Changelog

[8.2.0] - 2024-05-16

Added

Rudimentary plugin system added (will probably be replaced by something better in the future).
Added support for authorization plugins.
Improved caching for /count.
Now aborts searches if client disconnects.
Added CACHE_MAX_QUERY_DATA config variable, setting an optional max size per cached query data file.

Changed

Code was refactored into more manageable pieces.
Now prevents /corpus_config timeout.
Switched to using pymemcache instead of pylibmc.
Speeded up loading of YAML config files.
Optimized word picture SQL query. Now much faster!

Fixed

Fixed crashes during cache cleaning.
Fixed longstanding bug in timespan caching.
Fixed crash when there are no corpus config files.
Fixed crash when trying to access non-existent mode in corpus_config.
Fixed crash when using semicolons at the end of CQP queries.
Fixed bug in /count, where a pipe would be returned for unannotated tokens when top was used.
Fixed caching bug in /count leading to wrong relative total.

[8.1.0] - 2022-09-14

Added

Added /corpus_config endpoint, for serving corpus configuration used by the Korp frontend.
Added per_corpus and combined parameters to /count_time.
Added more information about word picture data to readme.

Fixed

Fixed some crashes related to caching.
Fix bug in /count_time when no corpora are within date range.
Made cache invalidation more reliable.

[8.0.0] - 2019-09-05

Added

Added OpenAPI specification.

Changed

/info has been split into two endpoints: /info and /corpus_info.
New improved /count format.
/loglike parameters group_by and group_by_struct are now optional.
Removed backward compatible parameters for all endpoints.
Better representation of structural attributes on token level in /query.