Add xvideos and xhamster

3 files changed, 152 insertions, 39 deletions

diff --git a/src/QuickMedia.cpp b/src/QuickMedia.cpp
index fd679ff..7402b97 100644
--- a/src/QuickMedia.cpp
+++ b/src/QuickMedia.cpp
@@ -67,6 +67,8 @@ static const std::pair<const char*, const char*> valid_plugins[] = {
     std::make_pair("soundcloud", "soundcloud_logo.png"),
     std::make_pair("pornhub", "pornhub_logo.png"),
     std::make_pair("spankbang", "spankbang_logo.png"),
+    std::make_pair("xvideos", "xvideos_logo.png"),
+    std::make_pair("xhamster", "xhamster_logo.png"),
     std::make_pair("4chan", "4chan_logo.png"),
     std::make_pair("nyaa.si", "nyaa_si_logo.png"),
     std::make_pair("matrix", "matrix_logo.png"),
@@ -444,7 +446,7 @@ namespace QuickMedia {
     static void usage() {
         fprintf(stderr, "usage: quickmedia <plugin> [--no-video] [--use-system-mpv-config] [--dir <directory>] [-e <window>]\n");
         fprintf(stderr, "OPTIONS:\n");
-        fprintf(stderr, "  plugin                      The plugin to use. Should be either launcher, 4chan, manga, manganelo, manganelos, mangatown, mangakatana, mangadex, pornhub, spankbang, youtube, spotify, soundcloud, nyaa.si, matrix, file-manager or stdin\n");
+        fprintf(stderr, "  plugin                      The plugin to use. Should be either launcher, 4chan, manga, manganelo, manganelos, mangatown, mangakatana, mangadex, pornhub, spankbang, xvideos, xhamster, youtube, spotify, soundcloud, nyaa.si, matrix, file-manager or stdin\n");
         fprintf(stderr, "  --no-video                  Only play audio when playing a video. Disabled by default\n");
         fprintf(stderr, "  --use-system-mpv-config     Use system mpv config instead of no config. Disabled by default\n");
         fprintf(stderr, "  --upscale-images            Upscale low-resolution manga pages using waifu2x-ncnn-vulkan. Disabled by default\n");
@@ -804,6 +806,90 @@ namespace QuickMedia {
             .related_media_thumbnail_handler({{"//div[class='right']//div[class='video-item']//img", "data-src", nullptr}});
     }
 
