private void SerializeRecognizeEntitiesResultCollection(ref Utf8JsonWriter json, RecognizeEntitiesResultCollection resultCollection)
{
json.WriteStartObject();
json.WriteStartArray("documents");
if (resultCollection.FirstOrDefault(r => r.Entities.Count > 0) != default)
{
foreach (var result in resultCollection)
{
if (!result.HasError)
{
json.WriteStartObject();
json.WriteString("id", result.Id);
json.WriteStartArray("entities");
foreach (var entity in result.Entities)
{
json.WriteStartObject();
json.WriteString("text", entity.Text);
json.WriteString("type", JsonSerializer.Serialize(entity.Category));
json.WriteString("subtype", JsonSerializer.Serialize(entity.SubCategory));
json.WriteNumber("offset", entity.GraphemeOffset);
json.WriteNumber("length", entity.GraphemeLength);
json.WriteNumber("score", entity.ConfidenceScore);
json.WriteEndObject();
}
json.WriteEndArray();
json.WriteEndObject();
}
}
}
json.WriteEndArray();
json.WriteStartArray("errors");
if (resultCollection.FirstOrDefault(r => r.HasError) != default)
{
foreach (var result in resultCollection)
{
if (result.HasError)
{
json.WriteStartObject();
json.WriteString("id", result.Id);
json.WriteStartObject("error");
json.WriteStartObject("innerError");
json.WriteString("message", result.Error.Message);
json.WriteEndObject();
json.WriteEndObject();
json.WriteEndObject();
}
}
}
json.WriteEndArray();
json.WriteString("modelVersion", resultCollection.ModelVersion);
json.WriteEndObject();
// TODO: add statistics if needed
}
public async Task TextWithStringIndexType()
{
TextAnalyticsClient client = GetClient();
string document = "👨 Microsoft the company.";
RecognizeEntitiesResultCollection responseWithUnicodeCodePoint = await client.RecognizeEntitiesBatchAsync(new List <string>() { document }, "en", new TextAnalyticsRequestOptions()
{
StringIndexType = StringIndexType.UnicodeCodePoint
});
RecognizeEntitiesResultCollection responseWithUtf16CodeUnit = await client.RecognizeEntitiesBatchAsync(new List <string>() { document }, "en");
var entitiesWithUnicodeCodePoint = responseWithUnicodeCodePoint.FirstOrDefault().Entities;
var entitiesWithUtf16CodeUnit = responseWithUtf16CodeUnit.FirstOrDefault().Entities;
Assert.AreEqual(1, entitiesWithUnicodeCodePoint.Count);
Assert.AreEqual("Microsoft", entitiesWithUnicodeCodePoint.FirstOrDefault().Text);
Assert.AreEqual(2, entitiesWithUnicodeCodePoint.FirstOrDefault().Offset);
Assert.AreEqual(9, entitiesWithUnicodeCodePoint.FirstOrDefault().Length);
Assert.AreEqual(1, entitiesWithUtf16CodeUnit.Count);
Assert.AreEqual("Microsoft", entitiesWithUtf16CodeUnit.FirstOrDefault().Text);
Assert.AreEqual(3, entitiesWithUtf16CodeUnit.FirstOrDefault().Offset);
Assert.AreEqual(9, entitiesWithUtf16CodeUnit.FirstOrDefault().Length);
}
public async Task RecognizeEntitiesWithSubCategoryTest()
{
TextAnalyticsRequestOptions options = new TextAnalyticsRequestOptions()
{
ModelVersion = "2020-04-01"
};
TextAnalyticsClient client = GetClient();
string document = "I had a wonderful trip to Seattle last week.";
RecognizeEntitiesResultCollection result = await client.RecognizeEntitiesBatchAsync(new List <string>() { document }, options : options);
var documentResult = result.FirstOrDefault();
Assert.IsFalse(documentResult.HasError);
Assert.GreaterOrEqual(documentResult.Entities.Count, 3);
foreach (CategorizedEntity entity in documentResult.Entities)
{
if (entity.Text == "last week")
{
Assert.AreEqual("DateRange", entity.SubCategory);
}
}
// Assert the options classes since overloads were added and the original now instantiates a RecognizeEntitiesOptions.
