Merge pull request #97 from matoous/md/comment-directives

feat(grammar): capture comment directives
tree-sitter · Dec 6, 2022 · 06432b8 · 06432b8
2 parents dfbef73 + e5f17e1
commit 06432b8
Show file tree

Hide file tree

Showing 5 changed files with 31,304 additions and 23,405 deletions.
diff --git a/corpus/source_files.txt b/corpus/source_files.txt
@@ -100,6 +100,54 @@ const d
   (comment)
   (const_declaration (const_spec (identifier))))
 
+================================================================================
+Directive comments
+================================================================================
+
+package main
+
+//extern test_extern
+func testExtern() {}
+
+//export text_export
+func textExport() {}
+
+//line test:123
+type testLine int
+
+//nolint:unused
+type Custom int
+
+---
+
+(source_file
+  (package_clause
+    (package_identifier))
+  (comment
+    (directive))
+  (function_declaration
+    (identifier)
+    (parameter_list)
+    (block))
+  (comment
+    (directive))
+  (function_declaration
+    (identifier)
+    (parameter_list)
+    (block))
+  (comment
+    (directive))
+  (type_declaration
+    (type_spec
+      (type_identifier)
+      (type_identifier)))
+  (comment
+    (directive))
+  (type_declaration
+    (type_spec
+      (type_identifier)
+      (type_identifier))))
+
 ============================================
 Non-ascii variable names
 ============================================

diff --git a/grammar.js b/grammar.js
@@ -895,15 +895,19 @@ module.exports = grammar({
     false: $ => 'false',
     iota: $ => 'iota',
 
+    // https://tip.golang.org/doc/comment#syntax
+    directive: $ => token(/(line |extern |export |[a-z0-9]+:[a-z0-9])/),
+
     // http://stackoverflow.com/questions/13014947/regex-to-match-a-c-style-multiline-comment/36328890#36328890
-    comment: $ => token(choice(
-      seq('//', /.*/),
+    comment: $ => choice(
+      seq('//', $.directive, /.*/),
+      seq('// ', /.*/),
       seq(
         '/*',
         /[^*]*\*+([^/*][^*]*\*+)*/,
         '/'
       )
-    ))
+    )
   }
 })
 

diff --git a/src/grammar.json b/src/grammar.json
@@ -6720,43 +6720,64 @@
       "type": "STRING",
       "value": "iota"
     },
-    "comment": {
+    "directive": {
       "type": "TOKEN",
       "content": {
-        "type": "CHOICE",
-        "members": [
-          {
-            "type": "SEQ",
-            "members": [
-              {
-                "type": "STRING",
-                "value": "//"
-              },
-              {
-                "type": "PATTERN",
-                "value": ".*"
-              }
-            ]
-          },
-          {
-            "type": "SEQ",
-            "members": [
-              {
-                "type": "STRING",
-                "value": "/*"
-              },
-              {
-                "type": "PATTERN",
-                "value": "[^*]*\\*+([^/*][^*]*\\*+)*"
-              },
-              {
-                "type": "STRING",
-                "value": "/"
-              }
-            ]
-          }
-        ]
+        "type": "PATTERN",
+        "value": "(line |extern |export |[a-z0-9]+:[a-z0-9])"
       }
+    },
+    "comment": {
+      "type": "CHOICE",
+      "members": [
+        {
+          "type": "SEQ",
+          "members": [
+            {
+              "type": "STRING",
+              "value": "//"
+            },
+            {
+              "type": "SYMBOL",
+              "name": "directive"
+            },
+            {
+              "type": "PATTERN",
+              "value": ".*"
+            }
+          ]
+        },
+        {
+          "type": "SEQ",
+          "members": [
+            {
+              "type": "STRING",
+              "value": "// "
+            },
+            {
+              "type": "PATTERN",
+              "value": ".*"
+            }
+          ]
+        },
+        {
+          "type": "SEQ",
+          "members": [
+            {
+              "type": "STRING",
+              "value": "/*"
+            },
+            {
+              "type": "PATTERN",
+              "value": "[^*]*\\*+([^/*][^*]*\\*+)*"
+            },
+            {
+              "type": "STRING",
+              "value": "/"
+            }
+          ]
+        }
+      ]
     }
   },
   "extras": [

diff --git a/src/node-types.json b/src/node-types.json
@@ -588,6 +588,21 @@
       }
     }
   },
+  {
+    "type": "comment",
+    "named": true,
+    "fields": {},
+    "children": {
+      "multiple": false,
+      "required": false,
+      "types": [
+        {
+          "type": "directive",
+          "named": true
+        }
+      ]
+    }
+  },
   {
     "type": "communication_case",
     "named": true,
@@ -2603,6 +2618,18 @@
     "type": "/",
     "named": false
   },
+  {
+    "type": "/*",
+    "named": false
+  },
+  {
+    "type": "//",
+    "named": false
+  },
+  {
+    "type": "// ",
+    "named": false
+  },
   {
     "type": "/=",
     "named": false
@@ -2695,10 +2722,6 @@
     "type": "chan",
     "named": false
   },
-  {
-    "type": "comment",
-    "named": true
-  },
   {
     "type": "const",
     "named": false
@@ -2715,6 +2738,10 @@
     "type": "defer",
     "named": false
   },
+  {
+    "type": "directive",
+    "named": true
+  },
   {
     "type": "else",
     "named": false