Merge branch 'trunk' of github.com:PDB-REDO/libcifpp into trunk

include/cif++/model.hpp

@@ -608,7 +608,13 @@ class sugar : public residue
 	sugar(sugar &&rhs);
 	sugar &operator=(sugar &&rhs);
 
-	int num() const { return std::stoi(m_auth_seq_id); }
+	int num() const {
+		int result;
+		auto r = std::from_chars(m_auth_seq_id.data(), m_auth_seq_id.data() + m_auth_seq_id.length(), result);
+		if (r.ec != std::errc())
+			throw std::runtime_error("The auth_seq_id should be a number for a sugar");
+		return result;
+	}
 	std::string name() const;
 
 	/// \brief Return the atom the C1 is linked to

include/cif++/parser.hpp

@@ -248,7 +248,7 @@ class sac_parser
 		SAVE
 	};
 
-	std::istream &m_source;
+	std::streambuf &m_source;
 
 	// Parser state
 	bool m_validate;
@@ -257,7 +257,7 @@ class sac_parser
 	CIFToken m_lookahead;
 	std::string m_token_value;
 	CIFValue mTokenType;
-	std::stack<int> m_buffer;
+	std::string m_buffer;	// retract buffer, used to be a stack<char>
 };
 
 // --------------------------------------------------------------------

src/file.cpp

@@ -115,8 +115,8 @@ void file::load_dictionary()
 		}
 	}
 
-	if (not m_validator)
-		load_dictionary("mmcif_pdbx.dic");	// TODO: maybe incorrect? Perhaps improve?
+	// if (not m_validator)
+	// 	load_dictionary("mmcif_pdbx.dic");	// TODO: maybe incorrect? Perhaps improve?
 }
 
 void file::load_dictionary(std::string_view name)

src/parser.cpp

@@ -47,7 +47,7 @@ namespace cif
 // --------------------------------------------------------------------
 
 sac_parser::sac_parser(std::istream &is, bool init)
-	: m_source(is)
+	: m_source(*is.rdbuf())
 {
 	m_validate = true;
 	m_line_nr = 1;
@@ -62,26 +62,26 @@ sac_parser::sac_parser(std::istream &is, bool init)
 // translation.
 int sac_parser::get_next_char()
 {
-	int result;
+	int result = std::char_traits<char>::eof();
 
-	if (m_buffer.empty())
-		result = m_source.get();
-	else
+	if (not m_buffer.empty())
 	{
-		result = m_buffer.top();
-		m_buffer.pop();
+		result = m_buffer.back();
+		m_buffer.pop_back();
 	}
+	else
+		result = m_source.sbumpc();
 
 	// very simple CR/LF translation into LF
 	if (result == '\r')
 	{
-		int lookahead = m_source.get();
+		int lookahead = m_source.sbumpc();
 		if (lookahead != '\n')
-			m_buffer.push(lookahead);
+			m_buffer.push_back(lookahead);
 		result = '\n';
 	}
 
-	m_token_value += static_cast<char>(result);
+	m_token_value.push_back(static_cast<char>(result));
 
 	if (result == '\n')
 		++m_line_nr;
@@ -106,7 +106,7 @@ void sac_parser::retract()
 	if (ch == '\n')
 		--m_line_nr;
 
-	m_buffer.push(ch);
+	m_buffer.push_back(ch);
 	m_token_value.pop_back();
 }
 
@@ -453,8 +453,6 @@ void sac_parser::match(CIFToken token)
 bool sac_parser::parse_single_datablock(const std::string &datablock)
 {
 	// first locate the start, as fast as we can
-	auto &sb = *m_source.rdbuf();
-
 	enum
 	{
 		start,
@@ -471,7 +469,7 @@ bool sac_parser::parse_single_datablock(const std::string &datablock)
 	std::string::size_type si = 0;
 	bool found = false;
 
-	for (auto ch = sb.sbumpc(); not found and ch != std::streambuf::traits_type::eof(); ch = sb.sbumpc())
+	for (auto ch = m_source.sbumpc(); not found and ch != std::streambuf::traits_type::eof(); ch = m_source.sbumpc())
 	{
 		switch (state)
 		{
@@ -544,8 +542,6 @@ sac_parser::datablock_index sac_parser::index_datablocks()
 	datablock_index index;
 
 	// first locate the start, as fast as we can
-	auto &sb = *m_source.rdbuf();
-
 	enum
 	{
 		start,
@@ -563,7 +559,7 @@ sac_parser::datablock_index sac_parser::index_datablocks()
 	std::string::size_type si = 0;
 	std::string datablock;
 
