/// <summary>
/// Extracts a string List from the char** returned from Mitie.
/// </summary>
/// <param name="charArrayPtr">The character array PTR.</param>
/// <returns></returns>
public static List <string> IntPtrToStringArrayAnsi(ref IntPtr charArrayPtr,
bool freeUnmanaged = false)
{
var tokenList = new List <string>();
IntPtr tempPtr = charArrayPtr;
while (true)
{
var stringPtr = (IntPtr)Marshal.PtrToStructure(tempPtr, typeof(IntPtr));
var managedString = Marshal.PtrToStringAnsi(stringPtr);
if (!string.IsNullOrEmpty(managedString))
{
tokenList.Add(managedString);
tempPtr = IntPtr.Add(tempPtr, IntPtr.Size);
// Note: Another way to increment pointer. Above is likely faster:
// charArrayPtr = new IntPtr(charArrayPtr.ToInt64() + IntPtr.Size);
}
else
{
break;
}
}
if (freeUnmanaged == true)
{
MarshaledWrapper.Exec_mitie_free(charArrayPtr);
charArrayPtr = IntPtr.Zero;
}
return(tokenList);
}
public void GetNumberOfNERDetections()
{
// Load the sample text to analyze
IntPtr ptr = new IntPtr();
var sentence = MarshaledWrapper.Exec_mitie_load_entire_file(SAMPLE_TEXT_FILE_NAME,
ref ptr);
// Tokenize the text
IntPtr tokens = new IntPtr();
MarshaledWrapper.Exec_mitie_tokenize(sentence, ref tokens, false);
// Load the sample *trained* ner model for Mitie
IntPtr ner = MarshaledWrapper.Exec_mitie_load_named_entity_extractor(SAMPLE_MITIE_NER_MODEL);
// Calculate the detections from the model
IntPtr detections = MarshaledWrapper.Exec_mitie_extract_entities(ner, tokens);
uint numDetections = MarshaledWrapper.Exec_mitie_ner_get_num_detections(detections);
Assert.True(numDetections == 23);
// Free unmanaged memory
MarshaledWrapper.Exec_mitie_free(tokens);
MarshaledWrapper.Exec_mitie_free(ner);
MarshaledWrapper.Exec_mitie_free(detections);
}
public void LoadEntireFile()
{
// Load the sample text to analyze
IntPtr ptr = new IntPtr();
var sentence = MarshaledWrapper.Exec_mitie_load_entire_file(SAMPLE_TEXT_FILE_NAME,
ref ptr);
Assert.NotNull((object)ptr); // Casted to avoid warning
}
public void GetTokens()
{
// Load the sample text to analyze
// Default overload used here is to release unmanaged memory
IntPtr ptr = new IntPtr();
var sentence = MarshaledWrapper.Exec_mitie_load_entire_file(SAMPLE_TEXT_FILE_NAME,
ref ptr);
IntPtr tokens = new IntPtr();
var tokenList = MarshaledWrapper.Exec_mitie_tokenize(sentence, ref tokens, true);
Assert.True(tokenList.Count == 212);
}
public void LoadNERModel()
{
// Load the sample text to analyze
IntPtr ptr = new IntPtr();
var sentence = MarshaledWrapper.Exec_mitie_load_entire_file(SAMPLE_TEXT_FILE_NAME,
ref ptr);
IntPtr tokens = new IntPtr();
MarshaledWrapper.Exec_mitie_tokenize(sentence, ref tokens, false);
IntPtr ner = MarshaledWrapper.Exec_mitie_load_named_entity_extractor(SAMPLE_MITIE_NER_MODEL);
Assert.NotNull((object)ner);
// Free unmanaged memory
MarshaledWrapper.Exec_mitie_free(tokens);
MarshaledWrapper.Exec_mitie_free(ner);
}
static void Main(string[] args)
{
IntPtr ptr = new IntPtr();
// Load a sentence from a sample file
var sentence = MarshaledWrapper.Exec_mitie_load_entire_file(@"..\..\..\..\sample1.txt",
ref ptr);
Console.WriteLine($"Loaded sentence -> {sentence}\n");
IntPtr tokens = new IntPtr();
IntPtr offsetTokens = new IntPtr();
List <uint> listOffsetTokens = new List <uint>();
// Gte the tokens and token offsets
var listTokens = MarshaledWrapper.Exec_mitie_tokenize_with_offsets(sentence,
ref listOffsetTokens,
ref tokens,
ref offsetTokens,
false);
foreach (var token in listTokens)
{
Console.WriteLine($"Token -> {token}");
}
Console.WriteLine("\n");
// Load the sample *trained* model
IntPtr entityExtractto = MarshaledWrapper.Exec_mitie_load_named_entity_extractor(@"..\..\..\..\ner_model.dat");
// Get the mumber of detections
IntPtr detections = MarshaledWrapper.Exec_mitie_extract_entities(entityExtractto, tokens);
long numDetections = MarshaledWrapper.Exec_mitie_ner_get_num_detections(detections);
Console.WriteLine($"Number of detections -> {numDetections}");
// Free the unmanaged memory
MarshaledWrapper.Exec_mitie_free(tokens);
MarshaledWrapper.Exec_mitie_free(offsetTokens);
MarshaledWrapper.Exec_mitie_free(entityExtractto);
MarshaledWrapper.Exec_mitie_free(detections);
}
/// <summary>
/// Marshals the long array to uint list.
/// </summary>
/// <param name="ptr">The PTR.</param>
/// <param name="length">The length.</param>
/// <param name="freeUnmanaged">if set to <c>true</c> [free unmanaged].</param>
/// <returns></returns>
public static List <uint> MarshalLongArrayToUintList(ref IntPtr ptr,
uint length,
bool freeUnmanaged = false)
{
List <uint> list = new List <uint>();
IntPtr tempPtr = ptr;
for (int i = 0; i < length; i++)
{
list.Add((uint)Marshal.PtrToStructure(tempPtr, typeof(uint)));
tempPtr = IntPtr.Add(tempPtr, IntPtr.Size);
}
if (freeUnmanaged == true)
{
MarshaledWrapper.Exec_mitie_free(ptr);
ptr = IntPtr.Zero;
}
return(list);
}
public void GetTokenOffsets()
{
// Load the sample text to analyze
IntPtr ptr = new IntPtr();
var sentence = MarshaledWrapper.Exec_mitie_load_entire_file(SAMPLE_TEXT_FILE_NAME,
ref ptr);
IntPtr tokens = new IntPtr();
IntPtr offsetTokens = new IntPtr();
List <uint> listOffsetTokens = new List <uint>();
// Tokenize with offsets
MarshaledWrapper.Exec_mitie_tokenize_with_offsets(sentence,
ref listOffsetTokens,
ref tokens,
ref offsetTokens);
Assert.True(listOffsetTokens.Count == 212);
// Free unmanaged memory
MarshaledWrapper.Exec_mitie_free(tokens);
MarshaledWrapper.Exec_mitie_free(offsetTokens);
}
