/**
* The contents of this file are subject to the license and copyright
* detailed in the LICENSE and NOTICE files at the root of the source
* tree and available online at
*
* http://www.dspace.org/license/
*/
package org.dspace.text.filter;
/**
* Abstract class for implementing initial article word filters
* Allows you to create new classes with their own rules for mapping
* languages to article word lists.
*
* @author Graham Triggs
*/
public abstract class InitialArticleWord implements TextFilter
{
/**
* When no language is passed, use null and let implementation decide what to do
*/
@Override
public String filter(String str)
{
return filter(str, null);
}
/**
* Do an initial definite/indefinite article filter on the passed string.
* On matching an initial word, can strip or move to the end, depending on the
* configuration of the implementing class.
*
* @param str The string to parse
* @param lang The language of the passed string
* @return String The filtered string
*/
@Override
public String filter(String str, String lang)
{
// Get the list of article words for this language
String[] articleWordArr = getArticleWords(lang);
// If we have an article word array, process the string
if (articleWordArr != null && articleWordArr.length > 0)
{
String initialArticleWord = null;
int curPos = 0;
int initialStart = -1;
int initialEnd = -1;
// Iterate through the characters until we find something significant, or hit the end
while (initialEnd < 0 && curPos < str.length())
{
// Have we found a significant character
if (Character.isLetterOrDigit(str.charAt(curPos)))
{
// Mark this as the cut point for the initial word
initialStart = curPos;
// Loop through the article words looking for a match
for (int idx = 0; initialEnd < 0 && idx < articleWordArr.length; idx++)
{
// Extract a fragment from the string to test
// Must be same length as the article word
if (idx > 1 && initialArticleWord != null)
{
// Only need to do so if we haven't already got one
// of the right length
if (initialArticleWord.length() != articleWordArr[idx].length())
{
initialArticleWord = extractText(str, curPos, articleWordArr[idx].length());
}
}
else
{
initialArticleWord = extractText(str, curPos, articleWordArr[idx].length());
}
// Does the fragment match an article word?
if (initialArticleWord!= null && initialArticleWord.equalsIgnoreCase(articleWordArr[idx]))
{
// Check to see if the next character in the source
// is a whitespace
boolean isNextWhitespace = Character.isWhitespace(
str.charAt(curPos + articleWordArr[idx].length())
);
// Check to see if the last character of the article word is a letter or digit
boolean endsLetterOrDigit = Character.isLetterOrDigit(initialArticleWord.charAt(initialArticleWord.length() - 1));
// If the last character of the article word is a letter or digit,
// then it must be followed by whitespace, if not, it can be anything
// Setting endPos signifies that we have found an article word
if (endsLetterOrDigit && isNextWhitespace)
{
initialEnd = curPos + initialArticleWord.length();
}
else if (!endsLetterOrDigit)
{
initialEnd = curPos + initialArticleWord.length();
}
}
}
// Quit the loop, as we have a significant character
break;
}
// Keep going
curPos++;
}
// If endPos is positive, then we've found an article word
if (initialEnd > 0)
{
// Find a cut point in the source string, removing any whitespace after the article word
int cutPos = initialEnd;
while (cutPos < str.length() && Character.isWhitespace(str.charAt(cutPos)))
{
cutPos++;
}
// Are we stripping the article word?
if (stripInitialArticle)
{
// Yes, simply return everything after the cut
return str.substring(cutPos);
}
else
{
// No - move the initial article word to the end
return new StringBuffer(str.substring(cutPos))
.append(wordSeparator)
.append(str.substring(initialStart, initialEnd))
.toString();
}
}
}
// Didn't do any processing, or didn't find an initial article word
// Return the original string
return str;
}
protected InitialArticleWord(boolean stripWord)
{
stripInitialArticle = stripWord;
}
protected InitialArticleWord()
{
stripInitialArticle = false;
}
/**
* Abstract method to get the list of words to use in the initial word filter
*
* @param lang The language to retrieve article words for
* @return An array of definite/indefinite article words
*/
protected abstract String[] getArticleWords(String lang);
// Separator to use when appending article to end
private String wordSeparator = ", ";
// Flag to signify initial article word should be removed
// If false, then the initial article word is appended to the end
private boolean stripInitialArticle = false;
/**
* Helper method to extract text from a string.
* Ensures that there is significant data (ie. non-whitespace)
* after the segment requested.
*
* @param str
* @param pos
* @param len
* @return
*/
private String extractText(String str, int pos, int len)
{
int testPos = pos + len;
while (testPos < str.length() && Character.isWhitespace(str.charAt(testPos)))
{
testPos++;
}
if (testPos < str.length())
{
return str.substring(pos, pos + len);
}
return null;
}
}