GH15943 Fixed defaults for compression in HDF5

doc/source/io.rst

@@ -4066,26 +4066,64 @@ Compression
 +++++++++++
 
 ``PyTables`` allows the stored data to be compressed. This applies to
-all kinds of stores, not just tables.
+all kinds of stores, not just tables. Two parameters are used to
+control compression: ``complevel`` and ``complib``.
+
+``complevel`` specifies if and how hard data is to be compressed.
+              ``complevel=0`` and ``complevel=None`` disables
+              compression and ``0<complevel<10`` enables compression.
+
+``complib`` specifies which compression library to use. If nothing is
+            specified the default library ``zlib`` is used. A
+            compression library usually optimizes for either good
+            compression rates or speed and the results will depend on
+            the type of data. Which type of
+            compression to choose depends on your specific needs and
+            data. The list of supported compression libraries:
+
+             - `zlib <http://zlib.net/>`_: The default compression library. A classic in terms of compression, achieves good compression rates but is somewhat slow.
+             - `lzo <http://www.oberhumer.com/opensource/lzo/>`_: Fast compression and decompression.
+             - `bzip2 <http://bzip.org/>`_: Good compression rates.
+             - `blosc <http://www.blosc.org/>`_: Fast compression and decompression.
+
+             .. versionadded:: 0.20.2
+
+                Support for alternative blosc compressors:
+
+                - `blosc:blosclz <http://www.blosc.org/>`_ This is the
+                  default compressor for ``blosc``
+                - `blosc:lz4
+                  <https://fastcompression.blogspot.dk/p/lz4.html>`_:
+                  A compact, very popular and fast compressor.
+                - `blosc:lz4hc
+                  <https://fastcompression.blogspot.dk/p/lz4.html>`_:
+                  A tweaked version of LZ4, produces better
+                  compression ratios at the expense of speed.
+                - `blosc:snappy <https://google.github.io/snappy/>`_:
+                  A popular compressor used in many places.
+                - `blosc:zlib <http://zlib.net/>`_: A classic;
+                  somewhat slower than the previous ones, but
+                  achieving better compression ratios.
+                - `blosc:zstd <https://facebook.github.io/zstd/>`_: An
+                  extremely well balanced codec; it provides the best
+                  compression ratios among the others above, and at
+                  reasonably fast speed.
+
+             If ``complib`` is defined as something other than the
+             listed libraries a ``ValueError`` exception is issued.
 
-- Pass ``complevel=int`` for a compression level (1-9, with 0 being no
-  compression, and the default)
-- Pass ``complib=lib`` where lib is any of ``zlib, bzip2, lzo, blosc`` for
-  whichever compression library you prefer.
+.. note::
 
-``HDFStore`` will use the file based compression scheme if no overriding
-``complib`` or ``complevel`` options are provided. ``blosc`` offers very
-fast compression, and is my most used. Note that ``lzo`` and ``bzip2``
-may not be installed (by Python) by default.
+   If the library specified with the ``complib`` option is missing on your platform,
+   compression defaults to ``zlib`` without further ado.
 
-Compression for all objects within the file
+Enable compression for all objects within the file:
 
 .. code-block:: python
 
-   store_compressed = pd.HDFStore('store_compressed.h5', complevel=9, complib='blosc')
+   store_compressed = pd.HDFStore('store_compressed.h5', complevel=9, complib='blosc:blosclz')
 
-Or on-the-fly compression (this only applies to tables). You can turn
-off file compression for a specific table by passing ``complevel=0``
+Or on-the-fly compression (this only applies to tables) in stores where compression is not enabled:
 
 .. code-block:: python
 

doc/source/whatsnew/v0.21.0.txt

@@ -46,13 +46,12 @@ Backwards incompatible API changes
 
 - Support has been dropped for Python 3.4 (:issue:`15251`)
 - The Categorical constructor no longer accepts a scalar for the ``categories`` keyword. (:issue:`16022`)
-
 - Accessing a non-existent attribute on a closed :class:`HDFStore` will now
   raise an ``AttributeError`` rather than a ``ClosedFileError`` (:issue:`16301`)
 - :func:`read_csv` now treats ``'null'`` strings as missing values by default (:issue:`16471`)
 - :func:`read_csv` now treats ``'n/a'`` strings as missing values by default (:issue:`16078`)
-
 - :class:`pandas.HDFStore`'s string representation is now faster and less detailed. For the previous behavior, use ``pandas.HDFStore.info()``. (:issue:`16503`).
+- Compression defaults in HDF stores now follow pytable standards. Default is no compression and if ``complib`` is missing and ``complevel`` > 0 ``zlib`` is used (:issue:`15943`)
 
 .. _whatsnew_0210.api:
 

pandas/core/generic.py

@@ -1278,10 +1278,10 @@ def to_hdf(self, path_or_buf, key, **kwargs):
             <http://pandas.pydata.org/pandas-docs/stable/io.html#query-via-data-columns>`__.
 
