updated uat file structure and tests

adsdata/file_defs.py

@@ -70,7 +70,7 @@
                            'extra_values': {'link_type': 'DATA', 'property': ['DATA']}, 'multiline': True,
                            'subparts': ['link_sub_type', 'item_count', ['url'], ['title']]}
 data_files['gpn'] = {'path': 'links/gpn/all.links', 'default_value': [], 'multiline': True}
-data_files['uat'] = {'path': 'links/uat/all.links', 'default_value': [], 'multiline': False}
+data_files['uat'] = {'path': 'links/uat/all.links', 'default_value': [], 'multiline': True}
 
 # file properties definitions required to generate metrics for CitationCapture records
 # use dict to hold each input files and their properties and idiosycrasies

adsdata/reader.py

@@ -99,13 +99,13 @@ def read_value_for(self, bibcode):
         # at this point, we have the first line with the bibcode in it
         # roll up possible other values on adjacent lines in file
         value = []
-        if self.file_info['path'] == 'links/gpn/all.links':
+        if 'gpn/' in self.file_info['path'] or 'uat/' in self.file_info['path']:
             value.append("/".join(self._get_rest(current_line).split("\t")))
         else:
             value.append(self._get_rest(current_line))
         current_line = self._readline()
         while self.file_info.get('multiline', False) and (current_line not in [None, '']) and (bibcode == self._get_bibcode(current_line)):
-            if self.file_info['path'] == 'links/gpn/all.links':
+            if 'gpn/' in self.file_info['path'] or 'uat/' in self.file_info['path']:
                 value.append("/".join(self._get_rest(current_line).split("\t")))
             else:
                 value.append(self._get_rest(current_line))

adsdata/tests/data1/config/links/uat/all.links

@@ -1,10 +1,16 @@
-2020AJ....160..175S	antapex	stellar classification	galaxy stellar disks	stellar interiors	zenith distance
-2020ApJ...892..106M	cosmic microwave background radiation	radio continuum emission	subgiant stars	x-ray telescopes	two-point correlation function	galaxy quenching	solid matter physics	comet nuclei	astronomy education	horizon
-2020ApJS..247...19M	f giant stars	origin of the universe	solar coronal radio emission	field stars	photodissociation reactions
-2020ApJS..249....7B	exoplanet evolution	gravitational instability	mesopause	nebulae	supernova evolution	astronomy data visualization	indigenous astronomy
-2021ApJ...909..204F	relativistic binary stars	stellar types
-2021ApJ...911..101D	astronomical sextants	celestial sphere	interstellar magnetic fields	lyman limit systems	m stars	plate tectonics	solar coronal streamers	center to limb observations
-2021ApJ...922..164M	lunar transient phenomena	protogalaxies	rapid stellar oscillations
-2022ApJ...934L...4T	proper motions	two-spectrum binary stars	stellar evolutionary models
-2022ApJ...941...95W	r stars	space observatories	spiral galaxies	three-body problem	virgo supercluster
-2023ApJ...954..108T	dust continuum emission	interstellar synchrotron emission	photoheliographs	protoplanetary nebulae	supergalactic longitude	chemical thermodynamics
+2020AJ....160..175S solar physics/solar motion/antapex	123
+2020AJ....160..175S observational astronomy/astronomical research/classification systems/stellar classification	123
+2020AJ....160..175S galactic and extra galactic astronomy/galaxy physics/galaxy disks/galaxy stellar disks	1523
+2020AJ....160..175S stellar astronomy/stellar physics/stellar structures/stellar interiors	1606
+2020ApJ...892..106M cosmology/origin of the universe/early universe/recombination (cosmology)/cosmic background radiation/cosmic microwave background radiation	322
+2020ApJ...892..106M cosmology/origin of the universe/big bang theory/recombination (cosmology)/cosmic background radiation/cosmic microwave background radiation	322
+2020ApJ...892..106M observational astronomy/astronomical methods/radio astronomy/cosmic noise/cosmic background radiation/cosmic microwave background radiation	322
+2020ApJ...892..106M cosmology/astronomical radiation sources/radio sources/radio continuum emission	5
+2020ApJ...892..106M interstellar medium/interstellar emissions/radio continuum emission	5
+2020ApJ...892..106M stellar astronomy/stellar types/stellar evolutionary types/evolved stars/subgiant stars	1646
+2021ApJ...911..101D observational astronomy/astronomical instrumentation/early astronomical instruments/astronomical sextants	93
+2021ApJ...911..101D observational astronomy/astronomical techniques/astrometry/astronomical coordinate systems/equatorial coordinate system/celestial sphere	467
+2021ApJ...922..164M solar system astronomy/lunar science/selenology/lunar features/lunar transient phenomena	976
+2021ApJ...922..164M galaxies/protogalaxies	573
+2021ApJ...922..164M stellar astronomy/stellar physics/stellar properties/stellar oscillations/rapid stellar oscillations	1363
+2021ApJ...922..164M stellar astronomy/astroseismology/stellar oscillations/rapid stellar oscillations	1363

