From 06d2c2de916bdbc3aa612de8ea33b5a172cca706 Mon Sep 17 00:00:00 2001
From: Inga <inga-lovinde@noreply.oomza.cutegay.software>
Date: Sun, 14 Jan 2024 16:39:19 +0000
Subject: [PATCH] implemented basic english spellchecker regex

---
 .../viteThesaurusRegexPlugin.test.ts          | 43 +++++++++++++++++++
 src/build-scripts/viteThesaurusRegexPlugin.ts | 25 +++++++++++
 2 files changed, 68 insertions(+)
 create mode 100644 src/build-scripts/viteThesaurusRegexPlugin.test.ts
 create mode 100644 src/build-scripts/viteThesaurusRegexPlugin.ts

diff --git a/src/build-scripts/viteThesaurusRegexPlugin.test.ts b/src/build-scripts/viteThesaurusRegexPlugin.test.ts
new file mode 100644
index 0000000..ec2952b
--- /dev/null
+++ b/src/build-scripts/viteThesaurusRegexPlugin.test.ts
@@ -0,0 +1,43 @@
+import t from 'tap';
+
+import { getRegexStringForThesaurusPath } from './viteThesaurusRegexPlugin.js';
+
+void t.test('extractWordsFromFile', async (t) => {
+    const regexString = await getRegexStringForThesaurusPath(
+        new URL('../../build-resources/th-en-x-basic.dat', import.meta.url),
+    );
+    const regex = new RegExp(regexString, 'gi');
+
+    const getMatches = (str: string) =>
+        [...str.matchAll(regex)].map((match) => match[0]);
+
+    t.strictSame(
+        getMatches(
+            'afterthought airplane another anybody anyhow anyone anything anywhere',
+        ),
+        [],
+    );
+
+    t.strictSame(
+        getMatches(
+            'afterthought airplane another xxx anybody anyhow anyone anything anywhere',
+        ),
+        ['xxx'],
+    );
+
+    t.strictSame(
+        getMatches(
+            'xx afterthought airplane another yy anybody anyhow anyone anything anywhere zzzz',
+        ),
+        ['xx', 'yy', 'zzzz'],
+    );
+
+    t.strictSame(getMatches('xx yy zzzz'), ['xx', 'yy', 'zzzz']);
+
+    t.strictSame(
+        getMatches(
+            "Thing to help people escape really fast if there's a problem and everything is on fire so they decide not to go to space",
+        ),
+        [],
+    );
+});
diff --git a/src/build-scripts/viteThesaurusRegexPlugin.ts b/src/build-scripts/viteThesaurusRegexPlugin.ts
new file mode 100644
index 0000000..15491b5
--- /dev/null
+++ b/src/build-scripts/viteThesaurusRegexPlugin.ts
@@ -0,0 +1,25 @@
+import type { PathLike } from 'node:fs';
+import { extractWordsFromFile } from './oooThesaurusParser.js';
+
+/*type ViteThesaurusRegexPluginOptions = {
+    inputOooThesaurusPath: PathLike;
+    outputRegexPath: string;
+};*/
+
+const WORD_CHARACTER_REGEX_FRAGMENT = "[a-zA-Z']";
+
+export const getRegexStringForThesaurusPath = async (
+    inputOooThesaurusPath: PathLike,
+) => {
+    const words = await extractWordsFromFile(inputOooThesaurusPath);
+    return `(?<!${WORD_CHARACTER_REGEX_FRAGMENT})${WORD_CHARACTER_REGEX_FRAGMENT}+(?!${WORD_CHARACTER_REGEX_FRAGMENT})(?<!(?<!${WORD_CHARACTER_REGEX_FRAGMENT})(${words.join(
+        '|',
+    )}))`;
+};
+
+/*export const viteThesaurusRegexPlugin = ({
+    inputOooThesaurusPath,
+    outputRegexPath,
+}: ViteThesaurusRegexPluginOptions) => {
+};
+*/