public static DepType getDepType(DependencyTree tree, int pIdx, int aIdx) {
Bunsetsu pBun = tree.getBunsetsuFromNodeId(pIdx);
Bunsetsu aBun = tree.getBunsetsuFromNodeId(aIdx);
Bunsetsu pBunHeadBun = tree.getBunsetsuFromId(pBun.getHead());
Bunsetsu aBunHeadBun = tree.getBunsetsuFromId(aBun.getHead());
if (pBun.getId() == aBun.getId()) {
return DepType.SAME_PHRASE;
} else if (pBun.getId() == aBunHeadBun.getId()) {
return DepType.DEP;
} else if (pBunHeadBun.getId() == aBunHeadBun.getId()) {
return DepType.DEP;
} else {
return DepType.ZERO_INTRA;
}
}
public static ArgPositionType getArgPositionType(DependencyTree tree, int aIdx) {
Bunsetsu aBun = tree.getBunsetsuFromNodeId(aIdx);
DependencyNode aBunHeadNode = aBun.getHeadNode();
if (aIdx == aBunHeadNode.getId()) {
return ArgPositionType.HEAD;
} else if (aIdx == aBunHeadNode.getId() - 1) {
return ArgPositionType.HEAD_LEFT;
} else if (aIdx == aBunHeadNode.getId() + 1) {
return ArgPositionType.HEAD_RIGHT;
} else {
return ArgPositionType.OTHER;
}
}
public static void main(String[] args) {
// String file = args[0];
IntInt2IntHashMap lEr = new IntInt2IntHashMap();
HashMap<String, TObjectIntHashMap<String>> stat =
new HashMap<String, TObjectIntHashMap<String>>();
try {
HashSet<String> headPat = PatternFileParser.parse(new File("ipa_head_pat.txt"));
HashSet<String> funcPat = PatternFileParser.parse(new File("ipa_func_pat.txt"));
CaboCha2Dep pipe = new CaboCha2Dep(System.in);
JapaneseDependencyTree2CaboCha caboChaOutPipe = new JapaneseDependencyTree2CaboCha();
caboChaOutPipe.setFormat(CaboChaFormat.OLD);
BufferedWriter writer = new BufferedWriter(new OutputStreamWriter(System.out, "utf-8"));
while (!pipe.eof()) {
DependencyTree tree = pipe.pipePerSentence();
if (tree == null) {
continue;
}
JapaneseDependencyTreeLib.setBunsetsuHead(tree, funcPat, headPat);
PredicateArgumentStructure[] pasList = tree.getPASList();
for (int j = 0; j < pasList.length; j++) {
int predId = pasList[j].getPredicateId();
String predType = pasList[j].predicateType;
int[] aIds = pasList[j].getIds();
String[] aLabels = pasList[j].getLabels();
for (int k = 0; k < aIds.length; k++) {
DepType dt = getDepType(tree, predId, aIds[k]);
ArgPositionType apt = getArgPositionType(tree, aIds[k]);
if (!stat.containsKey(aLabels[k])) {
stat.put(aLabels[k], new TObjectIntHashMap<String>());
}
TObjectIntHashMap<String> inner = stat.get(aLabels[k]);
if (!inner.containsKey(dt.toString() + ":" + apt.toString())) {
inner.put(dt.toString() + ":" + apt.toString(), 0);
}
inner.increment(dt.toString() + ":" + apt.toString());
aLabels[k] += ":" + dt + ":" + apt;
}
}
StringBuilder resultStr = new StringBuilder();
resultStr.append(caboChaOutPipe.pipePerSentence(tree));
writer.write(resultStr.toString());
}
// print statistics
for (Iterator it = stat.keySet().iterator(); it.hasNext(); ) {
String key = (String) it.next();
TObjectIntHashMap inner = stat.get(key);
for (TObjectIntIterator iit = inner.iterator(); iit.hasNext(); ) {
iit.advance();
System.err.print(key + "\t" + iit.key() + "\t" + iit.value() + "\n");
}
}
} catch (IOException e) {
e.printStackTrace();
}
}