Coverage for portality/view/doaj.py: 65%

1import json

2import os.path

3import re

4import urllib.error

5import urllib.parse

6import urllib.request

8from flask import Blueprint, request, make_response

9from flask import render_template, abort, redirect, url_for, send_file, jsonify

10from flask_login import current_user, login_required

12from portality.ui import exceptions as ui_exceptions

13from portality.bll import exceptions, exceptions as bll_exceptions

14from portality import constants

15from portality import dao

16from portality import models

17from portality import store

18from portality.bll import DOAJ

19from portality.core import app

20from portality.decorators import ssl_required, api_key_required

21from portality.lcc import lcc_jstree

22from portality.lib import plausible

23from portality.ui.messages import Messages

24from portality.ui import templates

25from portality.bll import DOAJ

27# ~~DOAJ:Blueprint~~

28blueprint = Blueprint('doaj', __name__)

31@blueprint.route("/")

32def home():

33 news = models.News.latest(app.config.get("FRONT_PAGE_NEWS_ITEMS", 5))

34 recent_journals = models.Journal.recent(max=16)

35 stats = DOAJ.siteService().site_statistics()

36 return render_template(templates.PUBLIC_INDEX, news=news, recent_journals=recent_journals, statistics=stats)

39@blueprint.route('/login/')

40def login():

41 return redirect(url_for('account.login'))

44@blueprint.route("/dismiss_site_note")

45def dismiss_site_note():

46 cont = request.values.get("continue")

47 if cont is not None:

48 resp = redirect(cont)

49 else:

50 resp = make_response()

51 # set a cookie that lasts for one year

52 resp.set_cookie(app.config.get("SITE_NOTE_KEY"), app.config.get("SITE_NOTE_COOKIE_VALUE"),

53 max_age=app.config.get("SITE_NOTE_SLEEP"), samesite=None, secure=True)

54 return resp

57@blueprint.route("/news/")

58def news():

59 # NOTE: On live this is also handled by the nginx redirect map, but this will strip those with parameters supplied

60 return redirect("https://blog.doaj.org")

63@blueprint.route("/fqw_hit", methods=['POST'])

64def fqw_hit():

65 page = request.form.get('embedding_page')

66 if page is not None:

67 plausible.send_event(app.config.get('ANALYTICS_CATEGORY_FQW', 'FQW'),

68 action=app.config.get('ANALYTICS_ACTION_FQW', 'hit'),

69 label=request.form.get('embedding_page'))

71 # No content response, whether data there or not.

72 return '', 204

75@blueprint.route("/search/journals", methods=["GET"])

76def journals_search():

77 return render_template(templates.PUBLIC_JOURNAL_SEARCH, lcc_tree=lcc_jstree)

80@blueprint.route("/search/articles", methods=["GET"])

81def articles_search():

82 return render_template(templates.PUBLIC_ARTICLE_SEARCH, lcc_tree=lcc_jstree)

85@blueprint.route("/search", methods=['GET'])

86def search():

87 # If there are URL params, check if we need to redirect to articles rather than journals

88 if request.values:

89 # Flat search the query params as string so we don't have to traverse all the way down the decoded json.

90 if re.search(r'\"_type\"\s*:\s*\"article\"', request.values.get('source', '')):

91 return redirect(url_for("doaj.articles_search"), 301)

92 return redirect(url_for("doaj.journals_search"), 301)

95@blueprint.route("/search", methods=['POST'])

96def search_post():

97 """ Redirect a query from the box on the index page to the search page. """

98 if request.form.get('origin') != 'ui':

99 abort(400) # bad request - we must receive searches from our own UI

100

101 ref = request.form.get("ref")

102 if ref is None:

103 abort(400) # Referrer is required

104

105 ct = request.form.get("content-type")

106 kw = request.form.get("keywords")

107 field = request.form.get("fields")

108

109 if kw is None:

110 kw = request.form.get("q") # back-compat for the simple search widget

111

112 # lhs for journals, rhs for articles

113 field_map = {

114 "all": (None, None),

115 "title": ("bibjson.title", "bibjson.title"),

116 "abstract": (None, "bibjson.abstract"),

117 "subject": ("index.classification", "index.classification"),

118 "author": (None, "bibjson.author.name"),

119 "issn": ("index.issn.exact", None),

120 "publisher": ("bibjson.publisher.name", None),

121 "country": ("index.country", None)

