Merge pull request #2490 from keflavich/alma_downloader_Aug11_2022

BUG: Fix for broken alma downloader

Author: bsipocz

CHANGES.rst

@@ -31,6 +31,7 @@ alma
 - Fixed a regression to handle arrays of string input for the ``query`` methods. [#2094]
 - Throws an error when an unsupported ``kwargs`` (or argument) is passed in to a function. [#2475]
 - New DataLink API handling. [#2493]
+- Fixed bug #2489 in which blank URLs were being sent to the downloader [#2490]
 
 
 astrometry.net

astroquery/alma/core.py

@@ -828,7 +828,15 @@ def retrieve_data_from_uid(self, uids, *, cache=True):
             raise TypeError("Datasets must be given as a list of strings.")
 
         files = self.get_data_info(uids)
-        file_urls = files['access_url']
+        # filter out blank access URLs
+        # it is possible for there to be length-1 lists
+        if len(files) == 1:
+            file_urls = files['access_url']
+            if isinstance(file_urls, str) and file_urls == '':
+                raise ValueError(f"Cannot download uid {uid} because it has no file")
+        else:
+            file_urls = [url for url in files['access_url'] if url]
+
         totalsize = files['content_length'].sum()*u.B
 
         # each_size, totalsize = self.data_size(files)

astroquery/alma/tests/test_alma_remote.py

@@ -149,7 +149,18 @@ def test_data_proprietary(self, alma):
         with pytest.raises(AttributeError):
             alma.is_proprietary('uid://NON/EXI/STING')
 
-    @pytest.mark.xfail(reason="Depends on PR 2438 (https://github.com/astropy/astroquery/pull/2438)")
+    def test_retrieve_data(self, temp_path, alma):
+        """
+        Regression test for issue 2490 (the retrieval step will simply fail if
+        given a blank line, so all we're doing is testing that it runs)
+        """
+        alma.cache_location = temp_path
+
+        # small solar TP-only data set (<1 GB)
+        uid = 'uid://A001/X87c/X572'
+
+        alma.retrieve_data_from_uid([uid])
+
     def test_data_info(self, temp_dir, alma):
         alma.cache_location = temp_dir