Merge pull request #6 from cto-af/export-js

Generate JS module from builder

builder.js

@@ -1164,4 +1164,59 @@ export class UnicodeTrieBuilder {
 
     return buf;
   }
+
+  /**
+   * @typedef {object} ModuleOptions
+   * @prop {string=} [version] Version of the source file, usually the Unicode
+   *   version.
+   * @prop {string=} [date] Date the source file was created.  Can be parsed
+   *   from most UCD files.
+   * @prop {string} [name="Trie"] Name exported from the module with the Trie
+   *   instance.
+   * @prop {string} [quot='"'] Quote.  Should be single or double.
+   * @prop {string} [semi=";"] Include semicolons? Pass in "" to disable.
+   * @prop {string} [pkg="@cto.af/unicode-trie"] Package name for this
+   *   package.  Mostly useful for internal tooling.
+   */
+
+  /**
+   * Create a string version of a JS module that will reconstitute this trie.
+   * Suitable for saving to a .mjs file.
+   *
+   * @param {ModuleOptions} [opts={}]
+   * @returns {string}
+   */
+  toModule(opts = {}) {
+    const { name, quot: q, semi: s, version, date, pkg } = {
+      name: "Trie",
+      quot: '"',
+      semi: ";",
+      pkg: "@cto.af/unicode-trie",
+      ...opts,
+    };
+    const buf = this.toBuffer();
+    let ret = `import { Buffer } from ${q}buffer${q}${s}\n`;
+    ret += `import { UnicodeTrie } from ${q}${pkg}${q}${s}\n\n`;
+    if (version) {
+      ret += `export const version = ${q}${version}${q}${s}\n`;
+    }
+    if (date) {
+      ret += `export const inputFileDate = new Date(${q}${new Date(date).toISOString()}${q})${s}\n`;
+    }
+    ret += `\
+export const generatedDate = new Date(${q}${new Date().toISOString()}${q})${s}
+export const ${name} = new UnicodeTrie(Buffer.from(
+  \`${buf.toString("base64").split(/(.{72})/).filter(x => x).join("\n   ")}\`,
+  ${q}base64${q}
+))${s}
+/**
+ * @type {Record<string, number>}
+ */
+export const names = Object.fromEntries(
+  ${name}.values.map((v, i) => [v, i])
+)${s}
+export const values = ${name}.values${s}
+`;
+    return ret;
+  }
 }

examples/genLineBreak.js

@@ -47,11 +47,6 @@ for (const line of txt.split("\n")) {
   }
 }
 
-const buf = t.toBuffer();
-
-await fs.writeFile("lineBreak.js", `\
-import { UnicodeTrie } from "../index.js";
-export const LineBreak = new UnicodeTrie(Buffer.from("${buf.toString("base64")}", "base64"));
-`);
+await fs.writeFile("lineBreak.js", t.toModule({ name: "LineBreak", pkg: "../index.js" }));
 
 console.log(`Ranges: ${ranges}\nSingle: ${single}\nTotal: ${total}`);

examples/getLineBreak.js

@@ -3,5 +3,6 @@
 import { LineBreak } from "./lineBreak.js";
 
 for (const num of process.argv.slice(2)) {
-  console.log(LineBreak.get(parseInt(num, 16)));
+  const n = parseInt(num, 16);
+  console.log(`U+${n.toString(16).padStart(4, "0")}:`, LineBreak.getString(n));
 }

test/test.js

@@ -294,4 +294,12 @@ describe("unicode trie", () => {
       }));
     }
   });
+
+  it("generates a module", () => {
+    const trie = new UnicodeTrieBuilder(0, 99);
+    let m = trie.toModule();
+    assert.match(m, /export const Trie/);
+    m = trie.toModule({ version: "1.0.0", date: 1, name: "Foo", quot: "'", semi: "" });
+    assert.match(m, /export const Foo/);
+  });
 });

types/builder.d.ts

@@ -78,6 +78,56 @@ export class UnicodeTrieBuilder {
      * @returns {Buffer}
      */
     toBuffer(): Buffer;
+    /**
+     * @typedef {object} ModuleOptions
+     * @prop {string=} [version] Version of the source file, usually the Unicode
+     *   version.
+     * @prop {string=} [date] Date the source file was created.  Can be parsed
+     *   from most UCD files.
+     * @prop {string} [name="Trie"] Name exported from the module with the Trie
+     *   instance.
+     * @prop {string} [quot='"'] Quote.  Should be single or double.
+     * @prop {string} [semi=";"] Include semicolons? Pass in "" to disable.
+     * @prop {string} [pkg="@cto.af/unicode-trie"] Package name for this
+     *   package.  Mostly useful for internal tooling.
+     */
+    /**
+     * Create a string version of a JS module that will reconstitute this trie.
+     * Suitable for saving to a .mjs file.
+     *
+     * @param {ModuleOptions} [opts={}]
+     * @returns {string}
+     */
+    toModule(opts?: {
+        /**
+         * Version of the source file, usually the Unicode
+         * version.
+         */
+        version?: string | undefined;
+        /**
+         * Date the source file was created.  Can be parsed
+         * from most UCD files.
+         */
+        date?: string | undefined;
+        /**
+         * Name exported from the module with the Trie
+         * instance.
+         */
+        name?: string | undefined;
+        /**
+         * Quote.  Should be single or double.
+         */
+        quot?: string | undefined;
+        /**
+         * Include semicolons? Pass in "" to disable.
+         */
+        semi?: string | undefined;
+        /**
+         * Package name for this
+         * package.  Mostly useful for internal tooling.
+         */
+        pkg?: string | undefined;
+    } | undefined): string;
     #private;
 }
 import { UnicodeTrie } from "./index.js";