CGSpace Notes

2021-04-01

• I wrote a script to query Sherpa’s API for our ISSNs: sherpa-issn-lookup.py
  • I’m curious to see how the results compare with the results from Crossref yesterday
• AReS Explorer was down since this morning, I didn’t see anything in the systemd journal
  • I simply took everything down with docker-compose and then back up, and then it was OK
  • Perhaps one of the containers crashed, I should have looked closer but I was in a hurry

2021-04-03

• Biruk from ICT contacted me to say that some CGSpace users still can’t log in
  • I guess the CGSpace LDAP bind account is really still locked after last week’s reset
  • He fixed the account and then I was finally able to bind and query:

$ ldapsearch -x -H ldaps://AZCGNEROOT2.CGIARAD.ORG:636/ -b "dc=cgiarad,dc=org" -D "cgspace-account" -W "(sAMAccountName=otheraccounttoquery)"

2021-04-04

• Check the index aliases on AReS Explorer to make sure they are sane before starting a new harvest:

$ curl -s 'http://localhost:9200/_alias/' | python -m json.tool | less

• Then set the openrxv-items-final index to read-only so we can make a backup:

$ curl -X PUT "localhost:9200/openrxv-items-final/_settings" -H 'Content-Type: application/json' -d'{"settings": {"index.blocks.write": true}}' 
{"acknowledged":true}%
$ curl -s -X POST http://localhost:9200/openrxv-items-final/_clone/openrxv-items-final-backup
{"acknowledged":true,"shards_acknowledged":true,"index":"openrxv-items-final-backup"}%
$ curl -X PUT "localhost:9200/openrxv-items-final/_settings" -H 'Content-Type: application/json' -d'{"settings": {"index.blocks.write": false}}'

• Then start a harvesting on AReS Explorer
• Help Enrico get some 2020 statistics for the Roots, Tubers and Bananas (RTB) community on CGSpace
  • He was hitting a bug on AReS and also he only needed stats for 2020, and AReS currently only gives all-time stats
• I cleaned up about 230 ISSNs on CGSpace in OpenRefine
  • I had exported them last week, then filtered for anything not looking like an ISSN with this GREL: isNotNull(value.match(/^\p{Alnum}{4}-\p{Alnum}{4}$/))
  • Then I applied them on CGSpace with the fix-metadata-values.py script:

$ ./ilri/fix-metadata-values.py -i /tmp/2021-04-01-ISSNs.csv -db dspace -u dspace -p 'fuuu' -f cg.issn -t 'correct' -m 253

• For now I only fixed obvious errors like “1234-5678.” and “e-ISSN: 1234-5678” etc, but there are still lots of invalid ones which need more manual work:
  • Too few characters
  • Too many characters
  • ISBNs
• Create the CGSpace community and collection structure for the new Accelerating Impacts of CGIAR Climate Research for Africa (AICCRA) and assign all workflow steps

2021-04-04

• The AReS Explorer harvesting from yesterday finished, and the results look OK, but actually the Elasticsearch indexes are messed up again:

$ curl -s 'http://localhost:9200/_alias/' | python -m json.tool
{
    "openrxv-items-final": {
        "aliases": {}
    },
    "openrxv-items-temp": {
        "aliases": {
            "openrxv-items": {}
        }
    },
...
}

• openrxv-items should be an alias of openrxv-items-final, not openrxv-temp… I will have to fix that manually
• Enrico asked for more information on the RTB stats I gave him yesterday
  • I remembered (again) that we can’t filter Atmire’s CUA stats by date issued
  • To show, for example, views/downloads in the year 2020 for RTB issued in 2020, we would need to use the DSpace statistics API and post a list of IDs and a custom date range
  • I tried to do that here by exporting the RTB community and extracting the IDs for items issued in 2020:

$ ~/dspace63/bin/dspace metadata-export -i 10568/80100 -f /tmp/rtb.csv
$ csvcut -c 'id,dcterms.issued,dcterms.issued[],dcterms.issued[en_US]' /tmp/rtb.csv | \
  sed '1d' | \
  csvsql --no-header --no-inference --query 'SELECT a AS id,COALESCE(b, "")||COALESCE(c, "")||COALESCE(d, "") AS issued FROM stdin' | \
  csvgrep -c issued -m 2020 | \
  csvcut -c id | \
  sed '1d' | \
  sort | \
  uniq

• So I remember in the future, this basically does the following:
  • Use csvcut to extract the id and all date issued columns from the CSV
  • Use sed to remove the header so we can refer to the columns using default a, b, c instead of their real names (which are tricky to match due to special characters)
  • Use csvsql to concatenate the various date issued columns (coalescing where null)
  • Use csvgrep to filter items by date issued in 2020
  • Use csvcut to extract the id column
  • Use sed to delete the header row
  • Use sort and uniq to filter out any duplicate IDs (there were three)
• Then I have a list of 296 IDs for RTB items issued in 2020
• I constructed a JSON file to post to the DSpace Statistics API:

{
  "limit": 100,
  "page": 0,
  "dateFrom": "2020-01-01T00:00:00Z",
  "dateTo": "2020-12-31T00:00:00Z",
  "items": [
"00358715-b70c-4fdd-aa55-730e05ba739e",
"004b54bb-f16f-4cec-9fbc-ab6c6345c43d",
"02fb7630-d71a-449e-b65d-32b4ea7d6904",
...
  ]
}

• Then I submitted the file three times (changing the page parameter):

$ curl -s -d @/tmp/2020-items.txt https://cgspace.cgiar.org/rest/statistics/items | json_pp > /tmp/page1.json
$ curl -s -d @/tmp/2020-items.txt https://cgspace.cgiar.org/rest/statistics/items | json_pp > /tmp/page2.json
$ curl -s -d @/tmp/2020-items.txt https://cgspace.cgiar.org/rest/statistics/items | json_pp > /tmp/page3.json

• Then I extracted the views and downloads in the most ridiculous way:

$ grep views /tmp/page*.json | grep -o -E '[0-9]+$' | sed 's/,//' | xargs | sed -e 's/ /+/g' | bc
30364
$ grep downloads /tmp/page*.json | grep -o -E '[0-9]+,' | sed 's/,//' | xargs | sed -e 's/ /+/g' | bc
9100

• For curiousity I did the same exercise for items issued in 2019 and got the following:
  • Views: 30721
  • Downloads: 10205