oga 0.1.0

data/ext/c/extconf.rb

@@ -0,0 +1,13 @@
+require 'mkmf'
+
+have_header('ruby.h')
+
+$CFLAGS << ' -Wextra -Wall -pedantic'
+
+if ENV['DEBUG']
+  $CFLAGS << ' -O0 -g'
+else
+  $CFLAGS << ' -O3'
+end
+
+create_makefile('liboga')

data/ext/c/lexer.c

@@ -0,0 +1,1518 @@
+
+#line 1 "ext/c/lexer.rl"
+#include "lexer.h"
+
+/*
+The following two macros allow the Ragel grammar to use generic function calls
+without relying on the setup of the C or Java lexer. Using these macros we can
+also pass along `self` to the callback functions without having to hard-code
+this in to the Ragel grammar.
+
+In the C lexer we don't need the `data` variable (since this is pulled in based
+on `ts` and `te`) so the macro ignores this argument.
+*/
+
+#define callback(name, data, encoding, start, stop) \
+    liboga_xml_lexer_callback(self, name, encoding, start, stop);
+
+#define callback_simple(name) \
+    liboga_xml_lexer_callback_simple(self, name);
+
+#define oga_ivar_get(owner, name) \
+    rb_ivar_get(owner, rb_intern(name))
+
+#define oga_ivar_set(owner, name, value) \
+    rb_ivar_set(owner, rb_intern(name), value)
+
+
+#line 26 "ext/c/lexer.rl"
+
+/**
+ * Calls a method defined in the Ruby side of the lexer. The String value is
+ * created based on the values of `ts` and `te` and uses the encoding specified
+ * in `encoding`.
+ *
+ * @example
+ *  rb_encoding *encoding = rb_enc_get(...);
+ *  liboga_xml_lexer_callback(self, "on_string", encoding, ts, te);
+ */
+void liboga_xml_lexer_callback(
+    VALUE self,
+    const char *name,
+    rb_encoding *encoding,
+    const char *ts,
+    const char *te
+)
+{
+    VALUE value  = rb_enc_str_new(ts, te - ts, encoding);
+    VALUE method = rb_intern(name);
+
+    rb_funcall(self, method, 1, value);
+}
+
+/**
+ * Calls a method defined in the Ruby side of the lexer without passing it any
+ * arguments.
+ *
+ * @example
+ *  liboga_xml_lexer_callback_simple(self, "on_cdata_start");
+ */
+void liboga_xml_lexer_callback_simple(VALUE self, const char *name)
+{
+    VALUE method = rb_intern(name);
+
+    rb_funcall(self, method, 0);
+}
+
+
+#line 69 "ext/c/lexer.c"
+static const int c_lexer_start = 35;
+static const int c_lexer_first_final = 35;
+static const int c_lexer_error = 0;
+
+static const int c_lexer_en_proc_ins_body = 41;
+static const int c_lexer_en_doctype = 43;
+static const int c_lexer_en_xml_decl = 55;
+static const int c_lexer_en_element_name = 60;
+static const int c_lexer_en_element_head = 62;
+static const int c_lexer_en_text = 64;
+static const int c_lexer_en_main = 35;
+
+
+#line 65 "ext/c/lexer.rl"
+
+/**
+ * Lexes the String specifies as the method argument. Token values have the
+ * same encoding as the input value.
+ *
+ * This method keeps track of an internal state using the instance variables
+ * `@act` and `@cs`.
+ */
+VALUE oga_xml_lexer_advance(VALUE self, VALUE data_block)
+{
+    /* Make sure that all data passed back to Ruby has the proper encoding. */
+    rb_encoding *encoding = rb_enc_get(data_block);
+
+    char *data_str_val = StringValuePtr(data_block);
+
+    const char *p    = data_str_val;
+    const char *pe   = data_str_val + strlen(data_str_val);
+    const char *eof  = pe;
+    const char *ts   = 0;
+    const char *te   = 0;
+    const char *mark = 0;
+
+    int act = NUM2INT(oga_ivar_get(self, "@act"));
+    int cs  = NUM2INT(oga_ivar_get(self, "@cs"));
+
+    
+#line 110 "ext/c/lexer.c"
+	{
+	if ( p == pe )
+		goto _test_eof;
+	goto _resume;
+
+_again:
+	switch ( cs ) {
+		case 35: goto st35;
+		case 36: goto st36;
+		case 1: goto st1;
+		case 2: goto st2;
+		case 3: goto st3;
+		case 4: goto st4;
+		case 5: goto st5;
+		case 6: goto st6;
+		case 7: goto st7;
+		case 8: goto st8;
+		case 9: goto st9;
+		case 10: goto st10;
+		case 11: goto st11;
+		case 12: goto st12;
+		case 37: goto st37;
+		case 13: goto st13;
+		case 14: goto st14;
+		case 15: goto st15;
+		case 16: goto st16;
+		case 17: goto st17;
+		case 18: goto st18;
+		case 19: goto st19;
+		case 20: goto st20;
+		case 21: goto st21;
+		case 22: goto st22;
+		case 23: goto st23;
+		case 24: goto st24;
+		case 38: goto st38;
+		case 39: goto st39;
+		case 40: goto st40;
+		case 41: goto st41;
+		case 42: goto st42;
+		case 43: goto st43;
+		case 0: goto st0;
+		case 25: goto st25;
+		case 26: goto st26;
+		case 44: goto st44;
+		case 45: goto st45;
+		case 46: goto st46;
+		case 47: goto st47;
+		case 48: goto st48;
+		case 49: goto st49;
+		case 50: goto st50;
+		case 51: goto st51;
+		case 52: goto st52;
+		case 53: goto st53;
+		case 54: goto st54;
+		case 27: goto st27;
+		case 28: goto st28;
+		case 55: goto st55;
+		case 56: goto st56;
+		case 29: goto st29;
+		case 57: goto st57;
+		case 30: goto st30;
+		case 58: goto st58;
+		case 59: goto st59;
+		case 60: goto st60;
+		case 61: goto st61;
+		case 62: goto st62;
+		case 31: goto st31;
+		case 32: goto st32;
+		case 33: goto st33;
+		case 63: goto st63;
+		case 34: goto st34;
+		case 64: goto st64;
+		case 65: goto st65;
+		case 66: goto st66;
+	default: break;
+	}
+
+	if ( ++p == pe )
+		goto _test_eof;
+_resume:
+	switch ( cs )
+	{
+tr0:
+	cs = 35;
+#line 280 "ext/ragel/base_lexer.rl"
+	{{p = ((te))-1;}{
+        p--;
+        cs = 64;
+    }}
+	goto _again;
+tr7:
+#line 56 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+        callback("on_comment", data, encoding, ts + 4, te - 3);
+    }}
+	goto st35;
+tr23:
+#line 72 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+        callback("on_cdata", data, encoding, ts + 9, te - 3);
+    }}
+	goto st35;
+tr26:
+#line 222 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+        callback_simple("on_element_end");
+    }}
+	goto st35;
+tr44:
+	cs = 35;
+#line 280 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+        p--;
+        cs = 64;
+    }}
+	goto _again;
+tr46:
+	cs = 35;
+#line 280 "ext/ragel/base_lexer.rl"
+	{te = p;p--;{
+        p--;
+        cs = 64;
+    }}
+	goto _again;
+tr48:
+	cs = 35;
+#line 216 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+        callback_simple("on_element_start");
+        p--;
+        cs = 60;
+    }}
+	goto _again;
+tr50:
+	cs = 35;
+#line 140 "ext/ragel/base_lexer.rl"
+	{te = p;p--;{
+        callback_simple("on_doctype_start");
+        cs = 43;
+    }}
+	goto _again;
+tr51:
+	cs = 35;
+#line 1 "NONE"
+	{	switch( act ) {
+	case 25:
+	{{p = ((te))-1;}
+        callback_simple("on_xml_decl_start");
+        cs = 55;
+    }
+	break;
+	case 28:
+	{{p = ((te))-1;}
+        callback_simple("on_proc_ins_start");
+        callback("on_proc_ins_name", data, encoding, ts + 2, te);
+
+        mark = te;
+
+        cs = 41;
+    }
+	break;
+	}
+	}
+	goto _again;
+tr52:
+	cs = 35;
+#line 90 "ext/ragel/base_lexer.rl"
+	{te = p;p--;{
+        callback_simple("on_proc_ins_start");
+        callback("on_proc_ins_name", data, encoding, ts + 2, te);
+
+        mark = te;
+
+        cs = 41;
+    }}
+	goto _again;
+st35:
+#line 1 "NONE"
+	{ts = 0;}
+	if ( ++p == pe )
+		goto _test_eof35;
+case 35:
+#line 1 "NONE"
+	{ts = p;}
+#line 295 "ext/c/lexer.c"
+	if ( (*p) == 60 )
+		goto tr45;
+	goto tr44;
+tr45:
+#line 1 "NONE"
+	{te = p+1;}
+	goto st36;
+st36:
+	if ( ++p == pe )
+		goto _test_eof36;
+case 36:
+#line 307 "ext/c/lexer.c"
+	switch( (*p) ) {
+		case 33: goto st1;
+		case 45: goto tr48;
+		case 47: goto st22;
+		case 63: goto st24;
+		case 95: goto tr48;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr48;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr48;
