private static string LoadDataDir()
{
// LUCENENET specific - reformatted with :, renamed from "analysis.data.dir"
string currentPath = SystemProperties.GetProperty("kuromoji:data:dir", AppDomain.CurrentDomain.BaseDirectory);
// If a matching directory path is found, set our DATA_DIR static
// variable. If it is null or empty after this process, we need to
// load the embedded files.
string candidatePath = System.IO.Path.Combine(currentPath, DATA_SUBDIR);
if (System.IO.Directory.Exists(candidatePath))
{
return(candidatePath);
}
while (new DirectoryInfo(currentPath).Parent != null)
{
try
{
candidatePath = System.IO.Path.Combine(new DirectoryInfo(currentPath).Parent.FullName, DATA_SUBDIR);
if (System.IO.Directory.Exists(candidatePath))
{
return(candidatePath);
}
currentPath = new DirectoryInfo(currentPath).Parent.FullName;
}
catch (SecurityException)
{
// ignore security errors
}
}
return(null); // This is the signal to load from local resources
}
/// <summary>
/// Get ICU configuration property value for the given name.
/// </summary>
/// <param name="name">The configuration property name.</param>
/// <param name="def">The default value.</param>
/// <returns>The configuration property value. If the property does not
/// exist, <paramref name="def"/> is returned.</returns>
public static string Get(string name, string def)
{
if (string.IsNullOrEmpty(name))
{
return(def);
}
string value = null;
// Try to get an environment variable first
value = SystemProperties.GetProperty(name, null);
if (value != null)
{
return(value);
}
try
{
value = global::ICU4N.ICUConfig.ResourceManager.GetString(name.Replace(".", "_"));
if (value != null)
{
return(value);
}
}
catch (MissingManifestResourceException)
{
}
return(def);
}
/// <summary>
/// Read both algorithm and config properties.
/// </summary>
/// <param name="algReader">From where to read algorithm and config properties.</param>
/// <exception cref="IOException">If there is a low-level I/O error.</exception>
public Config(TextReader algReader)
{
// read alg file to array of lines
IList <string> lines = new List <string>();
int lastConfigLine = 0;
string line;
while ((line = algReader.ReadLine()) != null)
{
lines.Add(line);
if (line.IndexOf('=') > 0)
{
lastConfigLine = lines.Count;
}
}
algReader.Dispose();
// copy props lines to string
MemoryStream ms = new MemoryStream();
TextWriter writer = new StreamWriter(ms);
for (int i = 0; i < lastConfigLine; i++)
{
writer.WriteLine(lines[i]);
}
// read props from string
this.props = new Dictionary <string, string>();
writer.Flush();
ms.Position = 0;
props.Load(ms);
// make sure work dir is set properly
string temp;
if (!props.TryGetValue("work.dir", out temp) || temp == null)
{
props["work.dir"] = SystemProperties.GetProperty("benchmark.work.dir", "work");
}
if (props.TryGetValue("print.props", out temp))
{
if (temp.Equals("true", StringComparison.OrdinalIgnoreCase))
{
PrintProps();
}
}
else if (DEFAULT_PRINT_PROPS)
{
PrintProps();
}
// copy algorithm lines
var sb = new StringBuilder();
for (int i = lastConfigLine; i < lines.Count; i++)
{
sb.Append(lines[i]);
sb.Append(NEW_LINE);
}
algorithmText = sb.ToString();
}
public virtual void ReadPreconfiguredEnvironmentTest()
{
string testKey = "tests:setup";
string testValue = "setup";
Assert.AreEqual(testValue, ConfigurationSettings.CurrentConfiguration[testKey]);
Assert.AreEqual(testValue, SystemProperties.GetProperty(testKey));
}
public virtual void TestRuntimeEnviromentOverrideSetting()
{
Assert.AreEqual("fr", ConfigurationSettings.CurrentConfiguration["tests:locale"]);
Assert.AreEqual("fr", SystemProperties.GetProperty("tests:locale"));
ConfigurationSettings.CurrentConfiguration["tests:locale"] = "en";
Assert.AreEqual("en", ConfigurationSettings.CurrentConfiguration["tests:locale"]);
Assert.AreEqual("en", SystemProperties.GetProperty("tests:locale"));
}
/// <summary>
/// Entry point to the DiffIt application.
