Implemented better DOCTYPE parsing, added more DOCTYPE tests

git-svn-id: http://pugixml.googlecode.com/svn/trunk@409 99668b35-9821-0410-8761-19e4c4f06640

2 files changed, 92 insertions, 29 deletions

diff --git a/tests/test_parse.cpp b/tests/test_parse.cpp
index 0719e5d..fb0dd23 100644
--- a/tests/test_parse.cpp
+++ b/tests/test_parse.cpp
@@ -497,35 +497,6 @@ TEST(parse_declaration_error)
 	CHECK(doc.load(STR("<?xml version='1?>"), parse_minimal | parse_declaration).status == status_bad_attribute);

 }

 

-TEST(parse_doctype_skip)

-{

-	xml_document doc;

-	CHECK(doc.load(STR("<!DOCTYPE doc>")) && !doc.first_child());

-	CHECK(doc.load(STR("<!DOCTYPE doc SYSTEM 'foo'>")) && !doc.first_child());

-	CHECK(doc.load(STR("<!DOCTYPE doc SYSTEM \"foo\">")) && !doc.first_child());

-	CHECK(doc.load(STR("<!DOCTYPE doc PUBLIC \"foo\" 'bar'>")) && !doc.first_child());

-	CHECK(doc.load(STR("<!DOCTYPE doc PUBLIC \"foo'\">")) && !doc.first_child());

-	CHECK(doc.load(STR("<!DOCTYPE doc SYSTEM 'foo' [<!ELEMENT foo 'ANY'>]>")) && !doc.first_child());

-

-	CHECK(doc.load(STR("<!DOCTYPE doc SYSTEM 'foo' [<!ELEMENT foo 'ANY'>]><node/>")));

-	CHECK_NODE(doc, STR("<node />"));

-}

-

-TEST(parse_doctype_error)

-{

-	xml_document doc;

-	CHECK(doc.load(STR("<!DOCTYPE")).status == status_bad_doctype);

-	CHECK(doc.load(STR("<!DOCTYPE doc")).status == status_bad_doctype);

-	CHECK(doc.load(STR("<!DOCTYPE doc SYSTEM 'foo")).status == status_bad_doctype);

-	CHECK(doc.load(STR("<!DOCTYPE doc SYSTEM \"foo")).status == status_bad_doctype);

-	CHECK(doc.load(STR("<!DOCTYPE doc PUBLIC \"foo\" 'bar")).status == status_bad_doctype);

-	CHECK(doc.load(STR("<!DOCTYPE doc PUBLIC \"foo'\"")).status == status_bad_doctype);

-	CHECK(doc.load(STR("<!DOCTYPE doc SYSTEM 'foo' [<!ELEMENT foo 'ANY")).status == status_bad_doctype);

-	CHECK(doc.load(STR("<!DOCTYPE doc SYSTEM 'foo' [<!ELEMENT foo 'ANY'>")).status == status_bad_doctype);

-	CHECK(doc.load(STR("<!DOCTYPE doc SYSTEM 'foo' [<!ELEMENT foo 'ANY'>]")).status == status_bad_doctype);

-	CHECK(doc.load(STR("<!DOCTYPE doc SYSTEM 'foo' [<!ELEMENT foo 'ANY'>] ")).status == status_bad_doctype);

-}

-

 TEST(parse_empty)

 {

 	xml_document doc;

diff --git a/tests/test_parse_doctype.cpp b/tests/test_parse_doctype.cpp
new file mode 100644
index 0000000..35015ff
--- /dev/null
+++ b/tests/test_parse_doctype.cpp
@@ -0,0 +1,92 @@
+#include "common.hpp"

+

+#include <string>

+

+bool test_doctype_wf(const std::basic_string<char_t>& decl)

+{

+	xml_document doc;

+

+	// standalone

+	if (!doc.load(decl.c_str()) || doc.first_child()) return false;

+

+	// pcdata pre/postfix

+	if (!doc.load(("a" + decl).c_str()) || doc.first_child()) return false;

+	if (!doc.load((decl + "b").c_str()) || doc.first_child()) return false;

+	if (!doc.load(("a" + decl + "b").c_str()) || doc.first_child()) return false;

+

+	// node pre/postfix

+	if (!doc.load(("<nodea/>" + decl).c_str()) || !test_node(doc, STR("<nodea />"), STR(""), format_raw)) return false;

+	if (!doc.load((decl + "<nodeb/>").c_str()) || !test_node(doc, STR("<nodeb />"), STR(""), format_raw)) return false;

+	if (!doc.load(("<nodea/>" + decl + "<nodeb/>").c_str()) || !test_node(doc, STR("<nodea /><nodeb />"), STR(""), format_raw)) return false;

+

+	return true;

+}

+

+bool test_doctype_nwf(const std::basic_string<char_t>& decl)

+{

+	xml_document doc;

+

+	// standalone

+	if (doc.load(decl.c_str()).status != status_bad_doctype) return false;

+

+	// pcdata postfix

+	if (doc.load((decl + "b").c_str()).status != status_bad_doctype) return false;

+

+	// node postfix

+	if (doc.load((decl + "<nodeb/>").c_str()).status != status_bad_doctype) return false;

+

+	return true;

