git.iko.soy/searx
on master help sign in
Merge pull request #724 from Athemis/master

[engine] PDBe (Protein Data Bank Europe)

commite2245611d78614555f59d0fe2cd4b94ce0b39b12 authorAdam Tauber <asciimoo@gmail.com> dateThu 13 Oct 2016, 11:19 (10 years ago) parent8b10eb63c58834
4 files changed+226
compare with master
unifiedsplit
AUTHORS.rst+1 searx/engines/pdbe.py+109 searx/settings.yml+7 tests/unit/engines/test_pdbe.py+109
AUTHORS.rst+1
@@ -59,3 +59,4 @@ generally made searx better:
5959- Harry Wood @harry-wood
6060- Thomas Renard @threnard
6161- Pydo `<https://github.com/pydo>`_
62+- Athemis `<https://github.com/Athemis>`_
searx/engines/pdbe.pynew file+109
@@ -0,0 +1,109 @@
1+"""
2+ PDBe (Protein Data Bank in Europe)
3+
4+ @website https://www.ebi.ac.uk/pdbe
5+ @provide-api yes (https://www.ebi.ac.uk/pdbe/api/doc/search.html),
6+ unlimited
7+ @using-api yes
8+ @results python dictionary (from json)
9+ @stable yes
10+ @parse url, title, content, img_src
11+"""
12+
13+from json import loads
14+from flask_babel import gettext
15+
16+categories = ['science']
17+
18+hide_obsolete = False
19+
20+# status codes of unpublished entries
21+pdb_unpublished_codes = ['HPUB', 'HOLD', 'PROC', 'WAIT', 'AUTH', 'AUCO', 'REPL', 'POLC', 'REFI', 'TRSF', 'WDRN']
22+# url for api query
23+pdbe_solr_url = 'https://www.ebi.ac.uk/pdbe/search/pdb/select?'
24+# base url for results
25+pdbe_entry_url = 'https://www.ebi.ac.uk/pdbe/entry/pdb/{pdb_id}'
26+# link to preview image of structure
27+pdbe_preview_url = 'https://www.ebi.ac.uk/pdbe/static/entry/{pdb_id}_deposited_chain_front_image-200x200.png'
28+
29+
30+def request(query, params):
31+
32+ params['url'] = pdbe_solr_url
33+ params['method'] = 'POST'
34+ params['data'] = {
35+ 'q': query,
36+ 'wt': "json" # request response in parsable format
37+ }
38+ return params
39+
40+
41+def construct_body(result):
42+ # set title
43+ title = result['title']
44+
45+ # construct content body
46+ content = """{title}<br />{authors} {journal} <strong>{volume}</strong>&nbsp;{page} ({year})"""
47+
48+ # replace placeholders with actual content
49+ try:
50+ if result['journal']:
51+ content = content.format(
52+ title=result['citation_title'],
53+ authors=result['entry_author_list'][0], journal=result['journal'], volume=result['journal_volume'],
54+ page=result['journal_page'], year=result['citation_year'])
55+ else:
56+ content = content.format(
57+ title=result['citation_title'],
58+ authors=result['entry_author_list'][0], journal='', volume='', page='', year=result['release_year'])
59+ img_src = pdbe_preview_url.format(pdb_id=result['pdb_id'])
60+ except (KeyError):
61+ content = None
62+ img_src = None
63+
64+ # construct url for preview image
65+ try:
66+ img_src = pdbe_preview_url.format(pdb_id=result['pdb_id'])
67+ except (KeyError):
68+ img_src = None
69+
70+ return [title, content, img_src]
71+
72+
73+def response(resp):
74+
75+ results = []
76+ json = loads(resp.text)['response']['docs']
77+
78+ # parse results
79+ for result in json:
80+ # catch obsolete entries and mark them accordingly
81+ if result['status'] in pdb_unpublished_codes:
82+ continue
83+ if hide_obsolete:
84+ continue
85+ if result['status'] == 'OBS':
86+ # expand title to add some sort of warning message
87+ title = gettext('{title}&nbsp;(OBSOLETE)').format(title=result['title'])
88+ superseded_url = pdbe_entry_url.format(pdb_id=result['superseded_by'])
89+
90+ # since we can't construct a proper body from the response, we'll make up our own
91+ msg_superseded = gettext("This entry has been superseded by")
92+ content = '<em>{msg_superseded} \<a href="{url}">{pdb_id}</a></em>'.format(
93+ msg_superseded=msg_superseded,
94+ url=superseded_url,
95+ pdb_id=result['superseded_by'], )
96+
97+ # obsoleted entries don't have preview images
98+ img_src = None
99+ else:
100+ title, content, img_src = construct_body(result)
101+
102+ results.append({
103+ 'url': pdbe_entry_url.format(pdb_id=result['pdb_id']),
104+ 'title': title,
105+ 'content': content,
106+ 'img_src': img_src
107+ })
108+
109+ return results
searx/settings.yml+7
@@ -339,6 +339,13 @@ engines:
339339 disabled : True
340340 shortcut : or
341341
342+ - name : pdbe
343+ engine : pdbe
344+ shortcut : pdb
345+# Hide obsolete PDB entries.
