Add OGP supports

2023-12-03 00:50:33 +11:00 · 2023-12-03 00:50:33 +11:00 · 6cb53f1bc1
parent dfe629aa2b
commit 6cb53f1bc1
7 changed files with 176 additions and 7 deletions
--- a/blankie/serializer.cpp
+++ b/blankie/serializer.cpp
@ -58,5 +58,6 @@ static inline bool is_autoclosing_tag(const char* tag) {
        || !strncmp(tag, "meta", 5)
        || !strncmp(tag, "img", 4)
        || !strncmp(tag, "br", 3)
+        || !strncmp(tag, "hr", 3)
        || !strncmp(tag, "input", 6);
 }
--- a/models.cpp
+++ b/models.cpp
@ -60,6 +60,11 @@ void from_json(const json& j, Account& account) {
    }
 }

+void from_json(const json& j, Size& size) {
+    j.at("width").get_to(size.width);
+    j.at("height").get_to(size.height);
+}
+
 void from_json(const json& j, Media& media) {
    j.at("type").get_to(media.type);
    j.at("url").get_to(media.url);
@ -69,6 +74,11 @@ void from_json(const json& j, Media& media) {
    if (!j.at("remote_url").is_null()) {
        media.remote_url = j["remote_url"].get<std::string>();
    }
+    if (media.type == "image" || media.type == "video" || media.type == "gifv") {
+        const json& meta = j.at("meta");
+        media.size = meta.at("original").get<Size>();
+        media.preview_size = meta.at("small").get<Size>();
+    }
    if (!j.at("description").is_null()) {
        media.description = j["description"].get<std::string>();
    }
--- a/models.h
+++ b/models.h
@ -51,11 +51,17 @@ struct Account {
    }
 };

+struct Size {
+    uint64_t width;
+    uint64_t height;
+};
 struct Media {
    std::string type;
    std::string url;
    std::optional<std::string> preview_url;
    std::optional<std::string> remote_url;
+    std::optional<Size> size;
+    std::optional<Size> preview_size;
    std::optional<std::string> description;
 };

@ -108,6 +114,7 @@ struct Instance {
 void from_json(const nlohmann::json& j, Emoji& emoji);
 void from_json(const nlohmann::json& j, AccountField& field);
 void from_json(const nlohmann::json& j, Account& account);
+void from_json(const nlohmann::json& j, Size& size);
 void from_json(const nlohmann::json& j, Media& media);
 void from_json(const nlohmann::json& j, PollOption& option);
 void from_json(const nlohmann::json& j, Poll& poll);
--- a/routes/status.cpp
+++ b/routes/status.cpp
@ -5,6 +5,8 @@
 #include "../models.h"

 static inline std::string make_title(const Post& post);
+static inline Nodes generate_ogp_nodes(const httplib::Request& req, const Post& post, const std::string& server);
+static inline void generate_media_ogp_nodes(Nodes& nodes, const Media& media, bool* has_video, bool* has_image);


 void status_route(const httplib::Request& req, httplib::Response& res) {
@ -48,7 +50,7 @@ void status_route(const httplib::Request& req, httplib::Response& res) {
        body.nodes.push_back(serialize_post(req, server, i));
    }

-    serve(req, res, make_title(*post), std::move(body));
+    serve(req, res, make_title(*post), std::move(body), generate_ogp_nodes(req, *post, server));
 }


