Interpret escapes in symbols

Author: mdaines

src/parser/index.js

@@ -1,5 +1,25 @@
 import { parser } from "./rules.js";
 
+const ESCAPES = {
+  "0": "\0",
+  "n": "\n",
+  "r": "\r",
+  "v": "\v",
+  "t": "\t",
+  "b": "\b",
+  "f": "\f",
+  "\n": "",
+  "\r": ""
+};
+
+function interpretEscapes(str) {
+  return str.replaceAll(/\\(?:x([0-9a-fA-F]{2})|u(?:\{([0-9a-fA-F]+)\}|([0-9a-fA-F]{4})))/g, function(_, hex, u1, u2) {
+    return String.fromCodePoint(parseInt(hex || u1 || u2, 16));
+  }).replaceAll(/\\(.)/gs, function(_, char) {
+    return ESCAPES[char] || char;
+  });
+}
+
 export default function(src) {
   let tree = parser.parse(src);
   let cursor = tree.cursor();
@@ -27,7 +47,7 @@ export default function(src) {
         production.push(symbol);
       }
     } else if (cursor.name === "QuotedSymbol") {
-      let symbol = src.slice(cursor.from + 1, cursor.to - 1);
+      let symbol = interpretEscapes(src.slice(cursor.from + 1, cursor.to - 1));
 
       if (typeof production === "undefined") {
         head = symbol;

test/parser.test.js

@@ -37,33 +37,56 @@ describe("parser", function() {
   });
 
   it("quoted symbols can contain escapes", function() {
-    assert.deepStrictEqual(parser(`"\\"" -> .`), { productions: [[`\\"`]] });
-    assert.deepStrictEqual(parser(`"\\'" -> .`), { productions: [[`\\'`]] });
-    assert.deepStrictEqual(parser(`'\\'' -> .`), { productions: [[`\\'`]] });
-    assert.deepStrictEqual(parser(`'\\"' -> .`), { productions: [[`\\"`]] });
-    assert.deepStrictEqual(parser(`"\\\\" -> .`), { productions: [[`\\\\`]] });
-
-    assert.deepStrictEqual(parser(`"\\b" -> .`), { productions: [[`\\b`]] });
-    assert.deepStrictEqual(parser(`"\\f" -> .`), { productions: [[`\\f`]] });
-    assert.deepStrictEqual(parser(`"\\n" -> .`), { productions: [[`\\n`]] });
-    assert.deepStrictEqual(parser(`"\\r" -> .`), { productions: [[`\\r`]] });
-    assert.deepStrictEqual(parser(`"\\t" -> .`), { productions: [[`\\t`]] });
-    assert.deepStrictEqual(parser(`"\\v" -> .`), { productions: [[`\\v`]] });
-    assert.deepStrictEqual(parser(`"\\0" -> .`), { productions: [[`\\0`]] });
-
-    assert.deepStrictEqual(parser(`"\\\n" -> .`), { productions: [[`\\\n`]] });
+    // QUOTATION MARK and APOSTROPHE
+    assert.deepStrictEqual(parser(`"\\"" -> .`), { productions: [[`"`]] });
+    assert.deepStrictEqual(parser(`"\\'" -> .`), { productions: [[`'`]] });
+    assert.deepStrictEqual(parser(`'\\'' -> .`), { productions: [[`'`]] });
+    assert.deepStrictEqual(parser(`'\\"' -> .`), { productions: [[`"`]] });
+    assert.deepStrictEqual(parser(`"\\\\" -> .`), { productions: [[`\\`]] });
+
+    // C-style escapes
+    assert.deepStrictEqual(parser(`"\\b" -> .`), { productions: [[`\b`]] });
+    assert.deepStrictEqual(parser(`"\\f" -> .`), { productions: [[`\f`]] });
+    assert.deepStrictEqual(parser(`"\\n" -> .`), { productions: [[`\n`]] });
+    assert.deepStrictEqual(parser(`"\\r" -> .`), { productions: [[`\r`]] });
+    assert.deepStrictEqual(parser(`"\\t" -> .`), { productions: [[`\t`]] });
+    assert.deepStrictEqual(parser(`"\\v" -> .`), { productions: [[`\v`]] });
+    assert.deepStrictEqual(parser(`"\\0" -> .`), { productions: [[`\0`]] });
+
+    // Identity escapes
+    assert.deepStrictEqual(parser(`"\\z" -> .`), { productions: [[`z`]] });
+
+    // Escaped line terminator
+    assert.deepStrictEqual(parser(`"\\\n" -> .`), { productions: [[`\n`]] });
+    assert.deepStrictEqual(parser(`"\\\r" -> .`), { productions: [[`\r`]] });
 
     // COPYRIGHT SIGN
-    assert.deepStrictEqual(parser(`"\\xA9" -> .`), { productions: [[`\\xA9`]] });
-    assert.deepStrictEqual(parser(`"\\xa9" -> .`), { productions: [[`\\xa9`]] });
-    assert.deepStrictEqual(parser(`"\\u00A9" -> .`), { productions: [[`\\u00A9`]] });
-    assert.deepStrictEqual(parser(`"\\u00a9" -> .`), { productions: [[`\\u00a9`]] });
+    assert.deepStrictEqual(parser(`"\\xA9" -> .`), { productions: [[`\xA9`]] });
+    assert.deepStrictEqual(parser(`"\\xa9" -> .`), { productions: [[`\xa9`]] });
+    assert.deepStrictEqual(parser(`"\\u00A9" -> .`), { productions: [[`\u00A9`]] });
+    assert.deepStrictEqual(parser(`"\\u00a9" -> .`), { productions: [[`\u00a9`]] });
+    assert.deepStrictEqual(parser(`"\\u00a9" -> .`), { productions: [[`\u{00A9}`]] });
+    assert.deepStrictEqual(parser(`"\\u00a9" -> .`), { productions: [[`\u{00a9}`]] });
 
     // BLACK HEART SUIT
-    assert.deepStrictEqual(parser(`"\\u2665" -> .`), { productions: [["\\u2665"]] });
+    assert.deepStrictEqual(parser(`"\\u2665" -> .`), { productions: [[`\u2665`]] });
+    assert.deepStrictEqual(parser(`"\\u{2665}" -> .`), { productions: [[`\u{2665}`]] });
 
     // TETRAGRAM FOR CENTRE
-    assert.deepStrictEqual(parser(`"\\u{1D306}" -> .`), { productions: [["\\u{1D306}"]] });
+    assert.deepStrictEqual(parser(`"\\u{1D306}" -> .`), { productions: [[`\u{1D306}`]] });
+
+    // AMPERSAND (\x and two hex digits)
+    assert.deepStrictEqual(parser(`"\\x2665" -> .`), { productions: [[`\x2665`]] });
+
+    // MODIFIER LETTER CAPITAL D (\u without braces)
+    assert.deepStrictEqual(parser(`"\\u1D306" -> .`), { productions: [[`\u1D306`]] });
+  });
+
+  it("quoted symbols can contain multiple escapes", function() {
+    assert.deepStrictEqual(
+      parser(`"\\" \\0 \\xA9 \\u00A9 \\u{2665}" -> .`),
+      { productions: [[`" \0 \xA9 \u00A9 \u{2665}`]] }
+    );
   });
 
   it("nonterminals don't need to be capitalized", function() {