Merge pull request #35 from bertbaron/develop

feat: grammar and highlighting fixes and improvements

CMakeLists.txt

@@ -0,0 +1,66 @@
+cmake_minimum_required(VERSION 3.13)
+
+project(tree-sitter-kcl
+        VERSION "0.1.0"
+        DESCRIPTION "KCL Configration and policy language"
+        HOMEPAGE_URL "https://github.com/kcl-lang/tree-sitter-kcl"
+        LANGUAGES C)
+
+option(BUILD_SHARED_LIBS "Build using shared libraries" ON)
+option(TREE_SITTER_REUSE_ALLOCATOR "Reuse the library allocator" OFF)
+
+set(TREE_SITTER_ABI_VERSION 15 CACHE STRING "Tree-sitter ABI version")
+if(NOT ${TREE_SITTER_ABI_VERSION} MATCHES "^[0-9]+$")
+    unset(TREE_SITTER_ABI_VERSION CACHE)
+    message(FATAL_ERROR "TREE_SITTER_ABI_VERSION must be an integer")
+endif()
+
+find_program(TREE_SITTER_CLI tree-sitter DOC "Tree-sitter CLI")
+
+add_custom_command(OUTPUT "${CMAKE_CURRENT_SOURCE_DIR}/src/parser.c"
+                   DEPENDS "${CMAKE_CURRENT_SOURCE_DIR}/src/grammar.json"
+                   COMMAND "${TREE_SITTER_CLI}" generate src/grammar.json
+                            --abi=${TREE_SITTER_ABI_VERSION}
+                   WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}"
+                   COMMENT "Generating parser.c")
+
+add_library(tree-sitter-kcl src/parser.c)
+if(EXISTS ${CMAKE_CURRENT_SOURCE_DIR}/src/scanner.c)
+  target_sources(tree-sitter-kcl PRIVATE src/scanner.c)
+endif()
+target_include_directories(tree-sitter-kcl
+                           PRIVATE src
+                           INTERFACE $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/bindings/c>
+                                     $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}>)
+
+target_compile_definitions(tree-sitter-kcl PRIVATE
+                           $<$<BOOL:${TREE_SITTER_REUSE_ALLOCATOR}>:TREE_SITTER_REUSE_ALLOCATOR>
+                           $<$<CONFIG:Debug>:TREE_SITTER_DEBUG>)
+
+set_target_properties(tree-sitter-kcl
+                      PROPERTIES
+                      C_STANDARD 11
+                      POSITION_INDEPENDENT_CODE ON
+                      SOVERSION "${TREE_SITTER_ABI_VERSION}.${PROJECT_VERSION_MAJOR}"
+                      DEFINE_SYMBOL "")
+
+configure_file(bindings/c/tree-sitter-kcl.pc.in
+               "${CMAKE_CURRENT_BINARY_DIR}/tree-sitter-kcl.pc" @ONLY)
+
+include(GNUInstallDirs)
+
+install(DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}/bindings/c/tree_sitter"
+        DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}"
+        FILES_MATCHING PATTERN "*.h")
+install(FILES "${CMAKE_CURRENT_BINARY_DIR}/tree-sitter-kcl.pc"
+        DESTINATION "${CMAKE_INSTALL_DATAROOTDIR}/pkgconfig")
+install(TARGETS tree-sitter-kcl
+        LIBRARY DESTINATION "${CMAKE_INSTALL_LIBDIR}")
+
+file(GLOB QUERIES queries/*.scm)
+install(FILES ${QUERIES}
+        DESTINATION "${CMAKE_INSTALL_DATADIR}/tree-sitter/queries/kcl")
+
+add_custom_target(ts-test "${TREE_SITTER_CLI}" test
+                  WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}"
+                  COMMENT "tree-sitter test")

