Added rate limiting and settings fixes.

7 files changed, 346 insertions, 1 deletions

diff --git a/src/Config.c b/src/Config.c
index 0c3fc1c..c4bd1f1 100644
--- a/src/Config.c
+++ b/src/Config.c
@@ -100,6 +100,16 @@ int load_config(const char *filename, Config *config) {
           strncpy(config->engines, value, sizeof(config->engines) - 1);
           config->engines[sizeof(config->engines) - 1] = '\0';
         }
+      } else if (strcmp(section, "rate_limit") == 0) {
+        if (strcmp(key, "search_requests") == 0) {
+          config->rate_limit_search_requests = atoi(value);
+        } else if (strcmp(key, "search_interval") == 0) {
+          config->rate_limit_search_interval = atoi(value);
+        } else if (strcmp(key, "images_requests") == 0) {
+          config->rate_limit_images_requests = atoi(value);
+        } else if (strcmp(key, "images_interval") == 0) {
+          config->rate_limit_images_interval = atoi(value);
+        }
       }
     }
   }
diff --git a/src/Config.h b/src/Config.h
index ce316f6..8e68eae 100644
--- a/src/Config.h
+++ b/src/Config.h
@@ -45,6 +45,10 @@ typedef struct {
   int cache_ttl_infobox;
   int cache_ttl_image;
   char engines[512];
+  int rate_limit_search_requests;
+  int rate_limit_search_interval;
+  int rate_limit_images_requests;
+  int rate_limit_images_interval;
 } Config;
 
