Fix HTML in the description not being parsed

https://pixwhile.zangetsu.kaizoku.cyou/artworks/87524102

CMakeLists.txt

@@ -27,7 +27,7 @@ add_link_options(${FLAGS})
 
 add_executable(${PROJECT_NAME} main.cpp misc.cpp config.cpp servehelper.cpp numberhelper.cpp pixivclient.cpp pixivmodels.cpp hiredis_wrapper.cpp
     routes/home.cpp routes/css.cpp routes/artworks.cpp routes/tags.cpp routes/guess_extension.cpp routes/users/common.cpp routes/users/illustrations.cpp
-    blankie/serializer.cpp blankie/escape.cpp blankie/murl.cpp)
+    blankie/serializer.cpp blankie/escape.cpp blankie/unescape.cpp blankie/murl.cpp)
 set_target_properties(${PROJECT_NAME}
     PROPERTIES
         CXX_STANDARD 20

blankie/murl.cpp

@@ -26,11 +26,6 @@
 // Looser than RFC 3986, but fragments might as well own everything
 #define FRAGMENT "([^ ]*)"
 
-#define FULL_HTTP_HTTPS_REGEX \
-    "(?:(https?)?:)?//" AUTHORITY \
-    PATH_ABEMPTY \
-    "(?:\\?" QUERY ")?" \
-    "(?:#" FRAGMENT ")?"
 #define HTTP_HTTPS_REGEX \
     "(?:(https?)?:)?(?://" AUTHORITY ")?" \
     PATH_ABEMPTY \
@@ -45,8 +40,7 @@ static std::string tolower(std::string str);
 namespace blankie {
 namespace murl {
 
-std::regex full_url_regex(FULL_HTTP_HTTPS_REGEX, std::regex::icase);
-std::regex url_regex(HTTP_HTTPS_REGEX, std::regex::icase);
+const std::regex url_regex(HTTP_HTTPS_REGEX, std::regex::icase);
 
 Url::Url(const std::string& str) {
     std::smatch sm;

blankie/murl.h

@@ -6,8 +6,6 @@
 namespace blankie {
 namespace murl {
 
-extern std::regex full_url_regex;
-
 struct Url {
     std::string scheme;
     std::string userinfo;

blankie/serializer.cpp

@@ -36,6 +36,8 @@ std::string Element::serialize() const {
             out += escape(*text);
         } else if (const std::string* str = std::get_if<std::string>(&node)) {
             out += escape(*str);
+        } else if (const HTMLString* html_str = std::get_if<HTMLString>(&node)) {
+            out += html_str->str;
         } else {
             throw std::runtime_error("Encountered unknown node");
         }

blankie/serializer.h

@@ -9,9 +9,15 @@ namespace blankie {
 namespace html {
 
 struct Element;
+struct HTMLString {
+    HTMLString() = default;
+    explicit HTMLString(std::string str_) : str(std::move(str_)) {}
+
+    std::string str;
+};
 
 typedef std::pair<const char*, std::string> Attribute;
-typedef std::variant<Element, const char*, std::string> Node;
+typedef std::variant<Element, const char*, std::string, HTMLString> Node;
 
 struct Element {
     const char* tag;

