@Override
public String toString() {
StringBuilder sb = new StringBuilder();
sb.append("Mapping of column names:\n");
for (ColumnNames columnName : ColumnNames.values()) {
sb.append(columnName.toString())
.append("\t")
.append(columnNameMap.get(columnName))
.append("\n");
}
sb.append("Mapping of data structures:\n");
for (DataStructures dataStruct : DataStructures.values()) {
sb.append(dataStruct.toString())
.append("\t")
.append(dataStructuresMap.get(dataStruct))
.append("\n");
}
sb.append("Split FEATS column: ").append(useSplitFeats).append("\n");
return sb.toString();
}
public void load(URL specModelURL, SpecificationModels featureSpecModels)
throws MaltChainedException {
BufferedReader br = null;
Pattern tabPattern = Pattern.compile("\t");
if (specModelURL == null) {
throw new FeatureException("The feature specification file cannot be found. ");
}
try {
br = new BufferedReader(new InputStreamReader(specModelURL.openStream()));
} catch (IOException e) {
throw new FeatureException(
"Could not read the feature specification file '" + specModelURL.toString() + "'. ", e);
}
if (br != null) {
int specModelIndex = featureSpecModels.getNextIndex();
String fileLine;
String items[];
StringBuilder featureText = new StringBuilder();
String splitfeats;
ArrayList<String> fileLines = new ArrayList<String>();
ArrayList<String> orderFileLines = new ArrayList<String>();
while (true) {
try {
fileLine = br.readLine();
} catch (IOException e) {
throw new FeatureException(
"Could not read the feature specification file '" + specModelURL.toString() + "'. ",
e);
}
if (fileLine == null) {
break;
}
if (fileLine.length() <= 1 && fileLine.trim().substring(0, 2).trim().equals("--")) {
continue;
}
fileLines.add(fileLine);
}
try {
br.close();
} catch (IOException e) {
throw new FeatureException(
"Could not close the feature specification file '" + specModelURL.toString() + "'. ",
e);
}
for (int j = 0; j < fileLines.size(); j++) {
orderFileLines.add(fileLines.get(j));
}
boolean deprel;
for (int j = 0; j < orderFileLines.size(); j++) {
deprel = false;
featureText.setLength(0);
splitfeats = "";
items = tabPattern.split(orderFileLines.get(j));
if (items.length < 2) {
throw new FeatureException(
"The feature specification file '"
+ specModelURL.toString()
+ "' must contain at least two columns.");
}
if (!(columnNameMap.containsKey(ColumnNames.valueOf(items[0].trim()))
|| columnNameMap.containsValue(items[0].trim()))) {
throw new FeatureException(
"Column one in the feature specification file '"
+ specModelURL.toString()
+ "' contains an unknown value '"
+ items[0].trim()
+ "'. ");
}
if (items[0].trim().equalsIgnoreCase("DEP") || items[0].trim().equalsIgnoreCase("DEPREL")) {
featureText.append("OutputColumn(DEPREL, ");
deprel = true;
} else {
if (columnNameMap.containsKey(ColumnNames.valueOf(items[0].trim()))) {
featureText
.append("InputColumn(")
.append(columnNameMap.get(ColumnNames.valueOf(items[0].trim())))
.append(", ");
} else if (columnNameMap.containsValue(items[0].trim())) {
featureText.append("InputColumn(").append(items[0].trim()).append(", ");
}
if (items[0].trim().equalsIgnoreCase("FEATS") && isUseSplitFeats()) {
splitfeats = "Split(";
}
}
if (!(items[1].trim().equalsIgnoreCase("STACK")
|| items[1].trim().equalsIgnoreCase("INPUT")
|| items[1].trim().equalsIgnoreCase("CONTEXT"))) {
throw new FeatureException(
"Column two in the feature specification file '"
+ specModelURL.toString()
+ "' should be either 'STACK', 'INPUT' or 'CONTEXT' (Covington), not '"
+ items[1].trim()
+ "'. ");
}
int offset = 0;
