Merge pull request #183 from taooceros/PinyinLibraryChange

Pinyin library change
This commit is contained in:
Jeremy Wu 2020-11-02 07:32:03 +11:00 committed by GitHub
commit aa3177d2d3
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
5 changed files with 69 additions and 183 deletions

View file

@ -1,178 +0,0 @@
using System;
using System.Collections.Concurrent;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using hyjiacan.util.p4n;
using hyjiacan.util.p4n.format;
using JetBrains.Annotations;
using Flow.Launcher.Infrastructure.Logger;
using Flow.Launcher.Infrastructure.Storage;
using Flow.Launcher.Infrastructure.UserSettings;
namespace Flow.Launcher.Infrastructure
{
public interface IAlphabet
{
string Translate(string stringToTranslate);
}
public class Alphabet : IAlphabet
{
private readonly HanyuPinyinOutputFormat Format = new HanyuPinyinOutputFormat();
private ConcurrentDictionary<string, string[][]> PinyinCache;
private BinaryStorage<Dictionary<string, string[][]>> _pinyinStorage;
private Settings _settings;
public void Initialize([NotNull] Settings settings)
{
_settings = settings ?? throw new ArgumentNullException(nameof(settings));
InitializePinyinHelpers();
}
private void InitializePinyinHelpers()
{
Format.setToneType(HanyuPinyinToneType.WITHOUT_TONE);
Stopwatch.Normal("|Flow Launcher.Infrastructure.Alphabet.Initialize|Preload pinyin cache", () =>
{
_pinyinStorage = new BinaryStorage<Dictionary<string, string[][]>>("Pinyin");
lock(_pinyinStorage)
{
var loaded = _pinyinStorage.TryLoad(new Dictionary<string, string[][]>());
PinyinCache = new ConcurrentDictionary<string, string[][]>(loaded);
}
// force pinyin library static constructor initialize
PinyinHelper.toHanyuPinyinStringArray('T', Format);
});
Log.Info($"|Flow Launcher.Infrastructure.Alphabet.Initialize|Number of preload pinyin combination<{PinyinCache.Count}>");
}
public string Translate(string str)
{
return ConvertChineseCharactersToPinyin(str);
}
public string ConvertChineseCharactersToPinyin(string source)
{
if (!_settings.ShouldUsePinyin)
return source;
if (string.IsNullOrEmpty(source))
return source;
if (!ContainsChinese(source))
return source;
var combination = PinyinCombination(source);
var pinyinArray=combination.Select(x => string.Join("", x));
var acronymArray = combination.Select(Acronym).Distinct();
var joinedSingleStringCombination = new StringBuilder();
var all = acronymArray.Concat(pinyinArray);
all.ToList().ForEach(x => joinedSingleStringCombination.Append(x));
return joinedSingleStringCombination.ToString();
}
public void Save()
{
if (!_settings.ShouldUsePinyin)
{
return;
}
lock(_pinyinStorage)
{
_pinyinStorage.Save(PinyinCache.ToDictionary(i => i.Key, i => i.Value));
}
}
private static string[] EmptyStringArray = new string[0];
private static string[][] Empty2DStringArray = new string[0][];
/// <summmary>
/// replace chinese character with pinyin, non chinese character won't be modified
/// Because we don't have words dictionary, so we can only return all possiblie pinyin combination
/// e.g. 音乐 will return yinyue and yinle
/// <param name="characters"> should be word or sentence, instead of single character. e.g. 微软 </param>
/// </summmary>
public string[][] PinyinCombination(string characters)
{
if (!_settings.ShouldUsePinyin || string.IsNullOrEmpty(characters))
{
return Empty2DStringArray;
}
if (!PinyinCache.ContainsKey(characters))
{
var allPinyins = new List<string[]>();
foreach (var c in characters)
{
var pinyins = PinyinHelper.toHanyuPinyinStringArray(c, Format);
if (pinyins != null)
{
var r = pinyins.Distinct().ToArray();
allPinyins.Add(r);
}
else
{
var r = new[] { c.ToString() };
allPinyins.Add(r);
}
}
var combination = allPinyins.Aggregate(Combination).Select(c => c.Split(';')).ToArray();
PinyinCache[characters] = combination;
return combination;
}
else
{
return PinyinCache[characters];
}
}
public string Acronym(string[] pinyin)
{
var acronym = string.Join("", pinyin.Select(p => p[0]));
return acronym;
}
public bool ContainsChinese(string word)
{
if (!_settings.ShouldUsePinyin)
{
return false;
}
if (word.Length > 40)
{
//Skip strings that are too long string for Pinyin conversion.
return false;
}
var chinese = word.Select(PinyinHelper.toHanyuPinyinStringArray)
.Any(p => p != null);
return chinese;
}
private string[] Combination(string[] array1, string[] array2)
{
if (!_settings.ShouldUsePinyin)
{
return EmptyStringArray;
}
var combination = (
from a1 in array1
from a2 in array2
select $"{a1};{a2}"
).ToArray();
return combination;
}
}
}

