Merge remote-tracking branch 'github/develop-cif2fasta' into develop

include/cif++/category.hpp

@@ -70,6 +70,26 @@ class duplicate_key_error : public std::runtime_error
 	}
 };
 
+/// @brief A missing_key_error is thrown when an attempt is made
+/// to create an index when one of the key fields is missing.
+class missing_key_error : public std::runtime_error
+{
+  public:
+    /**
+     * @brief Construct a new duplicate key error object
+     */
+	missing_key_error(const std::string &msg, const std::string &key)
+		: std::runtime_error(msg)
+		, m_key(key)
+	{
+	}
+
+	const std::string &get_key() const noexcept { return m_key; }
+
+  private:
+	std::string m_key;
+};
+
 /// @brief A multiple_results_error is throw when you request a single
 /// row using a query but the query contains more than exactly one row.
 class multiple_results_error : public std::runtime_error
@@ -516,7 +536,7 @@ class category
 	/// @param column The name of the column to return the value for
 	/// @return The value found
 	template <typename T>
-	T find1(condition &&cond, const char *column) const
+	T find1(condition &&cond, std::string_view column) const
 	{
 		return find1<T>(cbegin(), std::move(cond), column);
 	}
@@ -530,7 +550,7 @@ class category
 	/// @param column The name of the column to return the value for
 	/// @return The value found
 	template <typename T, std::enable_if_t<not is_optional_v<T>, int> = 0>
-	T find1(const_iterator pos, condition &&cond, const char *column) const
+	T find1(const_iterator pos, condition &&cond, std::string_view column) const
 	{
 		auto h = find<T>(pos, std::move(cond), column);
 
@@ -549,7 +569,7 @@ class category
 	/// @param column The name of the column to return the value for
 	/// @return The value found, can be empty if no row matches the condition
 	template <typename T, std::enable_if_t<is_optional_v<T>, int> = 0>
-	T find1(const_iterator pos, condition &&cond, const char *column) const
+	T find1(const_iterator pos, condition &&cond, std::string_view column) const
 	{
 		auto h = find<typename T::value_type>(pos, std::move(cond), column);
 
@@ -644,7 +664,7 @@ class category
 	/// @param column The column for which the value should be returned
 	/// @return The value found or a default constructed value if not found
 	template <typename T>
-	T find_first(condition &&cond, const char *column) const
+	T find_first(condition &&cond, std::string_view column) const
 	{
 		return find_first<T>(cbegin(), std::move(cond), column);
 	}
@@ -657,7 +677,7 @@ class category
 	/// @param column The column for which the value should be returned
 	/// @return The value found or a default constructed value if not found
 	template <typename T>
-	T find_first(const_iterator pos, condition &&cond, const char *column) const
+	T find_first(const_iterator pos, condition &&cond, std::string_view column) const
 	{
 		auto h = find<T>(pos, std::move(cond), column);
 
@@ -701,7 +721,7 @@ class category
 	/// @param cond The condition to search for
 	/// @return The value found or the minimal value for the type
 	template <typename T, std::enable_if_t<std::is_arithmetic_v<T>, int> = 0>
-	T find_max(const char *column, condition &&cond) const
+	T find_max(std::string_view column, condition &&cond) const
 	{
 		T result = std::numeric_limits<T>::min();
 
@@ -719,7 +739,7 @@ class category
 	/// @param column The column to use for the value
 	/// @return The value found or the minimal value for the type
 	template <typename T, std::enable_if_t<std::is_arithmetic_v<T>, int> = 0>
-	T find_max(const char *column) const
+	T find_max(std::string_view column) const
 	{
 		return find_max<T>(column, all());
 	}
@@ -730,7 +750,7 @@ class category
 	/// @param cond The condition to search for
 	/// @return The value found or the maximum value for the type
 	template <typename T, std::enable_if_t<std::is_arithmetic_v<T>, int> = 0>
-	T find_min(const char *column, condition &&cond) const
+	T find_min(std::string_view column, condition &&cond) const
 	{
 		T result = std::numeric_limits<T>::max();
 
