Add parsing for extended element counts in Chemkin thermo entries

Add unit tests for various element count syntaxes
Fix incorrect index for element count in columns 74-78
    (should start at column 73 using zero-indexing)

rmgpy/chemkin.pyx

@@ -80,7 +80,7 @@ def readThermoEntry(entry, Tmin=0, Tint=0, Tmax=0):
 
     comment = lines[0][len(species):24].strip()
     formula = {}
-    for i in [24, 29, 34, 39, 74]:
+    for i in [24, 29, 34, 39, 73]:
         element,count = lines[0][i:i+2].strip(), lines[0][i+2:i+5].strip()
         if element:
             try:
@@ -99,6 +99,21 @@ def readThermoEntry(entry, Tmin=0, Tint=0, Tmax=0):
                         raise
             if count != 0: # Some people put garbage elements in, with zero count. Ignore these. Allow negative counts though (eg. negative one electron)
                 formula[element] = count
+
+    # Parsing for extended elemental composition syntax, adapted from Cantera ck2cti.py
+    if lines[0].rstrip().endswith('&'):
+        complines = []
+        for i in range(len(lines)-1):
+            if lines[i].rstrip().endswith('&'):
+                complines.append(lines[i+1])
+            else:
+                break
+        lines = [lines[0]] + lines[i+1:]
+        elements = ' '.join(line.rstrip('&\n') for line in complines).split()
+        formula = {}
+        for i in range(0, len(elements), 2):
+            formula[elements[i].capitalize()] = int(elements[i+1])
+
     phase = lines[0][44]
     if phase.upper() != 'G':
         logging.warning("Was expecting gas phase thermo data for {0}. Skipping thermo data.".format(species))

rmgpy/chemkinTest.py

@@ -427,7 +427,7 @@ def test_mark_duplicate_reactions(self):
         self.assertEqual(duplicate_flags, expected_flags)
 
 
-class TestThermoWrite(unittest.TestCase):
+class TestThermoReadWrite(unittest.TestCase):
 
     def setUp(self):
         """This method is run once before each test."""
@@ -449,20 +449,42 @@ def setUp(self):
             comment=comment,
         )
 
-    def testWriteThermoBlock(self):
-        """Test that we can write a normal thermo block"""
-        species = Species(SMILES='CC')
-        species.thermo = self.nasa
+        # Chemkin entries for testing - note that the values are all the same
+        self.entry1 = """C2H6                    H   6C   2          G   300.000  3000.000  650.73      1
+-3.07954000E-01 2.45269000E-02-1.24130000E-05 3.07724000E-09-3.01467000E-13    2
+-1.06930000E+04 2.26280000E+01 4.03055000E+00-2.14171000E-03 4.90611000E-05    3
+-5.99027000E-08 2.38945000E-11-1.12576000E+04 3.56130000E+00                   4
+"""
 
-        expected = """C2H6                    H   6C   2          G   300.000  3000.000  650.73      1
+        self.entry2 = """CH3NO2X                 X   1H   3C   1O   2G   300.000  3000.000  650.73N   1 1
 -3.07954000E-01 2.45269000E-02-1.24130000E-05 3.07724000E-09-3.01467000E-13    2
 -1.06930000E+04 2.26280000E+01 4.03055000E+00-2.14171000E-03 4.90611000E-05    3
 -5.99027000E-08 2.38945000E-11-1.12576000E+04 3.56130000E+00                   4
 """
 
+        self.entry3 = """CH3NO2SX                                    G   300.000  3000.000  650.73      1&
+C 1 H 3 O 2 N 1 S 1 X 1
+-3.07954000E-01 2.45269000E-02-1.24130000E-05 3.07724000E-09-3.01467000E-13    2
+-1.06930000E+04 2.26280000E+01 4.03055000E+00-2.14171000E-03 4.90611000E-05    3
+-5.99027000E-08 2.38945000E-11-1.12576000E+04 3.56130000E+00                   4
+"""
+
+    def testWriteThermoBlock(self):
+        """Test that we can write a normal thermo block"""
+        species = Species(SMILES='CC')
+        species.thermo = self.nasa
+
         result = writeThermoEntry(species, verbose=False)
 
-        self.assertEqual(expected, result)
+        self.assertEqual(result, self.entry1)
+
+    def testReadThermoBlock(self):
+        """Test that we can read a normal thermo block"""
+        species, thermo, formula = readThermoEntry(self.entry1)
+
+        self.assertEqual(species, 'C2H6')
+        self.assertEqual(formula, {'H': 6, 'C': 2})
+        self.assertTrue(self.nasa.isIdenticalTo(thermo))
 
     def testWriteThermoBlock5Elem(self):
         """Test that we can write a thermo block for a species with 5 elements"""
@@ -478,15 +500,17 @@ def testWriteThermoBlock5Elem(self):
 """)
         species.thermo = self.nasa
 
-        expected = """CH3NO2X                 X   1H   3C   1O   2G   300.000  3000.000  650.73N   1 1
--3.07954000E-01 2.45269000E-02-1.24130000E-05 3.07724000E-09-3.01467000E-13    2
--1.06930000E+04 2.26280000E+01 4.03055000E+00-2.14171000E-03 4.90611000E-05    3
--5.99027000E-08 2.38945000E-11-1.12576000E+04 3.56130000E+00                   4
-"""
-
         result = writeThermoEntry(species, verbose=False)
 
-        self.assertEqual(expected, result)
+        self.assertEqual(result, self.entry2)
+
+    def testReadThermoBlock5Elem(self):
+        """Test that we can read a thermo block with 5 elements"""
+        species, thermo, formula = readThermoEntry(self.entry2)
+
+        self.assertEqual(species, 'CH3NO2X')
+        self.assertEqual(formula, {'X': 1, 'C': 1, 'O': 2, 'H': 3, 'N': 1})
+        self.assertTrue(self.nasa.isIdenticalTo(thermo))
 
     def testWriteThermoBlock6Elem(self):
         """Test that we can write a thermo block for a species with 6 elements"""
@@ -503,16 +527,17 @@ def testWriteThermoBlock6Elem(self):
 """)
         species.thermo = self.nasa
 
-        expected = """CH3NO2SX                                    G   300.000  3000.000  650.73      1&
-C 1 H 3 O 2 N 1 S 1 X 1
--3.07954000E-01 2.45269000E-02-1.24130000E-05 3.07724000E-09-3.01467000E-13    2
--1.06930000E+04 2.26280000E+01 4.03055000E+00-2.14171000E-03 4.90611000E-05    3
--5.99027000E-08 2.38945000E-11-1.12576000E+04 3.56130000E+00                   4
-"""
-
         result = writeThermoEntry(species, verbose=False)
 
-        self.assertEqual(expected, result)
+        self.assertEqual(result, self.entry3)
+
+    def testReadThermoBlock6Elem(self):
+        """Test that we can read a thermo block with 6 elements"""
+        species, thermo, formula = readThermoEntry(self.entry3)
+
+        self.assertEqual(species, 'CH3NO2SX')
+        self.assertEqual(formula, {'X': 1, 'C': 1, 'O': 2, 'H': 3, 'N': 1, 'S': 1})
+        self.assertTrue(self.nasa.isIdenticalTo(thermo))
 
 
 class TestReadReactionComments(unittest.TestCase):