Examples with FeatureMap gate.FeatureMap used on opensource projects

Example 11 with FeatureMap

use of gate.FeatureMap in project gate-core by GateNLP.

the class ParameterList method getDefaults.

// getInitimeDefaults()
/**
 * Get default parameter value set. Where more than one default
 * is possible amongst disjunctive parameters, only the first will be set.
 * To check if the default set is comprehensive,
 * use <TT>isFullyDefaulted()</TT>.
 * @see #isFullyDefaulted()
 */
public FeatureMap getDefaults(List<List<Parameter>> parameters) throws ParameterException {
    FeatureMap defaults = Factory.newFeatureMap();
    // each element of the parameters list is a list of (disjunctive) params
    Iterator<List<Parameter>> disjIter = parameters.iterator();
    // for each parameter disjunction in parameters
    disjIterLoop: while (disjIter.hasNext()) {
        // were any of this disj optional?
        boolean optional = false;
        // get an iterator for this disjunction of parameters
        List<Parameter> paramDisj = disjIter.next();
        Iterator<Parameter> paramsIter = paramDisj.iterator();
        // for each parameter in the disjunction
        while (paramsIter.hasNext()) {
            Parameter param = paramsIter.next();
            if (DEBUG)
                Out.prln("Examining " + param);
            if (!optional)
                optional = param.isOptional();
            // try and find a default value
            Object defaultValue = param.calculateDefaultValue();
            // no default found
            if (defaultValue == null) {
                // we've got at least one non-optional param unset
                if (!optional && !paramsIter.hasNext()) {
                    fullyDefaulted = false;
                }
            // valid default found - set it and continue with the next disj
            } else {
                defaults.put(param.getName(), defaultValue);
                continue disjIterLoop;
            }
        }
    // paramsIter
    }
    return defaults;
}

Example 12 with FeatureMap

use of gate.FeatureMap in project gate-core by GateNLP.

the class SerialDataStore method adopt.

// delete(lr)
/**
 * Adopt a resource for persistence.
 */
@Override
public LanguageResource adopt(LanguageResource lr) throws PersistenceException {
    // ignore security info
    // check the LR's current DS
    DataStore currentDS = lr.getDataStore();
    if (currentDS == null) {
        // an orphan - do the adoption
        LanguageResource res = lr;
        if (lr instanceof Corpus) {
            FeatureMap features1 = Factory.newFeatureMap();
            features1.put("transientSource", lr);
            try {
                // here we create the persistent LR via Factory, so it's registered
                // in GATE
                res = (LanguageResource) Factory.createResource("gate.corpora.SerialCorpusImpl", features1);
            // Here the transient corpus is not deleted from the CRI, because
            // this might not always be the desired behaviour
            // since we chose that it is for the GUI, this functionality is
            // now move to the 'Save to' action code in NameBearerHandle
            } catch (gate.creole.ResourceInstantiationException ex) {
                throw new GateRuntimeException(ex.getMessage());
            }
        }
        res.setDataStore(this);
        // let the world know
        fireResourceAdopted(new DatastoreEvent(this, DatastoreEvent.RESOURCE_ADOPTED, lr, null));
        return res;
    } else if (// adopted already here
    currentDS.equals(this))
        return lr;
    else {
        // someone else's child
        throw new PersistenceException("Can't adopt a resource which is already in a different datastore");
    }
}

Example 13 with FeatureMap

use of gate.FeatureMap in project gate-core by GateNLP.

the class CorpusBenchmarkTool method evaluateCorpus.

