Coverage for test / pubmed_processing_test.py: 99%
155 statements
« prev ^ index » next coverage.py v7.13.4, created at 2026-03-25 18:06 +0000
1# SPDX-FileCopyrightText: 2023-2026 Arcangelo Massari <arcangelo.massari@unibo.it>
2# SPDX-FileCopyrightText: 2024 Arianna Moretti <arianna.moretti4@unibo.it>
3#
4# SPDX-License-Identifier: ISC
6import csv
7import os
8import unittest
10from oc_ds_converter.lib.jsonmanager import *
12from oc_ds_converter.pubmed.pubmed_processing import PubmedProcessing
14BASE = os.path.join('test', 'pubmed_processing')
15IOD = os.path.join(BASE, 'iod')
16ALL_CASES_IOD = os.path.join(BASE, 'iod_all_cases')
17WANTED_PMIDS = os.path.join(BASE, 'wanted_pmids.csv')
18WANTED_PMIDS_FOLDER = os.path.join(BASE, 'wanted_pmids')
19JOURNALS_DICT = os.path.join(BASE, 'journals.json')
20DATA = os.path.join(BASE, '40228.csv')
21DATA_TEST_NAMES = os.path.join(BASE, 'CSVFile_iod_test.csv')
22DATA_DIR = BASE
23OUTPUT = os.path.join(BASE, 'meta_input')
24MULTIPROCESS_OUTPUT = os.path.join(BASE, 'multi_process_test')
25ZIP_INPUT = os.path.join(BASE, 'zip_test')
26PUBLISHERS_MAPPING = os.path.join(BASE, 'publishers.csv')
29class TestPubmedProcessing(unittest.TestCase):
30 maxDiff = None
32 def test_csv_creator(self):
33 pubmed_processor = PubmedProcessing(orcid_index=IOD, publishers_filepath_pubmed=None, journals_filepath=None)
34 data = open(DATA, newline="")
35 reader = csv.DictReader(data)
36 output = list()
37 for item in reader:
38 tabular_data = pubmed_processor.csv_creator(item)
39 if tabular_data:
40 output.append(tabular_data)
42 issnstr_1 = "[issn:0006-291X issn:1090-2104]"
43 issnstr_2 = "[issn:1090-2104 issn:0006-291X]"
44 expected_output = [
45 {'id': 'pmid:1 doi:10.1016/0006-2944(75)90147-7',
46 'title': 'Formate assay in body fluids: application in methanol poisoning.',
47 'author': 'A B Makar; K E McMartin; M Palese; T R Tephly',
48 'pub_date': '1975',
49 'venue': 'Biochemical medicine [issn:0006-2944]',
50 'volume': '',
51 'issue': '',
52 'page': '',
53 'type': 'journal article',
54 'publisher': 'Elsevier BV',
55 'editor': ''},
56 {'id': 'pmid:2 doi:10.1016/0006-291x(75)90482-9',
57 'title': 'Delineation of the intimate details of the backbone conformation of pyridine nucleotide coenzymes in aqueous solution.',
58 'author': 'K S Bose; Sarma, R H [orcid:0000-0000-0000-0000]',
59 'pub_date': '1975',
60 'venue': f'Biochemical and biophysical research communications {issnstr_1}',
61 'volume': '',
62 'issue': '',
63 'page': '',
64 'type': 'journal article',
65 'publisher': 'Elsevier BV',
66 'editor': ''},
67 {'id': 'pmid:3 doi:10.1016/0006-291x(75)90498-2',
68 'title': 'Metal substitutions incarbonic anhydrase: a halide ion probe study.',
69 'author': 'R J Smith; R G Bryant',
70 'pub_date': '1975',
71 'venue': f'Biochemical and biophysical research communications {issnstr_1}',
72 'volume': '',
73 'issue': '',
74 'page': '',
75 'type': 'journal article',
76 'publisher': 'Elsevier BV',
77 'editor': ''},
78 {'id': 'pmid:4 doi:10.1016/0006-291x(75)90506-9',
