feed_item.hpp

Go to the documentation of this file.

#pragma once
 
#include <pugixml.hpp>
#include <string>
#include <vector>
 
#include "utils.hpp"
 
using namespace pugi;
 
class FeedItem {
private:
    xml_node item_node;
    std::string website_url;
 
    std::string title;
    std::string content;
    std::string url;
    std::string media_url;
    std::string pub_date;
    std::string img_url;
    std::string author_name;
    std::string author_url;
 
    std::string extract_url();
 
    std::string extract_media_url();
 
    static inline const std::vector<ExtractionParam> __IMG_URL_PARAMS{
        {ExtractionParam::ParamType::ATTRIBUTE, {"thumbnail"}, "url"},
        {ExtractionParam::ParamType::ATTRIBUTE, {"media:thumbnail"}, "url"},
        {ExtractionParam::ParamType::ATTRIBUTE, {"group", "thumbnail"}, "url"},
        {ExtractionParam::ParamType::ATTRIBUTE, {"media:group", "thumbnail"}, "url"},
        {ExtractionParam::ParamType::ATTRIBUTE, {"media:group", "media:thumbnail"}, "url"}
    };
    std::string extract_img_url();
 
    void fix_url(std::string &s);
 
    static inline const std::vector<ExtractionParam> __CONTENT_PARAMS{
        {ExtractionParam::ParamType::CHILD, {"content"}},
        {ExtractionParam::ParamType::CHILD, {"content:encoded"}},
        {ExtractionParam::ParamType::CHILD, {"description"}},
        {ExtractionParam::ParamType::CHILD, {"summary"}},
        {ExtractionParam::ParamType::CHILD, {"media:group", "media:description"}}
    };
    static inline const std::vector<ExtractionParam> __PUB_DATE_PARAMS{
        {ExtractionParam::ParamType::CHILD, {"pubDate"}},
        {ExtractionParam::ParamType::CHILD, {"published"}},
        {ExtractionParam::ParamType::CHILD, {"updated"}},
        {ExtractionParam::ParamType::CHILD, {"date"}},
        {ExtractionParam::ParamType::CHILD, {"dc:date"}}
    };
    static inline const std::vector<ExtractionParam> __AUTHOR_NAME_PARAMS{
        {ExtractionParam::ParamType::CHILD, {"author", "name"}},
        {ExtractionParam::ParamType::CHILD, {"author"}},
        {ExtractionParam::ParamType::CHILD, {"dc:creator"}},
        {ExtractionParam::ParamType::CHILD, {"creator"}},
        {ExtractionParam::ParamType::CHILD, {"itunes:author"}},
    };
    static inline const std::vector<ExtractionParam> __AUTHOR_URL_PARAMS{
        {ExtractionParam::ParamType::CHILD, {"author", "uri"}}
    };
    void parse();
 
public:
    FeedItem(
            xml_node item_node, std::string website_url
    ): item_node{item_node}, website_url{website_url} {
        parse();
    }
 
    std::string get_title() { return title; }
    std::string get_content() { return content; }
    std::string get_url() { return url; }
    std::string get_media_url() { return media_url; }
    std::string get_pub_date() { return pub_date; }
    std::string get_img_url() { return img_url; }
    std::string get_author_name() { return author_name; }
    std::string get_author_url() { return author_url; }
 
    std::string to_json();
};