JPLSpec is even more down

CHANGES.rst

@@ -103,6 +103,7 @@ jplspec
 
 - Refactored to use linelists.core.  Added new ``get_molecule`` method [#3456]
 - Moved to linelists/.  astroquery.jplspec is now deprecated in favor of astroquery.linelists.jplspec [#3455]
+- Fallback added to go to the wayback machine [#3547]
 
 
 linelists.jplspec

astroquery/linelists/jplspec/__init__.py

@@ -15,6 +15,12 @@ class Conf(_config.ConfigNamespace):
         'https://spec.jpl.nasa.gov/cgi-bin/catform',
         'JPL Spectral Catalog URL.')
 
+    ftp_cat_server = _config.ConfigItem(
+        ['https://spec.jpl.nasa.gov/ftp/pub/catalog/',
+         'https://web.archive.org/web/20250630185813/https://spec.jpl.nasa.gov/ftp/pub/catalog/'],
+        'JPL FTP Catalog URL'
+    )
+
     timeout = _config.ConfigItem(
         60,
         'Time limit for connecting to JPL server.')

astroquery/linelists/jplspec/core.py

@@ -27,6 +27,7 @@ class JPLSpecClass(BaseQuery):
 
     # use the Configuration Items imported from __init__.py
     URL = conf.server
+    FTP_CAT_URL = conf.ftp_cat_server
     TIMEOUT = conf.timeout
 
     def __init__(self):
@@ -142,6 +143,7 @@ def query_lines(self, min_frequency, max_frequency, *,
                     parse_name_locally=False,
                     get_query_payload=False,
                     fallback_to_getmolecule=True,
+                    use_getmolecule=True,
                     cache=True):
         """
         Query the JPLSpec service for spectral lines.
@@ -153,6 +155,9 @@ def query_lines(self, min_frequency, max_frequency, *,
         governs whether `get_molecule` will be used when no results are returned
         by the query service.  This workaround is needed while JPLSpec's query
         tool is broken.
+
+        use_getmolecule is an option to force the query to use get_molecule.
+        It is needed if the JPL server is completely unresponsive.
         """
         response = self.query_lines_async(min_frequency=min_frequency,
                                           max_frequency=max_frequency,
@@ -166,11 +171,13 @@ def query_lines(self, min_frequency, max_frequency, *,
         if get_query_payload:
             return response
         else:
-            return self._parse_result(response, fallback_to_getmolecule=fallback_to_getmolecule)
+            return self._parse_result(response, fallback_to_getmolecule=fallback_to_getmolecule,
+                                      use_getmolecule=use_getmolecule)
 
     query_lines.__doc__ = process_asyncs.async_to_sync_docstr(query_lines_async.__doc__)
 
-    def _parse_result(self, response, *, verbose=False, fallback_to_getmolecule=False):
+    def _parse_result(self, response, *, verbose=False, fallback_to_getmolecule=False,
+                      use_getmolecule=True):
         """
         Parse a response into an `~astropy.table.Table`
 
@@ -201,8 +208,8 @@ def _parse_result(self, response, *, verbose=False, fallback_to_getmolecule=Fals
         QN":   Quantum numbers for the lower state.
         """
 
-        if 'Zero lines were found' in response.text:
-            if fallback_to_getmolecule:
+        if 'Zero lines were found' in response.text or use_getmolecule:
+            if fallback_to_getmolecule or use_getmolecule:
                 self.lookup_ids = build_lookup()
                 payload = parse_qs(response.request.body)
                 tbs = [self.get_molecule(mol) for mol in payload['Mol']]
@@ -320,7 +327,7 @@ def get_molecule(self, molecule_id, *, cache=True):
         molecule_str = parse_molid(molecule_id)
 
         # Construct the URL to the catalog file
-        url = f'https://spec.jpl.nasa.gov/ftp/pub/catalog/c{molecule_str}.cat'
+        url = f'{self.FTP_CAT_URL}/c{molecule_str}.cat'
 
