From fe6ff49ee8718404480bfff46aa23b152d1e81e1 Mon Sep 17 00:00:00 2001
From: Eugene Yokota <eed3si9n@gmail.com>
Date: Fri, 13 Jan 2023 01:36:12 -0500
Subject: [PATCH] Fix operator_identifier

Problem
-------
1. Currently `//` ends up matching `operator_identier`
2. There's also a bug in regex.

Solution
--------
1. Implement a workaround to avoid `//`.
2. Escape `-`, copying from the identifier regex.
---
 corpus/definitions.txt | 33 +++++++++++++++++----------------
 grammar.js             | 23 ++++++++++++++++++++++-
 script/smoke_test.sh   |  2 +-
 3 files changed, 40 insertions(+), 18 deletions(-)

diff --git a/corpus/definitions.txt b/corpus/definitions.txt
index a66cf30d1..e9d91ea40 100644
--- a/corpus/definitions.txt
+++ b/corpus/definitions.txt
@@ -58,6 +58,11 @@ def → = ???
 
 val test = id.##
 
+val x = y
+/////////
+// avoid matching slashes as operator
+/////////
+
 ---
 
 (compilation_unit
@@ -66,21 +71,13 @@ val test = id.##
     (type_parameters
       (covariant_type_parameter
         (identifier)))
-    (infix_type
-      (stable_type_identifier
-        (stable_identifier
-          (identifier)
-          (identifier))
+    (generic_type
+      (stable_type_identifier (stable_identifier (stable_identifier (identifier) (identifier)) (identifier))
         (type_identifier))
-      (operator_identifier)
-      (generic_type
-        (type_identifier
-          (MISSING _alpha_identifier))
-        (type_arguments
-          (type_identifier)))))
+      (type_arguments (type_identifier))))
   (val_definition
     (operator_identifier)
-    (postfix_expression
+    (field_expression
       (field_expression
         (field_expression
           (identifier)
@@ -89,14 +86,14 @@ val test = id.##
       (operator_identifier)))
   (val_definition
     (operator_identifier)
-    (postfix_expression
+    (field_expression
       (field_expression
         (identifier)
         (identifier))
       (operator_identifier)))
   (val_definition
     (operator_identifier)
-    (postfix_expression
+    (field_expression
       (field_expression
         (identifier)
         (identifier))
@@ -104,9 +101,13 @@ val test = id.##
   (function_definition (operator_identifier) (operator_identifier))
   (val_definition
     (identifier)
-    (postfix_expression
+    (field_expression
       (identifier)
-      (operator_identifier))))
+      (operator_identifier)))
+  (val_definition (identifier) (identifier))
+  (comment)
+  (comment)
+  (comment))
 
 ================================
 Package
diff --git a/grammar.js b/grammar.js
index d548f11a4..741413b32 100644
--- a/grammar.js
+++ b/grammar.js
@@ -1165,7 +1165,28 @@ module.exports = grammar({
 
     wildcard: $ => '_',
 
-    operator_identifier: $ => /[!#%&*+-\/:<=>?@'^\|‘~\p{Sm}\p{So}]+/,
+    /**
+     * Regex patterns created to avoid matching // comments.
+     * This could technically match illeagal tokens such as val ?// = 1
+     */
+    operator_identifier: $ => token(choice(
+      // single opchar
+      /[\-!#%&*+\/\\:<=>?@\u005e\u007c~\p{Sm}\p{So}]/,
+      seq(
+        // opchar minus slash
+        /[\-!#%&*+\\:<=>?@\u005e\u007c~\p{Sm}\p{So}]/,
+        // opchar*
+        repeat1(/[\-!#%&*+\/\\:<=>?@\u005e\u007c~\p{Sm}\p{So}]/),
+      ),
+      seq(
+        // opchar
+        /[\-!#%&*+\/\\:<=>?@\u005e\u007c~\p{Sm}\p{So}]/,
+        // opchar minus slash
+        /[\-!#%&*+\\:<=>?@\u005e\u007c~\p{Sm}\p{So}]/,
+        // opchar*
+        repeat(/[\-!#%&*+\/\\:<=>?@\u005e\u007c~\p{Sm}\p{So}]/),
+      ),
+    )),
 
     _non_null_literal: $ => 
       choice(
diff --git a/script/smoke_test.sh b/script/smoke_test.sh
index 5b372cd49..ae7b72a52 100755
--- a/script/smoke_test.sh
+++ b/script/smoke_test.sh
@@ -4,7 +4,7 @@
 
 SCALA_SCALA_LIBRARY_EXPECTED=100
 SCALA_SCALA_COMPILER_EXPECTED=66
-DOTTY_COMPILER_EXPECTED=74
+DOTTY_COMPILER_EXPECTED=66
 
 if [ ! -d "$SCALA_SCALA_DIR" ]; then
   echo "\$SCALA_SCALA_DIR must be set"