Try to fetch account information from posts without one

Example of one with: https://dlx.pink/notice/AbtdJkjioOo8ZSdDhw
Example of one without: https://dlx.pink/notice/AbD2kgNviafFEsebqq

models.cpp

@@ -7,6 +7,10 @@
 #include "models.h"
 #include "numberhelper.h"
 
+#define DOMAIN_RE "(?:[a-zA-Z0-9-]+\\.)+[a-zA-Z]{2,}"
+// https://docs.joinmastodon.org/methods/accounts/#422-unprocessable-entity
+#define USERNAME_RE "[a-zA-Z0-9_]+"
+
 using json = nlohmann::json;
 static time_t parse_rfc3339(const std::string& str);
 
@@ -28,15 +32,10 @@ void from_json(const json& j, AccountField& field) {
     }
 }
 
-static std::regex host_regex(R"EOF(https?://([a-z0-9-.]+)/.+)EOF", std::regex::ECMAScript | std::regex::icase);
+static std::regex host_regex("https?://(" DOMAIN_RE ")/.+", std::regex::ECMAScript | std::regex::icase);
 void from_json(const json& j, Account& account) {
     using namespace std::string_literals;
 
-    // https://dlx.pink/notice/AbtdJkjioOo8ZSdDhw
-    if (j.size() == 0) {
-        return;
-    }
-
     j.at("id").get_to(account.id);
     j.at("username").get_to(account.username);
     j.at("display_name").get_to(account.display_name);
@@ -117,6 +116,7 @@ void from_json(const json& j, Poll& poll) {
     j.at("emojis").get_to(poll.emojis);
 }
 
+static std::regex akkoma_status_url_regex("https?://(" DOMAIN_RE ")/(?:@|users/)(" USERNAME_RE ")/.+");
 void from_json(const json& j, Post& post) {
     j.at("id").get_to(post.id);
     post.created_at = parse_rfc3339(j.at("created_at").get_ref<const std::string&>());
@@ -142,13 +142,30 @@ void from_json(const json& j, Post& post) {
         post.reblog = std::make_unique<Post>();
         from_json(j["reblog"].get<json>(), *post.reblog.get());
     }
-    j.at("account").get_to(post.account);
     j.at("media_attachments").get_to(post.media_attachments);
     j.at("emojis").get_to(post.emojis);
     // https://social.kernel.org/@monsieuricon/Ac6oYwtLhess6uil1c
     if (j.contains("poll") && !j["poll"].is_null()) {
         post.poll = j["poll"].get<Poll>();
     }
+
+    // empty account with username accessible: https://dlx.pink/notice/AbtdJkjioOo8ZSdDhw
+    // empty account with username inaccesible: https://dlx.pink/notice/AbD2kgNviafFEsebqq
+    if (j.at("account").size()) {
+        j["account"].get_to(post.account);
+    } else {
+        std::smatch sm;
+        const std::string& url = j.at("url").get_ref<const std::string&>();
+        if (!std::regex_match(url, sm, akkoma_status_url_regex)) {
+            return;
+        }
+
+        post.account = {
+            .username = sm.str(2),
+            .server = sm.str(1),
+            .display_name = sm.str(2),
+        };
+    }
 }
 
 void from_json(const json& j, PostContext& context) {

routes/css.cpp

@@ -105,7 +105,12 @@ svg {
 .post-header span, .post-header time {
     margin-top: auto;
     margin-bottom: auto;
-    margin-left: 0.5em;
+}
+.post-header img {
+    margin-right: 0.5em;
+}
+.post-header time {
+    padding-left: 1em;
 }
 .post-time_header {
     margin-left: auto;

routes/status.cpp

@@ -33,6 +33,7 @@ void status_route(const httplib::Request& req, httplib::Response& res) {
     }
 
