cleanup the API, and added more tests

Author: Ko van der Sloot

include/frog/FrogAPI.h

@@ -108,11 +108,15 @@ class FrogAPI {
   ~FrogAPI();
   static std::string defaultConfigDir( const std::string& ="" );
   static std::string defaultConfigFile( const std::string& ="" );
-  void FrogFile( const std::string&, std::ostream&, const std::string& );
+  folia::Document *FrogFile( const std::string&, std::ostream& );
   void FrogServer( Sockets::ServerSocket &conn );
   void FrogInteractive();
   frog_data frog_sentence( std::vector<Tokenizer::Token>&,
 			   const size_t );
+  std::string Frogtostring( const std::string& );
+  std::string Frogtostringfromfile( const std::string& );
+
+ private:
   folia::Document *run_folia_engine( const std::string&,
 				     std::ostream& );
   folia::Document *run_text_engine( const std::string&,
@@ -122,10 +126,6 @@ class FrogAPI {
   folia::FoliaElement *append_to_folia( folia::FoliaElement *,
 					const frog_data&,
 					unsigned int& ) const;
-  std::string Frogtostring( const std::string& );
-  std::string Frogtostringfromfile( const std::string& );
-
- private:
   void add_ner_result( folia::Sentence *,
 		       const frog_data&,
 		       const std::vector<folia::Word*>& ) const;

src/Frog.cxx

@@ -618,7 +618,7 @@ int main(int argc, char *argv[]) {
         outS = new ofstream( outputFileName );
       }
       if ( fileNames.size() > 1 ){
-	LOG << "start procesessing " << fileNames.size() << " files..." << endl;
+	LOG << "start processing " << fileNames.size() << " files..." << endl;
       }
       for ( auto const& name : fileNames ){
 	string testName = testDirName + name;
@@ -688,12 +688,11 @@ int main(int argc, char *argv[]) {
 	  LOG << "running some extra Frog tests...." << endl;
 	  if ( testName.find( ".xml" ) != string::npos ){
 	    options.doXMLin = true;
+	    options.doXMLout = true;
 	  }
 	  else {
 	    options.doXMLin = false;
-	  }
-	  if ( !xmlOutName.empty() && xmlOutName.find( ".xml" ) != string::npos ){
-	    options.doXMLout = true;
+	    options.doXMLout = false;
 	  }
 	  LOG << "Start test: " << testName << endl;
 	  stringstream ss;
@@ -715,16 +714,84 @@ int main(int argc, char *argv[]) {
 	    LOG << "test OK!" << endl;
 	  }
 	  LOG << "Done with:" << testName << endl;
+
+	  //
+	  // also test FoLiA in en text out
+	  {
+	    if ( testName.find( ".xml" ) != string::npos ){
+	      options.doXMLin = true;
+	      options.doXMLout = false;
+	    }
+	    LOG << "Start test: " << testName << endl;
+	    stringstream ss;
+	    ifstream is( testName );
+	    string line;
+	    while ( getline( is, line ) ){
+	      ss << line << endl;
+	    }
+	    string s1 = frog.Frogtostring( ss.str() );
+	    *outS << "STRING 1 " << endl;
+	    *outS << s1 << endl;
+	    string s2 = frog.Frogtostringfromfile( testName );
+	    *outS << "STRING 2 " << endl;
+	    *outS << s2 << endl;
+	    if ( s1 != s2 ){
+	      LOG << "FAILED test :" << testName << endl;
+	    }
+	    else {
+	      LOG << "test OK!" << endl;
+	    }
+	    LOG << "Done with:" << testName << endl;
+	  }
+	  //
+	  // and even text in and FoLiA out
+	  {
+	    if ( testName.find( ".xml" ) == string::npos ){
+	      options.doXMLin = false;
+	      options.doXMLout = true;
+	    }
+	    LOG << "Start test: " << testName << endl;
+	    stringstream ss;
+	    ifstream is( testName );
+	    string line;
+	    while ( getline( is, line ) ){
+	      ss << line << endl;
+	    }
+	    options.docid = "test";
+	    string s1 = frog.Frogtostring( ss.str() );
+	    *outS << "STRING 1 " << endl;
+	    *outS << s1 << endl;
+	    string s2 = frog.Frogtostringfromfile( testName );
+	    *outS << "STRING 2 " << endl;
+	    *outS << s2 << endl;
+	    if ( s1 != s2 ){
+	      LOG << "FAILED test :" << testName << endl;
+	    }
+	    else {
+	      LOG << "test OK!" << endl;
+	    }
+	    LOG << "Done with:" << testName << endl;
+	  }
 	}
 	else {
+	  folia::Document *result = 0;
 	  try {
-	    frog.FrogFile( testName, *outS, xmlOutName );
+	    result = frog.FrogFile( testName, *outS );
 	  }
 	  catch ( exception& e ){
 	    LOG << "problem frogging: " << name << endl
 		<< e.what() << endl;
 	    continue;
 	  }
+	  if ( !xmlOutName.empty() ){
+	    if ( !result ){
+	      LOG << "FAILED to create FoLiA??" << endl;
+	    }
+	    else {
+	      result->save( xmlOutName );
+	      LOG << "FoLiA stored in " << xmlOutName << endl;
+	    }
+	  }
 	  if ( !outName.empty() ){
 	    LOG << "results stored in " << outName << endl;
 	    if ( outS != &cout ){