79 'title': 'Effect of chloroquine on cultured fibroblasts: release of lysosomal hydrolases and inhibition of their uptake.',
80 'author': 'U N Wiesmann; S DiDonato; N N Herschkowitz',
81 'pub_date': '1975',
82 'venue': f'Biochemical and biophysical research communications {issnstr_1}',
83 'volume': '',
84 'issue': '',
85 'page': '',
86 'type': 'journal article',
87 'publisher': 'Elsevier BV',
88 'editor': ''},
89 {'id': 'pmid:5 doi:10.1016/0006-291x(75)90508-2',
90 'title': 'Atomic models for the polypeptide backbones of myohemerythrin and hemerythrin.',
91 'author': 'W A Hendrickson; K B Ward',
92 'pub_date': '1975',
93 'venue': f'Biochemical and biophysical research communications {issnstr_1}',
94 'volume': '',
95 'issue': '',
96 'page': '',
97 'type': 'journal article',
98 'publisher': 'Elsevier BV',
99 'editor': ''}
100 ]
101 expected_output_alt= [
102 {'id': 'pmid:1 doi:10.1016/0006-2944(75)90147-7',
103 'title': 'Formate assay in body fluids: application in methanol poisoning.',
104 'author': 'A B Makar; K E McMartin; M Palese; T R Tephly',
105 'pub_date': '1975',
106 'venue': 'Biochemical medicine [issn:0006-2944]',
107 'volume': '',
108 'issue': '',
109 'page': '',
110 'type': 'journal article',
111 'publisher': 'Elsevier BV',
112 'editor': ''},
113 {'id': 'pmid:2 doi:10.1016/0006-291x(75)90482-9',
114 'title': 'Delineation of the intimate details of the backbone conformation of pyridine nucleotide coenzymes in aqueous solution.',
115 'author': 'K S Bose; Sarma, R H [orcid:0000-0000-0000-0000]',
116 'pub_date': '1975',
117 'venue': f'Biochemical and biophysical research communications {issnstr_2}',
118 'volume': '',
119 'issue': '',
120 'page': '',
121 'type': 'journal article',
122 'publisher': 'Elsevier BV',
123 'editor': ''},
124 {'id': 'pmid:3 doi:10.1016/0006-291x(75)90498-2',
125 'title': 'Metal substitutions incarbonic anhydrase: a halide ion probe study.',
126 'author': 'R J Smith; R G Bryant',
127 'pub_date': '1975',
128 'venue': f'Biochemical and biophysical research communications {issnstr_2}',
129 'volume': '',
130 'issue': '',
131 'page': '',
132 'type': 'journal article',
133 'publisher': 'Elsevier BV',
134 'editor': ''},
135 {'id': 'pmid:4 doi:10.1016/0006-291x(75)90506-9',
136 'title': 'Effect of chloroquine on cultured fibroblasts: release of lysosomal hydrolases and inhibition of their uptake.',
137 'author': 'U N Wiesmann; S DiDonato; N N Herschkowitz',
138 'pub_date': '1975',
139 'venue': f'Biochemical and biophysical research communications {issnstr_2}',
140 'volume': '',
141 'issue': '',
142 'page': '',
143 'type': 'journal article',
144 'publisher': 'Elsevier BV',
145 'editor': ''},
146 {'id': 'pmid:5 doi:10.1016/0006-291x(75)90508-2',
147 'title': 'Atomic models for the polypeptide backbones of myohemerythrin and hemerythrin.',
148 'author': 'W A Hendrickson; K B Ward',
149 'pub_date': '1975',
150 'venue': f'Biochemical and biophysical research communications {issnstr_2}',
151 'volume': '',
152 'issue': '',
153 'page': '',
154 'type': 'journal article',
155 'publisher': 'Elsevier BV',
156 'editor': ''}
157 ]
158 data.close()