     // https://dlx.pink/notice/AbtdJkjioOo8ZSdDhw
+    // https://dlx.pink/notice/AbD2kgNviafFEsebqq
     if (post->reblog && !post->reblog->account.id.empty()) {
         serve_redirect(req, res, get_origin(req) + '/' + server + "/@" + post->reblog->account.acct(false) + '/' + post->reblog->id, true);
         return;

servehelper.cpp

@@ -26,7 +26,7 @@ struct PostStatus {
     const char* icon_html;
     Node info_node;
 };
-static Element serialize_post(const httplib::Request& req, const std::string& server, const Post& post, bool main_post, const std::optional<PostStatus>& post_status);
+static Element serialize_post(const httplib::Request& req, const std::string& server, const Post& post, bool main_post, const std::optional<PostStatus>& post_status, const Post* reblogged = nullptr);
 static inline Element serialize_media(const Media& media);
 static inline Element serialize_poll(const httplib::Request& req, const Poll& poll);
 
@@ -177,7 +177,7 @@ Element serialize_post(const httplib::Request& req, const std::string& server, c
             fa_retweet,
             preprocess_html(req, post.account.emojis, post.account.display_name + " boosted"),
         };
-        return serialize_post(req, server, *post.reblog, main_post, post_status);
+        return serialize_post(req, server, *post.reblog, main_post, post_status, &post);
     } else if (pinned) {
         PostStatus post_status = {
             fa_thumbtack,
@@ -437,9 +437,17 @@ static inline std::vector<lxb_dom_node_t*> emojify(lxb_dom_document_t* document,
     return res;
 }
 
-static Element serialize_post(const httplib::Request& req, const std::string& server, const Post& post, bool main_post, const std::optional<PostStatus>& post_status) {
+static Element serialize_post(const httplib::Request& req, const std::string& server, const Post& post, bool main_post, const std::optional<PostStatus>& post_status, const Post* reblogged) {
     using namespace std::string_literals;
 
+    bool user_known = !post.account.id.empty();
+    bool user_ref_known = !post.account.username.empty() && !post.account.server.empty();
+    // `reblogged == nullptr` since a malicious server could take down the frontend
+    // by sending a post that is not a reblog with no account information
+    std::string post_url = user_known || reblogged == nullptr
+        ? get_origin(req) + '/' + server + "/@" + post.account.acct(false) + '/' + post.id + "#m"
+        : get_origin(req) + '/' + server + "/@" + reblogged->account.acct(false) + '/' + reblogged->id + "#m";
+
     std::string time_title = post.edited_at < 0
         ? full_time(post.created_at)
         : "Created: "s + full_time(post.created_at) + "\nEdited: " + full_time(post.edited_at);
@@ -478,14 +486,16 @@ static Element serialize_post(const httplib::Request& req, const std::string& se
 
     Element div("div", {{"class", "post"}}, {
         Element("div", {{"class", "post-header"}}, {
-            !post.account.id.empty() ? Element("a", {{"href", get_origin(req) + '/' + server + "/@" + post.account.acct(false)}}, {
-                Element("img", {{"class", "post-avatar"}, {"alt", "User profile picture"}, {"loading", "lazy"}, {"src", post.account.avatar_static}}, {}),
+            user_ref_known ? Element("a", {{"href", get_origin(req) + '/' + server + "/@" + post.account.acct(false)}}, {
+                !post.account.avatar_static.empty()
+                    ? Element("img", {{"class", "post-avatar"}, {"alt", "User profile picture"}, {"loading", "lazy"}, {"src", post.account.avatar_static}}, {})
+                    : Node(""),
                 Element("span", {
                     Element("b", {preprocess_html(req, post.account.emojis, post.account.display_name)}),
                     Element("br"), "@", post.account.acct(),
                 }),
             }) : Element("b", {"Unknown user"}),
-            Element("a", {{"class", "post-time_header"}, {"href", get_origin(req) + '/' + server + "/@" + post.account.acct(false) + '/' + post.id + "#m"}, {"title", time_title}}, {
+            Element("a", {{"class", "post-time_header"}, {"href", std::move(post_url)}, {"title", time_title}}, {
                 Element("time", {{"datetime", to_rfc3339(post.created_at)}}, {relative_time(post.created_at, current_time()), time_badge}),
             }),
         }),