cgspace-notes/2023-07.md at 8383cd466bd9e475c36e9fec05f6ddba99427b65

alanorth/cgspace-notes

Fork 0

mirror of https://github.com/alanorth/cgspace-notes.git synced 2024-11-09 16:45:45 +01:00

Alan Orth 293b500b26

content/posts/2023-07.md: minor grammar fix

2023-12-27 10:48:32 +03:00

21 KiB

Raw Blame History

title

date

author

2023-07-01

Export CGSpace to check for missing Initiative collection mappings
Start harvesting on AReS

2023-07-02

Minor edits to the crossref_doi_lookup.py script while running some checks from 22,000 CGSpace DOIs

2023-07-03

I analyzed the licenses declared by Crossref and found with high confidence that ~400 of ours were incorrect
- I took the more accurate ones from Crossref and updated the items on CGSpace
- I took a few hundred ISBNs as well for where we were missing them
- I also tagged ~4,700 items with missing licenses as "Copyrighted; all rights reserved" based on their Crossref license status being TDM, mostly from Elsevier, Wiley, and Springer
- Checking a dozen or so manually, I confirmed that if Crossref only has a TDM license then it's usually copyrighted (could still be open access, but we can't tell via Crossref)
- I would be curious to write a script to check the Unpaywall API for open access status...
- In the past I found that their license status was not very accurate, but the open access status might be more reliable
More minor work on the DSpace 7 item views
- I learned some new Angular template syntax
- I created a custom component to show Creative Commons licenses on the simple item page
- I also decided that I don't like the Impact Area icons as a component because they don't have any visual meaning

2023-07-04

Focus group meeting with CGSpace partners about DSpace 7
I added a themed file selection component to the CGSpace theme
- It displays the bistream description instead of the file name, just like we did in DSpace 6 XMLUI
I added a custom component to show share icons

2023-07-05

I spent some time trying to update OpenRXV from Angular 9 to 10 to 11 to 12 to 13
- Most things work but there are some minor bugs it seems
Mishell from CIP emailed me to say she was having problems approving an item on CGSpace
- Looking at PostgreSQL I saw there were a dozen or so locks that were several hours and even over one day old so I killed those processes and told her to try again

2023-07-06

Types meeting
I wrote a Python script to check Unpaywall for some information about DOIs

2023-07-7

Continue exploring Unpaywall data for some of our DOIs
- In the past I've found their licensing information to not be very reliable (preferring Crossref), but I think their open access status is more reliable, especially when the provider is listed as being the publisher
- Even so, sometimes the version can be "acceptedVersion", which is presumably the author's version, as opposed to the "publishedVersion", which means it's available as open access on the publisher's website
- I did some quality assurance and found ~100 that were marked as Limited Access, but should have been Open Access, and fixed a handful of licenses
Delete duplicate metadata as described in my DSpace issue from last year: https://github.com/DSpace/DSpace/issues/8253
Start working on some statistics on AGROVOC usage for my presenation next week
- I used the following SQL query to dump values from all subject fields and lower case them:

localhost/dspacetest= ☘ \COPY (SELECT DISTINCT(lower(text_value)) AS "subject" FROM metadatavalue WHERE dspace_object_id in (SELECT dspace_object_id FROM item) AND metadata_field_id IN (187, 120, 210, 122, 215, 127, 208, 124, 128, 123, 125, 135, 203, 236, 238, 119)) to /tmp/2023-07-07-cgspace-subjects.csv WITH CSV HEADER;
COPY 26443
Time: 2564.851 ms (00:02.565)

Then I extracted the subjects and looked them up against AGROVOC:

$ csvcut -c subject /tmp/2023-07-07-cgspace-subjects.csv | sed '1d' > /tmp/2023-07-07-cgspace-subjects.txt
$ ./ilri/agrovoc_lookup.py -i /tmp/2023-07-07-cgspace-subjects.txt -o /tmp/2023-07-07-cgspace-subjects-results.csv

I did some more tests with Angular 13 on OpenRXV and found out why the repository type dropdown wasn't working
- It was because of a missing 1-line JSON file in the data directory, which is runtime data, not code
- I copied the data directory from the production serve and rebuild and the site is working well now
- I did a full harvest with plugins and it worked!
- So it seems Angular 13.4.0 will work, yay

2023-07-08

Export CGSpace to check for missing Initiative collection mappings
- Start a harvest on AReS
The AGROVOC lookup finished, so I checked the number of matches:

$ csvgrep -c 'match type' -r '^.+$' ~/Downloads/2023-07-07-cgspace-subjects-resolved.csv | sed 1d | wc -l
12528

So that's 12,528 out of 26,443 unique terms (47.3%)
I did a LOT of work on the OpenRXV frontend build dependencies to bring more in line with Angular 13

2023-07-10

I did a lot more work on OpenRXV to test and update dependencies
I deployed the latest version on the production server

2023-07-12

CGSpace upgrade meeting with Americas and Africa group

2023-07-13

Michael Victor asked me to help Aditi extract some information from CGSpace
- She was interested in journal articles published between 2018 and 2023 with a range of subjects related to drought, flooding, resilience, etc
- I used an advanced query with some AGROVOC terms:

dcterms.issued:[2018 TO 2023] AND dcterms.type:"Journal Article" AND (dcterms.subject:flooding OR dcterms.subject:flood OR dcterms.subject:"extreme weather events" OR dcterms.subject:drought OR dcterms.subject:"drought resistance" OR dcterms.subject:"drought tolerance" OR dcterms.subject:"soil salinity" OR dcterms.subject:"pests of plants" OR dcterms.subject:pests OR dcterms.subject:heat OR dcterms.subject:fertilizers OR dcterms.subject:"fertilizer technology" OR dcterms.subject:"rice fields" OR dcterms.subject:"landscape conservation" OR dcterms.subject:"landscape restoration" OR dcterms.subject:livestock)

Interestingly, some variations of this same exact query produce no search results, and I see this error in the DSpace log:

org.dspace.discovery.SearchServiceException: org.apache.solr.search.SyntaxError: Cannot parse 'dcterms.issued:[2018 TO 2023] AND dcterms.type:"Journal Article" AND (dcterms.subject:flooding OR dcterms.subject:flood OR dcterms.subject:"extreme weather events" OR dcterms.subject:drought OR dcterms.subject:"drought resistance" OR dcterms.subject:"drought tolerance" OR dcterms.subject:"soil salinity" OR dcterms.subject:"pests of plants" OR dcterms.subject:pests OR dcterms.subject:heat OR dcterms.subject:fertilizers OR dcterms.subject:"fertilizer technology" OR dcterms.subject:"rice fields" OR dcterms.subject:livestock OR dcterms.subject:"landscape conservation" OR dcterms.subject:"landscape restoration\"\)': Lexical error at line 1, column 617.  Encountered: <EOF> after : "\"landscape restoration\\\"\\)"

It seems to be when there is a quoted search term at the end of the parenthesized group
- For what it's worth this same query worked fine on DSpace 7.6

2023-07-15

Export CGSpace to fix missing Initiative collection mappings
Start a harvest on AReS

2023-07-17

Rasika had sent me a list of new ORCID identifiers for new IWMI staff so I combined them with our existing list and ran resolve_orcids.py to refresh the names in our database
- I updated the list, updated names in the database, and tagged new authors with missing identifiers in existing items

2023-07-18

Meeting with IWMI, IRRI, and IITA colleagues about CGSpace upgrade plans
Maria from the Alliance mentioned having some submissions stuck on CGSpace
- I looked and found a number of locks stuck for many nineteen, eighteen, and more hours...
- I killed them and told her to try again

$ psql < locks-age.sql | less -S
$ psql < locks-age.sql | grep -E " (19|18|17|16|12):" | awk -F"|" '{print $10}' | sort -u | xargs kill

2023-07-19

I had to kill a bunch more locked processes in PostgreSQL, I'm not sure what's going on
After some discussion about an advanced search bug with Tim on Slack, I filed an issue on GitHub

2023-07-20

I added a new metadata field for CGIAR Impact Platforms (cg.subject.impactPlatform) to CGSpace

2023-07-22

Export CGSpace tp fix missing Initiative collections
Start a harvest on AReS

2023-07-24

Test Salem's new JavaScript-based DSpace Statistics API and send him some feedback
I noticed a few times that the Solr service on my DSpace 7 instance is getting OOM killed
- I had been using a 4g Solr heap, but maybe we don't need that much
- Tomcat is also using 4.6GB, and then there's PostgreSQL... so perhaps it's all a bit much on this system now

2023-07-25

Start testing exporting DSpace 6 Solr cores to import on DSpace 7:

$ chrt -b 0 dspace solr-export-statistics -i statistics

I'm curious how long it takes and how much data there will be
- The size of the Solr data directory is currently 82GB
- The export took about 2.5 hours and created 6,000 individual CSVs, one for each day of Solr stats
- The size of the exported CSVs is about 88GB
- I will copy just a few years to import on the DSpace 7 test server
- So importing these is going to require removing the Atmire custom fields:

$ dspace solr-import-statistics -i statistics
Exception: Error from server at http://localhost:8983/solr/statistics: ERROR: [doc=1a92472e-e39d-4602-9b4d-da022df8f233] unknown field 'containerCommunity'
org.apache.solr.client.solrj.impl.HttpSolrClient$RemoteSolrException: Error from server at http://localhost:8983/solr/statistics: ERROR: [doc=1a92472e-e39d-4602-9b4d-da022df8f233] unknown field 'containerCommunity'
        at org.apache.solr.client.solrj.impl.HttpSolrClient.executeMethod(HttpSolrClient.java:681)
        at org.apache.solr.client.solrj.impl.HttpSolrClient.request(HttpSolrClient.java:266)
        at org.apache.solr.client.solrj.impl.HttpSolrClient.request(HttpSolrClient.java:248)
        at org.apache.solr.client.solrj.SolrClient.request(SolrClient.java:1290)
        at org.dspace.util.SolrImportExport.importIndex(SolrImportExport.java:465)
        at org.dspace.util.SolrImportExport.main(SolrImportExport.java:148)
        at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
        at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:77)
        at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
        at java.base/java.lang.reflect.Method.invoke(Method.java:568)
        at org.dspace.app.launcher.ScriptLauncher.runOneCommand(ScriptLauncher.java:277)
        at org.dspace.app.launcher.ScriptLauncher.handleScript(ScriptLauncher.java:133)
        at org.dspace.app.launcher.ScriptLauncher.main(ScriptLauncher.java:98)

I will try using solr-import-export-json, which I've used in the past to skip Atmire custom fields in Solr:

$ chrt -b 0 ./run.sh -s http://localhost:8081/solr/statistics -a export -o /tmp/statistics-2022.json -f 'time:[2022-01-01T00\:00\:00Z TO 2022-12-31T23\:59\:59Z]' -k uid -S author_mtdt,author_mtdt_search,iso_mtdt_search,iso_mtdt,subject_mtdt,subject_mtdt_search,containerCollection,containerCommunity,containerItem,countryCode_ngram,countryCode_search,cua_version,dateYear,dateYearMonth,geoipcountrycode,geoIpCountryCode,ip_ngram,ip_search,isArchived,isInternal,isWithdrawn,containerBitstream,file_id,referrer_ngram,referrer_search,userAgent_ngram,userAgent_search,version_id,complete_query,complete_query_search,filterquery,ngram_query_search,ngram_simplequery_search,simple_query,simple_query_search,range,rangeDescription,rangeDescription_ngram,rangeDescription_search,range_ngram,range_search,actingGroupId,actorMemberGroupId,bitstreamCount,solr_update_time_stamp,bitstreamId,core_update_run_nb

Some users complained that CGSpace was slow and I found a handful of locks that were hours and days old...
- I killed those and told them to try again
After importing the Solr statistics into DSpace 7 I realized that my DSpace Statistics API will work fine
- I made some minor modifications to the Ansible infrastructure scripts to make sure it is enabled and then activated it on DSpace 7 Test

2023-07-26

Debugging lock issues on CGSpace
- I see the blocking PIDs for some long-held locks are "idle in transaction":

$ ps auxw | grep -E "(1864132|1659487)"
postgres 1659487  0.0  0.5 3269900 197120 ?      Ss   Jul25   0:03 postgres: 14/main: cgspace cgspace 127.0.0.1(61648) idle in transaction
postgres 1864132  0.1  0.7 3275704 254528 ?      Ss   07:27   0:08 postgres: 14/main: cgspace cgspace 127.0.0.1(36998) idle in transaction
postgres 1880388  0.0  0.0   9208  2432 pts/3    S+   08:48   0:00 grep -E (1864132|1659487)

I used some other scripts and found that those processes were executing the following statement:

select nextval ('public.tasklistitem_seq')

