fix: fix two-character elements (#1537)

reacnetgenerator/_reachtml.py

@@ -60,8 +60,11 @@ def _re(self, smi):
         [CH]
         >>> self._re('Na')
         [Na]
+        >>> self._re('[H]c(Cl)C([H])Cl')
+        [H][c]([Cl])[C]([H])[Cl]
         """
-        elements = "|".join([((an.upper() + "|" + an.lower()) if len(an)==1 else an) for an in self.atomname if an != 'H'])
+        Satom = sorted(self.atomname, key = lambda i:len(i), reverse=True)
+        elements = "|".join([((an.upper() + "|" + an.lower()) if len(an)==1 else an) for an in Satom if an != 'H'])
         smi = re.sub(r'(?<!\[)(' + elements + r')(?!H)', r'[\1]', smi)
         return smi.replace("[HH]", "[H]")
 

reacnetgenerator/test/test_reacnetgen.py

@@ -15,6 +15,7 @@
 from reacnetgenerator import ReacNetGenerator
 from reacnetgenerator._detect import _Detect
 from reacnetgenerator._hmmfilter import _HMMFilter
+from reacnetgenerator._reachtml import _HTMLResult
 from reacnetgenerator.gui import GUI
 from reacnetgenerator.utils import checksha256, download_multifiles, listtobytes
 from reacnetgenerator.commandline import parm2cmd
@@ -112,3 +113,14 @@ def test_benchmark_hmm(self, benchmark, reacnetgen_param):
         compressed_bytes = [listtobytes((5, 6)), listtobytes(
             ((5, 6, 1),)), listtobytes(index)]
         benchmark(hmmclass._getoriginandhmm, compressed_bytes)
+
+    def test_re(self, reacnetgen_param):
+        reacnetgen = ReacNetGenerator(**reacnetgen_param['rngparams'])
+        r = _HTMLResult(reacnetgen)
+        r.atomname = ['C', 'H', 'O', 'Na', 'Cl']
+        assert r._re('C'), '[C]'
+        assert r._re('[C]'), '[C]' 
+        assert r._re('[CH]'), '[CH]'
+        assert r._re('Na'), '[Na]'
+        assert r._re('[H]c(Cl)C([H])Cl'), '[H][c]([Cl])[C]([H])[Cl]'
+