diff --git a/cms/assets/img/ambassadors/johndove.jpg b/cms/assets/img/team/johndove.jpg similarity index 100% rename from cms/assets/img/ambassadors/johndove.jpg rename to cms/assets/img/team/johndove.jpg diff --git a/cms/data/ambassadors.yml b/cms/data/ambassadors.yml index 3f797bcd2d..f2a949b4f0 100644 --- a/cms/data/ambassadors.yml +++ b/cms/data/ambassadors.yml @@ -62,13 +62,6 @@ photo: "ivonne.jpg" coi: 2022: https://drive.google.com/file/d/1HnGhYbvbzL34guWOmIqcthcwAN8NADX1/view?usp=sharing - -- name: John G. Dove - region: North America - bio: "John has had a career in executive management, and is now an independent consultant and open access advocate who works with organisations seeking to accelerate their transition to open access. He advises both for-profits and non-profits, and has a particular interest in identifying the steps necessary to flip an entire discipline’s scholarly record to open access. His ambassador activities focus on increasing the support to DOAJ from the community. He served for six years on NISO’s Information Discovery and Interchange Topic Committee, and has written for Learned Publishing, Against the Grain, and Scholarly Kitchen. John serves on the Board of Trustees of his local public library in Revere, Massachusetts. He has a B.A. in Mathematics from Oberlin College." - photo: "johndove.jpg" - coi: - 2022: https://drive.google.com/file/d/1cWijl2xdmVjshsvaGTABOvC_chIIfuVA/view?usp=sharing - name: Mahmoud Khalifa region: Middle East and Persian Gulf diff --git a/cms/data/team.yml b/cms/data/team.yml index 781c84737e..e3a0b3df8e 100644 --- a/cms/data/team.yml +++ b/cms/data/team.yml @@ -6,17 +6,15 @@ photo: alejandra.png bio: 'Alejandra has a Bachelor’s degree in Information Science and a Master’s degree in Digital Media. She has around ten years of experience in information management, knowledge management and scholarly communication at national and international level.' coi: - 2016: https://drive.google.com/file/d/0ByRf6PVViI-mVFhjdkZWaFJZOTQ/view?usp=sharing&resourcekey=0-gMdxxHXyyJB9zFZIuh99QQ 2018: https://drive.google.com/file/d/0ByRf6PVViI-mRlRYTDBPRlZiWTRxQ3VMTUZpQnZ5ZkwyLVQ4/view?usp=sharing&resourcekey=0-mlQ6rSCEnr6RfpCwh_4SMw 2020: https://drive.google.com/file/d/1PF7Cc9vGAwWGqpqDo7nRULjxWIF2NR_Q/view?usp=sharing 2022: https://drive.google.com/file/d/1E45ycyctDfYkh65ZCM8PMtQsYym_eP6L/view?usp=sharing - name: Cenyu Shen - role: Quality Team Lead and Managing Editor + role: Deputy Head of Editorial (Quality) photo: cenyu.jpg bio: 'Cenyu holds a PhD in Information Systems Science at Hanken School of Economics in Finland. She has spent around seven years on Open Access research with a particular focus on gold OA publishing concerning journals and publishers outside the mainstream. She was one of three DOAJ ambassadors for China from 2016 to 2017.' coi: - 2016: https://drive.google.com/file/d/0B0fPCpIPjZlmNHZCQmxpUmN6bUEtYUx2VHZnbjVySS1fRTlr/view?usp=sharing&resourcekey=0-1TRIV1MEQMhdGbmCd7CbOA 2020: https://drive.google.com/file/d/1rm9fjOF3OHJ9lR9wEUyQBQTO2KdoNQcE/view?usp=sharing 2022: https://drive.google.com/file/d/1Mn_CR0twKxyFbbHxsLSrgeU984BNOLlS/view?usp=sharing @@ -25,7 +23,6 @@ photo: clara.jpg bio: 'Clara has 10 years experience in the scholarly publishing industry. She worked at Cambridge University Press as an Open Access Project Manager until 2015. She also works in science communication as a freelancer at the University Pompeu Fabra, Barcelona. Clara speaks Spanish, Catalan, English and some German and French. She loves cetaceans, freediving, cycling, and is an enthusiastic cook.' coi: - 2016: https://drive.google.com/file/d/0ByRf6PVViI-mbDFybndLbldEbFE/view?usp=sharing&resourcekey=0-lKZNFwvUNdVAGKatvnKiPg 2018: https://drive.google.com/file/d/1LHmZSZ6bwf6U71fNvIibJa6R1lquNhfR/view?usp=sharing 2020: https://drive.google.com/file/d/1v4duxnoTcNo4UbL_GBa5D1T8JtTl7oY1/view?usp=sharing 2022: https://drive.google.com/file/d/1hevYxG1102llDy-_i-onwKbDuOlBguA_/view?usp=sharing @@ -35,7 +32,6 @@ photo: dominic.jpg bio: 'Dominic has over 25 years of experience working with publisher and library communities. He is responsible for operations and development of the DOAJ platform. He acts as Committee chair for the Think. Check. Submit. initiative, of which DOAJ is a founding organisation. He represents DOAJ in Project JASPER, a cross-industry project working to ensure that journals are preserved for the long term. He also sits on the OASPA Board of Directors and serves as Secretary. Outside of work, he is reluctantly becoming an expert in the playparks of Stockholm with his twin sons.' coi: - 2016: https://drive.google.com/file/d/0ByRf6PVViI-mWmU0UHZqZm1xcDQ/view?usp=sharing&resourcekey=0-BmQKwWn6Vb9ot73Xie66aA 2018: https://drive.google.com/file/d/13XX_GUrw2xRmXARjRrTxegULPT8Redka/view?usp=sharing 2020: https://drive.google.com/file/d/1nxFOuAdXLb8A-LulhNpz9i5vSmr5DBwF/view?usp=sharing 2022: https://drive.google.com/file/d/1HBF9RLaIt3lFNG6WDcV08fQSMS_C6zwA/view?usp=sharing @@ -62,12 +58,18 @@ coi: 2022: https://drive.google.com/file/d/1-3xzwkHMclREgLhj_XNF5n6Nr4q2_bnw/view?usp=sharing +- name: John G. Dove + role: Advisor + photo: johndove.jpg + bio: "John has had a career in executive management, and is now an independent consultant and open access advocate who works with organisations seeking to accelerate their transition to open access. He advises both for-profits and non-profits, and has a particular interest in identifying the steps necessary to flip an entire discipline’s scholarly record to open access. His ambassador activities focus on increasing the support to DOAJ from the community. He served for six years on NISO’s Information Discovery and Interchange Topic Committee, and has written for Learned Publishing, Against the Grain, and Scholarly Kitchen. John serves on the Board of Trustees of his local public library in Revere, Massachusetts. He has a B.A. in Mathematics from Oberlin College." + coi: + 2022: https://drive.google.com/file/d/1cWijl2xdmVjshsvaGTABOvC_chIIfuVA/view?usp=sharing + - name: Judith Barnsby role: Head of Editorial photo: judith.jpg bio: 'Judith has 25 years of experience in the scholarly publishing industry, working for a range of non-profit society publishers and service providers before joining DOAJ. She has a keen interest in publishing standards and protocols, and has served on the board of CLOCKSS and as chair of the PALS (publisher and library solutions) working group in the UK. Judith loves books, especially detective fiction, and volunteers in her local library.' coi: - 2016: https://drive.google.com/file/d/0B0fPCpIPjZlmb3JmVkFYbjN5aTh1OUhLd2lZaEV0ZlFwbTZV/view?usp=sharing&resourcekey=0-o_PXKLk5UFbPk_-4B61jVA 2018: https://drive.google.com/file/d/0ByRf6PVViI-mV2lfMjByQjYxUkpMcXhuc2l5Q3ZDWlpiYUtZ/view?usp=sharing&resourcekey=0-6eiGIRal00eXvgJUTeN_lw 2020: https://drive.google.com/file/d/18MWTsze4cDQQRPHJl2XrYgHQvlxhsPZa/view?usp=sharing 2023: https://drive.google.com/file/d/1hUsVIY09N6WceSx1edTM-h516CJGkHcu/view?usp=share_link @@ -77,7 +79,6 @@ photo: Kamel.jpg bio: 'Kamel is Full Professor of Chemistry at the University of Bejaia, Algeria (ORCID). He gained his PhD in Process Engineering and Chemistry of Materials Science at the University of Setif, Algeria. Kamel joined DOAJ in 2016 as an Ambassador for North Africa. He is currently Creative Commons Algeria Chapter lead, director of the Laboratory of Organic Materials at the University of Bejaia and editor-in-chief of Algerian Journal of Natural Products. His scientific activity is focused on chemistry of Natural Products, scholarly communications and new developments in academic publishing. Father of 3 daughters, he likes travelling, healthy local foods & home-made snacks.' coi: - 2016: https://drive.google.com/file/d/0B0fPCpIPjZlmVEN4X1Q0RDdCams1NXhveW1HQmtMYU56bDE4/view?usp=sharing&resourcekey=0-wA1CGAbjB6FAX33gCDQmrA 2018: https://drive.google.com/file/d/1JdF2kh-fLXz8kPGN_3ijDt5y9K6s0hOQ/view?usp=sharing 2020: https://drive.google.com/file/d/1iXrjwLTNBXwKD2TwrPD9ApKL7O6uZ8Z7/view?usp=sharing 2022: https://drive.google.com/file/d/1cl18h_mYnNogYs8Rk-fhBTW6WKOTC2IF/view?usp=sharing @@ -94,28 +95,19 @@ photo: lars.jpg bio: 'Lars worked at Danish university libraries for two decades and was Director of Libraries at Lund University, Sweden from 2001 to 2011. He founded the DOAJ in 2003, and was Managing Director from 2013-2021. He has vast experience in change management, re-engineering of academic libraries, and development of information services for research & higher education. For two decades Lars has been a strong advocate of open access and for providing services to the open access movement. He is co-founder of OpenDOAR, the Directory of Open Access Books and Think. Check. Submit. Lars lives outside Copenhagen, and is married with 4 children and 4 grandchildren. He enjoys vegetable gardening, growing cacti and succulents, and playing internet chess.' coi: - 2016: https://drive.google.com/file/d/0ByRf6PVViI-mbmo2aU9NWkx5dGs/view?usp=sharing&resourcekey=0-mpdRgVU9UlFjC614-woDvg 2018: https://drive.google.com/file/d/1mm1a8nbY5MQX9loqIs2ZQuVN-73RfPuN/view?usp=sharing 2021: https://drive.google.com/file/d/1bNj5sqUsu4sRLmm_YOuh3JCSMERzQ1Ro/view?usp=sharing 2022: https://drive.google.com/file/d/1fRJtvci2_j4vad0C5N1pfqm2sHZQkFz3/view?usp=sharing - name: Leena Shah - role: Managing Editor and Ambassador + role: Deputy Head of Editorial (Workflow) and Ambassador photo: leena.jpg bio: "Leena joined the DOAJ team in 2016 as an Ambassador for India