adsdata/tests/test_process.py

@@ -53,7 +53,12 @@ def test_read(self):
             self.assertEqual(d['pub_openaccess'], {'pub_openaccess': False})
             self.assertEqual(d['refereed'], {'refereed': False})
             self.assertFalse(d['gpn'])
-            self.assertEqual(d['uat'], ['cosmic microwave background radiation', 'radio continuum emission', 'subgiant stars', 'x-ray telescopes', 'two-point correlation function', 'galaxy quenching', 'solid matter physics', 'comet nuclei', 'astronomy education', 'horizon'])
+            self.assertEqual(d['uat'], ['cosmology/origin of the universe/early universe/recombination (cosmology)/cosmic background radiation/cosmic microwave background radiation/322',
+                                        'cosmology/origin of the universe/big bang theory/recombination (cosmology)/cosmic background radiation/cosmic microwave background radiation/322',
+                                        'observational astronomy/astronomical methods/radio astronomy/cosmic noise/cosmic background radiation/cosmic microwave background radiation/322',
+                                        'cosmology/astronomical radiation sources/radio sources/radio continuum emission/5',
+                                        'interstellar medium/interstellar emissions/radio continuum emission/5',
+                                        'stellar astronomy/stellar types/stellar evolutionary types/evolved stars/subgiant stars/1646'])
 
         with Processor(compute_metrics=False) as processor, patch('adsputils.load_config', return_value={'INPUT_DATA_ROOT': './test/data1/config/'}):
             d = processor._read_next_bibcode('2024Icar..40815837G')

adsdata/tests/test_reader.py

@@ -336,6 +336,10 @@ def test_planetary_names(self):
 
     def test_uat(self):
         f = reader.NonbibFileReader('uat', data_files['uat'])
-        self.assertEqual({'uat': ['astronomical sextants', 'celestial sphere', 'interstellar magnetic fields', 'lyman limit systems', 'm stars', 'plate tectonics', 'solar coronal streamers', 'center to limb observations']}, f.read_value_for('2021ApJ...911..101D'))
-        self.assertEqual({'uat': ['lunar transient phenomena', 'protogalaxies', 'rapid stellar oscillations']}, f.read_value_for('2021ApJ...922..164M'))
+        self.assertEqual({'uat': ['observational astronomy/astronomical instrumentation/early astronomical instruments/astronomical sextants/93',
+                                  'observational astronomy/astronomical techniques/astrometry/astronomical coordinate systems/equatorial coordinate system/celestial sphere/467']}, f.read_value_for('2021ApJ...911..101D'))
+        self.assertEqual({'uat': ['solar system astronomy/lunar science/selenology/lunar features/lunar transient phenomena/976',
+                                    'galaxies/protogalaxies/573',
+                                    'stellar astronomy/stellar physics/stellar properties/stellar oscillations/rapid stellar oscillations/1363',
+                                    'stellar astronomy/astroseismology/stellar oscillations/rapid stellar oscillations/1363',]}, f.read_value_for('2021ApJ...922..164M'))