cgspace-notes/content/posts/2023-09.md

---
title: "September, 2023"
date: 2023-09-02T17:29:36+03:00
author: "Alan Orth"
categories: ["Notes"]
---

## 2023-09-02

- Export CGSpace to check for missing Initiative collection mappings
- Start a harvest on AReS

<!--more-->

## 2023-09-03

- I figured out how to use Altmetric and Dimensions badges in the DSpace Angular frontend
  - It still feels hacky, but using [AfterViewInit](https://stackoverflow.com/questions/41936631/how-to-trigger-the-function-after-dom-markup-is-loaded-in-angular-style-applicat), and importing the Altmetric `embed.js` in the component works
  - The style on mobile also needs work...

## 2023-09-06

- Discussion with Marie about finalizing the output types list on GitHub
  - I did some review and cleanup in preparation for publishing the new list

## 2023-09-07

- Export CGSpace to start doing a review of the metadata
- First I will start by extracting all items with DOIs, along with some fields I can compare against Crossref:

```console
$ csvgrep -c 'cg.identifier.doi[en_US]' -r 'doi.org' ~/Downloads/2023-09-07-cgspace.csv \
    | csvcut -c 'id,dc.title[en_US],dcterms.issued[en_US],dcterms.available[en_US],cg.issn[en_US],cg.isbn[en_US],cg.volume[en_US],cg.issue[en_US],cg.number[en_US],dcterms.extent[en_US],cg.identifier.doi[en_US],cg.reviewStatus[en_US],cg.isijournal[en_US],dcterms.license[en_US],dcterms.accessRights[en_US],dcterms.type[en_US],dc.identifier.uri[en_US]' \
    > /tmp/2023-09-07-cgspace-dois.csv
$ csvgrep -c 'cg.identifier.doi[en_US]' -r 'doi.org' ~/Downloads/2023-09-07-cgspace.csv | csvcut -c 'cg.identifier.doi[en_US]' | sed 1d > /tmp/2023-09-07-cgspace-dois.txt
```

- Then I resolved the DOIs from Crossref:

```console
$ ./ilri/crossref_doi_lookup.py -i /tmp/2023-09-07-cgspace-dois.txt -o /tmp/2023-09-07-cgspace-dois-results.csv -e a.orth@cgiar.org
```

- A user emailed to ask about uploading a 180MB PDF to CGSpace
  - I used GhostScript to try reducing it using the `screen`, `ebook` and `prepress` presets:

```console
$ gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 -dPDFSETTINGS=/screen -dNOPAUSE -dQUIET -dBATCH -sOutputFile=primer-screen.pdf Primer\ \(digital\)_Climate-\ smart\ and\ regenerative\ agriculture\ in\ climate\ change\ adaptation.pdf 
$ gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 -dPDFSETTINGS=/ebook -dNOPAUSE -dQUIET -dBATCH -sOutputFile=primer-ebook.pdf Primer\ \(digital\)_Climate-\ smart\ and\ regenerative\ agriculture\ in\ climate\ change\ adaptation.pdf
$ gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 -dPDFSETTINGS=/prepress -dNOPAUSE -dQUIET -dBATCH -sOutputFile=primer-prepress.pdf Primer\ \(digital\)_Climate-\ smart\ and\ regenerative\ agriculture\ in\ climate\ change\ adaptation.pdf
```

- The `prepress` one is 300DPI and looks visually identical to the original, so I proposed that we use that one

## 2023-09-08

- I did a review of the metadata for our items with DOIs, comparing with data from Crossref
  - I spot checked a handful of issue / online dates and licenses, and saw that Crossref's dates are always more accurate than ours when they differ
  - I also filled in some missing volumes, issues, ISSNs, and extents
  - This results in 14,000 changes to existing items, which will take several days to import unfortunately
  - After eight hours the first file is only about 2/3 finished... sigh
- Meet with Peter to discuss changes to the DSpace 7 test
  - Minor updates to submission forms and some new ideas for the home page and item page
- I figured out how to use a themed home page component and add a cards UI to our CGSpace theme

