feat(strings): add escape(), update unescape(), add tests

thi-ng · Sep 18, 2020 · e0d5f1e · e0d5f1e
1 parent ec94064
commit e0d5f1e
Show file tree

Hide file tree

Showing 4 changed files with 93 additions and 37 deletions.
diff --git a/packages/strings/src/escape.ts b/packages/strings/src/escape.ts
@@ -0,0 +1,78 @@
+import { U16, U32 } from "./radix";
+
+export const ESCAPES: Record<string, string> = {
+    0: "\0",
+    b: "\b",
+    t: "\t",
+    r: "\r",
+    v: "\v",
+    f: "\f",
+    n: "\n",
+    "'": "'",
+    '"': '"',
+    "\\": "\\",
+};
+
+export const ESCAPES_REV: Record<number, string> = {
+    0: "0",
+    8: "b",
+    9: "t",
+    10: "n",
+    11: "v",
+    12: "f",
+    13: "r",
+    33: "'",
+    34: '"',
+    92: "\\",
+};
+
+/**
+ * Escapes all non-ASCII characters (and well-known 0x0x control chars) using
+ * backslash escape sequences.
+ *
+ * @remarks
+ * - Well-known low-ASCII control chars will be escaped using simple `\`, e.g.
+ *   0x0a => `\n`
+ * - Non-BMP chars will be escaped using `\Uxxxxxxxx`
+ * - Chars outside 0x20 - 0x7e range will be escaped using `\uxxxxx`
+ *
+ * @param src
+ */
+export const escape = (src: string) =>
+    src
+        .replace(
+            /[\0\b\t\n\v\f\r'"\\]/g,
+            (x) => `\\${ESCAPES_REV[x.charCodeAt(0)!]}`
+        )
+        .replace(/[\ud800-\udfff]{2}/g, (x) => `\\U${U32(x.codePointAt(0)!)}`)
+        .replace(/[^\u0020-\u007e]/g, (x) => `\\u${U16(x.charCodeAt(0))}`);
+
+/**
+ * Replaces `\uxxxx` UTF-16 escapes, full `\Uxxxxxxxx` UTF-32 codepoint escapes
+ * and other well-known backslash escape sequences (see {@link ESCAPES}) with
+ * the characters they represent.
+ *
+ * @remarks
+ * Any unknown named escape sequences (e.g. `\1`) will remain as is.
+ *
+ * - https://en.wikipedia.org/wiki/UTF-16#Code_points_from_U+010000_to_U+10FFFF
+ * - https://www.unicode.org/charts/
+ * - https://www.branah.com/unicode-converter
+ *
+ * @example
+ * ```ts
+ * unescape("\\ud83d\\ude0e \\U0001f60e")
+ * // '😎'
+ * ```
+ *
+ * @param src
+ */
+export const unescape = (src: string) =>
+    src
+        .replace(/\\u([0-9a-fA-F]{4})/g, (_, id) =>
+            String.fromCharCode(parseInt(id, 16))
+        )
+        .replace(/\\U([0-9a-fA-F]{8})/g, (_, id) =>
+            String.fromCodePoint(parseInt(id, 16))
+        )
+        .replace(/\\([0btnvfr'"\\])/g, (_, id) => ESCAPES[id]);
diff --git a/packages/strings/src/index.ts b/packages/strings/src/index.ts
@@ -2,6 +2,7 @@ export * from "./api";
 export * from "./case";
 export * from "./center";
 export * from "./cursor";
+export * from "./escape";
 export * from "./float";
 export * from "./format";
 export * from "./groups";
@@ -20,7 +21,6 @@ export * from "./splice";
 export * from "./trim";
 export * from "./truncate";
 export * from "./truncate-left";
-export * from "./unescape";
 export * from "./units";
 export * from "./uuid";
 export * from "./wrap";
diff --git a/packages/strings/src/unescape.ts b/packages/strings/src/unescape.ts
diff --git a/packages/strings/test/escape.ts b/packages/strings/test/escape.ts
@@ -0,0 +1,14 @@
+import * as assert from "assert";
+import { escape, unescape } from "../src";
+
+const SRC = "\ta\nb😎c£\\\x00";
+
+describe("escape", () => {
+    it("escape", () => {
+        assert.strictEqual(escape(SRC), "\\ta\\nb\\U0001f60ec\\u00a3\\\\\\0");
+    });
+
+    it("roundtrip", () => {
+        assert.strictEqual(unescape(escape(SRC)), SRC);
+    });
+});