public static void main(String... args) throws IOException {
if (args == null || args.length < 5) {
System.out.printf(
"usage: java %s fichier.dgs outputPrefix imageType res policy%n",
FileSinkImages.class.getName());
System.exit(0);
}
FileSourceDGS dgs = new FileSourceDGS();
FileSinkImages fsi =
new FileSinkImages(
args[1],
OutputType.valueOf(args[2]),
Resolutions.valueOf(args[3]),
OutputPolicy.valueOf(args[4]));
dgs.addSink(fsi);
if (args.length > 5) fsi.addLogo(args[5], 0, 0);
fsi.setHighQuality();
fsi.setStyleSheet(
"graph { padding: 50px; fill-color: black; }"
+ "node { stroke-mode: plain; stroke-color: #3d5689,#639330,#8d4180,#97872f,#9c4432; stroke-width: 2px; fill-mode: dyn-plain; fill-color: #5782db,#90dd3e,#e069cb,#e0ce69,#e07c69; }"
+ "edge { fill-color: white; }");
boolean next = true;
dgs.begin(args[0]);
while (next) next = dgs.nextStep();
dgs.end();
}
/** @param args */
public static void main(String[] args) {
String docPath = null;
for (int i = 0; i < args.length; i++) { // iterate over cli parameter tokens
if (args[i].startsWith("-")) { // assume we found a switch
// get the relevant enum
CLISwitch sw = CLISwitch.getEnumFromSwitch(args[i]);
if (sw == null) { // unsupported CLI switch
logger.log(Level.WARNING, "Unsupported switch: " + args[i] + ". Quitting.");
System.exit(-1);
}
if (sw.getHasFollowingValue()) { // handle values for switches
if (args.length > i + 1
&& !args[i + 1].startsWith(
"-")) { // we still have an array index after this one and it's not a switch
sw.setValue(args[++i]);
} else { // value is missing or malformed
logger.log(
Level.WARNING, "Invalid or missing parameter after " + args[i] + ". Quitting.");
System.exit(-1);
}
} else { // activate the value-less switches
sw.setValue(null);
}
} else { // assume we found the document's path/name
docPath = args[i];
}
}
// display help dialog if HELP-switch is given
if (CLISwitch.HELP.getIsActive()) {
printHelp();
System.exit(0);
}
// start off with the verbosity recognition -- lots of the other
// stuff can be skipped if this is set too high
if (CLISwitch.VERBOSITY2.getIsActive()) {
logger.setLevel(Level.ALL);
logger.log(Level.INFO, "Verbosity: '-vv'; Logging level set to ALL.");
// output the found language resource folders
String languagesList = "";
for (String language : ResourceScanner.getInstance().getDetectedResourceFolders()) {
languagesList += System.getProperty("line.separator") + "- " + language;
}
logger.log(Level.INFO, "Listing detected language folders:" + languagesList);
} else if (CLISwitch.VERBOSITY.getIsActive()) {
logger.setLevel(Level.INFO);
logger.log(Level.INFO, "Verbosity: '-v'; Logging level set to INFO and above.");
} else {
logger.setLevel(Level.WARNING);
logger.log(
Level.INFO,
"Verbosity -v/-vv NOT FOUND OR RECOGNIZED; Logging level set to WARNING and above.");
}
// Check input encoding
String encodingType = null;
if (CLISwitch.ENCODING.getIsActive()) {
encodingType = CLISwitch.ENCODING.getValue().toString();
logger.log(Level.INFO, "Encoding '-e': " + encodingType);
} else {
// Encoding type not found
encodingType = CLISwitch.ENCODING.getValue().toString();
logger.log(Level.INFO, "Encoding '-e': NOT FOUND OR RECOGNIZED; set to 'UTF-8'");
}
// Check output format
OutputType outputType = null;
if (CLISwitch.OUTPUTTYPE.getIsActive()) {
outputType = OutputType.valueOf(CLISwitch.OUTPUTTYPE.getValue().toString().toUpperCase());
logger.log(Level.INFO, "Output '-o': " + outputType.toString().toUpperCase());
