First steps, remove CCP4 info from compound

include/cif++/Cif++.hpp

@@ -289,9 +289,10 @@ namespace detail
 		void swap(ItemReference& b);
 		
 		template<typename T = std::string>
-		T as() const
+		auto as() const
 		{
-			return item_value_as<T>::convert(*this);
+			using value_type = std::remove_cv_t<std::remove_reference_t<T>>;
+			return item_value_as<value_type>::convert(*this);
 		}
 
 		template<typename T>
@@ -337,11 +338,13 @@ namespace detail
 	template<typename T>
 	struct ItemReference::item_value_as<T, std::enable_if_t<std::is_floating_point_v<T>>>
 	{
-		static T convert(const ItemReference& ref)
+		using value_type = std::remove_reference_t<std::remove_cv_t<T>>;
+
+		static value_type convert(const ItemReference& ref)
 		{
-			T result = {};
+			value_type result = {};
 			if (not ref.empty())
-				result = static_cast<T>(std::stod(ref.c_str()));
+				result = static_cast<value_type>(std::stod(ref.c_str()));
 			return result;
 		}
 
@@ -376,7 +379,7 @@ namespace detail
 	};
 
 	template<typename T>
-	struct ItemReference::item_value_as<T, std::enable_if_t<std::is_integral_v<T> and std::is_unsigned_v<T>>>
+	struct ItemReference::item_value_as<T, std::enable_if_t<std::is_integral_v<T> and std::is_unsigned_v<T> and not std::is_same_v<T, bool>>>
 	{
 		static T convert(const ItemReference& ref)
 		{
@@ -417,7 +420,7 @@ namespace detail
 	};
 
 	template<typename T>
-	struct ItemReference::item_value_as<T, std::enable_if_t<std::is_integral_v<T> and std::is_signed_v<T>>>
+	struct ItemReference::item_value_as<T, std::enable_if_t<std::is_integral_v<T> and std::is_signed_v<T> and not std::is_same_v<T, bool>>>
 	{
 		static T convert(const ItemReference& ref)
 		{
@@ -482,6 +485,25 @@ namespace detail
 		}
 	};
 
+	template<typename T>
+	struct ItemReference::item_value_as<T, std::enable_if_t<std::is_same_v<T, bool>>>
+	{
+		static bool convert(const ItemReference& ref)
+		{
+			bool result = false;
+			if (not ref.empty())
+				result = iequals(ref.c_str(), "y");
+			return result;
+		}
+
+		static int compare(const ItemReference& ref, bool value, bool icase)
+		{
+			bool rv = convert(ref);
+			return value && rv ? 0
+				: (rv < value ? -1 : 1);
+		}
+	};
+
 	template<size_t N>
 	struct ItemReference::item_value_as<char[N]>
 	{

include/cif++/Structure.hpp

@@ -141,9 +141,6 @@ class Atom
 	
 	// access data in compound for this atom
 
-	// the energy-type field
-	std::string energyType() const;
-	
 	// convenience routine
 	bool isBackBone() const
 	{
@@ -243,10 +240,6 @@ class Residue
 	
 	bool isWater() const					{ return mCompoundID == "HOH"; }
 
-	bool isSugar() const;
-	bool isPyranose() const;
-	bool isFuranose() const;
-	
 	const Structure& structure() const		{ return *mStructure; }
 
 	bool empty() const						{ return mStructure == nullptr; }

src/PDB2Cif.cpp

@@ -457,7 +457,6 @@ class PDBFileParser
 		char					iCode;
 		int						numHetAtoms = 0;
 		std::string				text;
-		bool					sugar = false;
 		std::string				asymID;
 		std::vector<PDBRecord*>	atoms;
 		bool					processed = false;
@@ -467,14 +466,6 @@ class PDBFileParser
 		HET(const std::string& hetID, char chainID, int seqNum, char iCode, int numHetAtoms = 0, const std::string& text = {})
 			: hetID(hetID), chainID(chainID), seqNum(seqNum), iCode(iCode), numHetAtoms(numHetAtoms), text(text)
 		{
-			// just in case we don't have a CCP4 available
-			if (hetID == "MAN" or hetID == "BMA" or hetID == "NAG" or hetID == "NDG" or hetID == "FUC" or hetID == "FUL")
-				sugar = true;
-			else
-			{
-				auto compound = CompoundFactory::instance().create(hetID);
-				sugar = compound ? compound->isSugar() : false;
-			}
 		}
 
