inc/WikimediaApi.h

   1 //! \file
   2
   3 #ifndef WIKIWALKER_WIKIMEDIAAPI_H
   4 #define WIKIWALKER_WIKIMEDIAAPI_H
   5
   6 #include <string>
   7
   8 #include "ArticleCollection.h"
   9 #include "CurlWikiGrabber.h"
  10
  11 namespace WikiWalker
  12 {
  13   namespace WikimediaApiUtils
  14   {
  15     //! Represents URL components
  16     struct WikimediaUrlInfo {
  17       //! the API base URL
  18       std::string apiBaseUrl;
  19       //! the article title
  20       std::string articleTitle;
  21     };
  22
  23     /*! Parses a Wikimedia article URL and returns components.
  24      * \param articleUrl the Wikimedia article URL.
  25      * \returns a stuct containing base URL and title.
  26      */
  27     WikimediaUrlInfo parseArticleUrl(const std::string& articleUrl);
  28   }  // namespace WikimediaApiUtils
  29
  30   /*! \brief Represents the Wikimedia API.
  31    * \details the needed parts.
  32    * The API already converts the string/Json data to a
  33    * CollectionUtils::ArticleCollection.
  34    */
  35   class WikimediaApi
  36   {
  37   public:
  38     //! Usable generators
  39     enum class WikimediaGenerator {
  40       //! Don't use a generator
  41       NoGenerator,
  42       //! use the forwardlinks generator
  43       ForwardLinkGenerator
  44     };
  45
  46     /*! \brief Create a new instance of the WikimediaApi
  47      * \param baseUrl the API base URL of the Wikimedia instance, including the
  48      * protocol.
  49      * \todo Take "DataFetcher" or whatever to replace default CURL
  50      */
  51     explicit WikimediaApi(std::string baseUrl);
  52
  53     /*! Get the forward links from a specified title and store it in the
  54      * collection.
  55      *
  56      * \param title the title of the WikimediaArticle.
  57      * \param generator which generator to use to also fetch links in
  58      * articles linked from the specified artitle. ONLY this generator is used.
  59      * If you specify ForwardLinks, only the ForwardLinks generator is used. You
  60      * won't get the links of the article itself. Use NoGenerator for that.
  61      * \param[out] collection the collection to store into. If article already
  62      * exists in collection, it will be skipped (TODO: REALLY?).
  63      */
  64     void fetchForwardLinks(const std::string& title,
  65                            WikimediaGenerator generator,
  66                            CollectionUtils::ArticleCollection& collection);
  67
  68     /*! Get the backward links from a specified title and store it in the
  69      * collection.
  70      *
  71      * \param title the title of the WikimediaArticle.
  72      * \param generator which generator to use to also fetch links in
  73      * articles linked from the specified artitle.
  74      * \param[out] collection the collection to store into. If article already
  75      * exists in collection, it will be skipped (TODO: REALLY?).
  76      */
  77     void fetchBackwardLinks(const std::string& title,
  78                             WikimediaGenerator generator,
  79                             CollectionUtils::ArticleCollection& collection);
  80
  81   private:
  82     std::string baseUrl_;
  83     CurlWikiGrabber grabber_;
  84   };
  85 }  // namespace WikiWalker
  86
  87 #endif  // WIKIWALKER_WIKIMEDIAAPI_H