test/WikimediaJsonToArticleConverterTests.cpp

   1 #include <algorithm>
   2
   3 #include <UnitTest++/UnitTest++.h>
   4
   5 #include "Article.h"
   6 #include "WalkerException.h"
   7 #include "WikimediaJsonToArticleConverter.h"
   8
   9 SUITE(WikimediaJsonToArticleConverterTests)
  10 {
  11   using namespace WikiWalker;
  12   using namespace WikiWalker::CollectionUtils;
  13
  14   TEST(JsonDataWithOneLinkedArticle)
  15   {
  16     std::string testdata =
  17         R"({"batchcomplete":"","servedby":"mw1197","query":{"pages":[{"pageid":36669940,"ns":0,"title":"3PTT","links":[{"ns":0,"title":"Switch"}]}]}})";
  18
  19     ArticleCollection ac;
  20     WikimediaJsonToArticleConverter conv;
  21     auto cont = conv.convert(testdata, ac);
  22     CHECK(WikimediaJsonToArticleConverter::ContinuationStatus::
  23               ConversionCompleted == cont);
  24     CHECK_EQUAL("", conv.continuationData()["plcontinue"]);
  25     auto getArticle = CollectionUtils::get(ac, "3PTT");
  26     CHECK(getArticle != nullptr);
  27     CHECK_EQUAL(1, getArticle->countLinks());
  28     CHECK_EQUAL(2, ac.size());
  29   }
  30
  31   TEST(JsonDataWithInvalidArticle_Throws)
  32   {
  33     std::string testdata =
  34         R"({"batchcomplete":"","servedby":"mw1208","query":{"pages":[{"ns":0,"title":"FoObAr","missing":""}]}})";
  35
  36     ArticleCollection ac;
  37     WikimediaJsonToArticleConverter conv;
  38     auto ret = conv.convert(testdata, ac);
  39     CHECK(ret == WikiWalker::WikimediaJsonToArticleConverter::
  40                      ContinuationStatus::ConversionCompleted);
  41     auto art = CollectionUtils::get(ac, "FoObAr");
  42     CHECK(art != nullptr);
  43     CHECK(art->marked());
  44   }
  45
  46   TEST(JsonData_MoreLinks_HasContinueData)
  47   {
  48     std::string testdata =
  49         R"({"continue":{"plcontinue":"34419161|0|Jharkhand","continue":"||"},"servedby":"mw1283","query":{"pages":[{"pageid":34419161,"ns":0,"title":"Satar, Deoghar","links":[{"ns":0,"title":"Deoghar district"}]}]}})";
  50
  51     ArticleCollection ac;
  52     WikimediaJsonToArticleConverter conv;
  53     auto cont = conv.convert(testdata, ac);
  54     CHECK(cont == WikimediaJsonToArticleConverter::ContinuationStatus::
  55                       ConversionNeedsMoreData);
  56     CHECK_EQUAL("34419161|0|Jharkhand", conv.continuationData()["plcontinue"]);
  57     auto getArticle = CollectionUtils::get(ac, "Satar, Deoghar");
  58     CHECK(getArticle != nullptr);
  59     CHECK_EQUAL(1, getArticle->countLinks());
  60     CHECK_EQUAL(2, ac.size());
  61   }
  62
  63   TEST(JsonData_ContainsMultipleArticles)
  64   {
  65     std::string testdata =
  66         R"#({"batchcomplete": true,"query": {"normalized": [{"fromencoded": false,"from": "Zanfina_Ismajli","to": "Zanfina Ismajli"},{"fromencoded": false,"from": "Kleite_(Tochter_des_Danaos)","to": "Kleite (Tochter des Danaos)"}],"pages": [{"pageid": 2834303,"ns": 0,"title": "Zanfina Ismajli","links": [{"ns": 0,"title": "10. Mai"},{"ns": 0,"title": "1985"}]},{"pageid": 8086803,"ns": 0,"title": "Kleite (Tochter des Danaos)","links": [{"ns": 0,"title": "Aigyptos"},{"ns": 0,"title": "Altgriechische Sprache"}]}]},"limits": {"links": 500}})#";
  67     WikimediaJsonToArticleConverter conv;
  68     ArticleCollection ac;
  69     auto cont = conv.convert(testdata, ac);
  70     CHECK(WikimediaJsonToArticleConverter::ContinuationStatus::
  71               ConversionCompleted == cont);
  72
  73     CHECK_EQUAL(2, CollectionUtils::countAnalyzedArticles(ac));
  74     auto ptr = CollectionUtils::get(ac, "Zanfina Ismajli");
  75     CHECK(ptr != nullptr);
  76     ptr = CollectionUtils::get(ac, "Kleite (Tochter des Danaos)");
  77     CHECK(ptr != nullptr);
  78     CHECK_EQUAL(6, ac.size());
  79   }
  80
  81   TEST(JsonData_ConvertLinkshere)
  82   {
  83     std::string testdata =
  84         R"#({"batchcomplete":true,"query":{"pages":[{"pageid":2,"ns":0,"title":"Eins","linkshere":[{"ns":0,"title":"Zw\u00f6lf"},{"ns":0,"title":"Dreizehn"},{"ns":0,"title":"Ens"}]}]},"limits":{"linkshere":500}})#";
  85     WikimediaJsonToArticleConverter conv;
  86     ArticleCollection ac;
  87     auto cont = conv.convert(testdata, ac);
  88
  89     // exp: zwölf, dreizehn, ens --> eins
  90     CHECK(WikimediaJsonToArticleConverter::ContinuationStatus::
  91               ConversionCompleted == cont);
  92     CHECK_EQUAL(3, CollectionUtils::countAnalyzedArticles(ac));
  93     CHECK_EQUAL(4, ac.size());
  94
  95     auto ptr = CollectionUtils::get(ac, "Ens");
  96     REQUIRE CHECK(ptr != nullptr);
  97     CHECK(ptr->analyzed());
  98     ptr = CollectionUtils::get(ac, "Dreizehn");
  99     REQUIRE CHECK(ptr != nullptr);
 100     CHECK(ptr->analyzed());
 101     // better be save than sorry
 102     ptr = CollectionUtils::get(ac, "Zw\u00f6lf");
 103     REQUIRE CHECK(ptr != nullptr);
 104     CHECK(ptr->analyzed());
 105     ptr = CollectionUtils::get(ac, "Eins");
 106     REQUIRE CHECK(ptr != nullptr);
 107     CHECK(!ptr->analyzed());
 108   }
 109 }