Drop using CCP4 monomers library

version bump

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(libcifpp VERSION 5.2.5 LANGUAGES CXX)
+project(libcifpp VERSION 6.0.0 LANGUAGES CXX)
 
 
 list(PREPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/cmake")
 list(PREPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/cmake")
 
 

changelog

+Version 6.0.0
+- Drop the use of CCP4's monomer library for compound information
+
 Version 5.2.5
 Version 5.2.5
 - Correctly import the Eigen3 library
 - Correctly import the Eigen3 library
 
 

include/cif++/compound.hpp

@@ -44,11 +44,7 @@
 /// The data is loaded by default from a file called `components.cif`. This file
 /// The data is loaded by default from a file called `components.cif`. This file
 /// is located using load_resource. (See documentation on cif::load_resource for more information)
 /// is located using load_resource. (See documentation on cif::load_resource for more information)
 ///
 ///
-/// But if the CCP4 environment is available at runtime, the compound information
-/// may also be generated from the CCP4 monomer library.
-///
-/// Note that the information in CCP4 and CCD is not equal.
-///
+/// Note that since version 6 the CCP4 monomer library is no longer used.
 
 
 /// See also :doc:`/compound` for more information.
 /// See also :doc:`/compound` for more information.
 
 
@@ -157,10 +153,6 @@ class compound
 	float formula_weight() const { return m_formula_weight; } ///< Return the formula mass of the chemical component in Daltons.
 	float formula_weight() const { return m_formula_weight; } ///< Return the formula mass of the chemical component in Daltons.
 	int formal_charge() const { return m_formal_charge; }     ///< Return the formal charge on the chemical component.
 	int formal_charge() const { return m_formal_charge; }     ///< Return the formal charge on the chemical component.
 
 
-	/// The group record is only available in CCP4 monomer library files.
-	/// For CCD entries this value will always contain 'non-polymer'
-	std::string group() const { return m_group; }
-
 	const std::vector<compound_atom> &atoms() const { return m_atoms; } ///< Return the list of atoms for this compound
 	const std::vector<compound_atom> &atoms() const { return m_atoms; } ///< Return the list of atoms for this compound
 	const std::vector<compound_bond> &bonds() const { return m_bonds; } ///< Return the list of bonds for this compound
 	const std::vector<compound_bond> &bonds() const { return m_bonds; } ///< Return the list of bonds for this compound
 
 
@@ -176,8 +168,6 @@ class compound
 
 
   private:
   private:
 	friend class compound_factory_impl;
 	friend class compound_factory_impl;
-	friend class CCD_compound_factory_impl;
-	friend class CCP4_compound_factory_impl;
 
 
 	compound(cif::datablock &db);
 	compound(cif::datablock &db);
 	compound(cif::datablock &db, const std::string &id, const std::string &name, const std::string &type, const std::string &group);
 	compound(cif::datablock &db, const std::string &id, const std::string &name, const std::string &type, const std::string &group);
@@ -246,6 +236,8 @@ class compound_factory
 	CIFPP_EXPORT static const std::map<std::string, char> kAAMap, ///< Globally accessible static list of the default amino acids
 	CIFPP_EXPORT static const std::map<std::string, char> kAAMap, ///< Globally accessible static list of the default amino acids
 		kBaseMap;                                                 ///< Globally accessible static list of the default bases
 		kBaseMap;                                                 ///< Globally accessible static list of the default bases
 
 
+	void report_missing_compound(const std::string &compound_id);
+
   private:
   private:
 	compound_factory();
 	compound_factory();
 
 

include/cif++/utilities.hpp

@@ -366,6 +366,14 @@ std::unique_ptr<std::istream> load_resource(std::filesystem::path name);
 void add_file_resource(const std::string &name, std::filesystem::path dataFile);
 void add_file_resource(const std::string &name, std::filesystem::path dataFile);
 