/// <para>
/// This application takes one argument, the path to a file containing a
/// stemmer table. The program reads the file and generates the patch commands
/// for the stems.
/// </para>
/// </summary>
/// <param name="args">the path to a file containing a stemmer table</param>
public static void Main(string[] args)
{
int ins = Get(0, args[0]);
int del = Get(1, args[0]);
int rep = Get(2, args[0]);
int nop = Get(3, args[0]);
// LUCENENET specific - reformatted with :
string charset = SystemProperties.GetProperty("egothor:stemmer:charset", "UTF-8");
var stemmerTables = new List <string>();
// LUCENENET specific
// command line argument overrides environment variable or default, if supplied
for (int i = 1; i < args.Length; i++)
{
if ("-e".Equals(args[i], StringComparison.Ordinal) || "--encoding".Equals(args[i], StringComparison.Ordinal))
{
charset = args[i];
}
else
{
stemmerTables.Add(args[i]);
}
}
foreach (var stemmerTable in stemmerTables)
{
// System.out.println("[" + args[i] + "]");
Diff diff = new Diff(ins, del, rep, nop);
using (TextReader input = new StreamReader(new FileStream(stemmerTable, FileMode.Open, FileAccess.Read), Encoding.GetEncoding(charset)))
{
string line;
while ((line = input.ReadLine()) != null)
{
try
{
line = line.ToLowerInvariant();
StringTokenizer st = new StringTokenizer(line);
st.MoveNext();
string stem = st.Current;
Console.WriteLine(stem + " -a");
while (st.MoveNext())
{
string token = st.Current;
if (token.Equals(stem, StringComparison.Ordinal) == false)
{
Console.WriteLine(stem + " " + diff.Exec(token, stem));
}
}
}
catch (InvalidOperationException /*x*/)
{
// no base token (stem) on a line
}
}
}
}
}
public virtual void SetRuntimeEnvironmentTest()
{
string testKey = "tests:setting";
string testValue = "test.success";
ConfigurationSettings.CurrentConfiguration[testKey] = testValue;
Assert.AreEqual(testValue, ConfigurationSettings.CurrentConfiguration[testKey]);
Assert.AreEqual(testValue, SystemProperties.GetProperty(testKey));
}
// LUCENENET specific - changed the logic here to leave the
// ANALYSIS_DATA_DIR an empty string if it is not found. This
// allows us to skip loading files from disk if there are no files
// to load (and fixes LUCENE-1817 that prevents the on-disk files
// from ever being loaded).
