Next release uk language-packs check up #129

Proposal:

Follow-up to #127.

Before the next Docker release, verify that the release Docker ref/package set is prepared from current master after:

• fixed docker build to use manticore-language-packs with uk pack data #128
• merge commit: 7a460302615ef799d994a96d7703aae236822f1e

The goal is to make sure the next release image does not reintroduce the legacy Ukrainian lemmatizer stack that was removed from master.

Background

Docker release 25.0.0 was made from a release ref/tag, not directly from current master.

For 25.0.0, the release Dockerfile used:

manticore_25.0.0-26032712-ce3c27828__ARCH_64.deb
manticore-lemmatizer-uk_1.0.3__ARCH_64.deb

That was correct for that release at the time, but it means the next release preparation step must not copy the old manticore-lemmatizer-uk line forward.

The current Docker master has removed:

• manticore-lemmatizer-uk
• /usr/local/lib/manticore/lemmatize_uk.so
• Python 3.9
• pymorphy2
• pymorphy2-dicts-uk
• PYTHONWARNINGS=ignore::UserWarning:pymorphy2.analyzer

The native Ukrainian lemmatizer is provided by Manticore itself and requires uk.pak.

Relevant fixes:

• Built-in Ukrainian lemmatizer:

  • feat: built-in Ukrainian lemmatizer prototype manticoresearch#4414
  • merge commit: af564220b42ee8709bf811a8794570b34d30aa24

• Standalone language-packs package now includes uk.pak:

  • publish language packs to release and dev repos; added uk pack manticoresearch-language-packs#5
  • merge commit: eadde16d4108ea19b9f3b654d5219305e9d9d0ef
  • fixed package: manticore-language-packs 1.0.15-260430-eadde16

• Docker current/dev image removed the old stack:

  • fixed docker build to use manticore-language-packs with uk pack data #128
  • merge commit: 7a460302615ef799d994a96d7703aae236822f1e

Release Preparation Checks

When preparing the next Docker release ref/tag from master, inspect the release Dockerfile:

git show <new-release-ref>:Dockerfile | rg 'manticore-lemmatizer-uk|python3\.9|pymorphy2|PYTHONWARNINGS'

Expected result:

(no output)

If the release Dockerfile uses the Manticore bundle package, it should use only the new bundle package, for example:

manticore_<next-version>-<date>-<commit>__ARCH_64.deb

and must not add:

manticore-lemmatizer-uk_1.0.3__ARCH_64.deb

If the release Dockerfile uses split packages instead of the bundle package, the package set must include:

manticore-language-packs >= 1.0.15

or another package that owns:

/usr/share/manticore/uk.pak

Release Image Validation

Build the release/default image from the release ref:

docker build \
  --build-arg DEV=0 \
  --platform linux/amd64 \
  -t manticoresoftware/manticore:release-uk-check \
  .

Verify image contents:

docker run --rm --entrypoint sh manticoresoftware/manticore:release-uk-check -lc '
  set -e
  test -f /usr/share/manticore/uk.pak
  dpkg -S /usr/share/manticore/uk.pak
  ! dpkg -s manticore-lemmatizer-uk >/dev/null 2>&1
  test ! -e /usr/local/lib/manticore/lemmatize_uk.so
  ! command -v python3.9 >/dev/null 2>&1
  if find /usr/local/lib /usr/lib \( -iname "*pymorphy2*" -o -iname "*pymorphy2-dicts-uk*" -o -iname "*pymorphy2_dicts_uk*" \) 2>/dev/null | grep -q .; then
    echo "legacy pymorphy2 files found"
    exit 1
  fi
'

Note: use dpkg -S /usr/share/manticore/uk.pak, not only dpkg -L manticore-language-packs, because in a bundle-based release image uk.pak may be owned by the manticore bundle package.

Verify runtime Ukrainian morphology:

docker rm -f manticore-release-uk-check >/dev/null 2>&1 || true

docker run -d \
  --name manticore-release-uk-check \
  manticoresoftware/manticore:release-uk-check

timeout 60 sh -c '
  until docker logs manticore-release-uk-check 2>&1 | grep -q "accepting connections"; do
    sleep 1
  done
'

docker exec manticore-release-uk-check mysql -h0 -P9306 -e "
  CREATE TABLE test_uk (
    id bigint,
    content text
  )
  rt_mem_limit = '256M'
  morphology = 'lemmatize_uk'
  charset_table = '0..9, A..Z->a..z, _, a..z, U+0410..U+042F->U+0430..U+044F, U+0430..U+044F, U+0454, U+0456, U+0457, U+0491';
"

docker exec manticore-release-uk-check mysql -h0 -P9306 -e "
  INSERT INTO test_uk (id, content) VALUES (1, 'бігаю');
"

docker exec manticore-release-uk-check mysql -h0 -P9306 -e "
  CALL KEYWORDS('бігаю', 'test_uk');
"

docker rm -f manticore-release-uk-check

Expected normalized form:

бігати

Acceptance Criteria

• Release Dockerfile does not reference manticore-lemmatizer-uk.
• Release Dockerfile does not install Python 3.9 / pymorphy2 / pymorphy2-dicts-uk.
• Final release image contains /usr/share/manticore/uk.pak.
• /usr/share/manticore/uk.pak is owned by an installed package.
• Final release image does not contain /usr/local/lib/manticore/lemmatize_uk.so.
• morphology = 'lemmatize_uk' works in the release image.
• The same checks pass for expected release architectures, at least linux/amd64 and linux/arm64.

Notes

The current/dev Docker path and downstream test-kit path were already validated after #128. This ticket is only for the next release preparation/release image check.

Checklist:

^{To be completed by the assignee. Check off tasks that have been completed or are not applicable.}