 
 /**
 /**
+ * @brief List all the file resources added with cif::add_file_resource.
+ * 
+ * @param os The std::ostream to write the directories to
+ */
+
+void list_file_resources(std::ostream &os);
+
+/**
  * @brief Add a directory to the list of search directories. This list is
  * @brief Add a directory to the list of search directories. This list is
  * searched in a last-in-first-out order.
  * searched in a last-in-first-out order.
  * 
  * 
@@ -379,4 +387,12 @@ void add_file_resource(const std::string &name, std::filesystem::path dataFile);
 
 
 void add_data_directory(std::filesystem::path dataDir);
 void add_data_directory(std::filesystem::path dataDir);
 
 
+/**
+ * @brief List all the data directories, for error reporting on missing resources.
+ * 
+ * @param os The std::ostream to write the directories to
+ */
+
+void list_data_directories(std::ostream &os);
+
 } // namespace cif
 } // namespace cif

src/parser.cpp

@@ -608,6 +608,9 @@ sac_parser::datablock_index sac_parser::index_datablocks()
 	std::string::size_type si = 0;
 	std::string::size_type si = 0;
 	std::string datablock;
 	std::string datablock;
 
 
+	// Seek to beginning of file
+	m_source.pubseekpos(0);
+
 	for (auto ch = m_source.sbumpc(); ch != std::streambuf::traits_type::eof(); ch = m_source.sbumpc())
 	for (auto ch = m_source.sbumpc(); ch != std::streambuf::traits_type::eof(); ch = m_source.sbumpc())
 	{
 	{
 		switch (state)
 		switch (state)
@@ -667,7 +670,7 @@ sac_parser::datablock_index sac_parser::index_datablocks()
 
 
 			case data_name:
 			case data_name:
 				if (is_non_blank(ch))
 				if (is_non_blank(ch))
-					datablock.insert(datablock.end(), char(ch));
+					datablock.insert(datablock.end(), (char)std::toupper(ch));
 				else if (is_space(ch))
 				else if (is_space(ch))
 				{
 				{
 					if (not datablock.empty())
 					if (not datablock.empty())

src/pdb/pdb2cif.cpp

@@ -5146,7 +5146,7 @@ void PDBFileParser::ParseConnectivtyAnnotation()
 
 
 			getCategory("struct_conn")->emplace({
 			getCategory("struct_conn")->emplace({
 				{ "id", type + std::to_string(linkNr) },
 				{ "id", type + std::to_string(linkNr) },
-			{ "conn_type_id", type },
+				{ "conn_type_id", type },
 
 
 				// { "ccp4_link_id", ccp4LinkID },
 				// { "ccp4_link_id", ccp4LinkID },
 
 

src/utilities.cpp

@@ -845,6 +845,9 @@ class resource_pool
 
 
 	std::unique_ptr<std::istream> load(fs::path name);
 	std::unique_ptr<std::istream> load(fs::path name);
 
 
+	const auto data_directories() { return mDirs; }
+	const auto file_resources() { return mLocalResources; }
+
   private:
   private:
 	resource_pool();
 	resource_pool();
 
 
@@ -937,4 +940,22 @@ std::unique_ptr<std::istream> load_resource(std::filesystem::path name)
 	return resource_pool::instance().load(name);
 	return resource_pool::instance().load(name);
 }
 }
 
 
+void list_file_resources(std::ostream &os)
+{
+	auto &file_resources = resource_pool::instance().file_resources();
+
+	if (not file_resources.empty())
+	{
+		os << "\nThe following named resources were loaded:\n";
+		for (const auto &[name, path] : file_resources)
+			os << name << " -> " << std::quoted(path.string()) << '\n';
+	}
+}
+
+void list_data_directories(std::ostream &os)
+{
+	for (auto &p : resource_pool::instance().data_directories())
+		os << p << '\n';
+}
+
 } // namespace cif
 } // namespace cif

test/unit-v2-test.cpp

@@ -3483,3 +3483,11 @@ ATOM      7  CD  PRO A   1      15.762  13.216  43.724  1.00 30.71           C)"
 
 
 	auto f = cif::pdb::read(is);
 	auto f = cif::pdb::read(is);
 }
 }
+
+// --------------------------------------------------------------------
+
+BOOST_AUTO_TEST_CASE(compound_not_found_test_1)
+{
+	auto cmp = cif::compound_factory::instance().create("&&&");
+	BOOST_CHECK(cmp == nullptr);
+}
\ No newline at end of file