346+# Default is not to hide obsolete structures
347+# hide_obsolete : False
348+
342349 - name : photon
343350 engine : photon
344351 shortcut : ph
tests/unit/engines/test_pdbe.pynew file+109
@@ -0,0 +1,109 @@
1+import mock
2+from collections import defaultdict
3+from searx.engines import pdbe
4+from searx.testing import SearxTestCase
5+
6+
7+class TestPdbeEngine(SearxTestCase):
8+ def test_request(self):
9+ query = 'test_query'
10+ dicto = defaultdict(dict)
11+ params = pdbe.request(query, dicto)
12+ self.assertTrue('url' in params)
13+ self.assertTrue('ebi.ac.uk' in params['url'])
14+ self.assertTrue('data' in params)
15+ self.assertTrue('q' in params['data'])
16+ self.assertTrue(query in params['data']['q'])
17+ self.assertTrue('wt' in params['data'])
18+ self.assertTrue('json' in params['data']['wt'])
19+ self.assertTrue('method' in params)
20+ self.assertTrue(params['method'] == 'POST')
21+
22+ def test_response(self):
23+ self.assertRaises(AttributeError, pdbe.response, None)
24+ self.assertRaises(AttributeError, pdbe.response, [])
25+ self.assertRaises(AttributeError, pdbe.response, '')
26+ self.assertRaises(AttributeError, pdbe.response, '[]')
27+
28+ json = """
29+{
30+ "response": {
31+ "docs": [
32+ {
33+ "citation_title": "X-ray crystal structure of ferric Aplysia limacina myoglobin in different liganded states.",
34+ "citation_year": 1993,
35+ "entry_author_list": [
36+ "Conti E, Moser C, Rizzi M, Mattevi A, Lionetti C, Coda A, Ascenzi P, Brunori M, Bolognesi M"
37+ ],
38+ "journal": "J. Mol. Biol.",
39+ "journal_page": "498-508",
40+ "journal_volume": "233",
41+ "pdb_id": "2fal",
42+ "status": "REL",
43+ "title": "X-RAY CRYSTAL STRUCTURE OF FERRIC APLYSIA LIMACINA MYOGLOBIN IN DIFFERENT LIGANDED STATES"
44+ }
45+ ],
46+ "numFound": 1,
47+ "start": 0
48+ },
49+ "responseHeader": {
50+ "QTime": 0,
51+ "params": {
52+ "q": "2fal",
53+ "wt": "json"
54+ },
55+ "status": 0
56+ }
57+}
58+"""
59+
60+ response = mock.Mock(text=json)
61+ results = pdbe.response(response)
62+ self.assertEqual(type(results), list)
63+ self.assertEqual(len(results), 1)
64+ self.assertEqual(results[0]['title'],
65+ 'X-RAY CRYSTAL STRUCTURE OF FERRIC APLYSIA LIMACINA MYOGLOBIN IN DIFFERENT LIGANDED STATES')
66+ self.assertEqual(results[0]['url'], pdbe.pdbe_entry_url.format(pdb_id='2fal'))
67+ self.assertEqual(results[0]['img_src'], pdbe.pdbe_preview_url.format(pdb_id='2fal'))
68+ self.assertTrue('Conti E' in results[0]['content'])
69+ self.assertTrue('X-ray crystal structure of ferric Aplysia limacina myoglobin in different liganded states.' in
70+ results[0]['content'])
71+ self.assertTrue('1993' in results[0]['content'])
72+
73+ # Testing proper handling of PDB entries marked as obsolete
74+ json = """
75+{
76+ "response": {
77+ "docs": [
78+ {
79+ "citation_title": "Obsolete entry test",
80+ "citation_year": 2016,
81+ "entry_author_list": ["Doe J"],
82+ "journal": "J. Obs.",
83+ "journal_page": "1-2",
84+ "journal_volume": "1",
85+ "pdb_id": "xxxx",
86+ "status": "OBS",
87+ "title": "OBSOLETE ENTRY TEST",
88+ "superseded_by": "yyyy"
89+ }
90+ ],
91+ "numFound": 1,
92+ "start": 0
93+ },
94+ "responseHeader": {
95+ "QTime": 0,
96+ "params": {
97+ "q": "xxxx",
98+ "wt": "json"
99+ },
100+ "status": 0
101+ }
102+}
103+"""
104+ response = mock.Mock(text=json)
105+ results = pdbe.response(response)
106+ self.assertEqual(type(results), list)
107+ self.assertEqual(len(results), 1)
108+ self.assertEqual(results[0]['title'], 'OBSOLETE ENTRY TEST&nbsp;(OBSOLETE)')
109+ self.assertTrue(results[0]['content'].startswith('<em>This entry has been superseded by'))