From 7bda2cb529550fae678224102e1c5d0697dacf04 Mon Sep 17 00:00:00 2001
From: "arseny.kapoulkine"
 <arseny.kapoulkine@99668b35-9821-0410-8761-19e4c4f06640>
Date: Mon, 10 May 2010 19:15:44 +0000
Subject: Implemented better DOCTYPE parsing, added more DOCTYPE tests

git-svn-id: http://pugixml.googlecode.com/svn/trunk@409 99668b35-9821-0410-8761-19e4c4f06640
---
 src/pugixml.cpp              | 152 ++++++++++++++++++++++++++++++++-----------
 tests/test_parse.cpp         |  29 ---------
 tests/test_parse_doctype.cpp |  92 ++++++++++++++++++++++++++
 3 files changed, 206 insertions(+), 67 deletions(-)
 create mode 100644 tests/test_parse_doctype.cpp

diff --git a/src/pugixml.cpp b/src/pugixml.cpp
index aa10bbc..dcb25f7 100644
--- a/src/pugixml.cpp
+++ b/src/pugixml.cpp
@@ -1739,6 +1739,116 @@ namespace
 		{
 		}
 
+		// DOCTYPE consists of nested sections of the following possible types:
+		// <!-- ... -->, <? ... ?>, "...", '...'
+		// <![...]]>
+		// <!...>
+		// First group can not contain nested groups
+		// Second group can contain nested groups of the same type
+		// Third group can contain all other groups
+		xml_parse_result parse_doctype_primitive(char_t*& s, char_t* buffer_start)
+		{
+			if (*s == '"' || *s == '\'')
+			{
+				// quoted string
+				char_t ch = *s++;
+				SCANFOR(*s == ch);
+				if (!*s) THROW_ERROR(status_bad_doctype, s);
+
+				s++;
+			}
+			else if (s[0] == '<' && s[1] == '?')
+			{
+				// <? ... ?>
+				s += 2;
+				SCANFOR(s[0] == '?' && s[1] == '>'); // no need for ENDSWITH because ?> can't terminate proper doctype
+				if (!*s) THROW_ERROR(status_bad_doctype, s);
+
+				s += 2;
+			}
+			else if (s[0] == '<' && s[1] == '!' && s[2] == '-' && s[3] == '-')
+			{
+				s += 4;
+				SCANFOR(s[0] == '-' && s[1] == '-' && s[2] == '>'); // no need for ENDSWITH because --> can't terminate proper doctype
+				if (!*s) THROW_ERROR(status_bad_doctype, s);
+
+				s += 4;
+			}
+			else THROW_ERROR(status_bad_doctype, s);
+
+			THROW_ERROR(status_ok, s);
+		}
+
+		xml_parse_result parse_doctype_ignore(char_t*& s, char_t* buffer_start)
+		{
+			assert(s[0] == '<' && s[1] == '!' && s[2] == '[');
+			s++;
+
+			while (*s)
+			{
+				if (s[0] == '<' && s[1] == '!' && s[2] == '[')
+				{
+					// nested ignore section
+					xml_parse_result res = parse_doctype_ignore(s, buffer_start);
+
+					if (!res) return res;
+				}
+				else if (s[0] == ']' && s[1] == ']' && s[2] == '>')
+				{
+					// ignore section end
+					s += 3;
+
+					THROW_ERROR(status_ok, s);
+				}
+				else s++;
+			}
+
+			THROW_ERROR(status_bad_doctype, s);
+		}
+
+		xml_parse_result parse_doctype(char_t*& s, char_t* buffer_start, char_t endch, bool toplevel)
+		{
+			assert(s[0] == '<' && s[1] == '!');
+			s++;
+
+			while (*s)
+			{
+				if (s[0] == '<' && s[1] == '!' && s[2] != '-')
+				{
+					if (s[2] == '[')
+					{
+						// ignore
+						xml_parse_result res = parse_doctype_ignore(s, buffer_start);
+
+						if (!res) return res;
+					}
+					else
+					{
+						// some control group
+						xml_parse_result res = parse_doctype(s, buffer_start, endch, false);
+
+						if (!res) return res;
+					}
+				}
+				else if (s[0] == '<' || s[0] == '"' || s[0] == '\'')
+				{
+					// unknown tag (forbidden), or some primitive group
+					xml_parse_result res = parse_doctype_primitive(s, buffer_start);
+
+					if (!res) return res;
+				}
+				else if (*s == '>')
+				{
+					s++;
+
+					THROW_ERROR(status_ok, s);
+				}
+				else s++;
+			}
+
+			THROW_ERROR((toplevel && endch == '>') ? status_ok : status_bad_doctype, s);
+		}
+
 		xml_parse_result parse_exclamation(char_t*& ref_s, xml_node_struct* cursor, unsigned int optmsk, char_t* buffer_start, char_t endch)
 		{
 			// load into registers
@@ -1831,47 +1941,13 @@ namespace
 			}
 			else if (s[0] == 'D' && s[1] == 'O' && s[2] == 'C' && s[3] == 'T' && s[4] == 'Y' && s[5] == 'P' && ENDSWITH(s[6], 'E'))
 			{
-				if (s[6] != 'E') THROW_ERROR(status_bad_doctype, s); 
+				if (s[6] != 'E') THROW_ERROR(status_bad_doctype, s);
 
-			LOC_DOCTYPE:
-				SCANFOR(*s == '\'' || *s == '"' || *s == '[' || *s == '>');
-				if (*s == 0 && endch != '>') THROW_ERROR(status_bad_doctype, s);
+				s -= 2;
 
-				if (*s == '\'' || *s == '"') // '...SYSTEM "..."
-				{
-					ch = *s++;
-					SCANFOR(*s == ch);
-					if (*s == 0 && endch != '>') THROW_ERROR(status_bad_doctype, s);
+				xml_parse_result res = parse_doctype(s, buffer_start, endch, true);
 
-					s += (*s != 0);
-					goto LOC_DOCTYPE;
-				}
-
-				if(*s == '[') // '...[...'
-				{
-					++s;
-					unsigned int bd = 1; // Bracket depth counter.
-					while (*s!=0) // Loop till we're out of all brackets.
-					{
-						if (*s == ']') --bd;
-						else if (*s == '[') ++bd;
-						if (bd == 0) break;
-						++s;
-					}
-
-					if (bd != 0) THROW_ERROR(status_bad_doctype, s);
-				}
-
-				SCANFOR(*s == '>');
-
-				if (*s == 0)
-				{
-					if (endch != '>') THROW_ERROR(status_bad_doctype, s);
-				}
-				else
-				{
-					++s;
-				}
+				if (!res) return res;
 			}
 			else if (*s == 0 && endch == '-') THROW_ERROR(status_bad_comment, s);
 			else if (*s == 0 && endch == '[') THROW_ERROR(status_bad_cdata, s);
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");
+}
-- 
cgit v1.2.3