// generateCorpus
protected void evaluateCorpus(File fileDir, File processedDir, File markedDir, File errorDir) {
    // 1. check if we have input files and the processed Dir
    if (fileDir == null || !fileDir.exists())
        return;
    if (processedDir == null || !processedDir.exists())
        // if the user wants evaluation of marked and stored that's not possible
        if (isMarkedStored) {
            Out.prln("Cannot evaluate because no processed documents exist.");
            return;
        } else
            isMarkedClean = true;
    // create the error directory or clean it up if needed
    File errDir = null;
    if (isMoreInfoMode) {
        errDir = errorDir;
        if (errDir == null) {
            errDir = new File(currDir, ERROR_DIR_NAME);
        } else {
            // get rid of the directory, coz we wants it clean
            if (!Files.rmdir(errDir))
                Out.prln("cannot delete old error directory: " + errDir);
        }
        Out.prln("Create error directory: " + errDir + "<BR><BR>");
        errDir.mkdir();
    }
    // looked for marked texts only if the directory exists
    boolean processMarked = markedDir != null && markedDir.exists();
    if (!processMarked && (isMarkedStored || isMarkedClean)) {
        Out.prln("Cannot evaluate because no human-annotated documents exist.");
        return;
    }
    if (isMarkedStored) {
        evaluateMarkedStored(markedDir, processedDir, errDir);
        return;
    } else if (isMarkedClean) {
        evaluateMarkedClean(markedDir, fileDir, errDir);
        return;
    }
    Document persDoc = null;
    Document cleanDoc = null;
    Document markedDoc = null;
    // open the datastore and process each document
    try {
        // open the data store
        DataStore sds = Factory.openDataStore("gate.persist.SerialDataStore", processedDir.toURI().toURL().toExternalForm());
        List<String> lrIDs = sds.getLrIds("gate.corpora.DocumentImpl");
        for (int i = 0; i < lrIDs.size(); i++) {
            String docID = lrIDs.get(i);
            // read the stored document
            FeatureMap features = Factory.newFeatureMap();
            features.put(DataStore.DATASTORE_FEATURE_NAME, sds);
            features.put(DataStore.LR_ID_FEATURE_NAME, docID);
            FeatureMap hparams = Factory.newFeatureMap();
            // Gate.setHiddenAttribute(hparams, true);
            persDoc = (Document) Factory.createResource("gate.corpora.DocumentImpl", features, hparams);
            if (isMoreInfoMode) {
                StringBuffer errName = new StringBuffer(persDoc.getName());
                errName.replace(persDoc.getName().lastIndexOf("."), persDoc.getName().length(), ".err");
                Out.prln("<H2>" + "<a href=\"err/" + errName.toString() + "\">" + persDoc.getName() + "</a>" + "</H2>");
            } else
                Out.prln("<H2>" + persDoc.getName() + "</H2>");
            File cleanDocFile = new File(fileDir, persDoc.getName());
            // try reading the original document from clean
            if (!cleanDocFile.exists()) {
                Out.prln("Warning: Cannot find original document " + persDoc.getName() + " in " + fileDir);
            } else {
                FeatureMap params = Factory.newFeatureMap();
                params.put(Document.DOCUMENT_URL_PARAMETER_NAME, cleanDocFile.toURI().toURL());
                params.put(Document.DOCUMENT_ENCODING_PARAMETER_NAME, documentEncoding);
                // create the document
                cleanDoc = (Document) Factory.createResource("gate.corpora.DocumentImpl", params, hparams);
                cleanDoc.setName(persDoc.getName());
            }
            // try finding the marked document
            StringBuffer docName = new StringBuffer(persDoc.getName());
            if (!isMarkedDS) {
                docName.replace(persDoc.getName().lastIndexOf("."), docName.length(), ".xml");
                File markedDocFile = new File(markedDir, docName.toString());
                if (!processMarked || !markedDocFile.exists()) {
                    Out.prln("Warning: Cannot find human-annotated document " + markedDocFile + " in " + markedDir);
                } else {
                    FeatureMap params = Factory.newFeatureMap();
                    params.put(Document.DOCUMENT_URL_PARAMETER_NAME, markedDocFile.toURI().toURL());
                    params.put(Document.DOCUMENT_ENCODING_PARAMETER_NAME, documentEncoding);
                    // create the document
                    markedDoc = (Document) Factory.createResource("gate.corpora.DocumentImpl", params, hparams);
                    markedDoc.setName(persDoc.getName());
                }
            } else {
                // open marked from a DS
                // open the data store
                DataStore sds1 = Factory.openDataStore("gate.persist.SerialDataStore", markedDir.toURI().toURL().toExternalForm());
                List<String> lrIDs1 = sds1.getLrIds("gate.corpora.DocumentImpl");
                boolean found = false;
                int k = 0;
                // search for the marked doc with the same name
                while (k < lrIDs1.size() && !found) {
                    String docID1 = lrIDs1.get(k);
                    // read the stored document
                    FeatureMap features1 = Factory.newFeatureMap();
                    features1.put(DataStore.DATASTORE_FEATURE_NAME, sds1);
                    features1.put(DataStore.LR_ID_FEATURE_NAME, docID1);
                    Document tempDoc = (Document) Factory.createResource("gate.corpora.DocumentImpl", features1, hparams);
                    // check whether this is our doc
                    if (((String) tempDoc.getFeatures().get("gate.SourceURL")).endsWith(persDoc.getName())) {
                        found = true;
                        markedDoc = tempDoc;
                    } else
                        k++;
                }
            }
            evaluateDocuments(persDoc, cleanDoc, markedDoc, errDir);
            if (persDoc != null) {
                final gate.Document pd = persDoc;
                javax.swing.SwingUtilities.invokeLater(new Runnable() {

