/// <summary>
/// 是否存在指定前缀的词性
/// </summary>
/// <param name="prefix"></param>
/// <returns></returns>
public bool PrefixMatch(string prefix)
{
for (int i = 0; i < natures.Length; i++)
{
if (NatureHelper.StartsWith(natures[i], prefix))
{
return(true);
}
}
return(false);
}
private static bool Load(string path, Nature defNat, SortedDictionary <string, WordAttr> dict)
{
try
{
var splitter = new[] { ' ', '\t' };
if (path.EndsWith(".csv"))
{
splitter = new[] { ',' }
}
;
foreach (var line in File.ReadLines(path))
{
var segs = line.Split(splitter);
if (segs.Length == 0)
{
continue;
}
if (Config.NormalizeChar)
{
segs[0] = CharTable.Convert(segs[0]);
}
var natCount = (segs.Length - 1) / 2;
WordAttr attr;
if (natCount == 0)
{
attr = new WordAttr(defNat);
}
else
{
attr = new WordAttr(natCount);
for (int i = 0; i < natCount; i++)
{
attr.natures[i] = NatureHelper.GetOrCreate(segs[1 + (i << 1)]);
attr.freqs[i] = int.Parse(segs[(i + 1) << 1]);
attr.totalFreq += attr.freqs[i];
}
}
dict[segs[0]] = attr;
}
return(true);
}
catch (Exception e)
{
return(false);
}
}
private static bool Load()
{
if (LoadDat(Config.Custom_Dict_Path[0]))
{
return(true);
}
dat = new DoubleArrayTrie <WordAttr>();
var dict = new SortedDictionary <string, WordAttr>(StrComparer.Default);
try
{
for (var i = 0; i < Config.Custom_Dict_Path.Length; i++)
{
var p = Config.Custom_Dict_Path[i]; // 当前自定义词典文件路径
var defNat = Nature.n;
int spaceIdx = p.IndexOf(' ');
if (spaceIdx > 0)
{
// 有默认词性
var nat = p.Substring(spaceIdx + 1); // 空格之后为词性
p = p.Substring(0, spaceIdx); //
defNat = NatureHelper.GetOrCreate(nat);
}
Load(p, defNat, dict);
//bool success =
//if(!success)
// log warning "loading file failed: " + p
}
if (dict.Count == 0)
{
// log warning "no items loaded"
dict[Constants.TAG_OTHER] = null; // 当作空白占位符
}
dat.Build(dict);
SaveDat(Config.Custom_Dict_Path[0], dict);
return(true);
}
catch (Exception e)
{
return(false);
}
}
private static bool LoadDat(string path)
{
try
{
var ba = ByteArray.Create(path + Predefine.BIN_EXT);
if (ba == null)
{
return(false);
}
int size = ba.NextInt();
if (size < 0) // 一种兼容措施，当Size小于零表示文件头存储了-Size个用户词性
{
while (size < 0)
{
var customNat = ba.NextString();
NatureHelper.GetOrCreate(customNat); // register user-defined nature
size++;
}
size = ba.NextInt();
}
var attrs = new WordAttr[size];
for (int i = 0; i < size; i++) // 加载values
{
var totalFreq = ba.NextInt();
var len = ba.NextInt();
attrs[i] = new WordAttr(len);
attrs[i].totalFreq = totalFreq;
for (int j = 0; j < len; j++)
{
attrs[i].natures[j] = (Nature)ba.NextInt();
attrs[i].freqs[j] = ba.NextInt();
}
}
return(dat.Load(ba, attrs)); // 加载keys
}
catch (Exception e)
{
return(false);
}
}
public static Attribute Create(string natWithFreq)
{
try
{
var segs = natWithFreq.Split(' ');
int natCount = segs.Length / 2;
var attr = new Attribute(natCount);
for (int i = 0; i < natCount; i++)
{
attr.natures[i] = NatureHelper.GetOrCreate(segs[i << 1]); //LexiconUtil.Str2Nat(segs[2 * i], null);
attr.freqs[i] = int.Parse(segs[2 * i + 1]);
attr.totalFreq += attr.freqs[i];
}
return(attr);
}
catch (Exception e)
{
// log warning "creating Attribute failed";
return(null);
}
}