             Applicable only to format='table'.
-        complevel : int, 0-9, default 0
+        complevel : int, 0-9, default None
             Specifies a compression level for data.
             A value of 0 disables compression.
-        complib : {'zlib', 'lzo', 'bzip2', 'blosc', None}, default None
+        complib : {'zlib', 'lzo', 'bzip2', 'blosc'}, default 'zlib'
             Specifies the compression library to be used.
             As of v0.20.2 these additional compressors for Blosc are supported
             (default if no compressor specified: 'blosc:blosclz'):

pandas/io/pytables.py

@@ -411,10 +411,10 @@ class HDFStore(StringMixin):
             and if the file does not exist it is created.
         ``'r+'``
             It is similar to ``'a'``, but the file must already exist.
-    complevel : int, 0-9, default 0
+    complevel : int, 0-9, default None
             Specifies a compression level for data.
             A value of 0 disables compression.
-    complib : {'zlib', 'lzo', 'bzip2', 'blosc', None}, default None
+    complib : {'zlib', 'lzo', 'bzip2', 'blosc'}, default 'zlib'
             Specifies the compression library to be used.
             As of v0.20.2 these additional compressors for Blosc are supported
             (default if no compressor specified: 'blosc:blosclz'):
@@ -449,12 +449,15 @@ def __init__(self, path, mode=None, complevel=None, complib=None,
                 "complib only supports {libs} compression.".format(
                     libs=tables.filters.all_complibs))
 
+        if complib is None and complevel is not None:
+            complib = tables.filters.default_complib
+
         self._path = _stringify_path(path)
         if mode is None:
             mode = 'a'
         self._mode = mode
         self._handle = None
-        self._complevel = complevel
+        self._complevel = complevel if complevel else 0
         self._complib = complib
         self._fletcher32 = fletcher32
         self._filters = None
@@ -566,11 +569,8 @@ def open(self, mode='a', **kwargs):
         if self.is_open:
             self.close()
 
-        if self._complib is not None:
-            if self._complevel is None:
-                self._complevel = 9
-            self._filters = _tables().Filters(self._complevel,
-                                              self._complib,
+        if self._complevel and self._complevel > 0:
+            self._filters = _tables().Filters(self._complevel, self._complib,
                                               fletcher32=self._fletcher32)
 
         try:

pandas/tests/io/test_pytables.py

@@ -736,6 +736,59 @@ def test_put_compression_blosc(self):
             store.put('c', df, format='table', complib='blosc')
             tm.assert_frame_equal(store['c'], df)
 
+    def test_complibs_default_settings(self):
+        # GH15943
+        df = tm.makeDataFrame()
+
+        # Set complevel and check if complib is automatically set to
+        # default value
+        with ensure_clean_path(self.path) as tmpfile:
+            df.to_hdf(tmpfile, 'df', complevel=9)
+            result = pd.read_hdf(tmpfile, 'df')
+            tm.assert_frame_equal(result, df)
+
+            with tables.open_file(tmpfile, mode='r') as h5file:
+                for node in h5file.walk_nodes(where='/df', classname='Leaf'):
+                    assert node.filters.complevel == 9
+                    assert node.filters.complib == 'zlib'
+
+        # Set complib and check to see if compression is disabled
+        with ensure_clean_path(self.path) as tmpfile:
+            df.to_hdf(tmpfile, 'df', complib='zlib')
+            result = pd.read_hdf(tmpfile, 'df')
+            tm.assert_frame_equal(result, df)
+
+            with tables.open_file(tmpfile, mode='r') as h5file:
+                for node in h5file.walk_nodes(where='/df', classname='Leaf'):
+                    assert node.filters.complevel == 0
+                    assert node.filters.complib is None
+
+        # Check if not setting complib or complevel results in no compression
+        with ensure_clean_path(self.path) as tmpfile:
+            df.to_hdf(tmpfile, 'df')
+            result = pd.read_hdf(tmpfile, 'df')
+            tm.assert_frame_equal(result, df)
+
+            with tables.open_file(tmpfile, mode='r') as h5file:
+                for node in h5file.walk_nodes(where='/df', classname='Leaf'):
+                    assert node.filters.complevel == 0
+                    assert node.filters.complib is None
+
+        # Check if file-defaults can be overridden on a per table basis
+        with ensure_clean_path(self.path) as tmpfile:
+            store = pd.HDFStore(tmpfile)
+            store.append('dfc', df, complevel=9, complib='blosc')
+            store.append('df', df)
+            store.close()
+
+            with tables.open_file(tmpfile, mode='r') as h5file:
+                for node in h5file.walk_nodes(where='/df', classname='Leaf'):
+                    assert node.filters.complevel == 0
+                    assert node.filters.complib is None
+                for node in h5file.walk_nodes(where='/dfc', classname='Leaf'):
+                    assert node.filters.complevel == 9
+                    assert node.filters.complib == 'blosc'
+
     def test_complibs(self):
         # GH14478
         df = tm.makeDataFrame()