/**
* convert a list of APF files to XML files containing in-line markup for coreference. Takes the
* following arguments:
*
* <ul>
* <li>year: the year (2002, 2003, 2004, 2005) the APF file was created, which determines its
* format and file extension
* <li>ACEdir: the directory containing the text and APF files
* <li>corefDir: the directory containing the in-line coreference files
* <li>filelist: a file containing a list of document names
* <li>showTypes: (optional) if this argument is present, entity type and subtype information is
* included with each mention
* </ul>
*/
public static void main(String[] args) throws IOException {
if (!(args.length == 4 || args.length == 5)) {
System.err.println("APFtoCorefXML requires 4 or 5 arguments:");
System.err.println(" year apf-directory output-directory filelist [showTypes]");
System.exit(1);
}
String year = args[0];
apfExtension = ".apf.xml";
AceDocument.ace2004 = false;
AceDocument.ace2005 = false;
if (year.equals("2002")) {
apfExtension = ".sgm.tmx.rdc.xml";
} else if (year.equals("2003")) {
} else if (year.equals("2004")) {
AceDocument.ace2004 = true;
} else if (year.equals("2005")) {
AceDocument.ace2004 = true;
AceDocument.ace2005 = true;
} else {
System.err.println("Invalid year: must be 2002-2005");
System.exit(1);
}
ACEdir = args[1];
if (!ACEdir.endsWith("/")) ACEdir += "/";
outputDir = args[2];
if (!outputDir.endsWith("/")) outputDir += "/";
String fileList = args[3];
showTypes = args.length == 5;
processFileList(fileList);
}
/** analyze arguments to APFtoXML. */
public static void init(String[] args) throws IOException {
if (args.length == 0) argErr();
JetTest.encoding = "UTF-8";
year = args[0];
AceDocument.ace2004 = false;
AceDocument.ace2005 = false;
int requiredArgs = 6;
if (year.equals("2002")) {
} else if (year.equals("2003")) {
} else if (year.equals("2004")) {
requiredArgs = 8;
String gazFile = args[6];
String preDict = args[7];
gazetteer = new Gazetteer();
gazetteer.load(gazFile);
loadPreDict(preDict);
AceDocument.ace2004 = true;
} else if (year.equals("2005")) {
AceDocument.ace2004 = true;
AceDocument.ace2005 = true;
} else {
System.err.println("Invalid year: must be 2002-2005");
System.exit(1);
}
if (args.length <= requiredArgs) argErr();
ACEdir = args[1];
if (!ACEdir.endsWith("/")) ACEdir += "/";
outputDir = args[2];
if (!outputDir.endsWith("/")) outputDir += "/";
fileList = args[3];
apfExtension = args[4];
outputExtension = args[5];
for (int i = requiredArgs; i < args.length; i++) setFlag(args[i]);
}
private static void argErr() {
System.err.println("APFtoXML arguments:");
System.err.println(
" year apf-directory output-directory filelist apf-extension "
+ "output-extension [gazetteer pre-dictionary] flag ...");
System.err.println("gazetteer and pre-dictionary needed for year = 2004");
System.err.println("possible flags: sentences timex mentions extents types names");
System.exit(1);
}
public static void setFlag(String flag) {
if (flag.equals("sentences")
|| flag.equals("timex")
|| flag.equals("mentions")
|| flag.equals("extents")
|| flag.equals("types")
|| flag.equals("names")) {
flags.add(flag);
} else {
System.err.println("APFtoXML: invalid flag");
System.err.println("possible flags: sentences timex mentions extents types names");
System.exit(1);
}
}
/**
* process a set of documents through Jet in accordance with a Jet parameter file. Invoked by <br>
* ProcessDocuments propsFile docList inputDir inputSuffix outputDir outputSuffix
*
* @param propsFile Jet properties file
* @param docList file containing list of documents to be processed, 1 per line
* @param inputDir directory containing files to be processed
* @param inputSuffix file extension to be added to document name to obtain name of input file
* @param outputDir directory containing output files
* @param outputSuffix file extension to be added to document name to obtain name of output file
*/
public static void main(String[] args) throws IOException {
if (args.length != 6) {
System.err.println("ProcessDocuments requires 6 arguments:");
System.err.println(" propsFile docList inputDir inputSuffix outputDir outputSuffix");
System.exit(1);
}
String propsFile = args[0];
String docList = args[1];
String inputDir = args[2];
String inputSuffix = args[3];
String outputDir = args[4];
String outputSuffix = args[5];
// initialize Jet
System.out.println("Starting ACE Jet...");
JetTest.initializeFromConfig(propsFile);
// load ACE type dictionary
EDTtype.readTypeDict();
// turn off traces
Pat.trace = false;
Resolve.trace = false;
// ACE mode (provides additional antecedents ...)
Resolve.ACE = true;
String docName;
int docCount = 0;
BufferedReader docListReader = new BufferedReader(new FileReader(docList));
while ((docName = docListReader.readLine()) != null) {
docCount++;
String inputFile = docName + "." + inputSuffix;
ExternalDocument doc = new ExternalDocument("sgml", inputDir, inputFile);
doc.setAllTags(true);
doc.open();
String[] types = doc.getAnnotationTypes();
doc.setSGMLwrapMargin(0);
String outputFile = docName + "." + outputSuffix;
BufferedWriter writer = new BufferedWriter(new FileWriter(new File(outputDir, outputFile)));
// process document
Ace.monocase = Ace.allLowerCase(doc);
Control.processDocument(doc, writer, docCount == -1, docCount);
writer.close();
}
}
public static void main(String[] args) throws IOException, ClassNotFoundException {
// initialize Jet
if (args.length < 2) {
System.err.println("usage: java EntityFinder propfile datapath files ...");
System.exit(2);
}
JetTest.initializeFromConfig(args[0], args[1]);
Pat.trace = false;
Resolve.trace = false;
String script = JetTest.config.getProperty("processDocument");
if (script == null || script.length() == 0) {
Console.println("*** System error: no processDocument script.");
return;
}
for (int i = 2; i < args.length; i++) {
processFile(args[i], System.out);
}
}