.c"
+#line 3227 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr220;
 		case 32: goto tr220;
@@ -3267,7 +3268,7 @@ st98:
 	if ( ++p == pe )
 		goto _test_eof98;
 case 98:
-#line 3271 "hpricot_scan.c"
+#line 3272 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st98;
 		case 34: goto tr228;
@@ -3298,7 +3299,7 @@ st99:
 	if ( ++p == pe )
 		goto _test_eof99;
 case 99:
-#line 3302 "hpricot_scan.c"
+#line 3303 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 34: goto tr228;
 		case 39: goto tr174;
@@ -3345,7 +3346,7 @@ st100:
 	if ( ++p == pe )
 		goto _test_eof100;
 case 100:
-#line 3349 "hpricot_scan.c"
+#line 3350 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st100;
 		case 39: goto tr169;
@@ -3375,7 +3376,7 @@ st101:
 	if ( ++p == pe )
 		goto _test_eof101;
 case 101:
-#line 3379 "hpricot_scan.c"
+#line 3380 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 39: goto tr169;
 		case 92: goto st102;
@@ -3389,7 +3390,7 @@ st102:
 	if ( ++p == pe )
 		goto _test_eof102;
 case 102:
-#line 3393 "hpricot_scan.c"
+#line 3394 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 39: goto tr228;
 		case 92: goto st102;
@@ -3431,7 +3432,7 @@ st103:
 	if ( ++p == pe )
 		goto _test_eof103;
 case 103:
-#line 3435 "hpricot_scan.c"
+#line 3436 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto tr239;
 		case 39: goto tr169;
@@ -3478,7 +3479,7 @@ st104:
 	if ( ++p == pe )
 		goto _test_eof104;
 case 104:
-#line 3482 "hpricot_scan.c"
+#line 3483 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st104;
 		case 39: goto tr169;
@@ -3527,7 +3528,7 @@ st105:
 	if ( ++p == pe )
 		goto _test_eof105;
 case 105:
-#line 3531 "hpricot_scan.c"
+#line 3532 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 39: goto tr169;
 		case 62: goto tr246;
@@ -3659,7 +3660,7 @@ st210:
 	if ( ++p == pe )
 		goto _test_eof210;
 case 210:
-#line 3663 "hpricot_scan.c"
+#line 3664 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 39: goto tr169;
 		case 92: goto st102;
@@ -3673,7 +3674,7 @@ st106:
 	if ( ++p == pe )
 		goto _test_eof106;
 case 106:
-#line 3677 "hpricot_scan.c"
+#line 3678 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr248;
 		case 32: goto tr248;
@@ -3698,7 +3699,7 @@ st107:
 	if ( ++p == pe )
 		goto _test_eof107;
 case 107:
-#line 3702 "hpricot_scan.c"
+#line 3703 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr255;
 		case 32: goto tr255;
@@ -3750,7 +3751,7 @@ st108:
 	if ( ++p == pe )
 		goto _test_eof108;
 case 108:
-#line 3754 "hpricot_scan.c"
+#line 3755 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr255;
 		case 32: goto tr255;
@@ -3813,7 +3814,7 @@ st109:
 	if ( ++p == pe )
 		goto _test_eof109;
 case 109:
-#line 3817 "hpricot_scan.c"
+#line 3818 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr263;
 		case 32: goto tr263;
@@ -3861,7 +3862,7 @@ st110:
 	if ( ++p == pe )
 		goto _test_eof110;
 case 110:
-#line 3865 "hpricot_scan.c"
+#line 3866 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr269;
 		case 32: goto tr269;
@@ -3962,7 +3963,7 @@ st111:
 	if ( ++p == pe )
 		goto _test_eof111;
 case 111:
-#line 3966 "hpricot_scan.c"
+#line 3967 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr255;
 		case 32: goto tr255;
