Added datapoint bulk insert

This commit is contained in:
2026-01-28 22:21:51 +01:00
parent 5869eeabd6
commit 329af1c103
2 changed files with 71 additions and 7 deletions

View File

@@ -39,6 +39,19 @@ public class DatabaseHelper(ILogger<DatabaseHelper> logger)
helper.ExecuteSQLNonQuery(query.ToString(), parameters); helper.ExecuteSQLNonQuery(query.ToString(), parameters);
} }
public static int DatabaseInsertEmbeddingBulk(SQLHelper helper, List<(string hash, string model, byte[] embedding)> data)
{
return helper.BulkExecuteNonQuery(
"INSERT INTO embedding (id_datapoint, model, embedding) SELECT d.id, @model, @embedding FROM datapoint d WHERE d.hash = @hash",
data.Select(element => new object[] {
new MySqlParameter("@model", element.model),
new MySqlParameter("@embedding", element.embedding),
new MySqlParameter("@hash", element.hash)
})
);
}
public static int DatabaseInsertSearchdomain(SQLHelper helper, string name, SearchdomainSettings settings = new()) public static int DatabaseInsertSearchdomain(SQLHelper helper, string name, SearchdomainSettings settings = new())
{ {
Dictionary<string, dynamic> parameters = new() Dictionary<string, dynamic> parameters = new()
@@ -83,6 +96,20 @@ public class DatabaseHelper(ILogger<DatabaseHelper> logger)
); );
} }
public static int DatabaseInsertDatapoints(SQLHelper helper, List<(string name, ProbMethodEnum probmethod_embedding, SimilarityMethodEnum similarityMethod, string hash)> values, int id_entity)
{
return helper.BulkExecuteNonQuery(
"INSERT INTO datapoint (name, probmethod_embedding, similaritymethod, hash, id_entity) VALUES (@name, @probmethod_embedding, @similaritymethod, @hash, @id_entity)",
values.Select(element => new object[] {
new MySqlParameter("@name", element.name),
new MySqlParameter("@probmethod_embedding", element.probmethod_embedding),
new MySqlParameter("@similaritymethod", element.similarityMethod),
new MySqlParameter("@hash", element.hash),
new MySqlParameter("@id_entity", id_entity)
})
);
}
public static int DatabaseInsertDatapoint(SQLHelper helper, string name, ProbMethodEnum probmethod_embedding, SimilarityMethodEnum similarityMethod, string hash, int id_entity) public static int DatabaseInsertDatapoint(SQLHelper helper, string name, ProbMethodEnum probmethod_embedding, SimilarityMethodEnum similarityMethod, string hash, int id_entity)
{ {
Dictionary<string, dynamic> parameters = new() Dictionary<string, dynamic> parameters = new()
@@ -155,7 +182,7 @@ public class DatabaseHelper(ILogger<DatabaseHelper> logger)
helper.ExecuteSQLNonQuery("DELETE embedding.* FROM embedding JOIN datapoint dp ON id_datapoint = dp.id JOIN entity ON id_entity = entity.id WHERE entity.id_searchdomain = @searchdomain", parameters); helper.ExecuteSQLNonQuery("DELETE embedding.* FROM embedding JOIN datapoint dp ON id_datapoint = dp.id JOIN entity ON id_entity = entity.id WHERE entity.id_searchdomain = @searchdomain", parameters);
helper.ExecuteSQLNonQuery("DELETE datapoint.* FROM datapoint JOIN entity ON id_entity = entity.id WHERE entity.id_searchdomain = @searchdomain", parameters); helper.ExecuteSQLNonQuery("DELETE datapoint.* FROM datapoint JOIN entity ON id_entity = entity.id WHERE entity.id_searchdomain = @searchdomain", parameters);
helper.ExecuteSQLNonQuery("DELETE attribute.* FROM attribute JOIN entity ON id_entity = entity.id WHERE entity.id_searchdomain = @searchdomain", parameters); helper.ExecuteSQLNonQuery("DELETE FROM attribute WHERE id_entity IN (SELECT entity.id FROM entity WHERE id_searchdomain = @searchdomain)", parameters);
return helper.ExecuteSQLNonQuery("DELETE FROM entity WHERE entity.id_searchdomain = @searchdomain", parameters); return helper.ExecuteSQLNonQuery("DELETE FROM entity WHERE entity.id_searchdomain = @searchdomain", parameters);
} }

