fixes in pdbx validation, compound one letter code

src/compound.cpp

@@ -136,9 +136,14 @@ compound::compound(cif::datablock &db)
 	if (chemComp.size() != 1)
 	if (chemComp.size() != 1)
 		throw std::runtime_error("Invalid compound file, chem_comp should contain a single row");
 		throw std::runtime_error("Invalid compound file, chem_comp should contain a single row");
 
 
-	cif::tie(m_id, m_name, m_type, m_formula, m_formula_weight, m_formal_charge, m_one_letter_code, m_parent_id) =
+	std::string one_letter_code;
+
+	cif::tie(m_id, m_name, m_type, m_formula, m_formula_weight, m_formal_charge, one_letter_code, m_parent_id) =
 		chemComp.front().get("id", "name", "type", "formula", "formula_weight", "pdbx_formal_charge", "one_letter_code", "mon_nstd_parent_comp_id");
 		chemComp.front().get("id", "name", "type", "formula", "formula_weight", "pdbx_formal_charge", "one_letter_code", "mon_nstd_parent_comp_id");
 	
 	
+	if (one_letter_code.length() == 1)
+		m_one_letter_code = one_letter_code.front();
+
 	// The name should not contain newline characters since that triggers validation errors later on
 	// The name should not contain newline characters since that triggers validation errors later on
 	cif::replace_all(m_name, "\n", "");
 	cif::replace_all(m_name, "\n", "");
 
 

src/pdb/validate-pdbx.cpp

@@ -188,7 +188,7 @@ bool is_valid_pdbx_file(const file &file, std::string_view dictionary)
 				}
 				}
 			}
 			}
 
 
-			const auto &[seq, seq_can] = entity_poly.find1<std::optional<std::string>, std::optional<std::string>>("entity_id"_key == entity_id,
+			auto &&[seq, seq_can] = entity_poly.find1<std::optional<std::string>, std::optional<std::string>>("entity_id"_key == entity_id,
 				"pdbx_seq_one_letter_code", "pdbx_seq_one_letter_code_can");
 				"pdbx_seq_one_letter_code", "pdbx_seq_one_letter_code_can");
 			
 			
 			std::string::const_iterator si, sci, se, sce;
 			std::string::const_iterator si, sci, se, sce;
@@ -245,18 +245,29 @@ bool is_valid_pdbx_file(const file &file, std::string_view dictionary)
 				if (cif::VERBOSE > 0)
 				if (cif::VERBOSE > 0)
 					std::clog << "Warning: entity_poly has no sequence for entity_id " << entity_id << '\n';
 					std::clog << "Warning: entity_poly has no sequence for entity_id " << entity_id << '\n';
 			}
 			}
-			else if (not seq_match(false, seq->begin(), seq->end()))
+			else
+			{
+				seq->erase(std::remove_if(seq->begin(), seq->end(), [](char ch) { return std::isspace(ch); }), seq->end());
+
+				if (not seq_match(false, seq->begin(), seq->end()))
 					throw validation_error("Sequences do not match for entity " + entity_id);
 					throw validation_error("Sequences do not match for entity " + entity_id);
+			}
 
 
 			if (not seq_can.has_value())
 			if (not seq_can.has_value())
 			{
 			{
 				if (cif::VERBOSE > 0)
 				if (cif::VERBOSE > 0)
 					std::clog << "Warning: entity_poly has no sequence for entity_id " << entity_id << '\n';
 					std::clog << "Warning: entity_poly has no sequence for entity_id " << entity_id << '\n';
 			}
 			}
-			else if (not seq_match(true, seq_can->begin(), seq_can->end()))
+			else
+			{
+				seq_can->erase(std::remove_if(seq_can->begin(), seq_can->end(), [](char ch) { return std::isspace(ch); }), seq_can->end());
+				
+				if (not seq_match(true, seq_can->begin(), seq_can->end()))
 					throw validation_error("Canonical sequences do not match for entity " + entity_id);
 					throw validation_error("Canonical sequences do not match for entity " + entity_id);
 			}
 			}
 			
 			
+		}
+
 		result = true;
 		result = true;
 	}
 	}
 	catch (const std::exception &ex)
 	catch (const std::exception &ex)