Add a test for _store_data

sapphire/tests/test_data/README.md

@@ -0,0 +1,40 @@
+How to create the test data
+===========================
+
+
+Notes on recreating publicdb.h5 and publicdb_src.h5
+---------------------------------------------------
+
+This test data is simply 5 minutes of real data from station 501 downloaded
+from the raw data via the Public Database. Data from 2016-4-21 is used because
+it contains many types of data: comparator, config, events, singles, and
+weather. Because the download script downloads all blobs for the day, those
+need to be removed after downloading, and the file repacked otherwise the file
+remains large. This can be done using the following script:
+
+    >>> import tables
+    >>> from datetime import datetime
+    >>> from sapphire.publicdb import download_data
+    >>> with tables.open_file('publicdb_temp.h5', 'w') as data:
+    ...     download_data(data, '/station_501', 501, datetime(2016, 4, 21),
+    ...                   datetime(2016, 4, 21, 0, 1, 30), get_blobs=True)
+    ...     max_blobs_id = data.root.s501.events.col('traces').max()
+    ...     data.root.s501.blobs.truncate(max_blobs_id + 1)
+
+    $ ptrepack --complevel 9 --complib blosc publicdb_temp.h5 publicdb_src.h5
+    $ rm publicdb_temp.h5
+    $ cp publicdb_src.h5 publicdb_src_tmp.h5
+
+    >>> import tables
+    >>> from datetime import datetime
+    >>> from sapphire.publicdb import _store_data
+    >>> f = tables.Filters(complevel=1)
+    >>> with tables.open_file('publicdb.h5', 'w', filters=f) as data:
+    ...     _store_data(data, '/s501', 'publicdb_src_tmp.h5',
+    ...                 datetime(2016, 4, 21), datetime(2016, 4, 21, 0, 1))
+
+
+Other data
+----------
+
+The other data is created by the `esd_load_data.py` script one level up.

sapphire/tests/test_publicdb.py

@@ -1,11 +1,21 @@
 import unittest
 from datetime import datetime
 import logging
+import shutil
+import os
 
 from mock import sentinel, Mock, patch
 import tables
 
 from sapphire import publicdb
+from sapphire.tests.validate_results import validate_results
+
+from esd_load_data import create_tempfile_path
+
+
+self_path = os.path.dirname(__file__)
+test_data_src_path = os.path.join(self_path, 'test_data/publicdb_src.h5')
+test_data_path = os.path.join(self_path, 'test_data/publicdb.h5')
 
 
 class DownloadDataTest(unittest.TestCase):
@@ -40,10 +50,19 @@ def test_download_data(self, mock_server, mock_retrieve, mock_store):
                           sentinel.group, sentinel.station_id, start,
                           end, get_blobs=sentinel.blobs)
 
-    @unittest.skip('WIP')
     def test__store_data(self):
-        pass
-        # publicdb._store_data(dst_file, dst_group, src_filename, t0, t1)
+        # store data removes the source data when completed, so use a temp
+        tmp_src_path = create_tempfile_path()
+        shutil.copy(test_data_src_path, tmp_src_path)
+
+        output_path = create_tempfile_path()
+        start = datetime(2016, 4, 21)
+        end = datetime(2016, 4, 21, 0, 1)
+        filters = tables.Filters(complevel=1)
+        with tables.open_file(output_path, 'w', filters=filters) as datafile:
+            publicdb._store_data(datafile, '/s501', tmp_src_path, start, end)
+        validate_results(self, test_data_path, output_path)
+        os.remove(output_path)
 
     def test_datetimerange(self):
         combinations = [