Add named custom dictionary support

src/__tests__/__snapshots__/index.test.ts.snap

@@ -12,12 +12,22 @@ exports[`index.ts > was the change deliberate? 1`] = `
   "compressToUint8Array": [Function],
   "convertFromUint8Array": [Function],
   "convertToUint8Array": [Function],
+  "customBase16Dict": "0123456789ABCDEF",
+  "customBase32Dict": "ABCDEFGHIJKLMNOPQRSTUVWXYZ234567",
+  "customBase32HexDict": "0123456789ABCDEEFGHIJKLMNOPQRSTU",
+  "customBase36Dict": "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ",
+  "customBase58Dict": "123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz",
+  "customBase62Dict": "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz",
+  "customBase64Dict": "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=",
+  "customBase95Dict": " !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_\`abcdefghijklmnopqrstuvwxyz{|}~",
+  "customSafeDict": " !#$%&()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[]^_abcdefghijklmnopqrstuvwxyz{|}~",
   "decompress": [Function],
   "decompressFromBase64": [Function],
   "decompressFromCustom": [Function],
   "decompressFromEncodedURIComponent": [Function],
   "decompressFromUTF16": [Function],
   "decompressFromUint8Array": [Function],
+  "getCustomDictionary": [Function],
   "loadBinaryFile": [Function],
   "saveBinaryFile": [Function],
 }

src/cli.ts

@@ -19,7 +19,7 @@ program
     .description("Use lz-string to compress or decompress a file")
     .addOption(new Option("-d, --decompress", "if unset then this will compress"))
     .addOption(new Option("-e, --encoder <type>", "character encoding to use").choices(encoders).default("raw"))
-    .addOption(new Option("-c, --custom", "dictionary for custom encoder"))
+    .addOption(new Option("-c, --custom <dict>", "dictionary for custom encoder"))
     .addOption(new Option("-v, --verify", "verify before returning").default(true))
     .addOption(new Option("-o, --output <output-file>", "output file, otherwise write to stdout"))
     .addOption(new Option("-q, --quiet", "don't print any error messages"))

src/custom/__tests__/__snapshots__/index.test.ts.snap

@@ -3,6 +3,16 @@
 exports[`custom/index.ts > was the change deliberate? 1`] = `
 {
   "compressToCustom": [Function],
+  "customBase16Dict": "0123456789ABCDEF",
+  "customBase32Dict": "ABCDEFGHIJKLMNOPQRSTUVWXYZ234567",
+  "customBase32HexDict": "0123456789ABCDEEFGHIJKLMNOPQRSTU",
+  "customBase36Dict": "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ",
+  "customBase58Dict": "123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz",
+  "customBase62Dict": "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz",
+  "customBase64Dict": "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=",
+  "customBase95Dict": " !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_\`abcdefghijklmnopqrstuvwxyz{|}~",
+  "customSafeDict": " !#$%&()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[]^_abcdefghijklmnopqrstuvwxyz{|}~",
   "decompressFromCustom": [Function],
+  "getCustomDictionary": [Function],
 }
 `;

src/custom/__tests__/custom.test.ts

@@ -27,9 +27,8 @@ describe("custom", () => {
     });
 
     describe("base62", () => {
-        const base62 = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789";
-        const compress = (input: string | null) => compressToCustom(input, base62);
-        const decompress = (input: string | null) => decompressFromCustom(input, base62);
+        const compress = (input: string | null) => compressToCustom(input, "base62");
+        const decompress = (input: string | null) => decompressFromCustom(input, "base62");
 
         runTestSet<string>("", compress, decompress);
     });

src/custom/compressToCustom.ts

@@ -5,10 +5,19 @@
  */
 
 import { compress } from "../raw/compress";
+import { getCustomDictionary } from "./dictionaries";
 
+/**
+ * Custom dictionary support, this is not ready for production use yet, so lock
+ * the versions if using.
+ *
+ * @alpha
+ */
 export function compressToCustom(uncompressed: string | null, dict: string): string {
     if (uncompressed == null) return "";
 
+    dict = getCustomDictionary(dict);
+
     const compressed: string = compress(uncompressed);
     const charsPerUnicodeChar: number = Math.ceil(Math.log(65536) / Math.log(dict.length));
     let res: string = "";

src/custom/decompressFromCustom.ts

@@ -5,10 +5,20 @@
  */
 
 import { decompress } from "../raw/decompress";
+import { getCustomDictionary } from "./dictionaries";
 