+	} else
+		goto tr48;
+	goto tr46;
+st1:
+	if ( ++p == pe )
+		goto _test_eof1;
+case 1:
+	switch( (*p) ) {
+		case 45: goto st2;
+		case 68: goto st6;
+		case 91: goto st13;
+		case 100: goto st6;
+	}
+	goto tr0;
+st2:
+	if ( ++p == pe )
+		goto _test_eof2;
+case 2:
+	if ( (*p) == 45 )
+		goto st3;
+	goto tr0;
+st3:
+	if ( ++p == pe )
+		goto _test_eof3;
+case 3:
+	if ( (*p) == 45 )
+		goto st4;
+	goto st3;
+st4:
+	if ( ++p == pe )
+		goto _test_eof4;
+case 4:
+	if ( (*p) == 45 )
+		goto st5;
+	goto st3;
+st5:
+	if ( ++p == pe )
+		goto _test_eof5;
+case 5:
+	switch( (*p) ) {
+		case 45: goto st5;
+		case 62: goto tr7;
+	}
+	goto st3;
+st6:
+	if ( ++p == pe )
+		goto _test_eof6;
+case 6:
+	switch( (*p) ) {
+		case 79: goto st7;
+		case 111: goto st7;
+	}
+	goto tr0;
+st7:
+	if ( ++p == pe )
+		goto _test_eof7;
+case 7:
+	switch( (*p) ) {
+		case 67: goto st8;
+		case 99: goto st8;
+	}
+	goto tr0;
+st8:
+	if ( ++p == pe )
+		goto _test_eof8;
+case 8:
+	switch( (*p) ) {
+		case 84: goto st9;
+		case 116: goto st9;
+	}
+	goto tr0;
+st9:
+	if ( ++p == pe )
+		goto _test_eof9;
+case 9:
+	switch( (*p) ) {
+		case 89: goto st10;
+		case 121: goto st10;
+	}
+	goto tr0;
+st10:
+	if ( ++p == pe )
+		goto _test_eof10;
+case 10:
+	switch( (*p) ) {
+		case 80: goto st11;
+		case 112: goto st11;
+	}
+	goto tr0;
+st11:
+	if ( ++p == pe )
+		goto _test_eof11;
+case 11:
+	switch( (*p) ) {
+		case 69: goto st12;
+		case 101: goto st12;
+	}
+	goto tr0;
+st12:
+	if ( ++p == pe )
+		goto _test_eof12;
+case 12:
+	switch( (*p) ) {
+		case 9: goto st37;
+		case 32: goto st37;
+	}
+	goto tr0;
+st37:
+	if ( ++p == pe )
+		goto _test_eof37;
+case 37:
+	switch( (*p) ) {
+		case 9: goto st37;
+		case 32: goto st37;
+	}
+	goto tr50;
+st13:
+	if ( ++p == pe )
+		goto _test_eof13;
+case 13:
+	if ( (*p) == 67 )
+		goto st14;
+	goto tr0;
+st14:
+	if ( ++p == pe )
+		goto _test_eof14;
+case 14:
+	if ( (*p) == 68 )
+		goto st15;
+	goto tr0;
+st15:
+	if ( ++p == pe )
+		goto _test_eof15;
+case 15:
+	if ( (*p) == 65 )
+		goto st16;
+	goto tr0;
+st16:
+	if ( ++p == pe )
+		goto _test_eof16;
+case 16:
+	if ( (*p) == 84 )
+		goto st17;
+	goto tr0;
+st17:
+	if ( ++p == pe )
+		goto _test_eof17;
+case 17:
+	if ( (*p) == 65 )
+		goto st18;
+	goto tr0;
+st18:
+	if ( ++p == pe )
+		goto _test_eof18;
+case 18:
+	if ( (*p) == 91 )
+		goto st19;
+	goto tr0;
+st19:
+	if ( ++p == pe )
+		goto _test_eof19;
+case 19:
+	if ( (*p) == 93 )
+		goto st20;
+	goto st19;
+st20:
+	if ( ++p == pe )
+		goto _test_eof20;
+case 20:
+	if ( (*p) == 93 )
+		goto st21;
+	goto st19;
+st21:
+	if ( ++p == pe )
+		goto _test_eof21;
+case 21:
+	switch( (*p) ) {
+		case 62: goto tr23;
+		case 93: goto st21;
+	}
+	goto st19;
+st22:
+	if ( ++p == pe )
+		goto _test_eof22;
+case 22:
+	switch( (*p) ) {
+		case 45: goto st23;
+		case 95: goto st23;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto st23;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto st23;
+	} else
+		goto st23;
+	goto tr0;
+st23:
+	if ( ++p == pe )
+		goto _test_eof23;
+case 23:
+	switch( (*p) ) {
+		case 45: goto st23;
+		case 58: goto st22;
+		case 62: goto tr26;
+		case 95: goto st23;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto st23;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto st23;
+	} else
+		goto st23;
+	goto tr0;
+st24:
+	if ( ++p == pe )
+		goto _test_eof24;
+case 24:
+	switch( (*p) ) {
+		case 45: goto tr27;
+		case 95: goto tr27;
+		case 120: goto st39;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr27;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr27;
+	} else
+		goto tr27;
+	goto tr0;
+tr27:
+#line 1 "NONE"
+	{te = p+1;}
+#line 90 "ext/ragel/base_lexer.rl"
+	{act = 28;}
+	goto st38;
+tr54:
+#line 1 "NONE"
+	{te = p+1;}
