src/main.cpp

   1 //! \file main.cpp
   2
   3 #include <fstream>
   4 #include <iostream>
   5 #include <string>
   6
   7 #include "Article.h"
   8 #include "ArticleCollection.h"
   9 #include "ToGraphvizWriter.h"
  10 #include "WikiWalker.h"
  11 #include "version.h"
  12
  13 #include "config.h"
  14
  15 using CmdOpt = WikiWalker::CommandLineParserBase::CommandLineOptions;
  16
  17 //! limit for printing article links / contents
  18 const int printLimit = 10;
  19
  20 int main(int argc, char** argv)
  21 {
  22 #if defined(WW_USE_BOOST_PO)
  23   WikiWalker::BoostPoCommandLineParser cmdp;
  24 #elif defined(WW_USE_GETOPT)
  25   WikiWalker::GetoptCommandLineParser cmdp;
  26 #endif
  27
  28   try {
  29     cmdp.parse(argc, argv);
  30   } catch(std::exception& e) {
  31     std::cerr << std::endl << e.what() << std::endl;
  32     cmdp.printHelp();
  33     return -1;
  34   }
  35
  36   if(cmdp.hasSet(CmdOpt::Version)) {
  37     std::cout << "WikiWalker, version " << _WW_VERSION << std::endl;
  38     return 0;
  39   }
  40
  41   if(cmdp.hasSet(CmdOpt::Help)) {
  42     cmdp.printHelp();
  43     return 0;
  44   }
  45
  46   bool isUrlSet       = cmdp.hasSet(CmdOpt::URL);
  47   bool isCacheSet     = cmdp.hasSet(CmdOpt::JsonCache);
  48   bool isDotSet       = cmdp.hasSet(CmdOpt::DotOut);
  49   bool isDeepSet      = cmdp.hasSet(CmdOpt::FetchDeep);
  50   bool validRunConfig = isUrlSet || (isDotSet && isCacheSet);
  51
  52   if(!validRunConfig) {
  53     std::cerr << "Must either specify at least URL, "
  54               << "or dot and cache file." << std::endl;
  55     cmdp.printHelp();
  56     return 1;
  57   }
  58
  59   if(isUrlSet && isDeepSet && !isCacheSet) {
  60     std::cerr << "Please specify a cache file when using \"deep\" option"
  61               << std::endl;
  62     cmdp.printHelp();
  63     return 1;
  64   }
  65
  66   bool read_failed = false;
  67   WikiWalker::WikiWalker w;
  68
  69   if(isCacheSet) {
  70     try {
  71       std::string cachefile = cmdp.getValue(CmdOpt::JsonCache);
  72       w.readCache(cachefile);
  73     } catch(std::exception& e) {
  74       std::cout << e.what() << std::endl;
  75       read_failed = true;
  76     }
  77   }
  78
  79   if(isDeepSet) {
  80     w.deep(true);
  81   }
  82
  83   if(isUrlSet) {
  84     try {
  85       std::string url = cmdp.getValue(CmdOpt::URL);
  86       w.skipSslVerification(
  87           cmdp.hasSet(WikiWalker::CommandLineParserBase::CommandLineOptions::
  88                           SkipSslVerification));
  89       w.start(url);
  90     } catch(std::exception& e) {
  91       std::cout << "Error: " << e.what() << std::endl;
  92       return -1;
  93     }
  94   }
  95
  96   if(isCacheSet) {
  97     std::string cachefile = cmdp.getValue(CmdOpt::JsonCache);
  98     if(read_failed) {
  99       cachefile.append("_");
 100       std::cout << "Reading from cache failed, write to " << cachefile
 101                 << std::endl;
 102     }
 103     try {
 104       w.writeCache(cachefile);
 105     } catch(std::exception& e) {
 106       std::cout << "Error: " << e.what() << std::endl;
 107     }
 108   }
 109
 110   if(isDotSet) {
 111     const WikiWalker::CollectionUtils::ArticleCollection& ac = w.collection();
 112     std::string outfile = cmdp.getValue(CmdOpt::DotOut);
 113     WikiWalker::ToGraphvizWriter tgw;
 114     std::ofstream file(outfile, std::ios::trunc | std::ios::out);
 115
 116     if(file.fail()) {
 117       std::cerr << "Error opening dot out file for writing" << std::endl;
 118     } else {
 119       tgw.output(ac, file);
 120       file.flush();
 121
 122       if(file.bad() || file.fail()) {
 123         std::cerr << "Error during writing dot out file." << std::endl;
 124       }
 125
 126       file.close();
 127     }
 128   }
 129
 130   size_t numArt =
 131       WikiWalker::CollectionUtils::countAnalyzedArticles(w.collection());
 132   if(numArt > 10) {
 133     std::cout << "There are " << numArt << " analyzed articles."
 134               << " Not printing them. (Limit: " << printLimit << ")."
 135               << std::endl;
 136   } else {
 137     for(auto& a : w.collection()) {
 138       auto& art = a.second;
 139       if(art->marked()) {
 140         std::cout << "Article " << a.first << " is invalid or doesn't exist"
 141                   << std::endl;
 142       } else if(art->analyzed()) {
 143         std::cout << "Article " << a.first << " has " << art->countLinks()
 144                   << " links" << std::endl;
 145       }
 146     }
 147   }
 148
 149   return 0;
 150 }