                    @Override
                    public void run() {
                        Factory.deleteResource(pd);
                    }
                });
            }
            if (cleanDoc != null) {
                final gate.Document cd = cleanDoc;
                javax.swing.SwingUtilities.invokeLater(new Runnable() {

                    @Override
                    public void run() {
                        Factory.deleteResource(cd);
                    }
                });
            }
            if (markedDoc != null) {
                final gate.Document md = markedDoc;
                javax.swing.SwingUtilities.invokeLater(new Runnable() {

                    @Override
                    public void run() {
                        Factory.deleteResource(md);
                    }
                });
            }
        }
        // for loop through saved docs
        sds.close();
    } catch (java.net.MalformedURLException ex) {
        throw (GateRuntimeException) new GateRuntimeException("CorpusBenchmark: " + ex.getMessage()).initCause(ex);
    } catch (PersistenceException ex1) {
        throw (GateRuntimeException) new GateRuntimeException("CorpusBenchmark: " + ex1.getMessage()).initCause(ex1);
    } catch (ResourceInstantiationException ex2) {
        throw (GateRuntimeException) new GateRuntimeException("CorpusBenchmark: " + ex2.getMessage()).initCause(ex2);
    }
}

Example 14 with FeatureMap

use of gate.FeatureMap in project gate-core by GateNLP.

the class CorpusBenchmarkTool method evaluateMarkedClean.