159 self.assertTrue(output == expected_output or output == expected_output_alt)
161 def test_csv_creator_w_journal_dict(self):
162 pubmed_processor = PubmedProcessing(orcid_index=IOD, journals_filepath=JOURNALS_DICT)
163 data = open(DATA, newline="")
164 reader = csv.DictReader(data)
165 output = list()
166 for item in reader:
167 tabular_data = pubmed_processor.csv_creator(item)
168 if tabular_data:
169 output.append(tabular_data)
170 issnstr_1 = "[issn:0006-291X issn:1090-2104]"
171 issnstr_2 = "[issn:1090-2104 issn:0006-291X]"
172 expected_output = [
173 {'id': 'pmid:1 doi:10.1016/0006-2944(75)90147-7',
174 'title': 'Formate assay in body fluids: application in methanol poisoning.',
175 'author': 'A B Makar; K E McMartin; M Palese; T R Tephly',
176 'pub_date': '1975',
177 'venue': 'Biochemical medicine [issn:0006-2944]',
178 'volume': '',
179 'issue': '',
180 'page': '',
181 'type': 'journal article',
182 'publisher': 'Elsevier BV',
183 'editor': ''},
184 {'id': 'pmid:2 doi:10.1016/0006-291x(75)90482-9',
185 'title': 'Delineation of the intimate details of the backbone conformation of pyridine nucleotide coenzymes in aqueous solution.',
186 'author': 'K S Bose; Sarma, R H [orcid:0000-0000-0000-0000]',
187 'pub_date': '1975',
188 'venue': f'Biochemical and biophysical research communications {issnstr_2}',
189 'volume': '',
190 'issue': '',
191 'page': '',
192 'type': 'journal article',
193 'publisher': 'Elsevier BV',
194 'editor': ''},
195 {'id': 'pmid:3 doi:10.1016/0006-291x(75)90498-2',
196 'title': 'Metal substitutions incarbonic anhydrase: a halide ion probe study.',
197 'author': 'R J Smith; R G Bryant',
198 'pub_date': '1975',
199 'venue': f'Biochemical and biophysical research communications {issnstr_2}',
200 'volume': '',
201 'issue': '',
202 'page': '',
203 'type': 'journal article',
204 'publisher': 'Elsevier BV',
205 'editor': ''},
206 {'id': 'pmid:4 doi:10.1016/0006-291x(75)90506-9',
207 'title': 'Effect of chloroquine on cultured fibroblasts: release of lysosomal hydrolases and inhibition of their uptake.',
208 'author': 'U N Wiesmann; S DiDonato; N N Herschkowitz',
209 'pub_date': '1975',
210 'venue': f'Biochemical and biophysical research communications {issnstr_2}',
211 'volume': '',
212 'issue': '',
213 'page': '',
214 'type': 'journal article',
215 'publisher': 'Elsevier BV',
216 'editor': ''},
217 {'id': 'pmid:5 doi:10.1016/0006-291x(75)90508-2',
218 'title': 'Atomic models for the polypeptide backbones of myohemerythrin and hemerythrin.',
219 'author': 'W A Hendrickson; K B Ward',
220 'pub_date': '1975',
221 'venue': f'Biochemical and biophysical research communications {issnstr_2}',
222 'volume': '',
223 'issue': '',
224 'page': '',
225 'type': 'journal article',
226 'publisher': 'Elsevier BV',
227 'editor': ''
228 }
229 ]
230 expected_output_alt = [
231 {'id': 'pmid:1 doi:10.1016/0006-2944(75)90147-7',
232 'title': 'Formate assay in body fluids: application in methanol poisoning.',
233 'author': 'A B Makar; K E McMartin; M Palese; T R Tephly',
234 'pub_date': '1975',
235 'venue': 'Biochemical medicine [issn:0006-2944]',
