Compat, docs and dump chmod support for data frame formatter.

docs/contrib/index.rst

@@ -43,6 +43,7 @@ The following example shows how a package (e.g. the :py:mod:`~law.docker` packag
    matplotlib
    mercurial
    numpy
+   pandas
    profiling
    pyarrow
    rich

docs/contrib/pandas.rst

@@ -0,0 +1,13 @@
+pandas
+======
+
+.. automodule:: law.pandas
+
+.. contents::
+
+
+Class ``DataFrameFormatter``
+----------------------------
+
+.. autoclass:: DataFrameFormatter
+   :members:

law/cli/completion.sh

@@ -190,7 +190,7 @@ _law_complete() {
     # complete the "location" subcommand
     elif [ "${sub_cmd}" = "location" ]; then
         local words="help"
-        local contribs="arc awkward cms coffea docker dropbox gfal git glite hdf5 htcondor ipython keras lsf matplotlib mercurial numpy profiling pyarrow rich root singularity slack slurm tasks telegram tensorflow wlcg"
+        local contribs="arc awkward cms coffea docker dropbox gfal git glite hdf5 htcondor ipython keras lsf matplotlib mercurial numpy pandas profiling pyarrow rich root singularity slack slurm tasks telegram tensorflow wlcg"
         local inp="${cur##-}"
         inp="${inp##-}"
         COMPREPLY=( $( compgen -W "$( echo ${words} )" -P "--" -- "${inp}" ) $( compgen -W "$( echo ${contribs} )" -- "${inp}" ) )

law/contrib/pandas/formatter.py

@@ -7,10 +7,12 @@
 __all__ = ["DataFrameFormatter"]
 
 
-import pathlib
+import os
+
 from law.target.formatter import Formatter
 from law.target.file import get_path
 from law.logger import get_logger
+from law.util import no_value
 
 
 logger = get_logger(__name__)
@@ -29,6 +31,7 @@ def accepts(cls, path, mode):
     @classmethod
     def load(cls, path, *args, **kwargs):
         path = get_path(path)
+
         if path.endswith(".csv"):
             import pandas  # fmt: skip
             return pandas.read_csv(path, *args, **kwargs)
@@ -49,31 +52,38 @@ def load(cls, path, *args, **kwargs):
             import pandas  # fmt: skip
             return pandas.read_pickle(path, *args, **kwargs)
 
-        suffix = pathlib.Path(path).suffix
+        suffix = os.path.splitext(path)[1]
         raise NotImplementedError(
-            f'Suffix "{suffix}" not implemented in DataFrameFormatter',
+            "suffix \"{}\" not implemented in DataFrameFormatter".format(suffix),
         )
 
     @classmethod
     def dump(cls, path, obj, *args, **kwargs):
-        path = get_path(path)
+        _path = get_path(path)
+        perm = kwargs.pop("perm", no_value)
 
-        if path.endswith(".csv"):
-            return obj.to_csv(path, *args, **kwargs)
+        if _path.endswith(".csv"):
+            ret = obj.to_csv(_path, *args, **kwargs)
 
-        if path.endswith(".json"):
-            return obj.to_json(path, *args, **kwargs)
+        elif _path.endswith(".json"):
+            ret = obj.to_json(_path, *args, **kwargs)
 
-        if path.endswith(".parquet"):
-            return obj.to_parquet(path, *args, **kwargs)
+        elif _path.endswith(".parquet"):
+            ret = obj.to_parquet(_path, *args, **kwargs)
 
-        if path.endswith((".h5", ".hdf5")):
-            return obj.to_hdf(path, *args, **kwargs)
+        elif _path.endswith((".h5", ".hdf5")):
+            ret = obj.to_hdf(_path, *args, **kwargs)
 
-        if path.endswith((".pickle", ".pkl")):
-            return obj.to_pickle(path, *args, **kwargs)
+        elif _path.endswith((".pickle", ".pkl")):
+            ret = obj.to_pickle(_path, *args, **kwargs)
 
-        suffix = pathlib.Path(path).suffix
-        raise NotImplementedError(
-            f'Suffix "{suffix}" not implemented in DataFrameFormatter',
-        )
+        else:
+            suffix = os.path.splitext(path)[1]
+            raise NotImplementedError(
+                "suffix \"{}\" not implemented in DataFrameFormatter".format(suffix),
+            )
+
+        if perm != no_value:
+            cls.chmod(path, perm)
+
+        return ret