blankie/unescape.cpp

@@ -0,0 +1,125 @@
+#include <string>
+#include <climits>
+#include <stdexcept>
+
+#include "unescape.h"
+#include "unescape_data.h"
+
+// https://stackoverflow.com/a/42013433
+static inline size_t codepoint_to_utf8(char* buf, const unsigned long code) {
+    // you see, i don't care
+    // https://t.me/NightShadowsHangout/670534
+    #pragma GCC diagnostic push
+    #pragma GCC diagnostic ignored "-Wconversion"
+    if (code <= 0x7F) {
+        buf[0] = static_cast<char>(code);
+        return 1;
+    } else if (code <= 0x7FF) {
+        buf[0] = 0xC0 | static_cast<char>(code >> 6);   /* 110xxxxx */
+        buf[1] = 0x80 | (code & 0x3F);                  /* 10xxxxxx */
+        return 2;
+    } else if (code <= 0xFFFF) {
+        buf[0] = 0xE0 | static_cast<char>(code >> 12);  /* 1110xxxx */
+        buf[1] = 0x80 | ((code >> 6) & 0x3F);           /* 10xxxxxx */
+        buf[2] = 0x80 | (code & 0x3F);                  /* 10xxxxxx */
+        return 3;
+    } else if (code <= 0x10FFFF) {
+        buf[0] = 0xF0 | static_cast<char>(code >> 18);  /* 11110xxx */
+        buf[1] = 0x80 | ((code >> 12) & 0x3F);          /* 10xxxxxx */
+        buf[2] = 0x80 | ((code >> 6) & 0x3F);           /* 10xxxxxx */
+        buf[3] = 0x80 | (code & 0x3F);                  /* 10xxxxxx */
+        return 4;
+    #pragma GCC diagnostic pop
+    } else {
+        throw std::invalid_argument("codepoint passed is bigger than 0x10FFFF");
+    }
+}
+
+static inline bool isdigit(char ch) {
+    return ch >= '0' && ch <= '9';
+}
+
+static inline bool ishex(char ch) {
+    return isdigit(ch) || (ch >= 'a' && ch <= 'f') || (ch >= 'A' && ch <= 'F');
+}
+
+static inline unsigned long decode_numeric_entity(const std::string& entity) {
+    unsigned long codepoint;
+    char* last_converted_char;
+
+    errno = 0;
+    if (entity[1] == 'x' || entity[1] == 'X') {
+        if (entity.size() <= 2 || !ishex(entity[2])) {
+            return 0;
+        }
+        codepoint = strtoul(&entity.c_str()[2], &last_converted_char, 16);
+    } else {
+        if (entity.size() <= 1 || !isdigit(entity[1])) {
+            return 0;
+        }
+        codepoint = strtoul(&entity.c_str()[1], &last_converted_char, 10);
+    }
+
+    if ((codepoint == ULONG_MAX && errno == ERANGE) || last_converted_char[0] != '\0') {
+        return 0;
+    }
+
+    if (codepoint >= 0x80 && codepoint <= 0x9F) {
+        codepoint = windows1252_repl[codepoint - 0x80];
+    }
+    if (!codepoint || codepoint > 0x10FFFF || (codepoint >= 0xD800 && codepoint <= 0xDFFF)) {
+        codepoint = 0xFFFD;
+    }
+
+    return codepoint;
+}
+
+static inline unsigned long decode_string_entity(const std::string& entity) {
+    for (const Entity& i : string_entities) {
+        if (entity == i.string) {
+            return i.codepoint;
+        }
+    }
+
+    return 0;
+}
+
+static inline unsigned long decode_entity(std::string entity) {
+    return !entity.empty() && entity[0] == '#'
+        ? decode_numeric_entity(entity)
+        : decode_string_entity(entity);
+}
+
+namespace blankie {
+namespace html {
+
+[[nodiscard]] std::string unescape(const std::string& str) {
+    std::string output;
+    unsigned long codepoint;
+    char codepoint_buf[4];
+    size_t offset = 0, old_offset = 0, offset_end, codepoint_buf_size;
+
+    output.reserve(str.size());
+    while ((offset = str.find('&', offset)) != std::string::npos) {
+        offset_end = str.find(';', ++offset);
+        if (offset_end == std::string::npos) {
+            break;
+        }
+
+        codepoint = decode_entity(str.substr(offset, offset_end - offset));
+        if (codepoint) {
+            codepoint_buf_size = codepoint_to_utf8(codepoint_buf, codepoint);
+            output.append(str, old_offset, offset - old_offset - 1);
+            output.append(codepoint_buf, codepoint_buf_size);
+            old_offset = offset = offset_end + 1;
+        }
+    }
+
+    if (str.size() > old_offset) {
+        output.append(str, old_offset, std::string::npos);
+    }
+    return output;
+}
+
+} // namespace html
+} // namespace blankie

blankie/unescape.h

@@ -0,0 +1,9 @@
+#pragma once
+
+namespace blankie {
+namespace html {
+
+[[nodiscard]] std::string unescape(const std::string& str);
+
+} // namespace html
+} // namespace blankie

pixivmodels.cpp

@@ -117,8 +117,8 @@ void from_json(const nlohmann::json& j, Illust& illust) {
     illust_details.at("upload_timestamp").get_to(illust.upload_time);
 
     if (full_data) {
-        if (illust_details.contains("comment") && illust_details["comment"].is_string()) {
-            illust.comment = illust_details["comment"].get<std::string>();
+        if (illust_details.contains("comment_html") && illust_details["comment_html"].is_string()) {
+            illust.comment_html = illust_details["comment_html"].get<std::string>();
         }
         illust_details.at("display_tags").get_to(illust.tags);
     }
@@ -179,7 +179,7 @@ void from_json(const nlohmann::json& j, SearchResults& search_results) {
             // and i cba to use regex for it, especially when it's not even used in this context
             .upload_time = -1,
 