+    // Returns size_t(-1) if not found
+    static size_t find_end_of_json_array(const char *str, size_t start, size_t size) {
+        if(size <= start || str[start] != '[')
+            return size_t(-1);
+
+        bool inside_string = false;
+        bool escape = false;
+        int array_depth = 0;
+        for(size_t i = start; i < size; ++i) {
+            char c = str[i];
+            if(c == '"' && !escape) {
+                inside_string = !inside_string;
+            } else if(c == '\\') {
+                escape = !escape;
+            } else if(c == '[' && !inside_string && !escape) {
+                ++array_depth;
+            } else if(c == ']' && !inside_string && !escape) {
+                --array_depth;
+                if(array_depth == 0)
+                    return i + 1;
+            } else {
+                escape = false;
+            }
+        }
+
+        return size_t(-1);
+    }
+
+    static void add_xvideos_handlers(MediaGenericSearchPage *media_generic_search_page) {
+        media_generic_search_page->search_handler("https://www.xvideos.com/?k=%s&p=%p", 0)
+            .text_handler({{"//div[id='content']//div[class='thumb-under']//a", "title", "href", "/video"}})
+            .thumbnail_handler({{"//div[id='content']//div[class='thumb']//img", "data-src", "/videos/"}})
+            .related_media_custom_handler([](const std::string &html_source) {
+                std::vector<MediaRelatedItem> related_items;
+                size_t related_start = html_source.find("video_related=[");
+                if(related_start == std::string::npos)
+                    return related_items;
+
+                related_start += 14; // just before [
+                size_t json_end = find_end_of_json_array(html_source.c_str(), related_start, html_source.size());
+                if(json_end == size_t(-1))
+                    return related_items;
+
+                Json::Value json_root;
+                Json::CharReaderBuilder json_builder;
+                std::unique_ptr<Json::CharReader> json_reader(json_builder.newCharReader());
+                std::string json_errors;
+                if(!json_reader->parse(html_source.c_str() + related_start, html_source.c_str() + json_end, &json_root, &json_errors)) {
+                    fprintf(stderr, "Failed to parse xvideos related json, error: %s\n", json_errors.c_str());
+                    return related_items;
+                }
+
+                if(!json_root.isArray())
+                    return related_items;
+
+                for(const Json::Value &json_item : json_root) {
+                    if(!json_item.isObject())
+                        continue;
+
+                    const Json::Value &title_json = json_item["tf"];
+                    const Json::Value &url_json = json_item["u"];
+                    const Json::Value &thumbnail_url_json = json_item["i"];
+                    if(!title_json.isString() || !url_json.isString() || !thumbnail_url_json.isString())
+                        continue;
+
+                    MediaRelatedItem related_item;
+                    related_item.title = title_json.asString();
+                    related_item.url = url_json.asString();
+                    related_item.thumbnail_url = thumbnail_url_json.asString();
+                    related_items.push_back(std::move(related_item));
+                }
+
+                return related_items;
+            });
+    }
+
+    static void add_xhamster_handlers(MediaGenericSearchPage *media_generic_search_page) {
+        media_generic_search_page->search_handler("https://xhamster.com/search/%s?page=%p", 1)
+            .text_handler({{"//div[class='video-thumb-info']//a", "text", "href", "/videos/"}})
+            .thumbnail_handler({{"//img", "src", "/thumb-"}})
+            .related_media_text_handler({{"//div[class='video-thumb-info']//a", "text", "href", "/videos/"}})
+            .related_media_thumbnail_handler({{"//img", "src", "/thumb-"}});
+    }
+
     void Program::load_plugin_by_name(std::vector<Tab> &tabs, const char *start_dir) {
         if(!plugin_name || plugin_name[0] == '\0')
             return;
@@ -935,13 +1021,21 @@ namespace QuickMedia {
             tabs.push_back(Tab{create_body(), std::move(recommended_page), std::move(recommended_search_bar)});
             tabs.push_back(Tab{create_body(), std::make_unique<YoutubeSubscriptionsPage>(this), create_search_bar("Search...", SEARCH_DELAY_FILTER)});
         } else if(strcmp(plugin_name, "pornhub") == 0) {
-            auto search_page = std::make_unique<MediaGenericSearchPage>(this, "https://www.pornhub.com/", sf::Vector2i(320, 180));
+            auto search_page = std::make_unique<MediaGenericSearchPage>(this, "https://www.pornhub.com/", sf::Vector2i(320/1.5f, 180/1.5f));
             add_pornhub_handlers(search_page.get());
             tabs.push_back(Tab{create_body(), std::move(search_page), create_search_bar("Search...", 500)});
         } else if(strcmp(plugin_name, "spankbang") == 0) {
-            auto search_page = std::make_unique<MediaGenericSearchPage>(this, "https://spankbang.com/", sf::Vector2i(500/2, 281/2));
+            auto search_page = std::make_unique<MediaGenericSearchPage>(this, "https://spankbang.com/", sf::Vector2i(500/2.5f, 281/2.5f));
             add_spankbang_handlers(search_page.get());
             tabs.push_back(Tab{create_body(), std::move(search_page), create_search_bar("Search...", 500)});
+        } else if(strcmp(plugin_name, "xvideos") == 0) {
+            auto search_page = std::make_unique<MediaGenericSearchPage>(this, "https://www.xvideos.com/", sf::Vector2i(352/1.5f, 198/1.5f));
+            add_xvideos_handlers(search_page.get());
+            tabs.push_back(Tab{create_body(), std::move(search_page), create_search_bar("Search...", 500)});
+        } else if(strcmp(plugin_name, "xhamster") == 0) {
+            auto search_page = std::make_unique<MediaGenericSearchPage>(this, "https://xhamster.com/", sf::Vector2i(240, 135));
+            add_xhamster_handlers(search_page.get());
+            tabs.push_back(Tab{create_body(), std::move(search_page), create_search_bar("Search...", 500)});
         } else if(strcmp(plugin_name, "spotify") == 0) {
             tabs.push_back(Tab{create_body(), std::make_unique<SpotifyPodcastSearchPage>(this), create_search_bar("Search...", 350)});
             no_video = true;
diff --git a/src/plugins/MangaGeneric.cpp b/src/plugins/MangaGeneric.cpp
index b3326fc..48533b8 100644
--- a/src/plugins/MangaGeneric.cpp
+++ b/src/plugins/MangaGeneric.cpp
@@ -49,6 +49,20 @@ namespace QuickMedia {
             return quickmedia_html_node_get_attribute_value(node, field_name);
     }
 