+#line 182 "ext/ragel/base_lexer.rl"
+	{act = 25;}
+	goto st38;
+st38:
+	if ( ++p == pe )
+		goto _test_eof38;
+case 38:
+#line 572 "ext/c/lexer.c"
+	switch( (*p) ) {
+		case 45: goto tr27;
+		case 95: goto tr27;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr27;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr27;
+	} else
+		goto tr27;
+	goto tr51;
+st39:
+	if ( ++p == pe )
+		goto _test_eof39;
+case 39:
+	switch( (*p) ) {
+		case 45: goto tr27;
+		case 95: goto tr27;
+		case 109: goto st40;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr27;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr27;
+	} else
+		goto tr27;
+	goto tr52;
+st40:
+	if ( ++p == pe )
+		goto _test_eof40;
+case 40:
+	switch( (*p) ) {
+		case 45: goto tr27;
+		case 95: goto tr27;
+		case 108: goto tr54;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr27;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr27;
+	} else
+		goto tr27;
+	goto tr52;
+tr55:
+#line 107 "ext/ragel/base_lexer.rl"
+	{te = p+1;}
+	goto st41;
+tr57:
+#line 107 "ext/ragel/base_lexer.rl"
+	{te = p;p--;}
+	goto st41;
+tr58:
+	cs = 41;
+#line 100 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+            callback("on_text", data, encoding, mark, ts);
+            callback_simple("on_proc_ins_end");
+
+            cs = 35;
+        }}
+	goto _again;
+st41:
+#line 1 "NONE"
+	{ts = 0;}
+	if ( ++p == pe )
+		goto _test_eof41;
+case 41:
+#line 1 "NONE"
+	{ts = p;}
+#line 648 "ext/c/lexer.c"
+	if ( (*p) == 63 )
+		goto st42;
+	goto tr55;
+st42:
+	if ( ++p == pe )
+		goto _test_eof42;
+case 42:
+	if ( (*p) == 62 )
+		goto tr58;
+	goto tr57;
+tr30:
+#line 124 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+        callback("on_string", data, encoding, ts + 1, te - 1);
+    }}
+	goto st43;
+tr33:
+#line 154 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+            callback("on_doctype_inline", data, encoding, ts + 1, te - 1);
+        }}
+	goto st43;
+tr59:
+#line 163 "ext/ragel/base_lexer.rl"
+	{te = p+1;}
+	goto st43;
+tr61:
+	cs = 43;
+#line 169 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+            callback_simple("on_doctype_end");
+            cs = 35;
+        }}
+	goto _again;
+tr65:
+#line 1 "NONE"
+	{	switch( act ) {
+	case 3:
+	{{p = ((te))-1;}
+            callback("on_doctype_type", data, encoding, ts, te);
+        }
+	break;
+	case 7:
+	{{p = ((te))-1;}
+            callback("on_doctype_name", data, encoding, ts, te);
+        }
+	break;
+	}
+	}
+	goto st43;
+tr66:
+#line 165 "ext/ragel/base_lexer.rl"
+	{te = p;p--;{
+            callback("on_doctype_name", data, encoding, ts, te);
+        }}
+	goto st43;
+st43:
+#line 1 "NONE"
+	{ts = 0;}
+	if ( ++p == pe )
+		goto _test_eof43;
+case 43:
+#line 1 "NONE"
+	{ts = p;}
+#line 713 "ext/c/lexer.c"
+	switch( (*p) ) {
+		case 9: goto tr59;
+		case 32: goto tr59;
+		case 34: goto st25;
+		case 39: goto st26;
+		case 45: goto tr60;
+		case 62: goto tr61;
+		case 80: goto st45;
+		case 83: goto st50;
+		case 91: goto st27;
+		case 95: goto tr60;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr60;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr60;
+	} else
+		goto tr60;
+	goto st0;
+st0:
+cs = 0;
+	goto _out;
+st25:
+	if ( ++p == pe )
+		goto _test_eof25;
+case 25:
+	if ( (*p) == 34 )
+		goto tr30;
+	goto st25;
+st26:
+	if ( ++p == pe )
+		goto _test_eof26;
+case 26:
+	if ( (*p) == 39 )
+		goto tr30;
+	goto st26;
+tr60:
+#line 1 "NONE"
+	{te = p+1;}
+#line 165 "ext/ragel/base_lexer.rl"
+	{act = 7;}
+	goto st44;
+tr71:
+#line 1 "NONE"
+	{te = p+1;}
+#line 148 "ext/ragel/base_lexer.rl"
+	{act = 3;}
+	goto st44;
+st44:
+	if ( ++p == pe )
+		goto _test_eof44;
+case 44:
+#line 768 "ext/c/lexer.c"
+	switch( (*p) ) {
+		case 45: goto tr60;