private static void Init()
{
#if FEATURE_ENCODINGPROVIDERS
// Support for GB2312 encoding. See: https://docs.microsoft.com/en-us/dotnet/api/system.text.codepagesencodingprovider?view=netcore-2.0
var encodingProvider = System.Text.CodePagesEncodingProvider.Instance;
System.Text.Encoding.RegisterProvider(encodingProvider);
#endif
string dirName = "smartcn-data";
//string propName = "analysis.properties";
// Try the system property：-Danalysis.data.dir=/path/to/analysis-data
//ANALYSIS_DATA_DIR = System.getProperty("analysis.data.dir", "");
// LUCENENET specific - reformatted with :, renamed from "analysis.data.dir"
ANALYSIS_DATA_DIR = SystemProperties.GetProperty("smartcn:data:dir", "");
if (ANALYSIS_DATA_DIR.Length != 0)
{
return;
}
#if FEATURE_APPDOMAIN_BASEDIRECTORY
string currentPath = AppDomain.CurrentDomain.BaseDirectory;
#else
string currentPath = System.AppContext.BaseDirectory;
#endif
string candidatePath = System.IO.Path.Combine(currentPath, dirName);
if (Directory.Exists(candidatePath))
{
ANALYSIS_DATA_DIR = candidatePath;
return;
}
try
{
while (new DirectoryInfo(currentPath).Parent != null)
{
candidatePath = System.IO.Path.Combine(new DirectoryInfo(currentPath).Parent.FullName, dirName);
if (Directory.Exists(candidatePath))
{
ANALYSIS_DATA_DIR = candidatePath;
return;
}
currentPath = new DirectoryInfo(currentPath).Parent.FullName;
}
}
catch (SecurityException)
{
// ignore security errors
}
}
public void TestCustomConfigurationSettings()
{
Assert.AreEqual("banana", ConfigurationSettings.CurrentConfiguration["fruit"]);
Assert.AreEqual("banana", SystemProperties.GetProperty("fruit"));
Assert.AreEqual("lettuce", ConfigurationSettings.CurrentConfiguration["vegetable"]);
Assert.AreEqual("lettuce", SystemProperties.GetProperty("vegetable"));
Assert.AreEqual("yogurt", ConfigurationSettings.CurrentConfiguration["tests:goo"]);
Assert.AreEqual("yogurt", SystemProperties.GetProperty("tests:goo"));
Assert.AreEqual("pizza", ConfigurationSettings.CurrentConfiguration["tests:junk"]);
Assert.AreEqual("pizza", SystemProperties.GetProperty("tests:junk"));
}
public virtual void TestSetandUnset()
{
string testKey = "tests:locale";
string testValue_fr = "fr";
string testValue_en = "en";
//ConfigurationSettings.CurrentConfiguration[testKey] = testValue_fr;
Assert.AreEqual(testValue_fr, ConfigurationSettings.CurrentConfiguration["tests:locale"]);
Assert.AreEqual(testValue_fr, ConfigurationSettings.CurrentConfiguration[testKey]);
Assert.AreEqual(testValue_fr, SystemProperties.GetProperty(testKey));
ConfigurationSettings.CurrentConfiguration[testKey] = testValue_en;
Assert.AreEqual(testValue_en, ConfigurationSettings.CurrentConfiguration[testKey]);
Assert.AreEqual(testValue_en, SystemProperties.GetProperty(testKey));
ConfigurationSettings.CurrentConfiguration.Reload();
Assert.AreEqual(testValue_fr, ConfigurationSettings.CurrentConfiguration[testKey]);
Assert.AreEqual(testValue_fr, SystemProperties.GetProperty(testKey));
}
static ICUDebug()
{
// ICU4N TODO: Fix debugging so it makes sense in .NET
parameters = SystemProperties.GetProperty("ICUDebug");
debug = parameters != null;
help = debug && (parameters.Equals("") || parameters.IndexOf("help", StringComparison.Ordinal) != -1);
if (debug)
{
Console.Out.WriteLine("\nICUDebug=" + parameters);
}
// ICU4N TODO: Java? Need to translate
javaVersionString = SystemProperties.GetProperty("java.version", "0");
javaVersion = GetInstanceLenient(javaVersionString);
VersionInfo java14Version = VersionInfo.GetInstance("1.4.0");
isJDK14OrHigher = javaVersion.CompareTo(java14Version) >= 0;
}
static BinaryDictionary()
{
// LUCENENET specific - reformatted with :, renamed from "analysis.data.dir"
string currentPath = SystemProperties.GetProperty("kuromoji:data:dir",
#if NETSTANDARD1_6
System.AppContext.BaseDirectory
#else
AppDomain.CurrentDomain.BaseDirectory
#endif
);
// If a matching directory path is found, set our DATA_DIR static
// variable. If it is null or empty after this process, we need to
// load the embedded files.