View File

@@ -245,25 +245,30 @@ public class SearchdomainHelper(ILogger<SearchdomainHelper> logger, DatabaseHelp
else else
{ {
int id_entity = DatabaseHelper.DatabaseInsertEntity(helper, jsonEntity.Name, jsonEntity.Probmethod, _databaseHelper.GetSearchdomainID(helper, jsonEntity.Searchdomain)); int id_entity = DatabaseHelper.DatabaseInsertEntity(helper, jsonEntity.Name, jsonEntity.Probmethod, _databaseHelper.GetSearchdomainID(helper, jsonEntity.Searchdomain));
List<(string attribute, string value, int id_entity)> values = []; List<(string attribute, string value, int id_entity)> toBeInsertedAttributes = [];
foreach (KeyValuePair<string, string> attribute in jsonEntity.Attributes) foreach (KeyValuePair<string, string> attribute in jsonEntity.Attributes)
{ {
values.Add(new() { toBeInsertedAttributes.Add(new() {
attribute = attribute.Key, attribute = attribute.Key,
value = attribute.Value, value = attribute.Value,
id_entity = id_entity id_entity = id_entity
}); });
} }
DatabaseHelper.DatabaseInsertAttributes(helper, values); DatabaseHelper.DatabaseInsertAttributes(helper, toBeInsertedAttributes);
List<Datapoint> datapoints = []; List<Datapoint> datapoints = [];
List<(JSONDatapoint datapoint, string hash)> toBeInsertedDatapoints = [];
foreach (JSONDatapoint jsonDatapoint in jsonEntity.Datapoints) foreach (JSONDatapoint jsonDatapoint in jsonEntity.Datapoints)
{ {
string hash = Convert.ToBase64String(SHA256.HashData(Encoding.UTF8.GetBytes(jsonDatapoint.Text))); string hash = Convert.ToBase64String(SHA256.HashData(Encoding.UTF8.GetBytes(jsonDatapoint.Text)));
Datapoint datapoint = DatabaseInsertDatapointWithEmbeddings(helper, searchdomain, jsonDatapoint, id_entity, hash); toBeInsertedDatapoints.Add(new()
datapoints.Add(datapoint); {
datapoint = jsonDatapoint,
hash = hash
});
} }
List<Datapoint> datapoint = DatabaseInsertDatapointsWithEmbeddings(helper, searchdomain, toBeInsertedDatapoints, id_entity);
var probMethod = Probmethods.GetMethod(jsonEntity.Probmethod) ?? throw new ProbMethodNotFoundException(jsonEntity.Probmethod); var probMethod = Probmethods.GetMethod(jsonEntity.Probmethod) ?? throw new ProbMethodNotFoundException(jsonEntity.Probmethod);
Entity entity = new(jsonEntity.Attributes, probMethod, jsonEntity.Probmethod.ToString(), datapoints, jsonEntity.Name) Entity entity = new(jsonEntity.Attributes, probMethod, jsonEntity.Probmethod.ToString(), datapoints, jsonEntity.Name)
{ {
@@ -276,6 +281,38 @@ public class SearchdomainHelper(ILogger<SearchdomainHelper> logger, DatabaseHelp
} }
} }
public List<Datapoint> DatabaseInsertDatapointsWithEmbeddings(SQLHelper helper, Searchdomain searchdomain, List<(JSONDatapoint datapoint, string hash)> values, int id_entity)
{
List<Datapoint> result = [];
List<(string name, ProbMethodEnum probmethod_embedding, SimilarityMethodEnum similarityMethod, string hash)> toBeInsertedDatapoints = [];
List<(string hash, string model, byte[] embedding)> toBeInsertedEmbeddings = [];
foreach ((JSONDatapoint datapoint, string hash) value in values)
{
Datapoint datapoint = BuildDatapointFromJsonDatapoint(value.datapoint, id_entity, searchdomain, value.hash);
toBeInsertedDatapoints.Add(new()
{
name = datapoint.name,
probmethod_embedding = datapoint.probMethod.probMethodEnum,
similarityMethod = datapoint.similarityMethod.similarityMethodEnum,
hash = value.hash
});
foreach ((string, float[]) embedding in datapoint.embeddings)
{
toBeInsertedEmbeddings.Add(new()
{
hash = value.hash,
model = embedding.Item1,
embedding = BytesFromFloatArray(embedding.Item2)
});
}
result.Add(datapoint);
}
DatabaseHelper.DatabaseInsertDatapoints(helper, toBeInsertedDatapoints, id_entity);
DatabaseHelper.DatabaseInsertEmbeddingBulk(helper, toBeInsertedEmbeddings);
return result;
}
public Datapoint DatabaseInsertDatapointWithEmbeddings(SQLHelper helper, Searchdomain searchdomain, JSONDatapoint jsonDatapoint, int id_entity, string? hash = null) public Datapoint DatabaseInsertDatapointWithEmbeddings(SQLHelper helper, Searchdomain searchdomain, JSONDatapoint jsonDatapoint, int id_entity, string? hash = null)
{ {
if (jsonDatapoint.Text is null) if (jsonDatapoint.Text is null)