before becoming a Managing Editor. Prior to joining DOAJ she worked as a science librarian at Nanyang Technological University, Singapore, where she developed a keen interest in scholarly communication & open science. A recent addition to her interests is artificial intelligence in scholarly communication. Leena holds a Master’s degree in Information Studies and lives in Singapore. She loves watching sci-fi shows and is enthusiastic about travelling to new places." coi: - 2016: https://drive.google.com/file/d/0B0fPCpIPjZlmTHZuaEtMSDNIeUpKT2Fid19jVjVFTkRoUmdj/view?usp=sharing&resourcekey=0-KqvRVa30bQEUfqO-YA1L-g 2018: https://drive.google.com/file/d/1tifEjAIlU3txBw9DjIcRW9cZL7YG7_nU/view?usp=sharing 2020: https://drive.google.com/file/d/1zU-lLB5W54E_QUm5uto5tqB6cZl83TAJ/view?usp=sharing 2022: https://drive.google.com/file/d/19rw-naMJqHkI5T7aDIDPUkwPutBdDpDm/view?usp=sharing -- name: Luis Montilla - role: Managing Editor - photo: luis.jpeg - bio: "Luis is a marine ecologist with a passion for improving the quality of scientific publishing. After finishing his Masters in Venezuela, he spent three years in Italy completing his PhD studying marine microbial symbioses in seagrass beds. In his free time, he enjoys reading and watching movies." - coi: - 2023: https://drive.google.com/file/d/1IJhnV2Ht5t5jilaCAFzpuFdYk7UMOjN3/view?usp=sharing - - name: Mahmoud Khalifa role: Managing Editor and Ambassador photo: mahmoud-new.jpg @@ -151,7 +143,6 @@ photo: Rikard.jpg bio: 'Rikard has a Bachelor of Arts degree with a Major in Cultural Sciences and a specialization in publishing. He enjoys reading about philosophy and religion.' coi: - 2016: https://drive.google.com/file/d/0ByRf6PVViI-mdnJPdldOM0hUMFU/view?usp=sharing&resourcekey=0-8dJAtvm2n7vXV9NhqZYckw 2018: https://drive.google.com/file/d/1tOnW8L6TwolyLpIXwMKTITf9wGh_ukLb/view?usp=sharing 2020: https://drive.google.com/file/d/14c0RgpyD2Slzyh5s8LGvj5OwWbL4H8NX/view?usp=sharing 2023: https://drive.google.com/file/d/1HQIh1DlfhEutTWniXDGLYFVa9VxJ4OT9/view?usp=share_link @@ -161,7 +152,6 @@ photo: sonja.jpg bio: 'Sonja is a former Information Librarian from Lund University Library. She has a B.A. in English, Bulgarian and Russian from Lund University and specialises in applications for journals in the Slavonic languages.' coi: - 2016: https://drive.google.com/file/d/0ByRf6PVViI-mNUFoZWV4YnZ3bDg/view?usp=sharing&resourcekey=0-1JRid_DHRMKbgdzmVYL7NQ 2018: https://drive.google.com/file/d/1M5AGEDP79uk2olCcmVYjKCsmzL7tG2Vc/view?usp=sharing 2020: https://drive.google.com/file/d/1-4RJYScTs_zMBeD5zESNvCoIBCWTOWHR/view?usp=sharing 2022: https://drive.google.com/file/d/1soZtiW6gyVJPl7P_J60j2TL2Fqzl0QAs/view?usp=sharing @@ -178,7 +168,6 @@ photo: tom.jpg bio: 'Tom has a PhD in molecular microbiology and spent several years in Africa doing research on malaria, sleeping sickness and meningococcal epidemics. He has been actively advocating open access and open science since 2012 when he joined the Open Knowledge community and became a member of the DOAJ advisory board. His current research interests are development of quality systems for the assessment of scholarly journals and articles, and research in the area of soil microbiology in relation to soil health and human health.' coi: - 2016: https://drive.google.com/file/d/0ByRf6PVViI-mYUFZNDRISTZodUU/view?usp=sharing&resourcekey=0-g13FJaUJpdR_t2rMLEyzEQ 2018: https://drive.google.com/file/d/1x0w-a1TWQdJDKPtQpGhmDZSdA4BhFSpI/view?usp=sharing 2020: https://drive.google.com/file/d/1VyirUdc6FBNOujl938bHf1JCL1jLNwXV/view?usp=sharing 2022: https://drive.google.com/file/d/1ww7WHQEg1395bPn20Arb7LJn9lIROdBl/view?usp=sharing diff --git a/cms/pages/legal/terms.md b/cms/pages/legal/terms.md index ad872453eb..1b27b66e99 100644 --- a/cms/pages/legal/terms.md +++ b/cms/pages/legal/terms.md @@ -32,7 +32,7 @@ DOAJ uses a variety of licenses for the different parts of its website and the c + In our [OAI-PMH feed](/docs/oai-pmh) + In the [full data dump of all article metadata](/docs/public-data-dump/). -4. The *open source software* that DOAJ is built with is licensed under [an Apache license Version 2](https://github.com/DOAJ/doaj/blob/a6fc2bee499b5a8a1f24fb098acfb8e10bd72503/portality/static/vendor/select2-3.5.4/LICENSE). +4. The *open source software* that DOAJ is built with is licensed under [an Apache license Version 2](https://github.com/DOAJ/doaj/blob/develop/LICENSE). --- diff --git a/deploy/doaj_gunicorn_config.py b/deploy/doaj_gunicorn_config.py index f9425de5e5..a08dd6ef62 100644 --- a/deploy/doaj_gunicorn_config.py +++ b/deploy/doaj_gunicorn_config.py @@ -1,7 +1,7 @@ import multiprocessing bind = "0.0.0.0:5050" -workers = multiprocessing.cpu_count() * 8 + 1 +workers = multiprocessing.cpu_count() * 6 + 1 proc_name = 'doaj' max_requests = 1000 @@ -13,4 +13,4 @@ max_requests_jitter = 100 timeout = 40 -graceful_timeout = 40 \ No newline at end of file +graceful_timeout = 40 diff --git a/deploy/lambda/alert_backups_missing.py b/deploy/lambda/alert_backups_missing.py index 38a9edbc2e..566a361b9a 100644 --- a/deploy/lambda/alert_backups_missing.py +++ b/deploy/lambda/alert_backups_missing.py @@ -1,5 +1,5 @@ #!