122 }

123 default_field_opts = field_map.get(field, None)

124 default_field = None

125

126 route = ""

127 if not ct or ct == "journals":

128 route = url_for("doaj.journals_search")

129 if default_field_opts:

130 default_field = default_field_opts[0]

131 elif ct == "articles":

132 route = url_for("doaj.articles_search")

133 if default_field_opts:

134 default_field = default_field_opts[1]

135 else:

136 abort(400)

137

138 query = dao.Facetview2.make_query(kw, default_field=default_field, default_operator="AND")

139

140 return redirect(route + '?source=' + urllib.parse.quote(json.dumps(query)) + "&ref=" + urllib.parse.quote(ref))

141

142

143#############################################

144

145@blueprint.route("/csv")

146@plausible.pa_event(app.config.get('ANALYTICS_CATEGORY_JOURNALCSV', 'JournalCSV'),

147 action=app.config.get('ANALYTICS_ACTION_JOURNALCSV', 'Download'))

148def csv_data():

149 svc = DOAJ.journalService()

150 if current_user and not current_user.is_anonymous and current_user.has_role(constants.ROLE_PREMIUM_CSV):

151 jc = svc.get_premium_csv()

152 else:

153 jc = svc.get_free_csv()

154

155 if jc is None:

156 abort(404)

157

158 store_url = svc.get_temporary_url(jc)

159

160 if store_url.startswith("/"):

161 store_url = "/store" + store_url

162

163 return redirect(store_url, code=307)

164

165

166@blueprint.route("/sitemap.xml")

167def sitemap_legacy():

168 return redirect(url_for('doaj.sitemap', suffix="0"), 301)

169

170@blueprint.route("/sitemap_index.xml")

171def sitemap_index_legacy():

172 return redirect(url_for('doaj.sitemap', suffix="_index_0"), 301)

173

174@blueprint.route("/sitemap<suffix>.xml")

175def sitemap(suffix):

176 """

177 This route handles both sitemaps, of the form /sitemapN.xml, and sitemap indexes, of the form /sitemap_index_N.xml

178 :param suffix:

179 :return:

180 """

181 url = None

182 if suffix.startswith("_index_"):

183 n = suffix[len("_index_"):]

184 url = models.Cache.get_sitemap_index(n)

185 else:

186 url = models.Cache.get_sitemap(suffix)

187 if url is None:

188 abort(404)

189 if url.startswith("/"):

190 url = "/store" + url

191 return redirect(url, code=307)

192

193

194@blueprint.route("/public-data-dump/<record_type>")

195@login_required

196@plausible.pa_event(app.config.get('ANALYTICS_CATEGORY_PUBLICDATADUMP', 'PublicDataDump'),

197 action=app.config.get('ANALYTICS_ACTION_PUBLICDATADUMP', 'Download'))

198def public_data_dump_redirect(record_type):

199 if not current_user.has_role(constants.ROLE_PUBLIC_DATA_DUMP) and not current_user.has_role(constants.ROLE_PREMIUM_PDD):

200 abort(404)

201

202 svc = DOAJ.publicDataDumpService()

203 if current_user.has_role(constants.ROLE_PREMIUM_PDD):

204 dd = svc.get_premium_dump()

205 else:

206 dd = svc.get_free_dump()

207

208 if dd is None:

209 abort(404)

210

211 store_url = svc.get_temporary_url(dd, record_type)

212

213 if store_url.startswith("/"):

214 store_url = "/store" + store_url

215

216 return redirect(store_url, code=307)

217

218@blueprint.route("/store/<container>/<dir>/<filename>")

219def get_from_local_store_dir(container, dir, filename):

220 file = os.path.join(dir, filename)

221 return get_from_local_store(container, file)

222

223@blueprint.route("/store/<container>/<filename>")

224def get_from_local_store(container, filename):

225 if not app.config.get("STORE_LOCAL_EXPOSE", False):

226 abort(404)

227

228 from portality import store

229 localStore = store.StoreFactory.get(None)

230 file_handle = localStore.get(container, filename)

231 return send_file(file_handle, mimetype="application/octet-stream", as_attachment=True,

232 download_name=os.path.basename(filename))

233

234

235@blueprint.route('/autocomplete/<doc_type>/<field_name>', methods=["GET", "POST"])

236def autocomplete(doc_type, field_name):

237 prefix = request.args.get('q', '')

238 if not prefix:

239 return jsonify({'suggestions': [{"id": "",

240 "text": "No results found"}]}) # select2 does not understand 400, which is the correct code here...