+		case 95: goto tr60;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr60;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr60;
+	} else
+		goto tr60;
+	goto tr65;
+st45:
+	if ( ++p == pe )
+		goto _test_eof45;
+case 45:
+	switch( (*p) ) {
+		case 45: goto tr60;
+		case 85: goto st46;
+		case 95: goto tr60;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr60;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr60;
+	} else
+		goto tr60;
+	goto tr66;
+st46:
+	if ( ++p == pe )
+		goto _test_eof46;
+case 46:
+	switch( (*p) ) {
+		case 45: goto tr60;
+		case 66: goto st47;
+		case 95: goto tr60;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr60;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr60;
+	} else
+		goto tr60;
+	goto tr66;
+st47:
+	if ( ++p == pe )
+		goto _test_eof47;
+case 47:
+	switch( (*p) ) {
+		case 45: goto tr60;
+		case 76: goto st48;
+		case 95: goto tr60;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr60;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr60;
+	} else
+		goto tr60;
+	goto tr66;
+st48:
+	if ( ++p == pe )
+		goto _test_eof48;
+case 48:
+	switch( (*p) ) {
+		case 45: goto tr60;
+		case 73: goto st49;
+		case 95: goto tr60;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr60;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr60;
+	} else
+		goto tr60;
+	goto tr66;
+st49:
+	if ( ++p == pe )
+		goto _test_eof49;
+case 49:
+	switch( (*p) ) {
+		case 45: goto tr60;
+		case 67: goto tr71;
+		case 95: goto tr60;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr60;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr60;
+	} else
+		goto tr60;
+	goto tr66;
+st50:
+	if ( ++p == pe )
+		goto _test_eof50;
+case 50:
+	switch( (*p) ) {
+		case 45: goto tr60;
+		case 89: goto st51;
+		case 95: goto tr60;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr60;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr60;
+	} else
+		goto tr60;
+	goto tr66;
+st51:
+	if ( ++p == pe )
+		goto _test_eof51;
+case 51:
+	switch( (*p) ) {
+		case 45: goto tr60;
+		case 83: goto st52;
+		case 95: goto tr60;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr60;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr60;
+	} else
+		goto tr60;
+	goto tr66;
+st52:
+	if ( ++p == pe )
+		goto _test_eof52;
+case 52:
+	switch( (*p) ) {
+		case 45: goto tr60;
+		case 84: goto st53;
+		case 95: goto tr60;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr60;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr60;
+	} else
+		goto tr60;
+	goto tr66;
+st53:
+	if ( ++p == pe )
+		goto _test_eof53;
+case 53:
+	switch( (*p) ) {
+		case 45: goto tr60;
+		case 69: goto st54;
+		case 95: goto tr60;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr60;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr60;
+	} else
+		goto tr60;
+	goto tr66;
+st54:
+	if ( ++p == pe )
+		goto _test_eof54;
+case 54:
+	switch( (*p) ) {
+		case 45: goto tr60;
+		case 77: goto tr71;
+		case 95: goto tr60;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto tr60;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr60;
+	} else
+		goto tr60;
+	goto tr66;
+st27:
+	if ( ++p == pe )
+		goto _test_eof27;
+case 27:
+	goto st28;
+st28:
+	if ( ++p == pe )
+		goto _test_eof28;
+case 28:
+	if ( (*p) == 93 )
+		goto tr33;
+	goto st28;
+tr34:
+#line 201 "ext/ragel/base_lexer.rl"
+	{{p = ((te))-1;}}
+	goto st55;
+tr36:
+#line 124 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+        callback("on_string", data, encoding, ts + 1, te - 1);
+    }}
+	goto st55;
+tr76:
+#line 201 "ext/ragel/base_lexer.rl"
+	{te = p+1;}
+	goto st55;
+tr81:
+#line 201 "ext/ragel/base_lexer.rl"
+	{te = p;p--;}
+	goto st55;
+tr82:
+#line 195 "ext/ragel/base_lexer.rl"
+	{te = p;p--;{
+            callback("on_attribute", data, encoding, ts, te);