/usr/bin/env python3 -""" Steven Eardley 2020-02-07 for DOAJ - uploaded manually (todo: we should upload this in the release script) """ +""" Steven Eardley 2023-09-15 for DOAJ - uploaded manually (todo: we should upload this in the release script) """ # ~~BackupsMissing:Monitoring->Lambda:Technology~~ @@ -8,23 +8,25 @@ import json from datetime import datetime, timezone, timedelta -from portality.lib.dates import FMT_DATETIME_STD - s3 = boto3.client('s3') # Check the doaj elasticsearch snapshot bucket has been updated today (should happen daily at 0600 via background job) -buckets = ['doaj-index-backups'] +buckets = ['doaj-index-ipt-backups'] + # Check the doaj-nginx logs bucket has been updated today (should happen daily at 0630 via cron logrotate) -buckets += ['doaj-nginx-logs'] +# buckets += ['doaj-nginx-logs'] def lambda_handler(event, context): """ The main function executed by Lambda""" + start = datetime.utcnow() summary = {'success': [], 'fail': []} for b in buckets: + print('Checking bucket {0} was updated today'.format(b)) + # First check the bucket actually exists try: s3.head_bucket(Bucket=b) @@ -32,11 +34,13 @@ def lambda_handler(event, context): error_code = int(e.response['Error']['Code']) if error_code == 404: send_alert_email(b, last_mod=None) + raise # Then check the expected entry exists in the bucket's objects. files = list_bucket_keys(bucket_name=b) old_to_new = sorted(files, key=lambda f: f['LastModified']) newest = old_to_new[-1] + print('Latest backup is', newest) # If the newest file is older than 1 day old, our backups are not up to date. if datetime.now(timezone.utc) - newest['LastModified'] > timedelta(days=1): @@ -47,6 +51,8 @@ def lambda_handler(event, context): summary['success'].append(b) print(summary) # For the CloudWatch logs + print('Completed in', str(datetime.utcnow() - start)) + return str(summary) @@ -86,8 +92,8 @@ def send_alert_email(bucket, last_mod): msg = 'AWS backup error: bucket {b} is missing.'.format(b=bucket) else: msg = 'AWS backup error: bucket {b} has not been updated today - it was last modified on {t}.' \ - '\nYou may wish to check the corresponding logs.'.format(b=bucket, - t=last_mod.strftime(FMT_DATETIME_STD)) + '\nYou may wish to check the corresponding logs.'.format(b=bucket, t=last_mod.strftime( + '%Y-%m-%dT%H:%M:%SZ')) r = botocore.vendored.requests.post('https://api.mailgun.net/v3/doaj.org/messages', auth=('api', credentials.get('ERROR_MAIL_API_KEY', '')), diff --git a/doajtest/matrices/article_create_article/issn_validation_against_journal.matrix.csv b/doajtest/matrices/article_create_article/issn_validation_against_journal.matrix.csv new file mode 100644 index 0000000000..0d2f704aba --- /dev/null +++ b/doajtest/matrices/article_create_article/issn_validation_against_journal.matrix.csv @@ -0,0 +1,17 @@ +test_id,eissn,pissn,validated +1,eissn_in_doaj,pissn_in_doaj,yes +2,eissn_in_doaj,eissn_not_in_doaj, +3,eissn_in_doaj,pissn_not_in_doaj, +4,eissn_in_doaj,!eissn_in_doaj, +5,pissn_in_doaj,eissn_in_doaj, +6,pissn_in_doaj,eissn_not_in_doaj, +7,pissn_in_doaj,pissn_not_in_doaj, +8,pissn_in_doaj,!pissn_in_doaj, +9,eissn_not_in_doaj,eissn_in_doaj, +10,eissn_not_in_doaj,pissn_in_doaj, +11,eissn_not_in_doaj,pissn_not_in_doaj, +12,eissn_not_in_doaj,!eissn_not_in_doaj, +13,pissn_not_in_doaj,eissn_in_doaj, +14,pissn_not_in_doaj,pissn_in_doaj, +15,pissn_not_in_doaj,eissn_not_in_doaj, +16,pissn_not_in_doaj,!pissn_not_in_doaj, diff --git a/doajtest/matrices/article_create_article/issn_validation_against_journal.settings.csv b/doajtest/matrices/article_create_article/issn_validation_against_journal.settings.csv new file mode 100644 index 0000000000..a8eab3f4ce --- /dev/null +++ b/doajtest/matrices/article_create_article/issn_validation_against_journal.settings.csv @@ -0,0 +1,19 @@ +field,test_id,eissn,pissn,validated +type,index,generated,generated,conditional +deafult,,,,no +,,,, +values,,eissn_in_doaj,eissn_in_doaj,yes +values,,pissn_in_doaj,pissn_in_doaj,no +values,,eissn_not_in_doaj,eissn_not_in_doaj, +values,,pissn_not_in_doaj,pissn_not_in_doaj, +,,,, +,,,, +conditional validated,,eissn_in_doaj,pissn_in_doaj,yes +constraint eissn,,eissn_in_doaj,!eissn_in_doaj, +constraint eissn,,eissn_not_in_doaj,!eissn_not_in_doaj, +constraint eissn,,pissn_not_in_doaj,!pissn_not_in_doaj, +constraint eissn,,pissn_in_doaj,!pissn_in_doaj, +constraint pissn,,eissn_in_doaj,!eissn_in_doaj, +constraint