241

242 m = models.lookup_model(doc_type)

243 if not m:

244 return jsonify({'suggestions': [{"id": "",

245 "text": "No results found"}]}) # select2 does not understand 404, which is the correct code here...

246

247 size = request.args.get('size', 5)

248

249 filter_field = app.config.get("AUTOCOMPLETE_ADVANCED_FIELD_MAPS", {}).get(field_name)

250

251 suggs = []

252 if filter_field is None:

253 suggs = m.autocomplete(field_name, prefix, size=size)

254 else:

255 suggs = m.advanced_autocomplete(filter_field, field_name, prefix, size=size, prefix_only=False)

256

257 return jsonify({'suggestions': suggs})

258 # you shouldn't return lists top-level in a JSON response:

259 # http://flask.pocoo.org/docs/security/#json-security

260

261

262def is_issn_by_identifier(identifier):

263 return len(identifier) == 9

264

265

266def find_correct_redirect_identifier(identifier, bibjson) -> str:

267 """

268 return None if identifier is correct and no redirect is needed

269

270 :param identifier:

271 :param bibjson:

272 :return:

273 """

274 if is_issn_by_identifier(identifier): # the journal is referred to by an ISSN

275 # if there is an E-ISSN (and it's not the one in the request), redirect to it

276 eissn = bibjson.get_one_identifier(bibjson.E_ISSN)

277 if eissn and identifier != eissn:

278 return eissn

279

280 # if there's no E-ISSN, but there is a P-ISSN (and it's not the one in the request), redirect to the P-ISSN

281 if not eissn:

282 pissn = bibjson.get_one_identifier(bibjson.P_ISSN)

283 if pissn and identifier != pissn:

284 return pissn

285

286 # The journal has neither a PISSN or an EISSN. Yet somehow

287 # issn_ref is True, the request was referring to the journal

288 # by its ISSN. Not sure how this could ever happen, but just

289 # continue loading the data and do nothing else in such a

290 # case.

291

292 else: # the journal is NOT referred to by any ISSN

293

294 # if there is an E-ISSN, redirect to it

295 # if not, but there is a P-ISSN, redirect to it

296 # if neither ISSN is present, continue loading the page

297 issn = bibjson.get_one_identifier(bibjson.E_ISSN)

298 if not issn:

299 issn = bibjson.get_one_identifier(bibjson.P_ISSN)

300 if issn:

301 return issn

302

303 # let it continue loading if we only have the hex UUID for the journal (no ISSNs)

304 # and the user is referring to the toc page via that ID

305

306

307@blueprint.route("/toc/<identifier>")

308def toc(identifier=None):

309 """ Table of Contents page for a journal. identifier may be the journal id or an issn """

310 # If this route is changed, update JOURNAL_TOC_URL_FRAG in settings.py (partial ToC page link for journal CSV)

311 if identifier is None:

312 abort(400)

313

314 journalSvc = DOAJ.journalService()

315 try:

316 journal = journalSvc.find_best(identifier)

317 except bll_exceptions.ArgumentException:

318 abort(400)

319 except bll_exceptions.TooManyJournals:

320 abort(500)

321

322 if journal is None:

323 abort(404)

324

325 if journal.is_in_doaj() is False:

326 raise ui_exceptions.JournalWithdrawn()

327

328 # journal = find_toc_journal_by_identifier(identifier)

329 bibjson = journal.bibjson()

330 real_identifier = find_correct_redirect_identifier(identifier, bibjson)

331 if real_identifier:

332 return redirect(url_for('doaj.toc', identifier=real_identifier), 301)

333 else:

334 # now render all that information

335 return render_template(templates.PUBLIC_TOC_MAIN, journal=journal, bibjson=bibjson, tab="main")

336

337

338@blueprint.route("/toc/articles/<identifier>")

339def toc_articles_legacy(identifier=None):

340 return redirect(url_for('doaj.toc_articles', identifier=identifier, volume=1, issue=1), 301)

