Don’t inherit from the default Object for the token dictionary. (davidlukerice - #73)
Throw an error if
Search is instantiated without the required
uidFieldName constructor parameter.
Search uid field can now be an array (for nested/deep keys).
package.json to include correct files.
Performance tuning and removal of eager deopts.
Behind the scenes, this release also includes a rewrite from TypeScript to Flowtype. The external API should not be impacted by this rewrite however.
Fixed (hopefully) previous broken build.
Simple tokenizer now supports cyrillic. (De-Luxis - #21)
Fixed a bug in
TfIdfSearchIndex that caused errors when indexing certain reserved keywords (eg “constructor”).
Fixed tokenizer bug affecting IE <= 10 that caused prefix and substring token strategies incorrectly index terms.
This avoids potential stack overflow for large documents arrays.
Array parameter for nested values.
Search indexing supports non-string values (eg numbers).
Special thanks to @konradjurk for this release.
Small tweak to Node export check to avoid
module is not defined error for browser-based users.
Modified export to better support Node environment (thanks to @scommisso).
ISearchIndex interface in order to support TF-IDF (enabled by default).
IPruningStrategy; it didn’t seem like it added sufficient value to offset performance costs.
Udpated stop-words list to avoid filtering
Refactored stemming and stop-word support to be based on
ITokenizer decorators for better accuracy.
Updated README examples with more info.
JsSearch module wrapper around library and renamed
JsSearch class to
Added stemming support by way of the new
SimpleTokenizer and added better support for punctuation.
StopWordsIndexStrategyDecorator to support stop words filtering.