View file

@ -53,10 +53,10 @@
<PackageReference Include="Newtonsoft.Json" Version="12.0.3" />
<PackageReference Include="NLog.Schema" Version="4.7.0-rc1" />
<PackageReference Include="NLog.Web.AspNetCore" Version="4.9.0" />
<PackageReference Include="Pinyin4DotNet" Version="2016.4.23.4" />
<PackageReference Include="System.Drawing.Common" Version="4.7.0" />
<PackageReference Include="System.Runtime" Version="4.3.1" />
<PackageReference Include="PropertyChanged.Fody" Version="2.5.13" />
<PackageReference Include="ToolGood.Words.Pinyin" Version="3.0.1.4" />
</ItemGroup>
<ItemGroup>

View file

@ -0,0 +1,65 @@
using System;
using System.Collections.Concurrent;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using JetBrains.Annotations;
using Flow.Launcher.Infrastructure.Logger;
using Flow.Launcher.Infrastructure.Storage;
using Flow.Launcher.Infrastructure.UserSettings;
using ToolGood.Words.Pinyin;
using System.Threading.Tasks;
namespace Flow.Launcher.Infrastructure
{
public interface IAlphabet
{
string Translate(string stringToTranslate);
}
public class PinyinAlphabet : IAlphabet
{
private ConcurrentDictionary<string, string> _pinyinCache = new ConcurrentDictionary<string, string>();
private Settings _settings;
public void Initialize([NotNull] Settings settings)
{
_settings = settings ?? throw new ArgumentNullException(nameof(settings));
}
public string Translate(string content)
{
if (_settings.ShouldUsePinyin)
{
if (!_pinyinCache.ContainsKey(content))
{
if (WordsHelper.HasChinese(content))
{
var result = WordsHelper.GetPinyin(content, ";");
result = GetFirstPinyinChar(result) + result.Replace(";", "");
_pinyinCache[content] = result;
return result;
}
else
{
return content;
}
}
else
{
return _pinyinCache[content];
}
}
else
{
return content;
}
}
private string GetFirstPinyinChar(string content)
{
return string.Concat(content.Split(';').Select(x => x.First()));
}
}
}

View file

@ -29,7 +29,7 @@ namespace Flow.Launcher
private SettingWindowViewModel _settingsVM;
private readonly Updater _updater = new Updater(Flow.Launcher.Properties.Settings.Default.GithubRepo);
private readonly Portable _portable = new Portable();
private readonly Alphabet _alphabet = new Alphabet();
private readonly PinyinAlphabet _alphabet = new PinyinAlphabet();
private StringMatcher _stringMatcher;
[STAThread]

View file

@ -21,11 +21,11 @@ namespace Flow.Launcher
{
private readonly SettingWindowViewModel _settingsVM;
private readonly MainViewModel _mainVM;
private readonly Alphabet _alphabet;
private readonly PinyinAlphabet _alphabet;
#region Constructor
public PublicAPIInstance(SettingWindowViewModel settingsVM, MainViewModel mainVM, Alphabet alphabet)
public PublicAPIInstance(SettingWindowViewModel settingsVM, MainViewModel mainVM, PinyinAlphabet alphabet)
{
_settingsVM = settingsVM;
_mainVM = mainVM;
@ -76,7 +76,6 @@ namespace Flow.Launcher
_settingsVM.Save();
PluginManager.Save();
ImageLoader.Save();
_alphabet.Save();
}
public void ReloadAllPluginData()