public DocumentFactory(AnnotationSet annotationSet)
{
if (annotationSet == null)
throw new ArgumentNullException("annotationSet");
_annotationSet = annotationSet;
}
static void saveReport(WEMDocument document, string fileName,
AnnotationSet annotationSet)
{
using (StreamWriter writer = new StreamWriter(
new FileStream(fileName, FileMode.OpenOrCreate)))
{
foreach (var sen in document.Sentences)
{
foreach (var word in sen.Words)
{
if (word is Entity)
{
Entity entity = (Entity)word;
writer.Write(String.Format("[{0} \\ {1}] ",
entity.Content,
annotationSet[entity.EntityId]));
}
else
{
writer.Write(word.Content + ' ');
}
}
writer.WriteLine();
}
writer.Close();
}
}
public TextDocumentReader(AnnotationSet annotationSet)
{
_sententceFactory = new SentenceFactory(annotationSet);
InputLanguage = Language.SimplifiedChinese;
OutputEncoding = Encoding.UTF8;
}
/// <summary>
/// 初始化一个标注器的基类。
/// </summary>
/// <param name="annotationSet">该标注器所使用的标注集。</param>
protected AnnotatorBase(AnnotationSet annotationSet, Model model)
{
_annotationSet = annotationSet;
_model = model;
_initialized = false;
_isRunning = false;
}
static void Main(string[] args)
{
//try
//{
PFRDocumentReader reader = new PFRDocumentReader();
AnnotationSet wemas = new AnnotationSet();
wemas.Load("WEMAS.xml");
var dir = Directory.CreateDirectory("Results")
.CreateSubdirectory("PFR");
var repDir = Directory.CreateDirectory("Reports")
.CreateSubdirectory("PFR");
var comDir = Directory.CreateDirectory("Compare");
WEMDocument[] docs = null;
//using (Stream stream
// = new FileStream(
// "199801.txt", FileMode.Open, FileAccess.Read))
//{
// docs = reader.ReadDocuments(stream);
//}
//foreach (var doc in docs)
//{
// doc.Save(Path.Combine(dir.FullName, doc.Name + ".xml"));
// saveReport(
// doc,
// Path.Combine(repDir.FullName, doc.Name + ".txt"),
// wemas);
//}
DocumentFactory factory = new DocumentFactory(wemas);
docs = factory.GetDocumentsFromXml(
from docFile in dir.GetFiles()
select docFile.FullName);
foreach (var doc in docs)
{
doc.Save(Path.Combine(
comDir.FullName,
doc.Name + ".xml"));
}
//}
//catch (Exception e)
//{
// Console.WriteLine(e.Message);
//}
Console.WriteLine("Done.");
Console.ReadKey(true);
}
/// <summary>
/// Creates a new annotation set. Caller must have WRITE permission for theassociated dataset.The following fields are required: * datasetId * referenceSetIdAll other fields may be optionally specified, unless documented as beingserver-generated (for example, the `id` field).
/// Documentation https://developers.google.com/genomics/v1/reference/annotationsets/create
/// Generation Note: This does not always build corectly. Google needs to standardise things I need to figuer out which ones are wrong.
/// </summary>
/// <param name="service">Authenticated Genomics service.</param>
/// <param name="body">A valid Genomics v1 body.</param>
/// <returns>AnnotationSetResponse</returns>
public static AnnotationSet Create(GenomicsService service, AnnotationSet body)
{
try
{
// Initial validation.
if (service == null)
{
throw new ArgumentNullException("service");
}
if (body == null)
{
throw new ArgumentNullException("body");
}
// Make the request.
