From c564e79dbce6f9362ce605cbc3ffd77c430de0ff Mon Sep 17 00:00:00 2001
From: tastytea <tastytea@tastytea.de>
Date: Mon, 18 Jan 2021 08:19:22 +0100
Subject: [PATCH] Add spam filter.

---
 src/cgi.cpp | 26 ++++++++++++++++++++++++++
 src/cgi.hpp |  2 ++
 2 files changed, 28 insertions(+)

diff --git a/src/cgi.cpp b/src/cgi.cpp
index 581708a..255e096 100644
--- a/src/cgi.cpp
+++ b/src/cgi.cpp
@@ -35,6 +35,7 @@
 #include <iostream>
 #include <iterator>
 #include <map>
+#include <regex>
 #include <sstream>
 #include <stdexcept>
 #include <string>
@@ -83,6 +84,11 @@ entry_type parse_formdata()
     entry.description = cgi("description");
     entry.report_time = time::to_string(system_clock::now());
 
+    if (is_spam(entry))
+    {
+        throw SpamException{};
+    }
+
     std::uint8_t screenshot_counter{1};
     for (const auto &screenshot : cgi.getFiles())
     {
@@ -201,6 +207,26 @@ bool captcha_valid(std::uint8_t id, const string_view answer)
     return false;
 }
 
+bool is_spam(const entry_type &entry)
+{
+    using std::regex;
+    using std::regex_search;
+
+    std::ifstream file(files::get_datadir() / "spamfilter.lst");
+    if (file.good())
+    {
+        string line;
+        while (getline(file, line))
+        {
+            if (regex_search(entry.description, regex{line}))
+            {
+                return true;
+            }
+        }
+    }
+    return false;
+}
+
 } // namespace FediBlock::cgi
 
 const char *SpamException::what() const noexcept
diff --git a/src/cgi.hpp b/src/cgi.hpp
index 2151bb2..8f30302 100644
--- a/src/cgi.hpp
+++ b/src/cgi.hpp
@@ -56,6 +56,8 @@ string text2html(string text);
 // Check if the answer matches the solution we have for the id.
 bool captcha_valid(std::uint8_t id, string_view answer);
 
+bool is_spam(const entry_type &entry);
+
 } // namespace FediBlock::cgi
 
 class SpamException : public std::exception