From a8fd33ac012a79069ef1409503a2edcf3a585153 Mon Sep 17 00:00:00 2001
From: Michael Davis <mcarsondavis@gmail.com>
Date: Thu, 6 Jan 2022 09:00:00 -0600
Subject: [PATCH] add tree-sitter-regex (#1362)

* add tree-sitter-regex

* adapt regex highlights from upstream

* inject regex into elixir sigil_r/2 and sigil_R/2

* generate lang-support docs

* capture interesting nodes in character-ranges

* make $.character_class captures more consistent

* fix fallthrough behavior for character classes

* capture pattern characters as 'string'

* use latest tree-sitter-regex

* set elixir regex injections as combined

* add link to upstream queries

* inject regex in rust into 'Regex::new' raw string literals
---
 .gitmodules                              |  4 ++
 book/src/generated/lang-support.md       |  1 +
 helix-syntax/languages/tree-sitter-regex |  1 +
 languages.toml                           |  7 ++++
 runtime/queries/elixir/injections.scm    |  7 ++++
 runtime/queries/regex/highlights.scm     | 53 ++++++++++++++++++++++++
 runtime/queries/rust/injections.scm      | 14 +++++++
 7 files changed, 87 insertions(+)
 create mode 160000 helix-syntax/languages/tree-sitter-regex
 create mode 100644 runtime/queries/regex/highlights.scm

diff --git a/.gitmodules b/.gitmodules
index 9297708a6..f6a0fdc49 100644
--- a/.gitmodules
+++ b/.gitmodules
@@ -190,3 +190,7 @@
 	path = helix-syntax/languages/tree-sitter-git-rebase
 	url = https://github.com/the-mikedavis/tree-sitter-git-rebase.git
 	shallow = true
+[submodule "helix-syntax/languages/tree-sitter-regex"]
+	path = helix-syntax/languages/tree-sitter-regex
+	url = https://github.com/tree-sitter/tree-sitter-regex.git
+	shallow = true
diff --git a/book/src/generated/lang-support.md b/book/src/generated/lang-support.md
index a1fbf1724..daf8b0061 100644
--- a/book/src/generated/lang-support.md
+++ b/book/src/generated/lang-support.md
@@ -38,6 +38,7 @@
 | protobuf | ✓ |  | ✓ |  |
 | python | ✓ | ✓ | ✓ | `pylsp` |
 | racket |  |  |  | `racket` |
+| regex | ✓ |  |  |  |
 | ruby | ✓ |  | ✓ | `solargraph` |
 | rust | ✓ | ✓ | ✓ | `rust-analyzer` |
 | scala | ✓ |  | ✓ | `metals` |
diff --git a/helix-syntax/languages/tree-sitter-regex b/helix-syntax/languages/tree-sitter-regex
new file mode 160000
index 000000000..e1cfca3c7
--- /dev/null
+++ b/helix-syntax/languages/tree-sitter-regex
@@ -0,0 +1 @@
+Subproject commit e1cfca3c79896ff79842f057ea13e529b66af636
diff --git a/languages.toml b/languages.toml
index e8329fe71..3e2e7b15c 100644
--- a/languages.toml
+++ b/languages.toml
@@ -530,3 +530,10 @@ file-types = ["git-rebase-todo"]
 injection-regex = "git-rebase"
 comment-token = "#"
 indent = { tab-width = 2, unit = " " }
+
+[[language]]
+name = "regex"
+scope = "source.regex"
+injection-regex = "regex"
+file-types = ["regex"]
+roots = []
diff --git a/runtime/queries/elixir/injections.scm b/runtime/queries/elixir/injections.scm
index 321c90add..8370a0d8d 100644
--- a/runtime/queries/elixir/injections.scm
+++ b/runtime/queries/elixir/injections.scm
@@ -1,2 +1,9 @@
 ((comment) @injection.content
  (#set! injection.language "comment"))
+
+((sigil
+  (sigil_name) @_sigil_name
+  (quoted_content) @injection.content)
+ (#match? @_sigil_name "^(r|R)$")
+ (#set! injection.language "regex")
+ (#set! injection.combined))
diff --git a/runtime/queries/regex/highlights.scm b/runtime/queries/regex/highlights.scm
new file mode 100644
index 000000000..9376caa92
--- /dev/null
+++ b/runtime/queries/regex/highlights.scm
@@ -0,0 +1,53 @@
+; upstream: https://github.com/tree-sitter/tree-sitter-regex/blob/e1cfca3c79896ff79842f057ea13e529b66af636/queries/highlights.scm
+
+[
+  "("
+  ")"
+  "(?"
+  "(?:"
+  "(?<"
+  ">"
+  "["
+  "]"
+  "{"
+  "}"
+] @punctuation.bracket
+
+[
+  "*"
+  "+"
+  "|"
+  "="
+  "<="
+  "!"
+  "<!"
+  "?"
+] @operator
+
+[
+  (identity_escape)
+  (control_letter_escape)
+  (character_class_escape)
+  (control_escape)
+  (start_assertion)
+  (end_assertion)
+  (boundary_assertion)
+  (non_boundary_assertion)
+] @constant.character.escape
+
+(group_name) @property
+
+(count_quantifier
+  [
+    (decimal_digits) @constant.numeric
+    "," @punctuation.delimiter
+  ])
+
+(character_class
+  [
+    "^" @operator
+    (class_range "-" @operator)
+  ])
+
+(class_character) @constant.character
+(pattern_character) @string
diff --git a/runtime/queries/rust/injections.scm b/runtime/queries/rust/injections.scm
index d8382e499..77c70805e 100644
--- a/runtime/queries/rust/injections.scm
+++ b/runtime/queries/rust/injections.scm
@@ -10,3 +10,17 @@
   (token_tree) @injection.content)
  (#set! injection.language "rust")
  (#set! injection.include-children))
+
+(call_expression
+  function: (scoped_identifier
+    path: (identifier) @_regex (#eq? @_regex "Regex")
+    name: (identifier) @_new (#eq? @_new "new"))
+  arguments: (arguments (raw_string_literal) @injection.content)
+  (#set! injection.language "regex"))
+
+(call_expression
+  function: (scoped_identifier
+    path: (scoped_identifier (identifier) @_regex (#eq? @_regex "Regex") .)
+    name: (identifier) @_new (#eq? @_new "new"))
+  arguments: (arguments (raw_string_literal) @injection.content)
+  (#set! injection.language "regex"))