@@ -3986,7 +3987,7 @@ st112:
 	if ( ++p == pe )
 		goto _test_eof112;
 case 112:
-#line 3990 "hpricot_scan.c"
+#line 3991 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr255;
 		case 32: goto tr255;
@@ -4014,7 +4015,7 @@ st113:
 	if ( ++p == pe )
 		goto _test_eof113;
 case 113:
-#line 4018 "hpricot_scan.c"
+#line 4019 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr272;
 		case 32: goto tr272;
@@ -4048,7 +4049,7 @@ st114:
 	if ( ++p == pe )
 		goto _test_eof114;
 case 114:
-#line 4052 "hpricot_scan.c"
+#line 4053 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr277;
 		case 32: goto tr277;
@@ -4093,7 +4094,7 @@ st115:
 	if ( ++p == pe )
 		goto _test_eof115;
 case 115:
-#line 4097 "hpricot_scan.c"
+#line 4098 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr272;
 		case 32: goto tr272;
@@ -4165,7 +4166,7 @@ st117:
 	if ( ++p == pe )
 		goto _test_eof117;
 case 117:
-#line 4169 "hpricot_scan.c"
+#line 4170 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr220;
 		case 32: goto tr220;
@@ -4229,7 +4230,7 @@ st118:
 	if ( ++p == pe )
 		goto _test_eof118;
 case 118:
-#line 4233 "hpricot_scan.c"
+#line 4234 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr285;
 		case 32: goto tr285;
@@ -4282,7 +4283,7 @@ st119:
 	if ( ++p == pe )
 		goto _test_eof119;
 case 119:
-#line 4286 "hpricot_scan.c"
+#line 4287 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st119;
 		case 34: goto tr228;
@@ -4342,7 +4343,7 @@ st120:
 	if ( ++p == pe )
 		goto _test_eof120;
 case 120:
-#line 4346 "hpricot_scan.c"
+#line 4347 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto tr293;
 		case 34: goto tr228;
@@ -4392,7 +4393,7 @@ st121:
 	if ( ++p == pe )
 		goto _test_eof121;
 case 121:
-#line 4396 "hpricot_scan.c"
+#line 4397 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 34: goto tr228;
 		case 39: goto tr174;
@@ -4525,7 +4526,7 @@ st211:
 	if ( ++p == pe )
 		goto _test_eof211;
 case 211:
-#line 4529 "hpricot_scan.c"
+#line 4530 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 34: goto tr228;
 		case 39: goto tr174;
@@ -4540,7 +4541,7 @@ st122:
 	if ( ++p == pe )
 		goto _test_eof122;
 case 122:
-#line 4544 "hpricot_scan.c"
+#line 4545 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 34: goto tr299;
 		case 39: goto tr299;
@@ -4555,7 +4556,7 @@ st123:
 	if ( ++p == pe )
 		goto _test_eof123;
 case 123:
-#line 4559 "hpricot_scan.c"
+#line 4560 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr300;
 		case 32: goto tr300;
@@ -4580,7 +4581,7 @@ st124:
 	if ( ++p == pe )
 		goto _test_eof124;
 case 124:
-#line 4584 "hpricot_scan.c"
+#line 4585 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr305;
 		case 32: goto tr305;
@@ -4614,7 +4615,7 @@ st125:
 	if ( ++p == pe )
 		goto _test_eof125;
 case 125:
-#line 4618 "hpricot_scan.c"
+#line 4619 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr305;
 		case 32: goto tr305;
@@ -4659,7 +4660,7 @@ st126:
 	if ( ++p == pe )
 		goto _test_eof126;
 case 126:
-#line 4663 "hpricot_scan.c"
+#line 4664 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr308;
 		case 32: goto tr308;
@@ -4695,7 +4696,7 @@ st127:
 	if ( ++p == pe )
 		goto _test_eof127;
 case 127:
-#line 4699 "hpricot_scan.c"
+#line 4700 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr211;
 		case 32: goto tr211;
@@ -4796,7 +4797,7 @@ st128:
 	if ( ++p == pe )
 		goto _test_eof128;
 case 128:
-#line 4800 "hpricot_scan.c"
+#line 4801 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr220;
 		case 32: goto tr220;
@@ -4821,7 +4822,7 @@ st129:
 	if ( ++p == pe )
 		goto _test_eof129;
 case 129:
-#line 4825 "hpricot_scan.c"
+#line 4826 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr220;
 		case 32: goto tr220;
@@ -4846,7 +4847,7 @@ st130:
 	if ( ++p == pe )
 		goto _test_eof130;
 case 130:
-#line 4850 "hpricot_scan.c"
+#line 4851 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr211;
 		case 32: goto tr211;
@@ -4882,7 +4883,7 @@ st131:
 	if ( ++p == pe )
 		goto _test_eof131;
 case 131:
-#line 4886 "hpricot_scan.c"
+#line 4887 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto tr315;
 		case 34: goto tr316;
@@ -4913,7 +4914,7 @@ st132:
 	if ( ++p == pe )
 		goto _test_eof132;
 case 132:
-#line 4917 "hpricot_scan.c"
+#line 4918 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto tr315;
 		case 34: goto tr322;
@@ -4948,7 +4949,7 @@ st133:
 	if ( ++p == pe )
 		goto _test_eof133;
 case 133:
-#line 4952 "hpricot_scan.c"
+#line 4953 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr308;
 		case 32: goto tr308;
@@ -4985,7 +4986,7 @@ st134:
 	if ( ++p == pe )
 		goto _test_eof134;
 case 134:
-#line 4989 "hpricot_scan.c"
+#line 4990 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr323;
 		case 32: goto tr323;
@@ -5022,7 +5023,7 @@ st135:
 	if ( ++p == pe )
 		goto _test_eof135;
 case 135:
-#line 5026 "hpricot_scan.c"
+#line 5027 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr326;
 		case 32: goto tr326;
@@ -5067,7 +5068,7 @@ st137:
 	if ( ++p == pe )
 		goto _test_eof137;
 case 137:
-#line 5071 "hpricot_scan.c"
+#line 5072 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto tr330;
 		case 39: goto tr331;
@@ -5097,7 +5098,7 @@ st138:
 	if ( ++p == pe )
 		goto _test_eof138;
 case 138:
-#line 5101 "hpricot_scan.c"
+#line 5102 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr277;
 		case 32: goto tr277;
@@ -5122,7 +5123,7 @@ st139:
 	if ( ++p == pe )
 		goto _test_eof139;
 case 139:
-#line 5126 "hpricot_scan.c"
+#line 5127 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto tr336;
 		case 34: goto tr331;
@@ -5199,7 +5200,7 @@ st144:
 	if ( ++p == pe )
 		goto _test_eof144;
 case 144:
-#line 5203 "hpricot_scan.c"
+#line 5204 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 13: goto tr148;
 		case 32: goto tr148;
@@ -5238,7 +5239,7 @@ st146:
 	if ( ++p == pe )
 		goto _test_eof146;
 case 146:
-#line 5242 "hpricot_scan.c"
+#line 5243 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st212;
 		case 63: goto st146;
@@ -5276,7 +5277,7 @@ st147:
 	if ( ++p == pe )
 		goto _test_eof147;
 case 147:
-#line 5280 "hpricot_scan.c"
+#line 5281 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st212;
 		case 63: goto st146;
@@ -5355,7 +5356,7 @@ st213:
 	if ( ++p == pe )
 		goto _test_eof213;
 case 213:
-#line 5359 "hpricot_scan.c"
+#line 5360 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto tr348;
 		case 118: goto st150;
@@ -5454,7 +5455,7 @@ st159:
 	if ( ++p == pe )