src/FrogAPI.cxx

@@ -969,24 +969,32 @@ string FrogAPI::Frogtostring( const string& s ){
   if ( s.empty() ){
     return s;
   }
-  options.hide_timers = true;
   string tmp_file = tmpnam(0);
   ofstream os( tmp_file );
   os << s << endl;
   os.close();
-  stringstream ss;
-  FrogFile( tmp_file, ss, "" );
-  return ss.str();
+  return Frogtostringfromfile( tmp_file );
 }
 
-string FrogAPI::Frogtostringfromfile( const string& name ){
+string FrogAPI::Frogtostringfromfile( const string& infilename ){
   /// Parse a file, Frog it and return the result as a string.
   /// @name: The filename.
   /// @return the results of frogging. Depending of the current frog settings
   /// the inputfile can be interpreted as XML, an the ouput will be XML or
   /// tab separated
+  options.hide_timers = true;
+  bool old_val = options.noStdOut;
   stringstream ss;
-  FrogFile( name, ss, "" );
+  if ( options.doXMLout ){
+    options.noStdOut = true;
+  }
+  folia::Document *result = FrogFile( infilename, ss );
+  options.noStdOut = old_val;
+  if ( result ){
+    result->set_kanon( options.doKanon );
+    ss << result;
+    delete result;
+  }
   return ss.str();
 }
 
@@ -1665,7 +1673,7 @@ folia::Document *FrogAPI::run_folia_engine( const string& infilename,
     return 0;
   }
   if ( options.doXMLout ){
-    return engine.doc(true);
+    return engine.doc(true); //disconnect from the engine!
   }
   return 0;
 }
@@ -1707,43 +1715,23 @@ folia::Document *FrogAPI::run_text_engine( const string& infilename,
   return doc;
 }
 
-void FrogAPI::FrogFile( const string& infilename,
-			ostream& os,
-			const string& xmlOutF ) {
+folia::Document *FrogAPI::FrogFile( const string& infilename,
+				    ostream& os ){
+  folia::Document *result = 0;
   bool xml_in = options.doXMLin;
   if ( TiCC::match_back( infilename, ".xml.gz" )
        || TiCC::match_back( infilename, ".xml.bz2" )
        || TiCC::match_back( infilename, ".xml" ) ){
     // auto detect (compressed) xml.
     xml_in = true;
   }
-  string xmlOutFile = xmlOutF;
   timers.reset();
-  folia::Document *result = 0;
   if ( xml_in ){
-    // when the inputfile is .bz2 or .gz, we use the same compression on output
-    if ( !xmlOutFile.empty() ){
-      if ( TiCC::match_back( infilename, ".gz" ) ){
-	if ( !TiCC::match_back( xmlOutFile, ".gz" ) ){
-	  xmlOutFile += ".gz";
-	}
-      }
-      else if ( TiCC::match_back( infilename, ".bz2" ) ){
-	if ( !TiCC::match_back( xmlOutFile, ".bz2" ) ){
-	  xmlOutFile += ".bz2";
-	}
-      }
-    }
     result = run_folia_engine( infilename, os );
   }
   else {
     result = run_text_engine( infilename, os );
   }
-  if ( result ){
-    result->save( xmlOutFile, options.doKanon );
-    LOG << "resulting FoLiA doc saved in " << xmlOutFile << endl;
-    delete result;
-  }
   if ( !options.hide_timers ){
     LOG << "tokenisation took:  " << timers.tokTimer << endl;
     LOG << "CGN tagging took:   " << timers.tagTimer << endl;
@@ -1772,6 +1760,7 @@ void FrogAPI::FrogFile( const string& infilename,
     }
     LOG << "Frogging in total took: " << timers.frogTimer + timers.tokTimer << endl;
   }
+  return result;
 }
 
 // the functions below here are ONLY used by TSCAN.