I don't know why these can get blocked for hours without resolution, but for now I just killed them
- For what it's worth these sequences were removed in DSpace 7.0 along with the "traditional" item workflow—maybe that means we won't have such contention issues in DSpace 7!
I wrote a slightly longer regex to match locks that have been stuck for more than 1 hour based on the output of the locks-age.sql script and killed them:

$ psql < locks-age.sql | awk -F"|" '/ [[:digit:]][1-9]:[[:digit:]]{2}:[[:digit:]]{2}\./ {print $10}' | sort -u | xargs kill

I filed an issue for missing Altmetric badges on DSpace 7 Angular

2023-07-27

Export CGSpace to check countries, regions, types, and Initiatives
- There were a few minor issues in countries and regions, and I noticed 186 items without types!
- Then I ran the file through csv-metadata-quality to make sure items with countries have appropriate regions
Brief discussion about OpenRXV bugs and fixes with Moayad
I was toying with the idea of using an expanded whitespace check/fix based on ESLint's no-irregular-whitespace rule in csv-metadata-quality
- I found 176 items in CGSpace with such whitespace in their titles alone
- I compared the results of removing these characters and replacing them with a space
- In most cases removing it is the correct thing to do, for example "Pesticides : une arme à double tranchant" → "Pesticides: une arme à double tranchant"
- But in some items it is tricky, for example "L'environnement juridique est-il propice à la gestion" → "L'environnement juridique est-il propice àla gestion"
- I guess it would really need some good heuristics or a human to verify...
I upgraded OpenRXV to Angular v14

2023-07-28

After a bit more testing I merged the Angular v14 changes to OpenRXV master
I am getting an error trying to import the 2020 Solr statistics from CGSpace to DSpace 7:

Exception in thread "main" org.apache.solr.client.solrj.impl.BaseHttpSolrClient$RemoteSolrException: Error from server at http://localhost:8983/solr/statistics: ERROR: [doc=0008a7c1-e552-4a4e-93e4-4d23bf39964b] Error adding field 'workflowItemId'='0812be47-1bfe-45e2-9208-5bf10ee46f81' msg=For input string: "0812be47-1bfe-45e2-9208-5bf10ee46f81"
        at org.apache.solr.client.solrj.impl.HttpSolrClient.executeMethod(HttpSolrClient.java:745)
        at org.apache.solr.client.solrj.impl.HttpSolrClient.request(HttpSolrClient.java:259)
        at org.apache.solr.client.solrj.impl.HttpSolrClient.request(HttpSolrClient.java:240)
        at org.apache.solr.client.solrj.SolrRequest.process(SolrRequest.java:234)
        at org.apache.solr.client.solrj.SolrClient.add(SolrClient.java:102)
        at org.apache.solr.client.solrj.SolrClient.add(SolrClient.java:69)
        at org.apache.solr.client.solrj.SolrClient.add(SolrClient.java:82)
        at it.damore.solr.importexport.App.insertBatch(App.java:295)
        at it.damore.solr.importexport.App.lambda$writeAllDocuments$10(App.java:276)
        at it.damore.solr.importexport.BatchCollector.lambda$accumulator$0(BatchCollector.java:71)
        at java.base/java.util.stream.ReduceOps$3ReducingSink.accept(ReduceOps.java:169)
        at java.base/java.util.Iterator.forEachRemaining(Iterator.java:133)
        at java.base/java.util.Spliterators$IteratorSpliterator.forEachRemaining(Spliterators.java:1845)
        at java.base/java.util.stream.AbstractPipeline.copyInto(AbstractPipeline.java:509)
        at java.base/java.util.stream.AbstractPipeline.wrapAndCopyInto(AbstractPipeline.java:499)
        at java.base/java.util.stream.ReduceOps$ReduceOp.evaluateSequential(ReduceOps.java:921)
        at java.base/java.util.stream.AbstractPipeline.evaluate(AbstractPipeline.java:234)
        at java.base/java.util.stream.ReferencePipeline.collect(ReferencePipeline.java:682)
        at it.damore.solr.importexport.App.writeAllDocuments(App.java:252)
        at it.damore.solr.importexport.App.main(App.java:150)