+        }}
+	goto st55;
+tr83:
+	cs = 55;
+#line 189 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+            callback_simple("on_xml_decl_end");
+            cs = 35;
+        }}
+	goto _again;
+st55:
+#line 1 "NONE"
+	{ts = 0;}
+	if ( ++p == pe )
+		goto _test_eof55;
+case 55:
+#line 1 "NONE"
+	{ts = p;}
+#line 1014 "ext/c/lexer.c"
+	switch( (*p) ) {
+		case 34: goto tr77;
+		case 39: goto tr78;
+		case 45: goto st58;
+		case 63: goto st59;
+		case 95: goto st58;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto st58;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto st58;
+	} else
+		goto st58;
+	goto tr76;
+tr77:
+#line 1 "NONE"
+	{te = p+1;}
+	goto st56;
+st56:
+	if ( ++p == pe )
+		goto _test_eof56;
+case 56:
+#line 1039 "ext/c/lexer.c"
+	if ( (*p) == 34 )
+		goto tr36;
+	goto st29;
+st29:
+	if ( ++p == pe )
+		goto _test_eof29;
+case 29:
+	if ( (*p) == 34 )
+		goto tr36;
+	goto st29;
+tr78:
+#line 1 "NONE"
+	{te = p+1;}
+	goto st57;
+st57:
+	if ( ++p == pe )
+		goto _test_eof57;
+case 57:
+#line 1058 "ext/c/lexer.c"
+	if ( (*p) == 39 )
+		goto tr36;
+	goto st30;
+st30:
+	if ( ++p == pe )
+		goto _test_eof30;
+case 30:
+	if ( (*p) == 39 )
+		goto tr36;
+	goto st30;
+st58:
+	if ( ++p == pe )
+		goto _test_eof58;
+case 58:
+	switch( (*p) ) {
+		case 45: goto st58;
+		case 95: goto st58;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto st58;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto st58;
+	} else
+		goto st58;
+	goto tr82;
+st59:
+	if ( ++p == pe )
+		goto _test_eof59;
+case 59:
+	if ( (*p) == 62 )
+		goto tr83;
+	goto tr81;
+tr85:
+	cs = 60;
+#line 232 "ext/ragel/base_lexer.rl"
+	{te = p;p--;{
+            callback("on_element_name", data, encoding, ts, te);
+            cs = 62;
+        }}
+	goto _again;
+tr86:
+#line 228 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+            callback("on_element_ns", data, encoding, ts, te - 1);
+        }}
+	goto st60;
+st60:
+#line 1 "NONE"
+	{ts = 0;}
+	if ( ++p == pe )
+		goto _test_eof60;
+case 60:
+#line 1 "NONE"
+	{ts = p;}
+#line 1115 "ext/c/lexer.c"
+	switch( (*p) ) {
+		case 45: goto st61;
+		case 95: goto st61;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto st61;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto st61;
+	} else
+		goto st61;
+	goto st0;
+st61:
+	if ( ++p == pe )
+		goto _test_eof61;
+case 61:
+	switch( (*p) ) {
+		case 45: goto st61;
+		case 58: goto tr86;
+		case 95: goto st61;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto st61;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto st61;
+	} else
+		goto st61;
+	goto tr85;
+tr38:
+#line 243 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+            callback_simple("advance_line");
+        }}
+	goto st62;
+tr41:
+#line 124 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+        callback("on_string", data, encoding, ts + 1, te - 1);
+    }}
+	goto st62;
+tr43:
+	cs = 62;
+#line 266 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+            callback_simple("on_element_end");
+            cs = 35;
+        }}
+	goto _again;
+tr87:
+#line 241 "ext/ragel/base_lexer.rl"
+	{te = p+1;}
+	goto st62;
+tr91:
+	cs = 62;
+#line 260 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+            callback_simple("on_element_open_end");
+            cs = 35;
+        }}
+	goto _again;
+tr92:
+#line 252 "ext/ragel/base_lexer.rl"
+	{te = p;p--;{
+            callback("on_attribute", data, encoding, ts, te);
+        }}
+	goto st62;
+tr93:
+#line 248 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+            callback("on_attribute_ns", data, encoding, ts, te - 1);
+        }}
+	goto st62;
+st62:
+#line 1 "NONE"
+	{ts = 0;}
+	if ( ++p == pe )
+		goto _test_eof62;