if (items.length >= 3) {
try {
offset = new Integer(Integer.parseInt(items[2]));
} catch (NumberFormatException e) {
throw new FeatureException(
"The feature specification file '"
+ specModelURL.toString()
+ "' contains a illegal integer value. ",
e);
}
}
String functionArg = "";
if (items[1].trim().equalsIgnoreCase("CONTEXT")) {
if (offset >= 0) {
functionArg =
dataStructuresMap.get(DataStructures.valueOf("LEFTCONTEXT")) + "[" + offset + "]";
} else {
functionArg =
dataStructuresMap.get(DataStructures.valueOf("RIGHTCONTEXT"))
+ "["
+ Math.abs(offset + 1)
+ "]";
}
} else if (dataStructuresMap.containsKey(DataStructures.valueOf(items[1].trim()))) {
if (covington == true) {
if (dataStructuresMap
.get(DataStructures.valueOf(items[1].trim()))
.equalsIgnoreCase("Stack")) {
functionArg = "Left[" + offset + "]";
} else {
functionArg = "Right[" + offset + "]";
}
} else {
functionArg =
dataStructuresMap.get(DataStructures.valueOf(items[1].trim())) + "[" + offset + "]";
}
} else if (dataStructuresMap.containsValue(items[1].trim())) {
if (covington == true) {
if (items[1].trim().equalsIgnoreCase("Stack")) {
functionArg = "Left[" + offset + "]";
} else {
functionArg = "Right[" + offset + "]";
}
} else {
functionArg = items[1].trim() + "[" + offset + "]";
}
} else {
throw new FeatureException(
"Column two in the feature specification file '"
+ specModelURL.toString()
+ "' should not contain the value '"
+ items[1].trim());
}
int linearOffset = 0;
int headOffset = 0;
int depOffset = 0;
int sibOffset = 0;
int suffixLength = 0;
if (items.length >= 4) {
linearOffset = new Integer(Integer.parseInt(items[3]));
}
if (items.length >= 5) {
headOffset = new Integer(Integer.parseInt(items[4]));
}
if (items.length >= 6) {
depOffset = new Integer(Integer.parseInt(items[5]));
}
if (items.length >= 7) {
sibOffset = new Integer(Integer.parseInt(items[6]));
}
if (items.length >= 8) {
suffixLength = new Integer(Integer.parseInt(items[7]));
}
if (linearOffset < 0) {
linearOffset = Math.abs(linearOffset);
for (int i = 0; i < linearOffset; i++) {
functionArg = "pred(" + functionArg + ")";
}
} else if (linearOffset > 0) {
for (int i = 0; i < linearOffset; i++) {
functionArg = "succ(" + functionArg + ")";
}
}
if (headOffset >= 0) {
for (int i = 0; i < headOffset; i++) {
functionArg = "head(" + functionArg + ")";
}
} else {
throw new FeatureException(
"The feature specification file '"
+ specModelURL.toString()
+ "' should not contain a negative head function value. ");
}
if (depOffset < 0) {
depOffset = Math.abs(depOffset);
for (int i = 0; i < depOffset; i++) {
functionArg = "ldep(" + functionArg + ")";
}
} else if (depOffset > 0) {
for (int i = 0; i < depOffset; i++) {
functionArg = "rdep(" + functionArg + ")";
}
}
if (sibOffset < 0) {
sibOffset = Math.abs(sibOffset);
for (int i = 0; i < sibOffset; i++) {
functionArg = "lsib(" + functionArg + ")";
}
} else if (sibOffset > 0) {
for (int i = 0; i < sibOffset; i++) {
functionArg = "rsib(" + functionArg + ")";
}
}
if (deprel == true && (pppath == true || pplifted == true || ppcoveredRoot == true)) {
featureSpecModels.add(specModelIndex, mergePseudoProjColumns(functionArg));
} else {
if (suffixLength != 0) {
featureSpecModels.add(
specModelIndex,
"Suffix(" + featureText.toString() + functionArg + ")," + suffixLength + ")");
} else if (splitfeats.equals("Split(")) {
featureSpecModels.add(
specModelIndex, splitfeats + featureText.toString() + functionArg + "),\\|)");
} else {
featureSpecModels.add(specModelIndex, featureText.toString() + functionArg + ")");
}
}
}
}
}