Assert.IsFalse(options.IncludeStatistics);
Assert.AreEqual("2020-04-01", options.ModelVersion);
Assert.AreEqual(StringIndexType.Utf16CodeUnit, options.StringIndexType);
}
public async Task RecognizeEntitiesCategories()
{
TextAnalyticsClient client = GetClient();
const string document = "Bill Gates | Microsoft | New Mexico | 800-102-1100 | [email protected] | April 4, 1975 12:34 | April 4, 1975 | 12:34 | five seconds | 9 | third | 120% | €30 | 11m | 22 °C |" +
"Software Engineer | Wedding | Microsoft Surface laptop | Coding | 127.0.0.1 | https://github.com/azure/azure-sdk-for-net";
RecognizeEntitiesResultCollection response = await client.RecognizeEntitiesBatchAsync(new List <string>() { document }, "en", new TextAnalyticsRequestOptions()
{
ModelVersion = "2020-02-01"
});
var entities = response.FirstOrDefault().Entities.ToList();
Assert.AreEqual(21, entities.Count);
Assert.AreEqual(EntityCategory.Person, entities[0].Category);
Assert.AreEqual(EntityCategory.Organization, entities[1].Category);
Assert.AreEqual(EntityCategory.Location, entities[2].Category);
Assert.AreEqual(EntityCategory.PhoneNumber, entities[3].Category);
Assert.AreEqual(EntityCategory.Email, entities[4].Category);
Assert.AreEqual(EntityCategory.DateTime, entities[5].Category);
Assert.AreEqual(EntityCategory.DateTime, entities[6].Category);
Assert.AreEqual(EntityCategory.DateTime, entities[7].Category);
Assert.AreEqual(EntityCategory.DateTime, entities[8].Category);
Assert.AreEqual(EntityCategory.Quantity, entities[9].Category);
Assert.AreEqual(EntityCategory.Quantity, entities[10].Category);
Assert.AreEqual(EntityCategory.Quantity, entities[11].Category);
Assert.AreEqual(EntityCategory.Quantity, entities[12].Category);
Assert.AreEqual(EntityCategory.Quantity, entities[13].Category);
Assert.AreEqual(EntityCategory.Quantity, entities[14].Category);
Assert.AreEqual(EntityCategory.PersonType, entities[15].Category);
Assert.AreEqual(EntityCategory.Event, entities[16].Category);
Assert.AreEqual(EntityCategory.Product, entities[17].Category);
Assert.AreEqual(EntityCategory.Skill, entities[18].Category);
Assert.AreEqual(EntityCategory.IPAddress, entities[19].Category);
Assert.AreEqual(EntityCategory.Url, entities[20].Category);
}
private void SerializeRecognizeEntitiesResultCollection(ref Utf8JsonWriter json, RecognizeEntitiesResultCollection resultCollection)
{
json.WriteStartObject();
json.WriteStartArray("documents");
if (resultCollection.FirstOrDefault(r => r.NamedEntities.Count > 0) != default)
{
foreach (var result in resultCollection)
{
if (result.NamedEntities.Count > 0)
{
json.WriteStartObject();
json.WriteString("id", result.Id);
json.WriteStartArray("entities");
foreach (var entity in result.NamedEntities)
{
json.WriteStartObject();
json.WriteString("text", entity.Text);
json.WriteString("type", entity.Type);
json.WriteString("subtype", entity.SubType);
json.WriteNumber("offset", entity.Offset);
json.WriteNumber("length", entity.Length);
json.WriteNumber("score", entity.Score);
json.WriteEndObject();
}
json.WriteEndArray();
json.WriteEndObject();
}
}
}
json.WriteEndArray();
json.WriteStartArray("errors");
if (resultCollection.FirstOrDefault(r => r.ErrorMessage != default) != default)
{
foreach (var result in resultCollection)
{
if (result.ErrorMessage != null)
{
json.WriteStartObject();
json.WriteString("id", result.Id);
json.WriteStartObject("error");
json.WriteStartObject("innerError");
json.WriteString("message", result.ErrorMessage);
json.WriteEndObject();
json.WriteEndObject();
json.WriteEndObject();
}
}
}
json.WriteEndArray();
json.WriteString("modelVersion", resultCollection.ModelVersion);
json.WriteEndObject();
// TODO: add statistics if needed
}
public async Task EntitiesCategories()
{
TextAnalyticsClient client = GetClient();
const string document = "Bill Gates | Microsoft | New Mexico";
RecognizeEntitiesResultCollection response = await client.RecognizeEntitiesBatchAsync(new List <string>() { document }, "en", new TextAnalyticsRequestOptions()
{
ModelVersion = "2020-02-01"
});
var entities = response.FirstOrDefault().Entities.ToList();
Assert.AreEqual(3, entities.Count);
Assert.AreEqual(EntityCategory.Person, entities[0].Category);
Assert.AreEqual(EntityCategory.Organization, entities[1].Category);
Assert.AreEqual(EntityCategory.Location, entities[2].Category);
}
public async Task RecognizeEntitiesWithSubCategoryTest()
{
TextAnalyticsClient client = GetClient();
string document = "I had a wonderful trip to Seattle last week.";
RecognizeEntitiesResultCollection result = await client.RecognizeEntitiesBatchAsync(new List <string>() { document }, options : new TextAnalyticsRequestOptions()
{
ModelVersion = "2020-04-01"
});
var documentResult = result.FirstOrDefault();
Assert.IsFalse(documentResult.HasError);
Assert.GreaterOrEqual(documentResult.Entities.Count, 3);
foreach (CategorizedEntity entity in documentResult.Entities)
{
if (entity.Text == "last week")
{
Assert.AreEqual("DateRange", entity.SubCategory);
}
}
}
