Keep both oxygens and the acyl carbon of a peroxy ester

A peroxy ester rendered as a prefix lost an oxygen and gained a carbon.
`-C(=O)-O-O-tBu` came out as `tert-butoxycarbonyl`, which is
`-C(=O)-O-tBu`, and the acyl carbon stayed in the parent chain on top of
that, so `COC(=O)CCC(=O)OOC(C)(C)C` was named
`methyl 4-(tert-butoxycarbonyl)butanoate` -- ten carbons and four oxygens
for a molecule with nine and five.

Two causes.  `ester_prefix_handler` is shared with peroxy esters and spells
a single `…oxycarbonyl`; peroxy esters now get their own `…peroxycarbonyl`.
And `chain_external_carbonyl_groups` listed `ester` but not `peroxy_ester`,
so the acyl carbon was never moved out of the parent the way a plain ester's
is -- which is why the same molecule came out as a butanoate where the plain
ester gives a propanoate.

Of the three names this changes across 200k molecules, two go from wrong to
correct and none get worse; every probe round-trips through OPSIN.  One
benzoyl peroxy ester still fails to name, but it failed before this change
too.

Refutations 24 -> 23.  Both benchmarks otherwise unchanged.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

Author: AdrianM0

src/openclatura/data/namer_rules.json

@@ -492,6 +492,8 @@
       "ring_carboxylic_acid",
       "ester",
       "ring_carboxylate",
+      "peroxy_ester",
+      "ring_peroxy_ester",
       "amide",
       "ring_amide",
       "thioamide",

src/openclatura/functional_prefixes.py

@@ -6,7 +6,13 @@
 from dataclasses import dataclass
 
 from .assembly_parts import NameTokenBinding, SubstituentItem, rendered_substituent_text
-from .formatting import format_counted_prefixes, is_complex_prefix, oxy_prefix_from_branch, strip_outer_parentheses
+from .formatting import (
+    format_counted_prefixes,
+    format_multiplier,
+    is_complex_prefix,
+    oxy_prefix_from_branch,
+    strip_outer_parentheses,
+)
 from .group_atom_roles import amide_nitrogen, ester_or_peroxy_single_oxygen
 from .molecule import Molecule
 from .naming_protocols import RecursiveSubgraphNamer
@@ -70,7 +76,35 @@ def amide_prefix_from_group(
     return f"({format_counted_prefixes(sub_names)}{base})"
 
 
+def peroxy_ester_prefix_from_group(
+    mol: Molecule,
+    group: PerceivedGroup,
+    sub_exclude: set[int],
+    branch_namer: RecursiveSubgraphNamer,
+) -> str:
+    """Return a ``…peroxycarbonyl`` prefix for a peroxy ester.
+
+    ``oxycarbonyl`` spells one single-bonded oxygen, and a peroxy ester has
+    two, so ``-C(=O)-O-O-tBu`` came out as ``tert-butoxycarbonyl`` -- an oxygen
+    short, and a carbon long once the parent absorbed the acyl carbon anyway.
+    """
+
+    single_o = ester_or_peroxy_single_oxygen(mol, group)
+    if single_o is None:
+        return ""
+    r_group_c = next((n for n in mol.get_neighbors(single_o) if n not in group.atoms_involved), None)
+    if r_group_c is None:
+        return ""
+    branch_name = branch_namer(mol, r_group_c, sub_exclude | {single_o}, upstream_atom=single_o)
+    if not branch_name:
+        return ""
+    branch_name = strip_outer_parentheses(rendered_substituent_text(branch_name))
+    return f"({format_multiplier(branch_name, 1)}peroxycarbonyl)"
+
+
 def ester_prefix_handler(context: PrefixContext, group: PerceivedGroup) -> str:
+    if group.key in RULES.prefixes.peroxy_ester_groups:
+        return peroxy_ester_prefix_from_group(context.mol, group, context.sub_exclude, context.branch_namer)
     return ester_prefix_from_group(context.mol, group, context.sub_exclude, "carbonyl", context.branch_namer)
 
 

src/openclatura/tests/test_analysis.py

@@ -5677,3 +5677,14 @@ def test_azinic_acid_is_distinguished_from_a_nitro_group():
     assert name_smiles("CC=[N+]([O-])O") == "(ethylidene)azinic acid"
     assert name_smiles("c1ccccc1[N+](=O)[O-]") == "nitrobenzene"
     assert name_smiles("C[N+](=O)[O-]") == "nitromethane"
+
+
+def test_peroxy_ester_prefix_keeps_both_oxygens_and_its_carbon():
+    # "oxycarbonyl" spells one single-bonded oxygen and a peroxy ester has two,
+    # and the acyl carbon was left in the parent chain as well, so the prefix
+    # was an oxygen short and a carbon long.
+    assert name_smiles("OC(=O)CCC(=O)OOC") == "3-(methylperoxycarbonyl)propanoic acid"
+    assert name_smiles("COC(=O)CCC(=O)OOC(C)(C)C") == "methyl 3-((tert-butylperoxy)carbonyl)propanoate"
+    # The plain ester and the peroxy-ester suffix are both unaffected.
+    assert name_smiles("OC(=O)CCC(=O)OC") == "3-(methoxycarbonyl)propanoic acid"
+    assert name_smiles("CC(=O)OOC(C)(C)C") == "tert-butyl ethaneperoxoate"