+}

+

+#define TEST_DOCTYPE_WF(contents) CHECK(test_doctype_wf(STR(contents)))

+#define TEST_DOCTYPE_NWF(contents) CHECK(test_doctype_nwf(STR(contents)))

+

+TEST(parse_doctype_skip)

+{

+	TEST_DOCTYPE_WF("<!DOCTYPE doc>");

+	TEST_DOCTYPE_WF("<!DOCTYPE doc SYSTEM 'foo'>");

+	TEST_DOCTYPE_WF("<!DOCTYPE doc SYSTEM \"foo\">");

+	TEST_DOCTYPE_WF("<!DOCTYPE doc PUBLIC \"foo\" 'bar'>");

+	TEST_DOCTYPE_WF("<!DOCTYPE doc PUBLIC \"foo'\">");

+	TEST_DOCTYPE_WF("<!DOCTYPE doc SYSTEM 'foo' [<!ELEMENT foo 'ANY'>]>");

+}

+

+TEST(parse_doctype_error)

+{

+	TEST_DOCTYPE_NWF("<!DOCTYPE");

+	TEST_DOCTYPE_NWF("<!DOCTYPE doc");

+	TEST_DOCTYPE_NWF("<!DOCTYPE doc SYSTEM 'foo");

+	TEST_DOCTYPE_NWF("<!DOCTYPE doc SYSTEM \"foo");

+	TEST_DOCTYPE_NWF("<!DOCTYPE doc PUBLIC \"foo\" 'bar");

+	TEST_DOCTYPE_NWF("<!DOCTYPE doc PUBLIC \"foo'\"");

+	TEST_DOCTYPE_NWF("<!DOCTYPE doc SYSTEM 'foo' [<!ELEMENT foo 'ANY");

+	TEST_DOCTYPE_NWF("<!DOCTYPE doc SYSTEM 'foo' [<!ELEMENT foo 'ANY'>");

+	TEST_DOCTYPE_NWF("<!DOCTYPE doc SYSTEM 'foo' [<!ELEMENT foo 'ANY'>]");

+	TEST_DOCTYPE_NWF("<!DOCTYPE doc SYSTEM 'foo' [<!ELEMENT foo 'ANY'>] ");

+}

+

+// Examples from W3C recommendations

+TEST(parse_doctype_w3c_wf)

+{

+	TEST_DOCTYPE_WF("<!DOCTYPE greeting SYSTEM \"hello.dtd\">");

+	TEST_DOCTYPE_WF("<!DOCTYPE greeting [ <!ELEMENT greeting (#PCDATA)> ]>");

+	TEST_DOCTYPE_WF("<!DOCTYPE greeting [ <!ATTLIST list type    (bullets|ordered|glossary)  \"ordered\"> <!ATTLIST form method  CDATA   #FIXED \"POST\"> ]>");

+	TEST_DOCTYPE_WF("<!DOCTYPE greeting [ <!ENTITY % draft 'INCLUDE' > <!ENTITY % final 'IGNORE' > <![%draft;[ <!ELEMENT book (comments*, title, body, supplements?)> ]]> <![%final;[ <!ELEMENT book (title, body, supplements?)> ]]>]>");

+	TEST_DOCTYPE_WF("<!DOCTYPE greeting [ <!ENTITY open-hatch PUBLIC \"-//Textuality//TEXT Standard open-hatch boilerplate//EN\" \"http://www.textuality.com/boilerplate/OpenHatch.xml\"> ]>");

+	TEST_DOCTYPE_WF("<!DOCTYPE greeting [ <!ENTITY EndAttr \"27'\" > ]>");

+}

+

+TEST(parse_doctype_w3c_nwf)

+{

+	TEST_DOCTYPE_NWF("<!DOCTYPE greeting SYSTEM \"hello.dtd>");

+	TEST_DOCTYPE_NWF("<!DOCTYPE greeting SYSTEM");

+	TEST_DOCTYPE_NWF("<!DOCTYPE greeting [ <!ELEMENT greeting (#PCDATA)> ]");

+	TEST_DOCTYPE_NWF("<!DOCTYPE greeting [ <!ELEMENT greeting (#PCDATA)>");

+	TEST_DOCTYPE_NWF("<!DOCTYPE greeting [ <!ELEMENT greeting (#PCDATA");

+	TEST_DOCTYPE_NWF("<!DOCTYPE greeting [ ");

+	TEST_DOCTYPE_NWF("<!DOCTYPE greeting [ <!ATTLIST list type    (bullets|ordered|glossary)  \"ordered\"> ]");

+	TEST_DOCTYPE_NWF("<!DOCTYPE greeting [ <!ATTLIST list type    (bullets|ordered|glossary)  \"ordered\">");

+	TEST_DOCTYPE_NWF("<!DOCTYPE greeting [ <!ATTLIST list type    (bullets|ordered|glossary)  \"orde");

+	TEST_DOCTYPE_NWF("<!DOCTYPE greeting [ <!ATTLIST list type    (bullets|ordered|glossary) ");

+	TEST_DOCTYPE_NWF("<!DOCTYPE greeting [ <!ENTITY open-hatch PUBLIC \"-//Textuality//TEXT Standard open-hatch boilerplate//EN\" \"http://www.textuality.com/boilerplate/OpenHatch.x");

+}