Ahhhh, in DSpace 6 this field was a string in the Solr statistics schema, but in DSpace 7 it is an integer...?
- Oh, it seems to be an Atmire change in our DSpace 6... hmmm, so we need to ignore the workflowItemId field when exporting
  - Upstream: https://github.com/DSpace/DSpace/blob/dspace-6_x/dspace/solr/statistics/conf/schema.xml#L328
  - ILRI: https://github.com/ilri/DSpace/blob/6_x-prod/dspace/solr/statistics/conf/schema.xml#L344
- I am wondering if we can skip all these workflow fields since I don't think we are using any aspects of statistics related to workflows
- I diffed our Solr statistics schema with the one from vanilla DSpace 6 and got a list of all the fields that were different:

isInternal,workflowItemId,containerCommunity,containerCollection,containerItem,containerBitstream,dateYear,dateYearMonth,filterquery,complete_query,simple_query,complete_query_search,simple_query_search,ngram_query_search,ngram_simplequery_search,text,storage_statistics_type,storage_size,storage_nb_of_bitstreams,name,first_name,last_name,p_communities_id,p_communities_name,p_communities_map,p_group_id,p_group_name,p_group_map,group_id,group_name,group_map,parent_count,bitstreamId,bitstreamCount,actingGroupId,actorMemberGroupId,actingGroupParentId,rangeDescription,range,version_id,file_id,cua_version,core_update_run_nb,orphaned

I will combine it with the other fields I was skipping above and try the export again:

$ chrt -b 0 ./run.sh -s http://localhost:8081/solr/statistics -a export -o /tmp/statistics-2020.json -f 'time:[2020-01-01T00\:00\:00Z TO 2020-12-31T23\:59\:59Z]' -k uid -S actingGroupId,actingGroupParentId,actorMemberGroupId,author_mtdt,author_mtdt_search,bitstreamCount,bitstreamId,complete_query,complete_query_search,containerBitstream,containerCollection,containerCommunity,containerItem,core_update_run_nb,countryCode_ngram,countryCode_search,cua_version,dateYear,dateYearMonth,file_id,filterquery,first_name,geoipcountrycode,geoIpCountryCode,group_id,group_map,group_name,ip_ngram,ip_search,isArchived,isInternal,iso_mtdt,iso_mtdt_search,isWithdrawn,last_name,name,ngram_query_search,ngram_simplequery_search,orphaned,parent_count,p_communities_id,p_communities_map,p_communities_name,p_group_id,p_group_map,p_group_name,range,rangeDescription,rangeDescription_ngram,rangeDescription_search,range_ngram,range_search,referrer_ngram,referrer_search,simple_query,simple_query_search,solr_update_time_stamp,storage_nb_of_bitstreams,storage_size,storage_statistics_type,subject_mtdt,subject_mtdt_search,text,userAgent_ngram,userAgent_search,version_id,workflowItemId

Export a list of affiliations from the Initiatives community for Peter:

$ dspace metadata-export -i 10568/115087 -f /tmp/2023-07-28-initiatives.csv
$ csvcut -c 'cg.contributor.affiliation[en_US]' ~/Downloads/2023-07-28-initiatives.csv \
  | sed -e 1d -e 's/^"//' -e 's/"$//' -e 's/||/\n/g' -e '/^$/d'            \
  | sort | uniq -c | sort -hr                                              \
  | awk 'BEGIN { FS = "^[[:space:]]+[[:digit:]]+[[:space:]]+" } {print $2}'\
  | sed -e '1i cg.contributor.affiliation' -e 's/^\(.*\)$/"\1"/'           \
  > /tmp/2023-07-28-initiatives-affiliations.csv

This is a method I first used in 2023-01 to export affiliations ONLY used in items in the Initiatives community
- I did the same for authors and investors

2023-07-29

Export CGSpace to look for missing Initiative collection mappings
I found a bunch of locks waiting for many hours and killed them:

$ psql < locks-age.sql | awk -F"|" '$9 ~ / [[:digit:]][1-9]:[[:digit:]]{2}:[[:digit:]]{2}\./ {print $10}' | sort -u | xargs kill

This looks for a pattern matching something like 11:30:48.598436 in the age column (not 00:00:00) and kills them
Start a harvest on AReS

21 KiB Raw Blame History

2023-07-01

2023-07-02

2023-07-03

2023-07-04

2023-07-05

2023-07-06

2023-07-7

2023-07-08

2023-07-10

2023-07-12

2023-07-13

2023-07-15

2023-07-17

2023-07-18

2023-07-19

2023-07-20

2023-07-22

2023-07-24

2023-07-25

2023-07-26

2023-07-27

2023-07-28

2023-07-29

21 KiB

Raw Blame History