feat: parse html for url

HiddenUrl/Program.cs

@@ -0,0 +1,79 @@
+// See https://aka.ms/new-console-template for more information
+using System;
+using System.Net.Http;
+using System.Threading.Tasks;
+using HtmlAgilityPack;
+using System.Linq;
+
+
+var CHALLENGE_URL = "https://tns4lpgmziiypnxxzel5ss5nyu0nftol.lambda-url.us-east-1.on.aws/challenge";
+var PAGE_CACHE = "challenge.html";
+
+try
+{
+    string html;
+
+    if (File.Exists(PAGE_CACHE))
+    {
+        Console.WriteLine("Loading HTML from cache...");
+        html = await File.ReadAllTextAsync(PAGE_CACHE);
+    }
+    else
+    {
+        Console.WriteLine("Downloading HTML...");
+        try
+        {
+            using HttpClient client = new HttpClient();
+            html = await client.GetStringAsync(CHALLENGE_URL);
+
+            // Write to cache
+            await File.WriteAllTextAsync(PAGE_CACHE, html);
+            Console.WriteLine("HTML cached to file.");
+        }
+        catch (HttpRequestException e)
+        {
+            Console.WriteLine($"HTTP error: {e.Message}");
+            return;
+        }
+    }
+
+    var doc = new HtmlDocument();
+    doc.LoadHtml(html);
+
+    // Find all <section> with data-id starting with "92"
+    var sections = doc.DocumentNode.SelectNodes("//section[starts-with(@data-id, '92')]");
+    if (sections == null) return;
+
+    foreach (var section in sections)
+    {
+        // Find <article> children with data-class ending in "45"
+        var articles = section.SelectNodes(".//article[substring(@data-class, string-length(@data-class) - 1) = '45']");
+        if (articles == null) continue;
+
+        foreach (var article in articles)
+        {
+            // Find <div> children with data-tag containing "78"
+            var divs = article.SelectNodes(".//div[contains(@data-tag, '78')]");
+            if (divs == null) continue;
+
+            foreach (var div in divs)
+            {
+                // Find all <b> tags nested inside this div
+                var bTags = div.SelectNodes(".//b[contains(concat(' ', normalize-space(@class), ' '), ' ref ')]");
+                if (bTags == null) continue;
+
+                foreach (var b in bTags)
+                {
+                    Console.Write(b.GetAttributeValue("value", null));
+                }
+            }
+        }
+    }
+    // end with newline
+    Console.WriteLine();
+}
+catch
+{
+    Console.WriteLine("Unable to get HTML from page!");
+}
+