341

342

343@blueprint.route("/toc/<identifier>/articles")

344def toc_articles(identifier=None):

345 if identifier is None:

346 abort(400)

347

348 journalSvc = DOAJ.journalService()

349 try:

350 journal = journalSvc.find_best(identifier)

351 except bll_exceptions.ArgumentException:

352 abort(400)

353 except bll_exceptions.TooManyJournals:

354 abort(500)

355

356 if journal is None:

357 abort(404)

358

359 if journal.is_in_doaj() is False:

360 raise ui_exceptions.JournalWithdrawn()

361

362 bibjson = journal.bibjson()

363 articles_no = journal.article_stats()["total"]

364 real_identifier = find_correct_redirect_identifier(identifier, bibjson)

365 if real_identifier:

366 return redirect(url_for('doaj.toc_articles', identifier=real_identifier), 301)

367 else:

368 return render_template(templates.PUBLIC_TOC_ARTICLES, journal=journal, bibjson=bibjson, articles_no=articles_no, tab="articles")

369

370

371# ~~->Article:Page~~

372@blueprint.route("/article/<identifier>")

373def article_page(identifier=None):

374 # identifier must be the article id

375 article = models.Article.pull(identifier)

376

377 if article is None:

378 article = models.ArticleTombstone.pull(identifier)

379 if article:

380 raise ui_exceptions.TombstoneArticle()

381 else:

382 abort(404, description=Messages.ARTICLE_NOT_FOUND)

383

384 if not article.is_in_doaj():

385 raise ui_exceptions.ArticleFromWithdrawnJournal()

386

387 # find the related journal record

388 journal = article.get_journal()

389 if journal is None:

390 app.logger.exception(Messages.ARTICLE_ABANDONED_LOG.format(article_id=article.id))

391 abort(500, description=Messages.ARTICLE_ABANDONED_PUBLIC)

392 if journal.is_in_doaj() is False:

393 raise ui_exceptions.ArticleFromWithdrawnJournal()

394

395 # issns = article.bibjson().issns()

396 # more_issns = article.bibjson().journal_issns

397 # for issn in issns + more_issns:

398 # journals = models.Journal.find_by_issn(issn)

399 # if len(journals) == 0:

400 # app.logger.exception(Messages.ARTICLE_ABANDONED_LOG.format(article_id=article.id))

401 # abort(500, description=Messages.ARTICLE_ABANDONED_PUBLIC)

402 # try:

403 # journal = models.Journal.get_active_journal(journals)

404 # except exceptions.TooManyJournals:

405 # app.logger.exception(Messages.TOO_MANY_JOURNALS_LOG.format(identifier=identifier))

406 # abort(500, description=Messages.TOO_MANY_JOURNALS.format(identifier=identifier))

407 # except exceptions.JournalWithdrawn:

408 # raise exceptions.ArticleFromWithdrawnJournal

409

410 return render_template(templates.PUBLIC_ARTICLE, article=article, journal=journal, page={"highlight" : True})

411

412

413###############################################################

414# The various static endpoints

415###############################################################

416

417@blueprint.route("/googlebdb21861de30fe30.html")

418def google_webmaster_tools():

419 return 'google-site-verification: googlebdb21861de30fe30.html'

420

421

422@blueprint.route("/accessibility/")

423def accessibility():

424 return render_template(templates.STATIC_PAGE, page_frag="/legal/accessibility.html")

425

426

427@blueprint.route("/privacy/")

428def privacy():

429 return render_template(templates.STATIC_PAGE, page_frag="/legal/privacy.html")

430

431

432@blueprint.route("/contact/")

433def contact():

434 return render_template(templates.STATIC_PAGE, page_frag="/legal/contact.html")

435

436

437@blueprint.route("/terms/")

438def terms():

439 return render_template(templates.STATIC_PAGE, page_frag="/legal/terms.html")

440

441

442@blueprint.route("/code-of-conduct/")

443def conduct():

444 """

445 ~~Conduct:WebRoute~~

446 """

447 return render_template(templates.STATIC_PAGE, page_frag="/legal/code-of-conduct.html")

448

449

450@blueprint.route("/media/")

451def media():

452 """

453 ~~Media:WebRoute~~

454 """

455 return render_template(templates.STATIC_PAGE, page_frag="/legal/media.html")