 	};
@@ -4374,33 +4365,38 @@ void PDBFileParser::ConstructEntities()
 			mMod2parent.count(cc) ? mMod2parent[cc] : cc
 		);
 		
-		std::string formula = mFormuls[cc];
-		if (formula.empty() and compound != nullptr)
-			formula = compound->formula();
-		else
-		{
-			const std::regex rx(R"(\d+\((.+)\))");
-			std::smatch m;
-			if (std::regex_match(formula, m, rx))
-				formula = m[1].str();
-		}
-		
-		std::string name = mHetnams[cc];
-		if (name.empty() and compound != nullptr)
-			name = compound->name();
-
-		std::string type = "other";
+		std::string name;
+		std::string formula;
+		std::string type;
 		std::string nstd = ".";
+		std::string formulaWeight;
 		
 		if (compound != nullptr)
 		{
+			name = compound->name();
 			type = compound->type();
 			
-			if (type.empty())
-				type = "NON-POLYMER";
-
 			if (iequals(type, "L-peptide linking") or iequals(type, "peptide linking"))
 				nstd = "y";
+			
+			formula = compound->formula();
+			formulaWeight = std::to_string(compound->formulaWeight());
+		}
+
+		if (name.empty())
+			name = mHetnams[cc];
+
+		if (type.empty())
+			type = "NON-POLYMER";
+
+		if (formula.empty())
+		{
+			formula = mFormuls[cc];
+
+			const std::regex rx(R"(\d+\((.+)\))");
+			std::smatch m;
+			if (std::regex_match(formula, m, rx))
+				formula = m[1].str();
 		}
 
 		if (modResSet.count(cc))
@@ -4410,6 +4406,7 @@ void PDBFileParser::ConstructEntities()
 			{ "id", cc },
 			{ "name", name },
 			{ "formula", formula },
+			{ "formula_weight", formulaWeight },
 			{ "mon_nstd_flag", nstd },
 			{ "type", type }
 		});

src/Structure.cpp

@@ -566,16 +566,6 @@ int Atom::charge() const
 	return property<int>("pdbx_formal_charge");
 }
 
-std::string Atom::energyType() const
-{
-	std::string result;
-	
-	if (impl() and impl()->mCompound)
-		result = impl()->mCompound->getAtomByID(impl()->mAtomID).typeEnergy;
-	
-	return result;
-}
-
 float Atom::uIso() const
 {
 	float result;
@@ -1013,21 +1003,6 @@ bool Residue::isEntity() const
 	return a1.size() == a2.size();
 }
 
-bool Residue::isSugar() const
-{
-	return compound().isSugar();
-}
-
-bool Residue::isPyranose() const
-{
-	return cif::iequals(compound().group(), "pyranose");
-}
-
-bool Residue::isFuranose() const
-{
-	return cif::iequals(compound().group(), "furanose");
-}
-
 std::string Residue::authID() const
 {
 	std::string result;
@@ -2101,7 +2076,7 @@ void Structure::insertCompound(const std::string& compoundID, bool isEntity)
 {
 	auto compound = Compound::create(compoundID);
 	if (compound == nullptr)
-		throw std::runtime_error("Trying to insert unknown compound " + compoundID + " (not found in CCP4 monomers lib)");
+		throw std::runtime_error("Trying to insert unknown compound " + compoundID + " (not found in CCD)");
 
 	cif::Datablock& db = *mFile.impl().mDb;
 	
@@ -2113,6 +2088,7 @@ void Structure::insertCompound(const std::string& compoundID, bool isEntity)
 			{ "id", compoundID },
 			{ "name", compound->name() },
 			{ "formula", compound->formula() },
+			{ "formula_weight", compound->formulaWeight() },
 			{ "type", compound->type() }
 		});
 	}