string candidatePath = System.IO.Path.Combine(currentPath, DATA_SUBDIR);
if (System.IO.Directory.Exists(candidatePath))
{
DATA_DIR = candidatePath;
return;
}
while (new DirectoryInfo(currentPath).Parent != null)
{
try
{
candidatePath = System.IO.Path.Combine(new DirectoryInfo(currentPath).Parent.FullName, DATA_SUBDIR);
if (System.IO.Directory.Exists(candidatePath))
{
DATA_DIR = candidatePath;
return;
}
currentPath = new DirectoryInfo(currentPath).Parent.FullName;
}
catch (SecurityException)
{
// ignore security errors
}
}
}
/// <summary>
/// Entry point to the Compile application.
/// <para/>
/// This program takes any number of arguments: the first is the name of the
/// desired stemming algorithm to use (a list is available in the package
/// description) , all of the rest should be the path or paths to a file or
/// files containing a stemmer table to compile.
/// </summary>
/// <param name="args">the command line arguments</param>
public static void Main(string[] args)
{
if (args.Length < 1)
{
return;
}
// LUCENENET NOTE: This line does nothing in .NET
// and also does nothing in Java...what?
//args[0].ToUpperInvariant();
// Reads the first char of the first arg
backward = args[0][0] == '-';
int qq = (backward) ? 1 : 0;
bool storeorig = false;
if (args[0][qq] == '0')
{
storeorig = true;
qq++;
}
multi = args[0][qq] == 'M';
if (multi)
{
qq++;
}
// LUCENENET specific - reformatted with :
string charset = SystemProperties.GetProperty("egothor:stemmer:charset", "UTF-8");
var stemmerTables = new List <string>();
// LUCENENET specific
// command line argument overrides environment variable or default, if supplied
for (int i = 1; i < args.Length; i++)
{
if ("-e".Equals(args[i], StringComparison.Ordinal) || "--encoding".Equals(args[i], StringComparison.Ordinal))
{
charset = args[i];
}
else
{
stemmerTables.Add(args[i]);
}
}
char[] optimizer = new char[args[0].Length - qq];
for (int i = 0; i < optimizer.Length; i++)
{
optimizer[i] = args[0][qq + i];
}
foreach (var stemmerTable in stemmerTables)
{
// System.out.println("[" + args[i] + "]");
Diff diff = new Diff();
//int stems = 0; // not used
int words = 0;
AllocTrie();
Console.WriteLine(stemmerTable);
using (TextReader input = new StreamReader(
new FileStream(stemmerTable, FileMode.Open, FileAccess.Read), Encoding.GetEncoding(charset)))
{
string line;
while ((line = input.ReadLine()) != null)
{
try
{
line = line.ToLowerInvariant();
StringTokenizer st = new StringTokenizer(line);
st.MoveNext();
string stem = st.Current;
if (storeorig)
{
trie.Add(stem, "-a");
words++;
}
while (st.MoveNext())
{
string token = st.Current;
if (token.Equals(stem, StringComparison.Ordinal) == false)
{
trie.Add(token, diff.Exec(token, stem));
words++;
}
}
}
catch (InvalidOperationException /*x*/)
{
// no base token (stem) on a line
}
}
}
Optimizer o = new Optimizer();
Optimizer2 o2 = new Optimizer2();
Lift l = new Lift(true);
Lift e = new Lift(false);
Gener g = new Gener();
for (int j = 0; j < optimizer.Length; j++)
{
string prefix;
switch (optimizer[j])
{
case 'G':
trie = trie.Reduce(g);
prefix = "G: ";
break;
case 'L':
trie = trie.Reduce(l);
prefix = "L: ";
break;
case 'E':
trie = trie.Reduce(e);
prefix = "E: ";
break;
case '2':
trie = trie.Reduce(o2);
prefix = "2: ";
break;
case '1':
trie = trie.Reduce(o);
prefix = "1: ";
break;
default:
continue;
}
trie.PrintInfo(Console.Out, prefix + " ");
}
using (DataOutputStream os = new DataOutputStream(
new FileStream(stemmerTable + ".out", FileMode.OpenOrCreate, FileAccess.Write)))
{
os.WriteUTF(args[0]);
trie.Store(os);
}
}
}
// LUCENENET specific - changed the logic here to leave the
// ANALYSIS_DATA_DIR an empty string if it is not found. This
// allows us to skip loading files from disk if there are no files
// to load (and fixes LUCENE-1817 that prevents the on-disk files
// from ever being loaded).
