feat(perl)!: switch to tree-sitter-perl org's parser

5 files changed, 138 insertions, 169 deletions

diff --git a/lockfile.json b/lockfile.json
index 0d3748886..e2706fc6b 100644
--- a/lockfile.json
+++ b/lockfile.json
@@ -414,7 +414,7 @@
     "revision": "e01767921df18142055d97407595329d7629e643"
   },
   "perl": {
-    "revision": "79e88f64681660f3961939bf764d8f3b4bbb0d27"
+    "revision": "99bbcb0faddfc79db3fd2996cb56689a4685e78c"
   },
   "php": {
     "revision": "0e02e7fab7913a0e77343edb347c8f17cac1f0ba"
diff --git a/lua/nvim-treesitter/parsers.lua b/lua/nvim-treesitter/parsers.lua
index 8d12a69aa..526a34cd7 100644
--- a/lua/nvim-treesitter/parsers.lua
+++ b/lua/nvim-treesitter/parsers.lua
@@ -1240,10 +1240,11 @@ list.pem = {
 
 list.perl = {
   install_info = {
-    url = "https://github.com/ganezdragon/tree-sitter-perl",
-    files = { "src/parser.c", "src/scanner.cc" },
+    url = "https://github.com/tree-sitter-perl/tree-sitter-perl",
+    files = { "src/parser.c", "src/scanner.c" },
+    branch = "release",
   },
-  maintainers = { "@lcrownover" },
+  maintainers = { "@RabbiVeesh", "@LeoNerd" },
 }
 
 list.php = {
diff --git a/queries/perl/folds.scm b/queries/perl/folds.scm
index 7be36e5e0..df23f6317 100644
--- a/queries/perl/folds.scm
+++ b/queries/perl/folds.scm
@@ -1,11 +1,19 @@
-[
-  (function_definition)
-  (if_statement)
-  (unless_statement)
-  (while_statement)
-  (until_statement)
-  (for_statement_1)
-  (for_statement_2)
-  (standalone_block)
-  (pod_statement)
-] @fold
+(comment)+ @fold
+(pod) @fold
+
+; fold the block-typed package statements only
+(package_statement (block)) @fold
+
+[(subroutine_declaration_statement)
+ (conditional_statement)
+ (loop_statement)
+ (for_statement)
+ (cstyle_for_statement)
+ (block_statement)
+ (phaser_statement)] @fold
+
+(anonymous_subroutine_expression) @fold
+
+; perhaps folks want to fold these too?
+[(anonymous_array_expression)
+ (anonymous_hash_expression)] @fold
diff --git a/queries/perl/highlights.scm b/queries/perl/highlights.scm
index c7403b6e7..82d181183 100644
--- a/queries/perl/highlights.scm
+++ b/queries/perl/highlights.scm
@@ -1,198 +1,146 @@
-; Misc keywords
-[
-  "my" "our" "local"
-  "next" "last" "redo"
-  "goto"
-  "package"
-;  "do"
-;  "eval"
-] @keyword
+((source_file . (comment) @preproc)
+  (#lua-match? @preproc "^#!/"))
 
-; Keywords for including
 [ "use" "no" "require" ] @include
 
-; Keywords that mark conditional statements
 [ "if" "elsif" "unless" "else" ] @conditional
-(ternary_expression
-  ["?" ":"] @conditional.ternary)
 
-; Keywords that mark repeating loops
+(conditional_expression [ "?" ":" ] @conditional.ternary) 
+
 [ "while" "until" "for" "foreach" ] @repeat
 
-; Keyword for return expressions
-[ "return" ] @keyword.return
+"return" @keyword.return
 
-; Keywords for phaser blocks
-; TODO: Ideally these would be @keyword.phaser but vim-treesitter doesn't
-;   have such a thing yet
-[ "BEGIN" "CHECK" "UNITCHECK" "INIT" "END" ] @keyword.function
+"sub" @keyword.function
 
-; Keywords to define a function
-[ "sub" ] @keyword.function
+[ "map" "grep" ] @function.builtin
 
-; Keywords that are regular infix operators
-[
-  "and" "or" "not" "xor"
-  "eq" "ne" "lt" "le" "ge" "gt" "cmp"
-] @keyword.operator
+"package" @include
 
