searchengine/cpix/cpix/src/fileparser/textfileparser.cpp
changeset 1 6f2c1c46032b
parent 0 671dee74050a
equal deleted inserted replaced
0:671dee74050a 1:6f2c1c46032b
   137                                                    DEFAULT_ENCODING),
   137                                                    DEFAULT_ENCODING),
   138                                cpix_STORE_NO | cpix_INDEX_TOKENIZED));
   138                                cpix_STORE_NO | cpix_INDEX_TOKENIZED));
   139         doc->add(newField.get());
   139         doc->add(newField.get());
   140         newField.release();
   140         newField.release();
   141         
   141         
   142         wchar_t
   142         /* determine file size. if the stream is not seekable, the size will be -1
   143             excerpt[512];
   143          * Here if the file is empty then we put the path name contained in 
   144         getExcerptOfFile(excerpt,
   144          * "_docuid" as excerpt
       
   145          */
       
   146         FILE* textFilePtr = fopen(path, "r");
       
   147         fseek(textFilePtr, 0, SEEK_END);
       
   148         long long size = ftell(textFilePtr);
       
   149         fseek(textFilePtr, 0, SEEK_SET);
       
   150         fclose(textFilePtr);
       
   151         if (size > 0) {
       
   152 			wchar_t excerpt[512];
       
   153         	getExcerptOfFile(excerpt,
   145                          path,
   154                          path,
   146                          10, // max words
   155                          10, // max words
   147                          sizeof(excerpt) / sizeof(wchar_t));
   156                          sizeof(excerpt) / sizeof(wchar_t));
   148         
   157 			doc->setExcerpt(excerpt);
   149         doc->setExcerpt(excerpt);
   158 		
   150         doc->setAppClass(TEXTAPPCLASS);
   159 		} else {
       
   160             //For empty file setting the path as excerpt 
       
   161 			doc->setExcerpt(doc->get(LCPIX_DOCUID_FIELD));
       
   162 		}
       
   163         doc->setAppClass(CONTENTAPPCLASS);
   151         doc->setMimeType(LTEXTFILE_MIMETYPE);
   164         doc->setMimeType(LTEXTFILE_MIMETYPE);
   152         // Always perform generic file processing
   165         // Always perform generic file processing
   153         GenericFileProcessor(doc,path);
   166         GenericFileProcessor(doc,path);
   154 
   167 
   155         
   168