pissn,,eissn_not_in_doaj,!eissn_not_in_doaj, +constraint pissn,,pissn_not_in_doaj,!pissn_not_in_doaj, +constraint pissn,,pissn_in_doaj,!pissn_in_doaj, \ No newline at end of file diff --git a/doajtest/matrices/article_create_article/issn_validation_against_journal.settings.json b/doajtest/matrices/article_create_article/issn_validation_against_journal.settings.json new file mode 100644 index 0000000000..11d1012a96 --- /dev/null +++ b/doajtest/matrices/article_create_article/issn_validation_against_journal.settings.json @@ -0,0 +1,119 @@ +{ + "parameters": [ + { + "name": "test_id", + "type": "index" + }, + { + "name": "eissn", + "type": "generated", + "values": { + "eissn_in_doaj": { + "constraints": { + "pissn": { + "nor": [ + "eissn_in_doaj" + ] + } + } + }, + "pissn_in_doaj": { + "constraints": { + "pissn": { + "nor": [ + "pissn_in_doaj" + ] + } + } + }, + "eissn_not_in_doaj": { + "constraints": { + "pissn": { + "nor": [ + "eissn_not_in_doaj" + ] + } + } + }, + "pissn_not_in_doaj": { + "constraints": { + "pissn": { + "nor": [ + "pissn_not_in_doaj" + ] + } + } + } + } + }, + { + "name": "pissn", + "type": "generated", + "values": { + "eissn_in_doaj": {}, + "pissn_in_doaj": {}, + "eissn_not_in_doaj": {}, + "pissn_not_in_doaj": {}, + "!eissn_in_doaj": { + "constraints": { + "eissn": { + "or": [ + "eissn_in_doaj" + ] + } + } + }, + "!eissn_not_in_doaj": { + "constraints": { + "eissn": { + "or": [ + "eissn_not_in_doaj" + ] + } + } + }, + "!pissn_not_in_doaj": { + "constraints": { + "eissn": { + "or": [ + "pissn_not_in_doaj" + ] + } + } + }, + "!pissn_in_doaj": { + "constraints": { + "eissn": { + "or": [ + "pissn_in_doaj" + ] + } + } + } + } + }, + { + "name": "validated", + "type": "conditional", + "values": { + "yes": { + "conditions": [ + { + "eissn": { + "or": [ + "eissn_in_doaj" + ] + }, + "pissn": { + "or": [ + "pissn_in_doaj" + ] + } + } + ] + }, + "no": {} + } + } + ] +} \ No newline at end of file diff --git a/doajtest/unit/test_article_acceptable_and_permissions.py b/doajtest/unit/test_article_acceptable_and_permissions.py index eb4c04d4fb..5e0328635f 100644 --- a/doajtest/unit/test_article_acceptable_and_permissions.py +++ b/doajtest/unit/test_article_acceptable_and_permissions.py @@ -14,6 +14,11 @@ def is_acceptable_load_cases(): "test_id", {"test_id": []}) +def issn_validation_against_journal_load_sets(): + return load_parameter_sets(rel2abs(__file__, "..", "matrices", "article_create_article"), "issn_validation_against_journal", + "test_id", + {"test_id": []}) + class TestBLLPrepareUpdatePublisher(DoajTestCase): @@ -110,4 +115,73 @@ def test_has_permissions(self): assert failed_result["unowned"].sort() == [pissn, eissn].sort() # assert failed_result == {'success': 0, 'fail': 1, 'update': 0, 'new': 0, 'shared': [], # 'unowned': [pissn, eissn], - # 'unmatched': []}, "received: {}".format(failed_result) \ No newline at end of file + # 'unmatched': []}, "received: {}".format(failed_result) + + + @parameterized.expand(issn_validation_against_journal_load_sets) + def test_issn_validation_against_journal_load_sets(self, value, kwargs): + kwpissn = kwargs.get("pissn") + kweissn = kwargs.get("eissn") + validated = kwargs.get("validated") + + js = JournalFixtureFactory.make_many_journal_sources(2) + journal_in_doaj = Journal(**js[0]) + journal_in_doaj.set_in_doaj(True) + journal_in_doaj.bibjson().pissn = "1111-1111" + journal_in_doaj.bibjson().eissn = "2222-2222" + journal_in_doaj.save(blocking=True) + + journal_not_in_doaj = Journal(**js[1]) + journal_not_in_doaj.set_in_doaj(False) + journal_not_in_doaj.bibjson().pissn = "3333-3333" + journal_not_in_doaj.bibjson().eissn = "4444-4444" + journal_not_in_doaj.save(blocking=True) + + if (kwpissn == "pissn_in_doaj"): + pissn = journal_in_doaj.bibjson().pissn + elif (kwpissn == "eissn_in_doaj"): + pissn = journal_in_doaj.bibjson().eissn + elif (kwpissn == "pissn_not_in_doaj"): + pissn = journal_not_in_doaj.bibjson().pissn + else: + pissn = journal_not_in_doaj.bibjson().eissn + + if (kweissn == "pissn_in_doaj"): + eissn = journal_in_doaj.bibjson().pissn + elif (kweissn == "eissn_in_doaj"): + eissn = journal_in_doaj.bibjson().eissn + elif (kweissn == "pissn_not_in_doaj"): + eissn = journal_not_in_doaj.bibjson().pissn + else: + eissn = journal_not_in_doaj.bibjson().eissn + + + art_source = ArticleFixtureFactory.make_article_source(pissn=pissn, eissn=eissn) + article = Article(**art_source) + + if validated: + self.assertIsNone(self.svc.is_acceptable(article)) + + else: + with self.assertRaises(exceptions.ArticleNotAcceptable): + self.svc.is_acceptable(article) + + def test_check_validation_for_2_journals(self): + + js = JournalFixtureFactory.make_many_journal_sources(2, in_doaj=True) + journal_in_doaj = Journal(**js[0]) + journal_in_doaj.bibjson().pissn = "1111-1111" + journal_in_doaj.bibjson().eissn = "2222-2222" + journal_in_doaj.save(blocking=True) + + journal_not_in_doaj = Journal(**js[1]) + journal_not_in_doaj.bibjson().pissn = "3333-3333" + journal_not_in_doaj.bibjson().eissn = "4444-4444" + journal_not_in_doaj.save(blocking=True) + + + art_source = ArticleFixtureFactory.make_article_source(pissn="1111-1111", eissn="4444-4444") + article = Article(**art_source) + + with self.assertRaises(exceptions.ArticleNotAcceptable): + self.svc.is_acceptable(article) \ No newline at end of file diff --git a/doajtest/unit/test_models.py b/doajtest/unit/test_models.py index 5551cdcf5d..06175e6d76 100644 --- a/doajtest/unit/test_models.py +++ b/doajtest/unit/test_models.py @@ -1661,3 +1661,30 @@ def test_get_name_safe(self): # account does not exist assert models.Account.get_name_safe('not existing account id') == '' + def test_11_find_by_issn(self): + js = JournalFixtureFactory.make_many_journal_sources(2, in_doaj=True) + j1 = models.Journal(**js[0]) + j1.bibjson().pissn = "1111-1111" + j1.bibjson().eissn = "2222-2222" + j1.save(blocking=True) + + j2 = models.Journal(**js[1]) + j2.bibjson().pissn = "3333-3333" + j2.bibjson().eissn = "4444-4444" + j2.save(blocking=True) + + journals = models.Journal.find_by_issn(["1111-1111", "2222-2222"], True) + assert len(journals) == 1 + assert journals[0].id == j1.id + + journals = models.Journal.find_by_issn(["1111-1111", "3333-3333"], True) + assert len(journals) == 2 + assert journals[0].id == j1.id + assert journals[1].id == j2.id + + journals = models.Journal.find_by_issn_exact(["1111-1111", "2222-2222"], True) + assert len(journals) == 1 + assert journals[0].id == j1.id + + journals = models.Journal.find_by_issn_exact(["1111-1111", "3333-3333"], True) + assert len(journals) == 0 \ No newline at end of file diff --git a/portality/bll/services/article.py b/portality/bll/services/article.py index 7b55894d24..777d499636 100644 --- a/portality/bll/services/article.py +++ b/portality/bll/services/article.py @@ -159,9 +159,6 @@ def _validate_issns(article_bibjson: models.ArticleBibJSON): if len(pissn) > 1 or len(eissn) > 1: raise exceptions.ArticleNotAcceptable(message=Messages.EXCEPTION_TOO_MANY_ISSNS) - pissn = article_bibjson.get_one_identifier("pissn") - eissn = article_bibjson.get_one_identifier("eissn") - # no pissn or eissn if not pissn and not eissn: raise exceptions.ArticleNotAcceptable(message=Messages.EXCEPTION_NO_ISSNS) @@ -252,7 +249,8 @@ def has_permissions(self, account, article, limit_to_account): def is_acceptable(self, article: models.Article): """ conduct some deep validation on the article to make sure we will accept it - or the moment, this just means making sure it has a DOI and a fulltext + this just means making sure it has a DOI and a fulltext, and that its ISSNs + match a single journal """ try: bj = article.bibjson() @@ -266,12 +264,39 @@ def is_acceptable(self, article: models.Article): raise exceptions.ArticleNotAcceptable(message=Messages.EXCEPTION_NO_DOI_NO_FULLTEXT) self._validate_issns(bj) + journal = self.match_journal_with_validation(bj) # is journal in doaj (we do this check last as it has more performance impact) - journal = article.get_journal() if journal is None or not journal.is_in_doaj(): raise exceptions.ArticleNotAcceptable(message=Messages.EXCEPTION_ADDING_ARTICLE_TO_WITHDRAWN_JOURNAL) + @staticmethod + def match_journal_with_validation(article_bibjson: models.ArticleBibJSON): + pissn = article_bibjson.get_one_identifier("pissn") + eissn = article_bibjson.get_one_identifier("eissn") + + issns = [] + + if pissn is not None: + issns.append(pissn) + if eissn is not None: + issns.append(eissn) + + journal = models.Journal.find_by_issn_exact(issns, True) + + # check if only one journal matches pissn and eissn and if they are in the correct fields + # no need to check eissn, if pissn matches, pissn and eissn are different and only 1 journal has been found - then eissn matches too + if len(journal) != 1: + raise exceptions.ArticleNotAcceptable(message=Messages.EXCEPTION_MISMATCHED_ISSNS) + if pissn is not None: + if journal[0].bibjson().pissn != pissn: + raise exceptions.ArticleNotAcceptable(message=Messages.EXCEPTION_MISMATCHED_ISSNS) + if eissn is not None: + if journal[0].bibjson().eissn != eissn: + raise exceptions.ArticleNotAcceptable(message=Messages.EXCEPTION_MISMATCHED_ISSNS) + + return journal[0] + @staticmethod def is_legitimate_owner(article, owner): """ diff --git a/portality/forms/application_processors.py b/portality/forms/application_processors.py index 13a294d14d..1cd426c1f6 100644 --- a/portality/forms/application_processors.py +++ b/portality/forms/application_processors.py @@ -198,8 +198,11 @@ def _patch_target_note_id(self): for