From 9dd6f216ad18c09a28eaad67411d3a0e7f57a04f Mon Sep 17 00:00:00 2001
From: MohamedBassem <me@mbassem.com>
Date: Sat, 21 Sep 2024 17:49:47 +0000
Subject: feature(web): Add support for importing bookmarks from Pocket

---
 apps/web/lib/importBookmarkParser.ts | 68 ++++++++++++++++++++++++++++++++++++
 1 file changed, 68 insertions(+)
 create mode 100644 apps/web/lib/importBookmarkParser.ts

(limited to 'apps/web/lib/importBookmarkParser.ts')

diff --git a/apps/web/lib/importBookmarkParser.ts b/apps/web/lib/importBookmarkParser.ts
new file mode 100644
index 00000000..e1b21a79
--- /dev/null
+++ b/apps/web/lib/importBookmarkParser.ts
@@ -0,0 +1,68 @@
+// Copied from https://gist.github.com/devster31/4e8c6548fd16ffb75c02e6f24e27f9b9
+import * as cheerio from "cheerio";
+
+export interface ParsedBookmark {
+  title: string;
+  url?: string;
+  tags: string[];
+  addDate?: number;
+}
+
+export async function parseNetscapeBookmarkFile(
+  file: File,
+): Promise<ParsedBookmark[]> {
+  const textContent = await file.text();
+
+  if (!textContent.startsWith("<!DOCTYPE NETSCAPE-Bookmark-file-1>")) {
+    throw Error("The uploaded html file does not seem to be a bookmark file");
+  }
+
+  const $ = cheerio.load(textContent);
+
+  return $("a")
+    .map(function (_index, a) {
+      const $a = $(a);
+      const addDate = $a.attr("add_date");
+      let tags: string[] = [];
+      try {
+        tags = $a.attr("tags")?.split(",") ?? [];
+      } catch (e) {
+        /* empty */
+      }
+      return {
+        title: $a.text(),
+        url: $a.attr("href"),
+        tags: tags,
+        addDate: typeof addDate === "undefined" ? undefined : parseInt(addDate),
+      };
+    })
+    .get();
+}
+
+export async function parsePocketBookmarkFile(
+  file: File,
+): Promise<ParsedBookmark[]> {
+  const textContent = await file.text();
+
+  const $ = cheerio.load(textContent);
+
+  return $("a")
+    .map(function (_index, a) {
+      const $a = $(a);
+      const addDate = $a.attr("time_added");
+      let tags: string[] = [];
+      const tagsStr = $a.attr("tags");
+      try {
+        tags = tagsStr && tagsStr.length > 0 ? tagsStr.split(",") : [];
+      } catch (e) {
+        /* empty */
+      }
+      return {
+        title: $a.text(),
+        url: $a.attr("href"),
+        tags: tags,
+        addDate: typeof addDate === "undefined" ? undefined : parseInt(addDate),
+      };
+    })
+    .get();
+}
-- 
cgit v1.2.3-70-g09d2