Merge pull request #324 from johannescoetzee/fix/string-interpolation

fix: handle interpolations without spaces and $$ in interpolated strings
tree-sitter · Jul 23, 2023 · 8cdd8bd · 8cdd8bd
2 parents 8062487 + 6f9683b
commit 8cdd8bd
Show file tree

Hide file tree

Showing 4 changed files with 85 additions and 4 deletions.
diff --git a/corpus/literals.txt b/corpus/literals.txt
@@ -38,6 +38,15 @@ val string3 = raw"Not a new line \n${ha}"
 val string4 = s"""
 works even in multiline strings, ${name}
 """
+
+val string5 = s"$works${without}$spaces"
+
+val string6 = s"$a$b"
+
+val string7 = s"$$ $a"
+
+val string8 = s"$"$a"
+
 --------------------------------------------------------------------------------
 
 (compilation_unit
@@ -73,7 +82,43 @@ works even in multiline strings, ${name}
       (interpolated_string
         (interpolation
           (block
-            (identifier)))))))
+            (identifier))))))
+  (val_definition
+    (identifier)
+    (interpolated_string_expression
+      (identifier)
+      (interpolated_string
+        (interpolation
+          (identifier))
+        (interpolation
+          (block
+            (identifier)))
+        (interpolation
+          (identifier)))))
+  (val_definition
+    (identifier)
+    (interpolated_string_expression
+      (identifier)
+        (interpolated_string
+          (interpolation
+            (identifier))
+          (interpolation
+            (identifier)))))
+  (val_definition
+    (identifier)
+    (interpolated_string_expression
+      (identifier)
+      (interpolated_string
+        (interpolation
+          (identifier)))))
+  (val_definition
+    (identifier)
+    (interpolated_string_expression
+      (identifier)
+      (interpolated_string
+        (interpolation
+          (identifier))))))
+
 
 ================================================================================
 Integer literals

diff --git a/grammar.js b/grammar.js
@@ -1388,6 +1388,26 @@ module.exports = grammar({
     _alpha_identifier: $ =>
       /[\p{Lu}\p{Lt}\p{Nl}\p{Lo}\p{Lm}\$\p{Ll}_\u00AA\u00BB\u02B0-\u02B8\u02C0-\u02C1\u02E0-\u02E4\u037A\u1D78\u1D9B-\u1DBF\u2071\u207F\u2090-\u209C\u2C7C-\u2C7D\uA69C-\uA69D\uA770\uA7F8-\uA7F9\uAB5C-\uAB5F\$][\p{Lu}\p{Lt}\p{Nl}\p{Lo}\p{Lm}\$\p{Ll}_\u00AA\u00BB\u02B0-\u02B8\u02C0-\u02C1\u02E0-\u02E4\u037A\u1D78\u1D9B-\u1DBF\u2071\u207F\u2090-\u209C\u2C7C-\u2C7D\uA69C-\uA69D\uA770\uA7F8-\uA7F9\uAB5C-\uAB5F0-9\$_\p{Ll}]*(_[\-!#%&*+\/\\:<=>?@\u005e\u007c~\p{Sm}\p{So}]+)?/,
 
+    /**
+     * Despite what the lexical syntax suggests, the alphaid rule doesn't apply 
+     * to identifiers that aren't in blocks in interpolated strings (e.g. $foo).
+     * A more accurate description is given in 
+     * https://www.scala-lang.org/files/archive/spec/2.13/01-lexical-syntax.html
+     * where it states (regarding dollar sign escapes in interpolated strings) that
+     * """
+     * The simpler form consists of a ‘$’-sign followed by an identifier starting 
+     * with a letter and followed only by letters, digits, and underscore characters
+     * """
+     * where "letters" does not include the $ character.
+     *
+     * This rule is similar to the _alpha_identifier rule, with the differences 
+     * being that the $ character is excluded, along with the _(operator_chars)
+     * suffix and can be approximated as
+     * /[A-Za-z_][A-Z_a-z0-9]/;
+     */
+    _interpolation_identifier: $ =>
+      /[\p{Lu}\p{Lt}\p{Nl}\p{Lo}\p{Lm}\p{Ll}_\u00AA\u00BB\u02B0-\u02B8\u02C0-\u02C1\u02E0-\u02E4\u037A\u1D78\u1D9B-\u1DBF\u2071\u207F\u2090-\u209C\u2C7C-\u2C7D\uA69C-\uA69D\uA770\uA7F8-\uA7F9\uAB5C-\uAB5F][\p{Lu}\p{Lt}\p{Nl}\p{Lo}\p{Lm}\p{Ll}_\u00AA\u00BB\u02B0-\u02B8\u02C0-\u02C1\u02E0-\u02E4\u037A\u1D78\u1D9B-\u1DBF\u2071\u207F\u2090-\u209C\u2C7C-\u2C7D\uA69C-\uA69D\uA770\uA7F8-\uA7F9\uAB5C-\uAB5F0-9_\p{Ll}]*/,
+
     _backquoted_id: $ => /`[^\n`]+`/,
 
     _identifier: $ => choice($.identifier, $.operator_identifier),
@@ -1492,18 +1512,22 @@ module.exports = grammar({
 
     _interpolated_multiline_string_start: $ => '"""',
 
-    interpolation: $ => seq("$", choice($.identifier, $.block)),
+    _dollar_escape: $ => seq('$', choice('$', '"')),
+
+    _aliased_interpolation_identifier: $ => alias($._interpolation_identifier, $.identifier),
+
+    interpolation: $ => seq("$", choice($._aliased_interpolation_identifier, $.block)),
 
     interpolated_string: $ =>
       choice(
         seq(
           $._interpolated_string_start,
-          repeat(seq($._interpolated_string_middle, $.interpolation)),
+          repeat(seq($._interpolated_string_middle, choice($._dollar_escape, $.interpolation))),
           $._interpolated_string_end,
         ),
         seq(
           $._interpolated_multiline_string_start,
-          repeat(seq($._interpolated_multiline_string_middle, $.interpolation)),
+          repeat(seq($._interpolated_multiline_string_middle, choice($._dollar_escape, $.interpolation))),
           $._interpolated_multiline_string_end,
         ),
       ),

diff --git a/queries/scala/locals.scm b/queries/scala/locals.scm
@@ -27,3 +27,4 @@
   name: (identifier) @local.definition)
 
 (identifier) @local.reference
+
diff --git a/test/highlight/basics.scala b/test/highlight/basics.scala
@@ -79,5 +79,16 @@ object Hello {
 //     ^method
   val hello2 = c"some $meth"
 //                      ^method
+  val hello3 = s"$$$meth$hello2%"
+//               ^string
+//                 ^punctuation.special
+//                  ^method
+//                      ^punctuation.special
+//                       ^variable
+//                             ^string
+  val hello4 = s"$"$hello3"
+//               ^string
+//                 ^punctuation.special
+//                  ^variable
 }