-export function decompressFromCustom(compressed: string | null, dict: string): string | null {
+/**
+ * Custom dictionary support, this is not ready for production use yet, so lock
+ * the versions if using.
+ *
+ * @alpha
+ */
+export function decompressFromCustom(compressed: string | null, dict: string) {
     if (compressed == null) return "";
     if (compressed == "") return null;
+
+    dict = getCustomDictionary(dict);
+
     if (dict.length < 2) return null;
 
     const charsPerUnicodeChar: number = Math.ceil(Math.log(65536) / Math.log(dict.length));

src/custom/dictionaries.ts

@@ -0,0 +1,75 @@
+/*
+ * SPDX-FileCopyrightText: 2013 Pieroxy <[email protected]>
+ *
+ * SPDX-License-Identifier: MIT
+ */
+
+/**
+ * Otherwise known as hexadecimal.
+ */
+export const customBase16Dict = "0123456789ABCDEF";
+
+/**
+ * Base32 dictionary as defined in rfc4648:
+ * https://datatracker.ietf.org/doc/html/rfc4648#section-6
+ */
+export const customBase32Dict = "ABCDEFGHIJKLMNOPQRSTUVWXYZ234567";
+
+/**
+ * Base64 dictionary using the more well-known extended hex format:
+ * https://datatracker.ietf.org/doc/html/rfc4648#section-7
+ */
+export const customBase32HexDict = "0123456789ABCDEEFGHIJKLMNOPQRSTU";
+
+/**
+ * Base36 dictionary.
+ */
+export const customBase36Dict = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ";
+
+/**
+ * Base58 dictionary.
+ */
+export const customBase58Dict = "123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz";
+
+/**
+ * Base62 dictionary.
+ */
+export const customBase62Dict = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz";
+
+/**
+ * Base64 dictionary:
+ * https://datatracker.ietf.org/doc/html/rfc4648#section-4
+ */
+export const customBase64Dict = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
+
+/**
+ * A simple dictionary for safe source-code inclusion. Effectively base91, as
+ * it contains base95 with all three quote types and the escape character
+ * removed.
+ */
+export const customSafeDict =
+    " !#$%&()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[]^_abcdefghijklmnopqrstuvwxyz{|}~";
+
+/**
+ * Base95 dictionary.
+ * (Two escaped characters in source).
+ */
+export const customBase95Dict =
+    " !\"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~";
+
+/**
+ * If the dictionary passed is a known named dictionary then return that,
+ * otherwise return the source and assume it's correct.
+ */
+export const getCustomDictionary = (dict: string) =>
+    ({
+        safe: customSafeDict,
+        base16: customBase16Dict,
+        base32: customBase32Dict,
+        base32hex: customBase32HexDict,
+        base36: customBase36Dict,
+        base58: customBase58Dict,
+        base62: customBase62Dict,
+        base64: customBase64Dict,
+        base95: customBase95Dict,
+    }[dict]) || dict;

src/custom/index.ts

@@ -6,3 +6,15 @@
 
 export { compressToCustom } from "./compressToCustom";
 export { decompressFromCustom } from "./decompressFromCustom";
+export {
+    customSafeDict,
+    customBase16Dict,
+    customBase32Dict,
+    customBase32HexDict,
+    customBase36Dict,
+    customBase58Dict,
+    customBase62Dict,
+    customBase64Dict,
+    customBase95Dict,
+    getCustomDictionary,
+} from "./dictionaries";

src/index.ts

@@ -7,7 +7,20 @@
 import { _compress } from "./_compress";
 import { _decompress } from "./_decompress";
 import { compressToBase64, decompressFromBase64 } from "./base64";
-import { compressToCustom, decompressFromCustom } from "./custom";
+import {
+    compressToCustom,
+    customBase16Dict,
+    customBase32Dict,
+    customBase32HexDict,
+    customBase36Dict,
+    customBase58Dict,
+    customBase62Dict,
+    customBase64Dict,
+    customBase95Dict,
+    customSafeDict,
+    decompressFromCustom,
+    getCustomDictionary,
+} from "./custom";
 import { compressToEncodedURIComponent, decompressFromEncodedURIComponent } from "./encodedURIComponent";
 import { loadBinaryFile, saveBinaryFile } from "./node";
 import { compress, decompress } from "./raw";
@@ -30,12 +43,22 @@ export default {
     compressToUTF16,
     convertFromUint8Array,
     convertToUint8Array,
+    customSafeDict,
+    customBase16Dict,
+    customBase32Dict,
+    customBase32HexDict,
+    customBase36Dict,
+    customBase58Dict,
+    customBase62Dict,
+    customBase64Dict,
+    customBase95Dict,
     decompress,
     decompressFromBase64,
     decompressFromCustom,
     decompressFromEncodedURIComponent,
     decompressFromUint8Array,
     decompressFromUTF16,
+    getCustomDictionary,
     loadBinaryFile,
     saveBinaryFile,
 };

tsconfig.json

@@ -11,7 +11,7 @@
         "skipLibCheck": true,
         "sourceMap": true,
         "strict": true,
-        "target": "ESNext",
+        "target": "ESNext"
     },
-    "include": ["src"],
+    "include": ["src"]
 }