 int load_config(const char *filename, Config *config);
diff --git a/src/Limiter/RateLimit.c b/src/Limiter/RateLimit.c
new file mode 100644
index 0000000..3c6bbff
--- /dev/null
+++ b/src/Limiter/RateLimit.c
@@ -0,0 +1,193 @@
+#include "RateLimit.h"
+#include <arpa/inet.h>
+#include <beaker.h>
+#include <ctype.h>
+#include <netinet/in.h>
+#include <pthread.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/socket.h>
+#include <sys/un.h>
+#include <time.h>
+#include <unistd.h>
+
+typedef struct RateLimitEntry {
+  char client_key[64];
+  char scope[32];
+  time_t window_start;
+  time_t last_seen;
+  int count;
+  struct RateLimitEntry *next;
+} RateLimitEntry;
+
+extern __thread int current_client_socket;
+extern __thread char current_request_buffer[];
+
+static pthread_mutex_t rate_limit_mutex = PTHREAD_MUTEX_INITIALIZER;
+static RateLimitEntry *rate_limit_entries = NULL;
+
+static int is_blank_char(char c) {
+  return c == ' ' || c == '\t' || c == '\r' || c == '\n';
+}
+
+static void trim_copy(char *dest, size_t dest_size, const char *src,
+                      size_t src_len) {
+  while (src_len > 0 && is_blank_char(*src)) {
+    src++;
+    src_len--;
+  }
+
+  while (src_len > 0 && is_blank_char(src[src_len - 1])) {
+    src_len--;
+  }
+
+  if (dest_size == 0)
+    return;
+
+  if (src_len >= dest_size)
+    src_len = dest_size - 1;
+
+  memcpy(dest, src, src_len);
+  dest[src_len] = '\0';
+}
+
+static void get_client_key(char *client_key, size_t client_key_size) {
+  const char *header = strstr(current_request_buffer, "X-Forwarded-For:");
+  if (!header)
+    return;
+
+  header += strlen("X-Forwarded-For:");
+  const char *line_end = strpbrk(header, "\r\n");
+  size_t line_len = line_end ? (size_t)(line_end - header) : strlen(header);
+  const char *comma = memchr(header, ',', line_len);
+  size_t value_len = comma ? (size_t)(comma - header) : line_len;
+
+  trim_copy(client_key, client_key_size, header, value_len);
+}
+
+static void get_client_key_from_socket(char *client_key,
+                                       size_t client_key_size) {
+  struct sockaddr_storage addr;
+  socklen_t addr_len = sizeof(addr);
+
+  if (getpeername(current_client_socket, (struct sockaddr *)&addr, &addr_len) !=
+      0) {
+    return;
+  }
+
+  if (addr.ss_family == AF_INET) {
+    struct sockaddr_in *ipv4 = (struct sockaddr_in *)&addr;
+    inet_ntop(AF_INET, &ipv4->sin_addr, client_key, client_key_size);
+  } else if (addr.ss_family == AF_INET6) {
+    struct sockaddr_in6 *ipv6 = (struct sockaddr_in6 *)&addr;
+    inet_ntop(AF_INET6, &ipv6->sin6_addr, client_key, client_key_size);
+  } else if (addr.ss_family == AF_UNIX) {
+    snprintf(client_key, client_key_size, "unix:%d", current_client_socket);
+  }
+}
+
+void rate_limit_get_client_key(char *client_key, size_t client_key_size) {
+  if (!client_key || client_key_size == 0)
+    return;
+
+  client_key[0] = '\0';
+  get_client_key(client_key, client_key_size);
+
+  if (client_key[0] == '\0') {
+    get_client_key_from_socket(client_key, client_key_size);
+  }
+
+  if (client_key[0] == '\0') {
+    snprintf(client_key, client_key_size, "nun");
+  }
+}
+
+static void prune_stale_entries(time_t now) {
+  RateLimitEntry **cursor = &rate_limit_entries;
+
+  while (*cursor) {
+    RateLimitEntry *entry = *cursor;
+    if (now - entry->last_seen > 9999) {
+      *cursor = entry->next;
+      free(entry);
+      continue;
+    }
+    cursor = &entry->next;
+  }
+}
+
+static RateLimitEntry *find_entry(const char *client_key, const char *scope) {
+  for (RateLimitEntry *entry = rate_limit_entries; entry; entry = entry->next) {
+    if (strcmp(entry->client_key, client_key) == 0 &&
+        strcmp(entry->scope, scope) == 0) {
+      return entry;
+    }
+  }
+  return NULL;
+}
+
+static RateLimitEntry *create_entry(const char *client_key, const char *scope,
+                                    time_t now) {
+  RateLimitEntry *entry = (RateLimitEntry *)calloc(1, sizeof(RateLimitEntry));
+  if (!entry)
+    return NULL;
+
+  snprintf(entry->client_key, sizeof(entry->client_key), "%s", client_key);
+  snprintf(entry->scope, sizeof(entry->scope), "%s", scope);
+  entry->window_start = now;
+  entry->last_seen = now;
+  entry->next = rate_limit_entries;
+  rate_limit_entries = entry;
+  return entry;
+}
+
+RateLimitResult rate_limit_check(const char *scope,
+                                 const RateLimitConfig *config) {
+  RateLimitResult result = {.limited = 0, .retry_after_seconds = 0};
+
+  if (!scope || !config || config->max_requests <= 0 ||
+      config->interval_seconds <= 0) {
+    return result;
+  }
+
+  char client_key[64];
+  time_t now = time(NULL);
+
+  rate_limit_get_client_key(client_key, sizeof(client_key));
+
+  pthread_mutex_lock(&rate_limit_mutex);
+
+  prune_stale_entries(now);
+
+  RateLimitEntry *entry = find_entry(client_key, scope);
+  if (!entry) {
+    entry = create_entry(client_key, scope, now);
+    if (!entry) {
+      pthread_mutex_unlock(&rate_limit_mutex);
+      return result;
+    }
+  }
+
+  entry->last_seen = now;
+
+  if (now - entry->window_start >= config->interval_seconds) {
+    entry->window_start = now;
+    entry->count = 0;
+  }
+
+  if (entry->count >= config->max_requests) {
+    result.limited = 1;
+    result.retry_after_seconds =
+        config->interval_seconds - (int)(now - entry->window_start);
+    if (result.retry_after_seconds < 1) {
+      result.retry_after_seconds = 1;
+    }
+    pthread_mutex_unlock(&rate_limit_mutex);
+    return result;
+  }
+
+  entry->count++;
+  pthread_mutex_unlock(&rate_limit_mutex);
+  return result;
+}
diff --git a/src/Limiter/RateLimit.h b/src/Limiter/RateLimit.h
new file mode 100644
index 0000000..fabd05d
--- /dev/null
+++ b/src/Limiter/RateLimit.h
@@ -0,0 +1,20 @@
+#ifndef RATE_LIMIT_H
+#define RATE_LIMIT_H
+
+#include <stddef.h>
+
+typedef struct {
+  int max_requests;
+  int interval_seconds;
+} RateLimitConfig;
+
+typedef struct {
+  int limited;
+  int retry_after_seconds;
+} RateLimitResult;
+
+void rate_limit_get_client_key(char *client_key, size_t client_key_size);
+RateLimitResult rate_limit_check(const char *scope,
+                                 const RateLimitConfig *config);
+
+#endif
diff --git a/src/Main.c b/src/Main.c
index 326b5ae..c3a607a 100644
--- a/src/Main.c
+++ b/src/Main.c
@@ -55,7 +55,11 @@ int main() {
                 .cache_ttl_search = DEFAULT_CACHE_TTL_SEARCH,
                 .cache_ttl_infobox = DEFAULT_CACHE_TTL_INFOBOX,
                 .cache_ttl_image = DEFAULT_CACHE_TTL_IMAGE,
-                .engines = ""};
+                .engines = "",
+                .rate_limit_search_requests = 0,
+                .rate_limit_search_interval = 0,
+                .rate_limit_images_requests = 0,
+                .rate_limit_images_interval = 0};
 