+    static void body_items_prepend_website_url(BodyItems &body_items, const std::string &website_url) {
+        for(auto &body_item : body_items) {
+            if(string_starts_with(body_item->url, "//"))
+                body_item->url = "https://" + body_item->url.substr(2);
+            else if(string_starts_with(body_item->url, "/"))
+                body_item->url = website_url + body_item->url.substr(1);
+
+            if(string_starts_with(body_item->thumbnail_url, "//"))
+                body_item->thumbnail_url = "https://" + body_item->thumbnail_url.substr(2);
+            else if(string_starts_with(body_item->thumbnail_url, "/"))
+                body_item->thumbnail_url = website_url + body_item->thumbnail_url.substr(1);
+        }
+    }
+
     static int html_append_search(QuickMediaHtmlSearch *html_search, const char *html_query, HtmlSearchUserdata *search_userdata) {
         return quickmedia_html_find_nodes_xpath(html_search, html_query,
             [](QuickMediaHtmlNode *node, void *userdata) {
@@ -239,17 +253,7 @@ namespace QuickMedia {
             }
         }
 
-        for(auto &body_item : result_items) {
-            if(string_starts_with(body_item->url, "//"))
-                body_item->url = "https://" + body_item->url.substr(2);
-            else if(string_starts_with(body_item->url, "/"))
-                body_item->url = website_url + body_item->url.substr(1);
-
-            if(string_starts_with(body_item->thumbnail_url, "//"))
-                body_item->thumbnail_url = "https://" + body_item->thumbnail_url.substr(2);
-            else if(string_starts_with(body_item->thumbnail_url, "/"))
-                body_item->thumbnail_url = website_url + body_item->thumbnail_url.substr(1);
-        }
+        body_items_prepend_website_url(result_items, website_url);
 
         cleanup:
         quickmedia_html_search_deinit(&html_search);
@@ -333,17 +337,7 @@ namespace QuickMedia {
             }
         }
 
-        for(auto &body_item : chapters_items) {
-            if(string_starts_with(body_item->url, "//"))
-                body_item->url = "https://" + body_item->url.substr(2);
-            else if(string_starts_with(body_item->url, "/"))
-                body_item->url = website_url + body_item->url.substr(1);
-
-            if(string_starts_with(body_item->thumbnail_url, "//"))
-                body_item->thumbnail_url = "https://" + body_item->thumbnail_url.substr(2);
-            else if(string_starts_with(body_item->thumbnail_url, "/"))
-                body_item->thumbnail_url = website_url + body_item->thumbnail_url.substr(1);
-        }
+        body_items_prepend_website_url(chapters_items, website_url);
 
         for(auto &it : creators) {
             if(string_starts_with(it.second, "//"))
diff --git a/src/plugins/MediaGeneric.cpp b/src/plugins/MediaGeneric.cpp
index c911e6b..19ad87c 100644
--- a/src/plugins/MediaGeneric.cpp
+++ b/src/plugins/MediaGeneric.cpp
@@ -18,7 +18,21 @@ namespace QuickMedia {
             return quickmedia_html_node_get_attribute_value(node, field_name);
     }
 
-    static PluginResult fetch_page_results(const std::string &url, const std::string &website_url, const std::vector<MediaTextQuery> &text_queries, const std::vector<MediaThumbnailQuery> &thumbnail_queries, BodyItems &result_items) {
+    static void body_items_prepend_website_url(BodyItems &body_items, const std::string &website_url) {
+        for(auto &body_item : body_items) {
+            if(string_starts_with(body_item->url, "//"))
+                body_item->url = "https://" + body_item->url.substr(2);
+            else if(string_starts_with(body_item->url, "/"))
+                body_item->url = website_url + body_item->url.substr(1);
+
+            if(string_starts_with(body_item->thumbnail_url, "//"))
+                body_item->thumbnail_url = "https://" + body_item->thumbnail_url.substr(2);
+            else if(string_starts_with(body_item->thumbnail_url, "/"))
+                body_item->thumbnail_url = website_url + body_item->thumbnail_url.substr(1);
+        }
+    }
+
+    static PluginResult fetch_page_results(const std::string &url, const std::string &website_url, const std::vector<MediaTextQuery> &text_queries, const std::vector<MediaThumbnailQuery> &thumbnail_queries, MediaRelatedCustomHandler *custom_handler, BodyItems &result_items) {
         std::vector<CommandArg> args;
         if(!website_url.empty())
             args.push_back({ "-H", "referer: " + website_url });
@@ -30,6 +44,18 @@ namespace QuickMedia {
         if(website_data.empty())
             return PluginResult::OK;
 
+        if(custom_handler && *custom_handler) {
+            std::vector<MediaRelatedItem> media_related_items = (*custom_handler)(website_data);
+            for(MediaRelatedItem &media_related_item : media_related_items) {
+                auto body_item = BodyItem::create(strip(media_related_item.title));
+                body_item->url = std::move(media_related_item.url);
+                body_item->thumbnail_url = std::move(media_related_item.thumbnail_url);
+                result_items.push_back(std::move(body_item));
+            }
+            body_items_prepend_website_url(result_items, website_url);
+            return PluginResult::OK;
+        }
+
         QuickMediaHtmlSearch html_search;
         int result = quickmedia_html_search_init(&html_search, website_data.c_str());
         if(result != 0)
@@ -73,17 +99,7 @@ namespace QuickMedia {
             }
         }
 
-        for(auto &body_item : result_items) {
-            if(string_starts_with(body_item->url, "//"))
-                body_item->url = "https://" + body_item->url.substr(2);
-            else if(string_starts_with(body_item->url, "/"))
-                body_item->url = website_url + body_item->url.substr(1);
-
-            if(string_starts_with(body_item->thumbnail_url, "//"))
-                body_item->thumbnail_url = "https://" + body_item->thumbnail_url.substr(2);
-            else if(string_starts_with(body_item->thumbnail_url, "/"))
-                body_item->thumbnail_url = website_url + body_item->thumbnail_url.substr(1);
-        }
+        body_items_prepend_website_url(result_items, website_url);
 
         cleanup:
         quickmedia_html_search_deinit(&html_search);
@@ -112,7 +128,7 @@ namespace QuickMedia {
         std::string url = search_query.search_template;
         string_replace_all(url, "%s", url_param_encode(str));
         string_replace_all(url, "%p", std::to_string(search_query.page_start + page));
-        return fetch_page_results(url, website_url, text_queries, thumbnail_queries, result_items);
+        return fetch_page_results(url, website_url, text_queries, thumbnail_queries, nullptr, result_items);
     }
 
     PluginResult MediaGenericSearchPage::submit(const std::string&, const std::string &url, std::vector<Tab> &result_tabs) {
@@ -121,7 +137,7 @@ namespace QuickMedia {
     }
 
     PluginResult MediaGenericSearchPage::get_related_media(const std::string &url, BodyItems &result_items) {
-        return fetch_page_results(url, website_url, related_media_text_queries, related_media_thumbnail_queries, result_items);
+        return fetch_page_results(url, website_url, related_media_text_queries, related_media_thumbnail_queries, &related_custom_handler, result_items);
     }
 
     MediaGenericSearchPage& MediaGenericSearchPage::search_handler(const char *search_template, int page_start) {
@@ -142,11 +158,20 @@ namespace QuickMedia {
 
     MediaGenericSearchPage& MediaGenericSearchPage::related_media_text_handler(std::vector<MediaTextQuery> queries) {
         related_media_text_queries = std::move(queries);
+        related_custom_handler = nullptr;
         return *this;
     }
 
     MediaGenericSearchPage& MediaGenericSearchPage::related_media_thumbnail_handler(std::vector<MediaThumbnailQuery> queries) {
         related_media_thumbnail_queries = std::move(queries);
+        related_custom_handler = nullptr;
+        return *this;
+    }
+
+    MediaGenericSearchPage& MediaGenericSearchPage::related_media_custom_handler(MediaRelatedCustomHandler handler) {
+        related_custom_handler = handler;
+        related_media_text_queries.clear();
+        related_media_thumbnail_queries.clear();
         return *this;
     }