// evaluateMarkedStored
protected void evaluateMarkedClean(File markedDir, File cleanDir, File errDir) {
    Document persDoc = null;
    Document cleanDoc = null;
    Document markedDoc = null;
    File[] cleanDocs = cleanDir.listFiles();
    for (int i = 0; i < cleanDocs.length; i++) {
        if (!cleanDocs[i].isFile())
            continue;
        // try reading the original document from clean
        FeatureMap params = Factory.newFeatureMap();
        try {
            params.put(Document.DOCUMENT_URL_PARAMETER_NAME, cleanDocs[i].toURI().toURL());
        } catch (java.net.MalformedURLException ex) {
            Out.prln("Cannot create document from file: " + cleanDocs[i].getAbsolutePath());
            continue;
        }
        // params.put(Document.DOCUMENT_ENCODING_PARAMETER_NAME, "");
        params.put(Document.DOCUMENT_ENCODING_PARAMETER_NAME, documentEncoding);
        FeatureMap hparams = Factory.newFeatureMap();
        // create the document
        try {
            cleanDoc = (Document) Factory.createResource("gate.corpora.DocumentImpl", params, hparams, cleanDocs[i].getName());
        } catch (gate.creole.ResourceInstantiationException ex) {
            Out.prln("Cannot create document from file: " + cleanDocs[i].getAbsolutePath());
            continue;
        }
        if (isMoreInfoMode) {
            StringBuffer errName = new StringBuffer(cleanDocs[i].getName());
            errName.replace(cleanDocs[i].getName().lastIndexOf("."), cleanDocs[i].getName().length(), ".err");
            Out.prln("<H2>" + "<a href=\"err/" + errName.toString() + "\">" + cleanDocs[i].getName() + "</a>" + "</H2>");
        } else
            Out.prln("<H2>" + cleanDocs[i].getName() + "</H2>");
        // try finding the marked document
        if (!isMarkedDS) {
            StringBuffer docName = new StringBuffer(cleanDoc.getName());
            docName.replace(cleanDoc.getName().lastIndexOf("."), docName.length(), ".xml");
            File markedDocFile = new File(markedDir, docName.toString());
            if (!markedDocFile.exists()) {
                Out.prln("Warning: Cannot find human-annotated document " + markedDocFile + " in " + markedDir);
                continue;
            } else {
                params = Factory.newFeatureMap();
                try {
                    params.put(Document.DOCUMENT_URL_PARAMETER_NAME, markedDocFile.toURI().toURL());
                } catch (java.net.MalformedURLException ex) {
                    Out.prln("Cannot create document from file: " + markedDocFile.getAbsolutePath());
                    continue;
                }
                // params.put(Document.DOCUMENT_ENCODING_PARAMETER_NAME, "");
                params.put(Document.DOCUMENT_ENCODING_PARAMETER_NAME, documentEncoding);
                // create the document
                try {
                    markedDoc = (Document) Factory.createResource("gate.corpora.DocumentImpl", params, hparams, cleanDoc.getName());
                } catch (gate.creole.ResourceInstantiationException ex) {
                    Out.prln("Cannot create document from file: " + markedDocFile.getAbsolutePath());
                    continue;
                }
            }
        // if markedDoc exists
        } else {
            try {
                // open marked from a DS
                // open the data store
                DataStore sds1 = Factory.openDataStore("gate.persist.SerialDataStore", markedDir.toURI().toURL().toExternalForm());
                List<String> lrIDs1 = sds1.getLrIds("gate.corpora.DocumentImpl");
                boolean found = false;
                int k = 0;
                // search for the marked doc with the same name
                while (k < lrIDs1.size() && !found) {
                    String docID1 = lrIDs1.get(k);
                    // read the stored document
                    FeatureMap features1 = Factory.newFeatureMap();
                    features1.put(DataStore.DATASTORE_FEATURE_NAME, sds1);
                    features1.put(DataStore.LR_ID_FEATURE_NAME, docID1);
                    Document tempDoc = (Document) Factory.createResource("gate.corpora.DocumentImpl", features1, hparams);
                    // check whether this is our doc
                    if (((String) tempDoc.getFeatures().get("gate.SourceURL")).endsWith(cleanDoc.getName())) {
                        found = true;
                        markedDoc = tempDoc;
                    } else
                        k++;
                }
            } catch (java.net.MalformedURLException ex) {
                Out.prln("Error finding marked directory " + markedDir.getAbsolutePath());
            } catch (gate.persist.PersistenceException ex1) {
                Out.prln("Error opening marked as a datastore (-marked_ds specified)");
            } catch (gate.creole.ResourceInstantiationException ex2) {
                Out.prln("Error opening marked as a datastore (-marked_ds specified)");
            }
        }
        try {
            evaluateDocuments(persDoc, cleanDoc, markedDoc, errDir);
        } catch (gate.creole.ResourceInstantiationException ex) {
            ex.printStackTrace();
            Out.prln("Evaluate failed on document: " + cleanDoc.getName());
        }
        if (persDoc != null) {
            final gate.Document pd = persDoc;
            javax.swing.SwingUtilities.invokeLater(new Runnable() {