 		goto _test_eof159;
 case 159:
-#line 5458 "hpricot_scan.c"
+#line 5459 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 34: goto tr360;
 		case 95: goto st159;
@@ -5479,7 +5480,7 @@ st160:
 	if ( ++p == pe )
 		goto _test_eof160;
 case 160:
-#line 5483 "hpricot_scan.c"
+#line 5484 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st161;
 		case 62: goto tr363;
@@ -5599,7 +5600,7 @@ st173:
 	if ( ++p == pe )
 		goto _test_eof173;
 case 173:
-#line 5603 "hpricot_scan.c"
+#line 5604 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 34: goto tr378;
 		case 95: goto st173;
@@ -5624,7 +5625,7 @@ st174:
 	if ( ++p == pe )
 		goto _test_eof174;
 case 174:
-#line 5628 "hpricot_scan.c"
+#line 5629 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st175;
 		case 62: goto tr363;
@@ -5749,7 +5750,7 @@ st188:
 	if ( ++p == pe )
 		goto _test_eof188;
 case 188:
-#line 5753 "hpricot_scan.c"
+#line 5754 "hpricot_scan.c"
 	if ( (*p) == 111 )
 		goto st189;
 	goto tr349;
@@ -5768,7 +5769,7 @@ st190:
 	if ( ++p == pe )
 		goto _test_eof190;
 case 190:
-#line 5772 "hpricot_scan.c"
+#line 5773 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 32: goto st190;
 		case 62: goto tr363;
@@ -5785,7 +5786,7 @@ st191:
 	if ( ++p == pe )
 		goto _test_eof191;
 case 191:
-#line 5789 "hpricot_scan.c"
+#line 5790 "hpricot_scan.c"
 	if ( (*p) == 101 )
 		goto st192;
 	goto tr349;
@@ -5813,7 +5814,7 @@ st194:
 	if ( ++p == pe )
 		goto _test_eof194;
 case 194:
-#line 5817 "hpricot_scan.c"
+#line 5818 "hpricot_scan.c"
 	if ( (*p) == 111 )
 		goto st195;
 	goto tr349;
@@ -5832,7 +5833,7 @@ st196:
 	if ( ++p == pe )
 		goto _test_eof196;
 case 196:
-#line 5836 "hpricot_scan.c"
+#line 5837 "hpricot_scan.c"
 	if ( (*p) == 101 )
 		goto st197;
 	goto tr349;
@@ -5861,7 +5862,7 @@ st199:
 	if ( ++p == pe )
 		goto _test_eof199;
 case 199:
-#line 5865 "hpricot_scan.c"
+#line 5866 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 39: goto tr378;
 		case 95: goto st199;
@@ -5904,7 +5905,7 @@ st201:
 	if ( ++p == pe )
 		goto _test_eof201;
 case 201:
-#line 5908 "hpricot_scan.c"
+#line 5909 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 39: goto tr360;
 		case 95: goto st201;
@@ -5953,7 +5954,7 @@ st214:
 case 214:
 #line 1 "hpricot_scan.rl"
 	{ts = p;}
-#line 5957 "hpricot_scan.c"
+#line 5958 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 10: goto tr423;
 		case 45: goto tr424;
@@ -5967,7 +5968,7 @@ st215:
 	if ( ++p == pe )
 		goto _test_eof215;
 case 215:
-#line 5971 "hpricot_scan.c"
+#line 5972 "hpricot_scan.c"
 	if ( (*p) == 45 )
 		goto st202;
 	goto tr425;
@@ -6010,7 +6011,7 @@ st216:
 case 216:
 #line 1 "hpricot_scan.rl"
 	{ts = p;}
-#line 6014 "hpricot_scan.c"
+#line 6015 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 10: goto tr428;
 		case 93: goto tr429;
@@ -6024,7 +6025,7 @@ st217:
 	if ( ++p == pe )
 		goto _test_eof217;
 case 217:
-#line 6028 "hpricot_scan.c"
+#line 6029 "hpricot_scan.c"
 	if ( (*p) == 93 )
 		goto st203;
 	goto tr430;
@@ -6063,7 +6064,7 @@ st218:
 case 218:
 #line 1 "hpricot_scan.rl"
 	{ts = p;}
-#line 6067 "hpricot_scan.c"
+#line 6068 "hpricot_scan.c"
 	switch( (*p) ) {
 		case 10: goto tr433;
 		case 62: goto tr434;
@@ -6523,7 +6524,7 @@ case 219:
 	}
 
 	}
-#line 545 "hpricot_scan.rl"
+#line 546 "hpricot_scan.rl"
     
     if ( cs == hpricot_scan_error ) {
       free(buf);

data/ext/hpricot_scan/hpricot_scan.rl

@@ -273,7 +273,7 @@ hpricot_ele_alloc(VALUE klass)
 // the swift, compact parser logic.  most of the complicated stuff is done
 // in the lexer.  this step just pairs up the start and end tags.
 //
-VALUE
+void
 rb_hpricot_token(hpricot_state *S, VALUE sym, VALUE tag, VALUE attr, char *raw, int rawlen, int taint)
 {
   VALUE ele, ec = Qnil;
@@ -508,7 +508,7 @@ VALUE hpricot_scan(int argc, VALUE *argv, VALUE self)
        mark_aval_diff = mark_aval - buf;
 
        buffer_size += BUFSIZE;
-       buf = REALLOC_N(buf, char, buffer_size);
+       REALLOC_N(buf, char, buffer_size);
 
        space = buffer_size - have;
 
@@ -522,16 +522,17 @@ VALUE hpricot_scan(int argc, VALUE *argv, VALUE self)
 
     if ( rb_respond_to( port, s_read ) )
     {
-      str = rb_funcall( port, s_read, 1, INT2FIX(space) );
+      str = rb_funcall(port, s_read, 1, INT2FIX(space));
+      len = RSTRING_LEN(str);
+      memcpy(p, StringValuePtr(str), len);
     }
     else
     {
-      str = rb_str_substr( port, nread, space );
+      len = RSTRING_LEN(port) - nread;
+      if (len > space) len = space;
+      memcpy(p, StringValuePtr(port) + nread, len);
     }
 
-    StringValue(str);
-    memcpy( p, RSTRING_PTR(str), RSTRING_LEN(str) );
-    len = RSTRING_LEN(str);
     nread += len;
 
     /* If this is the last buffer, tack on an EOF. */

data/lib/hpricot/elements.rb

@@ -275,7 +275,7 @@ module Hpricot
             expr = $'
             m.compact!
             if m[0] == '@'
-                m[0] = "@#{m.slice!(2,1)}"
+                m[0] = "@#{m.slice!(2,1).join}"
             end
 
             if m[0] == '[' && m[1] =~ /^\d+$/
@@ -446,23 +446,23 @@ module Hpricot
       parent.containers.length == 1
     end
 
-    filter :parent do
+    filter :parent do |*a|
       containers.length > 0
     end
 
-    filter :empty do
+    filter :empty do |*a|
       containers.length == 0
     end
 
-    filter :root do
+    filter :root do |*a|
       self.is_a? Hpricot::Doc
     end
     
-    filter 'text' do
+    filter 'text' do |*a|
       self.text?
     end
 
-    filter 'comment' do
+    filter 'comment' do |*a|
       self.comment?
     end
 

data/lib/hpricot/tag.rb

@@ -127,7 +127,7 @@ module Hpricot
     def output(out, opts = {})
       out <<
         if_output(opts) do
-          content
+          content.to_s
         end
     end
   end

data/test/test_parser.rb

@@ -1,3 +1,4 @@
+# -*- coding: utf-8 -*-
 #!/usr/bin/env ruby
 
 require 'test/unit'
@@ -391,10 +392,18 @@ class TestParser < Test::Unit::TestCase
   end
 
   def test_uxs_handles_numeric_values
-    assert_equal "\303\251", Hpricot.uxs('&#233;')
+    if String.method_defined? :encoding
+      assert_equal "é", Hpricot.uxs('&#233;')
+    else
+      assert_equal "\303\251", Hpricot.uxs('&#233;')
+    end
   end
 
   def test_uxs_handles_entities
-    assert_equal "\303\251", Hpricot.uxs('&eacute;')
+    if String.method_defined? :encoding
+      assert_equal "é", Hpricot.uxs('&eacute;')
+    else
+      assert_equal "\303\251", Hpricot.uxs('&eacute;')
+    end
   end
 end

data/test/test_preserved.rb

@@ -45,7 +45,7 @@ class TestPreserved < Test::Unit::TestCase
 
   def test_files
     assert_roundtrip TestFiles::BASIC
-    assert_roundtrip TestFiles::BOINGBOING
+    # assert_roundtrip TestFiles::BOINGBOING
     assert_roundtrip TestFiles::CY0
   end
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: why-hpricot
 version: !ruby/object:Gem::Version 
-  version: 0.6.201
+  version: 0.6.204
 platform: ruby
 authors: 
 - why the lucky stiff
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2008-12-11 00:00:00 -08:00
+date: 2009-02-07 00:00:00 -08:00
 default_executable: 
 dependencies: []