note in self.target.notes: note_date = dates.parse(note['date']) if not note.get('author_id') and note_date > dates.before_now(60): - note['author_id'] = current_user.id - + try: + note['author_id'] = current_user.id + except AttributeError: + # Skip if we don't have a current_user + pass class NewApplication(ApplicationProcessor): @@ -307,7 +310,6 @@ def patch_target(self): if (self.target.owner is None or self.target.owner == "") and (self.source.owner is not None): self.target.set_owner(self.source.owner) - def finalise(self, account, save_target=True, email_alert=True): """ account is the administrator account carrying out the action @@ -326,7 +328,6 @@ def finalise(self, account, save_target=True, email_alert=True): elif not j.is_in_doaj(): raise Exception(Messages.EXCEPTION_EDITING_WITHDRAWN_JOURNAL) - # if we are allowed to finalise, kick this up to the superclass super(AdminApplication, self).finalise() @@ -813,7 +814,6 @@ def patch_target(self): if (self.target.owner is None or self.target.owner == "") and (self.source.owner is not None): self.target.set_owner(self.source.owner) - def finalise(self): # FIXME: this first one, we ought to deal with outside the form context, but for the time being this # can be carried over from the old implementation diff --git a/portality/lib/plausible.py b/portality/lib/plausible.py index 2aa602d986..90b1b8f46b 100644 --- a/portality/lib/plausible.py +++ b/portality/lib/plausible.py @@ -62,7 +62,7 @@ def send_event(goal: str, on_completed=None, **props_kwargs): def _send(): resp = requests.post(plausible_api_url, json=payload, headers=headers) if resp.status_code >= 300: - logger.warning(f'send plausible event api fail. [{resp.status_code}][{resp.text}]') + logger.warning(f'Send plausible event API fail. snd: [{resp.url}] [{headers}] [{payload}] rcv: [{resp.status_code}] [{resp.text}]') if on_completed: on_completed(resp) diff --git a/portality/models/v2/journal.py b/portality/models/v2/journal.py index ac1ce42585..efa6aa53e4 100644 --- a/portality/models/v2/journal.py +++ b/portality/models/v2/journal.py @@ -70,6 +70,22 @@ def find_by_issn(cls, issns, in_doaj=None, max=10): records = [cls(**r.get("_source")) for r in result.get("hits", {}).get("hits", [])] return records + @classmethod + def find_by_issn_exact(cls, issns, in_doaj=None, max=2): + """ + Finds journal that matches given issns exactly - if no data problems should always be only 1 + """ + if not isinstance(issns, list): + issns = [issns] + if len(issns) > 2: + return [] + q = JournalQuery() + q.find_by_issn_exact(issns, in_doaj=in_doaj, max=max) + result = cls.query(q=q.query) + # create an array of objects, using cls rather than Journal, which means subclasses can use it too + records = [cls(**r.get("_source")) for r in result.get("hits", {}).get("hits", [])] + return records + @classmethod def issns_by_owner(cls, owner, in_doaj=None): q = IssnQuery(owner, in_doaj=in_doaj) @@ -922,6 +938,16 @@ class JournalQuery(object): } } + must_query = { + "track_total_hits": True, + "query": { + "bool": { + "must": [ + ] + } + } + } + all_doaj = { "track_total_hits": True, "query": { @@ -947,6 +973,14 @@ def find_by_issn(self, issns, in_doaj=None, max=10): self.query["query"]["bool"]["must"].append({"term": {"admin.in_doaj": in_doaj}}) self.query["size"] = max + def find_by_issn_exact(self, issns, in_doaj=None, max=10): + self.query = deepcopy(self.must_query) + for issn in issns: + self.query["query"]["bool"]["must"].append({"term": {"index.issn.exact": issn}}) + if in_doaj is not None: + self.query["query"]["bool"]["must"].append({"term": {"admin.in_doaj": in_doaj}}) + self.query["size"] = max + def all_in_doaj(self): q = deepcopy(self.all_doaj) if self.minified: diff --git a/portality/scripts/journals_update_via_csv.py b/portality/scripts/journals_update_via_csv.py index 298b7c817b..c696068a85 100644 --- a/portality/scripts/journals_update_via_csv.py +++ b/portality/scripts/journals_update_via_csv.py @@ -82,6 +82,7 @@ reader = csv.DictReader(g, fieldnames=header_row) # verify header row with current CSV headers, report errors + # TODO: Include 'Owner' field - but we should probably base this process off the AdminCSV too. expected_headers = JournalFixtureFactory.csv_headers() # Always perform a match check on supplied headers, not counting order @@ -155,6 +156,14 @@ if len(updates) > 0: [print(upd) for upd in updates] + # Check we have the expected owner (if supplied) before proceeding to create an update request + own = row.get('Owner') + if own is not None: + if own.strip().lower() != j.owner.strip().lower(): + print('ABORTING - supplied owner {0} mismatches journal owner {1}.'