Merge pull request #48 from TREX-CoE/improve-overwriting

Improve force overwriting for existing output files

.github/workflows/test.yml

@@ -45,6 +45,9 @@ jobs:
           trexio convert-to -t cartesian -o trexio_orca_h2o.h5 trexio_orca_h2o_sph.h5
           trexio convert-to -t cartesian -o trexio_pyscf_h2o.h5 trexio_pyscf_h2o_sph.h5
           echo "==  Done conversion  =="
+          # dummy checks for overwriting
+          trexio convert-from -t gaussian  -i data/chbrclf.log -b hdf5 -w False trexio_gaussi.h5 || echo "Failure as it should"
+          trexio convert-from -t gaussian  -i data/chbrclf.log -b hdf5 -w True  trexio_gaussi.h5 && echo "Success as it should"
           echo "=== Check TREXIO file converted from GAMESS ==="
           trexio check-mos -n 50 trexio_gamess.h5 > mos-res
           grep "Norm of the error" < mos-res | grep -Eo "([0-9]+\.[0-9]*|\.?[0-9]+)([eE][+-][0-9]+)?" > error-res

src/trexio_tools/converters/convert_back_end.py

@@ -88,10 +88,7 @@ def run_converter(filename_from, filename_to, back_end_to, back_end_from=None, o
     """The high-level converter function."""
 
     import os
-    try:
-        import trexio
-    except ImportError as exc:
-        raise ImportError("trexio Python package is not installed.") from exc
+    import trexio
 
     try:
         # For proper python package
@@ -115,6 +112,9 @@ def run_converter(filename_from, filename_to, back_end_to, back_end_from=None, o
         else:
             raise ValueError(f'Output file {filename_to} already exists. Consider using the `-w` argument.')
 
+    if back_end_from is None:
+        back_end_from = trexio.TREXIO_AUTO
+
     with trexio.File(filename_from, 'r', back_end_from) as trexio_file_from:
         with trexio.File(filename_to, 'w', back_end_to) as trexio_file_to:
             data_handler(trexio_file_from, trexio_file_to)

src/trexio_tools/converters/convert_from.py

@@ -11,10 +11,7 @@
 from .orca_to_trexio import orca_to_trexio as run_orca
 from .crystal_to_trexio import crystal_to_trexio as run_crystal
 
-try:
-    import trexio
-except ImportError as exc:
-    raise ImportError("trexio Python package is not installed.") from exc
+import trexio
 
 try:
     from resultsFile import getFile, a0, get_lm
@@ -759,13 +756,6 @@ def run_molden(trexio_file, filename, normalized_basis=True, multiplicity=None,
 
 def run(trexio_filename, filename, filetype, back_end, spin=None, motype=None):
 
-    if os.path.exists(trexio_filename):
-        print(f"TREXIO file {trexio_filename} already exists and will be removed before conversion.")
-        if back_end == trexio.TREXIO_HDF5:
-            os.remove(trexio_filename)
-        else:
-            raise NotImplementedError(f"Please remove the {trexio_filename} directory manually.")
-
     if "pyscf" not in filetype.lower():
         trexio_file = trexio.File(trexio_filename, mode='w', back_end=back_end)
 

src/trexio_tools/converters/convert_to.py

@@ -9,11 +9,7 @@
 from . import cart_sphe as cart_sphe
 import numpy as np
 
-try:
-    import trexio
-except:
-    print("Error: The TREXIO Python library is not installed")
-    sys.exit(1)
+import trexio
 
 """
 Converter from trexio to fcidump
@@ -606,11 +602,8 @@ def run(trexio_filename, filename, filetype, spin_order):
         run_spherical(trexio_file, filename)
 #    elif filetype.lower() == "normalized_aos":
 #        run_normalized_aos(trexio_file, filename)
-#    elif filetype.lower() == "gamess":
-#        run_resultsFile(trexio_file, filename)
     elif filetype.lower() == "fcidump":
         run_fcidump(trexio_file, filename, spin_order)
-#    elif filetype.lower() == "molden":
     else:
         raise NotImplementedError(f"Conversion from TREXIO to {filetype} is not supported.")
 

src/trexio_tools/converters/orca_to_trexio.py

@@ -28,12 +28,6 @@ def orca_to_trexio(
     with open(orca_json, 'r') as f:
         data = json.load(f)
 
