@Override
/**
* Implementation of contains method for items for Basic File Persistence.
*
* <p>This works by building the target file name based on a field name from within the item and
* then checks if the file exists.
*
* @param item to perform check for
* @return true if the file exists (and hence the item is persisted)
* @throws PersistenceException if the contains check fails
*/
public boolean contains(Item item) throws PersistenceException {
try {
Map<String, Object> contents = item.getContents();
Set<String> keys = contents.keySet();
if (keys.isEmpty()) {
throw new PersistenceException("Provided item has no content");
}
String[] keysArray = (String[]) keys.toArray();
_uuid = ItemNameTools.getUUID(keysArray[0]);
String targetFileForCheck = this.getTargetLocation();
File checkFile = new File(targetFileForCheck);
return checkFile.exists();
} catch (Exception exc) {
throw new PersistenceException("Contains check failed due to " + exc.toString());
}
}
@Override
public List<Item> generate(File inputFile, String source)
throws GenerationException, IOException {
FileValidation.standardChecks(inputFile);
List<Item> generated = this.generate(new FileInputStream(inputFile), source);
List<Item> enriched = new Vector<Item>();
for (Item item : generated) {
// Basic File Comparitors, FILE_NAME, FILE_SIZE and FILE_MODIFIED_UTC
// TODO - check if this is enough
item.addComparitor(StandardAspects.FILE_FILENAME);
item.addComparitor(StandardAspects.FILE_SIZE_BYTES);
item.addComparitor(StandardAspects.FILE_MODIFIED_UTC);
enriched.add(AspectEnrichment.fileEnrichment(item, inputFile));
}
return enriched;
}
@Override
public List<Item> generate(String inputString, String source) throws GenerationException {
List<Item> items = new Vector<Item>();
Item item = new Item();
item = this.setMandatoryAspects(item, source);
// TODO
// Add use of utils to remove punctuation and tidy textual content
inputString = TextUtils.clean(inputString);
String[] tokens = inputString.split("[ ]");
item.addString("text_token_count", Integer.toString(tokens.length));
item.addString("text_content", inputString);
Vector<String> uniqueTokens = new Vector<String>();
for (String token : tokens) {
if (!uniqueTokens.contains(token)) uniqueTokens.add(token);
}
item.addString("text_unique_token_count", Integer.toString(uniqueTokens.size()));
StringBuffer uniques = null;
for (String token : uniqueTokens) {
if (uniques == null) {
uniques = new StringBuffer(token);
} else {
uniques.append(" " + token);
}
}
item.addString("text_unique_tokens", uniques.toString());
items.add(item);
return items;
}
@Override
public List<Item> generate(URL url, String source) throws GenerationException {
try {
HttpURLConnection connection = (HttpURLConnection) url.openConnection();
List<Item> generated = this.generate(connection.getInputStream(), source);
List<Item> enriched = new Vector<Item>();
for (Item item : generated) {
// URL basic Text, comparitors are URL and MODIFIED
// TODO - This may not work with dynamic pages, check and fix
item.addComparitor(StandardAspects.URL_URL);
item.addComparitor(StandardAspects.URL_MODIFIED_UTC);
enriched.add(AspectEnrichment.urlEnrichment(item, connection));
}
return enriched;
} catch (Exception exc) {
throw new GenerationException(exc);
}
}
@Override
public boolean persistItem(Item item, boolean overwrite) throws PersistenceException {
// Caveat - this method uses the *provided* uuid as an indicator of file location. This
// abstracts the item contents from the target storage indicator which could be an overhead,
// but it allows the system to persist vanilla items (non-qualified items) as well as fully
// qualified.
//
// Also for sake of simplicity the persister just uses String objects from the aspects - the
// non-String
// objects are discarded **BAD UTH**
//
// For fully qualified items simply call the item name utils to get the uuid and use that as
// the provided parameter to the object.
// First check parameters have been setup correctly
if (_uuid == null && _targetDirectory == null) {
throw new PersistenceException("Persister not initialised (missing parameters).");
}
// Prepare the target file
String filename = this.getTargetLocation();
// If overwrite then remove the file if it exists
if (overwrite) {
@SuppressWarnings("unused")
boolean success = this.removeItem(item);
} else {
if (this.itemExists(item)) {
throw new PersistenceException("Overwrite disabled and item is already persisted.");
}
}
// Debatable behaviour model - open file now, write aspects as extracted
try {
PrintWriter out = new PrintWriter(new FileOutputStream(filename));
// Manually map the created date as a separate field
out.print("CREATED:::" + Long.toString(item.getCreationUTC()) + "\n");
// Manually store the comparitors for re-constituting the item
List<String> comparitors = item.getComparitors();
StringBuffer compBuffer = null;
for (String comparitor : comparitors) {
if (compBuffer == null) {
compBuffer = new StringBuffer("COMPARITORS:::" + comparitor);
} else {
compBuffer.append("," + comparitor);
}
}
out.print(compBuffer.toString() + "\n");
// Now split the aspects and discard the non-java.util.String ones (for now, **BAD UTH**)
Map<String, Object> contents = item.getContents();
for (String key : contents.keySet()) {
Object value = contents.get(key);
if (value.getClass().getCanonicalName().equals("java.lang.String")) {
String payload = (String) value;
// Convert the data to store-able
String output = key + ":::" + payload + "\n";
out.print(output);
}
}
out.close();
} catch (Exception exc) {
throw new PersistenceException("File output failure due to " + exc.toString());
}
return false;
}