236 'volume': '',
237 'issue': '',
238 'page': '',
239 'type': 'journal article',
240 'publisher': 'Elsevier BV',
241 'editor': ''},
242 {'id': 'pmid:2 doi:10.1016/0006-291x(75)90482-9',
243 'title': 'Delineation of the intimate details of the backbone conformation of pyridine nucleotide coenzymes in aqueous solution.',
244 'author': 'K S Bose; Sarma, R H [orcid:0000-0000-0000-0000]',
245 'pub_date': '1975',
246 'venue': f'Biochemical and biophysical research communications {issnstr_1}',
247 'volume': '',
248 'issue': '',
249 'page': '',
250 'type': 'journal article',
251 'publisher': 'Elsevier BV',
252 'editor': ''},
253 {'id': 'pmid:3 doi:10.1016/0006-291x(75)90498-2',
254 'title': 'Metal substitutions incarbonic anhydrase: a halide ion probe study.',
255 'author': 'R J Smith; R G Bryant',
256 'pub_date': '1975',
257 'venue': f'Biochemical and biophysical research communications {issnstr_1}',
258 'volume': '',
259 'issue': '',
260 'page': '',
261 'type': 'journal article',
262 'publisher': 'Elsevier BV',
263 'editor': ''},
264 {'id': 'pmid:4 doi:10.1016/0006-291x(75)90506-9',
265 'title': 'Effect of chloroquine on cultured fibroblasts: release of lysosomal hydrolases and inhibition of their uptake.',
266 'author': 'U N Wiesmann; S DiDonato; N N Herschkowitz',
267 'pub_date': '1975',
268 'venue': f'Biochemical and biophysical research communications {issnstr_1}',
269 'volume': '',
270 'issue': '',
271 'page': '',
272 'type': 'journal article',
273 'publisher': 'Elsevier BV',
274 'editor': ''},
275 {'id': 'pmid:5 doi:10.1016/0006-291x(75)90508-2',
276 'title': 'Atomic models for the polypeptide backbones of myohemerythrin and hemerythrin.',
277 'author': 'W A Hendrickson; K B Ward',
278 'pub_date': '1975',
279 'venue': f'Biochemical and biophysical research communications {issnstr_1}',
280 'volume': '',
281 'issue': '',
282 'page': '',
283 'type': 'journal article',
284 'publisher': 'Elsevier BV',
285 'editor': ''
286 }
287 ]
289 data.close()
290 self.assertTrue(output == expected_output or output == expected_output_alt)
292 def test_orcid_finder(self):
293 pubmed_processor = PubmedProcessing(orcid_index=IOD)
294 orcid_found = pubmed_processor.orcid_finder('10.1016/0006-291x(75)90482-9')
295 expected_output = {'0000-0000-0000-0000': 'sarma, r h'}
296 self.assertEqual(orcid_found, expected_output)
298 def test_get_agents_strings_list(self):
299 pubmed_processor = PubmedProcessing(orcid_index=ALL_CASES_IOD, )
300 agents_list = [{'role': 'author', 'name': 'Arianna Moretti', 'family': '', 'given': ''}, {'role': 'author', 'name': 'S Peroni', 'family': '', 'given': ''}, {'role': 'author', 'name': 'Chiara D Giambattista', 'family': '', 'given': ''}]
301 ag = pubmed_processor.get_agents_strings_list("10.3000/1000000001", agents_list)
302 self.assertEqual(ag[0], ['Moretti, Arianna [orcid:0000-0001-5486-7070]', 'Peroni, Silvio [orcid:0000-0003-0530-4305]', 'Di Giambattista, Chiara [orcid:0000-0001-8665-095X]'] )
304 def test_get_venue_name_with_extended_map(self):
305 item = {
306 "journal": "Biochim Biophys Acta"}