-; Variables
 [
-  (scalar_variable)
-  (array_variable)
-  (hash_variable)
-] @variable
+  "do"
+  "my" "our" "local"
+  "last" "next" "redo" "goto"
+  "undef"
+] @keyword
+
+(_ operator: _ @operator)
+"\\" @operator
+
+(yadayada) @exception
+
+(phaser_statement phase: _ @keyword.phaser)
 
-; Special builtin variables
 [
-  (special_scalar_variable)
-  (special_array_variable)
-  (special_hash_variable)
-  (special_literal)
-  (super)
-] @variable.builtin
+  "or" "and"
+  "eq" "ne" "cmp" "lt" "le" "ge" "gt"
+  "isa"
+] @keyword.operator
 
-((scalar_variable) @variable.builtin
- (#eq? @variable.builtin "$#ARGV"))
+(eof_marker) @preproc
+(data_section) @comment
+
+(pod) @text
 
-; Integer numbers
 [
-  (integer)
-  (hexadecimal)
+  (number)
+  (version)
 ] @number
 
-; Float numbers
 [
-  (floating_point)
-  (scientific_notation)
-] @float
-
-; version sortof counts as a kind of multipart integer
-(version) @constant
+  (string_literal) 
+  (interpolated_string_literal) 
+  (quoted_word_list) 
+  (command_string) 
+  (heredoc_content)
+  (replacement)
+  (transliteration_content)
+] @string
 
-; Package names are types
-(package_name) @type
+[
+  (heredoc_token)
+  (command_heredoc_token)
+  (heredoc_end)
+] @label
 
-; The special SUPER:: could be called a namespace. It isn't really but it
-;   should highlight differently and we might as well do it this way
-(super) @namespace
+[(escape_sequence) (escaped_delimiter)] @string.escape
 
-; Comments are comments
-(comments) @comment @spell
+(_ modifiers: _ @character.special)
+[  
+ (quoted_regexp)
+ (match_regexp)
+ (regexp_content)
+] @string.regex
 
-((source_file . (comments) @preproc)
-  (#lua-match? @preproc "^#!/"))
+(autoquoted_bareword _?) @string.special
 
-; POD should be handled specially with its own embedded subtype but for now
-;   we'll just have to do this.
-(pod_statement) @text
+(use_statement (package) @type)
+(package_statement (package) @type)
+(require_expression (bareword) @type)
 
-(method_invocation
-  function_name: (identifier) @method.call)
-(call_expression
-  function_name: (identifier) @function.call)
+(subroutine_declaration_statement name: (bareword) @function)
+(attribute_name) @attribute
+(attribute_value) @string
 
-; Built-in functions
-((call_expression
-  function_name: (identifier) @function.builtin)
- (#any-of? @function.builtin
-   "print" "printf" "sprintf" "say"
-   "push" "pop" "shift" "unshift" "splice"
-   "exists" "delete" "keys" "values"
-   "each"))
+(label) @label
 
-[
-  (map)
-  (grep)
-  (bless)
-] @function.builtin
+(statement_label label: _ @label)
 
-;; ----------
+(relational_expression operator: "isa" right: (bareword) @type)
 
-(use_constant_statement
-  constant: (identifier) @constant)
+(function_call_expression (function) @function.call)
+(method_call_expression (method) @method.call)
+(method_call_expression invocant: (bareword) @type)
 
-(named_block_statement
-  function_name: (identifier) @function)
+(func0op_call_expression function: _ @function.builtin)
+(func1op_call_expression function: _ @function.builtin)
 
-(function_definition
-  name: (identifier) @function)
+([(function)(expression_statement (bareword))] @function.builtin
+ (#set! "priority" 101)
+ (#lua-match? @function.builtin
+   "^(accept|atan2|bind|binmode|bless|crypt|chmod|chown|connect|die|dbmopen|exec|fcntl|flock|getpriority|getprotobynumber|gethostbyaddr|getnetbyaddr|getservbyname|getservbyport|getsockopt|glob|index|ioctl|join|kill|link|listen|mkdir|msgctl|msgget|msgrcv|msgsend|opendir|print|printf|push|pack|pipe|return|rename|rindex|read|recv|reverse|say|select|seek|semctl|semget|semop|send|setpgrp|setpriority|seekdir|setsockopt|shmctl|shmread|shmwrite|shutdown|socket|socketpair|split|sprintf|splice|substr|system|symlink|syscall|sysopen|sysseek|sysread|syswrite|tie|truncate|unlink|unpack|utime|unshift|vec|warn|waitpid|formline|open|sort)$"
+))
 
