src/ArticleCollection.cpp

   1 //! \file ArticleCollection.cpp
   2
   3 #include "ArticleCollection.h"
   4
   5 #include <algorithm>
   6 #include <utility>
   7
   8 #include "Article.h"
   9 #include "WalkerException.h"
  10
  11 namespace WikiWalker
  12 {
  13   namespace CollectionUtils
  14   {
  15     static bool articleIsAnalyzed(ArticleCollection::const_reference x)
  16     {
  17       auto art = x.second;
  18       if(art == nullptr) {
  19         return false;
  20       }
  21       return art->analyzed();
  22     }
  23
  24     size_t countAnalyzedArticles(const ArticleCollection& collection)
  25     {
  26       return std::count_if(
  27           collection.begin(), collection.end(), articleIsAnalyzed);
  28     }
  29
  30     bool add(ArticleCollection& collection, std::shared_ptr<Article> article)
  31     {
  32       auto ret = collection.insert(std::make_pair(article->title(), article));
  33       return ret.second;
  34     }
  35
  36     void merge(ArticleCollection& collection,
  37                const ArticleCollection& other,
  38                MergeStrategy strategy)
  39     {
  40       switch(strategy) {
  41         case MergeStrategy::IgnoreDuplicates:
  42           collection.insert(other.begin(), other.end());
  43           break;
  44         case MergeStrategy::AlwaysOverwrite:
  45           for(const auto& art : other) {
  46             collection[art.first] = art.second;
  47           }
  48           break;
  49         case MergeStrategy::UseArticleWithMoreLinks:
  50           for(const auto& art : other) {
  51             auto articleInThis = collection[art.first];
  52             auto& otherArt     = art.second;
  53
  54             if(articleInThis == nullptr ||
  55                (!articleInThis->analyzed() && otherArt->analyzed())) {
  56               collection[art.first] = art.second;
  57               continue;
  58             }
  59
  60             if(!otherArt->analyzed()) {
  61               continue;
  62             }
  63
  64             auto linksInThis  = articleInThis->countLinks();
  65             auto linksInOther = otherArt->countLinks();
  66             if(linksInOther > linksInThis) {
  67               collection[art.first] = art.second;
  68             }
  69           }
  70           break;
  71         default:
  72           throw WalkerException("Not supported");
  73       }
  74     }
  75
  76     /*! \todo I'm not exactly happy with returning a shared_ptr. Return
  77      * reference instead? - but how to say "not found" then? */
  78     std::shared_ptr<Article> get(const ArticleCollection& collection,
  79                                  const std::string& title)
  80     {
  81       auto it = collection.find(title);
  82
  83       if(collection.end() == it) {
  84         return nullptr;
  85       }
  86
  87       return it->second;
  88     }
  89   }  // namespace CollectionUtils
  90
  91 }  // namespace WikiWalker