456

457

458@blueprint.route("/support/")

459def support():

460 return render_template(templates.STATIC_PAGE, page_frag="/support/index.html")

461

462

463@blueprint.route("/support/sponsors/")

464def sponsors():

465 return render_template(templates.STATIC_PAGE, page_frag="/support/sponsors.html")

466

467

468@blueprint.route("/support/publisher-supporters/")

469def publisher_supporters():

470 return render_template(templates.STATIC_PAGE, page_frag="/support/publisher-supporters.html")

471

472

473@blueprint.route("/support/supporters/")

474def supporters():

475 return render_template(templates.STATIC_PAGE, page_frag="/support/supporters.html")

476

477

478@blueprint.route("/support/funders/")

479def funders():

480 return render_template(templates.STATIC_PAGE, page_frag="/support/funders.html")

481

482

483@blueprint.route("/support/thank-you/")

484def application_thanks():

485 return render_template(templates.STATIC_PAGE, page_frag="/support/thank-you.html")

486

487

488@blueprint.route("/apply/guide/")

489def guide():

490 return render_template(templates.STATIC_PAGE, page_frag="/apply/guide.html")

491

492

493@blueprint.route("/apply/transparency/")

494def transparency():

495 return render_template(templates.STATIC_PAGE, page_frag="/apply/transparency.html")

496

497

498@blueprint.route("/apply/why-index/")

499def why_index():

500 return render_template(templates.STATIC_PAGE, page_frag="/apply/why-index.html")

501

502

503@blueprint.route("/apply/publisher-responsibilities/")

504def publisher_responsibilities():

505 return render_template(templates.STATIC_PAGE, page_frag="/apply/publisher-responsibilities.html")

506

507

508@blueprint.route("/apply/copyright-and-licensing/")

509def copyright_and_licensing():

510 return render_template(templates.STATIC_PAGE, page_frag="/apply/copyright-and-licensing.html")

511

512

513@blueprint.route("/docs/oai-pmh/")

514def oai_pmh():

515 return render_template(templates.STATIC_PAGE, page_frag="/docs/oai-pmh.html")

516

517

518@blueprint.route('/docs/api/')

519def docs():

520 major_version = app.config.get("CURRENT_API_MAJOR_VERSION")

521 return redirect(url_for('api_v' + major_version + '.docs'))

522

523

524@blueprint.route("/docs/xml/")

525def xml():

526 return render_template(templates.STATIC_PAGE, page_frag="/docs/xml.html")

527

528

529@blueprint.route("/docs/widgets/")

530def widgets():

531 return render_template(templates.STATIC_PAGE, page_frag="/docs/widgets.html", base_url=app.config.get('BASE_URL'))

532

533

534@blueprint.route("/docs/public-data-dump/")

535def public_data_dump():

536 if current_user and not current_user.is_anonymous and (

537 current_user.has_role(constants.ROLE_PUBLIC_DATA_DUMP) or current_user.has_role(constants.ROLE_PREMIUM_PDD)):

538 dds = DOAJ.publicDataDumpService()

539 if current_user.has_role(constants.ROLE_PREMIUM_PDD):

540 dd = dds.get_premium_dump()

541 else:

542 dd = dds.get_free_dump()

543 return render_template(templates.STATIC_PAGE, page_frag="/docs/pdd-access.html", pdd=dd)

544 return render_template(templates.STATIC_PAGE, page_frag="/docs/pdd-contact.html")

545

546

547@blueprint.route("/docs/openurl/")

548def openurl():

549 return render_template(templates.STATIC_PAGE, page_frag="/docs/openurl.html")

550

551

552@blueprint.route("/docs/faq/")

553def faq():

554 return render_template(templates.STATIC_PAGE, page_frag="/docs/faq.html")

555

556@blueprint.route("/docs/journal-csv")

557def journal_csv():

558 svc = DOAJ.journalService()

559 if not current_user.is_anonymous and current_user.has_role(constants.ROLE_PREMIUM_CSV):

560 jc = svc.get_premium_csv()

561 else:

562 jc = svc.get_free_csv()

563 return render_template(templates.STATIC_PAGE, page_frag="/docs/journal-csv.html", jc=jc)

564

565@blueprint.route("/docs/premium")

566def premium():

567 return render_template(templates.STATIC_PAGE, page_frag="/docs/premium.html")

