From f11d3f2ac641bce64c5a674c0995cf85ffc7e37e Mon Sep 17 00:00:00 2001
From: zyachel <aricla@protonmail.com>
Date: Mon, 8 Apr 2024 00:22:51 +0530
Subject: [PATCH] fix(parse): don't bail out on encountering weird characters

replace all \x3C(<) with its valid unique escape sequence
---
 fetchers/answersFetcher.js |  3 ++-
 fetchers/fetcher.js        | 12 +++++-------
 utils/parse.js             | 11 +++++++++++
 3 files changed, 18 insertions(+), 8 deletions(-)
 create mode 100644 utils/parse.js

diff --git a/fetchers/answersFetcher.js b/fetchers/answersFetcher.js
index 733a768..39d7526 100644
--- a/fetchers/answersFetcher.js
+++ b/fetchers/answersFetcher.js
@@ -5,6 +5,7 @@
 import * as cheerio from 'cheerio';
 import getAxiosInstance from '../utils/getAxiosInstance.js';
 import AppError from '../utils/AppError.js';
+import parse from '../utils/parse.js';
 
 ////////////////////////////////////////////////////////
 //                     FUNCTION
@@ -31,7 +32,7 @@ const answersFetcher = async (resourceStr, lang) => {
       if (!matches) return;
 
       // brittle logic, but works
-      const matchedPart = JSON.parse(JSON.parse(matches[1])).data;
+      const matchedPart = JSON.parse(parse(matches[1])).data;
 
       // only question block has this word
       if (typeof matchedPart.question?.viewerHasAnswered !== 'undefined') {
diff --git a/fetchers/fetcher.js b/fetchers/fetcher.js
index d3b7c8d..89b0166 100644
--- a/fetchers/fetcher.js
+++ b/fetchers/fetcher.js
@@ -5,6 +5,7 @@
 import * as cheerio from 'cheerio';
 import getAxiosInstance from '../utils/getAxiosInstance.js';
 import AppError from '../utils/AppError.js';
+import parse from '../utils/parse.js';
 
 ////////////////////////////////////////////////////////
 //                     FUNCTION
@@ -18,19 +19,16 @@ import AppError from '../utils/AppError.js';
  * await fetcher('topic/Space-Physics'); // will return 'space physics' topic object
  * await fetcher('profile/Charlie-Cheever'); // will return object containing information about charlie cheever
  */
-const fetcher = async (
-  resourceStr,
-  { keyword, lang, toEncode = true }
-) => {
+const fetcher = async (resourceStr, { keyword, lang, toEncode = true }) => {
   try {
     // as url might contain unescaped chars. so, encoding it right away
     const str = toEncode ? encodeURIComponent(resourceStr) : resourceStr;
     const axiosInstance = getAxiosInstance(lang);
     const res = await axiosInstance.get(str);
-    
+
     const $ = cheerio.load(res.data);
 
-    const regex = new RegExp(`"{\\\\"data\\\\":\\{\\\\"${keyword}.*\\}"`); // equivalent to /"\{\\"data\\":\{\\"searchConnection.*\}"/
+    const regex = new RegExp(String.raw`"{\\"data\\":\{\\"${keyword}.*?\}"`);
 
     let rawData;
     $('body script').each((i, el) => {
@@ -45,7 +43,7 @@ const fetcher = async (
 
     if (!rawData) throw new AppError("couldn't retrieve data", 500);
 
-    return JSON.parse(rawData);
+    return parse(rawData);
   } catch (err) {
     const statusCode = err.response?.status;
     if (statusCode === 404) throw new AppError('Not found', 404);
diff --git a/utils/parse.js b/utils/parse.js
new file mode 100644
index 0000000..697dd9b
--- /dev/null
+++ b/utils/parse.js
@@ -0,0 +1,11 @@
+const invalidLessThan = /\\x3C/g;
+const validLessThan = '\\u003C';
+
+/**
+ * parses and corrects invalid escape sequences
+ *  @param {string} data
+ * @returns {Record<PropertyKey, any>}
+ */
+const parse = data => JSON.parse(data.replace(invalidLessThan, validLessThan));
+
+export default parse;