Add --extra option to ck2yaml

interfaces/cython/cantera/ck2yaml.py

@@ -13,8 +13,11 @@
             [--transport=<filename>]
             [--surface=<filename>]
             [--name=<name>]
+            [--extra=<filename>]
             [--output=<filename>]
             [--permissive]
+            [--quiet]
+            [--no-validate]
             [-d | --debug]
 
 Example:
@@ -34,6 +37,10 @@
 The '--permissive' option allows certain recoverable parsing errors (e.g.
 duplicate transport data) to be ignored. The '--name=<name>' option
 is used to override default phase names (i.e. 'gas').
+
+The '--extra=<filename>' option takes a YAML file as input. This option can be
+used to add to the file description, or to define custom fields that are
+included in the YAML output.
 """
 
 from collections import defaultdict, OrderedDict
@@ -742,7 +749,8 @@ def __init__(self):
         self.species_dict = {}  # bulk and surface species
         self.surfaces = []
         self.reactions = []
-        self.headerLines = []
+        self.header_lines = []
+        self.extra = {}  # for extra entries
         self.files = []  # input file names
 
     def warn(self, message):
@@ -1328,6 +1336,35 @@ def parse_expression(expression, dest):
 
         return reaction, revReaction
 
+    def load_extra_file(self, path):
+        """
+        Load YAML-formatted entries from ``path`` on disk.
+        """
+        with open(path, 'rt', encoding="utf-8") as stream:
+            yml = yaml.round_trip_load(stream)
+
+        # do not overwrite reserved field names
+        reserved = {'generator', 'input-files', 'cantera-version', 'date',
+                    'units', 'phases', 'species', 'reactions'}
+        reserved &= set(yml.keys())
+        if reserved:
+            raise InputError("The YAML file '{}' provided as '--extra' input "
+                "must not redefine reserved field name: "
+                "'{}'".format(path, reserved))
+
+        # replace header lines
+        if 'description' in yml:
+            if isinstance(yml['description'], str):
+                if self.header_lines:
+                    self.header_lines += ['']
+                self.header_lines += yml.pop('description').split('\n')
+            else:
+                raise InputError("The alternate description provided in "
+                    "'{}' needs to be a string".format(path))
+
+        # remainder
+        self.extra = yml
+
     def load_chemkin_file(self, path, skip_undeclared_species=True, surface=False):
         """
         Load a Chemkin-format input file from ``path`` on disk.
@@ -1353,10 +1390,13 @@ def readline():
             line, comment = readline()
             advance = True
             inHeader = True
+            header = []
+            indent = 80
             while line is not None:
                 tokens = line.split() or ['']
                 if inHeader and not line.strip():
-                    self.headerLines.append(comment.rstrip())
+                    header.append(comment.rstrip())
+                    indent = min(indent, re.search('[^ ]', comment).start())
 
                 if tokens[0].upper().startswith('ELEM'):
                     inHeader = False
@@ -1738,6 +1778,9 @@ def readline():
                 else:
                     advance = True
 
+        for h in header:
+            self.header_lines.append(h[indent:])
+
         self.check_duplicate_reactions()
 
         for index, reaction in enumerate(self.reactions):
@@ -1838,8 +1881,8 @@ def write_yaml(self, name='gas', out_name='mech.yaml'):
                 if surf.reactions:
                     n_reacting_phases += 1
 
-            # header from original file
-            desc = '\n'.join(line.rstrip() for line in self.headerLines)
+            # Write header lines
+            desc = '\n'.join(line.rstrip() for line in self.header_lines)
             desc = desc.strip('\n')
             desc = textwrap.dedent(desc)
             if desc.strip():
@@ -1857,6 +1900,13 @@ def write_yaml(self, name='gas', out_name='mech.yaml'):
                 metadata.yaml_set_comment_before_after_key('generator', before='\n')
             emitter.dump(metadata, dest)
 
+            # Write extra entries
+            if self.extra:
+                extra = BlockMap(self.extra)
+                key = list(self.extra.keys())[0]
+                extra.yaml_set_comment_before_after_key(key, before='\n')
+                emitter.dump(extra, dest)
+
             units = FlowMap([('length', 'cm'), ('time', 's')])
             units['quantity'] = self.output_quantity_units
             units['activation-energy'] = self.output_energy_units
@@ -1946,8 +1996,8 @@ def write_yaml(self, name='gas', out_name='mech.yaml'):
 
     @staticmethod
     def convert_mech(input_file, thermo_file=None, transport_file=None,
-                     surface_file=None, phase_name='gas', out_name=None,
-                     quiet=False, permissive=None):
+                     surface_file=None, phase_name='gas', extra_file=None,
+                     out_name=None, quiet=False, permissive=None):
 
         parser = Parser()
         if quiet:
@@ -2013,6 +2063,19 @@ def convert_mech(input_file, thermo_file=None, transport_file=None,
                                 surface_file, parser.line_number, err))
                 raise
 
+        if extra_file:
+            parser.files.append(extra_file)
+            extra_file = os.path.expanduser(extra_file)
+            if not os.path.exists(extra_file):
+                raise IOError('Missing input file: {0!r}'.format(extra_file))
+            try:
+                # Read input mechanism files
+                parser.load_extra_file(extra_file)
+            except Exception as err:
+                logging.warning("\nERROR: Unable to parse '{0}':\n{1}\n".format(
+                                extra_file, err))
+                raise
+
         if out_name:
             out_name = os.path.expanduser(out_name)
         else:
@@ -2027,15 +2090,16 @@ def convert_mech(input_file, thermo_file=None, transport_file=None,
         return surface_names
 