@@ -748,7 +768,7 @@ class category
 	/// @param column The column to use for the value
 	/// @return The value found or the maximum value for the type
 	template <typename T, std::enable_if_t<std::is_arithmetic_v<T>, int> = 0>
-	T find_min(const char *column) const
+	T find_min(std::string_view column) const
 	{
 		return find_min<T>(column, all());
 	}
@@ -756,8 +776,17 @@ class category
 	/// @brief Return whether a row exists that matches condition @a cond
 	/// @param cond The condition to match
 	/// @return True if a row exists
+	[[deprecated("Use contains instead")]]
 	bool exists(condition &&cond) const
 	{
+		return contains(std::move(cond));
+	}
+
+	/// @brief Return whether a row exists that matches condition @a cond
+	/// @param cond The condition to match
+	/// @return True if a row exists
+	bool contains(condition &&cond) const
+	{
 		bool result = false;
 
 		if (cond)
@@ -922,6 +951,11 @@ class category
 			{ return prefix + std::to_string(nr + 1); });
 	}
 
+	/// @brief Generate a new, unique value for a item named @a tag
+	/// @param tag The name of the item
+	/// @return a new unique value
+	std::string get_unique_value(std::string_view tag);
+
 	// --------------------------------------------------------------------
 
 	/// \brief Update a single column named @a tag in the rows that match \a cond to value \a value

include/cif++/condition.hpp

@@ -934,6 +934,16 @@ struct key
 	{
 	}
 
+	/**
+	 * @brief Construct a new key object using @a itemTag as name
+	 * 
+	 * @param itemTag 
+	 */
+	explicit key(std::string_view itemTag)
+		: m_item_tag(itemTag)
+	{
+	}
+
 	key(const key &) = delete;
 	key &operator=(const key &) = delete;
 

include/cif++/item.hpp

@@ -556,7 +556,7 @@ struct item_handle::item_value_as<T, std::enable_if_t<std::is_arithmetic_v<T> an
 
 		auto txt = ref.text();
 
-		if (txt.empty())
+		if (ref.empty())
 			result = 1;
 		else
 		{

include/cif++/iterator.hpp

@@ -562,22 +562,23 @@ class conditional_iterator_proxy
 
 		reference operator*()
 		{
-			return *mBegin;
+			return *m_begin;
 		}
 
 		pointer operator->()
 		{
-			return &*mBegin;
+			m_current = *m_begin;
+			return &m_current;
 		}
 
 		conditional_iterator_impl &operator++()
 		{
-			while (mBegin != mEnd)
+			while (m_begin != m_end)
 			{
-				if (++mBegin == mEnd)
+				if (++m_begin == m_end)
 					break;
 				
-				if (m_condition->operator()(mBegin))
+				if (m_condition->operator()(m_begin))
 					break;
 			}
 
@@ -591,18 +592,22 @@ class conditional_iterator_proxy
 			return result;
 		}
 
-		bool operator==(const conditional_iterator_impl &rhs) const { return mBegin == rhs.mBegin; }
-		bool operator!=(const conditional_iterator_impl &rhs) const { return mBegin != rhs.mBegin; }
+		bool operator==(const conditional_iterator_impl &rhs) const { return m_begin == rhs.m_begin; }
+		bool operator!=(const conditional_iterator_impl &rhs) const { return m_begin != rhs.m_begin; }
+
+		bool operator==(const row_iterator &rhs) const { return m_begin == rhs; }
+		bool operator!=(const row_iterator &rhs) const { return m_begin != rhs; }
 
 		template <typename IRowType, typename... ITs>
-		bool operator==(const iterator_impl<IRowType, ITs...> &rhs) const { return mBegin == rhs; }
+		bool operator==(const iterator_impl<IRowType, ITs...> &rhs) const { return m_begin == rhs; }
 
 		template <typename IRowType, typename... ITs>
-		bool operator!=(const iterator_impl<IRowType, ITs...> &rhs) const { return mBegin != rhs; }
+		bool operator!=(const iterator_impl<IRowType, ITs...> &rhs) const { return m_begin != rhs; }
 
 	  private:
-		CategoryType *mCat;
-		base_iterator mBegin, mEnd;
+		CategoryType *m_cat;
+		base_iterator m_begin, m_end;
+		value_type m_current;
 		const condition *m_condition;
 	};
 
