fix(plugin): recognize literal escapes and C string suffixes

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

8 files changed, 80 insertions, 28 deletions

diff --git a/plugin/AGENTS/ARCH.md b/plugin/AGENTS/ARCH.md
index 31c5974..90e9614 100644
--- a/plugin/AGENTS/ARCH.md
+++ b/plugin/AGENTS/ARCH.md
@@ -82,10 +82,11 @@ only a subset of its syntax:
   ordinary parentheses. `TYPE Name = C_TYPE("...")` is a single-line alias
   declaration, so it does not open an indentation block.
 - The compiler decodes C++17 byte escapes and Unicode escapes in literals; strings
-  encode Unicode as UTF-8 and CHAR Unicode escapes must be ASCII. The plugin's
-  grammar and quote masking still ignore escapes and retain legacy `'''` handling;
-  the compiler now requires an escaped quote character (`'\''`). Neither supports
-  multiline literal continuation.
+  encode Unicode as UTF-8 and CHAR Unicode escapes must be ASCII. Highlighting
+  uses the C++ escape scope and string-suffix scope for adjacent uppercase C.
+  Grammar and shared quote masking honor escaped quotes and backslashes; quote
+  characters use `'\''`, with no legacy three-quote special case. Neither
+  supports multiline literal continuation.
 - Blocks: IF/THEN/ELSE/ENDIF, WHILE/DO/ENDWHILE, REPEAT/UNTIL, FOR/NEXT,
   FUNCTION/ENDFUNCTION, PROCEDURE/ENDPROCEDURE, record TYPE/ENDTYPE and
   TRY/CATCH/ENDTRY. CATCH aligns with TRY and starts another indented body;
diff --git a/plugin/ThirdPartyNotices.md b/plugin/ThirdPartyNotices.md
index 581724f..6bb976c 100644
--- a/plugin/ThirdPartyNotices.md
+++ b/plugin/ThirdPartyNotices.md
@@ -6,7 +6,9 @@ Microsoft VS Code’s built-in C++ extension. Exception keywords reuse its
 `keyword.control.throw.cpp` scopes; ENDTRY uses the TRY scope. Selection, class,
 access, constant, template, operator and external-declaration highlighting also
 reuse C++ scopes. OVERLOAD borrows a declaration modifier, #LIB borrows #pragma
-and INCLUDE_OBJ borrows #include; ellipses use the C++ variadic scope.
+and INCLUDE_OBJ borrows #include; ellipses use the C++ variadic scope. Literal
+escapes reuse `constant.character.escape`; the adjacent C string suffix borrows
+`keyword.other.suffix.literal.user-defined.string.cpp`.
 
 https://github.com/microsoft/vscode/tree/main/extensions/cpp
 
diff --git a/plugin/examples/highlighting.cps b/plugin/examples/highlighting.cps
index b5deaee..f80fd8f 100644
--- a/plugin/examples/highlighting.cps
+++ b/plugin/examples/highlighting.cps
@@ -17,7 +17,10 @@ DECLARE Letter : CHAR
 Ready <- TRUE AND NOT FALSE
 Total <- -3.5 + 2 * 4 / 2
 Text <- "CIE // [()]" & " pseudocode"
-Letter <- 'A'
+Letter <- '\''
+Text <- "Quote: \"; slash: \\; byte: \x41; Unicode: \u0042"
+DECLARE RawText : C_STRING
+RawText <- "Native text"C
 
 FOR Index <- 1 TO 3 STEP 1
 	Values[Index] <- Index MOD 2
diff --git a/plugin/src/lexical.js b/plugin/src/lexical.js
index ff01166..fff8091 100644
--- a/plugin/src/lexical.js
+++ b/plugin/src/lexical.js
@@ -1,7 +1,7 @@
 'use strict';
 
 // Preserve UTF-16 offsets and newlines, masking CIE quotes and // comments.
-// Unlike C++, CIE has no escapes or multiline literals.
+// Backslash escapes protect the next character; literals end at each newline.
 function codeOnly(text) {
 	let quote = null;
 	let comment = false;
@@ -15,12 +15,11 @@ function codeOnly(text) {
 		} else if (comment) {
 			result.push(' ');
 		} else if (quote) {
-			if (char === quote) quote = null;
 			result.push(' ');
-		} else if (text.slice(i, i + 3) === "'''") {
-			// The compiler accepts a quote character as a three-quote CHAR literal.
-			result.push(' ', ' ', ' ');
-			i += 2;
+			if (char === '\\' && i + 1 < text.length && !/[\r\n]/.test(text[i + 1])) {
+				result.push(' ');
+				i++;
+			} else if (char === quote) quote = null;
 		} else if (char === '"' || char === "'") {
 			quote = char;
 			result.push(' ');
diff --git a/plugin/syntaxes/cie-pseudocode.tmLanguage.json b/plugin/syntaxes/cie-pseudocode.tmLanguage.json
index d0830a8..cdf175a 100644
--- a/plugin/syntaxes/cie-pseudocode.tmLanguage.json
+++ b/plugin/syntaxes/cie-pseudocode.tmLanguage.json
@@ -160,40 +160,58 @@
     "strings": {
       "patterns": [
         {
-          "match": "'''",
-          "name": "string.quoted.single.cie-pseudocode"
-        },
-        {
           "begin": "\"",
-          "end": "\"|$",
+          "end": "(\")(C)?|$",
           "beginCaptures": {
             "0": {
               "name": "punctuation.definition.string.begin.cie-pseudocode"
             }
           },
           "endCaptures": {
-            "0": {
+            "1": {
               "name": "punctuation.definition.string.end.cie-pseudocode"
+            },
+            "2": {
+              "name": "keyword.other.suffix.literal.user-defined.string.cpp"
             }
           },
           "name": "string.quoted.double.cie-pseudocode",
-          "patterns": []
+          "patterns": [
+            {
+              "include": "#escapes"
+            }
+          ]
         },
         {
           "begin": "'",
-          "end": "'|$",
+          "end": "(')|$",
           "beginCaptures": {
             "0": {
               "name": "punctuation.definition.string.begin.cie-pseudocode"
             }
           },
           "endCaptures": {
-            "0": {
+            "1": {
               "name": "punctuation.definition.string.end.cie-pseudocode"
             }
           },
           "name": "string.quoted.single.cie-pseudocode",
-          "patterns": []
+          "patterns": [
+            {
+              "include": "#escapes"
+            }
+          ]
+        }
+      ]
+    },
+    "escapes": {
+      "patterns": [
+        {
+          "match": "\\\\(?:['\"?\\\\abfnrtv]|[0-7]{1,3}|x[0-9A-Fa-f]+|u[0-9A-Fa-f]{4}|U[0-9A-Fa-f]{8})",
+          "name": "constant.character.escape"
+        },
+        {
+          "match": "\\\\[^\\r\\n]"
         }
       ]
     }
diff --git a/plugin/test/brackets.test.js b/plugin/test/brackets.test.js
index e144335..a62e657 100644
--- a/plugin/test/brackets.test.js
+++ b/plugin/test/brackets.test.js
@@ -24,8 +24,11 @@ test('comments and quotes are ignored with correct UTF-16 offsets after Unicode
 	assert.deepEqual(unmatchedBrackets(unfinished), [unfinished.length - 1]);
 	const unicode = 'OUTPUT "Hello😀"\r\nA[';
 	assert.deepEqual(unmatchedBrackets(unicode), [unicode.length - 1]);
-	// Backslashes do not escape CIE quotes.