private static void Init()
{
#if NETSTANDARD
// Support for GB2312 encoding. See: https://docs.microsoft.com/en-us/dotnet/api/system.text.codepagesencodingprovider?view=netcore-2.0
var encodingProvider = System.Text.CodePagesEncodingProvider.Instance;
System.Text.Encoding.RegisterProvider(encodingProvider);
#endif
string dirName = "analysis-data";
//string propName = "analysis.properties";
// Try the system property：-Danalysis.data.dir=/path/to/analysis-data
//ANALYSIS_DATA_DIR = System.getProperty("analysis.data.dir", "");
ANALYSIS_DATA_DIR = SystemProperties.GetProperty("analysis.data.dir", "");
if (ANALYSIS_DATA_DIR.Length != 0)
{
return;
}
#if NETSTANDARD1_6
string currentPath = System.AppContext.BaseDirectory;
#else
string currentPath = AppDomain.CurrentDomain.BaseDirectory;
#endif
//FileInfo[] cadidateFiles = new FileInfo[] { new FileInfo(currentPath + "/" + dirName),
// new FileInfo(currentPath + "/bin/" + dirName)/*, new FileInfo("./" + propName),
// new FileInfo("./lib/" + propName)*/ };
//for (int i = 0; i < cadidateFiles.Length; i++)
//{
// FileInfo file = cadidateFiles[i];
// if (file.Exists)
// {
// ANALYSIS_DATA_DIR = file.FullName;
// //if (file.isDirectory())
// //{
// // ANALYSIS_DATA_DIR = file.getAbsolutePath();
// //}
// //else if (file.isFile() && GetAnalysisDataDir(file).Length != 0)
// //{
// // ANALYSIS_DATA_DIR = GetAnalysisDataDir(file);
// //}
// break;
// }
//}
string candidatePath = System.IO.Path.Combine(currentPath, dirName);
if (Directory.Exists(candidatePath))
{
ANALYSIS_DATA_DIR = candidatePath;
return;
}
try
{
while (new DirectoryInfo(currentPath).Parent != null)
{
candidatePath = System.IO.Path.Combine(new DirectoryInfo(currentPath).Parent.FullName, dirName);
if (Directory.Exists(candidatePath))
{
ANALYSIS_DATA_DIR = candidatePath;
return;
}
currentPath = new DirectoryInfo(currentPath).Parent.FullName;
}
}
catch (SecurityException)
{
// ignore security errors
}
//for (int i = 0; i < cadidateDirectories.Count; i++)
//{
// DirectoryInfo dir = cadidateDirectories[i];
// if (dir.Exists)
// {
// ANALYSIS_DATA_DIR = dir.FullName;
// break;
// }
//}
//if (ANALYSIS_DATA_DIR.Length == 0)
//{
// // Dictionary directory cannot be found.
// throw new Exception("WARNING: Can not find lexical dictionary directory!"
// + " This will cause unpredictable exceptions in your application!"
// + " Please refer to the manual to download the dictionaries.");
//}
}
public virtual void TestXMLConfiguration()
{
// TODO - not working with XML.
Assert.AreEqual("0x00000010", ConfigurationSettings.CurrentConfiguration["tests:seed"]);
Assert.AreEqual("0x00000010", SystemProperties.GetProperty("tests:seed"));
}
