1.1.2

Add IntegerSeparatorTokenizer and DecimalSeparatorTokenizer

1.1.1

strvalued tokens in searching

This fixes a bug where if the user searches an integer, strpos handles the number as a character instead of a string.

1.1.0

Features

• Added a new searchField and searchToken functions.
  These functions allows you to autocomplete with any field in the engine (like title). The old suggest method is renamed suggestToken but still exists as deprecated.

    function suggestField($field, $value, $wrapSpan = false) { /* ... */}

  The engine will provide you suggestions for your $value by using the passed $field.
  The third parameter, $wrapSpan, will render differently depending of what you have passed :

  • true : $values that are present in suggestions will be wrapped by a tag.
  • any string : the span will have a class corresponding to the passed value

Deprecations

• suggest : Please use suggestToken instead.

1.0.1

(AdminPanel) Fix connex search results not having header

1.0.0

The first Stable Release is here !

No much evolution since the last beta, except now the fieldSearch is behaving as he should (searching with incomplete tokens) and files are saved as base64 names instead of a direct character, allowing unahtorized character to be indexed without flaws.
Migration instructions are found at the bottom of this release note

0.7 to 1.0 Changelog

Features

Connex Search

Connex search is, as his name suggests, an additional search for document that might interest the user based on the returned documents. The engine will try to match what's common between the resulted documents and will do a second search with this data. It'll provide you a list of tokens he found and also documents that resulted from the second search.

More informations about the Connex Search

Administration Panel

An Administration Panel has been added to the project with a nice list of features that allows you to check your configuration and what's inside your index.

You can clear the cache, check if your schema is valid, test what kind of facets you have, test if the tokenization is correct, do some search, try the connex search and even edit your documents directly !

More informations about the Administration Panel

Some Indexing optimizations

The old indexing process was quite slow when we hit several thousands of documents. The problem is not entirely solved but at least it's way faster than before !

I'm considering other solutions and will find a more optimized way to store the index data.

Breakings

• QueryBuilder's queries (deprecated in 0.7) has been removed and now will trigger Exceptions.
  You'll want to update your code by using the new QuerySegments:

  use VFou\Search\Query\QuerySegment;
  
  $seg = QuerySegment::and(
      QuerySegment::not(QuerySegment::exactSearch('genre','Drama')),
      QuerySegment::or(
          QuerySegment::exactSearch('year',"2011"),
          QuerySegment::and(
              QuerySegment::greaterSearch('rating',"8"),
              QuerySegment::exactSearch('author','nobody')
          )
      )
  );

Migrate from 0.7 to 1.0

Better Before Updating

If you're using the QueryBuilder's search methods (fieldSearch, exactSearch, etc... not orderBy, limit, etc...) you must migrate to their equivalent in QuerySegment's static methods.

After Updating

You will need to reindex your documents to enable the new features and avoid any unexpected error because of missing files.

0.7.12

fix fieldSearch not searching after incomplete tokens

1.0-beta7

Changes

• the Admin Panel's run function now return the HTML to be displayed instead of echoing everything

1.0-beta6

Changes

• Added two parameters to the run function that allows the user to define an uri and a base url for the admin panel

1.0-beta5

Features

• Added an Admin Panel to ease the management of your index and debug possible schemas and types issues

Note : I'm by far not proud of the code behind this admin panel. I wanted to avoid using external libraries for anything in this panel. Everything is in flat PHP with simple templates. There is not much css either (and i'm really not good at this so i'll pass my turn).

1.0-beta4

Fixes :

• 'all' file is not written
• included 'all' file into token deletion
• prevent a divizion by zero with the connex search

Changes :

• WordSeparatorTokenizer now keep the original words as a token