## 2023-09-09

- I can't believe that almost 18 hours later the first CSV import with 5,000 changes is not done...
- Run all system updates on CGSpace and reboot it, as it had been two months since the last time

## 2023-09-10

- Minor work on the DSpace 7 home page

## 2023-09-11

- Export CGSpace to check for missing Initiative collection mappings
- Start a harvest on AReS

## 2023-09-12

- Minor work on DSpace 7 home page
- Minor work on CG Core types
  - I published a new HTML version of the updated IPtypes and archived the current version as v2.0.0 so we can still reference it

## 2023-09-13

- Stefano reminded me about the updated OAI MODS mappings on CGSpace so I re-applied them on DSpace Test and updated the OAI index so he could confirm
  - Now I'm ready to put it on CGSpace if he confirms
- I created a basic theme for CIP on DSpace 7
  - While doing that I noticed that a bunch of CIP bitstreams didn't have the latest 500px thumbnails so I re-ran filter-media on a handful of their collections
- I had two occurrences of an OOM kill of the Tomcat 9 java process on DSpace 7 test tonight
  - Once while doing a Discovery index, the other while doing filter media

## 2023-09-15

- Discuss issues with the Altmetric API with the Altmetric support team
  - Apparently we can use a different API, the [Explorer API](https://www.altmetric.com/explorer/documentation/api), since we already have access to the Explorer dashboard
- I reduced the Solr heap size on DSpace 7 from 3GB to 2GB
  - Apparentlty I already did this from 4GB to 3GB a few months ago
  - The Solr admin interface was showing Solr taking ~1GB of RAM so I think this should be safe
- Mark on DSpace Slack said he uses PM2's `--max-memory-restart` so the processes restart when they hit the limit
  - Also, he said he had to reduce `cache:serverSide:botCache:max` from 1000 to 500 to cache less SSR pages in memory
- I decided to try deploying DSpace 7 Test on a Hetzner server with 64GB RAM, 6 CPUs, and 2x512GB NVMe SSD

## 2023-09-16

- Export CGSpace to check for missing Initiative collection mappings
- Start a harvest on AReS
- Configure the privacy policy page on DSpace 7 using a themed component with the text from our DSpace 6 site
- I realized that for all my custom Angular components I should be using `routerLink` instead of `href` when I am constructing links
  - The `routerLink` routes within the single page application and saves state, while the `href` reloads the page
  - Using the `routerLink` way is faster and results in less flashing and jumping in the page when navigating
  - See: https://stackoverflow.com/a/61588147

## 2023-09-17

- I added an About page to DSpace 7 Test using similar logic to the privacy page

## 2023-09-18

- I filed a GitHub issue for being unable to navigate dropdown lists using the keyboard on the dspace-angular submission form: https://github.com/DSpace/dspace-angular/issues/2500
- I filed a GitHub issue for the search filters capitalizing metadata values: https://github.com/DSpace/dspace-angular/issues/2501

## 2023-09-19

- Complete migration of DSpace 7 Test from Linode to Hetzner
- Export some years of Solr stats from CGSpace to import on the new DSpace 7 Test:

```console
$ chrt -b 0 ./run.sh -s http://localhost:8081/solr/statistics -a export -o /tmp/statistics-2020-2022.json -f 'time:[2020-01-01T00\:00\:00Z TO 2022-12-31T23\: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
```

- Ben sent me an export of ILRI presentations from Slideshare and asked if we could see if any are missing on CGSpace
  - First I exported CGSpace and extracted the `cg.identifier.url` column so I could normalize all Slideshare URLs to use "https://www.slideshare.net" instead of localized variants (es.slideshare.net, fr.slideshare.net, etc) as well as non-https links and links with query params and slashes at the end
  - This was about 250 URLs
- I extracted the URL field from both our list and the Slideshare list and then used [GNU `join` to print non-matched lines](https://unix.stackexchange.com/questions/274548/join-two-files-each-with-two-columns-including-non-matching-lines):

```console
$ join -t, -v 2 -11 -21 -o auto /tmp/cgspace-ilri-slideshare-sorted-only-urls-sorted.csv /tmp/ilri-slideshare-sorted-sorted.csv | wc -l
542
```

