Project

General

Profile

« Previous | Next » 

Revision 31116

Added by Marek Horst about 10 years ago

#757 introducing article type extraction along with unit test. Article type will be required for filtering out pmc duplicates and leaving only proper types

View differences:

modules/icm-iis-ingest-pmc/trunk/src/test/java/eu/dnetlib/iis/ingest/pmc/citations/PmcXmlHandlerTest.java
1 
package eu.dnetlib.iis.ingest.pmc.citations;
2 

  		




  
  3
  
    
import static org.junit.Assert.assertEquals;


  		




  
  4
  
    

  		




  
  5
  
    
import java.io.InputStream;


  		




  
  6
  
    

  		




  
  7
  
    
import javax.xml.parsers.SAXParser;


  		




  
  8
  
    
import javax.xml.parsers.SAXParserFactory;


  		




  
  9
  
    

  		




  
  10
  
    
import org.junit.Test;


  		




  
  11
  
    

  		




  
  12
  
    
/**


  		




  
  13
  
    
 * {@link PmcXmlHandler} test class.


  		




  
  14
  
    
 * @author mhorst


  		




  
  15
  
    
 *


  		




  
  16
  
    
 */


  		




  
  17
  
    
public class PmcXmlHandlerTest {


  		




  
  18
  
    

  		




  
  19
  
    
	@Test


  		




  
  20
  
    
	public void testParsing() throws Exception {


  		




  
  21
  
    
		String filePath = "/eu/dnetlib/iis/ingest/pmc/citations/data/pmc_example.xml";


  		




  
  22
  
    
		InputStream inputStream = null;


  		




  
  23
  
    
		try {


  		




  
  24
  
    
			SAXParser saxParser = SAXParserFactory.newInstance().newSAXParser();


  		




  
  25
  
    
			PmcXmlHandler pmcXmlHandler = new PmcXmlHandler();


  		




  
  26
  
    
			saxParser.parse(inputStream = PmcXmlHandler.class.getResourceAsStream(filePath), 


  		




  
  27
  
    
					pmcXmlHandler);


  		




  
  28
  
    
			assertEquals("10629213", pmcXmlHandler.getArticleId());


  		




  
  29
  
    
			assertEquals("correction", pmcXmlHandler.getArticleType());


  		




  
  30
  
    
		} finally {


  		




  
  31
  
    
			if (inputStream!=null) {


  		




  
  32
  
    
				inputStream.close();


  		




  
  33
  
    
			}


  		




  
  34
  
    
		}


  		




  
  35
  
    
	}


  		




  
  36
  
    
}


  		




  0
  37
  
    

  		












  

    
      modules/icm-iis-ingest-pmc/trunk/src/test/resources/eu/dnetlib/iis/ingest/pmc/citations/data/pmc_example.xml
    
  





  
  1
  
    
<article xmlns:xlink="http://www.w3.org/1999/xlink" article-type="correction"><front><journal-meta><journal-id journal-id-type="nlm-ta">J Cell Biol</journal-id><journal-title>The Journal of Cell Biology</journal-title><issn pub-type="ppub">0021-9525</issn><issn pub-type="epub">1540-8140</issn><publisher><publisher-name>The Rockefeller University Press</publisher-name></publisher></journal-meta><article-meta><article-id pub-id-type="pmcid">2174284</article-id><article-id pub-id-type="publisher-id">cor1</article-id><article-id pub-id-type="pmid">10629213</article-id><article-categories><subj-group subj-group-type="heading"><subject>Correction</subject></subj-group></article-categories><pub-date pub-type="ppub"><day>24</day><month>1</month><year>2000</year></pub-date><volume>148</volume><issue>2</issue><fpage>1</fpage><lpage>1</lpage><permissions><copyright-statement>&#x000a9; 2000 The Rockefeller University Press</copyright-statement><copyright-year>2000</copyright-year><copyright-holder>The Rockefeller University Press</copyright-holder></permissions><related-article related-article-type="corrected-article" vol="145" page="515" id="N0x2b4fc80N0x3b573b0" xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="10225953" ext-link-type="pubmed"/></article-meta></front><body><p><italic>Streptomyces carpinensis</italic> (Fig. 3 C) was incorrectly listed in the legend as <italic>Streptomyces antibioticus</italic>. This error does not affect the conclusions of the paper.</p></body></article>


