Merge pull request #8 from pierre-24/append_data

Append data to existing file, if any

docs/use.rst

@@ -596,6 +596,7 @@ The output depends on the value of ``-V``, which can be:
       There is no way to change this behavior.
     + By default, the program also include the base tensors calculated in the process.
       The ``-S`` option prevents this (that may be useful in the case of electric field differentiation)
+    + If you want to add results to existing ``molecule_nd.h5`` file, you can use the ```--append`` option.
     + Projection over normal mode of all the geometrical derivatives is requested via the ``-p`` option, but you can also request that the cartesian hessian used to do so is different, with the ``-H`` option (which accepts FCHK and dalton archives with cartesian hessian in it as argument).
 
 

nachos/bake.py

@@ -7,6 +7,7 @@
 
 from qcip_tools import derivatives_g, derivatives
 from qcip_tools.chemistry_files import helpers, PropertyNotDefined, PropertyNotPresent
+from qcip_tools.chemistry_files.chemistry_datafile import ChemistryDataFile, BadChemistryDataFile
 
 import nachos
 from nachos.core import files, baking
@@ -110,6 +111,9 @@ def get_arguments_parser():
         type=treat_romberg_arg,
         help='Bypass detection and force a value in the triangle. Must be of the form `k;m`.')
 
+    arguments_parser.add_argument(
+        '-a', '--append', action='store_true', help='Append to existing H5 file')
+
     return arguments_parser
 
 
@@ -144,8 +148,17 @@ def main():
     storage = files.ComputationalResults(recipe, directory=recipe_directory)
     storage.read(args.data)
 
+    original_cf = None
+    if args.append:
+        with open(args.output) as f:
+            try:
+                original_cf = ChemistryDataFile()
+                original_cf.read(f)
+            except BadChemistryDataFile as e:
+                return exit_failure('Cannot append data to `{}`: {}'.format(args.output, e))
+
     # go and bake
-    baker = baking.Baker(recipe, storage, directory=recipe_directory)
+    baker = baking.Baker(recipe, storage, directory=recipe_directory, original_cf=original_cf)
     only = None
 
     if args.only:

nachos/core/baking.py

@@ -13,6 +13,14 @@ class BadBaking(Exception):
     pass
 
 
+def _equal_molecules_or_raise(mol1, mol2):
+    """Compare two geometries
+    """
+
+    if [a.symbol for a in mol1] != [a.symbol for a in mol2]:
+        raise BadBaking('not the same geometries: atomic symbols are different')
+
+
 class Baker:
     """Baker class to finally perform the numerical differentiation
 
@@ -22,9 +30,11 @@ class Baker:
     :type storage: nachos.core.files.ComputationalResults
     :param directory: working directory
     :type directory: str
+    :param original_cf: Originak chemistry file to append to
+    :type original_cf: qcip_tools.chemistry_files.chemistry_datafile.ChemistryDataFile
     """
 
-    def __init__(self, recipe, storage, directory='.'):
+    def __init__(self, recipe, storage, directory='.', original_cf=None):
         self.recipe = recipe
 
         if not os.path.isdir(directory):
@@ -33,6 +43,14 @@ def __init__(self, recipe, storage, directory='.'):
         self.directory = directory
         self.storage = storage
 
+        if original_cf is not None:
+            _equal_molecules_or_raise(self.recipe.geometry, original_cf.molecule)
+            self.original_cf = original_cf
+        else:
+            self.original_cf = chemistry_datafile.ChemistryDataFile.from_molecule(
+                self.recipe.geometry, 'nachos ND result'
+            )
+
         if self.storage.check() != ([], []):
             raise BadBaking('The storage (h5 file) does not fulfill the recipe!')
 
@@ -73,7 +91,7 @@ def bake(self, only=None, out=sys.stdout, verbosity_level=0, copy_zero_field_bas
             raise BadBaking('no differentiation requested!')
 
         bases.sort(key=lambda a: a[1], reverse=True)
-        f = chemistry_datafile.ChemistryDataFile.from_molecule(self.recipe.geometry, 'nachos ND result')
+        f = self.original_cf
         dof = 3 * len(self.recipe.geometry)
 
         if copy_zero_field_basis:

tests/tests_nachos_bake.py

@@ -44,9 +44,9 @@ def test_bake_gaussian_F(self):
 
         storage = files.ComputationalResults(r, directory=directory)
         storage.read(storage_path)
-        baker = baking.Baker(r, storage, directory=directory)
 
         # with energy:
+        baker = baking.Baker(r, storage, directory=directory)
         cf_with_energy = baker.bake(only=[(derivatives.Derivative(), 3)])
 
         self.assertIn('F', cf_with_energy.derivatives)
@@ -64,6 +64,7 @@ def test_bake_gaussian_F(self):
             electrical_derivatives['FFF']['static'], cf_with_energy.derivatives['FFF']['static'])
 
         # with mu:
+        baker = baking.Baker(r, storage, directory=directory)
         cf_with_mu = baker.bake(only=[(derivatives.Derivative('F'), 2)])
 
         self.assertIn('FF', cf_with_mu.derivatives)
@@ -79,6 +80,7 @@ def test_bake_gaussian_F(self):
             delta=.01)
 
         # with alpha:
+        baker = baking.Baker(r, storage, directory=directory)
         cf_with_alpha = baker.bake(only=[(derivatives.Derivative('FF'), 1)])
 
         self.assertIn('FFF', cf_with_alpha.derivatives)
@@ -89,6 +91,7 @@ def test_bake_gaussian_F(self):
             cf_with_alpha.derivatives['FFF']['static'])
 
         # dynamic
+        baker = baking.Baker(r, storage, directory=directory)
         cf_with_alpha = baker.bake(only=[(derivatives.Derivative('dD'), 1)])
 
         self.assertIn('dDF', cf_with_alpha.derivatives)
@@ -159,9 +162,8 @@ def test_bake_gaussian_G(self):
 
         geometrical_derivatives = fchk.property('geometrical_derivatives')
 
-        baker = baking.Baker(r, storage, directory=directory)
-
         # with energy:
+        baker = baking.Baker(r, storage, directory=directory)
         cf_with_energy = baker.bake(only=[(derivatives.Derivative(), 2)])
 
         self.assertIn('G', cf_with_energy.derivatives)
@@ -175,6 +177,7 @@ def test_bake_gaussian_G(self):
             geometrical_derivatives['GG'], cf_with_energy.derivatives['GG'])
 
         # try with all
+        baker = baking.Baker(r, storage, directory=directory)
         cf = baker.bake()
 
         self.assertIn('G', cf.derivatives)
@@ -212,6 +215,7 @@ def test_bake_gaussian_G(self):
                 math.fabs(ph[i, i]), mwh.frequencies[i] ** 2, places=5)
 
         # now, bake and steal results from zero field
+        baker = baking.Baker(r, storage, directory=directory)
         cf_with_copy = baker.bake(only=[(derivatives.Derivative(), 0)], copy_zero_field_basis=True)
 
         # gradient and hessian ...