-    import os
-    try:
-        os.remove(filename)
-    except:
-        print(f"File {filename} does not exist.")
-
     # trexio back end handling
     if back_end.lower() == "hdf5":
         trexio_back_end = trexio.TREXIO_HDF5

src/trexio_tools/trexio_run.py

@@ -6,7 +6,7 @@
       trexio check-basis      [-n N_POINTS]  [-b BACK_END]  TREXIO_FILE
       trexio check-mos        [-n N_POINTS]  [-b BACK_END]  TREXIO_FILE
       trexio convert-to       -t TYPE -o OUTPUT_FILE [-y SPIN_ORDER]  TREXIO_FILE
-      trexio convert-from     -t TYPE -i INPUT_FILE  [-b BACK_END]  [-x MO_TYPE]  [-m MULTIPLICITY]  TREXIO_FILE
+      trexio convert-from     -t TYPE -i INPUT_FILE  [-b BACK_END]  [-x MO_TYPE]  [-m MULTIPLICITY]  [-w OVERWRITE]  TREXIO_FILE
       trexio convert-backend  -i INPUT_FILE  -o OUTPUT_FILE  -b BACK_END  -j TREX_JSON_FILE  [-s BACK_END_FROM]  [-w OVERWRITE]
       trexio (-h | --help)
 
@@ -19,7 +19,7 @@
       -s, --back_end_from=BACK_END  [hdf5 | text | auto]  The input TREXIO back end.  [default: auto]
       -m, --multiplicity=MULTIPLICITY  Spin multiplicity for the Crystal converter.
       -j, --json=TREX_JSON_FILE     TREX configuration file (in JSON format).
-      -w, --overwrite=OVERWRITE     Overwrite flag for the conversion of back ends.  [default: True]
+      -w, --overwrite=OVERWRITE     Overwrite the output TREXIO file if it already exists.  [default: True]
       -t, --type=TYPE               [gaussian | gamess | pyscf | orca | crystal | fcidump | molden | cartesian ] File format.
       -x, --motype=MO_TYPE          Type of the molecular orbitals. For example, GAMESS has RHF, MCSCF, GUGA, and Natural as possible MO types.
       -y, --spin_order=TYPE         [block | interleave] How to organize spin orbitals when converting to FCIDUMP [default: block]
@@ -30,24 +30,46 @@
 import os
 
 
-def main(filename=None, args=None):
+def remove_trexio_file(filename:str, overwrite:bool) -> None:
+    """Remove the TREXIO file/directory if it exists."""
+    if os.path.exists(filename):
+        if overwrite:
+            # dummy check
+            if '*' in filename:
+                raise ValueError(f'TREXIO filename {filename} contains * symbol. Are you sure?')
+            # check that the file is actually TREXIO file
+            try:
+                is_trexio = False
+                with trexio.File(filename, 'r', trexio.TREXIO_AUTO) as tfile:
+                    if trexio.has_metadata_package_version(tfile):
+                        is_trexio = True
+
+                if is_trexio: os.system(f'rm -rf -- {filename}')
+
+            except:
+                raise Exception(f'Output file {filename} exists but it is not a TREXIO file. Are you sure?')
+        else:
+            raise Exception(f'Output file {filename} already exists but overwrite option is not provided. Consider using the `-w` CLI argument.')
+
+    return
+
+
+def main(filename=None, args=None) -> None:
     """Main entry point"""
 
     if filename is None and args is None:
         args = docopt(__doc__)
         filename = args["TREXIO_FILE"]
 
-    if args["--n_points"]:
-       n_points = int(args["--n_points"])
-    else:
-       n_points = 81
+    n_points = int(args["--n_points"]) if args["--n_points"] else 81
+
+    overwrite = not str(args["--overwrite"]).lower() in ['false', '0']
 
     if args["convert-backend"]:
-        if str(args["--overwrite"]).lower() in ['false', '0']:
-            overwrite = False
-        else:
-            overwrite = True
-            print(f'File {args["--output"]} will be overwritten.')
+        remove_trexio_file(args["--output"], overwrite)
+
+    if args["convert-from"]:
+        remove_trexio_file(args["TREXIO_FILE"], overwrite)
 
     if args["--back_end"]:
         if str(args["--back_end"]).lower() == "hdf5":
@@ -74,9 +96,7 @@ def main(filename=None, args=None):
     else:
         back_end_from = trexio.TREXIO_AUTO
 
-    spin = None
-    if args["--multiplicity"]:
-        spin = int(args["--multiplicity"])
+    spin = int(args["--multiplicity"]) if args["--multiplicity"] else None
 
     if args["check-basis"]:
         trexio_file = trexio.File(filename, 'r', back_end=back_end)
@@ -109,7 +129,6 @@ def main(filename=None, args=None):
             args["--output"],
             back_end_to=back_end,
             back_end_from=back_end_from,
-            overwrite=overwrite,
             json_filename=args["--json"]
             )