public void initialize(UimaContext context) throws ResourceInitializationException {
super.initialize(context);
this.dirPath = (String) context.getConfigParameterValue(PARAM_DIR_PATH);
this.baseData = new File(this.dirPath);
// set up the file system.
//
// corpus
// |
// +-- categoryLabel_<train|test|eval>.txt
//
// each category contains one line per document with line breaks stripped away
//
try {
if (baseData.exists()) {
Map<String, File> filesToZip = Converters.recursivelyListFiles(this.baseData);
try {
SimpleDateFormat df = new SimpleDateFormat("MM-dd-yy-HHmm");
String suffix = df.format(new Date());
File targetZip =
new File(
this.baseData.getParentFile().getParentFile().getName()
+ "_"
+ this.baseData.getParentFile().getName()
+ "_"
+ this.baseData.getName()
+ "_"
+ suffix
+ ".zip");
Converters.zipIt(filesToZip, targetZip);
} catch (Exception e) {
}
Converters.recursivelyDeleteFiles(this.baseData);
}
this.baseData.mkdirs();
} catch (Exception e) {
throw new ResourceInitializationException(e);
}
}
@Test
public void test() throws Exception {
String[] args = {PdfDirWatcher.READ_SECTION_TEXT, inputFile.getPath(), outputFile.getPath()};
WatchDirectory.main(args);
// Listen for changes over 60 seconds
for (int i = 0; i < 60; i++) {
Thread.sleep(1000);
System.out.print(i + " secs ... ");
if (i == 5) {
Converters.copyFile(f1, f2);
Converters.copyFile(f1, f3);
}
if (i == 45) {
f3.delete();
}
}
}
protected void setUp() throws Exception {
super.setUp();
URL u = this.getClass().getClassLoader().getResource("sampleData/plos/8_8");
inputFile = new File(u.getPath());
outputFile = new File(inputFile.getParent() + "/temp");
Converters.recursivelyDeleteFiles(outputFile);
outputFile.mkdir();
u = this.getClass().getClassLoader().getResource("sampleData/plos/8_8/pbio.1000441.pdf");
f1 = new File(u.getPath());
f2 = new File(f1.getParent() + "/temp.pdf");
f2.delete();
f3 = new File(f1.getParent() + "/temp2.pdf");
f3.delete();
}
protected void tearDown() throws Exception {
super.tearDown();
f2.delete();
f3.delete();
Converters.recursivelyDeleteFiles(outputFile);
}
public static void main(String args[]) throws Exception {
LapdfEngine engine = new LapdfEngine();
if (args.length < 1) {
System.err.println(USAGE);
System.exit(1);
}
String inputFileOrDirPath = args[0];
String outputDirPath = "";
String ruleFilePath = "";
File inputFileOrDir = new File(inputFileOrDirPath);
if (!inputFileOrDir.exists()) {
System.err.println(USAGE);
System.err.println("Input file / dir '" + inputFileOrDirPath + "' does not exist.");
System.err.println("Please include full path");
System.exit(1);
}
// output folder is set.
if (args.length > 1) {
outputDirPath = args[1];
} else {
outputDirPath = "-";
}
if (outputDirPath.equals("-")) {
if (inputFileOrDir.isDirectory()) {
outputDirPath = inputFileOrDirPath;
} else {
outputDirPath = inputFileOrDir.getParent();
}
}
File outDir = new File(outputDirPath);
if (!outDir.exists()) {
outDir.mkdir();
}
// output folder is set.
File ruleFile = null;
if (args.length > 2) {
ruleFilePath = args[2];
} else {
ruleFilePath = "-";
}
if (ruleFilePath.equals("-")) {
ruleFile = Converters.extractFileFromJarClasspath("rules/general.drl");
} else {
ruleFile = new File(ruleFilePath);
}
if (!ruleFile.exists()) {
System.err.println(USAGE);
System.err.println(ruleFilePath + " does not exist.");
System.err.println("Please include full path");
}
if (inputFileOrDir.isDirectory()) {
Pattern patt = Pattern.compile("\\.pdf$");
Map<String, File> inputFiles = Converters.recursivelyListFiles(inputFileOrDir, patt);
Iterator<String> it = inputFiles.keySet().iterator();
while (it.hasNext()) {
String key = it.next();
File pdf = inputFiles.get(key);
String pdfStem = pdf.getName();
pdfStem = pdfStem.replaceAll("\\.pdf", "");
String outImgPath =
Converters.mimicDirectoryStructure(inputFileOrDir, outDir, pdf).getPath();
outImgPath = outImgPath.replaceAll("\\.pdf", "_secImgs");
File outImgDir = new File(outImgPath);
if (!outImgDir.exists()) outImgDir.mkdir();
try {
LapdfDocument lapdf = engine.blockifyPdfFile(pdf);
engine.classifyDocument(lapdf, ruleFile);
engine.renderImageOutlines(lapdf, outImgDir, pdfStem, LapdfMode.CLASSIFY);
} catch (Exception e) {
e.printStackTrace();
}
}
} else {
String pdfStem = inputFileOrDir.getName();
pdfStem = pdfStem.replaceAll("\\.pdf$", "");
String outImgPath = outDir.getPath() + "/" + pdfStem + "_secImgs";
File outImgDir = new File(outImgPath);
if (!outImgDir.exists()) outImgDir.mkdir();
LapdfDocument lapdf = engine.blockifyPdfFile(inputFileOrDir);
engine.classifyDocument(lapdf, ruleFile);
engine.renderImageOutlines(lapdf, outImgDir, pdfStem, LapdfMode.CLASSIFY);
}
}