- Index the text columns of the market/solution Content Blocks and the
vitec_items/vitec_faq_items collection children; story documents now
include their body (inline tt_content children via
tx_vitec_usecase_content, same per-element extraction as pages) -
stories were indexed with subtitle and teaser only until now
- Upgrade EXT:solr 14.0.0-RC1 -> 14.0.2 final (the Solr image moved to a
fresh volume because the configset changed; synonyms restored via REST)
- News decay with age by publication date: crdate is the import date, so
news map datetime to created; boost function gated to news only
- Results carry `image`: processed 640px tile image per type,
root-relative, null for pages and downloads
- Spec v1.21: 7.16 states the /_json prefix and lists `solution` in the
type vocabulary; footer had been stuck at v1.19
- suggest-demo.html fetches through /_json after the URL swap
Solr 10 dropped the ExtractingRequestHandler, so a dedicated Tika
container on the Solr VPS (Caddy route /tika/*, own basic-auth
credential; Tika itself has no auth) extracts file text during
indexing: TikaDownloadContentIndexer listens on
BeforeDocumentIsProcessedForIndexingEvent, resolves the file through
the existing DownloadFileResolver and appends the text (mime
whitelist, 30 MB cap, 100k chars, fail-soft) to the document's content
field. Extractions are cached in var/tika-cache keyed on
path+size+mtime - a full re-index of 177 downloads drops from 2:12 min
to 27 s, replacing a file re-extracts naturally. Datasheet
specifications ("625i", "genlock") are now searchable. Spec v1.15.
The search JSON gains tab support: the new `filter` GET parameter
restricts results to one document type (product, news, download, story,
page, market); the response carries `filter` (active value or null) and
`facets.type` with per-type counts and active flags. Counts stay
complete while a filter is active (keepAllFacetsOnSelection), so tabs
never collapse - except on an empty filtered result, documented in spec
v1.12 clause 7.16.
The parameter is named `filter` because `type` is TYPO3's reserved
page-type parameter and crashes page resolution; like q and page it is
excluded from cHash validation. Facets come from EXT:solr's native
faceting, serialized generically by SearchJsonRenderer - a future
category facet only needs TypoScript.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Apache Solr 10 runs on a dedicated VPS behind a Caddy HTTPS proxy
(vitecsolr.evomedien.de) because the managed webserver only allows
outgoing standard ports. Credentials stay out of git: the site config
carries %env()% placeholders resolved via putenv() in the git-ignored
config/system/additional.php.
- composer: apache-solr-for-typo3/solr 14.0.0-RC1 (the only line
compatible with TYPO3 14; final 14.0.0 will arrive via composer update)
- config.yaml: read connection https/443, core_en per language,
env placeholder credentials; .gitignore covers additional.php
- setup.typoscript: config.index_enable = 1 (page indexing silently
refuses without it), solr_pi_results JSON renderer registration,
results highlighting, and content field extraction from tt_content
rows - the headless JSON output has no TYPO3SEARCH markers, so the
default page content extraction indexed an empty content field
- SearchJsonRenderer (renderer #27): JSON output for the search plugin
on /search. GET q/page in; { query, page, resultsPerPage, numFound,
totalPages, results[], suggestions } out. Disables the page cache per
request: config.no_cache is gone in TYPO3 v14, and q/page are
excluded from cHash, so cached variants would collide
- ext_localconf.php: q and page added to cacheHash excludedParameters
- SolrIndexCommand (vitec:solr-index): works the index queue from the
CLI with connection diagnostics and a --debug single-step mode -
EXT:solr 14 ships no console commands and the backend button indexes
one item per click
orasche
