Gitiles
Code Review Sign In
korap.ids-mannheim.de / KorAP / deliko / 105913d26f2c1fcf9fdf00a2c125d6f5d1f2620e
commit105913d26f2c1fcf9fdf00a2c125d6f5d1f2620e[log] [tgz]
authorMarc Kupietz <kupietz@ids-mannheim.de>Thu May 09 21:34:13 2024 +0200
committerMarc Kupietz <kupietz@ids-mannheim.de>Thu May 09 21:34:13 2024 +0200
tree8270859c52a167510d3d2427c014d6ea57ffafef
parent54ab7e5da863a8cdd646637e80937548ce8fd213 [diff]
CI: skip index build and deploy jobs on merge requests
1 file changed
tree: 8270859c52a167510d3d2427c014d6ea57ffafef
  1. html/
  2. lib/
  3. scripts/
  4. test/
  5. xslt/
  6. .gitignore
  7. .gitlab-ci.yml
  8. korap4dnb-compose.yml
  9. krill-korap4dnb.cfg
  10. Makefile
  11. Readme.md
Readme.md

EPub to KorAP (via TEI I5) conversion

Run

Generate an I5 corpus from the included test data

make -j $(nproc) target/dnb18.i5.xml SRC_DIR=test/resources/DNB YEARS=18

Generate all I5 corpora

make -j $(nproc) i5  SRC_DIR=test/resources/DNB

Generate a KorAP-XML ZIP

Prerequisite: KorAP-XML-CoNLL-U

make -j $(nproc) target/dnb18.zip SRC_DIR=test/resources/DNB YEARS=18

Run KorAP

Adjust the following line in your korap4dnb-compose.yml to point to your index (it is in target/dnb.index by default, but should better be copied to a safe place):

      - "${PWD}/target/dnb.index:/kustvakt/index:z"

and start the docker:

docker compose -p korap4dnb --profile=lite -f korap4dnb-compose.yml up -d

Stop KorAP

docker compose -p korap4dnb down

To generate Annotations

Install prerequisite korap/conllu2treetagger and korap/conllu2spacy docker images if not present:

docker image inspect korap/conllu2treetagger:latest || curl -Ls 'https://gitlab.ids-mannheim.de/KorAP/CoNLL-U-Treetagger/-/jobs/artifacts/master/raw/conllu2treetagger.xz?job=build-docker-image' | docker load

docker image inspect korap/conllu2spacy:latest || curl -Ls https://corpora.ids-mannheim.de/tools/conllu2spacy.tar.xz | docker load

Make annotations fro dnb20:

make -j $(nproc) target/dnb20.marmot-malt.zip target/dnb20.spacy.zip target/dnb20.tree_tagger.zip

To build KorAP index (also directly)

Build KorAP all, up to the deployable index:

make -j $(nproc) all

News

  • 2024-04-19

    • SRC_DIR now defaults to the production sample!
    • ISBN number recognition should be fixed now
    • ignore faulty xhtml input files and conversion errors – just issue a warning
    • docker compose now uses http default port 80 externally

  • 2024-04-15

    • added pass2 and pass3 to xslt conversion to …
      • fix div, p, hi, ref … nestings
      • remove empty elements
      • join subsequent hi elements
    • improved korapxml2krill performance by using all cores (-1 does not work here)
    • sanitized the Makefile and dropped YY variable, use YEARS instead

  • 2024-04-10

    • multiple authors (and non-authors) are now correctly handled
    • some more .(x)html files are now dropped (toc, cover, etc.)
    • PRELIMINARY support for splitting everything into annual volumes
      • use make YY=22 to select 2022
      • does not yet work for the index!

  • 2024-03-24

    • slow udpipe2 dropped
    • added marmot POS and morpho-syntactic annotations
    • added malt dependency annotations

  • 2024-03-18

    • added make deploy to install new index and restart local KorAP@DNB instance (also available as ci target)
    • added show-server-logs and show-server-status make targets to monitor the local KorAP@DNB instance

  • 2024-03-17

    • added make all to build all targets, including the index

  • 2024-03-16

    • CI/CD pipeline added
    • first working pipeline for EPub ⮕ TEI I5 ⮕ KorAP-XML ⮕ (UDPipe+TreeTagger+Spacy) ⮕ Krill ⮕ KorAP-JSON

  • 2024-03-15: DNB test data added

  • 2024-03-08: example EPub and I5 added from DeReKo KJL corpus: Christiane F. ; Kai Hermann ; Horst Rieck: Wir Kinder vom Bahnhof Zoo in the folder test/resources/ – do not distribute (copyrighted data)