v3.3.6

Created new discord server for development/support discussion! Discord

• Add indexRoot variable to javascript template (#429)
• Update libmupdf to 1.23.4, update libcurl to 8.4.0

v3.3.4

Changes:

• Rework the sist2-admin UX (#389)
• Rework the sist2 index schema, should improve performance for larger datasets
• Implement support for embeddings search (using OpenBLAS for SQLite, and the new KNN query for Elasticsearch)
• Completely rework user scripts (See scripting.md). Uses Python instead of Elasticsearch's "painless" language.

Bug fixes:

• Fix error with incremental search (#406)
• Fix sist2-admin issue when scan fails (#415)
• Fix path error when scanning the entire disk (/) (#426)
• Fix problem where thread pool would not terminate (#416, #397, #399)
• Fix "Do not verify SSL connections to Elasticsearch" (#420)

v3.1.4

Bug fixes:

• Fix stats page (#387)

v3.1.3

Bug fixes!

• Fix thumbnails for PNG and various other media files (#382)
• ARM64 build is functional again (#383)
• Fix wrong encoding when serving files through sist2 frontend (#378)
• Fix wrong date shown in date picker (#376)
• Upgrade mupdf version, fixes various problems when parsing EPUB files (#350)
• Fix DESCENDING search direction pagination when using SQLite search backend (#388)

v3.1.0

New features:

• Polish localisation in web UI (thanks @MarioSob) (#180)
• Thumbnails are saved in WEBP format instead of JPEG (#362)
• Optimization: Only fetch ES aggregations on the first request (#357)
• Optimization: Skip encrypted files when no passphrase is supplied (#355)

Breaking changes:

• --thumbnail-quality parameter now takes values from 0-100 instead of 2-32

v3.0.7

Changes:

• SQLite alternative search backend (#130)
• /f/ endpoint now reads the document path from the sqlite index rather than Elasticsearch

v3.0.6

Bugfixes:

• Fix tesseract search paths (#352)

v3.0.4

New features:

• Add simple web app (sist2-admin) to configure/schedule scan jobs and frontends
• Simpler deployment with docker-compose
• Add support for Auth0 authentication 86ca9f1
• German localisation
• Add configurable "featured fields" line 20adcce
• Use child processes when scanning:
  • Segfaults no longer crash the entire process
  • Much faster ebook scanning (~10x faster with 24 cores & NVMe)
• Add client-side NER support dc39c0e

Breaking changes:

• --thumbnail-quality argument now takes number from 2-31 instead of 1-31
• Index format is now a single SQLite file. Older indices are not compatible with 3.X
• --incremental scan is always done in-place

Bug fixes & tweaks:

• Fix thumbnails in incremental scan c18557e
• Fix mtime ES mapping c0b8a9c
• Various OCR fixes b9f0086
• Various lightbox fixes 9972e21 707bac8
• Rework badge styling 6f75964
• Handle & skip zip bombs when scanning fa14efb
• Fix support for ES 8.X 886fa72
• Fix date picker error 9813646

v2.12.1

New features:

• Add support for Elasticsearch 8.X.X (#266)
• Add --es-insecure-ssl option to skip SSL certificate verification

Bug fixes:

• Fix libmagic mime guessing again
• Remove unused ES settings (#287)

v2.12.0

Breaking changes:

• Reworked how document ID are stored (older index formats are NOT compatible with v2.12.X)

New features:

• Keyboard shortcuts for lightbox (#258)
• Add option to disable CSS animations in the lightbox
• Add filter bar in tag picker (#276)
• Enable highlight in document info modal, remove /d/ endpoint (#281)

Bug fixes:

• Fix raw thumbnails (#259)
• Error handling in get_es_version() (#270)
• Fix query args updating outside of the search page
• Do blank search on page reload when media tab auto-reload is disabled
• Do not store duration or bitrate when the value is 0 or for images
• Fix some segmentation fault errors for large .DOC files
• Fix exec-script & fix memory leak in exec_args_validate (#264)
• Dynamically link libmagic for docker builds, enables mime guessing (#265)
• Fix bug with max_analyzed_offset (#282)

Other:

• Update to mongoose 7.6
• Build libmobi cmake, update to 0.10