Prepend "> " to <blockquote> elements

https://tech.lgbt/@mia@void.rehab/111500676785694526

htmlhelper.cpp

@@ -9,8 +9,9 @@
 
 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 void preprocess_blockquote(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 void get_text_content(lxb_dom_node_t* node, std::string& out, size_t blockquote_depth);
 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);
 
@@ -51,16 +52,14 @@ 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);
+    get_text_content(child, out, 0);
 
     if (!out.empty()) {
         size_t remove_from = out.size();
         while (remove_from && out[remove_from - 1] == '\n') {
             remove_from--;
         }
-        // Don't engulf everything, otherwise it crashes
+        if (out.size() > remove_from) {
-        // https://ruby.social/@CoralineAda/109951421922797743
-        if (out.size() > remove_from && remove_from != 0) {
             out.erase(remove_from);
         }
     }
@@ -69,12 +68,8 @@ std::string get_text_content(lxb_dom_node_t* child) {
         while (out.size() > remove_to && out[remove_to] == '\n') {
             remove_to++;
         }
-        // Don't engulf everything, otherwise it crashes
-        // https://ruby.social/@CoralineAda/109951421922797743
-        if (out.size() > remove_to) {
         out.erase(0, remove_to);
     }
-    }
 
     return out;
 }
@@ -101,6 +96,9 @@ static inline void preprocess_html(const httplib::Request& req, const std::strin
     if (strncmp(tag_name, "A", 2) == 0) {
         // Proprocess links
         preprocess_link(req, domain_name, element);
+    } else if (strncmp(tag_name, "BLOCKQUOTE", 11) == 0) {
+        // Prepend "> " to <blockquote>
+        preprocess_blockquote(element);
     }
 
     // Walk through the element's children
@@ -171,6 +169,29 @@ static inline void preprocess_link(const httplib::Request& req, const std::strin
     }
 }
 
+// https://tech.lgbt/@mia@void.rehab/111500676785694526
+static inline void preprocess_blockquote(lxb_dom_element_t* element) {
+    using namespace std::string_literals;
+
+    lxb_dom_node_t* child = lxb_dom_node_first_child(lxb_dom_interface_node(element));
+    while (child) {
+        if (child->type == LXB_DOM_NODE_TYPE_ELEMENT) {
+            preprocess_blockquote(lxb_dom_interface_element(child));
+        } else if (child->type == LXB_DOM_NODE_TYPE_TEXT) {
+            size_t len;
+            const char* text = reinterpret_cast<const char*>(lxb_dom_node_text_content(child, &len));
+            std::string new_text = "> "s + std::string(text, len);
+
+            lxb_dom_node_t* new_child = lxb_dom_interface_node(lxb_dom_document_create_text_node(child->owner_document, reinterpret_cast<const lxb_char_t*>(new_text.data()), new_text.size()));
+            lxb_dom_node_insert_after(child, new_child);
+            lxb_dom_node_remove(child);
+            child = new_child;
+        }
+
+        child = lxb_dom_node_next(child);
+    }
+}
+
 static std::regex unhandled_link_re("\\bunhandled-link\\b");
 static inline bool should_fix_link(lxb_dom_element_t* element, const std::string& element_cls) {
     // https://vt.social/@LucydiaLuminous/111448085044245037
@@ -212,13 +233,16 @@ static inline bool should_fix_link(lxb_dom_element_t* element, const std::string
     return child == nullptr;
 }
 
-static inline void get_text_content(lxb_dom_node_t* node, std::string& out) {
+static inline void get_text_content(lxb_dom_node_t* node, std::string& out, size_t blockquote_depth) {
-    bool is_br = false, is_p = false;
+    bool is_br = false, is_p = false, is_blockquote = false;
 
     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));
 
+        for (size_t i = 0; i < blockquote_depth; i++) {
+            out += "> ";
+        }
         out.append(text, len);
     } else if (node->type == LXB_DOM_NODE_TYPE_ELEMENT) {
         lxb_dom_element_t* element = lxb_dom_interface_element(node);
@@ -226,26 +250,29 @@ static inline void get_text_content(lxb_dom_node_t* node, std::string& out) {
 
         is_p = strncmp(tag_name, "P", 2) == 0;
         is_br = strncmp(tag_name, "BR", 3) == 0;
+        is_blockquote = strncmp(tag_name, "BLOCKQUOTE", 11) == 0;
     }
 
-    if (is_p || is_br) {
+    if (is_p || is_br || is_blockquote) {
         out.push_back('\n');
     }
 
     lxb_dom_node_t* child = lxb_dom_node_first_child(node);
     while (child) {
-        get_text_content(child, out);
+        get_text_content(child, out, blockquote_depth + is_blockquote);
 
         child = lxb_dom_node_next(child);
     }
 
-    if (is_p) {
+    if (is_p || is_blockquote) {
         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);
+    size_t len;
+    const char* text = reinterpret_cast<const char*>(lxb_dom_node_text_content(child, &len));
+    std::vector<lxb_dom_node_t*> nodes = emojify(child->owner_document, std::string(text, len), emojis);
 
     lxb_dom_node_insert_after(child, nodes[0]);
     lxb_dom_node_destroy(child);