.format(own, j.owner)) + writer.writerow([j.id, ' | '.join(updates), 'COULD NOT UPDATE - Owner mismatch. Expected {0} Got {1}'.format(own, j.owner)]) + continue + # Create an update request for this journal update_req = None jlock = None @@ -204,7 +213,7 @@ # Add note to UR if supplied if note: - fc.target.add_note(note) + fc.target.add_note(note, author_id=sys_acc.id) if not args.manual_review: # This is the update request, in 'update request' state diff --git a/portality/settings.py b/portality/settings.py index f4c645782c..99b00a4025 100644 --- a/portality/settings.py +++ b/portality/settings.py @@ -9,7 +9,7 @@ # Application Version information # ~~->API:Feature~~ -DOAJ_VERSION = "6.3.16" +DOAJ_VERSION = "6.4.0" API_VERSION = "3.0.1" ###################################### @@ -429,7 +429,7 @@ HUEY_SCHEDULE = { "sitemap": {"month": "*", "day": "*", "day_of_week": "*", "hour": "8", "minute": "0"}, "reporting": {"month": "*", "day": "1", "day_of_week": "*", "hour": "0", "minute": "0"}, - "journal_csv": {"month": "*", "day": "*", "day_of_week": "*", "hour": "*", "minute": "35"}, + "journal_csv": {"month": "*", "day": "*", "day_of_week": "*", "hour": "*", "minute": "20"}, "read_news": {"month": "*", "day": "*", "day_of_week": "*", "hour": "*", "minute": "30"}, "article_cleanup_sync": {"month": "*", "day": "2", "day_of_week": "*", "hour": "0", "minute": "0"}, "async_workflow_notifications": {"month": "*", "day": "*", "day_of_week": "1", "hour": "5", "minute": "0"}, diff --git a/portality/templates/includes/contribution_rates.html b/portality/templates/includes/contribution_rates.html index 29f28a8426..821fefede9 100644 --- a/portality/templates/includes/contribution_rates.html +++ b/portality/templates/includes/contribution_rates.html @@ -260,6 +260,7 @@

For institutions that are members of:

The definition as to Large/Small is at the discretion of the consortium.

+ diff --git a/portality/templates/publisher/preservation.html b/portality/templates/publisher/preservation.html index c87a2fcb97..611eb380af 100644 --- a/portality/templates/publisher/preservation.html +++ b/portality/templates/publisher/preservation.html @@ -25,10 +25,10 @@

Guidance before uploading your file

  1. Only the full text of articles whose metadata is already uploaded to DOAJ can be sent to us. Check that your article metadata appears in DOAJ first.
  2. Only articles for journals indexed in DOAJ can be uploaded. -
  3. Collect the full texts into a package consisting of folders and files.
    4. -
  4. Compress the package into a ZIP file.
    5. -
  5. Upload the zipped package (on this page).
    6. -
  6. Check that the file has uploaded correctly in the History of Uploads section and is not bigger than 50MB.
    7. +
  7. Collect the full text files into a package containing folders and files.
    8. +
  8. Compress the package into a ZIP file. Keep the name of the file simple: avoid spaces, hyphens, underscores, special characters, etc
    9. +
  9. Upload the zipped package (on this page). It may not be bigger than 50MB.
    10. +
  10. Check that the file has uploaded correctly in the History of Uploads section.

The package must have the following structure:

diff --git a/portality/ui/messages.py b/portality/ui/messages.py index ac7f9163bc..8eabd73f80 100644 --- a/portality/ui/messages.py +++ b/portality/ui/messages.py @@ -61,6 +61,7 @@ class Messages(object): EXCEPTION_NO_CONTRIBUTORS_EXPLANATION = "DOAJ requires at least one author for each article." EXCEPTION_TOO_MANY_ISSNS = "Too many ISSNs. Only 2 ISSNs are allowed: one Print ISSN and one Online ISSN." + EXCEPTION_MISMATCHED_ISSNS = "ISSNs provided don't match any journal." EXCEPTION_ISSNS_OF_THE_SAME_TYPE = "Both ISSNs have the same type: {type}" EXCEPTION_IDENTICAL_PISSN_AND_EISSN = "The Print and Online ISSNs supplied are identical. If you supply 2 ISSNs they must be different." EXCEPTION_NO_ISSNS = "Neither Print ISSN nor Online ISSN has been supplied. DOAJ requires at least one ISSN." diff --git a/portality/view/doaj.py b/portality/view/doaj.py index 5ac2f67f69..1292728293 100644 --- a/portality/view/doaj.py +++ b/portality/view/doaj.py @@ -198,8 +198,13 @@ def public_data_dump_redirect(record_type): if not current_user.has_role(constants.ROLE_PUBLIC_DATA_DUMP): abort(404) - target_data = models.Cache.get_public_data_dump().get(record_type, {}) - if target_data is None: + # Make sure the PDD exists + pdd = models.Cache.get_public_data_dump() + if pdd is None: + abort(404) + + target_data = pdd.get(record_type, {}) + if not target_data: abort(404) main_store = store.StoreFactory.get(constants.STORE__SCOPE__PUBLIC_DATA_DUMP) diff --git a/setup.py b/setup.py index a83bf6daf2..2c36468273 100644 --- a/setup.py +++ b/setup.py @@ -5,7 +5,7 @@ setup( name='doaj', - version='6.3.16', + version='6.4.0', packages=find_packages(), install_requires=[ "awscli==1.20.50",