README.md

@@ -24,7 +24,7 @@ tree-sitter g
 
 ### Testing
 
-Add test suites in [./test/corpus/](./test/corpus/) and run the test command
+Add test suites in [./test/corpus/](./test/corpus/) or [./test/highlight/](./test/highlight/) and run the test command
 
 ```shell
 tree-sitter t -u

grammar.js

@@ -178,7 +178,7 @@ module.exports = grammar({
     $._dedent,
     $.string_start,
     $._string_content,
-    $.escape_interpolation,
+    '${',
     $.string_end,
 
     // Mark comments as external tokens so that the external scanner is always
@@ -296,7 +296,8 @@ module.exports = grammar({
       field('name', $.parameter),
       optional(seq(
         '(',
-        field('base', $.identifier),
+        commaSep1(field('base', $.identifier)),
+        optional(','),
         ')'
       )),
       optional(seq(
@@ -475,17 +476,6 @@ module.exports = grammar({
       field('body', $._suite),
     ),
 
-    rule_stmt: $ => seq(
-      'rule',
-      field('name', $.identifier),
-      optional(seq(
-        'for',
-        field('protocol', $.identifier),
-      )),
-      ':',
-      field('body', $._suite),
-    ),
-
     check_stmt: $ => prec.left(seq(
       'check',
       ':',
@@ -945,19 +935,28 @@ module.exports = grammar({
 
     string: $ => seq(
       $.string_start,
-      repeat($.string_content),
+     repeat(choice($.interpolation, $.string_content)),
       $.string_end,
     ),
 
     string_content: $ => prec.right(repeat1(
       choice(
-        $.escape_interpolation,
         $.escape_sequence,
         $._not_escape_sequence,
         $._string_content,
       ))),
 
-    escape_interpolation: _ => token.immediate(/\$\{[^}]*\}/),
+    interpolation: $ => seq(
+      '${',
+      field('expression', $._simple_expr),
+      optional(field("format_spec", $.format_spec)),
+      '}',
+    ),
+
+    format_spec: $ => seq(
+      ':',
+      token(prec(1, /[^}\n]+/)),
+    ),
 
     escape_sequence: _ => token.immediate(prec(1, seq(
       '\\',
@@ -968,14 +967,13 @@ module.exports = grammar({
         /\d{3}/,
         /\r?\n/,
         /['"abfrntv\\]/,
-        /N\{[^}]+\}/,
+        /N\{[^}]+}/,
+        /\$/,
       ),
     ))),
 
     _not_escape_sequence: _ => token.immediate('\\'),
 
-    _string_content: _ => token.immediate(/[^\\{}]+/),
-
     integer: $ => token(choice(
       seq(
         choice('0x', '0X'),

package.json

@@ -38,16 +38,16 @@
     "node-gyp-build": "^4.8.1"
   },
   "peerDependencies": {
-    "tree-sitter": "^0.21.1"
+    "tree-sitter": "^0.22.4"
   },
   "peerDependenciesMeta": {
     "tree_sitter": {
       "optional": true
     }
   },
   "devDependencies": {
-    "tree-sitter-cli": "^0.23.0",
-    "prebuildify": "^6.0.1"
+    "prebuildify": "^6.0.1",
+    "tree-sitter-cli": "^0.24.7"
   },
   "tree-sitter": [
     {

queries/highlights.scm

@@ -1,5 +1,8 @@
 (identifier) @variable
 
+; Reset highlighting in string interpolations
+(interpolation) @none
+
 (import_stmt
   (dotted_name
     (identifier) @namespace))
@@ -44,10 +47,24 @@
 
 (comment) @comment
 (string) @string
-(escape_sequence) @escape
+(escape_sequence) @string.escape
+
+(schema_stmt
+  body: (block
+    .
+    (string
+      (string_content) @string.documentation)))
+
+(decorator
+  (identifier) @attribute)
+
+(call_expr
+  function: (identifier) @function)
 
 (call_expr
-    (identifier) @function)
+  function: (selector_expr
+    (select_suffix
+      (identifier) @function)))
 
 [
   (integer)
@@ -57,7 +74,7 @@
 [
   (true)
   (false)
-  (none) 
+  (none)
   (undefined)
 ] @constant.builtin
 
@@ -82,6 +99,19 @@
   "type"
 ] @keyword
 
+[
+  "("
+  ")"
+  "["
+  "]"
+  "{"
+  "}"
+] @punctuation.bracket
+
+(interpolation
+  "${" @punctuation.special
+  "}" @punctuation.special)
+
 [
   "+"
   "-"
@@ -103,7 +133,6 @@
   "=="
   "!="
   "@"
-;  "\\"
   "and"
   "or"
   "not"
@@ -112,3 +141,26 @@
   "="
   ":"
 ] @operator
+
+; second argument is a regex in all regex functions with at least two arguments
+(call_expr
+  function: (selector_expr
+    (identifier) @_regex)
+  arguments: (argument_list
+    (_)
+    .
+    (string
+      (string_content) @string.regexp))
+  (#eq? @_regex "regex"))
+
+; first argument is a regex in 'regex.compile' function
+(call_expr
+  .
+  function: (selector_expr
+    (identifier) @_regex
+    (select_suffix
+      (identifier) @_fn (#eq? @_fn "compile")))
+  arguments: (argument_list
+    (string
+      (string_content) @string.regexp))
+  (#eq? @_regex "regex"))