Preparing to integrate parts of Unihan.

This commit is contained in:
GoldenCrystal
2014-11-15 00:58:03 +01:00
parent d7ada9e413
commit 533c9e6d1b
15 changed files with 213 additions and 53 deletions
@@ -8,11 +8,11 @@ using System.Unicode;
namespace System.Unicode.Builder
{
public sealed class FileUcdSource : IUcdSource
public sealed class FileDataSource : IDataSource
{
private readonly string baseDirectory;
public FileUcdSource(string baseDirectory)
public FileDataSource(string baseDirectory)
{
this.baseDirectory = Path.GetFullPath(baseDirectory);
}
@@ -8,19 +8,19 @@ using System.Threading.Tasks;
namespace System.Unicode.Builder
{
public class HttpUcdSource : IUcdSource
public class HttpDataSource : IDataSource
{
public static readonly Uri UnicodeCharacterDataUri = new Uri("http://www.unicode.org/Public/UCD/latest/ucd/", UriKind.Absolute);
private readonly HttpClient httpClient;
private readonly Uri baseUri;
public HttpUcdSource()
public HttpDataSource()
: this(UnicodeCharacterDataUri)
{
}
public HttpUcdSource(Uri baseUri)
public HttpDataSource(Uri baseUri)
{
this.httpClient = new HttpClient();
this.baseUri = baseUri;
@@ -7,7 +7,7 @@ using System.Threading.Tasks;
namespace System.Unicode.Builder
{
public interface IUcdSource : IDisposable
public interface IDataSource : IDisposable
{
Task<Stream> OpenDataFileAsync(string fileName);
}
+33 -25
View File
@@ -13,7 +13,7 @@ namespace System.Unicode.Builder
{
public const string UnihanDirectoryName = "Unihan";
public const string UnihanArchiveName = "Unihan.zip";
public const string UcdDirectoryName = "UCD";
public const string directoryName = "UCD";
public const string UcdArchiveName = "UCD.zip";
public static readonly string[] ucdRequiredFiles = new[]
@@ -22,32 +22,39 @@ namespace System.Unicode.Builder
"PropList.txt",
"DerivedCoreProperties.txt",
"Blocks.txt",
"Jamo.txt"
"Jamo.txt",
};
private static byte[] DownloadUcdArchive()
public static readonly string[] unihanRequiredFiles = new[]
{
"Unihan_NumericValues.txt",
"Unihan_Readings.txt",
"Unihan_Variants.txt",
};
private static byte[] DownloadDataArchive(string archiveName)
{
using (var httpClient = new HttpClient())
{
return httpClient.GetByteArrayAsync(HttpUcdSource.UnicodeCharacterDataUri + UcdArchiveName).Result;
return httpClient.GetByteArrayAsync(HttpDataSource.UnicodeCharacterDataUri + archiveName).Result;
}
}
internal static IUcdSource GetUcdSource(bool? shouldDownload, bool? shouldSaveArchive, bool? shouldExtract)
internal static IDataSource GetDataSource(string archiveName, string directoryName, string[] requiredFiles, bool? shouldDownload, bool? shouldSaveArchive, bool? shouldExtract)
{
string baseDirectory = Environment.CurrentDirectory;
string ucdDirectory = Path.Combine(baseDirectory, UcdDirectoryName);
string ucdArchiveFileName = Path.Combine(baseDirectory, UcdArchiveName);
string dataDirectory = Path.Combine(baseDirectory, Program.directoryName);
string dataArchiveFileName = Path.Combine(baseDirectory, archiveName);
if (shouldDownload != true)
{
bool hasValidDirectory = Directory.Exists(ucdDirectory);
bool hasValidDirectory = Directory.Exists(dataDirectory);
if (hasValidDirectory)
{
foreach (string ucdRequiredFile in ucdRequiredFiles)
foreach (string requiredFile in requiredFiles)
{
if (!File.Exists(Path.Combine(ucdDirectory, ucdRequiredFile)))
if (!File.Exists(Path.Combine(dataDirectory, requiredFile)))
{
hasValidDirectory = false;
break;
@@ -57,47 +64,47 @@ namespace System.Unicode.Builder
if (hasValidDirectory)
{
return new FileUcdSource(ucdDirectory);
return new FileDataSource(dataDirectory);
}
if (File.Exists(ucdArchiveFileName))
if (File.Exists(dataArchiveFileName))
{
if (shouldExtract == true)
{
ZipFile.ExtractToDirectory(ucdArchiveFileName, ucdDirectory);
return new FileUcdSource(ucdDirectory);
ZipFile.ExtractToDirectory(dataArchiveFileName, dataDirectory);
return new FileDataSource(dataDirectory);
}
else
{
return new ZipUcdSource(File.OpenRead(ucdArchiveFileName));
return new ZipDataSource(File.OpenRead(dataArchiveFileName));
}
}
}
if (shouldDownload != false)
{
var ucdArchiveData = DownloadUcdArchive();
var dataArchiveData = DownloadDataArchive(archiveName);
if (shouldSaveArchive == true)
{
var stream = File.Open(ucdArchiveFileName, FileMode.Create, FileAccess.ReadWrite, FileShare.Read);
var stream = File.Open(dataArchiveFileName, FileMode.Create, FileAccess.ReadWrite, FileShare.Read);
try
{
stream.Write(ucdArchiveData, 0, ucdArchiveData.Length);
ucdArchiveData = null; // Release the reference now, since we won't need it anymore.