         # Request the catalog file
         response = self._request(method='GET', url=url,

astroquery/linelists/jplspec/tests/test_jplspec.py

@@ -23,8 +23,10 @@ def data_path(filename):
 
 class MockResponseSpec:
 
-    def __init__(self, filename):
+    def __init__(self, filename, body=''):
         self.filename = data_path(filename)
+        self.request = Mock()
+        self.request.body = body
 
     @property
     def text(self):
@@ -76,8 +78,8 @@ def test_input_multi():
 
 def test_query():
 
-    response = MockResponseSpec(file1)
-    tbl = JPLSpec._parse_result(response)
+    response = MockResponseSpec(file1, body='Mol=28001 CO')
+    tbl = JPLSpec._parse_result(response, fallback_to_getmolecule=False, use_getmolecule=False)
     assert isinstance(tbl, Table)
     assert len(tbl) == 8
     assert set(tbl.keys()) == set(['FREQ', 'ERR', 'LGINT', 'DR', 'ELO', 'GUP',
@@ -94,8 +96,8 @@ def test_query():
 
 def test_query_truncated():
 
-    response = MockResponseSpec(file2)
-    tbl = JPLSpec._parse_result(response)
+    response = MockResponseSpec(file2, body='Mol=28001 CO')
+    tbl = JPLSpec._parse_result(response, fallback_to_getmolecule=False, use_getmolecule=False)
     assert isinstance(tbl, Table)
     assert len(tbl) == 6
     assert set(tbl.keys()) == set(['FREQ', 'ERR', 'LGINT', 'DR', 'ELO', 'GUP',
@@ -109,8 +111,8 @@ def test_query_truncated():
 
 def test_query_multi():
 
-    response = MockResponseSpec(file3)
-    tbl = JPLSpec._parse_result(response)
+    response = MockResponseSpec(file3, body='Mol=28001 CO')
+    tbl = JPLSpec._parse_result(response, fallback_to_getmolecule=False, use_getmolecule=False)
     assert isinstance(tbl, Table)
     assert len(tbl) == 208
     assert set(tbl.keys()) == set(['FREQ', 'ERR', 'LGINT', 'DR', 'ELO', 'GUP',
@@ -230,7 +232,7 @@ def test_fallback_to_getmolecule_with_empty_response():
 
     # Test with fallback disabled - should raise EmptyResponseError
     with pytest.raises(EmptyResponseError, match="Response was empty"):
-        JPLSpec._parse_result(mock_response, fallback_to_getmolecule=False)
+        JPLSpec._parse_result(mock_response, fallback_to_getmolecule=False, use_getmolecule=False)
 
     # Test with fallback enabled - should call get_molecule
     molecules = {'18003': ('H2O', {'FREQ': [100.0, 200.0]})}

astroquery/linelists/jplspec/tests/test_jplspec_remote.py

@@ -13,7 +13,8 @@ def test_remote():
                               max_frequency=1000 * u.GHz,
                               min_strength=-500,
                               molecule="18003 H2O",
-                              fallback_to_getmolecule=False)
+                              fallback_to_getmolecule=False,
+                              use_getmolecule=False)
     assert isinstance(tbl, Table)
     assert len(tbl) == 36
     assert set(tbl.keys()) == set(['FREQ', 'ERR', 'LGINT', 'DR', 'ELO', 'GUP',
@@ -62,7 +63,8 @@ def test_remote_regex():
                               max_frequency=1000 * u.GHz,
                               min_strength=-500,
                               molecule=("28001", "28002", "28003"),
-                              fallback_to_getmolecule=False)
+                              fallback_to_getmolecule=False,
+                              use_getmolecule=False)
     assert isinstance(tbl, Table)
     assert len(tbl) == 16
     assert set(tbl.keys()) == set(['FREQ', 'ERR', 'LGINT', 'DR', 'ELO', 'GUP',