Skip to content

Bump dependencies and upgrade to Tika 4.1.0 - #8141

Open
labkey-adam wants to merge 8 commits into
developfrom
fb_dependencies_oct2026
Open

labkey-adam wants to merge 8 commits into
developfrom
fb_dependencies_oct2026

Conversation

@labkey-adam

@labkey-adam labkey-adam commented Oct 6, 2026 •

Copy link
Copy Markdown
Contributor

Rationale

Latest versions are best

Changes

  • Update many dependencies to their latest versions
  • Deprecate support for PostgreSQL 14.x since it's reaching end-of-life soon (Nov 12). Test deprecation bit on PG versions.
  • Update Tika 3.3.2 -> 4.1.0. Major upgrade entails significant configuration, logging, and API changes.
  • Add support for .eml and .sas7bdat files

@labkey-adam labkey-adam changed the title Bump dependencies Bump dependencies and upgrade to Tika 4.1.0 Oct 7, 2026
@labkey-adam labkey-adam self-assigned this Oct 7, 2026
@labkey-adam

labkey-adam commented Oct 7, 2026 •

Copy link
Copy Markdown
Contributor Author

Tika parser coverage: what was added and what was deliberately left out

Added to the search module:

  • tika-parser-mail-commons: shared email utilities that the mail module below requires. Without it, .eml and mbox parsing fail with NoClassDefFoundError.
  • tika-parser-mail-module: indexes .eml and mbox email. No new third-party libraries, since mime4j was already shipped.
  • parso: enables Tika's SAS .sas7bdat parser, which we already ship. SAS datasets are common in clinical and biomedical research.

TikaTestCase now covers .eml (including To/Cc indexing) and .sas7bdat using new sample files in the testAutomation PR, so the two PRs must merge together.

Not added: parsers we ship whose libraries we deliberately omit. LuceneSearchServiceImpl excludes them so they're never loaded, and any other parser that fails to load is still logged.

  • MATLAB .mat (jmatio): Tika identifies many non-MATLAB .mat files as MATLAB, so enabling it would add parse errors.
  • RAR (junrar): archive expansion cost and a history of denial-of-service CVEs.
  • Access .mdb/.accdb (jackcess): about 1.4 MB for limited demand.
  • Outlook .pst (java-libpst): entire mailboxes in one file, with size and privacy concerns.

Not added: Tika modules we don't ship.

  • Scientific (HDF, NetCDF, GRIB): a heavy dependency tree (netCDF, Apache SIS, GeoAPI, JAXB) for mostly metadata; see Issue 38386.
  • SQLite: an 11.7 MB native driver.
  • OCR: requires Tesseract installed on every server.
  • Audio/video, fonts, CAD, RSS, web archives and S/MIME: metadata-only or niche for our users.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant