Coverage for portality/lib/query_filters.py: 13%
176 statements
« prev ^ index » next coverage.py v6.4.2, created at 2022-07-20 16:12 +0100
1from flask_login import current_user
2from portality.core import app
3from portality import models, constants
4from copy import deepcopy
6# General utilities
7###################
10def remove_fields(query: dict, fields_to_remove: list):
11 q = deepcopy(query)
12 for del_attr in fields_to_remove:
13 if del_attr in q:
14 del q[del_attr]
15 return q
18# query sanitisers
19##################
21def public_query_validator(q):
22 # no deep paging
23 if q.from_result() > 10000:
24 return False
26 if q.size() > 200:
27 return False
29 return True
32# query filters
33###############
35def remove_search_limits(query: dict):
36 return remove_fields(query, ['size', 'from'])
38def only_in_doaj(q):
39 q.clear_match_all()
40 q.add_must_filter({"term": {"admin.in_doaj": True}})
41 return q
44def owner(q):
45 q.clear_match_all()
46 q.add_must_filter({"term" : {"admin.owner.exact" : current_user.id}})
47 return q
50def update_request(q):
51 q.clear_match_all()
52 q.add_must_filter({"range" : {"created_date" : {"gte" : app.config.get("UPDATE_REQUESTS_SHOW_OLDEST")}}})
53 q.add_must_filter({"term" : {"admin.application_type.exact" : constants.APPLICATION_TYPE_UPDATE_REQUEST}})
54 return q
57def not_update_request(q):
58 q.clear_match_all()
59 q.add_must_filter({"term" : {"admin.application_type.exact" : constants.APPLICATION_TYPE_NEW_APPLICATION}})
60 return q
63def associate(q):
64 q.clear_match_all()
65 q.add_must_filter({"term" : {"admin.editor.exact" : current_user.id}})
66 return q
69def editor(q):
70 gnames = []
71 groups = models.EditorGroup.groups_by_editor(current_user.id)
72 for g in groups:
73 gnames.append(g.name)
74 q.clear_match_all()
75 q.add_must({"terms" : {"admin.editor_group.exact" : gnames}})
76 return q
79def private_source(q):
80 q.add_include(["admin.application_status", "admin.ticked",
81 "admin.seal", "last_updated", "created_date", "id", "bibjson"])
82 return q
85def public_source(q):
86 q.add_include(["admin.ticked", "admin.seal", "last_updated",
87 "created_date", "id", "bibjson"])
88 return q
91def strip_facets(q):
92 q.clear_facets()
93 return q
96def es_type_fix(q):
97 # FIXME: document this and will need attention for ES 7 upgrade
98 ctx = q.as_dict()
99 if "query" not in ctx:
100 return q
101 ctx = ctx["query"]
102 if "filtered" not in ctx:
103 return q
104 ctx = ctx["filtered"]
105 if "filter" not in ctx:
106 return q
107 ctx = ctx["filter"]
108 if "bool" not in ctx:
109 return q
110 ctx = ctx["bool"]
111 if "must" not in ctx:
112 return q
113 ctx = ctx["must"]
114 for m in ctx:
115 if "term" in m and "_type" in m["term"]:
116 m["term"] = {"es_type" : m["term"]["_type"]}
117 return q
120def last_update_fallback(q):
121 s = q.sort()
122 if s is None or len(s) == 0:
123 return q
125 add_created_sort = False
126 sort_order = None
127 for sortby in s:
128 if "last_manual_update" in sortby:
129 sort_order = sortby["last_manual_update"].get("order")
130 add_created_sort = True
131 break
133 if add_created_sort:
134 params = {}
135 if sort_order is not None:
136 params["order"] = sort_order
137 s.append({"created_date" : params})
139 q.set_sort(s)
140 return q
143# results filters
144#################
146def public_result_filter(results, unpacked=False):
147 # Dealing with single unpacked result
148 if unpacked:
149 if "admin" in results:
150 for k in list(results["admin"]):
151 if k not in ["ticked", "seal"]:
152 del results["admin"][k]
153 return results
155 # Dealing with a list of es results
156 if "hits" not in results:
157 return results
158 if "hits" not in results["hits"]:
159 return results
161 for hit in results["hits"]["hits"]:
162 if "_source" in hit:
163 if "admin" in hit["_source"]:
164 for k in list(hit["_source"]["admin"]):
165 if k not in ["ticked", "seal"]:
166 del hit["_source"]["admin"][k]
168 return results
171def prune_author_emails(results, unpacked=False):
172 # Dealing with single unpacked ES result
173 if unpacked:
174 if "bibjson" in results:
175 if "author" in results["bibjson"]:
176 for a in results["bibjson"]["author"]:
177 if "email" in a:
178 del a["email"]
179 return results
181 # Dealing with a list of ES results
182 if "hits" not in results:
183 return results
184 if "hits" not in results["hits"]:
185 return results
187 for hit in results["hits"]["hits"]:
188 if "_source" in hit:
189 if "bibjson" in hit["_source"]:
190 if "author" in hit["_source"]["bibjson"]:
191 for a in hit["_source"]["bibjson"]["author"]:
192 if "email" in a:
193 del a["email"]
195 return results
198def publisher_result_filter(results, unpacked=False):
199 allowed_admin = ["ticked", "seal", "in_doaj", "related_applications", "current_application", "current_journal", "application_status"]
200 # Dealing with single unpacked ES result
201 if unpacked:
202 if "admin" in results:
203 for k in list(results["admin"]):
204 if k not in allowed_admin:
205 del results["admin"][k]
206 return results
208 # Dealing with a list of ES results
209 if "hits" not in results:
210 return results
211 if "hits" not in results["hits"]:
212 return results
214 for hit in results["hits"]["hits"]:
215 if "_source" in hit:
216 if "admin" in hit["_source"]:
217 for k in list(hit["_source"]["admin"]):
218 if k not in allowed_admin:
219 del hit["_source"]["admin"][k]
221 return results
224def add_fqw_facets(results, unpacked=False):
225 if unpacked:
226 return results
228 facets = {
229 "index.license.exact": {
230 "_type": "terms",
231 "missing": 0,
232 "total": 0,
233 "other": 0,
234 "terms": []
235 },
236 "bibjson.journal.title.exact": {
237 "_type": "terms",
238 "missing": 0,
239 "total": 0,
240 "other": 0,
241 "terms": []
242 },
243 "bibjson.archiving_policy.policy.exact": {
244 "_type": "terms",
245 "missing": 0,
246 "total": 0,
247 "other": 0,
248 "terms": []
249 },
250 "created_date": {
251 "_type": "date_histogram",
252 "entries": []
253 },
254 "index.country.exact": {
255 "_type": "terms",
256 "missing": 0,
257 "total": 0,
258 "other": 0,
259 "terms": []
260 },
261 "index.date": {
262 "_type": "date_histogram",
263 "entries": []
264 },
265 "bibjson.journal.volume.exact": {
266 "_type": "terms",
267 "missing": 0,
268 "total": 0,
269 "other": 0,
270 "terms": []
271 },
272 "index.publisher.exact": {
273 "_type": "terms",
274 "missing": 0,
275 "total": 0,
276 "other": 0,
277 "terms": []
278 },
279 "index.has_seal.exact": {
280 "_type": "terms",
281 "missing": 0,
282 "total": 0,
283 "other": 0,
284 "terms": []
285 },
286 "index.has_apc.exact": {
287 "_type": "terms",
288 "missing": 0,
289 "total": 0,
290 "other": 0,
291 "terms": []
292 },
293 "bibjson.editorial_review.process.exact": {
294 "_type": "match",
295 "missing": 0,
296 "total": 0,
297 "other": 0,
298 "terms": []
299 },
300 "index.language.exact": {
301 "_type": "match",
302 "missing": 0,
303 "total": 0,
304 "other": 0,
305 "terms": []
306 },
307 "bibjson.journal.number.exact": {
308 "_type": "terms",
309 "missing": 0,
310 "total": 0,
311 "other": 0,
312 "terms": []
313 },
314 "index.date_toc_fv_month": {
315 "_type": "date_histogram",
316 "entries": []
317 },
318 "index.classification.exact": {
319 "_type": "terms",
320 "missing": 0,
321 "total": 0,
322 "other": 0,
323 "terms": []
324 },
325 "_type": {
326 "_type": "terms",
327 "missing": 0,
328 "total": 0,
329 "other": 0,
330 "terms": []
331 },
332 "index.issn.exact": {
333 "_type": "terms",
334 "missing": 0,
335 "total": 0,
336 "other": 0,
337 "terms": []
338 }
339 }
341 results["aggregations"] = facets
342 return results
345def fqw_back_compat(results, unpacked=False):
346 if unpacked:
347 return results
349 # Dealing with a list of ES results
350 if "hits" not in results:
351 return results
352 if "hits" not in results["hits"]:
353 return results
355 for hit in results["hits"]["hits"]:
356 if hit.get("_source").get("es_type") != "journal":
357 continue
359 identifiers = []
360 bj = hit.get("_source", {}).get("bibjson", {})
361 if bj.get("pissn"):
362 identifiers.append({"type" : "pissn", "id" : bj.get("pissn")})
363 if bj.get("eissn"):
364 identifiers.append({"type" : "eissn", "id" : bj.get("eissn")})
366 bj["identifier"] = identifiers
368 return results