} else {
// Output type not found
outputType = (OutputType) CLISwitch.OUTPUTTYPE.getValue();
logger.log(
Level.INFO,
"Output '-o': NOT FOUND OR RECOGNIZED; set to " + outputType.toString().toUpperCase());
}
// Check language
Language language = null;
if (CLISwitch.LANGUAGE.getIsActive()) {
language = Language.getLanguageFromString((String) CLISwitch.LANGUAGE.getValue());
if (language == Language.WILDCARD
&& !ResourceScanner.getInstance()
.getDetectedResourceFolders()
.contains(language.getName())) {
logger.log(
Level.SEVERE,
"Language '-l': " + CLISwitch.LANGUAGE.getValue() + " NOT RECOGNIZED; aborting.");
printHelp();
System.exit(-1);
} else {
logger.log(Level.INFO, "Language '-l': " + language.getName());
}
} else {
// Language not found
language = Language.getLanguageFromString((String) CLISwitch.LANGUAGE.getValue());
logger.log(
Level.INFO, "Language '-l': NOT FOUND; set to " + language.toString().toUpperCase());
}
// Check type
DocumentType type = null;
if (CLISwitch.DOCTYPE.getIsActive()) {
try {
if (CLISwitch.DOCTYPE
.getValue()
.equals("narrative")) { // redirect "narrative" to "narratives"
CLISwitch.DOCTYPE.setValue("narratives");
}
type = DocumentType.valueOf(CLISwitch.DOCTYPE.getValue().toString().toUpperCase());
} catch (IllegalArgumentException e) {
logger.log(
Level.WARNING,
"Type '-t': NOT RECOGNIZED. These are the available options: "
+ Arrays.asList(DocumentType.values()));
System.exit(-1);
}
logger.log(Level.INFO, "Type '-t': " + type.toString().toUpperCase());
} else {
// Type not found
type = (DocumentType) CLISwitch.DOCTYPE.getValue();
logger.log(Level.INFO, "Type '-t': NOT FOUND; set to " + type.toString().toUpperCase());
}
// Check document creation time
Date dct = null;
if (CLISwitch.DCT.getIsActive()) {
try {
DateFormat formatter = new SimpleDateFormat("yyyy-MM-dd");
dct = formatter.parse(CLISwitch.DCT.getValue().toString());
logger.log(Level.INFO, "Document Creation Time '-dct': " + dct.toString());
} catch (Exception e) {
// DCT was not parseable
logger.log(Level.WARNING, "Document Creation Time '-dct': NOT RECOGNIZED. Quitting.");
printHelp();
System.exit(-1);
}
} else {
if ((type == DocumentType.NEWS) || (type == DocumentType.COLLOQUIAL)) {
// Dct needed
dct = (Date) CLISwitch.DCT.getValue();
logger.log(
Level.INFO,
"Document Creation Time '-dct': NOT FOUND; set to local date ("
+ dct.toString()
+ ").");
} else {
logger.log(Level.INFO, "Document Creation Time '-dct': NOT FOUND; skipping.");
}
}
// Handle locale switch
String locale = (String) CLISwitch.LOCALE.getValue();
Locale myLocale = null;
if (CLISwitch.LOCALE.getIsActive()) {
// check if the requested locale is available
for (Locale l : Locale.getAvailableLocales()) {
if (l.toString().toLowerCase().equals(locale.toLowerCase())) myLocale = l;
}
try {
Locale.setDefault(myLocale); // try to set the locale
logger.log(Level.INFO, "Locale '-locale': " + myLocale.toString());
} catch (Exception e) { // if the above fails, spit out error message and available locales
logger.log(
Level.WARNING,
"Supplied locale parameter couldn't be resolved to a working locale. Try one of these:");
logger.log(
Level.WARNING,
Arrays.asList(Locale.getAvailableLocales()).toString()); // list available locales
printHelp();
System.exit(-1);
}
} else {
// no -locale parameter supplied: just show default locale
logger.log(
Level.INFO,
"Locale '-locale': NOT FOUND, set to environment locale: "
