Remove dependency on tidy, fix ph, support all 4chan markup

Go back to previous page when failing to fetch number of pages

1 files changed, 40 insertions, 33 deletions

diff --git a/src/plugins/NyaaSi.cpp b/src/plugins/NyaaSi.cpp
index 5d9e41b..d4667af 100644
--- a/src/plugins/NyaaSi.cpp
+++ b/src/plugins/NyaaSi.cpp
@@ -291,17 +291,18 @@ namespace QuickMedia {
             return PluginResult::NET_ERR;
 
         QuickMediaHtmlSearch html_search;
-        int result = quickmedia_html_search_init(&html_search, website_data.c_str());
+        int result = quickmedia_html_search_init(&html_search, website_data.c_str(), website_data.size());
         if(result != 0)
             goto cleanup;
 
         result = quickmedia_html_find_nodes_xpath(&html_search, "//h3[class='panel-title']",
-            [](QuickMediaHtmlNode *node, void *userdata) {
+            [](QuickMediaMatchNode *node, void *userdata) {
                 std::string *title = (std::string*)userdata;
-                const char *text = quickmedia_html_node_get_text(node);
-                if(title->empty() && text) {
-                    *title = text;
+                QuickMediaStringView text = quickmedia_html_node_get_text(node);
+                if(title->empty() && text.data) {
+                    title->assign(text.data, text.size);
                 }
+                return 0;
             }, &title);
 
         if(result != 0)
@@ -314,16 +315,17 @@ namespace QuickMedia {
         }
 
         result = quickmedia_html_find_nodes_xpath(&html_search, "//div[class='panel-body']//div[class='row']//a",
-            [](QuickMediaHtmlNode *node, void *userdata) {
+            [](QuickMediaMatchNode *node, void *userdata) {
                 ResultItemExtra *item_data = (ResultItemExtra*)userdata;
-                const char *href = quickmedia_html_node_get_attribute_value(node, "href");
-                const char *text = quickmedia_html_node_get_text(node);
-                if(item_data->result_items->empty() && href && text && strncmp(href, "/user/", 6) == 0) {
+                QuickMediaStringView href = quickmedia_html_node_get_attribute_value(node, "href");
+                QuickMediaStringView text = quickmedia_html_node_get_text(node);
+                if(item_data->result_items->empty() && href.data && text.data && href.size >= 6 && memcmp(href.data, "/user/", 6) == 0) {
                     auto body_item = BodyItem::create("");
-                    body_item->set_description("Submitter: " + strip(text));
-                    body_item->url = "https://" + *item_data->domain + "/" + std::string(href);
+                    body_item->set_description("Submitter: " + std::string(text.data, text.size));
+                    body_item->url = "https://" + *item_data->domain + "/" + std::string(href.data, href.size);
                     item_data->result_items->push_back(std::move(body_item));
                 }
+                return 0;
             }, &result_item_extra);
 
         if(result != 0)
@@ -335,17 +337,18 @@ namespace QuickMedia {
             result_items.push_back(std::move(body_item));
         }
 
-        result_items.front()->set_title(strip(title));
+        result_items.front()->set_title(title);
 
         result = quickmedia_html_find_nodes_xpath(&html_search, "//div[id='torrent-description']",
-            [](QuickMediaHtmlNode *node, void *userdata) {
+            [](QuickMediaMatchNode *node, void *userdata) {
                 std::string *description = (std::string*)userdata;
-                const char *text = quickmedia_html_node_get_text(node);
-                if(description->empty() && text) {
-                    std::string desc = strip(text);
+                QuickMediaStringView text = quickmedia_html_node_get_text(node);
+                if(description->empty() && text.data) {
+                    std::string desc(text.data, text.size);
                     html_unescape_sequences(desc);
                     *description = std::move(desc);
                 }
+                return 0;
             }, &description);
 
         if(result != 0)
@@ -355,12 +358,13 @@ namespace QuickMedia {
             result_items.front()->set_description(result_items.front()->get_description() + "\nDescription:\n" + description);
 
         result = quickmedia_html_find_nodes_xpath(&html_search, "//div[class='container']//a",
-            [](QuickMediaHtmlNode *node, void *userdata) {
+            [](QuickMediaMatchNode *node, void *userdata) {
                 std::string *magnet_url = (std::string*)userdata;
-                const char *href = quickmedia_html_node_get_attribute_value(node, "href");
-                if(magnet_url->empty() && href && strncmp(href, "magnet:?", 8) == 0) {
-                    *magnet_url = href;
+                QuickMediaStringView href = quickmedia_html_node_get_attribute_value(node, "href");
+                if(magnet_url->empty() && href.data && href.size >= 8 && memcmp(href.data, "magnet:?", 8) == 0) {
+                    magnet_url->assign(href.data, href.size);
                 }
+                return 0;
             }, &magnet_url);
 
         if(result != 0)
@@ -377,15 +381,16 @@ namespace QuickMedia {
         comments_start_index = result_items.size();
 
         result = quickmedia_html_find_nodes_xpath(&html_search, "//div[id='comments']//a",
-            [](QuickMediaHtmlNode *node, void *userdata) {
+            [](QuickMediaMatchNode *node, void *userdata) {
                 auto *item_data = (BodyItems*)userdata;
-                const char *href = quickmedia_html_node_get_attribute_value(node, "href");
-                const char *text = quickmedia_html_node_get_text(node);
-                if(href && text && strncmp(href, "/user/", 6) == 0) {
-                    auto body_item = BodyItem::create(strip(text));
+                QuickMediaStringView href = quickmedia_html_node_get_attribute_value(node, "href");
+                QuickMediaStringView text = quickmedia_html_node_get_text(node);
+                if(href.data && text.data && href.size >= 6 && memcmp(href.data, "/user/", 6) == 0) {
+                    auto body_item = BodyItem::create(std::string(text.data, text.size));
                     //body_item->url = "https://nyaa.si/" + std::string(href);
                     item_data->push_back(std::move(body_item));
                 }
+                return 0;
             }, &result_items);
 
         if(result != 0 || result_items.size() == comments_start_index)
@@ -396,14 +401,15 @@ namespace QuickMedia {
         body_item_image_context.index = comments_start_index;
 
         result = quickmedia_html_find_nodes_xpath(&html_search, "//div[id='comments']//img[class='avatar']",
-            [](QuickMediaHtmlNode *node, void *userdata) {
+            [](QuickMediaMatchNode *node, void *userdata) {
                 auto *item_data = (BodyItemContext*)userdata;
-                const char *src = quickmedia_html_node_get_attribute_value(node, "src");
-                if(src && item_data->index < item_data->body_items->size()) {
-                    (*item_data->body_items)[item_data->index]->thumbnail_url = src;
+                QuickMediaStringView src = quickmedia_html_node_get_attribute_value(node, "src");
+                if(src.data && item_data->index < item_data->body_items->size()) {
+                    (*item_data->body_items)[item_data->index]->thumbnail_url.assign(src.data, src.size);
                     (*item_data->body_items)[item_data->index]->thumbnail_size = sf::Vector2i(120, 120);
                     item_data->index++;
                 }
+                return 0;
             }, &body_item_image_context);
 
         if(result != 0)
@@ -412,15 +418,16 @@ namespace QuickMedia {
         body_item_image_context.index = comments_start_index;
 
         result = quickmedia_html_find_nodes_xpath(&html_search, "//div[id='comments']//div[class='comment-content']",
-            [](QuickMediaHtmlNode *node, void *userdata) {
+            [](QuickMediaMatchNode *node, void *userdata) {
                 auto *item_data = (BodyItemContext*)userdata;
-                const char *text = quickmedia_html_node_get_text(node);
-                if(text && item_data->index < item_data->body_items->size()) {
-                    std::string desc = strip(text);
+                QuickMediaStringView text = quickmedia_html_node_get_text(node);
+                if(text.data && item_data->index < item_data->body_items->size()) {
+                    std::string desc(text.data, text.size);
                     html_unescape_sequences(desc);
                     (*item_data->body_items)[item_data->index]->set_description(std::move(desc));
                     item_data->index++;
                 }
+                return 0;
             }, &body_item_image_context);
 
         cleanup: