public boolean load(String fileName, boolean isReset) {
File file = new File(fileName);
if (!file.canRead()) return false; // fail while opening the file
try {
byte[] b = null;
DataInputStream in = new DataInputStream(new FileInputStream(file));
// 读取长度
tableLen = GFCommon.bytes2int(Utility.readBytes(in, 4), false);
logger.debug("tableLen:" + tableLen);
// 读取符号标志
symbolTable = new int[tableLen];
for (int i = 0; i < tableLen; i++) {
b = Utility.readBytes(in, 4);
symbolTable[i] = GFCommon.bytes2int(b, false);
logger.debug("symbolTable[" + i + "]:" + symbolTable[i]);
}
long fileLen = file.length();
long curLen = 4 + tableLen * 4;
while (curLen < fileLen) {
logger.debug("tagContext:");
TagContext tc = new TagContext();
// 读取关键词
b = Utility.readBytes(in, 4);
int key = GFCommon.bytes2int(b);
curLen += 4;
logger.debug("\tkey:" + key);
// 读取总词频
b = Utility.readBytes(in, 4);
curLen += 4;
int totalFreq = GFCommon.bytes2int(b, false);
logger.debug("\ttotalFreq:" + totalFreq);
// 读取词频
int[] tagFreq = new int[tableLen];
for (int i = 0; i < tableLen; i++) {
b = Utility.readBytes(in, 4);
curLen += 4;
tagFreq[i] = GFCommon.bytes2int(b, false);
logger.debug("\ttagFreq[" + i + "]:" + tagFreq[i]);
}
// 读取上下文数组
int[][] contextArray = new int[tableLen][tableLen];
for (int i = 0; i < tableLen; i++) {
String pr = "";
logger.debug("\tcontextArray[" + i + "]");
for (int j = 0; j < tableLen; j++) {
b = Utility.readBytes(in, 4);
curLen += 4;
contextArray[i][j] = GFCommon.bytes2int(b, false);
pr += " " + contextArray[i][j];
}
logger.debug("\t\t" + pr);
}
tc.setTotalFreq(totalFreq);
tc.setKey(key);
tc.setTagFreq(tagFreq);
tc.setContextArray(contextArray);
tcList.add(tc);
}
in.close();
} catch (FileNotFoundException e) {
logger.debug("FileNotFoundException:{}", e);
} catch (IOException e) {
logger.debug("IOException:{}", e);
}
return true;
}
