improved parser. is_non_quoted string

CMakeLists.txt

@@ -25,7 +25,7 @@
 cmake_minimum_required(VERSION 3.16)
 cmake_minimum_required(VERSION 3.16)
 
 
 # set the project name
 # set the project name
-project(cifpp VERSION 5.0.5 LANGUAGES CXX)
+project(cifpp VERSION 5.0.6 LANGUAGES CXX)
 
 
 list(PREPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/cmake")
 list(PREPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/cmake")
 
 

changelog

+Version 5.0.6
+- Fix file::contains, using iequals
+
 Version 5.0.5
 Version 5.0.5
 - Fix code to work on 32 bit machines
 - Fix code to work on 32 bit machines
 
 

include/cif++/parser.hpp

@@ -27,6 +27,7 @@
 #pragma once
 #pragma once
 
 
 #include <map>
 #include <map>
+#include <regex>
 
 
 #include <cif++/row.hpp>
 #include <cif++/row.hpp>
 
 
@@ -93,8 +94,6 @@ class sac_parser
 
 
 	static bool is_unquoted_string(std::string_view text)
 	static bool is_unquoted_string(std::string_view text)
 	{
 	{
-		auto s = text.begin();
-
 		bool result = true;
 		bool result = true;
 		for (auto ch : text)
 		for (auto ch : text)
 		{
 		{
@@ -104,14 +103,10 @@ class sac_parser
 			break;
 			break;
 		}
 		}
 
 
-		// but be careful it does not contain e.g. stop_
-		if (result)
-		{
-			static const std::regex reservedRx(R"((^(?:data|save)|.*(?:loop|stop|global))_.+)", std::regex_constants::icase);
-			result = not std::regex_match(text.begin(), text.end(), reservedRx);
-		}
+		static const std::regex kReservedRx(R"(loop_|stop_|global_|data_\S+|save_\S+)", std::regex_constants::icase);
 
 
-		return result;
+		// but be careful it does not contain e.g. stop_
+		return result and not std::regex_match(text.begin(), text.end(), kReservedRx);
 	}
 	}
 
 
   protected:
   protected:

src/file.cpp

@@ -125,7 +125,7 @@ void file::load_dictionary(std::string_view name)
 
 
 bool file::contains(std::string_view name) const
 bool file::contains(std::string_view name) const
 {
 {
-	return std::find_if(begin(), end(), [name](const datablock &db) { return db.name() == name; }) != end();
+	return std::find_if(begin(), end(), [name](const datablock &db) { return iequals(db.name(), name); }) != end();
 }
 }
 
 
 datablock &file::operator[](std::string_view name)
 datablock &file::operator[](std::string_view name)

src/parser.cpp

@@ -380,18 +380,13 @@ sac_parser::CIFToken sac_parser::get_next_token()
 				{
 				{
 					std::string s = to_lower_copy(m_token_value);
 					std::string s = to_lower_copy(m_token_value);
 
 
-					if (s == "global_")
-						result = CIFToken::GLOBAL;
-					else if (s == "stop_")
-						result = CIFToken::STOP;
-					else if (s == "loop_")
-						result = CIFToken::LOOP;
-					else if (s == "data_")
+					if (s == "data_")
 					{
 					{
 						state = State::DATA;
 						state = State::DATA;
 						continue;
 						continue;
 					}
 					}
-					else if (s == "save_")
+					
+					if (s == "save_")
 					{
 					{
 						state = State::SAVE;
 						state = State::SAVE;
 						continue;
 						continue;
@@ -405,6 +400,12 @@ sac_parser::CIFToken sac_parser::get_next_token()
 
 
 					if (m_token_value == ".")
 					if (m_token_value == ".")
 						mTokenType = CIFValue::Inapplicable;
 						mTokenType = CIFValue::Inapplicable;
+					else if (iequals(m_token_value, "global_"))
+						result = CIFToken::GLOBAL;
+					else if (iequals(m_token_value, "stop_"))
+						result = CIFToken::STOP;
+					else if (iequals(m_token_value, "loop_"))
+						result = CIFToken::LOOP;
 					else if (m_token_value == "?")
 					else if (m_token_value == "?")
 					{
 					{
 						mTokenType = CIFValue::Unknown;
 						mTokenType = CIFValue::Unknown;
@@ -786,6 +787,9 @@ void sac_parser::parse_save_frame()
 
 
 void parser::produce_datablock(const std::string &name)
 void parser::produce_datablock(const std::string &name)
 {
 {
+	if (VERBOSE >= 4)
+		std::cerr << "producing data_" << name << std::endl;
+
 	const auto &[iter, ignore] = m_file.emplace(name);
 	const auto &[iter, ignore] = m_file.emplace(name);
 	m_datablock = &(*iter);
 	m_datablock = &(*iter);
 }
 }
@@ -801,7 +805,7 @@ void parser::produce_category(const std::string &name)
 
 
 void parser::produce_row()
 void parser::produce_row()
 {
 {
-	if (VERBOSE >= 4)
+	if (VERBOSE >= 4 and m_category != nullptr)
 		std::cerr << "producing row for category " << m_category->name() << std::endl;
 		std::cerr << "producing row for category " << m_category->name() << std::endl;
 
 
 	if (m_category == nullptr)
 	if (m_category == nullptr)

test/unit-v2-test.cpp

@@ -2322,15 +2322,21 @@ _test.text ??
 
 
 BOOST_AUTO_TEST_CASE(output_test_1)
 BOOST_AUTO_TEST_CASE(output_test_1)
 {
 {
+	cif::VERBOSE = 5;
+
 	auto data1 = R"(
 	auto data1 = R"(
 data_Q
 data_Q
 loop_
 loop_
 _test.text
 _test.text
-"stop_the_crap"
+stop_the_crap
 'and stop_ this too'
 'and stop_ this too'
 'data_dinges'
 'data_dinges'
-'blablaglobal_bla'
+blablaglobal_bla
 boo.data_.whatever
 boo.data_.whatever
+'data_.whatever'
+'stop_'
+'loop_'
+'global_'
 )"_cf;
 )"_cf;
 
 
 	auto &db1 = data1.front();
 	auto &db1 = data1.front();
@@ -2341,11 +2347,15 @@ boo.data_.whatever
 		const char *s;
 		const char *s;
 		bool q;
 		bool q;
 	} kS[] = {
 	} kS[] = {
-		{ "stop_the_crap", false },
+		{ "stop_the_crap", true },
 		{ "and stop_ this too", false },
 		{ "and stop_ this too", false },
 		{ "data_dinges", false },
 		{ "data_dinges", false },
-		{ "blablaglobal_bla", false },
-		{ "boo.data_.whatever", true }
+		{ "blablaglobal_bla", true },
+		{ "boo.data_.whatever", true },
+		{ "data_.whatever", false },
+		{ "stop_", false },
+		{ "loop_", false },
+		{ "global_", false }
 	};
 	};
 
 
 	BOOST_CHECK_EQUAL(test1.size(), sizeof(kS) / sizeof(T));
 	BOOST_CHECK_EQUAL(test1.size(), sizeof(kS) / sizeof(T));