private void cmdManaged_Click(object sender, EventArgs e)
{
//some lang eg. Thai, Lao, need dictionary breaking
//we use dic data from icu-project
//1. create dictionary based breaking engine
//TODO: dic should be read once
var dicProvider = new IcuSimpleTextFileDictionaryProvider()
{
DataDir = "../../../icu58/brkitr_src/dictionaries"
};
CustomBreakerBuilder.Setup(dicProvider);
CustomBreaker breaker1 = CustomBreakerBuilder.NewCustomBreaker();
char[] test = this.textBox1.Text.ToCharArray();
this.listBox1.Items.Clear();
breaker1.BreakWords(test, 0, test.Length);
foreach (BreakSpan span in breaker1.GetBreakSpanIter())
{
string s = new string(test, span.startAt, span.len);
this.listBox1.Items.Add(span.startAt + " " + s);
}
}
private void cmdManaged_Click(object sender, EventArgs e)
{
CustomBreakerBuilder.DataDir = "../../../icu58/brkitr_src/dictionaries";
CustomBreaker breaker1 = CustomBreakerBuilder.NewCustomBreaker();
//2. create dictionary based breaking engine
//
char[] test = this.textBox1.Text.ToCharArray();
this.listBox1.Items.Clear();
breaker1.BreakWords(test, 0);
foreach (var span in breaker1.GetBreakSpanIter())
{
string s = new string(test, span.startAt, span.len);
this.listBox1.Items.Add(span.startAt + " " + s);
}
}
void ParseWithManaged(int ntimes)
{
//-------------------
CustomBreakerBuilder.DataDir = "../../../icu58/brkitr_src/dictionaries";
CustomBreaker breaker1 = CustomBreakerBuilder.NewCustomBreaker();
char[] test = this.textBox1.Text.ToCharArray();
//-------------
for (int i = ntimes - 1; i >= 0; --i)
{
breaker1.BreakWords(test, 0);
foreach (var span in breaker1.GetBreakSpanIter())
{
}
}
}
public void WordKindTest()
{
var breaker = new CustomBreaker { ThrowIfCharOutOfRange = true };
var breakList = new List<BreakSpan>();
#warning Use `breaker.BreakWords("«Maître leçon»", breakList);` once #156 is merged
breaker.BreakWords("«Maître leçon»");
breakList.AddRange(breaker.GetBreakSpanIter());
Assert.AreEqual(breakList.Count, 5);
void BreakSpanEqual(BreakSpan actual, BreakSpan expected)
{
Assert.AreEqual(expected.startAt, actual.startAt);
Assert.AreEqual(expected.len, actual.len);
Assert.AreEqual(expected.wordKind, actual.wordKind);
}
BreakSpanEqual(breakList[0], new BreakSpan { startAt = 0, len = 1, wordKind = WordKind.Punc });
BreakSpanEqual(breakList[1], new BreakSpan { startAt = 1, len = 6, wordKind = WordKind.Text });
BreakSpanEqual(breakList[2], new BreakSpan { startAt = 7, len = 1, wordKind = WordKind.Whitespace });
BreakSpanEqual(breakList[3], new BreakSpan { startAt = 8, len = 5, wordKind = WordKind.Text });
BreakSpanEqual(breakList[4], new BreakSpan { startAt = 13, len = 1, wordKind = WordKind.Punc });
}
private void cmdManaged_Click(object sender, EventArgs e)
{
//some lang eg. Thai, Lao, need dictionary breaking
//we use dic data from icu-project
//1. create dictionary based breaking engine
CustomBreaker breaker1 = CustomBreakerBuilder.NewCustomBreaker();
char[] test = this.textBox1.Text.ToCharArray();
this.listBox1.Items.Clear();
breaker1.BreakWords(test, 0, test.Length);
foreach (BreakSpan span in breaker1.GetBreakSpanIter())
{
string s = new string(test, span.startAt, span.len);
this.listBox1.Items.Add(span.startAt + " " + s);
}
}
void ParseWithManaged(int ntimes)
{
//-------------------
var dicProvider = new IcuSimpleTextFileDictionaryProvider()
{
DataDir = "../../../icu58/brkitr_src/dictionaries"
};
CustomBreakerBuilder.Setup(dicProvider);
CustomBreaker breaker1 = CustomBreakerBuilder.NewCustomBreaker();
char[] test = this.textBox1.Text.ToCharArray();
//-------------
for (int i = ntimes - 1; i >= 0; --i)
{
breaker1.BreakWords(test, 0, test.Length);
foreach (var span in breaker1.GetBreakSpanIter())
{
}
}
}