Update _filter_xyz to ignore lines after cartesian coordinates in sta…

…ndard xyz files
MolSSI · Apr 13, 2022 · 9c22bdc · 9c22bdc
1 parent d04a380
commit 9c22bdc
Show file tree

Hide file tree

Showing 2 changed files with 223 additions and 1 deletion.
diff --git a/qcelemental/molparse/from_string.py b/qcelemental/molparse/from_string.py
@@ -671,7 +671,7 @@ def process_variable(matchobj):
 xyz1strict = re.compile(r"\A" + r"(?P<nat>\d+)" + r"\Z")
 SIMPLENUCLEUS = r"""((?P<E>[A-Z]{1,3})|(?P<Z>\d{1,3}))"""
 atom_cartesian_strict = re.compile(
-    r"\A" + r"(?P<nucleus>" + SIMPLENUCLEUS + r")" + SEP + CARTXYZ + r"\Z", re.IGNORECASE | re.VERBOSE
+    r"\A" + r"(?P<nucleus>" + SIMPLENUCLEUS + r")" + SEP + CARTXYZ + ".*" + r"\Z", re.IGNORECASE | re.VERBOSE
 )
 
 xyz1 = re.compile(r"\A" + r"(?P<nat>\d+)" + r"[\s,]*" + r"((?P<ubohr>(bohr|au))|(?P<uang>ang))?" + r"\Z", re.IGNORECASE)
@@ -737,9 +737,17 @@ def process_atom_cartesian(matchobj):
         for iln, line in enumerate(string.split("\n")):
             line = line.strip()
             if iln == 0:
+                try:
+                    num_atoms = int(line)
+                except ValueError:
+                    # Not a standard xyz format; continue with regular process
+                    num_atoms = None
                 line = re.sub(xyz1strict, "", line)
             elif iln == 1:
                 continue
+            elif num_atoms and iln > num_atoms + 1:
+                # If standard xyz ignore everything after cartesian coords
+                break
             else:
                 line = re.sub(atom_cartesian_strict, process_atom_cartesian, line)
             if line:

diff --git a/qcelemental/tests/test_molparse_from_string.py b/qcelemental/tests/test_molparse_from_string.py
@@ -5,6 +5,7 @@
 
 import qcelemental
 from qcelemental.models import Molecule
+from qcelemental.molparse.from_string import _filter_xyz
 from qcelemental.testing import compare, compare_molrecs, compare_recursive, compare_values, tnm
 
 _arrays_prov_stamp = {"creator": "QCElemental", "version": "1.0", "routine": "qcelemental.molparse.from_arrays"}
@@ -921,6 +922,219 @@ def test_xyzp_qm_7e():
     assert compare_molrecs(fullans, final["qm"], tnm() + ": full qm")
 