+case 62:
+#line 1 "NONE"
+	{ts = p;}
+#line 1199 "ext/c/lexer.c"
+	switch( (*p) ) {
+		case 9: goto tr87;
+		case 10: goto tr38;
+		case 13: goto st31;
+		case 32: goto tr87;
+		case 34: goto st32;
+		case 39: goto st33;
+		case 45: goto st63;
+		case 47: goto st34;
+		case 61: goto tr87;
+		case 62: goto tr91;
+		case 95: goto st63;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto st63;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto st63;
+	} else
+		goto st63;
+	goto st0;
+st31:
+	if ( ++p == pe )
+		goto _test_eof31;
+case 31:
+	if ( (*p) == 10 )
+		goto tr38;
+	goto st0;
+st32:
+	if ( ++p == pe )
+		goto _test_eof32;
+case 32:
+	if ( (*p) == 34 )
+		goto tr41;
+	goto st32;
+st33:
+	if ( ++p == pe )
+		goto _test_eof33;
+case 33:
+	if ( (*p) == 39 )
+		goto tr41;
+	goto st33;
+st63:
+	if ( ++p == pe )
+		goto _test_eof63;
+case 63:
+	switch( (*p) ) {
+		case 45: goto st63;
+		case 58: goto tr93;
+		case 95: goto st63;
+	}
+	if ( (*p) < 65 ) {
+		if ( 48 <= (*p) && (*p) <= 57 )
+			goto st63;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto st63;
+	} else
+		goto st63;
+	goto tr92;
+st34:
+	if ( ++p == pe )
+		goto _test_eof34;
+case 34:
+	if ( (*p) == 62 )
+		goto tr43;
+	goto st0;
+tr96:
+	cs = 64;
+#line 1 "NONE"
+	{	switch( act ) {
+	case 0:
+	{{goto st0;}}
+	break;
+	case 23:
+	{{p = ((te))-1;}
+            callback("on_text", data, encoding, ts, te);
+            cs = 35;
+        }
+	break;
+	}
+	}
+	goto _again;
+tr97:
+	cs = 64;
+#line 306 "ext/ragel/base_lexer.rl"
+	{te = p;p--;{
+            callback("on_text", data, encoding, ts, te);
+            cs = 35;
+        }}
+	goto _again;
+tr98:
+	cs = 64;
+#line 296 "ext/ragel/base_lexer.rl"
+	{te = p+1;{
+            callback("on_text", data, encoding, ts, mark);
+
+            p    = mark - 1;
+            mark = 0;
+
+            cs = 35;
+        }}
+	goto _again;
+st64:
+#line 1 "NONE"
+	{ts = 0;}
+#line 1 "NONE"
+	{act = 0;}
+	if ( ++p == pe )
+		goto _test_eof64;
+case 64:
+#line 1 "NONE"
+	{ts = p;}
+#line 1314 "ext/c/lexer.c"
+	if ( (*p) == 60 )
+		goto tr95;
+	goto tr94;
+tr94:
+#line 1 "NONE"
+	{te = p+1;}
+#line 296 "ext/ragel/base_lexer.rl"
+	{ mark = p; }
+#line 306 "ext/ragel/base_lexer.rl"
+	{act = 23;}
+	goto st65;
+st65:
+	if ( ++p == pe )
+		goto _test_eof65;
+case 65:
+#line 1330 "ext/c/lexer.c"
+	if ( (*p) == 60 )
+		goto tr95;
+	goto tr94;
+tr95:
+#line 296 "ext/ragel/base_lexer.rl"
+	{ mark = p; }
+	goto st66;
+st66:
+	if ( ++p == pe )
+		goto _test_eof66;
+case 66:
+#line 1342 "ext/c/lexer.c"
+	switch( (*p) ) {
+		case 33: goto tr98;
+		case 45: goto tr98;
+		case 60: goto tr95;
+		case 63: goto tr98;
+		case 95: goto tr98;
+	}
+	if ( (*p) < 65 ) {
+		if ( 47 <= (*p) && (*p) <= 57 )
+			goto tr98;
+	} else if ( (*p) > 90 ) {
+		if ( 97 <= (*p) && (*p) <= 122 )
+			goto tr98;
+	} else
+		goto tr98;
+	goto tr94;
+	}
+	_test_eof35: cs = 35; goto _test_eof; 
+	_test_eof36: cs = 36; goto _test_eof; 
+	_test_eof1: cs = 1; goto _test_eof; 
+	_test_eof2: cs = 2; goto _test_eof; 
+	_test_eof3: cs = 3; goto _test_eof; 
+	_test_eof4: cs = 4; goto _test_eof; 
+	_test_eof5: cs = 5; goto _test_eof; 
+	_test_eof6: cs = 6; goto _test_eof; 
+	_test_eof7: cs = 7; goto _test_eof; 
+	_test_eof8: cs = 8; goto _test_eof; 
+	_test_eof9: cs = 9; goto _test_eof; 
+	_test_eof10: cs = 10; goto _test_eof; 
+	_test_eof11: cs = 11; goto _test_eof; 
+	_test_eof12: cs = 12; goto _test_eof; 
+	_test_eof37: cs = 37; goto _test_eof; 
+	_test_eof13: cs = 13; goto _test_eof; 
+	_test_eof14: cs = 14; goto _test_eof; 
+	_test_eof15: cs = 15; goto _test_eof; 
+	_test_eof16: cs = 16; goto _test_eof; 
+	_test_eof17: cs = 17; goto _test_eof; 
+	_test_eof18: cs = 18; goto _test_eof; 
+	_test_eof19: cs = 19; goto _test_eof; 
