Examples with EmailDocumentHandler gate.email.EmailDocumentHandler used on opensource projects

Example 1 with EmailDocumentHandler

use of gate.email.EmailDocumentHandler in project gate-core by GateNLP.

the class EmailDocumentFormat method unpackMarkup.

/**
 * Unpack the markup in the document. This converts markup from the
 * native format (e.g. EMAIL) into annotations in GATE format.
 * Uses the markupElementsMap to determine which elements to convert, and
 * what annotation type names to use.
 * It always tryes to parse te doc's content. It doesn't matter if the
 * sourceUrl is null or not.
 *
 * @param doc The gate document you want to parse.
 */
@Override
public void unpackMarkup(gate.Document doc) throws DocumentFormatException {
    if ((doc == null) || (doc.getSourceUrl() == null && doc.getContent() == null)) {
        throw new DocumentFormatException("GATE document is null or no content found. Nothing to parse!");
    }
    // End if
    setNewLineProperty(doc);
    // create an EmailDocumentHandler
    EmailDocumentHandler emailDocHandler = null;
    emailDocHandler = new gate.email.EmailDocumentHandler(doc, this.markupElementsMap, this.element2StringMap);
    StatusListener statusListener = new StatusListener() {

        @Override
        public void statusChanged(String text) {
            // this is implemented in DocumentFormat.java and inherited here
            fireStatusChanged(text);
        }
    };
    // Register a status listener with it
    emailDocHandler.addStatusListener(statusListener);
    try {
        // Call the method that creates annotations on the gate document
        emailDocHandler.annotateMessages();
        // Process the body annotations and search for paragraphs
        AnnotationSet bodyAnnotations = doc.getAnnotations(GateConstants.ORIGINAL_MARKUPS_ANNOT_SET_NAME).get("body");
        if (bodyAnnotations != null && !bodyAnnotations.isEmpty()) {
            Iterator<Annotation> iter = bodyAnnotations.iterator();
            while (iter.hasNext()) {
                Annotation a = iter.next();
                annotateParagraphs(doc, a.getStartNode().getOffset().intValue(), a.getEndNode().getOffset().intValue(), GateConstants.ORIGINAL_MARKUPS_ANNOT_SET_NAME);
            }
        // End while
        }
    // End if
    } catch (IOException e) {
        throw new DocumentFormatException("Couldn't create a buffered reader ", e);
    } catch (InvalidOffsetException e) {
        throw new DocumentFormatException(e);
    } finally {
        emailDocHandler.removeStatusListener(statusListener);
    }
// End try
}