#2331 - System should throw an error in case of wrong IUBcode

Fix code. Add UT.
epam · Oct 17, 2024 · 1810087 · 1810087
1 parent 13d2d36
commit 1810087
Show file tree

Hide file tree

Showing 3 changed files with 17 additions and 0 deletions.
diff --git a/api/tests/integration/ref/formats/idt_to_ket.py.out b/api/tests/integration/ref/formats/idt_to_ket.py.out
@@ -30,6 +30,7 @@ idt_unresolved.ket:SUCCEED
 idt_unresolved_many.ket:SUCCEED
 idt_unsplit.ket:SUCCEED
 Test '!+A-$#12w12r23e32e33': got expected error 'Invalid symbols in the sequence: !,-,$,#,1,2,w,1,2,2,3,e,3,2,e,3,3'
+Test '(YY:00330067)': got expected error 'Invalid mixed base - only numerical index allowed.'
 Test '+/5Phos/A': got expected error 'Sugar prefix could not be used with modified monomer.'
 Test '/': got expected error 'Unexpected end of data'
 Test '//': got expected error 'Invalid modification: empty string.'

diff --git a/api/tests/integration/tests/formats/idt_to_ket.py b/api/tests/integration/tests/formats/idt_to_ket.py
@@ -101,6 +101,7 @@ def find_diff(a, b):
     "/52MOErA//32MOErA/*": "Monomer /32MOErA/ doesn't have phosphate, so '*' couldn't be applied.",
     "/3Phos/*": "Symbol '*' could be placed only between two nucleotides/nucleosides.",
     "r(B1:50003000)(B1)": "Unknown mixed base 'B1'",
+    "(YY:00330067)": "Invalid mixed base - only numerical index allowed.",
 }
 for idt_seq in sorted(idt_errors.keys()):
     error = idt_errors[idt_seq]

diff --git a/core/indigo-core/molecule/src/sequence_loader.cpp b/core/indigo-core/molecule/src/sequence_loader.cpp
@@ -1081,17 +1081,32 @@ void SequenceLoader::loadIdt(KetDocument& document)
                     if (mixed_base.back() == ')')
                     {
                         mixed_base = idt_alias.substr(1, idt_alias.size() - 2);
+                        auto check_mixed_base = [](const std::string& base) {
+                            if (base.size() < 2)
+                                return;
+                            auto count = base.substr(1, base.size() - 1);
+                            for (auto ch : count)
+                            {
+                                if (!std::isdigit(ch))
+                                    throw Error("Invalid mixed base - only numerical index allowed.");
+                            }
+                        };
                         if (auto pos = mixed_base.find(':'); pos != std::string::npos)
                         {
                             auto ratios_str = mixed_base.substr(pos + 1, mixed_base.size() - pos - 1);
                             mixed_base = mixed_base.substr(0, pos);
+                            check_mixed_base(mixed_base);
                             if (ratios_str.size() != 8)
                                 throw Exception("Invalid IDT variant monomer %s", idt_alias.c_str());
                             ratios.emplace(std::array<float, 4>{std::stof(ratios_str.substr(0, 2)), std::stof(ratios_str.substr(2, 2)),
                                                                 std::stof(ratios_str.substr(4, 2)), std::stof(ratios_str.substr(6, 2))});
                             idt_alias = '(' + mixed_base + ')';
                             mixed_base = mixed_base[0];
                         }
+                        else
+                        {
+                            check_mixed_base(mixed_base);
+                        }
                     }
                     if (sugar == "R" && RNA_DNA_MIXED_BASES.count(mixed_base) == 0)
                         idt_alias = 'r' + idt_alias;