Demo implementation for migration of ES to SOLR

[RayyanSeliya]

Problem

i have built a small solr demo for the bookbrainz “es -> solr” migration so we can validate the end-to-end approach (schema/analyzers, indexing, and website search queries) before doing a full migration and get to know where we can have a potential errors migrating and tackle those in the upcoming full migration

this pr is for draft demo for review/feedback not for production !

the demo focuses on:

• multilingual search behavior (icu + char mapping)
• work–author relationship lookup + type-filtered results

Solution

this PR adds a runnable solr configset + a minimal solr search implementation for the website.

key parts:

• solr-config/conf/:
  • schema.xml (field types + analyzers)
  • solrconfig.xml (request handlers, including /autocomplete)
  • mapping-chars.txt (pre-token normalization)
• solr setup + indexing:
  • docker-compose.solr.yml
  • setup-solr-with-icu.sh
  • scripts/index-solr-test-data.js
  • demo dataset loader used by the website auto-index path: src/common/helpers/search-test-data.ts
• website integration:
  • src/common/helpers/search-solr.ts (solr-based search logic)
  • src/common/helpers/search-switch.ts (toggles solr vs elastic via USE_SOLR)
  • src/server/app.js + src/server/routes/search.tsx wired to use the switch

notes / constraints:

• solr data for this demo is hardcoded (17 documents).
• autocomplete and fuzzy/typo behavior are not fully “product-ready” yet:
  • autocomplete logic exists (solr /autocomplete handler) but is not integrated into the main website search ui yet
  • fuzzy/typo behavior still needs refinement to be reliable and safe

manual verification

i tested the website search with the demo dataset and recorded a walkthrough video.

tested (high level):

• author searches: lovecraft, tolkien, murakami, austen, asimov
• work searches: cthulhu, lord rings, kafka, pride prejudice, foundation
• edition / identifiers: cthulhu 1928, anniversary, kafka vintage, isbn 978-0-486-27204-8, ids like nm0522454, Q892
• publishers / series / groups: penguin, vintage, foundation series, trilogy, LOTR
• multilingual searches through the website: ラヴクラフト, トールキン, 村上, Остин, 指輪物語
• work–author relationship + type filter: searching lovecraft shows the author and “the call of cthulhu”, and filtering Type: Work narrows results

video:

demo-solr-results.online-video-cutter.com.1.mp4

AI usage

• i did not use any ai
• i have used ai in this pr (drafting parts of the indexing scripts and some code)
  • i used ai for faster drafting, but the query/indexing logic was taken from and mirrored to match the existing solr docs and the current elasticsearch + mb* implementations:
    • metabrainz/sir
    • metabrainz/mbsssss
    • metabrainz/mb-solr
  • i used ai tools for coding
  • i used ai tools for communication
  • i understand all the changes made in this pr

Action

1. i have run the code and manually tested the changes
2. if needed, reviewers can run the demo using:
   1. bash setup-solr-with-icu.sh
   2. USE_SOLR=true ./develop.sh
   3. open http://localhost:9099 and try the search queries listed above

[RayyanSeliya]

ignore the lint errors :)

[faizanakhtar123]

@RayyanSeliya nice