+ Locale.getDefault().toString());
}
// Read configuration from file
String configPath = CLISwitch.CONFIGFILE.getValue().toString();
try {
logger.log(Level.INFO, "Configuration path '-c': " + configPath);
readConfigFile(configPath);
logger.log(Level.FINE, "Config initialized");
} catch (Exception e) {
e.printStackTrace();
logger.log(
Level.WARNING,
"Config could not be initialized! Please supply the -c switch or "
+ "put a config.props into this directory.");
printHelp();
System.exit(-1);
}
// Set the preprocessing POS tagger
POSTagger posTagger = null;
if (CLISwitch.POSTAGGER.getIsActive()) {
try {
posTagger = POSTagger.valueOf(CLISwitch.POSTAGGER.getValue().toString().toUpperCase());
} catch (IllegalArgumentException e) {
logger.log(
Level.WARNING,
"Given POS Tagger doesn't exist. Please specify a valid one as listed in the help.");
printHelp();
System.exit(-1);
}
logger.log(Level.INFO, "POS Tagger '-pos': " + posTagger.toString().toUpperCase());
} else {
// Type not found
posTagger = (POSTagger) CLISwitch.POSTAGGER.getValue();
logger.log(
Level.INFO,
"POS Tagger '-pos': NOT FOUND OR RECOGNIZED; set to "
+ posTagger.toString().toUpperCase());
}
// Set whether or not to use the Interval Tagger
Boolean doIntervalTagging = false;
if (CLISwitch.INTERVALS.getIsActive()) {
doIntervalTagging = CLISwitch.INTERVALS.getIsActive();
logger.log(Level.INFO, "Interval Tagger '-it': " + doIntervalTagging.toString());
} else {
logger.log(
Level.INFO,
"Interval Tagger '-it': NOT FOUND OR RECOGNIZED; set to " + doIntervalTagging.toString());
}
// make sure we have a document path
if (docPath == null) {
logger.log(Level.WARNING, "No input file given; aborting.");
printHelp();
System.exit(-1);
}
// Run HeidelTime
RandomAccessFile aFile = null;
MappedByteBuffer buffer = null;
FileChannel inChannel = null;
PrintWriter pwOut = null;
try {
logger.log(Level.INFO, "Reading document using charset: " + encodingType);
aFile = new RandomAccessFile(docPath, "r");
inChannel = aFile.getChannel();
buffer = inChannel.map(FileChannel.MapMode.READ_ONLY, 0, inChannel.size());
buffer.load();
byte[] inArr = new byte[(int) inChannel.size()];
for (int i = 0; i < buffer.limit(); i++) {
inArr[i] = buffer.get();
}
// double-newstring should not be necessary, but without this, it's not running on Windows (?)
String input = new String(new String(inArr, encodingType).getBytes("UTF-8"), "UTF-8");
HeidelTimeStandalone standalone =
new HeidelTimeStandalone(language, type, outputType, null, posTagger, doIntervalTagging);
String out = standalone.process(input, dct);
// Print output always as UTF-8
pwOut = new PrintWriter(new OutputStreamWriter(System.out, "UTF-8"));
pwOut.println(out);
} catch (Exception e) {
e.printStackTrace();
} finally {
if (pwOut != null) {
pwOut.close();
}
if (buffer != null) {
buffer.clear();
}
if (inChannel != null) {
try {
inChannel.close();
} catch (IOException e) {
}
}
if (aFile != null) {
try {
aFile.close();
} catch (IOException e) {
}
}
}
}
public static void main(String... args) throws IOException {
HashMap<Option, String> options = new HashMap<Option, String>();
LinkedList<String> others = new LinkedList<String>();
for (Option option : Option.values())
if (option.defaultValue != null) options.put(option, option.defaultValue);
if (args != null && args.length > 0) {
Pattern valueGetter = Pattern.compile("^--\\w[\\w-]*\\w?(?:=(?:\"([^\"]*)\"|([^\\s]*)))$");
for (int i = 0; i < args.length; i++) {
