Rewrite pixiv URLs to Pixwhile

blankie/murl.cpp

@@ -1,5 +1,6 @@
 #include <regex>
 #include <vector>
+#include <ctype.h>
 #include <climits>
 
 #include "murl.h"
@@ -14,7 +15,7 @@
 // A lot looser than RFC 3986, but things will go very wrong very quickly if I comply
 // Hostname does not allow for stuff like "//The", but it is not important here
 #define HOSTCHAR "[\\w\\d\\-.]"
-#define HOST "(\\[[\\da-f:.]+\\]|" HOSTCHAR "+(?:\\." HOSTCHAR "+)+)"
+#define HOST "(\\[[\\da-f:.]+\\]|localhost|" HOSTCHAR "+(?:\\." HOSTCHAR "+)+)"
 #define PORT "(\\d*)"
 #define AUTHORITY "(?:" USERINFO "@)?" HOST "(?::" PORT ")?"
 
@@ -38,6 +39,7 @@
     "(?:#" FRAGMENT ")?"
 
 static void handle_segment(std::vector<std::string>& segments, const std::string& str, size_t offset, size_t length);
+static std::string tolower(std::string str);
 
 namespace blankie {
 namespace murl {
@@ -87,6 +89,10 @@ std::string Url::to_string() const {
     return str;
 }
 
+bool Url::is_host_equal(std::string other) const {
+    return tolower(this->hostname) == tolower(std::move(other));
+}
+
 std::string normalize_path(const std::string& str) {
     std::vector<std::string> segments;
     std::string res;
@@ -143,3 +149,10 @@ static void handle_segment(std::vector<std::string>& segments, const std::string
         segments.push_back(str.substr(offset, length));
     }
 }
+
+static std::string tolower(std::string str) {
+    for (size_t i = 0; i < str.size(); i++) {
+        str[i] = static_cast<char>(tolower(str[i]));
+    }
+    return str;
+}

blankie/murl.h

@@ -32,6 +32,9 @@ struct Url {
         return res;
     }
 
+    // NOT SECURE!
+    bool is_host_equal(std::string other) const;
+
     std::string to_string() const;
 };
 

routes/artworks.cpp

@@ -9,8 +9,8 @@
 static inline Element generate_user_link(const httplib::Request& req, const Config& config, const Illust& illust);
 static inline Element generate_images(const httplib::Request& req, const Config& config, const Illust& illust);
 static inline Element generate_preview_images(const httplib::Request& req, const Config& config, const Illust& illust);
-static inline std::vector<blankie::html::Node> parse_description_line(std::string str);
+static inline std::vector<blankie::html::Node> parse_description_line(const httplib::Request& req, const Config& config, std::string str);
-static inline Element generate_description(const std::string& description);
+static inline Element generate_description(const httplib::Request& req, const Config& config, const std::string& description);
 static inline Element generate_illust_tags(const Illust& illust);
 
 static inline bool is_true(const std::string& str);
@@ -45,7 +45,7 @@ void artworks_route(const httplib::Request& req, httplib::Response& res, const C
         Element("br")
     });
     if (illust.comment) {
-        body.nodes.push_back(generate_description(*illust.comment));
+        body.nodes.push_back(generate_description(req, config, *illust.comment));
     }
     body.nodes.push_back(generate_illust_tags(illust));
     body.nodes.push_back(Element("p", {time_to_string(illust.upload_time)}));
@@ -112,8 +112,7 @@ static inline Element generate_preview_images(const httplib::Request& req, const
     return div;
 }
 
-// TODO proxy urls
+static inline std::vector<blankie::html::Node> parse_description_line(const httplib::Request& req, const Config& config, std::string str) {
-static inline std::vector<blankie::html::Node> parse_description_line(std::string str) {
     std::vector<blankie::html::Node> nodes;
     std::smatch sm;
 
@@ -121,7 +120,13 @@ static inline std::vector<blankie::html::Node> parse_description_line(std::strin
         if (sm.prefix().length()) {
             nodes.push_back(sm.prefix());
         }
-        nodes.push_back(Element("a", {{"href", sm.str(0)}}, {sm.str(0)}));
+
+        blankie::murl::Url url(sm.str(0));
+        std::string url_str = url.is_host_equal("pixiv.net") || url.is_host_equal("www.pixiv.net")
+            ? proxy_url(get_origin(req, config), std::move(url))
+            : url.to_string();
+        nodes.push_back(Element("a", {{"href", url_str}}, {url_str}));
+
         str = sm.suffix();
     }
     if (!str.empty()) {
@@ -131,7 +136,7 @@ static inline std::vector<blankie::html::Node> parse_description_line(std::strin
     return nodes;
 }
 
-static inline Element generate_description(const std::string& description) {
+static inline Element generate_description(const httplib::Request& req, const Config& config, const std::string& description) {
     Element p("p");
     size_t pos = 0;
     size_t last_pos = 0;
@@ -139,7 +144,7 @@ static inline Element generate_description(const std::string& description) {
         if (!p.nodes.empty()) {
             p.nodes.push_back(Element("br"));
         }
-        std::vector<blankie::html::Node> nodes = parse_description_line(std::move(str));
+        std::vector<blankie::html::Node> nodes = parse_description_line(req, config, std::move(str));
         p.nodes.insert(p.nodes.end(), nodes.begin(), nodes.end());
     };
 

servehelper.cpp

@@ -72,20 +72,22 @@ std::string get_origin(const httplib::Request& req, const Config& config) {
     return origin;
 }
 
+std::string proxy_url(blankie::murl::Url base, blankie::murl::Url url) {
+    if (!url.path.empty() && url.path[0] != '/') {
+        base.path += '/';
+    }
+    base.path += blankie::murl::normalize_path(std::move(url.path));
+    if (!base.query.empty() && !url.query.empty()) {
+        base.query += '&';
+    }
+    base.query += std::move(url.query);
+    base.fragment = std::move(url.fragment);
+    return base.to_string();
+}
+
 std::string proxy_image_url(const Config& config, blankie::murl::Url url) {
-    if (url.hostname == "s.pximg.net") {
+    if (url.is_host_equal("s.pximg.net")) {
         return url.to_string();
     }
-
+    return proxy_url(config.image_proxy_url, std::move(url));
-    blankie::murl::Url new_url = config.image_proxy_url;
-    if (!url.path.empty() && url.path[0] != '/') {
-        new_url.path += '/';
-    }
-    new_url.path += blankie::murl::normalize_path(std::move(url.path));
-    if (!new_url.query.empty() && !url.query.empty()) {
-        new_url.query += '&';
-        new_url.query += std::move(url.query);
-    }
-    new_url.fragment = std::move(url.fragment);
-    return new_url.to_string();
 }

servehelper.h

@@ -14,4 +14,5 @@ void serve_error(const httplib::Request& req, httplib::Response& res, const Conf
         std::string title, std::optional<std::string> subtitle = std::nullopt, std::optional<std::string> info = std::nullopt);
 void serve_redirect(const httplib::Request& req, httplib::Response& res, const Config& config, std::string url);
 std::string get_origin(const httplib::Request& req, const Config& config);
+std::string proxy_url(blankie::murl::Url base, blankie::murl::Url url);
 std::string proxy_image_url(const Config& config, blankie::murl::Url url);