indexed CCD file

include/cif++/CifParser.hpp

@@ -96,6 +96,10 @@ inline bool isUnquotedString(const char* s)
 std::tuple<std::string,std::string> splitTagName(const std::string& tag);
 
 // --------------------------------------------------------------------
+
+using DatablockIndex = std::map<std::string,std::size_t>;
+
+// --------------------------------------------------------------------
 // sac Parser, analogous to SAX Parser (simple api for xml)
 
 class SacParser
@@ -142,7 +146,10 @@ class SacParser
 	CIFToken getNextToken();
 	void match(CIFToken token);
 
-	bool parseFile(const std::string& datablock);
+	bool parseSingleDatablock(const std::string& datablock);
+
+	DatablockIndex indexDatablocks();
+	bool parseSingleDatablock(const std::string& datablock, const DatablockIndex &index);
 
 	void parseFile();
 	void parseGlobal();
@@ -199,7 +206,7 @@ class SacParser
 class Parser : public SacParser
 {
   public:
-	Parser(std::istream& is, File& f);
+	Parser(std::istream& is, File& f, bool init = true);
 
 	virtual void produceDatablock(const std::string& name);
 	virtual void produceCategory(const std::string& name);

src/Cif++.cpp

@@ -3073,7 +3073,7 @@ void File::load(std::istream& is, const std::string& datablock)
 	setValidator(nullptr);
 
 	Parser p(is, *this);
-	p.parseFile(datablock);
+	p.parseSingleDatablock(datablock);
 	
 	if (saved != nullptr)
 	{

src/CifParser.cpp

@@ -523,7 +523,104 @@ SacParser::CIFToken SacParser::getNextToken()
 	return result;
 }
 
-bool SacParser::parseFile(const std::string& datablock)
+DatablockIndex SacParser::indexDatablocks()
+{
+	DatablockIndex index;
+
+	// first locate the start, as fast as we can
+	auto &sb = *mData.rdbuf();
+
+	enum {
+		start, comment, string, string_quote, qstring, data, data_name
+	} state = start;
+
+	int quote = 0;
+	bool bol = true;
+	const char dblk[] = "data_";
+	std::string::size_type si = 0;
+	std::string datablock;
+	bool found = false;
+
+	while (sb.in_avail() > 0 and not found)
+	{
+		int ch = sb.sbumpc();
+		switch (state)
+		{
+			case start:
+				switch (ch)
+				{
+					case '#':	state = comment;	break;
+					case 'd':
+					case 'D':
+						state = data;
+						si = 1;
+						break;
+					case '\'':
+					case '"':
+						state = string;
+						quote = ch;
+						break;
+					case ';':
+						if (bol)
+							state = qstring;
+						break;
+				}
+				break;
+			
+			case comment:
+				if (ch == '\n')
+					state = start;
+				break;
+
+			case string:
+				if (ch == quote)
+					state = string_quote;
+				break;
+			
+			case string_quote:
+				if (std::isspace(ch))
+					state = start;
+				else
+					state = string;
+				break;
+			
+			case qstring:
+				if (ch == ';' and bol)
+					state = start;
+				break;
+			
+			case data:
+				if (dblk[si] == 0 and isNonBlank(ch))
+				{
+					datablock = { static_cast<char>(ch) };
+					state = data_name;
+				}
+				else if (dblk[si++] != ch)
+					state = start;
+				break;
+			
+			case data_name:
+				if (isNonBlank(ch))
+					datablock.insert(datablock.end(), ch);
+				else if (isspace(ch))
+				{
+					if (not datablock.empty())
+						index[datablock] = mData.tellg();
+					
+					state = start;
+				}
+				else
+					state = start;
+				break;
+		}
+
+		bol = (ch == '\n');
+	}
+
+	return index;
+}
+
+bool SacParser::parseSingleDatablock(const std::string& datablock)
 {
 	// first locate the start, as fast as we can
 	auto &sb = *mData.rdbuf();
@@ -607,6 +704,25 @@ bool SacParser::parseFile(const std::string& datablock)
 	return found;
 }
 
+bool SacParser::parseSingleDatablock(const std::string& datablock, const DatablockIndex &index)
+{
+	bool result = false;
+
+	auto i = index.find(datablock);
+	if (i != index.end())
+	{
+		mData.seekg(i->second);
+
+		produceDatablock(datablock);
+		mLookahead = getNextToken();
+		parseDataBlock();
+
+		result = true;
+	}
+
+	return result;
+}
+
 void SacParser::parseFile()
 {
 	while (mLookahead != eCIFTokenEOF)
@@ -728,8 +844,8 @@ void SacParser::parseSaveFrame()
 
 // --------------------------------------------------------------------
 
-Parser::Parser(std::istream& is, File& f)
-	: SacParser(is), mFile(f), mDataBlock(nullptr)
+Parser::Parser(std::istream& is, File& f, bool init)
+	: SacParser(is, init), mFile(f), mDataBlock(nullptr)
 {
 }
 

src/Compound.cpp

@@ -42,6 +42,7 @@
 #include "cif++/CifUtils.hpp"
 #include "cif++/Compound.hpp"
 #include "cif++/Point.hpp"
+#include "cif++/CifParser.hpp"
 
 namespace ba = boost::algorithm;
 namespace fs = std::filesystem;
@@ -455,6 +456,8 @@ class CCDCompoundFactoryImpl : public CompoundFactoryImpl
 	CCDCompoundFactoryImpl() {}
 
 	Compound *create(std::string id) override;
+
+	cif::DatablockIndex mIndex;
 };
 
 Compound *CCDCompoundFactoryImpl::create(std::string id)
@@ -467,14 +470,34 @@ Compound *CCDCompoundFactoryImpl::create(std::string id)
 	if (not ccd)
 		throw std::runtime_error("Could not locate the CCD components.cif file, please make sure the software is installed properly and/or use the update-dictionary-script to fetch the data.");
 
+	cif::File file;
+
+	if (mIndex.empty())
+	{
+		if (cif::VERBOSE)
+		{
+			std::cout << "Creating component index " << "...";
+			std::cout.flush();
+		}
+
+		cif::Parser parser(*ccd, file, false);
+		mIndex = parser.indexDatablocks();
+
+		if (cif::VERBOSE)
+			std::cout << " done" << std::endl;
+		
+		// reload the resource, perhaps this should be improved...
+		ccd = cif::loadResource("components.cif");
+	}
+
 	if (cif::VERBOSE)
 	{
 		std::cout << "Loading component " << id << "...";
 		std::cout.flush();
 	}
 
-	cif::File file;
-	file.load(*ccd, id);
+	cif::Parser parser(*ccd, file, false);
+	parser.parseSingleDatablock(id, mIndex);
 
 	if (cif::VERBOSE)
 		std::cout << " done" << std::endl;