if (args[i].matches("^--\\w[\\w-]*\\w?(=(\"[^\"]*\"|[^\\s]*))?$")) {
boolean found = false;
for (Option option : Option.values()) {
if (args[i].startsWith("--" + option.fullopts + "=")) {
Matcher m = valueGetter.matcher(args[i]);
if (m.matches()) {
options.put(option, m.group(1) == null ? m.group(2) : m.group(1));
}
found = true;
break;
}
}
if (!found) {
System.err.printf("unknown option: %s%n", args[i].substring(0, args[i].indexOf('=')));
System.exit(1);
}
} else if (args[i].matches("^-\\w$")) {
boolean found = false;
for (Option option : Option.values()) {
if (args[i].equals("-" + option.shortopts)) {
options.put(option, args[++i]);
break;
}
}
if (!found) {
System.err.printf("unknown option: %s%n", args[i]);
System.exit(1);
}
} else {
others.addLast(args[i]);
}
}
} else {
usage();
System.exit(0);
}
LinkedList<String> errors = new LinkedList<String>();
if (others.size() == 0) {
errors.add("dgs file name missing.");
}
String imagePrefix;
OutputType outputType = null;
OutputPolicy outputPolicy = null;
Resolution resolution = null;
Quality quality = null;
String logo;
String stylesheet;
imagePrefix = options.get(Option.IMAGE_PREFIX);
try {
outputType = OutputType.valueOf(options.get(Option.IMAGE_TYPE));
} catch (IllegalArgumentException e) {
errors.add("bad image type: " + options.get(Option.IMAGE_TYPE));
}
try {
outputPolicy = OutputPolicy.valueOf(options.get(Option.OUTPUT_POLICY));
} catch (IllegalArgumentException e) {
errors.add("bad output policy: " + options.get(Option.OUTPUT_POLICY));
}
try {
quality = Quality.valueOf(options.get(Option.QUALITY));
} catch (IllegalArgumentException e) {
errors.add("bad quality: " + options.get(Option.QUALITY));
}
logo = options.get(Option.LOGO);
stylesheet = options.get(Option.STYLESHEET);
try {
resolution = Resolutions.valueOf(options.get(Option.IMAGE_RESOLUTION));
} catch (IllegalArgumentException e) {
Pattern p = Pattern.compile("^\\s*(\\d+)\\s*x\\s*(\\d+)\\s*$");
Matcher m = p.matcher(options.get(Option.IMAGE_RESOLUTION));
if (m.matches()) {
resolution =
new CustomResolution(Integer.parseInt(m.group(1)), Integer.parseInt(m.group(2)));
} else {
errors.add("bad resolution: " + options.get(Option.IMAGE_RESOLUTION));
}
}
if (stylesheet != null && stylesheet.length() < 1024) {
File test = new File(stylesheet);
if (test.exists()) {
FileReader in = new FileReader(test);
char[] buffer = new char[128];
String content = "";
while (in.ready()) {
int c = in.read(buffer, 0, 128);
content += new String(buffer, 0, c);
}
stylesheet = content;
in.close();
}
}
{
File test = new File(others.peek());
if (!test.exists()) errors.add(String.format("file \"%s\" does not exist", others.peek()));
}
if (errors.size() > 0) {
System.err.printf("error:%n");
for (String error : errors) System.err.printf("- %s%n", error);
System.exit(1);
}
FileSourceDGS dgs = new FileSourceDGS();
FileSinkImages fsi = new FileSinkImages(imagePrefix, outputType, resolution, outputPolicy);
dgs.addSink(fsi);
if (logo != null) fsi.addLogo(logo, 0, 0);
fsi.setQuality(quality);
if (stylesheet != null) fsi.setStyleSheet(stylesheet);
boolean next = true;
dgs.begin(others.get(0));
while (next) next = dgs.nextStep();
dgs.end();
}
public void setParserEvaluationTypes(final String string) throws EvalError, CorpusError {
if (string.equals("")) {
evalTypes.add(OutputType.CLUMP);
evalTypes.add(OutputType.NPS);
} else for (final String s : string.split(",")) evalTypes.add(OutputType.valueOf(s));
}