  		




  0
  2
  
    

  		












  

    
      modules/icm-iis-ingest-pmc/trunk/src/main/java/eu/dnetlib/iis/ingest/pmc/citations/PmcXmlHandler.java
    
  





  50
  50
  
    

  		




  51
  51
  
    
//	attributes


  		




  52
  52
  
    
	private static final String ATTR_PUB_TYPE = "pub-id-type";


  		




  
  53
  
    
	private static final String ATTR_ARTICLE_TYPE = "article-type";


  		




  53
  54
  
    
	


  		




  54
  55
  
    
	private static final String PUB_ID_TYPE_PMID = "pmid";


  		




  55
  56
  
    

  		




  ......




  67
  68
  
    
	


  		




  68
  69
  
    
	private String articleIdType = null;


  		




  69
  70
  
    
	


  		




  
  71
  
    
	private String articleType = null;


  		




  
  72
  
    
	


  		




  70
  73
  
    
	boolean containsTextChild = false;


  		




  71
  74
  
    
	


  		




  72
  
  
    
	private int counter = 0;


  		




  
  75
  
    
	boolean rootElement = true;


  		




  73
  76
  
    
	


  		




  74
  77
  
    
	/**


  		




  75
  78
  
    
	 * Default constructor.


  		




  ......




  82
  85
  
    
	@Override


  		




  83
  86
  
    
	public void startDocument() throws SAXException {


  		




  84
  87
  
    
		this.parents = new Stack<String>();


  		




  85
  
  
    
		this.counter = 0;


  		




  86
  88
  
    
		clearAllFields();


  		




  87
  89
  
    
	}


  		




  88
  90
  
    

  		




  89
  91
  
    
	@Override


  		




  90
  92
  
    
	public void startElement(String uri, String localName, String qName,


  		




  91
  93
  
    
			Attributes attributes) throws SAXException {


  		




  92
  
  
    
		if (isWithinElement(qName, ELEM_ARTICLE_ID, ELEM_ARTICLE_META)) {


  		




  
  94
  
    
		if (rootElement) {


  		




  
  95
  
    
			rootElement = false;


  		




  
  96
  
    
			this.articleType = attributes.getValue(ATTR_ARTICLE_TYPE);


  		




  
  97
  
    
		} else if (isWithinElement(qName, ELEM_ARTICLE_ID, ELEM_ARTICLE_META)) {


  		




  93
  98
  
    
			this.currentValue = new StringBuilder();


  		




  94
  99
  
    
			this.articleIdType = attributes.getValue(ATTR_PUB_TYPE);


  		




  95
  100
  
    
		}


  		




  ......




  111
  116
  
    
	private void clearAllFields() {


  		




  112
  117
  
    
		this.articleId = null;


  		




  113
  118
  
    
		this.articleIdType = null;


  		




  
  119
  
    
		this.rootElement = true;


  		




  114
  120
  
    
	}


  		




  115
  121
  
    
	


  		




  116
  122
  
    
	boolean isWithinElement(String qName,


  		




  ......




  123
  129
  
    
	public void endDocument() throws SAXException {


  		




  124
  130
  
    
		parents.clear();


  		




  125
  131
  
    
		parents = null;


  		




  126
  
  
    
		log.debug("total number of processed records: " + counter);


  		




  127
  132
  
    
	}


  		




  128
  133
  
    

  		




  129
  134
  
    
	@Override


  		




  ......




  137
  142
  
    
	public String getArticleId() {


  		




  138
  143
  
    
		return articleId;


  		




  139
  144
  
    
	}


  		




  
  145
  
    

  		




  
  146
  
    
	public String getArticleType() {


  		




  
  147
  
    
		return articleType;


  		




  
  148
  
    
	}


  		




  140
  149
  
    
	


  		




  141
  150
  
    
}


  		












Also available in:  Unified diff