@ -69,3 +71,64 @@ static inline std::string make_title(const Post& post) {

    return title;
 }
+
+static inline Nodes generate_ogp_nodes(const httplib::Request& req, const Post& post, const std::string& server) {
+    using namespace std::string_literals;
+    std::string url = get_origin(req) + '/' + server + "/@" + post.account.acct(false) + '/' + post.id;
+    bool has_video = false, has_image = false;
+
+    Nodes nodes({
+        // left-to-right override--thank https://anarres.family/@alice@mk.nyaa.place
+        Element("meta", {{"property", "og:title"}, {"content", post.account.display_name + "\u202d (@" + post.account.acct() + ')'}}, {}),
+        Element("meta", {{"property", "og:site_name"}, {"content", "Coyote"}}, {}),
+        Element("meta", {{"property", "og:url"}, {"content", std::move(url)}}, {}),
+    });
+    if (!post.sensitive) {
+        nodes.push_back(Element("meta", {{"property", "og:description"}, {"content", get_text_content(post.content)}}, {}));
+
+        for (const Media& media : post.media_attachments) {
+            generate_media_ogp_nodes(nodes, media, &has_video, &has_image);
+        }
+    } else if (!post.spoiler_text.empty()) {
+        nodes.push_back(Element("meta", {{"property", "og:description"}, {"content", "CW: "s + post.spoiler_text}}, {}));
+    }
+
+    const char* type = !post.sensitive && has_video
+        ? "video"
+        : !post.sensitive && has_image ? "image" : "article";
+    nodes.push_back(Element("meta", {{"property", "og:type"}, {"content", type}}, {}));
+
+    return nodes;
+}
+
+static inline void generate_media_ogp_nodes(Nodes& nodes, const Media& media, bool* has_video, bool* has_image) {
+    if (media.type == "image") {
+        *has_image = true;
+        nodes.push_back(Element("meta", {{"property", "og:image"}, {"content", media.preview_url.value_or(media.url)}}, {}));
+
+        std::optional<Size> size = media.preview_size ? media.preview_size : media.size;
+        if (size) {
+            nodes.push_back(Element("meta", {{"property", "og:image:width"}, {"content", std::to_string(size->width)}}, {}));
+            nodes.push_back(Element("meta", {{"property", "og:image:height"}, {"content", std::to_string(size->height)}}, {}));
+        }
+
+        if (media.description) {
+            nodes.push_back(Element("meta", {{"property", "og:image:alt"}, {"content", *media.description}}, {}));
+        }
+    } else if (media.type == "video" || media.type == "gifv") {
+        *has_video = true;
+        nodes.push_back(Element("meta", {{"property", "og:video"}, {"content", media.preview_url.value_or(media.url)}}, {}));
+
+        std::optional<Size> size = media.preview_size ? media.preview_size : media.size;
+        if (size) {
+            nodes.push_back(Element("meta", {{"property", "og:video:width"}, {"content", std::to_string(size->width)}}, {}));
+            nodes.push_back(Element("meta", {{"property", "og:video:height"}, {"content", std::to_string(size->height)}}, {}));
+        }
+
+        if (media.description) {
+            nodes.push_back(Element("meta", {{"property", "og:video:alt"}, {"content", *media.description}}, {}));
+        }
+    } else if (media.type == "audio") {
+        nodes.push_back(Element("meta", {{"property", "og:audio"}, {"content", media.url}}, {}));
+    }
+}
--- a/routes/user.cpp
+++ b/routes/user.cpp
@ -12,6 +12,8 @@ static inline Element user_header(const httplib::Request& req, const std::string
 static inline Element user_link_field(const httplib::Request& req, const Account& account, const AccountField& field);
 static inline Element sorting_method_link(const httplib::Request& req, const std::string& server, const Account& account, PostSortingMethod current_method, PostSortingMethod new_method);

+static inline Nodes generate_ogp_nodes(const httplib::Request& req, const Account& account, const std::optional<std::string>& max_id, PostSortingMethod sorting_method);
+

 void user_route(const httplib::Request& req, httplib::Response& res) {
    using namespace std::string_literals;
@ -66,7 +68,7 @@ void user_route(const httplib::Request& req, httplib::Response& res) {
        body.nodes.push_back(Element("p", {{"class", "more_posts"}}, {"There are no more posts"}));
    }

-    serve(req, res, account->display_name + " (@" + account->acct() + ')', std::move(body));
+    serve(req, res, account->display_name + " (@" + account->acct() + ')', std::move(body), generate_ogp_nodes(req, *account, max_id, sorting_method));
 }


@ -161,3 +163,27 @@ static inline Element sorting_method_link(const httplib::Request& req, const std
    }
    return a;
 }
+
+
+static inline Nodes generate_ogp_nodes(const httplib::Request& req, const Account& account, const std::optional<std::string>& max_id, PostSortingMethod sorting_method) {
+    std::string url = get_origin(req) + '/' + account.server + "/@" + account.acct(false) + sorting_method_suffixes[sorting_method];
+    if (max_id) {
+        url += "?max_id=";
+        url += *max_id;
+    }
+
+    std::string note = get_text_content(account.note_html);
+
+    Nodes nodes({
+        // left-to-right override--thank https://anarres.family/@alice@mk.nyaa.place
+        Element("meta", {{"property", "og:title"}, {"content", account.display_name + "\u202d (@" + account.acct() + ')'}}, {}),
+        Element("meta", {{"property", "og:type"}, {"content", "website"}}, {}),
+        Element("meta", {{"property", "og:site_name"}, {"content", "Coyote"}}, {}),
+        Element("meta", {{"property", "og:url"}, {"content", std::move(url)}}, {}),
+        Element("meta", {{"property", "og:image"}, {"content", account.avatar}}, {}),
+    });
+    if (!note.empty()) {
+        nodes.push_back(Element("meta", {{"property", "og:description"}, {"content", std::move(note)}}, {}));
+    }
+    return nodes;
+}
--- a/servehelper.cpp
+++ b/servehelper.cpp
@ -18,6 +18,7 @@
 static inline void preprocess_html(const httplib::Request& req, const std::string& domain_name, const std::vector<Emoji>& emojis, lxb_dom_element_t* element);
 static inline void preprocess_link(const httplib::Request& req, const std::string& domain_name, lxb_dom_element_t* element);
 static inline bool should_fix_link(lxb_dom_element_t* element, const std::string& element_cls);
+static inline void get_text_content(lxb_dom_node_t* node, std::string& out);
 static inline lxb_dom_node_t* emojify(lxb_dom_node_t* child, const std::vector<Emoji>& emojis);
 static inline std::vector<lxb_dom_node*> emojify(lxb_dom_document_t* document, std::string str, const std::vector<Emoji>& emojis);