 
+@pytest.mark.parametrize(
+    "string,elbl,geom",
+    (
+        (
+            """5
+gdb 1	157.7118	157.70997	157.70699	0.	13.21	-0.3877	0.1171	0.5048	35.3641	0.044749	-40.47893	-40.476062	-40.475117	-40.498597	6.469	
+C	-0.0126981359	 1.0858041578	 0.0080009958	-0.535689
+H	 0.002150416	-0.0060313176	 0.0019761204	 0.133921
+H	 1.0117308433	 1.4637511618	 0.0002765748	 0.133922
+H	-0.540815069	 1.4475266138	-0.8766437152	 0.133923
+H	-0.5238136345	 1.4379326443	 0.9063972942	 0.133923
+1341.307	1341.3284	1341.365	1562.6731	1562.7453	3038.3205	3151.6034	3151.6788	3151.7078
+C	C	
+InChI=1S/CH4/h1H4	InChI=1S/CH4/h1H4""",
+            ["C", "H", "H", "H", "H"],
+            [
+                -0.0126981359,
+                1.0858041578,
+                0.0080009958,
+                0.002150416,
+                -0.0060313176,
+                0.0019761204,
+                1.0117308433,
+                1.4637511618,
+                0.0002765748,
+                -0.540815069,
+                1.4475266138,
+                -0.8766437152,
+                -0.5238136345,
+                1.4379326443,
+                0.9063972942,
+            ],
+        ),
+        (
+            """20
+        gdb 52625	3.48434	0.81389	0.77349	4.0931	85.49	-0.2471	0.0275	0.2746	1578.0163	0.16756	-403.158572	-403.147447	-403.146503	-403.196607	38.82
+        C	 0.0219866132	 1.4617007325	 0.0778162941	-0.5003
+        C	 0.0172170008	 0.0062570163	 0.0278221402	 0.060798
+        C	 0.0082754818	-1.1968179556	-0.0219490036	 0.058561
+        C	-0.0048543985	-2.6632285567	-0.0864436684	 0.086487
+        C	-1.4625346892	-3.1775528493	-0.0739515886	-0.401401
+        C	 0.724612823	-3.1404242978	-1.3630324305	-0.401392
+        N	 0.7132087245	-3.1481618818	 1.1179206119	-0.295483
+        C	 0.9247284349	-4.4479813471	 1.4397739679	 0.162556
+        O	 0.5626675182	-5.4147891002	 0.7982523365	-0.336127
+        H	 1.0431199948	 1.8581192318	 0.0749725881	 0.150933
+        H	-0.4997253695	 1.8847111976	-0.7874860262	 0.154682
+        H	-0.4780438734	 1.8320081388	 0.979321682	 0.150927
+        H	-1.9940013133	-2.8054928472	-0.9536444371	 0.11928
+        H	-1.4636215253	-4.2685570577	-0.0833553058	 0.154594
+        H	-1.981857154	-2.824763824	 0.8204749775	 0.116725
+        H	 1.749444309	-2.7614376976	-1.3787041159	 0.11672
+        H	 0.7443784569	-4.2310758294	-1.3847506905	 0.154572
+        H	 0.2030559276	-2.7681902401	-2.2485640369	 0.11929
+        H	 1.0701803562	-2.4436691628	 1.7431541049	 0.251085
+        H	 1.4874481325	-4.5389119798	 2.3916335805	 0.077494
+        9.6558	74.9555	86.4308	130.4187	209.6956	213.4664	245.0078	277.3227	292.8775	293.3837	344.8375	364.9338	411.1024	508.5101	542.9334	570.3714	618.7374	820.0082	829.2715	936.286	986.4059	1028.4996	1033.1508	1051.8084	1054.6924	1076.9172	1172.1103	1179.8358	1222.2743	1289.1699	1387.6509	1408.9379	1414.4085	1431.5059	1476.1153	1477.0789	1478.6118	1481.2338	1494.3428	1511.202	1519.6039	1792.8486	2370.6229	2920.0767	3029.2922	3050.3711	3054.2985	3090.8073	3092.1476	3124.8526	3128.7105	3148.8786	3152.1387	3639.6323
+        CC#CC(C)(C)NC=O	CC#CC(C)(C)NC=O
+        InChI=1S/C7H11NO/c1-4-5-7(2,3)8-6-9/h6H,1-3H3,(H,8,9)	InChI=1S/C7H11NO/c1-4-5-7(2,3)8-6-9/h6H,1-3H3,(H,8,9)
+        """,
+            ["C", "C", "C", "C", "C", "C", "N", "C", "O", "H", "H", "H", "H", "H", "H", "H", "H", "H", "H", "H"],
+            [
+                0.0219866132,
+                1.4617007325,
+                0.0778162941,
+                0.0172170008,
+                0.0062570163,
+                0.0278221402,
+                0.0082754818,
+                -1.1968179556,
+                -0.0219490036,
+                -0.0048543985,
+                -2.6632285567,
+                -0.0864436684,
+                -1.4625346892,
+                -3.1775528493,
+                -0.0739515886,
+                0.724612823,
+                -3.1404242978,
+                -1.3630324305,
+                0.7132087245,
+                -3.1481618818,
+                1.1179206119,
+                0.9247284349,
+                -4.4479813471,
+                1.4397739679,
+                0.5626675182,
+                -5.4147891002,
+                0.7982523365,
+                1.0431199948,
+                1.8581192318,
+                0.0749725881,
+                -0.4997253695,
+                1.8847111976,
+                -0.7874860262,
+                -0.4780438734,
+                1.8320081388,
+                0.979321682,
+                -1.9940013133,
+                -2.8054928472,