@@ -673,13 +678,13 @@ iterator_proxy<Category, Ts...>::iterator_proxy(Category &cat, row_iterator pos,
 template <typename Category, typename... Ts>
 conditional_iterator_proxy<Category, Ts...>::conditional_iterator_impl::conditional_iterator_impl(
 	Category &cat, row_iterator pos, const condition &cond, const std::array<uint16_t, N> &cix)
-	: mCat(&cat)
-	, mBegin(pos, cix)
-	, mEnd(cat.end(), cix)
+	: m_cat(&cat)
+	, m_begin(pos, cix)
+	, m_end(cat.end(), cix)
 	, m_condition(&cond)
 {
 	if (m_condition == nullptr or m_condition->empty())
-		mBegin = mEnd;
+		m_begin = m_end;
 }
 
 template <typename Category, typename... Ts>

include/cif++/validate.hpp

@@ -202,7 +202,7 @@ struct category_validator
 	}
 
 	/// @brief Add item_validator @a v to the list of item validators
-	void addItemValidator(item_validator &&v);
+	void add_item_validator(item_validator &&v);
 
 	/// @brief Return the item_validator for item @a tag, may return nullptr
 	const item_validator *get_validator_for_item(std::string_view tag) const;

src/category.cpp

@@ -672,7 +672,7 @@ void category::set_validator(const validator *v, datablock &db)
 				std::ostringstream msg;
 				msg << "Cannot construct index since the key field" << (missing.size() > 1 ? "s" : "") << " "
 							<< cif::join(missing, ", ") << " in " << m_name << " " << (missing.size() == 1 ? "is" : "are") << " missing\n";
-				throw std::runtime_error(msg.str());
+				throw missing_key_error(msg.str(), *missing.begin());
 			}
 		}
 	}
@@ -860,7 +860,7 @@ bool category::validate_links() const
 			auto cond = get_parents_condition(r, *parent);
 			if (not cond)
 				continue;
-			if (not parent->exists(std::move(cond)))
+			if (not parent->contains(std::move(cond)))
 			{
 				++missing;
 				if (VERBOSE and first_missing_rows.size() < 5)
@@ -997,7 +997,7 @@ bool category::has_children(row_handle r) const
 
 	for (auto &&[childCat, link] : m_child_links)
 	{
-		if (not childCat->exists(get_children_condition(r, *childCat)))
+		if (not childCat->contains(get_children_condition(r, *childCat)))
 			continue;
 
 		result = true;
@@ -1013,7 +1013,7 @@ bool category::has_parents(row_handle r) const
 
 	for (auto &&[parentCat, link] : m_parent_links)
 	{
-		if (not parentCat->exists(get_parents_condition(r, *parentCat)))
+		if (not parentCat->contains(get_parents_condition(r, *parentCat)))
 			continue;
 
 		result = true;
@@ -1214,7 +1214,7 @@ void category::erase_orphans(condition &&cond, category &parent)
 		if (not cond(r))
 			continue;
 		
-		if (parent.exists(get_parents_condition(r, parent)))
+		if (parent.contains(get_parents_condition(r, parent)))
 			continue;
 
 		if (VERBOSE > 1)
@@ -1263,7 +1263,7 @@ std::string category::get_unique_id(std::function<std::string(int)> generator)
 	{
 		for (;;)
 		{
-			if (not exists(key(id_tag) == result))
+			if (not contains(key(id_tag) == result))
 				break;
 			
 			result = generator(static_cast<int>(m_last_unique_num++));
@@ -1273,6 +1273,36 @@ std::string category::get_unique_id(std::function<std::string(int)> generator)
 	return result;
 }
 
+std::string category::get_unique_value(std::string_view tag)
+{
+	std::string result;
+
+	if (m_validator and m_cat_validator)
+	{
+		auto iv = m_cat_validator->get_validator_for_item(tag);
+
+		if (iv and iv->m_type and iv->m_type->m_primitive_type == DDL_PrimitiveType::Numb)
+		{
+			uint64_t v = find_max<uint64_t>(tag);
+			result = std::to_string(v + 1);
+		}
+	}
+
+	if (result.empty())
+	{
+		// brain-dead implementation
+		for (size_t ix = 0; ix < size(); ++ix)
+		{
+			// result = m_name + "-" + std::to_string(ix);
+			result = cif_id_for_number(ix);
+			if (not contains(key(tag) == result))
+				break;
+		}
+	}
+
+	return result;
+}
+
 void category::update_value(const std::vector<row_handle> &rows, std::string_view tag, std::string_view value)
 {
 	using namespace std::literals;
@@ -1376,7 +1406,7 @@ void category::update_value(const std::vector<row_handle> &rows, std::string_vie
 						check = std::move(check) && key(ck) == parent[pk].text();
 				}
 