 (function) @function
 
-[
-  "(" ")"
-  "[" "]"
-  "{" "}"
-  (standard_input_to_variable)
-] @punctuation.bracket
-
-[ "`" "\"" ] @punctuation.special
-
-[
-  "=~"
-  "!~"
-  "="
-  "=="
-  "+"
-  "-"
-  "."
-  "//"
-  "||"
-  "&&"
-  "<<"
-  (arrow_operator)
-  (hash_arrow_operator)
-  (array_dereference)
-  (hash_dereference)
-  (to_reference)
-  (type_glob)
-  (hash_access_variable)
-] @operator
+(ERROR) @error
 
-[
-  (regex_option)
-  (regex_option_for_substitution)
-  (regex_option_for_transliteration)
-] @parameter
+(_
+  "{" @punctuation.special
+  (varname)
+  "}" @punctuation.special)
 
-(type_glob
-  (identifier) @variable)
+(varname 
+  (block
+    "{" @punctuation.special 
+    "}" @punctuation.special))
 
-[
-  (word_list_qw)
-  (command_qx_quoted)
-  (string_single_quoted)
-  (string_double_quoted)
-  (string_qq_quoted)
-  (bareword)
-  (transliteration_tr_or_y)
-] @string
 
-[
- (heredoc_start_identifier)
- (heredoc_end_identifier)
-] @label
+(
+  [(varname) (filehandle)] @variable.builtin
+  (#lua-match? @variable.builtin "^((ENV|ARGV|INC|ARGVOUT|SIG|STDIN|STDOUT|STDERR)|[_ab]|\\W|\\d+|\\^.*)$")
+)
 
-(heredoc_body_statement) @text.literal
+(scalar) @variable.scalar
+(scalar_deref_expression [ "$" "*"] @variable.scalar)
+[(array) (arraylen)] @variable.array
+(array_deref_expression [ "@" "*"] @variable.array)
+(hash) @variable.hash
+(hash_deref_expression [ "%" "*"] @variable.hash)
 
-[
-  (pattern_matcher)
-  (regex_pattern_qr)
-  (patter_matcher_m)
-  (substitution_pattern_s)
-] @string.regex
+(array_element_expression array:(_) @variable.array)
+(slice_expression array:(_) @variable.array)
+(keyval_expression array:(_) @variable.array)
 
-(escape_sequence) @string.escape
+(hash_element_expression hash:(_) @variable.hash)
+(slice_expression hash:(_) @variable.hash)
+(keyval_expression hash:(_) @variable.hash)
 
-[
-  ","
-  (semi_colon)
-  (start_delimiter)
-  (end_delimiter)
-  (ellipsis_statement)
-] @punctuation.delimiter
+(comment) @comment
 
-(function_attribute) @field
+([ "=>" "," ";" "->" ] @punctuation.delimiter)
 
-(function_signature) @type
+(
+  [ "[" "]" "{" "}" "(" ")" ] @punctuation.bracket
+  ; priority hack so nvim + ts-cli behave the same
+  (#set! "priority" 90))
diff --git a/queries/perl/injections.scm b/queries/perl/injections.scm
index cab5f53d5..ba5e45f45 100644
--- a/queries/perl/injections.scm
+++ b/queries/perl/injections.scm
@@ -1,2 +1,14 @@
-((comments) @injection.content
+; an injections.scm file for nvim-treesitter
+((comment) @injection.content
  (#set! injection.language "comment"))
+ 
+((pod) @injection.content
+ (#set! injection.language "pod"))
+
+((substitution_regexp
+  (replacement) @injection.content
+  (substitution_regexp_modifiers) @_modifiers)
+    ; match if there's a single `e` in the modifiers list
+  (#lua-match? @_modifiers "e")
+  (#not-lua-match? @_modifiers "e.*e")
+  (#set! injection.language "perl"))