+                -0.9536444371,
+                -1.4636215253,
+                -4.2685570577,
+                -0.0833553058,
+                -1.981857154,
+                -2.824763824,
+                0.8204749775,
+                1.749444309,
+                -2.7614376976,
+                -1.3787041159,
+                0.7443784569,
+                -4.2310758294,
+                -1.3847506905,
+                0.2030559276,
+                -2.7681902401,
+                -2.2485640369,
+                1.0701803562,
+                -2.4436691628,
+                1.7431541049,
+                1.4874481325,
+                -4.5389119798,
+                2.3916335805,
+            ],
+        ),
+        (
+            """17
+        gdb 107395	3.22974	1.27619	1.028	3.0629	69.23	-0.225	0.0454	0.2704	1155.6964	0.136978	-455.1439	-455.135732	-455.134788	-455.176691	31.549
+        N	 0.0166534686	 1.2958609713	-0.1502735485	-0.580455
+        C	-0.0682337319	-0.0600800983	-0.0096861238	 0.457461
+        N	-1.1305816717	-0.7597824714	 0.1390863864	-0.408126
+        C	-0.7009986293	-2.1439574926	 0.1108213578	 0.311595
+        C	-1.4984924036	-3.0589993793	 1.002599169	-0.138474
+        O	-2.8585678608	-3.0949107355	 0.6105794873	-0.417767
+        C	-0.0617590849	-2.6053899658	-1.1855277252	-0.354308
+        C	 0.8014538633	-2.0927793706	-0.0638522254	 0.002142
+        O	 1.1296621	-0.7079465651	-0.1012706145	-0.24761
+        H	 0.8924648972	 1.705797935	 0.136582908	 0.26608
+        H	-0.8059467993	 1.7929265086	 0.153937696	 0.270518
+        H	-1.1201801933	-4.0847553601	 0.9397586373	 0.099132
+        H	-1.3951151651	-2.7241068393	 2.0475874391	 0.085667
+        H	-3.1252104475	-2.1733539114	 0.5051511429	 0.295168
+        H	-0.0217120799	-3.6759992339	-1.3586546368	 0.121153
+        H	-0.1831965908	-1.9943417386	-2.0738763382	 0.124725
+        H	 1.5549266689	-2.687136563	 0.4367361986	 0.113098
+        70.6081	153.9333	180.5159	278.5491	345.0174	381.8663	413.9812	430.6681	448.7324	482.9536	553.6703	659.1349	739.6911	741.9888	783.4372	872.8339	876.0968	943.3731	959.9083	980.6717	997.5511	1046.3298	1069.8255	1090.0307	1128.0369	1148.246	1187.611	1267.6525	1296.0396	1370.3317	1379.5593	1424.1337	1443.8219	1468.4333	1503.1705	1614.9407	1733.6758	2971.4118	3070.4849	3127.1322	3195.3399	3220.8339	3586.8344	3701.0432	3786.2797
+        NC1=NC2(CO)CC2O1	NC1=N[C@@]2(CO)C[C@H]2O1
+        InChI=1S/C5H8N2O2/c6-4-7-5(2-8)1-3(5)9-4/h3,8H,1-2H2,(H2,6,7)	InChI=1S/C5H8N2O2/c6-4-7-5(2-8)1-3(5)9-4/h3,8H,1-2H2,(H2,6,7)/t3-,5-/m1/s1
+        """,
+            ["N", "C", "N", "C", "C", "O", "C", "C", "O", "H", "H", "H", "H", "H", "H", "H", "H"],
+            [
+                0.0166534686,
+                1.2958609713,
+                -0.1502735485,
+                -0.0682337319,
+                -0.0600800983,
+                -0.0096861238,
+                -1.1305816717,
+                -0.7597824714,
+                0.1390863864,
+                -0.7009986293,
+                -2.1439574926,
+                0.1108213578,
+                -1.4984924036,
+                -3.0589993793,
+                1.002599169,
+                -2.8585678608,
+                -3.0949107355,
+                0.6105794873,
+                -0.0617590849,
+                -2.6053899658,
+                -1.1855277252,
+                0.8014538633,
+                -2.0927793706,
+                -0.0638522254,
+                1.1296621,
+                -0.7079465651,
+                -0.1012706145,
+                0.8924648972,
+                1.705797935,
+                0.136582908,
+                -0.8059467993,
+                1.7929265086,
+                0.153937696,
+                -1.1201801933,
+                -4.0847553601,
+                0.9397586373,
+                -1.3951151651,
+                -2.7241068393,
+                2.0475874391,
+                -3.1252104475,
+                -2.1733539114,
+                0.5051511429,
+                -0.0217120799,
+                -3.6759992339,
+                -1.3586546368,
+                -0.1831965908,
+                -1.9943417386,
+                -2.0738763382,
+                1.5549266689,
+                -2.687136563,
+                0.4367361986,
+            ],
+        ),
+    ),
+)
+def test_xyz_gdb_format(string, elbl, geom):
+    unprocessed, processed = _filter_xyz(string, strict=True)
+
+    assert not unprocessed
+    assert processed["elbl"] == elbl
+    assert processed["geom"] == geom
+
+
 subject8 = """\
 3
    stuffs