-            .comment = std::nullopt,
+            .comment_html = std::nullopt,
             .tags = std::move(tags),
             .images = {get_illust_images(i, std::nullopt)},
             .page_count = i.at("pageCount").get<size_t>()

pixivmodels.h

@@ -50,7 +50,7 @@ struct Illust {
     bool ai_generated;
     time_t upload_time;
 
-    std::optional<std::string> comment;
+    std::optional<std::string> comment_html;
     std::vector<Tag> tags;
     std::vector<Images> images;
     size_t page_count;

routes/artworks.cpp

@@ -2,6 +2,8 @@
 
 #include "routes.h"
 #include "../blankie/murl.h"
+#include "../blankie/escape.h"
+#include "../blankie/unescape.h"
 #include "../servehelper.h"
 #include "../numberhelper.h"
 #include "../pixivclient.h"
@@ -9,10 +11,9 @@
 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 Nodes parse_description_line(const httplib::Request& req, const Config& config, std::string str);
-static inline Element generate_description(const httplib::Request& req, const Config& config, const std::string& description);
 static inline Element generate_illust_tags(const httplib::Request& req, const Config& config, const Illust& illust);
-static inline std::string generate_description_text(const httplib::Request& req, const Config& config, std::string description);
+static inline blankie::html::HTMLString fix_description_links(const httplib::Request& req, const Config& config, blankie::html::HTMLString str);
+static inline std::string generate_description_text(const httplib::Request& req, const Config& config, blankie::html::HTMLString description);
 static inline Nodes generate_ogp_nodes(const httplib::Request& req, const Config& config, const Illust& illust, bool preview);
 
 static inline bool is_true(const std::string& str);
@@ -48,8 +49,8 @@ void artworks_route(const httplib::Request& req, httplib::Response& res, const C
         !preview ? generate_images(req, config, illust) : generate_preview_images(req, config, illust),
         Element("br")
     });
-    if (illust.comment) {
-        body.nodes.push_back(generate_description(req, config, *illust.comment));
+    if (illust.comment_html) {
+        body.nodes.push_back(Element("p", {fix_description_links(req, config, blankie::html::HTMLString(*illust.comment_html))}));
     }
     body.nodes.push_back(generate_illust_tags(req, config, illust));
     body.nodes.push_back(Element("p", {time_to_string(illust.upload_time)}));
@@ -122,61 +123,6 @@ static inline Element generate_preview_images(const httplib::Request& req, const
     return div;
 }
 
-static inline Nodes parse_description_line(const httplib::Request& req, const Config& config, std::string str) {
-    Nodes nodes;
-    std::smatch sm;
-
-    while (std::regex_search(str, sm, blankie::murl::full_url_regex)) {
-        std::string prefix = sm.prefix();
-        std::string url_str = sm.str(0);
-        std::string suffix = sm.suffix();
-
-        if (prefix.ends_with('(') && url_str.ends_with(')')) {
-            url_str.pop_back();
-            suffix.insert(0, 1, ')');
-        }
-        if (!prefix.empty()) {
-            nodes.push_back(std::move(prefix));
-        }
-
-        blankie::murl::Url url(std::move(url_str));
-        url_str = url.is_host_equal("pixiv.net") || url.is_host_equal("www.pixiv.net")
-            ? proxy_pixiv_url(req, config, std::move(url))
-            : url.to_string();
-        nodes.push_back(Element("a", {{"href", url_str}}, {url_str}));
-
-        str = std::move(suffix);
-    }
-    if (!str.empty()) {
-        nodes.push_back(std::move(str));
-    }
-
-    return nodes;
-}
-
-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;
-    auto add = [&](std::string str) {
-        if (!p.nodes.empty()) {
-            p.nodes.push_back(Element("br"));
-        }
-        Nodes nodes = parse_description_line(req, config, std::move(str));
-        p.nodes.insert(p.nodes.end(), nodes.begin(), nodes.end());
-    };
-
-    while ((pos = description.find('\n', pos)) != std::string::npos) {
-        add(description.substr(last_pos, pos - last_pos));
-        last_pos = ++pos;
-    }
-    if (description.size() > last_pos) {
-        add(description.substr(last_pos));
-    }
-
-    return p;
-}
-
 static inline Element generate_illust_tags(const httplib::Request& req, const Config& config, const Illust& illust) {
     Element div("div", {{"class", "illust-tags"}}, {});
 