-	assert.deepEqual(unmatchedBrackets('"x\\"['), [4]);
-	const quoteChar = "OUTPUT ''' [";
+	assert.deepEqual(unmatchedBrackets(String.raw`OUTPUT "x\"[(//"`), []);
+	const backslash = String.raw`OUTPUT "x\\" [`;
+	assert.deepEqual(unmatchedBrackets(backslash), [backslash.length - 1]);
+	const quoteChar = String.raw`OUTPUT '\'' [`;
 	assert.deepEqual(unmatchedBrackets(quoteChar), [quoteChar.length - 1]);
+	const escaped = 'OUTPUT "😀\\"["\r\nA[';
+	assert.deepEqual(unmatchedBrackets(escaped), [escaped.length - 1]);
 });
diff --git a/plugin/test/highlighting.test.js b/plugin/test/highlighting.test.js
index bfdf0a2..8be3d9c 100644
--- a/plugin/test/highlighting.test.js
+++ b/plugin/test/highlighting.test.js
@@ -122,10 +122,29 @@ test('comments, strings, lowercase and keyword prefixes cannot become keywords',
 		assert.ok(!scopesAt(text, text.split(' ')[0]).some(
 			s => /^(keyword|storage)\./.test(s)), text);
 	}
-	// CIE has neither backslash escapes nor multiline string literals.
+	// An unfinished literal must not carry its string scope onto the next line.
 	let state = grammar.tokenizeLine('OUTPUT "unfinished').ruleStack;
 	assert.ok(grammar.tokenizeLine('IF X THEN', state).tokens[0].scopes.includes(
 		'keyword.control.cie-pseudocode'));
-	assert.ok(scopesAt('OUTPUT "x\\" + 1', '+').includes('keyword.operator.cie-pseudocode'));
-	assert.ok(scopesAt("OUTPUT ''' + 1", '+').includes('keyword.operator.cie-pseudocode'));
+});
+
+test('literal escapes protect quotes and reuse C++ escape and suffix scopes', () => {
+	for (const escape of [String.raw`\'`, String.raw`\"`, String.raw`\\`, String.raw`\?`,
+		...['a', 'b', 'f', 'n', 'r', 't', 'v', '0', '101', 'x4142', 'u0041', 'U0001F600']
+			.map(value => `\\${value}`)]) {
+		const text = `OUTPUT "${escape} IF // [" + 1`;
+		assert.ok(scopesAt(text, escape).includes('constant.character.escape'), escape);
+		assert.ok(scopesAt(text, 'IF').includes('string.quoted.double.cie-pseudocode'), escape);
+		assert.ok(scopesAt(text, '+').includes('keyword.operator.cie-pseudocode'), escape);
+	}
+	assert.ok(scopesAt(String.raw`OUTPUT '\'' + 1`, '+').includes('keyword.operator.cie-pseudocode'));
+	assert.ok(scopesAt(String.raw`OUTPUT '\x41'`, String.raw`\x41`).includes('constant.character.escape'));
+	assert.ok(scopesAt('"text"C', 'C').includes('keyword.other.suffix.literal.user-defined.string.cpp'));
+	for (const text of ['"text" C', '"text"c', "'x'C"]) {
+		assert.ok(!scopesAt(text, text.at(-1)).includes('keyword.other.suffix.literal.user-defined.string.cpp'));
+	}
+	for (const text of ['OUTPUT "unfinished', 'OUTPUT "unfinished\\', "OUTPUT '''"]) {
+		const state = grammar.tokenizeLine(text).ruleStack;
+		assert.ok(grammar.tokenizeLine('IF Ready THEN', state).tokens[0].scopes.includes('keyword.control.cie-pseudocode'));
+	}
 });
diff --git a/plugin/test/indentation.test.js b/plugin/test/indentation.test.js
index 1a9d91e..9140ed3 100644
--- a/plugin/test/indentation.test.js
+++ b/plugin/test/indentation.test.js
@@ -75,3 +75,10 @@ test('blank lines are empty when formatted and indented at the caret after Enter
 		assert.deepEqual(indentationEdits(lines, line, line, true), [{ line, length: 3, text: '\t' }]);
 	}
 });
+
+test('escaped quotes keep block keywords inside literals masked', () => {
+	assert.deepEqual(lineDepths([
+		String.raw`IF Text = "quoted \" THEN"`,
+		'THEN', String.raw`OUTPUT "escaped \" ENDIF //"`, 'ENDIF'
+	]), [0, 0, 1, 0]);
+});