IndxSearchLib 5.0.5

• net10.0

  • CommunityToolkit.HighPerformance (>= 8.4.0)
  • Microsoft.Data.Sqlite (>= 10.0.3)
  • Microsoft.Extensions.Logging.Abstractions (>= 10.0.3)
  • SQLitePCLRaw.bundle_e_sqlite3 (>= 3.0.3)

- Fixed: **a document inserted or updated is ranked right straight away.** After an insert, an
       update or a field update, a word new to the index, or one that had appeared only in the first
       searchable field, was scored with the first field's weight and boost until the next `Index()`.
       A word found only in a low-weight field such as a description could then lift the document as
       far as a title match would. It now counts with the field it is in.
     - Fixed: **apostrophes no longer split a word or stand in its way.** "Children’s" is now the word
       "childrens" and "don't" is "dont", for every apostrophe people type (' ’ ‘ ʼ `). Before, the
       straight ' split the word ("gardener" + "s") and the typographic ’ stayed inside it, so a search
       without the apostrophe, or with a typo, missed or ranked low a title such as "The Gardener’s
       Daughter". The default text rules always meant to remove the apostrophe; they ran too late to
       reach the words.
     - Changed: **the text rules now apply to whole words and to near-exact matching,** not only to the
       index's internal fragments. In the default rules ':' and ';' now become a space rather than being
       removed. If you have your own `TextNormalizer` rule that removes a separator, such as `(":", "")`,
       it now joins the words on either side; map it to a space instead.
     - Changed: **ranking moves for text with apostrophes, colons and semicolons,** as the fixes above
       take effect. An index picks this up at its next `Index()`, which the server runs whenever it loads
       a dataset, so there is nothing to migrate. One deliberate trade-off: French or Italian elision such
       as "l'homme" becomes one word, "lhomme".