-				if (childCat->exists(std::move(check))) // phew..., narrow escape
+				if (childCat->contains(std::move(check))) // phew..., narrow escape
 					continue;
 
 				// create the actual copy, if we can...
@@ -1405,7 +1435,7 @@ void category::update_value(const std::vector<row_handle> &rows, std::string_vie
 void category::update_value(row *row, uint16_t column, std::string_view value, bool updateLinked, bool validate)
 {
 	// make sure we have an index, if possible
-	if (m_index == nullptr and m_cat_validator != nullptr)
+	if ((updateLinked or validate) and m_index == nullptr and m_cat_validator != nullptr)
 		m_index = new category_index(*this);
 
 	auto &col = m_columns[column];
@@ -1444,7 +1474,7 @@ void category::update_value(row *row, uint16_t column, std::string_view value, b
 	if (not value.empty())
 		row->append(column, { value });
 
-	if (reinsert)
+	if (reinsert and m_index != nullptr)
 		m_index->insert(*this, row);
 
 	// see if we need to update any child categories that depend on this value

src/dictionary_parser.cpp

@@ -87,7 +87,7 @@ class dictionary_parser : public parser
 				error("Undefined category '" + iv.first);
 
 			for (auto &v : iv.second)
-				const_cast<category_validator *>(cv)->addItemValidator(std::move(v));
+				const_cast<category_validator *>(cv)->add_item_validator(std::move(v));
 		}
 
 		// check all item validators for having a typeValidator

src/model.cpp

@@ -1783,7 +1783,7 @@ atom &structure::emplace_atom(atom &&atom)
 	std::string symbol = atom.get_property("type_symbol");
 
 	using namespace cif::literals;
-	if (not atom_type.exists("symbol"_key == symbol))
+	if (not atom_type.contains("symbol"_key == symbol))
 		atom_type.emplace({ { "symbol", symbol } });
 
 	return m_atoms.emplace_back(std::move(atom));
@@ -1969,7 +1969,7 @@ void structure::change_residue(residue &res, const std::string &newCompound,
 
 		// create rest
 		auto &chemComp = m_db["chem_comp"];
-		if (not chemComp.exists(key("id") == newCompound))
+		if (not chemComp.contains(key("id") == newCompound))
 		{
 			chemComp.emplace({{"id", newCompound},
 				{"name", compound->name()},
@@ -2702,7 +2702,7 @@ void structure::cleanup_empty_categories()
 	for (auto chemComp : chem_comp)
 	{
 		std::string compID = chemComp["id"].as<std::string>();
-		if (atomSite.exists("label_comp_id"_key == compID or "auth_comp_id"_key == compID))
+		if (atomSite.contains("label_comp_id"_key == compID or "auth_comp_id"_key == compID))
 			continue;
 
 		obsoleteChemComps.push_back(chemComp);
@@ -2719,7 +2719,7 @@ void structure::cleanup_empty_categories()
 	for (auto entity : entities)
 	{
 		std::string entityID = entity["id"].as<std::string>();
-		if (atomSite.exists("label_entity_id"_key == entityID))
+		if (atomSite.contains("label_entity_id"_key == entityID))
 			continue;
 
 		obsoleteEntities.push_back(entity);

src/pdb/validate-pdbx.cpp

@@ -183,7 +183,7 @@ bool is_valid_pdbx_file(const file &file, std::string_view dictionary)
 						"label_asym_id"_key == asym_id and
 						"label_seq_id"_key == seq_id and not std::move(cond);
 					
-					if (atom_site.exists(std::move(cond)))
+					if (atom_site.contains(std::move(cond)))
 						throw validation_error("An atom_site record exists that has no parent in the poly seq scheme categories");
 				}
 			}

src/validate.cpp

@@ -233,7 +233,7 @@ void item_validator::operator()(std::string_view value) const
 
 // --------------------------------------------------------------------
 
-void category_validator::addItemValidator(item_validator &&v)
+void category_validator::add_item_validator(item_validator &&v)
 {
 	if (v.m_mandatory)
 		m_mandatory_fields.insert(v.m_tag);

test/unit-v2-test.cpp

@@ -568,7 +568,7 @@ _test.value
 	auto &test = db["test"];
 	REQUIRE(test.size() == 5);
 
-	REQUIRE(test.exists("value"_key == cif::null));
+	REQUIRE(test.contains("value"_key == cif::null));
 	REQUIRE(test.find("value"_key == cif::null).size() == 2);
 }