-	for (auto ch = sb.sbumpc(); ch != std::streambuf::traits_type::eof(); ch = sb.sbumpc())
+	for (auto ch = m_source.sbumpc(); ch != std::streambuf::traits_type::eof(); ch = m_source.sbumpc())
 	{
 		switch (state)
 		{
@@ -626,7 +622,7 @@ sac_parser::datablock_index sac_parser::index_datablocks()
 				else if (isspace(ch))
 				{
 					if (not datablock.empty())
-						index[datablock] = m_source.tellg();
+						index[datablock] = m_source.pubseekoff(0, std::ios_base::cur, std::ios_base::in);
 
 					state = start;
 				}
@@ -648,7 +644,7 @@ bool sac_parser::parse_single_datablock(const std::string &datablock, const data
 	auto i = index.find(datablock);
 	if (i != index.end())
 	{
-		m_source.seekg(i->second);
+		m_source.pubseekpos(i->second, std::ios_base::in);
 
 		produce_datablock(datablock);
 		m_lookahead = get_next_token();

src/pdb/cif2pdb.cpp

@@ -704,16 +704,17 @@ class Ff : public FBase
 		else
 		{
 			std::string s{ text() };
-			try
-			{
-				os << std::stod(s);
-			}
-			catch (const std::exception &ex)
+
+			double d = 0;
+			auto r = cif::from_chars(s.data(), s.data() + s.length(), d);
+			if (r.ec != std::errc())
 			{
-				if (VERBOSE >= 0)
+				if (VERBOSE > 0)
 					std::cerr << "Failed to write '" << s << "' as a double, this indicates an error in the code for writing PDB files" << std::endl;
 				os << s;
 			}
+			else
+				os << d;
 		}
 	}
 };

src/pdb/pdb2cif.cpp

@@ -1136,16 +1136,15 @@ void PDBFileParser::PreParseInput(std::istream &is)
 	{
 		std::string cs = lookahead.substr(offset, len);
 		cif::trim(cs);
-		int result;
+		int result = 0;
 
-		try
-		{
-			result = cs.empty() ? 0 : stoi(cs);
-		}
-		catch (...)
+		if (not cs.empty())
 		{
-			throw std::runtime_error("Continuation std::string '" + cs + "' is not valid");
+			auto r = std::from_chars(cs.data(), cs.data() + cs.length(), result);
+			if (r.ec != std::errc())
+				throw std::runtime_error("Continuation std::string '" + cs + "' is not valid");
 		}
+
 		return result;
 	};
 
@@ -1402,7 +1401,12 @@ void PDBFileParser::PreParseInput(std::istream &is)
 			link.symOpB = cur->vS(67, 72);    //	67 - 72         SymOP          sym2            Symmetry operator atom 2.
 
 			if (type == "LINK") //	 1 -  6         Record name    "LINK  "
-				link.distance = std::stof(cur->vF(74, 78));
+			{
+				auto f = cur->vF(74, 78);
+				auto r = cif::from_chars(f.data(), f.data() + f.length(), link.distance);
+				if (r.ec != std::errc() and cif::VERBOSE > 0)
+					std::cerr << "Error parsing link distance at line " << cur->mLineNr << std::endl;
+			}
 			//	74 – 78         Real(5.2)      Length          Link distance
 
 			mLinks.push_back(link);
@@ -5107,11 +5111,10 @@ void PDBFileParser::ParseConnectivtyAnnotation()
 			if (mRec->is("LINK  "))
 			{
 				distance = vS(74, 78);
-				try
-				{
-					stod(distance);
-				}
-				catch (const std::invalid_argument &)
+
+				double d;
+				auto r = cif::from_chars(distance.data(), distance.data() + distance.length(), d);
+				if (r.ec != std::errc())
 				{
 					if (cif::VERBOSE > 0)
 						std::cerr << "Distance value '" << distance << "' is not a valid float in LINK record" << std::endl;
@@ -6211,6 +6214,10 @@ file read(std::istream &is)
 			result.load(is);
 	}
 
+	// Must be a PDB like file, right?
+	if (result.get_validator() == nullptr)
+		result.load_dictionary("mmcif_pdbx.dic");
+
 	return result;
 }
 

src/pdb/pdb2cif_remark_3.cpp

@@ -1207,13 +1207,15 @@ void Remark3Parser::storeCapture(const char *category, std::initializer_list<con
 			}
 			else if (iequals(category, "pdbx_refine_tls_group"))
 			{
-				std::string tlsGroupID;
+				std::string tlsID;
 				if (not mDb["pdbx_refine_tls"].empty())
-					tlsGroupID = mDb["pdbx_refine_tls"].back()["id"].as<std::string>();
+					tlsID = mDb["pdbx_refine_tls"].back()["id"].as<std::string>();
+				std::string tlsGroupID = cat.get_unique_id("");
 
-				cat.emplace({ { "pdbx_refine_id", mExpMethod },
+				cat.emplace({
+					{ "pdbx_refine_id", mExpMethod },
 					{ "id", tlsGroupID },
-					{ "refine_tls_id", tlsGroupID } });
+					{ "refine_tls_id", tlsID } });
 			}
 			else if (iequals(category, "pdbx_refine_tls"))
 			{
@@ -1425,10 +1427,9 @@ bool Remark3Parser::parse(const std::string &expMethod, PDBRecord *r, cif::datab
 	sort(scores.begin(), scores.end());
 
 	bool guessProgram = scores.empty() or scores.front().score < 0.9f;
-	;
 	if (guessProgram)
 	{
-		if (cif::VERBOSE >= 0)
+		if (cif::VERBOSE > 0)
 			std::cerr << "Unknown or untrusted program in REMARK 3, trying all parsers to see if there is a match" << std::endl;
 
 		tryParser(new BUSTER_TNT_Remark3Parser("BUSTER-TNT", expMethod, r, db));