Refactoring pinyin

1. use custom patched pinyin library Pinyin4Net
2. fix memory leak on startup: 360mb -> 160mb when using vs15 debugger
This commit is contained in:
bao-qian
2016-04-24 00:37:25 +01:00
parent 952b5fab89
commit b600bce23f
16 changed files with 198 additions and 691 deletions

View File

@@ -1,31 +1,58 @@
using System;
using JetBrains.Annotations;
using System.Collections.Generic;
using System.Linq;
using Wox.Infrastructure;
using Wox.Infrastructure.Logger;
namespace Wox.Infrastructure
{
public class StringMatcher
public static class StringMatcher
{
/// <summary>
/// Check if a candidate is match with the source
/// </summary>
/// <param name="source"></param>
/// <param name="candidate"></param>
/// <returns>Match score</returns>
public static int Match(string source, string candidate)
public static int Score(string source, string target)
{
if (string.IsNullOrEmpty(source) || string.IsNullOrEmpty(candidate)) return 0;
FuzzyMatcher matcher = FuzzyMatcher.Create(candidate);
int score = matcher.Evaluate(source).Score;
if (score > 0) return score;
score = matcher.Evaluate(source.Unidecode()).Score;
return score;
if (!string.IsNullOrEmpty(source) && !string.IsNullOrEmpty(target))
{
FuzzyMatcher matcher = FuzzyMatcher.Create(target);
var score = matcher.Evaluate(source).Score;
return score;
}
else
{
return 0;
}
}
public static bool IsMatch(string source, string candidate)
public static int ScoreForPinyin(string source, string target)
{
return Match(source, candidate) > 0;
if (!string.IsNullOrEmpty(source) && !string.IsNullOrEmpty(target))
{
FuzzyMatcher matcher = FuzzyMatcher.Create(target);
//todo happlebao currently generate pinyin on every query, should be generate on startup/index
if (Alphabet.ContainsChinese(source))
{
var combination = Alphabet.PinyinComination(source);
var pinyinScore = combination.Select(pinyin => matcher.Evaluate(string.Join("", pinyin)).Score)
.Max();
var acronymScore = combination.Select(Alphabet.Acronym)
.Select(pinyin => matcher.Evaluate(pinyin).Score)
.Max();
var score = Math.Max(pinyinScore, acronymScore);
return score;
}
else
{
return 0;
}
}
else
{
return 0;
}
}
public static bool IsMatch(string source, string target)
{
return Score(source, target) > 0;
}
}
}