Custom weighting is now a 0 -> 10 float, meaning you can specify data-pagefind-weight="0.5"
Word splitting now covers all punctuation, making code indexing pretty good
<h1 id="myComponent">Hello</h1> -> h1 id my component hello h1
Compound words split using the above method are now ranked lower based on how many parts there were
GitHub will now index as github at weight 1.0, and git + hub each at weight 0.5
Excerpt calculation is now substantially better, and picks the 'best' excerpt region (which it previously didn't)
This excerpt calculation now also takes word weighting into consideration, meaning compound words are implicitly de-prioritized in excerpts, and higher weighted words like headings are preferred (or dense regions of normal text)
data-pagefind-weight="0.5"
<h1 id="myComponent">Hello</h1>
->h1 id my component hello h1
GitHub
will now index asgithub
at weight 1.0, andgit
+hub
each at weight 0.5