307 pubmed_processor = PubmedProcessing(orcid_index=None, journals_filepath=JOURNALS_DICT)
308 venue_name = pubmed_processor.get_venue_name(item, "59")
309 self.assertEqual(venue_name, "Biochimica et biophysica acta [issn:0006-3002]")
311 def test_get_venue_name_with_extended_map_ISSN(self):
312 item = {
313 "journal": "Biochem Biophys Res Commun"}
314 pubmed_processor = PubmedProcessing(orcid_index=None, journals_filepath=JOURNALS_DICT)
315 venue_name = pubmed_processor.get_venue_name(item, "2")
316 alt_1 = "Biochemical and biophysical research communications [issn:0006-291X issn:1090-2104]"
317 alt_2 = "Biochemical and biophysical research communications [issn:1090-2104 issn:0006-291X]"
318 self.assertTrue(venue_name == alt_1 or venue_name == alt_2)
320 def test_get_venue_name_no_extended_map(self):
321 item = {
322 "journal": "Biochim Biophys Acta"}
323 pubmed_processor = PubmedProcessing(orcid_index=None, journals_filepath=None)
324 venue_name = pubmed_processor.get_venue_name(item, "59")
325 self.assertEqual(venue_name, "Biochimica et biophysica acta [issn:0006-3002]")
327 def test_issn_worker(self):
328 input = 'ISSN 0006-291X'
329 output = list()
330 PubmedProcessing.issn_worker(input, output)
331 expected_output = ['issn:0006-291X']
332 self.assertEqual(output, expected_output)
334 def test_id_worker(self):
335 field_issn = '00062952'
336 field_isbn = ['978-3-905673-82-1']
337 issn_list = list()
338 isbn_list = list()
339 PubmedProcessing.id_worker(field_issn, issn_list, PubmedProcessing.issn_worker)
340 PubmedProcessing.id_worker(field_isbn, isbn_list, PubmedProcessing.isbn_worker)
341 expected_issn_list = ['issn:0006-2952']
342 expected_isbn_list = ['isbn:9783905673821']
343 self.assertEqual((issn_list, isbn_list), (expected_issn_list, expected_isbn_list))
345 def test_find_homonyms(self):
346 ra_list_1 = [{'role': 'author', 'name': 'K S Bose', 'family': '', 'given': ''},
347 {'role': 'author', 'name': 'R H Sarma', 'family': '', 'given': ''}]
348 ra_list_2 = [{'role': 'author', 'name': 'K S Bose', 'family': '', 'given': ''},
349 {'role': 'author', 'name': 'R H Bose', 'family': '', 'given': ''}]
350 ra_list_3 = [{'role': 'author', 'name': 'Anna Maria Rossi', 'family': '', 'given': ''},
351 {'role': 'author', 'name': 'Katia Rossi Bianchi', 'family': '', 'given': ''},
352 {'role': 'author', 'name': 'K Rossi B', 'family': '', 'given': ''}]
353 ra_list_4 = [{'role': 'author', 'name': 'R J Smith', 'family': '', 'given': ''},
354 {'role': 'author', 'name': 'R J Smith Bryant', 'family': '', 'given': ''},
355 {'role': 'author', 'name': 'Ronald Bryant', 'family': '', 'given': ''}]
356 pubmed_processor_ra = PubmedProcessing(orcid_index=IOD, publishers_filepath_pubmed=None, journals_filepath=None)
357 homonyms_l1 = pubmed_processor_ra.find_homonyms(ra_list_1)
358 homonyms_l2 = pubmed_processor_ra.find_homonyms(ra_list_2)
359 homonyms_l3 = pubmed_processor_ra.find_homonyms(ra_list_3)
360 homonyms_l4 = pubmed_processor_ra.find_homonyms(ra_list_4)
362 self.assertEqual(homonyms_l1, {})