return(service.Annotationsets.Create(body).Execute());
}
catch (Exception ex)
{
throw new Exception("Request Annotationsets.Create failed.", ex);
}
}
public void Start()
{
if (startCheck())
{
AnnotationSet wemas;
ICTCLASAnnotator ictclasAno;
CRFPPAnnotator crfppAno;
Corpus corpus;
Console.BackgroundColor = ConsoleColor.DarkBlue;
Console.Clear();
writeTitle();
ConsoleColor color = Console.ForegroundColor;
Console.ForegroundColor = ConsoleColor.Green;
try
{
Console.WriteLine(MessageInitializing);
wemas = new AnnotationSet();
wemas.Load(_annotationSet);
ictclasAno = new ICTCLASAnnotator(wemas, null);
crfppAno = new CRFPPAnnotator(wemas,
new Model(_crfppModel));
crfppAno.SetCRFPPRootPath(_crfppDir);
corpus = new Corpus(Encoding.GetEncoding(_xmlEncoding));
Console.Write(MessageLoadingCorpus);
savePosition();
var documents = from file in Directory.GetFiles(_corpusDir)
where file.ToUpper().EndsWith(".TXT")
select file;
if (documents.Count() == 0)
{
Console.WriteLine();
Console.WriteLine(MessageEmptyCorpus);
Console.WriteLine(MessageDone);
Console.ForegroundColor = color;
pause();
return;
}
_sentenceFactory
= new SentenceFactory(wemas);
_sentenceFactory.InputLanguage
= Language.SimplifiedChinese;
_sentenceFactory.OutputEncoding = corpus.Encoding;
int count = 0;
foreach (var doc in documents)
{
Sentence[] sens = _sentenceFactory.GetSentences(
new FileStream(doc, FileMode.Open));
WEMDocument wDoc = new WEMDocument(corpus.Encoding);
wDoc.Name = doc;
foreach (var sen in sens)
{
wDoc.AddSentence(sen);
}
corpus.AddDocument(wDoc);
rewrite(String.Format("{0}%", (int)((double)
++count / documents.Count() * 100)));
}
Console.WriteLine("共加载{0}篇文档。",
corpus.Documents.Count);
Console.Write(MessageAnnotatingSeg);
savePosition();
bool done = false;
ictclasAno.AnnotationProgressChanged += (s, ea) =>
{
rewrite(String.Format("{0}%",
Math.Round(ea.ProgressInDouble, 2) * 100));
};
ictclasAno.AnnotationCompleted += (s, ea) =>
{
if (ea.Error == null)
{
rewrite("100%");
Console.WriteLine(MessageAnnotatingEnt);
crfppAno.AnnotationProgressChanged += (sen, args) =>
{
Console.WriteLine(args.Message);
};
crfppAno.AnnotationCompleted += (sen, args) =>
{
if (args.Error == null)
{
Console.WriteLine(
MessageAnnotationFinished);
if (_reportsEnabled)
Console.Write(MessageSavingDocsAndReps);
else
Console.Write(MessageSavingDocuments);
savePosition();
int sCount = 0;
foreach (var doc in corpus.Documents)
{
string reportFileName;
string fileName = documents.ElementAt(
corpus.Documents.IndexOf(doc));
FileInfo fi = new FileInfo(fileName);
reportFileName = fileName = fi.Name;
fileName = fileName.Remove(
fileName.Length - 3);
fileName = _outputDir + fileName
+ "xml";
doc.Save(fileName);
if (_reportsEnabled)
saveReport(doc,
_reportsDir + reportFileName,
wemas);
rewrite(String.Format("{0}%",
(int)((double)++sCount
/ corpus.Documents.Count * 100)));
}
Console.WriteLine(MessageDone);
done = true;
}
else
{
Console.ForegroundColor = ConsoleColor.Red;
Console.WriteLine();
Console.WriteLine(MessageUnhandledException);
Console.WriteLine(args.Error.Message);
Console.WriteLine(args.Error.StackTrace);
done = true;
}
};
if (crfppAno.Initialize())
{
crfppAno.ProcessCorpusAsync(corpus);
}
else
{
Console.ForegroundColor = ConsoleColor.Red;
Console.WriteLine(MessageEntInitFailed);
done = true;
return;
}
}
else
{
Console.ForegroundColor = ConsoleColor.Red;
Console.WriteLine();
Console.WriteLine(MessageUnhandledException);
Console.WriteLine(ea.Error.Message);
Console.WriteLine(ea.Error.StackTrace);
done = true;
}
};
if (ictclasAno.Initialize())
{
ictclasAno.ProcessCorpusAsync(corpus);
while (!done)
Thread.Sleep(50);
}
else
{
Console.ForegroundColor = ConsoleColor.Red;
Console.WriteLine(MessageSegInitFailed);
}
}
catch (Exception ex)
{
Console.ForegroundColor = ConsoleColor.Red;
Console.WriteLine();
Console.WriteLine(MessageUnhandledException);
Console.WriteLine(ex.Message);
}
Console.ForegroundColor = color;
//pause();
return;
}
else
{
//();
return;
}
}
static void ProcessCommand(string command)
{
string[] blocks = command.Split();
switch (blocks[0].ToLower())
{
case "":
return;
case "exit":
if (blocks.Length == 1)
_exitFlag = true;
else
