Try another approach

stsewd · Dec 29, 2023 · f2318ba · f2318ba
1 parent aefcc28
commit f2318ba
Show file tree

Hide file tree

Showing 15 changed files with 647 additions and 790 deletions.
diff --git a/Makefile b/Makefile
@@ -8,15 +8,9 @@ serve: build
 	npm run wasm
 	npm run web
 
-release: build format
+release: build
 	npm run wasm
 	# GitHub pages doesn't like symbolic links
 	cp tree-sitter-comment.wasm docs/js/tree-sitter-comment.wasm
 
-format:
-	clang-format -i \
-	  --style="{BasedOnStyle: webkit, IndentWidth: 2}" \
-	  src/scanner.c \
-	  src/tree_sitter_comment/*
-
-.PHONY: build test serve release format
+.PHONY: build test serve release
diff --git a/binding.gyp b/binding.gyp
@@ -9,7 +9,6 @@
       "sources": [
         "src/parser.c",
         "bindings/node/binding.cc",
-        "src/scanner.c"
       ],
       "cflags_c": [
         "-std=c99",

diff --git a/docs/js/tree-sitter-comment.wasm b/docs/js/tree-sitter-comment.wasm
diff --git a/grammar.js b/grammar.js
@@ -40,34 +40,41 @@ const STOP_CHARS = [
 module.exports = grammar({
   name: "comment",
 
-  externals: ($) => [
-    $.name,
-    $.invalid_token
+  conflicts: ($) => [
+    [$._simple_tag, $._tag_with_annotation],
   ],
 
   rules: {
     source: ($) => repeat(
       choice(
         $.tag,
-        $._full_uri,
-        alias($._text, "text"),
+        $.uri,
+        $._text,
       ),
     ),
 
-    tag: ($) => seq(
-      $.name,
-      optional($._user),
-      ":",
+    tag: ($) => choice(
+      $._simple_tag,
+      $._normal_tag,
+      $._tag_with_annotation,
     ),
 
-    _user: ($) => seq(
+    _simple_tag: ($) => alias($._tag_name, $.name),
+
+    _normal_tag: ($) => seq(
+      alias($._tag_name, $.name),
+      token.immediate(":"),
+    ),
+
+    _tag_with_annotation: ($) => seq(
+      alias($._tag_name, $.name),
       "(",
-      alias(/[^()]+/, $.user),
+      optional(alias(/[^()]+/, $.annotation)),
       ")",
+      token.immediate(":"),
     ),
 
-    // This token is split into two parts so the end character isn't included in the URI itself.
-    _full_uri: ($) => seq($.uri, choice(alias($._end_char, "text"), /\s/)),
+    _tag_name: ($) => /[A-Z]([A-Z0-9_-]*[A-Z0-9])?/,
 
     // This token needs to be single regex, otherwise a partial match will result in an error.
     uri: ($) => get_uri_regex(),

diff --git a/src/grammar.json b/src/grammar.json
@@ -12,94 +12,112 @@
           },
           {
             "type": "SYMBOL",
-            "name": "_full_uri"
+            "name": "uri"
           },
           {
-            "type": "ALIAS",
-            "content": {
-              "type": "SYMBOL",
-              "name": "_text"
-            },
-            "named": false,
-            "value": "text"
+            "type": "SYMBOL",
+            "name": "_text"
           }
         ]
       }
     },
     "tag": {
-      "type": "SEQ",
+      "type": "CHOICE",
       "members": [
         {
           "type": "SYMBOL",
-          "name": "name"
+          "name": "_simple_tag"
         },
         {
-          "type": "CHOICE",
-          "members": [
-            {
-              "type": "SYMBOL",
-              "name": "_user"
-            },
-            {
-              "type": "BLANK"
-            }
-          ]
+          "type": "SYMBOL",
+          "name": "_normal_tag"
         },
         {
-          "type": "STRING",
-          "value": ":"
+          "type": "SYMBOL",
+          "name": "_tag_with_annotation"
         }
       ]
     },
-    "_user": {
+    "_simple_tag": {
+      "type": "ALIAS",
+      "content": {
+        "type": "SYMBOL",
+        "name": "_tag_name"
+      },
+      "named": true,
+      "value": "name"
+    },
+    "_normal_tag": {
       "type": "SEQ",
       "members": [
-        {
-          "type": "STRING",
-          "value": "("
-        },
         {
           "type": "ALIAS",
           "content": {
-            "type": "PATTERN",
-            "value": "[^()]+"
+            "type": "SYMBOL",
+            "name": "_tag_name"
           },
           "named": true,
-          "value": "user"
+          "value": "name"
         },
         {
-          "type": "STRING",
-          "value": ")"
+          "type": "IMMEDIATE_TOKEN",
+          "content": {
+            "type": "STRING",
+            "value": ":"
+          }
         }
       ]
     },
-    "_full_uri": {
+    "_tag_with_annotation": {
       "type": "SEQ",
       "members": [
         {
-          "type": "SYMBOL",
-          "name": "uri"
+          "type": "ALIAS",
+          "content": {
+            "type": "SYMBOL",
+            "name": "_tag_name"
+          },
+          "named": true,
+          "value": "name"
+        },
+        {
+          "type": "STRING",
+          "value": "("
         },
         {
           "type": "CHOICE",
           "members": [
             {
               "type": "ALIAS",
               "content": {
-                "type": "SYMBOL",
-                "name": "_end_char"
+                "type": "PATTERN",
+                "value": "[^()]+"
               },
-              "named": false,
-              "value": "text"
+              "named": true,
+              "value": "annotation"
             },
             {
-              "type": "PATTERN",
-              "value": "\\s"
+              "type": "BLANK"
             }
           ]
+        },
+        {
+          "type": "STRING",
+          "value": ")"
+        },
+        {
+          "type": "IMMEDIATE_TOKEN",
+          "content": {
+            "type": "STRING",
+            "value": ":"
+          }
         }
       ]
     },
+    "_tag_name": {
+      "type": "PATTERN",
+      "value": "[A-Z]([A-Z0-9_-]*[A-Z0-9])?"
+    },
     "uri": {
       "type": "PATTERN",
       "value": "https?:\\/\\/([^\\s\\.,:;!\\?\\\\'\"\\}\\]\\)>]|[\\.,:;!\\?\\\\'\"\\}\\]\\)>][a-zA-Z0-9])+"
@@ -262,18 +280,14 @@
       "value": "\\s"
     }
   ],
-  "conflicts": [],
-  "precedences": [],
-  "externals": [
-    {
-      "type": "SYMBOL",
-      "name": "name"
-    },
-    {
-      "type": "SYMBOL",
-      "name": "invalid_token"
-    }
+  "conflicts": [
+    [
+      "_simple_tag",
+      "_tag_with_annotation"
+    ]
   ],
+  "precedences": [],
+  "externals": [],
   "inline": [],
   "supertypes": []
 }

diff --git a/src/node-types.json b/src/node-types.json
@@ -27,21 +27,16 @@
       "required": true,
       "types": [
         {
-          "type": "name",
+          "type": "annotation",
           "named": true
         },
         {
-          "type": "user",
+          "type": "name",
           "named": true
         }
       ]
     }
   },
-  {
-    "type": "text",
-    "named": false,
-    "fields": {}
-  },
   {
     "type": "!",
     "named": false
@@ -111,15 +106,15 @@
     "named": false
   },
   {
-    "type": "name",
+    "type": "annotation",
     "named": true
   },
   {
-    "type": "uri",
+    "type": "name",
     "named": true
   },
   {
-    "type": "user",
+    "type": "uri",
     "named": true
   },
   {