   if (load_config("config.ini", &cfg) != 0) {
     fprintf(stderr, "[WARN] Could not load config file, using defaults\n");
diff --git a/src/Routes/Images.c b/src/Routes/Images.c
index 03eb280..98fd3f4 100644
--- a/src/Routes/Images.c
+++ b/src/Routes/Images.c
@@ -1,10 +1,21 @@
 #include "Images.h"
+#include "../Cache/Cache.h"
+#include "../Limiter/RateLimit.h"
 #include "../Scraping/ImageScraping.h"
 #include "../Utility/Unescape.h"
 #include "../Utility/Utility.h"
 #include "Config.h"
 
+static char *build_images_request_cache_key(const char *query, int page,
+                                            const char *client_key) {
+  char scope_key[BUFFER_SIZE_MEDIUM];
+  snprintf(scope_key, sizeof(scope_key), "images_request:%s",
+           client_key ? client_key : "unknown");
+  return cache_compute_key(query, page, scope_key);
+}
+
 int images_handler(UrlParams *params) {
+  extern Config global_config;
   TemplateContext ctx = new_context();
   char *raw_query = "";
   int page = 1;
@@ -52,12 +63,55 @@ int images_handler(UrlParams *params) {
     return -1;
   }
 
+  char client_key[BUFFER_SIZE_SMALL];
+  rate_limit_get_client_key(client_key, sizeof(client_key));
+
+  char *request_cache_key =
+      build_images_request_cache_key(raw_query, page, client_key);
+  int request_is_cached = 0;
+
+  if (request_cache_key && get_cache_ttl_image() > 0) {
+    char *cached_marker = NULL;
+    size_t cached_marker_size = 0;
+
+    if (cache_get(request_cache_key, (time_t)get_cache_ttl_image(),
+                  &cached_marker, &cached_marker_size) == 0) {
+      request_is_cached = 1;
+    }
+
+    free(cached_marker);
+  }
+
+  if (!request_is_cached) {
+    RateLimitConfig rate_limit_config = {
+        .max_requests = global_config.rate_limit_images_requests,
+        .interval_seconds = global_config.rate_limit_images_interval,
+    };
+    RateLimitResult rate_limit_result =
+        rate_limit_check("images", &rate_limit_config);
+    if (rate_limit_result.limited) {
+      char response[256];
+      snprintf(response, sizeof(response),
+               "<h1>Slow down!</h1><p>Too many image searches from you!</p>");
+      send_response(response);
+      free(request_cache_key);
+      free(display_query);
+      free_context(&ctx);
+      return -1;
+    }
+
+    if (request_cache_key && get_cache_ttl_image() > 0) {
+      cache_set(request_cache_key, "1", 1);
+    }
+  }
+
   ImageResult *results = NULL;
   int result_count = 0;
 
   if (scrape_images(raw_query, page, &results, &result_count) != 0 ||
       !results) {
     send_response("<h1>Error fetching images</h1>");
+    free(request_cache_key);
     free(display_query);
     free_context(&ctx);
     return -1;
@@ -72,6 +126,7 @@ int images_handler(UrlParams *params) {
     if (inner_counts)
       free(inner_counts);
     free_image_results(results, result_count);
+    free(request_cache_key);
     free(display_query);
     free_context(&ctx);
     return -1;
@@ -106,6 +161,7 @@ int images_handler(UrlParams *params) {
   free(inner_counts);
 