- Important to note that you must use GNU `sort` on the fiels first, as I had tried sorting in vim and it didn't satisfy `join`
- So it seems there are 542 Slideshare presentations we are missing

## 2023-09-20

- Regarding the incorrect city in Solr statistics, I see we have 1,600,000 of them
  - Before filing a GitHub issue, I want to check if they maybe come from an Atmire module, as I see them clustered around two particular CUA versions:

```json
{
  "responseHeader": {
    "status": 0,
    "QTime": 2760,
    "params": {
      "q": "city:com.maxmind.geoip2.record.City*",
      "facet.field": "cua_version",
      "indent": "true",
      "rows": "0",
      "wt": "json",
      "facet": "true",
      "_": "1695192301927"
    }
  },
  "response": {
    "numFound": 1661863,
    "start": 0,
    "docs": []
  },
  "facet_counts": {
    "facet_queries": {},
    "facet_fields": {
      "cua_version": [
        "6.x-4.1.10-ilri-RC7",
        1112186,
        "6.x-4.1.10-ilri-RC5",
        451180,
        "6.x-4.1.10-ilri-RC9",
        0
      ]
    },
    "facet_dates": {},
    "facet_ranges": {},
    "facet_intervals": {}
  }
}
```

- I migrated AReS from Linode to Hetzner
- I asked on Slack and someone told me that we need to edit `src/app/menu.resolver.ts` to add new drop down menus to the top navbar
  - It works, though is unfortunate that we can't do it in a theme

## 2023-09-21

- More minor work on DSpace 7 home page and menus
- Meeting to discuss types and DSpace 7 migration plans
- Create a DSpace 7 theme for IITA

## 2023-09-22

- Create a DSpace 7 theme for IWMI
- I had some issues with pm2 on the new DSpace 7 Test
  - It seems to be due to mixing systemd starting versus manually starting / stopping...
  - After reading the discussion in [this pm2 issue](https://github.com/Unitech/pm2/issues/2914) I realize that we probably need to use `--no-daemon` to have systemd fully manage the processes without pm2 trying to save state

## 2023-09-23

- Export CGSpace to check for missing Initiative collection mappings
- Start a harvest on AReS

## 2023-09-25

- CGSpace metadata and community / collection cleanup
- Review some patches on DSpace Angular
- Create a basic Alliance theme for DSpace 7

## 2023-09-27

- I realized that we can get controlled vocabularies from DSpace 7's REST API, for both value-pairs and hierarchical controlled vocabularies, ie:

https://dspace7test.ilri.org/server/api/submission/vocabularies/common_iso_languages/entries

## 2023-09-29

- Meeting with Aditi and others to discuss plan for using CGSpace to do a systematic review of CGIAR research on climate change
- I cleaned up metadata for a hundred or so items, and realized we will need to do more to make sure abstracts and open access status are correct since there will be a laser focus on the metadata

## 2023-09-30

- Export CGSpace to check for missing Initiative collection mappings
- Still working on checking Unpaywall for access rights and licenses for our DOIs
- Regarding Unpaywall's "evidence" metadata about whether an item is open access or not, after looking at dozens of items manually:
  - evidence: "oa journal (via doaj)"    <---- yes
  - evidence: "open (via free article)"  <---- hmmm, not always correct
  - evidence: "open (via page says license)" <--- noooo, can't rely on that
  - evidence: "open (via page says Open Access)"  <---- yes...?
  - evidence: "open (via free pdf)"      <---- hmmm, not always correct
  - evidence: "oa journal (via publisher name)" <---- noooo
- I updated access status for about four hundred more items based on this, and licenses for a dozen or so

<!-- vim: set sw=2 ts=2: -->