tree-grepper

Works like grep, but uses tree-sitter to search for structure instead of strings. Here's a longer introduction to the tool as a blog post.

Installing

Use nix to install:

cachix use tree-grepper # if you have cachix installed
nix-env -if https://github.com/BrianHicks/tree-grepper/archive/refs/heads/main.tar.gz

Usage

Use it like grep (or really, more like ack/ag/pt/rg.)

$ tree-grepper -q elm '(import_clause (import) (upper_case_qid)@name)'
./src/Main.elm:4:7:name:Browser
./src/Main.elm:6:7:name:Html
./src/Main.elm:8:7:name:Html.Events
...

By default, tree-grepper will output one match per (newline-delimited) line. The columns here are filename, row, column, match name, and match text.

Note, however, that if your query includes a match with newlines in the text they will be included in the output! If this causes problems for your use case, try asking for JSON output (-f json) instead.

tree-grepper uses Tree-sitter's s-expressions to find matches. See the tree-sitter docs on queries for what all you can do there.

We add one important thing on top of the standard query stuff (including #eq? and #match?): if you name a pattern starting with an underscore, it will not be returned in the output. This is primarily useful for filtering out matches you don't really care about. For example, to match JavaScript calls to require but not other functions, you could do this:

(call_expression (identifier)@_fn (arguments . (string)@import .) (#eq? @_fn require))

In addition to text output, we support JSON output for scripting: just specify -f json. You also get more info (the match's end location and node kind) by asking for JSON output.

Tree View

You can discover the node names your language uses by using --show-tree languagename path/to/file. When you do this, tree-grepper will parse the file and print out an indented tree view. The format like this:

file 1:1
  module_declaration 1:1
    module 1:1: module
    upper_case_qid 1:8
      upper_case_identifier 1:8: Math
    exposing_list 1:13
      exposing 1:13: exposing
      ( 1:22: (
      exposed_value 1:23
        lower_case_identifier 1:23: average
      , 1:30: ,
      exposed_value 1:32
        lower_case_identifier 1:32: percentOf
      ) 1:41: )

Each line takes the format {node name} {location} {source, if present}. Source is only shown for the leaf-most nodes on the tree to avoid printing a huge block. However, tree-grepper can extract text from any of these nodes.

You can use the node names in queries. For example:

• tree-grepper -q elm (exposed_value) would have matches on average and percentOf.
• tree-grepper -q elm (module_declration) would match on the whole declaration, module Math exposing (average, percentOf)

Supported Languages

• C
• C++
• Elixir
• Elm
• Go
• Haskell
• Java
• JavaScript
• Markdown
• PHP
• Python
• Ruby
• Rust
• TypeScript

... and your favorite? We're open to PRs for adding whatever language you'd like!

For development, there's a nix-shell setup that'll get you everything you need. Set up nix (just Nix, not NixOS) and then run nix-shell in the root of this repository.

After that, you just need to add a tree-sitter grammar to the project. The tree-sitter project keeps an up-to-date list, so you may not even need to write your own!

Note: when you're adding grammars, please keep things in alphabetical order.

1. Add your grammar as an input in flakes.nix, following the template of the ones already there. You'll need to add an entry in inputs and another in the updateVendor script.
2. Run direnv reload to make sure you have the latest changes, then update-vendor to get your grammar in the right place. Make sure the repo content under vendor/YOUR-GRAMMAR looks how you expect.
3. Set up compilation in build.rs by following the pattern there.
4. Set up a new target in src/language.rs by following the patterns there.
5. Add the language to the list of supported languages in this readme.

License

See LICENSE in the source.