#include "rapidxml.hpp"
#include <cstring>

inline bool is_content_root(const char* name, std::size_t len) {
    return (len == 8 && std::memcmp(name, "abstract", 8) == 0) ||
           (len == 4 && std::memcmp(name, "body", 4) == 0);
}

size_t sum_text(const rapidxml::xml_node<>* node) {
    size_t total = 0;
    for (auto* child = node->first_node(); child; child = child->next_sibling()) {
        if (child->type() == rapidxml::node_type::node_data) {
            total += child->value_size();
        } else if (child->type() == rapidxml::node_type::node_element) {
            total += sum_text(child);
        }
    }
    return total;
}

size_t walk(const rapidxml::xml_node<>* node) {
    size_t total = 0;
    for (auto* child = node->first_node(); child; child = child->next_sibling()) {
        if (child->type() != rapidxml::node_type::node_element) continue;
        if (is_content_root(child->name(), child->name_size())) {
            total += sum_text(child);
        } else {
            total += walk(child);
        }
    }
    return total;
}

size_t pubmed_len(char* buffer, size_t /*len*/) {
    rapidxml::xml_document<> doc;
    doc.parse<rapidxml::parse_no_entity_translation |
              rapidxml::parse_no_string_terminators>(buffer);
    return walk(&doc);
}

size_t pubmed_len_with_entities(char* buffer, size_t /*len*/) {
    rapidxml::xml_document<> doc;
    doc.parse<rapidxml::parse_no_string_terminators>(buffer);
    return walk(&doc);
}