+	_test_eof20: cs = 20; goto _test_eof; 
+	_test_eof21: cs = 21; goto _test_eof; 
+	_test_eof22: cs = 22; goto _test_eof; 
+	_test_eof23: cs = 23; goto _test_eof; 
+	_test_eof24: cs = 24; goto _test_eof; 
+	_test_eof38: cs = 38; goto _test_eof; 
+	_test_eof39: cs = 39; goto _test_eof; 
+	_test_eof40: cs = 40; goto _test_eof; 
+	_test_eof41: cs = 41; goto _test_eof; 
+	_test_eof42: cs = 42; goto _test_eof; 
+	_test_eof43: cs = 43; goto _test_eof; 
+	_test_eof25: cs = 25; goto _test_eof; 
+	_test_eof26: cs = 26; goto _test_eof; 
+	_test_eof44: cs = 44; goto _test_eof; 
+	_test_eof45: cs = 45; goto _test_eof; 
+	_test_eof46: cs = 46; goto _test_eof; 
+	_test_eof47: cs = 47; goto _test_eof; 
+	_test_eof48: cs = 48; goto _test_eof; 
+	_test_eof49: cs = 49; goto _test_eof; 
+	_test_eof50: cs = 50; goto _test_eof; 
+	_test_eof51: cs = 51; goto _test_eof; 
+	_test_eof52: cs = 52; goto _test_eof; 
+	_test_eof53: cs = 53; goto _test_eof; 
+	_test_eof54: cs = 54; goto _test_eof; 
+	_test_eof27: cs = 27; goto _test_eof; 
+	_test_eof28: cs = 28; goto _test_eof; 
+	_test_eof55: cs = 55; goto _test_eof; 
+	_test_eof56: cs = 56; goto _test_eof; 
+	_test_eof29: cs = 29; goto _test_eof; 
+	_test_eof57: cs = 57; goto _test_eof; 
+	_test_eof30: cs = 30; goto _test_eof; 
+	_test_eof58: cs = 58; goto _test_eof; 
+	_test_eof59: cs = 59; goto _test_eof; 
+	_test_eof60: cs = 60; goto _test_eof; 
+	_test_eof61: cs = 61; goto _test_eof; 
+	_test_eof62: cs = 62; goto _test_eof; 
+	_test_eof31: cs = 31; goto _test_eof; 
+	_test_eof32: cs = 32; goto _test_eof; 
+	_test_eof33: cs = 33; goto _test_eof; 
+	_test_eof63: cs = 63; goto _test_eof; 
+	_test_eof34: cs = 34; goto _test_eof; 
+	_test_eof64: cs = 64; goto _test_eof; 
+	_test_eof65: cs = 65; goto _test_eof; 
+	_test_eof66: cs = 66; goto _test_eof; 
+
+	_test_eof: {}
+	if ( p == eof )
+	{
+	switch ( cs ) {
+	case 36: goto tr46;
+	case 1: goto tr0;
+	case 2: goto tr0;
+	case 3: goto tr0;
+	case 4: goto tr0;
+	case 5: goto tr0;
+	case 6: goto tr0;
+	case 7: goto tr0;
+	case 8: goto tr0;
+	case 9: goto tr0;
+	case 10: goto tr0;
+	case 11: goto tr0;
+	case 12: goto tr0;
+	case 37: goto tr50;
+	case 13: goto tr0;
+	case 14: goto tr0;
+	case 15: goto tr0;
+	case 16: goto tr0;
+	case 17: goto tr0;
+	case 18: goto tr0;
+	case 19: goto tr0;
+	case 20: goto tr0;
+	case 21: goto tr0;
+	case 22: goto tr0;
+	case 23: goto tr0;
+	case 24: goto tr0;
+	case 38: goto tr51;
+	case 39: goto tr52;
+	case 40: goto tr52;
+	case 42: goto tr57;
+	case 44: goto tr65;
+	case 45: goto tr66;
+	case 46: goto tr66;
+	case 47: goto tr66;
+	case 48: goto tr66;
+	case 49: goto tr66;
+	case 50: goto tr66;
+	case 51: goto tr66;
+	case 52: goto tr66;
+	case 53: goto tr66;
+	case 54: goto tr66;
+	case 56: goto tr81;
+	case 29: goto tr34;
+	case 57: goto tr81;
+	case 30: goto tr34;
+	case 58: goto tr82;
+	case 59: goto tr81;
+	case 61: goto tr85;
+	case 63: goto tr92;
+	case 65: goto tr96;
+	case 66: goto tr97;
+	}
+	}
+
+	_out: {}
+	}
+
+#line 91 "ext/c/lexer.rl"
+
+    oga_ivar_set(self, "@act", INT2NUM(act));
+    oga_ivar_set(self, "@cs", INT2NUM(cs));
+
+    return Qnil;
+}
+
+/**
+ * Resets the internal state of the lexer.
+ */
+VALUE oga_xml_lexer_reset(VALUE self)
+{
+    oga_ivar_set(self, "@act", INT2NUM(0));
+    oga_ivar_set(self, "@cs", INT2NUM(c_lexer_start));
+
+    return Qnil;
+}
+
+
+#line 111 "ext/c/lexer.rl"
+
+
+void Init_liboga_xml_lexer()
+{
+    VALUE mOga   = rb_const_get(rb_cObject, rb_intern("Oga"));
+    VALUE mXML   = rb_const_get(mOga, rb_intern("XML"));
+    VALUE cLexer = rb_define_class_under(mXML, "Lexer", rb_cObject);
+
+    rb_define_method(cLexer, "advance_native", oga_xml_lexer_advance, 1);
+    rb_define_method(cLexer, "reset_native", oga_xml_lexer_reset, 0);
+}