363 self.assertCountEqual(homonyms_l2, {'K S Bose': ['R H Bose'], 'R H Bose': ['K S Bose']})
364 self.assertCountEqual(homonyms_l3, {'Anna Maria Rossi': ['Katia Rossi Bianchi', 'K Rossi B'], 'Katia Rossi Bianchi': ['Anna Maria Rossi', 'K Rossi B'], 'K Rossi B': ['Anna Maria Rossi', 'Katia Rossi Bianchi']})
365 self.assertCountEqual(homonyms_l4, {'R J Smith': ['R J Smith Bryant'], 'R J Smith Bryant': ['Ronald Bryant', 'R J Smith'], 'Ronald Bryant': ['R J Smith Bryant']})
367 def test_compute_affinity(self):
368 pubmed_processor_ra_ca = PubmedProcessing(orcid_index=IOD, publishers_filepath_pubmed=None, journals_filepath=None)
369 target_full_names = "Anna Cristiana Cardinali Santelli"
370 ra_list_1 = ["Anna Cardinali", "Anna C. Santelli", "A.Cristiana Santelli Cardinali", "Anna Cristiana Santelli Cardinali", "Anna C S C", "Anna cristiana Santelli CARDINALI"]
371 ra_list_2 = ["Anna Cardinali", "Anna C. Santelli", "A.Cristiana Santelli Cardinali", "Anna C S C", "Anna cristiana Santelli CARDINALI"]
372 ra_list_3 = ["Anna Cardinali", "Anna C. Santelli", "A.Cristiana Santelli Cardinali", "Anna C S C"]
373 ra_list_4 = ["Anna Cardinali", "Anna C. Santelli", "Anna C S C"]
374 ra_list_5 = ["Anna Cardinali", "Anna C S C"]
375 ra_list_6 = ["Filippo C. Moroni", "Silvia C."]
376 ra_list_7 = ["Olga Santelli", " Vincenzo Cardinali"]
377 ra_list_8 = ["Olga Santelli", " Vincenzo Cardinali", "Carla Anna Cardinali Santelli"]
379 self.assertEqual(pubmed_processor_ra_ca.compute_affinity(target_full_names, ra_list_1), "Anna Cristiana Santelli Cardinali")
380 self.assertEqual(pubmed_processor_ra_ca.compute_affinity(target_full_names, ra_list_2), "Anna cristiana Santelli CARDINALI")
381 self.assertEqual(pubmed_processor_ra_ca.compute_affinity(target_full_names, ra_list_3), "A.Cristiana Santelli Cardinali")
382 self.assertEqual(pubmed_processor_ra_ca.compute_affinity(target_full_names, ra_list_4), "Anna C. Santelli")
383 self.assertEqual(pubmed_processor_ra_ca.compute_affinity(target_full_names, ra_list_5), "Anna Cardinali")
384 self.assertEqual(pubmed_processor_ra_ca.compute_affinity(target_full_names, ra_list_6), "")
385 self.assertEqual(pubmed_processor_ra_ca.compute_affinity(target_full_names, ra_list_7), "")
386 self.assertEqual(pubmed_processor_ra_ca.compute_affinity(target_full_names, ra_list_8), "Carla Anna Cardinali Santelli")
389 def test_redis_db(self):
390 pubmed_processor = PubmedProcessing(orcid_index=IOD, journals_filepath=JOURNALS_DICT)
391 inp_ent = {'pmid': '5', 'doi': '10.1016/0006-291x(75)90508-2',
392 'title': 'Atomic models for the polypeptide backbones of myohemerythrin and hemerythrin.',
393 'authors': 'W A Hendrickson, K B Ward', 'year': '1975', 'journal': 'Biochem Biophys Res Commun',
394 'cited_by': '7118409 6768892 2619971 2190210 3380793 20577584 8372226 7012375 856811 678527 33255345 33973855 402092 7012894 1257769 861288 1061139 3681996', 'references': '4882249 5059118 14834145 1056020 5509841'}
395 exp_res = {'id': 'pmid:5 doi:10.1016/0006-291x(75)90508-2', 'title': 'Atomic models for the polypeptide backbones of myohemerythrin and hemerythrin.',
396 'author': 'W A Hendrickson; K B Ward', 'pub_date': '1975', 'venue': 'Biochemical and biophysical research communications [issn:0006-291X issn:1090-2104]',
397 'volume': '', 'issue': '', 'page': '', 'type': 'journal article', 'publisher': 'Elsevier BV', 'editor': ''}
399 tabular_data = pubmed_processor.csv_creator(inp_ent)
400 self.assertEqual(exp_res, tabular_data)
402 inp_ent_fake_doi = {'pmid': '5', 'doi': '10.1016/a_fake_doi',
403 'title': 'Atomic models for the polypeptide backbones of myohemerythrin and hemerythrin.',
404 'authors': 'W A Hendrickson, K B Ward', 'year': '1975', 'journal': 'Biochem Biophys Res Commun',
405 'cited_by': '7118409 6768892 2619971 2190210 3380793 20577584 8372226 7012375 856811 678527 33255345 33973855 402092 7012894 1257769 861288 1061139 3681996', 'references': '4882249 5059118 14834145 1056020 5509841'}
407 tabular_data_no_redis_data = pubmed_processor.csv_creator(inp_ent_fake_doi)
408 self.assertEqual(tabular_data_no_redis_data['id'], 'pmid:5')
410 pubmed_processor.BR_redis.sadd('doi:10.1016/a_fake_doi', 'omid:000101')
412 tabular_data_w_redis_data = pubmed_processor.csv_creator(inp_ent_fake_doi)
413 self.assertEqual(tabular_data_w_redis_data['id'], 'pmid:5 doi:10.1016/a_fake_doi')
415 pubmed_processor.doi_m.storage_manager.delete_storage()
416 pubmed_processor.BR_redis.flushdb()
417 pubmed_processor.RA_redis.flushdb()
419 def test_get_citations(self):
420 inp_ent = {'pmid': '5', 'doi': '10.1016/0006-291x(75)90508-2',
421 'title': 'Atomic models for the polypeptide backbones of myohemerythrin and hemerythrin.',
422 'authors': 'W A Hendrickson, K B Ward', 'year': '1975', 'journal': 'Biochem Biophys Res Commun',
423 'cited_by': '7118409 6768892 2619971 2190210 3380793 20577584 8372226 7012375 856811 678527 33255345 33973855 402092 7012894 1257769 861288 1061139 3681996', 'references': '4882249 5059118 14834145 1056020 5509841'}
425 pubmed_processor = PubmedProcessing(orcid_index=IOD,
426 journals_filepath=JOURNALS_DICT)
428 citation_list = pubmed_processor.get_citations("pmid:5", inp_ent)
429 exp_citation_list = [('pmid:5', 'pmid:1056020'), ('pmid:5', 'pmid:4882249'), ('pmid:5', 'pmid:14834145'), ('pmid:5', 'pmid:5509841'), ('pmid:5', 'pmid:5059118')]
431 self.assertEqual(set(citation_list), set(exp_citation_list))
434 def test_get_citing_pmid(self):
435 inp_ent_meta = {'id': 'pmid:5 doi:10.1016/0006-291x(75)90508-2', 'title': 'Atomic models for the polypeptide backbones of myohemerythrin and hemerythrin.',
436 'author': 'W A Hendrickson; K B Ward', 'pub_date': '1975', 'venue': 'Biochemical and biophysical research communications [issn:0006-291X issn:1090-2104]',
437 'volume': '', 'issue': '', 'page': '', 'type': 'journal article', 'publisher': 'Elsevier BV', 'editor': ''}
439 pubmed_processor = PubmedProcessing(orcid_index=IOD,
440 journals_filepath=JOURNALS_DICT)
442 out_citing_pmid = pubmed_processor.get_citing_pmid((inp_ent_meta))
443 exp_out_citing_pmid = 'pmid:5'
445 self.assertEqual(out_citing_pmid, exp_out_citing_pmid)
448if __name__ == '__main__':
449 unittest.main()