Faster initial indexing for big sets of simple documents.

New build method using meson/ninja.

MacOS: index the Finder tags and comments.

Linux: use libmagic by default, no more executing "file".

Miscellaneous fixes and improvements.

Restructured text tokenizer, easing the integration of external modules.

Support using Jieba to tokenize Chinese text.

GUI: preview uses either webkit or webengine instead of a QTextBrowser

Miscellaneous fixes and improvements.

Full support for external indexers adding data to a Recoll index from a separate script.

Case-insensitive main configuration file parameter names.

Miscellaneous fixes and improvements.

Change librecoll library installation location for better sharing, esp. with the KDE modules.

Multiple small enhancements and fixes.

Option for automatic expansion of probably misspelled queries.

Other small improvements, esp. for processing audio files.

Miscellaneous small improvements.

GUI: new side pane for quick filtering on dates or directories.

Windows: real time indexing.

Was released as 1.31 only because of the new "issub:" query language feature, else it’s only bug fixes…​

New tools to check that things are indexed or why they are not.

Miscellaneous GUI improvements.

Will help with discovering shortcuts (and of course changing them).

Allow selectively hiding most GUI elements, ultimately only keeping the search entry and result areas.

Reorganized menus.

Support external word splitters, allowing more language-sensitive term extraction for Korean.

Windows: Python extension now available, thanks to the MSVC build (forced by the MSVC-built official Python distribution).

Windows: use aspell for spelling suggestions, index Outlook .pst files, and interface with the task scheduler to schedule batch indexing.

All Python scripts now use Python3 by default. They are still supposedly compatible with Python2.

Reworked simple search entry area, with new completion popup.

Windows: the installation now includes an embedded Python3, no need for a separate Python installation.

Will work with any Xapian 1.2 or 1.4 version, but was necessitated by changes in Xapian 1.4.

Stores the document plain texts as part of the index. The storage space needed increases by around 10-20%, but it was necessary for generating snippets with the new Xapian.

New snippets generator, which produces more readable snippets in general, with preserved character case and punctuation.

Partial real-time indexing: it is now possible to specify that only part of a file tree will be monitored for changes. This should mostly alleviate the need to have separate indexes for the batch and real-time data. See the monitordirs configuration variable documentation.

Executing recollindex while a real time indexer is running triggers a batch incremental pass.

All document type handlers are now compatible with Python3 (but still use Python2).

Option to have clickable === links inside the preview window (and generate them for plain text).

Replace a bunch of locally grown hacks with c11 standard features, esp. std::thread, std::mutex etc. As c11 support is now mandatory, get rid of compatibility code (e.g. switches to use c++11 smart pointers or the local version).

Convert the logging facility to use c++ streams instead of stdio.

Allow execm input handlers to set arbitrary data fields. Previously, the only way for handlers to provide metadata was through HTML meta fields.

fn and cfn queryaliases enabled by default.

A new autotools-based build system replaces the old mostly homegrown one for Unix-like platforms.

Recoll was ported to Windows, where the build is based on MinGW and Qt Creator.

New synonyms expansion feature.

A Bison-based query parser replaces the old regexp-based one and allows parenthized sub-expressions and easier future expansions.

Avoid retrying to index previously indexed files if nothing seems to have changed in the filters.

Allow saving queries to files and reload them later. Available both for simple and advanced queries, and based on XML files.

Improve indexing speed by always using vfork() for spawning external commands.

GUI gets "close to system tray" function.

Open With results list popup menu entry.

fieldname:term1,term2 and fieldname:term1/term2 shortcuts for AND/OR searches inside fields.

Query fragments tool.

Better handling of compound terms like mail addresses.

Selection on source collection type (Web history / File system).

Configurable GUI geometry.

Different handling of container file / subdocuments file name searches.

Simultaneous -e -i options to recollindex.

Better indexing performance through multithreading.

Display list of subdocuments (e.g. attachments) for a given result.

Collapsed duplicate results display === link.

Path translation facility (for portable indexes).

Caches last uncompressed file (e.g. for fast compressed mbox access).

Partial recursive reindex option to command line indexer.

Can import tags from external application.

Extended attributes indexing is on by default.

New Python interface for data access. API re-modeled against newer Python Database API 2.0.

Shared librecoll.so.

Optional index configuration for case and diacritics sensitivity.

Advanced search history.

Page-level access when opening PDFs, and snippets window.

Use Xapian Synonyms tables for query expansion.

Language-dependant unaccenting.

GUI dialogs for indexing schedule setup.

Phrase-based _dir:</i> filtering, accepting path fragments. Size filtering.

Python module default install and Unity Lens.

Result list switched to WebKit: drops qt3 support.

Indexing always performed by separate process.

Dynamic category filters (defined as language fragments).

Images displayed in preview window.

Save to file popup in preview.

Fields in advanced search.

Searches anchored to beginning or end of text.

Saving the result table to CSV.

Improved indentation in preview window.

Optional spreadsheet presentation of results.

Clear search history.

Better thunderbird mbox format handling.