@@ -198,33 +144,63 @@ static inline Element generate_illust_tags(const httplib::Request& req, const Co
     return div;
 }
 
-static inline std::string generate_description_text(const httplib::Request& req, const Config& config, std::string description) {
+const std::regex start_link_regex("<a href=\"([^\"]+?)\"(?: target=\"_blank\")?>");
+static inline blankie::html::HTMLString fix_description_links(const httplib::Request& req, const Config& config, blankie::html::HTMLString str) {
+    using namespace std::string_literals;
+
+    blankie::html::HTMLString out;
+    std::smatch sm;
+
+    out.str.reserve(str.str.size());
+    while (std::regex_search(str.str, sm, start_link_regex)) {
+        out.str += sm.prefix();
+
+        std::string url_str;
+        blankie::murl::Url url(sm.str(1));
+        if (url.is_host_equal("pixiv.net") || url.is_host_equal("www.pixiv.net")) {
+            url_str = proxy_pixiv_url(req, config, std::move(url));
+        } else if (url.path == "/jump.php") {
+            url_str = blankie::murl::unescape(std::move(url.query));
+        } else {
+            url_str = url.to_string();
+        }
+        out.str += "<a href=\""s + blankie::html::escape(std::move(url_str)) + "\">";
+
+        str.str = sm.suffix();
+    }
+    out.str += std::move(str.str);
+
+    return out;
+}
+
+const std::regex link_regex("<a href=\"([^\"]+?)\">.+?</a>");
+const std::regex tag_regex("<(.+?)>");
+static inline std::string generate_description_text(const httplib::Request& req, const Config& config, blankie::html::HTMLString description) {
+    description = fix_description_links(req, config, std::move(description));
+
     std::string new_description;
     std::smatch sm;
 
-    new_description.reserve(description.size());
-    while (std::regex_search(description, sm, blankie::murl::full_url_regex)) {
-        std::string prefix = sm.prefix();
-        std::string url_str = sm.str(0);
-        std::string suffix = sm.suffix();
-
-        if (prefix.ends_with('(') && url_str.ends_with(')')) {
-            url_str.pop_back();
-            suffix.insert(0, 1, ')');
-        }
-        new_description += std::move(prefix);
-
-        blankie::murl::Url url(std::move(url_str));
-        url_str = url.is_host_equal("pixiv.net") || url.is_host_equal("www.pixiv.net")
-            ? proxy_pixiv_url(req, config, std::move(url))
-            : url.to_string();
-        new_description += std::move(url_str);
-
-        description = std::move(suffix);
+    new_description.reserve(description.str.size());
+    while (std::regex_search(description.str, sm, link_regex)) {
+        new_description += sm.prefix();
+        new_description += sm.str(1);
+        description.str = sm.suffix();
     }
-    new_description += std::move(description);
+    new_description += std::move(description.str);
 
-    return new_description;
+    description.str = std::move(new_description);
+    new_description.reserve(description.str.size());
+    while (std::regex_search(description.str, sm, tag_regex)) {
+        new_description += sm.prefix();
+        if (sm.str(1) == "br /") {
+            new_description += '\n';
+        }
+        description.str = sm.suffix();
+    }
+    new_description += std::move(description.str);
+
+    return blankie::html::unescape(std::move(new_description));
 }
 
 static inline Nodes generate_ogp_nodes(const httplib::Request& req, const Config& config, const Illust& illust, bool preview) {
@@ -238,8 +214,8 @@ static inline Nodes generate_ogp_nodes(const httplib::Request& req, const Config
         Element("meta", {{"property", "og:site_name"}, {"content", "Pixwhile"}}, {}),
         Element("meta", {{"property", "og:url"}, {"content", std::move(url)}}, {})
     });
-    if (illust.comment) {
-        nodes.push_back(Element("meta", {{"property", "og:description"}, {"content", generate_description_text(req, config, *illust.comment)}}, {}));
+    if (illust.comment_html) {
+        nodes.push_back(Element("meta", {{"property", "og:description"}, {"content", generate_description_text(req, config, blankie::html::HTMLString(*illust.comment_html))}}, {}));
     }
 
     // i don't even know what multiple og:images do anymore