Project

General

Profile

« Previous | Next » 

Revision 56610

Added by Claudio Atzori over 5 years ago

do not log or propagate erroneous and potentially huge xml records. They might clog the information system

View differences:

modules/dnet-modular-collector-service/trunk/src/main/java/eu/dnetlib/data/collector/plugins/oaisets/OaiSetsIterator.java
97 97 
		try {
98 98 
			doc = reader.read(new StringReader(xml));
99 99 
		} catch (DocumentException e) {
100 
			log.warn("Error parsing xml, I try to clean it: " + xml, e);
100 
			log.warn("Error parsing xml, I try to clean it", e);
101 101 
			final String cleaned = XmlCleaner.cleanAllEntities(xml);
102 102 
			try {
103 103 
				doc = reader.read(new StringReader(cleaned));
104 104 
			} catch (DocumentException e1) {
105 
				throw new CollectorServiceException("Error parsing cleaned document:" + cleaned, e1);
105 
				throw new CollectorServiceException("Error parsing cleaned document", e1);
106 106 
			}
107 107 
		}
108 108 

  		




  ......




  126
  126
  
    
		}


  		




  127
  127
  
    
		if (sawAllSets) {


  		




  128
  128
  
    
			log.warn("URL " + baseUrl + " keeps returning the same OAI sets. Please contact the repo admin.");


  		




  129
  
  
    
			System.out.println("URL " + baseUrl + " keeps returning the same OAI sets. Please contact the repo admin.");


  		




  130
  129
  
    
			return null;


  		




  131
  130
  
    
		} else return doc.valueOf("//*[local-name()='resumptionToken']");


  		




  132
  131
  
    
	}


  		












  

    
      modules/dnet-modular-collector-service/trunk/src/main/java/eu/dnetlib/data/collector/plugins/oai/OaiIterator.java
    
  





  134
  134
  
    
        try {


  		




  135
  135
  
    
            doc = reader.read(new StringReader(xml));


  		




  136
  136
  
    
        } catch (DocumentException e) {


  		




  137
  
  
    
            log.warn("Error parsing xml, I try to clean it: " + xml, e);


  		




  
  137
  
    
            log.warn("Error parsing xml, I try to clean it", e);


  		




  138
  138
  
    
            final String cleaned = XmlCleaner.cleanAllEntities(xml);


  		




  139
  139
  
    
            try {


  		




  140
  140
  
    
                doc = reader.read(new StringReader(cleaned));


  		




  141
  141
  
    
            } catch (DocumentException e1) {


  		




  142
  142
  
    
                final String resumptionToken = extractResumptionToken(xml);


  		




  143
  143
  
    
                if (resumptionToken == null)


  		




  144
  
  
    
                    throw new CollectorServiceException("Error parsing cleaned document:" + cleaned, e1);


  		




  
  144
  
    
                    throw new CollectorServiceException("Error parsing cleaned document", e1);


  		




  145
  145
  
    
                return resumptionToken;


  		




  146
  146
  
    
            }


  		




  147
  147
  
    
        }


  		












Also available in:  Unified diff