Console.WriteLine("Unknown Parameters.");
return;
case "clear":
if (blocks.Length == 1)
Console.Clear();
else
Console.WriteLine("Unknown Parameters.");
return;
case "genas":
AnnotationDictionary dict = new AnnotationDictionary();
AnnotationSet ans = new AnnotationSet("WEMAS");
dict.Load("WEMAS.xml");
foreach (var entry in dict._Dictionary)
{
ans.SetAnnotationDescription(entry.Key, entry.Value);
}
foreach (var sep in dict.SentenceSeparators)
{
ans.AddSentenceSeparator(sep);
}
ans.Save("WEMAS.wemas");
return;
case "testas":
AnnotationSet ans2 = new AnnotationSet("WEMAS");
ans2.Load("WEMAS.wemas");
ans2.Description =
"WEB ENTITY MINER Default Annotation Set.";
ans2.Save("WEMAS2.wemas");
return;
case "testws":
AnnotationSet wemas = new AnnotationSet();
wemas.Load("WEMAS.wemas");
ICTCLASAnnotator ano = new ICTCLASAnnotator(wemas, null);
CRFPPAnnotator crfAno = new CRFPPAnnotator(wemas,
new Model(
AppDomain.CurrentDomain.BaseDirectory
+ "model.crfppmodel"));
Corpus c = new Corpus(Encoding.UTF8);
SentenceFactory.AnnotationSet = wemas;
SentenceFactory.InputLanguage = Language.SimplifiedChinese;
SentenceFactory.OutputEncoding = Encoding.UTF8;
var sens = SentenceFactory.GetSentences(
new FileStream("utf8.txt", FileMode.Open, FileAccess.Read));
WEMDocument doc = new WEMDocument(Encoding.UTF8);
foreach (var sen in sens)
{
doc.AddSentence(sen);
}
c.AddDocument(doc);
try
{
ano.ProcessCorpus(c);
crfAno.ProcessCorpus(c);
}
catch (Exception ex)
{
Console.WriteLine("Unhandled Exception:\n{0}",
ex.Message);
}
doc.Save("RESULT.xml");
foreach (var sententce in doc.Sentences)
{
foreach (var word in sententce.Words)
{
if (word is Entity)
{
Console.WriteLine("{0}/ENTITY:{1}",
word.Content,
wemas[((Entity)word).EntityId]);
}
else
{
Console.WriteLine("{0}", word.Content);
}
}
}
return;
default:
Console.WriteLine("Unknown Command: '{0}'.", blocks[0]);
return;
}
}
/// <summary>
/// Updates an annotation set. The update must respect all mutabilityrestrictions and other invariants described on the annotation set resource.Caller must have WRITE permission for the associated dataset.
/// Documentation https://developers.google.com/genomics/v1/reference/annotationsets/update
/// Generation Note: This does not always build corectly. Google needs to standardise things I need to figuer out which ones are wrong.
/// </summary>
/// <param name="service">Authenticated Genomics service.</param>
/// <param name="annotationSetId">The ID of the annotation set to be updated.</param>
/// <param name="body">A valid Genomics v1 body.</param>
/// <param name="optional">Optional paramaters.</param>
/// <returns>AnnotationSetResponse</returns>
public static AnnotationSet Update(GenomicsService service, string annotationSetId, AnnotationSet body, AnnotationsetsUpdateOptionalParms optional = null)
{
try
{
// Initial validation.
if (service == null)
{
throw new ArgumentNullException("service");
}
if (body == null)
{
throw new ArgumentNullException("body");
}
if (annotationSetId == null)
{
throw new ArgumentNullException(annotationSetId);
}
// Building the initial request.
var request = service.Annotationsets.Update(body, annotationSetId);
// Applying optional parameters to the request.
request = (AnnotationsetsResource.UpdateRequest)SampleHelpers.ApplyOptionalParms(request, optional);
// Requesting data.
return(request.Execute());
}
catch (Exception ex)
{
throw new Exception("Request Annotationsets.Update failed.", ex);
}
}
public Task Update(string skillId, string annotationId, AnnotationSet set)
{
return(Client.Update(skillId, annotationId, set));
}
// Constructor: Set readonlies, get other metaobjects
internal ProjectionMetaObject()
{
annotations = new AnnotationSet();
behaviors = new BehaviorSet ();
}
private static AnnotationSet AnnotationSet(params object[] annotations)
{
var set = new AnnotationSet();
foreach (var annotation in annotations)
set.Apply(annotation);
return set;
}
public ICTCLASAnnotator(AnnotationSet annotationSet, Model model)
: base(annotationSet, model)
{
}
public CRFPPAnnotator(AnnotationSet annotationSet, Model model)
: base(annotationSet, model)
{
CRFPPHelper.RootPath =
AppDomain.CurrentDomain.BaseDirectory + "CRFPP";
}