fixed the test

Author: keflavich

astroquery/alma/core.py

@@ -31,6 +31,7 @@
     _gen_science_sql, _gen_spec_res_sql, ALMA_DATE_FORMAT
 from . import conf, auth_urls
 from astroquery.utils.commons import ASTROPY_LT_4_1
+from astroquery.exceptions import CorruptDataWarning
 
 __all__ = {'AlmaClass', 'ALMA_BANDS'}
 
@@ -750,7 +751,6 @@ def download_files(self, files, savedir=None, cache=True,
 
             if verify_only:
                 existing_file_length = os.stat(filename).st_size
-                print(f"{filename}: {existing_file_length}")
                 if 'content-length' in check_filename.headers:
                     length = int(check_filename.headers['content-length'])
                     if length == 0:
@@ -761,9 +761,9 @@ def download_files(self, files, savedir=None, cache=True,
                         log.info(f"Found cached file {filename} with size {existing_file_length} < expected "
                                  f"size {length}.  The download should be continued.")
                     elif existing_file_length > length:
-                        print("Issuing a warning")
                         warnings.warn(f"Found cached file {filename} with size {existing_file_length} > expected "
-                                      f"size {length}.  The download is likely corrupted.")
+                                      f"size {length}.  The download is likely corrupted.",
+                                      CorruptDataWarning)
                     else:
                         raise ValueError("It should not be possible to reach this state.")
                 else:

astroquery/alma/tests/test_alma_remote.py

@@ -13,6 +13,7 @@
 from astropy import coordinates
 from astropy import units as u
 
+from astroquery.exceptions import CorruptDataWarning
 from astroquery.utils.commons import ASTROPY_LT_4_1
 from .. import Alma
 
@@ -674,10 +675,10 @@ def test_verify_html_file(alma, caplog):
 
     # download the file
     result = alma.download_files(['https://almascience.nao.ac.jp/dataPortal/member.uid___A001_X1284_X1353.qa2_report.html'])
-    assert result
+    assert 'member.uid___A001_X1284_X1353.qa2_report.html' in result[0]
 
     result = alma.download_files(['https://almascience.nao.ac.jp/dataPortal/member.uid___A001_X1284_X1353.qa2_report.html'], verify_only=True)
-    assert result
+    assert 'member.uid___A001_X1284_X1353.qa2_report.html' in result[0]
     local_filepath = result[0]
     existing_file_length = 66336
     assert f"Found cached file {local_filepath} with expected size {existing_file_length}." in caplog.text
@@ -686,24 +687,25 @@ def test_verify_html_file(alma, caplog):
     with open(local_filepath, 'ab') as fh:
         fh.write(b"Extra Text")
 
-    print("Trying the failed version now")
     caplog.clear()
-    with warnings.catch_warnings(record=True) as ww:
+    length = 66336
+    existing_file_length = length + 10
+    with pytest.warns(expected_warning=CorruptDataWarning,
+            match=f"Found cached file {local_filepath} with size {existing_file_length} > expected size {length}.  The download is likely corrupted."):
         result = alma.download_files(['https://almascience.nao.ac.jp/dataPortal/member.uid___A001_X1284_X1353.qa2_report.html'], verify_only=True)
-        assert result
-        length = 66336
-        existing_file_length = length + 10
-        print(f"WARNING: {str(ww)}")
-        print(f"caplog: {caplog.text}")
-        assert f"Found cached file {local_filepath} with size {existing_file_length} > expected size {length}.  The download is likely corrupted." in str(ww)
+    assert 'member.uid___A001_X1284_X1353.qa2_report.html' in result[0]
 
     # manipulate the file: make it small
     with open(local_filepath, 'wb') as fh:
         fh.write(b"Empty Text")
 
     caplog.clear()
     result = alma.download_files(['https://almascience.nao.ac.jp/dataPortal/member.uid___A001_X1284_X1353.qa2_report.html'], verify_only=True)
-    assert result
+    assert 'member.uid___A001_X1284_X1353.qa2_report.html' in result[0]
     length = 66336
     existing_file_length = 10
     assert f"Found cached file {local_filepath} with size {existing_file_length} < expected size {length}.  The download should be continued." in caplog.text
+
+    # cleanup: we don't want `test_download_html_file` to fail if this test is re-run
+    if os.path.exists(local_filepath):
+        os.remove(local_filepath)

astroquery/exceptions.py

@@ -8,7 +8,7 @@
 __all__ = ['TimeoutError', 'InvalidQueryError', 'RemoteServiceError',
            'TableParseError', 'LoginError', 'ResolverError',
            'NoResultsWarning', 'LargeQueryWarning', 'InputWarning',
-           'AuthenticationWarning', 'MaxResultsWarning']
+           'AuthenticationWarning', 'MaxResultsWarning', 'CorruptDataWarning']
 
 
 class TimeoutError(Exception):
@@ -98,6 +98,14 @@ class MaxResultsWarning(AstropyWarning):
     pass
 
 
+class CorruptDataWarning(AstropyWarning):
+    """
+    Astroquery warning class to be issued when there is a sign that the
+    (partially) downloaded data are corrupt.
+    """
+    pass
+
+
 class EmptyResponseError(ValueError):
     """
     Astroquery error class to be raised when the query returns an empty result