Support for cifv1.0 (empty category names)

changelog

 Version 5.0.1
 - Fix loading dictionaries
+- Support for cifv1.0 files
 
 Version 5.0.0
 - Total rewrite of cif part

include/cif++/category.hpp

@@ -86,7 +86,7 @@ class category
 
 	const std::string &name() const { return m_name; }
 
-	iset fields() const;
+	iset key_fields() const;
 
 	std::set<uint16_t> key_field_indices() const;
 
@@ -482,6 +482,8 @@ class category
 		return get_column_ix(name) < m_columns.size();
 	}
 
+	iset get_columns() const;
+
 	// --------------------------------------------------------------------
 
 	void sort(std::function<int(row_handle,row_handle)> f);

src/category.cpp

@@ -410,7 +410,7 @@ category_index::entry *category_index::insert(entry *h, row *v)
 		row_handle rh(m_category, *v);
 
 		std::ostringstream os;
-		for (auto col : m_category.fields())
+		for (auto col : m_category.key_fields())
 		{
 			if (rh[col])
 				os << col << ": " << std::quoted(rh[col].text()) << "; ";
@@ -686,7 +686,17 @@ category::~category()
 
 // --------------------------------------------------------------------
 
-iset category::fields() const
+iset category::get_columns() const
+{
+	iset result;
+
+	for (auto &col : m_columns)
+		result.insert(col.m_name);
+
+	return result;
+}
+
+iset category::key_fields() const
 {
 	if (m_validator == nullptr)
 		throw std::runtime_error("No Validator specified");
@@ -1853,7 +1863,10 @@ void category::write(std::ostream &os, const std::vector<uint16_t> &order, bool
 		for (auto cix : order)
 		{
 			auto &col = m_columns[cix];
-			os << '_' << m_name << '.' << col.m_name << ' ' << '\n';
+			os << '_';
+			if (not m_name.empty())
+				os << m_name << '.';
+			os << col.m_name << ' ' << '\n';
 			columnWidths[cix] = 2;
 		}
 
@@ -1941,7 +1954,10 @@ void category::write(std::ostream &os, const std::vector<uint16_t> &order, bool
 		{
 			auto &col = m_columns[cix];
 
-			os << '_' << m_name << '.' << col.m_name << std::string(l - col.m_name.length() - m_name.length() - 2, ' ');
+			os << '_';
+			if (not m_name.empty())
+				os << m_name << '.';
+			os << col.m_name << std::string(l - col.m_name.length() - m_name.length() - 2, ' ');
 
 			std::string_view s;
 			auto iv = m_head->get(cix);
@@ -1978,29 +1994,44 @@ bool category::operator==(const category &rhs) const
 //	if (tagsA != tagsB)
 //		std::cout << "Unequal number of fields" << std::endl;
 
+	const category_validator *catValidator = nullptr;
+
 	auto validator = a.get_validator();
-	auto catValidator = validator->get_validator_for_category(a.name());
-	if (catValidator == nullptr)
-		throw std::runtime_error("missing cat validator");
+	if (validator != nullptr)
+		catValidator = validator->get_validator_for_category(a.name());
 	
 	typedef std::function<int(std::string_view,std::string_view)> compType;
 	std::vector<std::tuple<std::string,compType>> tags;
-	auto keys = catValidator->m_keys;
+	std::vector<std::string> keys;
 	std::vector<size_t> keyIx;
 	
-	for (auto& tag: a.fields())
+	if (catValidator == nullptr)
 	{
-		auto iv = catValidator->get_validator_for_item(tag);
-		if (iv == nullptr)
-			throw std::runtime_error("missing item validator");
-		auto tv = iv->m_type;
-		if (tv == nullptr)
-			throw std::runtime_error("missing type validator");
-		tags.push_back(std::make_tuple(tag, std::bind(&cif::type_validator::compare, tv, std::placeholders::_1, std::placeholders::_2)));
-		
-		auto pred = [tag](const std::string& s) -> bool { return cif::iequals(tag, s) == 0; };
-		if (find_if(keys.begin(), keys.end(), pred) == keys.end())
-			keyIx.push_back(tags.size() - 1);
+		for (auto& tag: a.get_columns())
+		{
+			tags.push_back(std::make_tuple(tag, [](std::string_view va, std::string_view vb) { return va.compare(vb); }));
+			keyIx.push_back(keys.size());
+			keys.push_back(tag);
+		}
+	}
+	else
+	{
+		keys = catValidator->m_keys;
+
+		for (auto& tag: a.key_fields())
+		{
+			auto iv = catValidator->get_validator_for_item(tag);
+			if (iv == nullptr)
+				throw std::runtime_error("missing item validator");
+			auto tv = iv->m_type;
+			if (tv == nullptr)
+				throw std::runtime_error("missing type validator");
+			tags.push_back(std::make_tuple(tag, std::bind(&cif::type_validator::compare, tv, std::placeholders::_1, std::placeholders::_2)));
+			
+			auto pred = [tag](const std::string& s) -> bool { return cif::iequals(tag, s) == 0; };
+			if (find_if(keys.begin(), keys.end(), pred) == keys.end())
+				keyIx.push_back(tags.size() - 1);
+		}
 	}
 	
 	// a.reorderByIndex();

src/condition.cpp

@@ -32,7 +32,7 @@ namespace cif
 
 iset get_category_fields(const category &cat)
 {
-	return cat.fields();
+	return cat.key_fields();
 }
 
 uint16_t get_column_ix(const category &cat, std::string_view col)

src/parser.cpp

@@ -695,7 +695,8 @@ void sac_parser::parse_global()
 
 void sac_parser::parse_datablock()
 {
-	std::string cat;
+	static const std::string kUnitializedCategory("<invalid>");
+	std::string cat = kUnitializedCategory;	// intial value acts as a guard for empty category names
 
 	while (m_lookahead == CIFToken::LOOP or m_lookahead == CIFToken::Tag or m_lookahead == CIFToken::SAVE)
 	{
@@ -703,7 +704,7 @@ void sac_parser::parse_datablock()
 		{
 			case CIFToken::LOOP:
 			{
-				cat.clear(); // should start a new category
+				cat = kUnitializedCategory; // should start a new category
 
 				match(CIFToken::LOOP);
 
@@ -714,7 +715,7 @@ void sac_parser::parse_datablock()
 					std::string catName, itemName;
 					std::tie(catName, itemName) = split_tag_name(m_token_value);
 
-					if (cat.empty())
+					if (cat == kUnitializedCategory)
 					{
 						produce_category(catName);
 						cat = catName;
@@ -800,6 +801,9 @@ void parser::produce_row()
 	if (VERBOSE >= 4)
 		std::cerr << "producing row for category " << m_category->name() << std::endl;
 
+	if (m_category == nullptr)
+		error("inconsistent categories in loop_");
+
 	m_category->emplace({});
 	m_row = m_category->back();
 	// m_row.lineNr(m_line_nr);
@@ -810,7 +814,7 @@ void parser::produce_item(const std::string &category, const std::string &item,
 	if (VERBOSE >= 4)
 		std::cerr << "producing _" << category << '.' << item << " -> " << value << std::endl;
 
-	if (not iequals(category, m_category->name()))
+	if (m_category == nullptr or not iequals(category, m_category->name()))
 		error("inconsistent categories in loop_");
 
 	m_row[item] = m_token_value;

src/pdb/pdb2cif_remark_3.cpp

@@ -1483,7 +1483,7 @@ bool Remark3Parser::parse(const std::string &expMethod, PDBRecord *r, cif::datab
 					auto r1 = cat1.front();
 					auto r2 = cat2.front();
 
-					for (auto column : cat1.fields())
+					for (auto column : cat1.key_fields())
 						r2[column] = r1[column].text();
 				}
 			}

src/text.cpp

@@ -224,9 +224,10 @@ std::tuple<std::string, std::string> split_tag_name(std::string_view tag)
 
 	auto s = tag.find('.');
 	if (s == std::string::npos)
-		throw std::runtime_error("tag does not contain dot (" + std::string{ tag } + ')');
-	return std::tuple<std::string, std::string>{
-		tag.substr(1, s - 1), tag.substr(s + 1)};
+		// throw std::runtime_error("tag does not contain dot (" + std::string{ tag } + ')');
+		return std::tuple<std::string, std::string>{ "", tag.substr(1) };
+	else
+		return std::tuple<std::string, std::string>{tag.substr(1, s - 1), tag.substr(s + 1)};
 }
 
 // --------------------------------------------------------------------

test/unit-v2-test.cpp

@@ -2970,3 +2970,91 @@ _cat_1.id_2
 	for (const auto &[key, test] : TESTS)
 		BOOST_CHECK_EQUAL((bool)cat1[key], test);
 }
+
+// --------------------------------------------------------------------
+
+BOOST_AUTO_TEST_CASE(cifv1_0_1)
+{
+	auto f = R"(data_TEST
+#
+loop_
+_id
+_name
+1 aap
+2 noot
+3 mies
+4 ?
+5 .
+    )"_cf;
+
+	auto &db = f.front();
+
+	auto &cat = db[""];
+
+	for (auto r : cat)
+	{
+		int id;
+		std::optional<std::string> name;
+
+		cif::tie(id, name) = r.get("id", "name");
+
+		switch (id)
+		{
+			case 1: BOOST_CHECK_EQUAL(*name, "aap"); break;
+			case 2: BOOST_CHECK_EQUAL(*name, "noot"); break;
+			case 3: BOOST_CHECK_EQUAL(*name, "mies"); break;
+			default: BOOST_CHECK(name.has_value() == false);
+		}
+	}
+
+	std::stringstream ss;
+	ss << db;
+
+	auto f2 = cif::file(ss);
+	auto &db2 = f2.front();
+
+	BOOST_TEST(db == db2);
+}
+
+// BOOST_AUTO_TEST_CASE(cifv1_0_2)
+// {
+// 	BOOST_CHECK_THROW(R"(data_TEST
+// #
+// _version 1.0
+// loop_
+// _id
+// _name
+// 1 aap
+// 2 noot
+// 3 mies
+// 4 ?
+// 5 .
+//     )"_cf, cif::parse_error);
+// }
+
+BOOST_AUTO_TEST_CASE(cifv1_0_3)
+{
+	auto f = R"(data_TEST
+#
+_version 1.0
+_date    today
+)"_cf;
+
+	auto &db = f.front();
+
+	auto &cat = db[""];
+	BOOST_CHECK(not cat.empty());
+
+	auto r = cat.front();
+	BOOST_CHECK_EQUAL(r["version"].as<std::string>(), "1.0");
+	BOOST_CHECK_EQUAL(r["date"].as<std::string>(), "today");
+
+	std::stringstream ss;
+	ss << db;
+
+	auto f2 = cif::file(ss);
+	auto &db2 = f2.front();
+
+	BOOST_TEST(db == db2);
+}
+