568

569@blueprint.route("/about/")

570def about():

571 return render_template(templates.STATIC_PAGE, page_frag="/about/index.html")

572

573

574

575@blueprint.route("/at-20/")

576def at_20():

577 return render_template(templates.STATIC_PAGE, page_frag="/about/at-20.html")

582@blueprint.route("/about/ambassadors/")

583def ambassadors():

584 return render_template(templates.STATIC_PAGE, page_frag="/about/ambassadors.html")

585

586

587@blueprint.route("/about/advisory-board-council/")

588def abc():

589 return render_template(templates.STATIC_PAGE, page_frag="/about/advisory-board-council.html")

590

591

592@blueprint.route("/about/editorial-policy-advisory-group/")

593def epag():

594 return render_template(templates.STATIC_PAGE, page_frag="/about/editorial-policy-advisory-group.html")

595

596

597@blueprint.route("/about/volunteers/")

598def volunteers():

599 return render_template(templates.STATIC_PAGE, page_frag="/about/volunteers.html")

600

601

602@blueprint.route("/about/team/")

603def team():

604 return render_template(templates.STATIC_PAGE, page_frag="/about/team.html")

605

606

607@blueprint.route("/preservation/")

608def preservation():

609 return render_template(templates.STATIC_PAGE, page_frag="/preservation/index.html")

610

611

612# LEGACY ROUTES

613@blueprint.route("/subjects")

614def subjects():

615 # return render_template("doaj/subjects.html", subject_page=True, lcc_jstree=json.dumps(lcc_jstree))

616 return redirect(url_for("doaj.journals_search"), 301)

617

618

619@blueprint.route("/application/new")

620def old_application():

621 return redirect(url_for("apply.public_application", **request.args), code=308)

622

623

624@blueprint.route("/<cc>/mejorespracticas")

625@blueprint.route("/<cc>/boaspraticas")

626@blueprint.route("/<cc>/bestpractice")

627@blueprint.route("/<cc>/editionsavante")

628@blueprint.route("/bestpractice")

629@blueprint.route("/oainfo")

630def bestpractice(cc=None):

631 return redirect(url_for("doaj.transparency", **request.args), code=308)

632

633

634@blueprint.route("/membership")

635def membership():

636 return redirect(url_for("doaj.support", **request.args), code=308)

637

638

639@blueprint.route("/publishermembers")

640def old_sponsors():

641 return redirect(url_for("doaj.sponsors", **request.args), code=308)

642

643

644@blueprint.route("/members")

645def members():

646 return redirect(url_for("doaj.supporters", **request.args), code=308)

647

648

649@blueprint.route('/features')

650def features():

651 return redirect(url_for("doaj.xml", **request.args), code=308)

652

653

654# @blueprint.route('/widgets')

655# def old_widgets():

656# return redirect(url_for("doaj.widgets", **request.args), code=308)

657

658

659# @blueprint.route("/public-data-dump/<record_type>")

660# def old_public_data_dump(record_type):

661# return redirect(url_for("doaj.public_data_dump", **request.args), code=308)

662

663

664@blueprint.route("/openurl/help")

665def old_openurl():

666 return redirect(url_for("doaj.openurl", **request.args), code=308)

667

668

669@blueprint.route("/faq")

670def old_faq():

671 return redirect(url_for("doaj.faq", **request.args), code=308)

672

673

674@blueprint.route("/publishers")

675def publishers():

676 return redirect(url_for("doaj.guide", **request.args), code=308)

677

678

679# Redirects necessitated by new templates

680@blueprint.route("/password-reset/")

681def new_password_reset():

682 return redirect(url_for('account.forgot'), code=301)

683

684

685@blueprint.route("/u/<alias>")

686@plausible.pa_event(app.config.get('ANALYTICS_CATEGORY_URLSHORT', 'Urlshort'),

687 action=app.config.get('ANALYTICS_ACTION_URLSHORT_REDIRECT', 'Redirect'))

688def shortened_url(alias):

689 short = DOAJ.shortUrlService().find_url_by_alias(alias)

690 if short:

691 return redirect(short.url)

692

693 app.logger.debug(f"Shortened URL not found: [{alias}]")

694 abort(404)

Coverage for portality / view / doaj.py: 65%

402 statements