                @Override
                public void run() {
                    Factory.deleteResource(pd);
                }
            });
        }
        if (cleanDoc != null) {
            final gate.Document cd = cleanDoc;
            javax.swing.SwingUtilities.invokeLater(new Runnable() {

                @Override
                public void run() {
                    Factory.deleteResource(cd);
                }
            });
        }
        if (markedDoc != null) {
            final gate.Document md = markedDoc;
            javax.swing.SwingUtilities.invokeLater(new Runnable() {

                @Override
                public void run() {
                    Factory.deleteResource(md);
                }
            });
        }
    }
// for loop through clean docs
}

Example 15 with FeatureMap

use of gate.FeatureMap in project gate-core by GateNLP.

the class XmlDocumentHandler method startElement.

// endDocument();
/**
 * This method is called when the SAX parser encounts the beginning of an
 * XML element.
 */
/**
 * @param uri - namespace uri
 * @param localName - local, unprefixed element name
 * @param qName - fully qualified, prefixed element name
 * @param atts
 * @throws SAXException
 */
@Override
public void startElement(String uri, String localName, String qName, Attributes atts) throws SAXException {
    // call characterActions
    if (readCharacterStatus) {
        readCharacterStatus = false;
        charactersAction(new String(contentBuffer).toCharArray(), 0, contentBuffer.length());
    }
    // so far is a multiple of ELEMENTS_RATE
    if ((++elements % ELEMENTS_RATE) == 0) {
        fireStatusChangedEvent("Processed elements : " + elements);
    }
    Integer customObjectId = null;
    // Construct a SimpleFeatureMapImpl from the list of attributes
    FeatureMap fm = Factory.newFeatureMap();
    /**
     * Use localName rather than qName and add the namespace prefix and uri
     *  as features if global flag is set
     */
    String elemName = qName;
    boolean hasNSUri = (uri != null && !uri.isEmpty());
    if (deserializeNamespaceInfo && hasNSUri) {
        elemName = localName;
        StringTokenizer strToken = new StringTokenizer(qName, ":");
        if (strToken.countTokens() > 1) {
            String nsPrefix = strToken.nextToken();
            fm.put(namespaceURIFeature, uri);
            fm.put(namespacePrefixFeature, nsPrefix);
        }
    }
    // Get the name and the value of the attributes and add them to a FeaturesMAP
    for (int i = 0; i < atts.getLength(); i++) {
        String attName = atts.getLocalName(i);
        String attValue = atts.getValue(i);
        String attUri = atts.getURI(i);
        if (attUri != null && Gate.URI.equals(attUri)) {
            if ("gateId".equals(attName)) {
                customObjectId = Integer.parseInt(attValue);
            }
            // End if
            if ("annotMaxId".equals(attName)) {
                customObjectsId = Integer.parseInt(attValue);
            }
            // End if
            if ("matches".equals(attName)) {
                StringTokenizer strTokenizer = new StringTokenizer(attValue, ";");
                List<Integer> list = new ArrayList<Integer>();
                // Take all tokens,create Integers and add them to the list
                while (strTokenizer.hasMoreTokens()) {
                    String token = strTokenizer.nextToken();
                    list.add(Integer.valueOf(token));
                }
                // End while
                fm.put(attName, list);
            }
        // End if
        } else {
            fm.put(atts.getQName(i), attValue);
        }
    // End if
    }
    // End for
    // create the START index of the annotation
    Long startIndex = Long.valueOf(tmpDocContent.length());
    // initialy the Start index is equal with End index
    CustomObject obj = new CustomObject(customObjectId, elemName, fm, startIndex, startIndex);
    // put this object into the stack
    stack.push(obj);
}