src/WikiWalker.cpp

   1 //! \file WikiWalker.cpp
   2
   3 #include "WikiWalker.h"
   4
   5 #include <cassert>
   6 #include <fstream>
   7 #include <iostream>
   8
   9 #include "LUrlParser.h"
  10
  11 #include "JsonSerializer.h"
  12 #include "StringUtils.h"
  13 #include "WalkerException.h"
  14 #include "WikimediaApi.h"
  15
  16 // since the class is names like the namespace, this is a bit awkward...
  17
  18 namespace WikiWalker
  19 {
  20   void WikiWalker::start(const std::string& url)
  21   {
  22     auto info = WikimediaApiUtils::parseArticleUrl(url);
  23     WikimediaApi wapi(info.apiBaseUrl);
  24
  25     wapi.fetchForwardLinks(info.articleTitle,
  26                            WikimediaApi::WikimediaGenerator::NoGenerator,
  27                            articleSet_);
  28
  29     if(fetchGenerator_) {
  30       wapi.fetchForwardLinks(
  31           info.articleTitle,
  32           WikimediaApi::WikimediaGenerator::ForwardLinkGenerator,
  33           articleSet_);
  34     }
  35   }
  36
  37   void WikiWalker::readCache(const std::string& cacheFile)
  38   {
  39     JsonSerializer jser;
  40     std::ifstream cache(cacheFile);
  41
  42     // assumption: having write-only access to a file is so rare that I don't
  43     // care also, currently the file is used for both read and write, so
  44     // initially it won't exist.
  45     if(!cache.is_open()) {
  46       return;
  47     }
  48
  49     jser.deserialize(articleSet_, cache);
  50
  51     // doesn't work anymore since jsoncpp class changed
  52     // assert(cache.eof());
  53
  54     if(cache.fail()) {
  55       cache.close();
  56       throw WalkerException("Error reading from file");
  57     }
  58   }
  59
  60   void WikiWalker::writeCache(const std::string& cacheFile)
  61   {
  62     JsonSerializer w;
  63
  64     std::ofstream cache(cacheFile, std::ios::trunc);
  65
  66     if(!cache.is_open()) {
  67       throw WalkerException("Error writing to cache file. Check permissions.");
  68     }
  69
  70     w.serialize(articleSet_, cache);
  71
  72     if(cache.fail() || cache.bad()) {
  73       cache.close();
  74       throw WalkerException("I/O eception when writing to cache file");
  75     }
  76
  77     cache.flush();
  78     cache.close();
  79   }
  80 }  // namespace WikiWalker