一部のクエリパラメータを消すように

packages/lib/package.json

@@ -16,9 +16,11 @@
   "dependencies": {
     "@mozilla/readability": "^0.5.0",
     "@read-stack/tsconfig": "workspace:*",
+    "@types/picomatch": "^2.3.3",
     "cheerio": "1.0.0-rc.12",
     "iconv-lite": "^0.6.3",
     "jsdom": "^23.0.1",
+    "picomatch": "^4.0.1",
     "zod": "^3.22.4"
   },
   "devDependencies": {

packages/lib/src/crawler/index.ts

@@ -1,6 +1,7 @@
 import { fetchArticleFromNote } from '@/crawler/note';
 import { fetchArticleFromQiita } from '@/crawler/qiita';
 import { fetchUsingReadability } from '@/crawler/readability';
+import { trimUrl } from '@/crawler/trim';
 import { fetchArticleFromZenn } from '@/crawler/zenn';
 
 export interface ArticleResponse {
@@ -12,19 +13,21 @@ export interface ArticleResponse {
 
 // server-side only
 export const fetchArticle = async (
-  url: string,
+  url: string
 ): Promise<ArticleResponse | null> => {
-  if (/https?:\/\/note.com\/[^/]+\/n\/[^/]+/.exec(url)) {
-    return fetchArticleFromNote(url);
+  const trimmedUrl = trimUrl(url);
+
+  if (/https?:\/\/note.com\/[^/]+\/n\/[^/]+/.exec(trimmedUrl)) {
+    return fetchArticleFromNote(trimmedUrl);
   }
 
-  if (/https?:\/\/qiita.com\/[^/]+\/items\/[^/]+/.exec(url)) {
-    return fetchArticleFromQiita(url);
+  if (/https?:\/\/qiita.com\/[^/]+\/items\/[^/]+/.exec(trimmedUrl)) {
+    return fetchArticleFromQiita(trimmedUrl);
   }
 
-  if (/https?:\/\/zenn.dev\/[^/]+\/articles\/[^/]+/.exec(url)) {
-    return fetchArticleFromZenn(url);
+  if (/https?:\/\/zenn.dev\/[^/]+\/articles\/[^/]+/.exec(trimmedUrl)) {
+    return fetchArticleFromZenn(trimmedUrl);
   }
 
-  return fetchUsingReadability(url);
+  return fetchUsingReadability(trimmedUrl);
 };

packages/lib/src/crawler/trim.ts

@@ -0,0 +1,37 @@
+import { isMatch } from 'picomatch';
+
+export interface TrimData {
+  host: string[];
+  queries: string[];
+}
+
+const trimData: TrimData[] = [
+  {
+    host: ['*'],
+    queries: [
+      'utm_source',
+      'utm_medium',
+      'utm_campaign',
+      'utm_term',
+      'utm_content',
+    ],
+  },
+  {
+    host: ['twitter.com', 'x.com'],
+    queries: ['s', 't'],
+  },
+];
+
+export const trimUrl = (url: string): string => {
+  const urlObj = new URL(url);
+  const data = trimData.filter((d) => isMatch(urlObj.host, d.host));
+
+  data.forEach(({ queries }) => {
+    queries.forEach((query) => {
+      // eslint-disable-next-line drizzle/enforce-delete-with-where -- drizzle関係なし
+      urlObj.searchParams.delete(query);
+    });
+  });
+
+  return urlObj.toString();
+};