stream.Write(dataArchiveData, 0, dataArchiveData.Length);
dataArchiveData = null; // Release the reference now, since we won't need it anymore.
if (shouldExtract == true)
{
using (var archive = new ZipArchive(stream, ZipArchiveMode.Read, false))
{
archive.ExtractToDirectory(ucdDirectory);
archive.ExtractToDirectory(dataDirectory);
return new FileUcdSource(ucdDirectory);
return new FileDataSource(dataDirectory);
}
}
else
{
return new ZipUcdSource(stream);
return new ZipDataSource(stream);
}
}
catch
@@ -108,7 +115,7 @@ namespace System.Unicode.Builder
}
else
{
return new ZipUcdSource(new MemoryStream(ucdArchiveData));
return new ZipDataSource(new MemoryStream(dataArchiveData));
}
}
@@ -119,9 +126,10 @@ namespace System.Unicode.Builder
{
UnicodeInfoBuilder data;
using (var ucdSource = GetUcdSource(null, null, null))
using (var ucdSource = GetDataSource(UcdArchiveName, directoryName, ucdRequiredFiles, null, null, null))
using (var unihanSource = GetDataSource(UnihanArchiveName, UnihanDirectoryName, unihanRequiredFiles, null, null, null))
{
data = UnicodeDataProcessor.BuildDataAsync(ucdSource).Result;
data = UnicodeDataProcessor.BuildDataAsync(ucdSource, unihanSource).Result;
}
using (var stream = new DeflateStream(File.Create("ucd.dat"), CompressionLevel.Optimal, false))
@@ -27,7 +27,7 @@ namespace System.Unicode.Builder
return string.IsNullOrEmpty(s) ? null : s;
}
public static async Task<UnicodeInfoBuilder> BuildDataAsync(IUcdSource ucdSource)
public static async Task<UnicodeInfoBuilder> BuildDataAsync(IDataSource ucdSource, IDataSource unihanSource)
{
var builder = new UnicodeInfoBuilder(new Version(7, 0));
@@ -39,7 +39,7 @@ namespace System.Unicode.Builder
return builder;
}
private static async Task ProcessUnicodeDataFile(IUcdSource ucdSource, UnicodeInfoBuilder builder)
private static async Task ProcessUnicodeDataFile(IDataSource ucdSource, UnicodeInfoBuilder builder)
{
using (var reader = new UnicodeDataFileReader(await ucdSource.OpenDataFileAsync(UnicodeDataFileName).ConfigureAwait(false)))
{
@@ -156,7 +156,7 @@ namespace System.Unicode.Builder
}
}
private static async Task ProcessPropListFile(IUcdSource ucdSource, UnicodeInfoBuilder builder)
private static async Task ProcessPropListFile(IDataSource ucdSource, UnicodeInfoBuilder builder)
{
using (var reader = new UnicodeDataFileReader(await ucdSource.OpenDataFileAsync(PropListFileName).ConfigureAwait(false)))
{
@@ -173,7 +173,7 @@ namespace System.Unicode.Builder
}
}
private static async Task ProcessDerivedCorePropertiesFile(IUcdSource ucdSource, UnicodeInfoBuilder builder)
private static async Task ProcessDerivedCorePropertiesFile(IDataSource ucdSource, UnicodeInfoBuilder builder)
{
using (var reader = new UnicodeDataFileReader(await ucdSource.OpenDataFileAsync(DerivedCorePropertiesFileName).ConfigureAwait(false)))
{
@@ -190,7 +190,7 @@ namespace System.Unicode.Builder
}
}
private static async Task ProcessBlocksFile(IUcdSource ucdSource, UnicodeInfoBuilder builder)
private static async Task ProcessBlocksFile(IDataSource ucdSource, UnicodeInfoBuilder builder)
{
using (var reader = new UnicodeDataFileReader(await ucdSource.OpenDataFileAsync(BlocksFileName).ConfigureAwait(false)))
{
@@ -54,16 +54,16 @@
<ItemGroup>
<Compile Include="BinaryWriterExtensions.cs" />
<Compile Include="CharacterDecompositionMapping.cs" />
<Compile Include="FileUcdSource.cs" />
<Compile Include="HttpUcdSource.cs" />
<Compile Include="IUcdSource.cs" />
<Compile Include="FileDataSource.cs" />
<Compile Include="HttpDataSource.cs" />
<Compile Include="IDataSource.cs" />
<Compile Include="Program.cs" />
<Compile Include="Properties\AssemblyInfo.cs" />
<Compile Include="UnicodeCharacterDataBuilder.cs" />
<Compile Include="UnicodeInfoBuilder.cs" />
<Compile Include="UnicodeDataFileReader.cs" />
<Compile Include="UnicodeDataProcessor.cs" />
<Compile Include="ZipUcdSource.cs" />
<Compile Include="ZipDataSource.cs" />
</ItemGroup>
<ItemGroup>
<None Include="..\System.Unicode.snk">
@@ -8,11 +8,11 @@ using System.Threading.Tasks;
namespace System.Unicode.Builder
{
public sealed class ZipUcdSource : IUcdSource
public sealed class ZipDataSource : IDataSource
{
private readonly ZipArchive archive;
public ZipUcdSource(Stream stream)
public ZipDataSource(Stream stream)
{
archive = new ZipArchive(stream, ZipArchiveMode.Read, false);
}