   free_image_results(results, result_count);
+  free(request_cache_key);
   free(display_query);
   free_context(&ctx);
 
diff --git a/src/Routes/Search.c b/src/Routes/Search.c
index 81e43d4..170b488 100644
--- a/src/Routes/Search.c
+++ b/src/Routes/Search.c
@@ -1,9 +1,11 @@
 #include "Search.h"
+#include "../Cache/Cache.h"
 #include "../Infobox/Calculator.h"
 #include "../Infobox/CurrencyConversion.h"
 #include "../Infobox/Dictionary.h"
 #include "../Infobox/UnitConversion.h"
 #include "../Infobox/Wikipedia.h"
+#include "../Limiter/RateLimit.h"
 #include "../Scraping/Scraping.h"
 #include "../Utility/Display.h"
 #include "../Utility/Unescape.h"
@@ -378,7 +380,17 @@ static char *build_search_href(const char *query, const char *engine_id,
   return href;
 }
 
+static char *build_search_request_cache_key(const char *query,
+                                            const char *engine_id, int page,
+                                            const char *client_key) {
+  char scope_key[BUFFER_SIZE_MEDIUM];
+  snprintf(scope_key, sizeof(scope_key), "search_request:%s:%s",
+           engine_id ? engine_id : "all", client_key ? client_key : "unknown");
+  return cache_compute_key(query, page, scope_key);
+}
+
 int results_handler(UrlParams *params) {
+  extern Config global_config;
   TemplateContext ctx = new_context();
   char *raw_query = "";
   const char *selected_engine_id = "all";
@@ -474,6 +486,47 @@ int results_handler(UrlParams *params) {
     }
   }
 
+  char client_key[BUFFER_SIZE_SMALL];
+  rate_limit_get_client_key(client_key, sizeof(client_key));
+
+  char *request_cache_key = build_search_request_cache_key(
+      raw_query, selected_engine_id, page, client_key);
+  int request_is_cached = 0;
+
+  if (request_cache_key && get_cache_ttl_search() > 0) {
+    char *cached_marker = NULL;
+    size_t cached_marker_size = 0;
+
+    if (cache_get(request_cache_key, (time_t)get_cache_ttl_search(),
+                  &cached_marker, &cached_marker_size) == 0) {
+      request_is_cached = 1;
+    }
+
+    free(cached_marker);
+  }
+
+  if (engine_idx > 0 && !request_is_cached) {
+    RateLimitConfig rate_limit_config = {
+        .max_requests = global_config.rate_limit_search_requests,
+        .interval_seconds = global_config.rate_limit_search_interval,
+    };
+    RateLimitResult rate_limit_result =
+        rate_limit_check("search", &rate_limit_config);
+    if (rate_limit_result.limited) {
+      char response[256];
+      snprintf(response, sizeof(response),
+               "<h1>Slow down!</h1><p>Too many searches from you!</p>");
+      send_response(response);
+      free(request_cache_key);
+      free_context(&ctx);
+      return -1;
+    }
+
+    if (request_cache_key && get_cache_ttl_search() > 0) {
+      cache_set(request_cache_key, "1", 1);
+    }
+  }
+
   int filter_engine_count = 0;
   for (int i = 0; i < ENGINE_COUNT; i++) {
     if (ENGINE_REGISTRY[i].enabled)
@@ -551,6 +604,7 @@ int results_handler(UrlParams *params) {
             }
           }
         }
+        free(request_cache_key);
         free_context(&ctx);
         if (redirect_url) {
           send_redirect(redirect_url);
@@ -569,6 +623,7 @@ int results_handler(UrlParams *params) {
         }
       }
     }
+    free(request_cache_key);
     free_context(&ctx);
     send_response("<h1>No results found</h1>");
     return 0;
@@ -668,6 +723,7 @@ int results_handler(UrlParams *params) {
           }
         }
       }
+      free(request_cache_key);
       free_context(&ctx);
       return 0;
     }
@@ -817,6 +873,8 @@ int results_handler(UrlParams *params) {
     }
   }
 
+  free(request_cache_key);
+
   if (page == 1) {
     for (int i = 0; i < HANDLER_COUNT; i++) {
       if (infobox_data[i].success) {