Change defaults to be more unicodey

robsimmons · Dec 27, 2024 · 689d937 · 689d937
1 parent 444331b
commit 689d937
Show file tree

Hide file tree

Showing 3 changed files with 21 additions and 22 deletions.
diff --git a/src/client.test.ts b/src/client.test.ts
@@ -51,8 +51,8 @@ test('Basic operation', () => {
 
 test('String escapes', () => {
   expect(
-    solutions(new Dusa('res "\\0\\b\\f\\n\\r\\t\\v\\\'\\"\\\\\\x12\\u{12}\\u{2601}".')),
-  ).toStrictEqual(['res "\\x00\\x08\\x0c\\n\\x0d\\x09\\x0b\'\\"\\\\\\x12\\x12\\u{2601}"']);
+    solutions(new Dusa('res "\\0\\b\\f\\n\\r\\t\\v\\\'\\"\\\\\\x12\\u{12}☁\\u{2601}".')),
+  ).toStrictEqual(['res "\\x00\\x08\\x0c\\n\\x0d\\x09\\x0b\'\\"\\\\\\x12\\x12☁☁"']);
 
   expect(runForDusaError('a is "\\u{d901}".\n')).toStrictEqual([
     'Cannot encode lone surrogate \\u{d901}',

diff --git a/src/datastructures/data.test.ts b/src/datastructures/data.test.ts
@@ -41,8 +41,8 @@ test('Internalizing basic types', () => {
     '1',
     '-1',
     '"abc"',
-    '"\\"\\u{1f98a}\\""',
-    '"\'\\n\\x09\\x9d\\u{126}\\x00\\\\"',
+    '"\\"🦊\\""',
+    '"\'\\n\\x09\\x9dĦ\\x00\\\\"',
     'a',
     'b',
     'c',

diff --git a/src/datastructures/data.ts b/src/datastructures/data.ts
@@ -161,28 +161,27 @@ export function escapeString(input: string): string {
   const escaped = [];
   let i = 0;
   while (i < input.length) {
-    if (input.codePointAt(i)! > 0xffff) {
-      escaped.push(`\\u{${input.codePointAt(i)!.toString(16)}}`);
-      i += 2;
-    } else {
-      const ch = input.charAt(i);
-      if (ch.charCodeAt(0) > 0xff) {
-        escaped.push(`\\u{${input.charCodeAt(i).toString(16)}}`);
-      } else if (ch.match(/[ !#-[\]-~]/)) {
-        escaped.push(ch);
-      } else if (ch === '\\') {
+    const code = input.codePointAt(i)!;
+    const ch = String.fromCodePoint(code);
+    switch (ch) {
+      case '\\':
         escaped.push('\\\\');
-      } else if (ch === '"') {
+        break;
+      case '"':
         escaped.push('\\"');
-      } else if (ch === '\n') {
+        break;
+      case '\n':
         escaped.push('\\n');
-      } else if (ch.charCodeAt(0) >= 16) {
-        escaped.push(`\\x${input.charCodeAt(i).toString(16)}`);
-      } else {
-        escaped.push(`\\x0${input.charCodeAt(i).toString(16)}`);
-      }
-      i += 1;
+        break;
+      default:
+        if (code < 32 || (128 <= code && code < 160)) {
+          escaped.push(`\\x${input.charCodeAt(i).toString(16).padStart(2, '0')}`);
+        } else {
+          escaped.push(ch);
+        }
     }
+    i += 1;
+    if (code > 0xffff) i += 1;
   }
   return escaped.join('');
 }