 
-def convert_mech(input_file, thermo_file=None, transport_file=None, surface_file=None,
-                 phase_name='gas', out_name=None, quiet=False, permissive=None):
+def convert_mech(input_file, thermo_file=None, transport_file=None,
+                 surface_file=None, phase_name='gas', extra_file=None,
+                 out_name=None, quiet=False, permissive=None):
     return Parser.convert_mech(input_file, thermo_file, transport_file, surface_file,
-                               phase_name, out_name, quiet, permissive)
+                               phase_name, extra_file, out_name, quiet, permissive)
 
 def main(argv):
 
     longOptions = ['input=', 'thermo=', 'transport=', 'surface=', 'name=',
-                   'output=', 'permissive', 'help', 'debug', 'quiet',
+                   'extra=', 'output=', 'permissive', 'help', 'debug', 'quiet',
                    'no-validate', 'id=']
 
     try:
@@ -2068,7 +2132,7 @@ def main(argv):
     if '--id' in options:
         phase_name = options.get('--id', 'gas')
         logging.warning("\nFutureWarning: "
-                        "option '--id=...' is superseded by '--name=...'")
+                        "Option '--id=...' will be replaced by '--name=...'")
     else:
         phase_name = options.get('--name', 'gas')
 
@@ -2077,6 +2141,8 @@ def main(argv):
               ' must be provided.\nRun "ck2yaml.py --help" to see usage help.')
         sys.exit(1)
 
+    extra_file = options.get('--extra')
+
     if '--output' in options:
         out_name = options['--output']
         if not out_name.endswith('.yaml') and not out_name.endswith('.yml'):
@@ -2087,8 +2153,8 @@ def main(argv):
         out_name = os.path.splitext(thermo_file)[0] + '.yaml'
 
     surfaces = Parser.convert_mech(input_file, thermo_file, transport_file,
-                                   surface_file, phase_name, out_name,
-                                   quiet, permissive)
+                                   surface_file, phase_name, extra_file,
+                                   out_name, quiet, permissive)
 
     # Do full validation by importing the resulting mechanism
     if not input_file:

interfaces/cython/cantera/test/data/extra.yaml

@@ -0,0 +1,9 @@
+description: |-
+  This is an alternative description.
+
+foo:
+- spam
+- eggs
+
+bar:
+  spam: eggs

interfaces/cython/cantera/test/test_convert.py

@@ -7,10 +7,16 @@
 import cantera as ct
 from cantera import ck2cti, ck2yaml, cti2yaml, ctml2yaml
 
+try:
+    import ruamel_yaml as yaml
+except ImportError:
+    from ruamel import yaml
+
 
 class converterTestCommon:
     def convert(self, inputFile, thermo=None, transport=None,
-                surface=None, output=None, quiet=True, permissive=None):
+                surface=None, output=None, extra=None,
+                quiet=True, permissive=None):
         if output is None:
             output = inputFile[:-4]  # strip '.inp'
         if inputFile is not None:
@@ -21,11 +27,14 @@ def convert(self, inputFile, thermo=None, transport=None,
             transport = pjoin(self.test_data_dir, transport)
         if surface is not None:
             surface = pjoin(self.test_data_dir, surface)
+        if extra is not None:
+            extra = pjoin(self.test_data_dir, extra)
         output = pjoin(self.test_work_dir, output + self.ext)
         if os.path.exists(output):
             os.remove(output)
         self._convert(inputFile, thermo=thermo, transport=transport,
-            surface=surface, output=output, quiet=quiet, permissive=permissive)
+            surface=surface, output=output, extra=extra,
+            quiet=quiet, permissive=permissive)
 
     def checkConversion(self, refFile, testFile):
         ref = ct.Solution(refFile)
@@ -415,8 +424,8 @@ class ck2ctiTest(converterTestCommon, utilities.CanteraTest):
     ext = '.cti'
     InputError = ck2cti.InputParseError
 
-    def _convert(self, inputFile, *, thermo, transport, surface, output, quiet,
-                 permissive):
+    def _convert(self, inputFile, *, thermo, transport, surface, output, extra,
+                 quiet, permissive):
         ck2cti.convertMech(inputFile, thermoFile=thermo,
             transportFile=transport, surfaceFile=surface, outName=output,
             quiet=quiet, permissive=permissive)
@@ -430,11 +439,25 @@ class ck2yamlTest(converterTestCommon, utilities.CanteraTest):
     ext = '.yaml'
     InputError = ck2yaml.InputError
 
-    def _convert(self, inputFile, *, thermo, transport, surface, output, quiet,
-                 permissive):
+    def _convert(self, inputFile, *, thermo, transport, surface, output, extra,
+                 quiet, permissive):
         ck2yaml.convert_mech(inputFile, thermo_file=thermo,
             transport_file=transport, surface_file=surface, out_name=output,
-            quiet=quiet, permissive=permissive)
+            extra_file=extra, quiet=quiet, permissive=permissive)
+
+    def test_extra(self):
+        self.convert('gri30.inp', thermo='gri30_thermo.dat',
+                     transport='gri30_tran.dat', output='gri30_extra',
+                     extra='extra.yaml')
+
+        output = pjoin(self.test_work_dir, 'gri30_extra' + self.ext)
+        with open(output, 'rt', encoding="utf-8") as stream:
+            yml = yaml.safe_load(stream)
+
+        desc = yml['description'].split('\n')[-1]
+        self.assertEqual(desc, 'This is an alternative description.')
+        for key in ['foo', 'bar']:
+            self.assertIn(key, yml.keys())
 
 
 class CtmlConverterTest(utilities.CanteraTest):