v1.0.0-alpha.9

Changes

• Add support for #gsub! Neovim directive in 726b39b

  The neovim #gsub! directive allows replacing the part of a string that
  matches a given lua-pattern.

  Support for this directive has been added in order to expand support for
  nvim-treesitter's large corpus of injection queries.

• Add pruner.injection.indented for embeddings in e149980

  This directive instructs Pruner to treat the surrounding whitespace in
  nix-style embeddings as not part of the injection content.

  This could theoretically be completely replaced with a #trim!
  directive, but the semantics aren't entirely the same.

  The #trim! directive would remove all leading/trailing
  whitespace/newlines while nix-style embeddings only treat the first
  leading/trailing whitespace as excluded.

  For now, this is added as an experimental property and will most likely
  be removed/replaced with just #trim! in the future.

• Add support for #trim! Neovim directive in 2f8cb04

  The Neovim #trim! directive allows trimming a captures range to remove
  leading or trailing newlines and/or whitespace.

• Support the injection.combined capture property in 83bf691

  Currently Pruner completely fails to handle embedded regions that
  contain templating. In these cases treesitter grammars typically produce
  multiple injected region fragments which surround the templated host
  regions.

  This patch adds support for the injection.combined query capture
  property which instructs the parser to combine all related regions into
  a contiguous whole.

• Add support for language aliases in 30d1301

  Currently Pruner requires embedded regions to be annotated with their
  full, canonical language name. Many languages however make use of
  aliases like the file extension to tag an embedded region.

  In these cases Pruner won't know how to handle the region, even if it
  would be able to handle that language in its canonical form.

  To resolve this, this patch introduces a new config entry called
  language_aliases where mappings between a canonical language and it's
  aliases can be defined.

  In the future we can also ship Pruner with pre-defined mappings for
  common languages.

  [language_aliases]
  typescript = ["ts", "tsx"]

• Add support for pruner-ignore in 3986550
• Add @pruner.ignore.marker query captures in 07559cc

  The current pruner-ignore marker heuristic relies on the underlying
  grammar defining nodes as type = "comment". Surprisingly, not all
  grammars do this!

  Most notably Markdown, which marks comment blocks as html_block and relies
  on the injected grammar to define this document region as a comment.

  This breaks pruners ignore marker heuristics.

  The solution is to add support for pruner.ignore.marker captures
  discovered via ignore.scm queries. This allows users to define their
  own markers for ignoring nodes.

• Add @pruner.ignore query captures in 577cf6d

  This works like @pruner.ignore.marker but allows directly defining the
  target node to ignore instead of a marker which is then used as an
  anchor to discover the ignore target.

Full Changelog: v1.0.0-alpha.8...v1.0.0-alpha.9