@ -194,6 +195,35 @@ Element serialize_post(const httplib::Request& req, const std::string& server, c
    }
 }

+std::string get_text_content(lxb_dom_node_t* child) {
+    std::string out;
+    get_text_content(child, out);
+
+    if (!out.empty()) {
+        size_t remove_from = out.size();
+        while (remove_from && out[remove_from - 1] == '\n') {
+            remove_from--;
+        }
+        if (out.size() > remove_from) {
+            out.erase(remove_from);
+        }
+    }
+    if (!out.empty()) {
+        size_t remove_to = 0;
+        while (out.size() > remove_to && out[remove_to] == '\n') {
+            remove_to++;
+        }
+        out.erase(0, remove_to);
+    }
+
+    return out;
+}
+
+std::string get_text_content(blankie::html::HTMLString str) {
+    LXB::HTML::Document document(str.str);
+    return get_text_content(document.body());
+}
+
 blankie::html::HTMLString preprocess_html(const httplib::Request& req, const std::string& domain_name, const std::vector<Emoji>& emojis, const blankie::html::HTMLString& str) {
    LXB::HTML::Document document(str.str);
    preprocess_html(req, domain_name, emojis, document.body_element());
@ -316,11 +346,40 @@ static inline bool should_fix_link(lxb_dom_element_t* element, const std::string
    return child == nullptr;
 }

-static inline lxb_dom_node_t* emojify(lxb_dom_node_t* child, const std::vector<Emoji>& emojis) {
-    size_t text_content_len;
-    const char* text_content = reinterpret_cast<const char*>(lxb_dom_node_text_content(child, &text_content_len));
+static inline void get_text_content(lxb_dom_node_t* node, std::string& out) {
+    bool is_br = false, is_p = false;

-    std::vector<lxb_dom_node_t*> nodes = emojify(child->owner_document, std::string(text_content, text_content_len), emojis);
+    if (node->type == LXB_DOM_NODE_TYPE_TEXT) {
+        size_t len;
+        const char* text = reinterpret_cast<const char*>(lxb_dom_node_text_content(node, &len));
+
+        out.append(text, len);
+    } else if (node->type == LXB_DOM_NODE_TYPE_ELEMENT) {
+        lxb_dom_element_t* element = lxb_dom_interface_element(node);
+        const char* tag_name = reinterpret_cast<const char*>(lxb_dom_element_tag_name(element, nullptr));
+
+        is_p = strncmp(tag_name, "P", 2) == 0;
+        is_br = strncmp(tag_name, "BR", 3) == 0;
+    }
+
+    if (is_p || is_br) {
+        out.push_back('\n');
+    }
+
+    lxb_dom_node_t* child = lxb_dom_node_first_child(node);
+    while (child) {
+        get_text_content(child, out);
+
+        child = lxb_dom_node_next(child);
+    }
+
+    if (is_p) {
+        out.push_back('\n');
+    }
+}
+
+static inline lxb_dom_node_t* emojify(lxb_dom_node_t* child, const std::vector<Emoji>& emojis) {
+    std::vector<lxb_dom_node_t*> nodes = emojify(child->owner_document, get_text_content(child), emojis);

    lxb_dom_node_insert_after(child, nodes[0]);
    lxb_dom_node_destroy(child);
@ -460,7 +519,7 @@ static inline Element serialize_media(const Media& media) {
                video.attributes.push_back({"poster", *media.preview_url});
            }
            return video;
-        } else if (media.type == "unknown" && media.remote_url) {
+        } else if (media.type == "unknown") {
            if (media.remote_url) {
                // https://botsin.space/@lina@vt.social/111053598696451525
                return Element("a", {{"href", *media.remote_url}}, {"Media is not available from this instance, view externally"});
--- a/servehelper.h
+++ b/servehelper.h
@ -4,6 +4,7 @@

 #include <httplib/httplib.h>
 #include "blankie/serializer.h"
+#include "lxb_wrapper.h"
 struct Post; // forward declaration from models.h
 struct Emoji; // forward declaration from models.h
 class CurlUrl; // forward declaration from curlu_wrapper.h
@ -24,5 +25,7 @@ bool should_send_304(const httplib::Request& req, uint64_t hash);

 Element serialize_post(const httplib::Request& req, const std::string& server, const Post& post, bool pinned = false, bool main_post = false);

+std::string get_text_content(lxb_dom_node_t* child);
+std::string get_text_content(blankie::html::HTMLString str);
 blankie::html::HTMLString preprocess_html(const httplib::Request& req, const std::string& domain_name, const std::vector<Emoji>& emojis, const blankie::html::HTMLString& str);
 blankie::html::HTMLString preprocess_html(const httplib::Request& req, const std::vector<Emoji>& emojis, const std::string& str);