使用 DingoClient 连接 DingoDB

为提升速度，DingoDB 提供功能全面而强大的API，支持对数据库进行 DDL 或 DML 操作。

示例

Dingo sdk 的功能：SDK 文档

1. Document_index

from dingodb import SDKDocumentDingoDB, SDKClient
from dingodb.common.document_rep import DocumentType, DocumentColumn, DocumentSchema


addrs = "127.0.0.1:22001,127.0.0.1:22002,127.0.0.1:22003"
sdk_client = SDKClient(addrs)

x = SDKDocumentDingoDB(sdk_client)
print(x)

index_name = "document_index_test"

删除索引

delete_index_out = x.delete_index(index_name)
print(delete_index_out)
import time

创建索引

scheme =  DocumentSchema()
col = DocumentColumn("text", DocumentType.STRING)
scheme.add_document_column(col)
col = DocumentColumn("i64", DocumentType.INT64)
scheme.add_document_column(col)
col = DocumentColumn("f64", DocumentType.DOUBLE)
scheme.add_document_column(col)
col = DocumentColumn("bytes", DocumentType.BYTES)
scheme.add_document_column(col)
col = DocumentColumn("bool", DocumentType.BOOL)
scheme.add_document_column(col)
col = DocumentColumn("datetime", DocumentType.DATETIME)
scheme.add_document_column(col)

create_index_out = x.create_index(index_name, scheme, 3, operand=[5, 10, 20])
# create_index_out = x.create_index(index_name, scheme, 3)
print(create_index_out)
time.sleep(5)

创建训练数据和查询数据

ids = [3, 5, 7, 9, 11, 13, 15, 17, 19, 21]
documents = [
    {"text" : "Ancient empires rise and fall, shaping history's course.", "i64" : 1003, "f64" : 1003.0, "bytes" : "bytes_data_3","bool":True, "datetime": "2021-01-01T00:00:00Z"},
    {"text" : "Artistic expressions reflect diverse cultural heritages.", "i64" : 1005, "f64" : 1005.0, "bytes" : "bytes_data_5","bool":False, "datetime": "2021-01-01T00:00:00Z"},
    {"text" : "Social movements transform societies, forging new paths.", "i64" : 1007, "f64" : 1007.0, "bytes" : "bytes_data_7","bool":True, "datetime": "2022-01-01T00:00:00Z"},
    {"text" : "Economies fluctuate, reflecting the complex interplay of global forces.", "i64" : 1009, "f64" : 1009.0, "bytes" : "bytes_data_9","bool":False, "datetime": "2022-01-01T00:00:00Z"},
    {"text" : "Strategic military campaigns alter the balance of power.", "i64" : 1011, "f64" : 1011.0, "bytes" : "bytes_data_11","bool":True, "datetime": "2023-01-01T00:00:00Z"},
    {"text" : "Quantum leaps redefine understanding of physical laws.", "i64" : 1013, "f64" : 1013.0, "bytes" : "bytes_data_13","bool":False, "datetime": "2023-01-01T00:00:00Z"},
    {"text" : "Chemical reactions unlock mysteries of nature.", "i64" : 1015, "f64" : 1015.0, "bytes" : "bytes_data_15","bool":True, "datetime": "2024-01-01T00:00:00Z"},
    {"text" : "Philosophical debates ponder the essence of existence.", "i64" : 1017, "f64" : 1017.0, "bytes" : "bytes_data_17","bool":False, "datetime": "2024-01-01T00:00:00Z"},
    {"text" : "Marriages blend traditions, celebrating love's union.", "i64" : 1019, "f64" : 1019.0, "bytes" : "bytes_data_19","bool":True, "datetime": "2025-01-01T00:00:00Z"},
    {"text" : "Explorers discover uncharted territories, expanding world maps.", "i64" : 1021, "f64" : 10021.0, "bytes" : "bytes_data_21","bool":False, "datetime": "2025-01-01T00:00:00Z"}
]

document_add_out = x.document_add(index_name, documents, ids)
print(document_add_out)
print(document_add_out.to_dict())

文档搜索

document_search_out = x.document_search(index_name, "discover", 5, with_scalar_data=True)
print(document_search_out)
print(document_search_out.to_dict())

document_search_out = x.document_search(index_name, "of", 3, with_scalar_data=True)
print(document_search_out)

document_search_out = x.document_search(index_name, "of", 5,[13, 15], with_scalar_data=True)
print(document_search_out)

document_search_out = x.document_search_all(index_name, "of", with_scalar_data=True, query_limit=4096)
print(document_search_out)

document_search_out = x.document_search(index_name, r"(text:'of' AND i64: >= 1013)", 5,  [9, 11, 13, 15], with_scalar_data=True)
print(document_search_out)

document_search_out = x.document_search(index_name, r"( bool:true)", 5,  [9, 11, 13, 15], with_scalar_data=True)
print(document_search_out)

document_search_out = x.document_search(index_name, r"(datetime:'2023-01-01T00:00:00Z' )", 5,  [3,5,7,9, 11, 13, 15], with_scalar_data=True)
print(document_search_out)

文档查询

document_query_out = x.document_query(index_name, ids, True, ["text", "i64"])
print(document_query_out)

获取文本边框

document_get_border_out = x.document_get_border(index_name, True)
print(document_get_border_out)
document_get_border_out = x.document_get_border(index_name, False)
print(document_get_border_out)

文档扫描查询

document_scan_query_out = x.document_scan_query(index_name, ids[0], ids[-1], False, 2)
print(document_scan_query_out)
document_scan_query_out = x.document_scan_query(index_name, ids[-1], ids[0], True, 2)
print(document_scan_query_out)

document_scan_query_out = x.document_scan_query(index_name, ids[0], ids[-1] + 10, False, 100, True, ["text", "i64"])
print(document_scan_query_out)

文档索引度量

document_metrics_out = x.document_index_metrics(index_name)
print(document_metrics_out)

文本数量

document_count_out = x.document_count(index_name, 0, 19)
print(document_count_out)

文本删除

document_delete_out = x.document_delete(index_name, ids)
print(document_delete_out)

2. Document_regex_index

导入

from dingodb import SDKDocumentDingoDB, SDKClient
from dingodb.common.document_rep import DocumentType, DocumentColumn, DocumentSchema


addrs = "127.0.0.1:22001,127.0.0.1:22002,127.0.0.1:22003"
sdk_client = SDKClient(addrs)

x = SDKDocumentDingoDB(sdk_client)
print(x)

index_name = "document_regex_index_test"

删除索引

delete_index_out = x.delete_index(index_name)
print(delete_index_out)

创建索引

import time

scheme =  DocumentSchema()
col = DocumentColumn("title", DocumentType.STRING)
scheme.add_document_column(col)
col = DocumentColumn("text", DocumentType.STRING)
scheme.add_document_column(col)

create_index_out = x.create_index(index_name, scheme, 3, operand=[5, 10, 20])
# create_index_out = x.create_index(index_name, scheme, 3)
print(create_index_out)
time.sleep(5)

创建训练数据和查询数据

ids = [1, 2, 3]
documents = [
    {"title" : "a", "text" : "The Diary of Muadib"},
    {"title" : "bb", "text" : "A Dairy Cow"},
    {"title" : "ccc", "text" : "The Diary of a Young Girl"}
]

document_add_out = x.document_add(index_name, documents, ids)
print(document_add_out)

全文索引的正则查询

#  base64encode Dia.* to RGlhLioq
#  text contains "Dia"
document_search_out = x.document_search(index_name, "text:RE [RGlhLio=]", 5, False, with_scalar_data=True)
print(document_search_out)

文本搜索长度

#  base64encode (.{0,2})  to KC57MCwyfSk=
#  title length <= 2
document_search_out = x.document_search(index_name, "title:RE [KC57MCwyfSk=]", 5, False, with_scalar_data=True)
print(document_search_out)

文件搜索

#  base64encode (.{0,2})  to KC57MCwyfSk=
#  base64encode Dia.* to RGlhLioq
#  title length <= 2 and text contains "Dia"
document_search_out = x.document_search(index_name, "title:RE [KC57MCwyfSk=] AND text:RE [RGlhLio=]", 5, False, with_scalar_data=True)
print(document_search_out)

3. Rawkv

导入

import numpy as np
import os

from dingodb import SDKRawKVDingoDB, SDKClient

# need to create region ( range(wa,wc) for this example ) before using SDKRawKVDingoDB

addrs = "127.0.0.1:22001,127.0.0.1:22002,127.0.0.1:22003"
sdk_client = SDKClient(addrs)
x = SDKRawKVDingoDB(sdk_client)
print(x)

输入

x.rawkv_put("wb01", "value1")

批量输入

x.rawkv_batch_put([
    ("wb02", "value2"),
    ("wb03", "value3"),
    ("wb04", "value4"),
    ("wb05", "value5"),
    ("wb06", "value6"),
    ("wb07", "value7"),
    ("wb08", "value8"),
    ("wb09", "value9"),
    ("wb10", "value10"),
    ("wb11", "value11"),
    ("wb12", "value12"),
    ("wb13", "value13"),
    ("wb14", "value14"),
    ("wb15", "value15")])

获取

x.rawkv_get("wb01")

转为字典输出

[kv.to_dict() for kv in x.rawkv_batch_get([
    "wb01",
    "wb02",
    "wb03",
    "wb04",
    "wb05",
    "wb06",
    "wb07",
    "wb08",
    "wb09",
    "wb10",
    "wb11",
    "wb12",
    "wb13",
    "wb14",
    "wb15"
])]

x.rawkv_put_if_absent("wb16", "value16")

删除

x.rawkv_delete("wb16")

x.rawkv_batch_delete([
    "wb01",
    "wb02",
])

[kv.to_dict() for kv in x.rawkv_batch_put_if_absent([
    ("wb01", "value1"),
    ("wb02", "value2"),
    ("wb03", "value3"),
    ("wb04", "value4"),
    ("wb05", "value5"),
    ("wb06", "value6"),
    ("wb07", "value7"),
    ("wb08", "value8"),
    ("wb09", "value9"),
    ("wb10", "value10"),
    ("wb11", "value11"),
    ("wb12", "value12"),
    ("wb13", "value13"),
    ("wb14", "value14"),
    ("wb15", "value15")])]

[kv.to_dict() for kv in x.rawkv_scan("wb", "wc", 15)]

删除分区

x.rawkv_delete_range("wa", "wc")

4. Region-creator

导入

import numpy as np
import os

from dingodb import SDKRegionCreatorDingoDB, SDKClient

addrs = "127.0.0.1:22001,127.0.0.1:22002,127.0.0.1:22003"
sdk_client = SDKClient(addrs)
x = SDKRegionCreatorDingoDB(sdk_client)
print(x)

删除分区

x.drop_region(80031)

创建分区

x.create_region_id(10)

x.create_region("test1","wb00000000","wc000000",80033)

5. Vector_index

import numpy as np
import os

from dingodb import SDKVectorDingoDB, SDKClient
from dingodb.common.vector_rep import ScalarType, ScalarColumn, ScalarSchema

addrs = "127.0.0.1:22001,127.0.0.1:22002,127.0.0.1:22003"
sdk_client = SDKClient(addrs)
x = SDKVectorDingoDB(sdk_client)
print(x)

index_name = "test_index_grpc1"

删除索引

x.delete_index(index_name)

创建索引

# help(x.create_index)
#x.create_index(index_name, 8, "binary_flat", "hamming", 3, operand=[5,10,15,20])
col = ScalarColumn("id",ScalarType.DOUBLE,True)
sca = ScalarSchema()
sca.add_scalar_column(col)
x.create_index_with_schema(index_name, 16,sca, "binary_ivf_flat", "hamming", 3, operand=[100,500,1500,3000,6000])

创建训练数据和查询数据

d = 16                           # dimension
bd = 2                     #binary dimension
nb = 4                      # database size
np.random.seed(1234)             # make reproducible
xb = np.random.randint(0, 255, (nb, bd))  # 生成范围为 0-255 的随机整数
print(xb.shape)
xb[:, 0] += np.arange(nb) 
print(xb)
print(xb.shape)

ids = [1, 2, 3, 4]
datas = [{"id": 50}, {"id": 120}, {"id": 130}, {"id": 4.40}]
vectors = xb.tolist()

新增

for i in range(10):
    x.vector_add(index_name, datas, vectors, ids,"binary")

删除

x.vector_delete(index_name,ids)

自增ID获取

x.vector_get_auto_increment_id(index_name)

自增ID更新

x.vector_update_auto_increment_id(index_name,16)

插入

x.vector_upsert(index_name, datas, vectors, ids,"binary")

搜索

# vector_search
x.vector_search(index_name, vectors[0],value_type="binary")
# return 
    # error RuntimeError

使用 pre_filter 或 post_filter 搜索

x.vector_search(index_name, vectors[0], 10, {"meta_expr": {"id": 1}},value_type="binary")

用 id 获取索引

x.vector_get(index_name, [1, 2, 6])

添加：扫描

x.vector_scan(index_name, 20, 60,is_reverse=True,end_id=0)

添加计数

x.vector_count(index_name)

度量

x.vector_metrics(index_name)

获取最大索引行数

x.get_max_index_row(index_name)

删除索引

# delete_index
x.delete_index(index_name)

6. Vector_binary_index

导入

import numpy as np
import os

from dingodb import SDKVectorDingoDB, SDKClient
from dingodb.common.vector_rep import ScalarType, ScalarColumn, ScalarSchema

addrs = "127.0.0.1:22001,127.0.0.1:22002,127.0.0.1:22003"
sdk_client = SDKClient(addrs)
x = SDKVectorDingoDB(sdk_client)
print(x)

index_name = "test_index_grpc1"

删除索引

x.delete_index(index_name)

创建索引

# help(x.create_index)
#x.create_index(index_name, 8, "binary_flat", "hamming", 3, operand=[5,10,15,20])
col = ScalarColumn("id",ScalarType.DOUBLE,True)
sca = ScalarSchema()
sca.add_scalar_column(col)
x.create_index_with_schema(index_name, 16,sca, "binary_ivf_flat", "hamming", 3, operand=[100,500,1500,3000,6000])

创建训练数据和查询数据

d = 16                           # dimension
bd = 2                     #binary dimension
nb = 4                      # database size
np.random.seed(1234)             # make reproducible
xb = np.random.randint(0, 255, (nb, bd))  # 生成范围为 0-255 的随机整数
print(xb.shape)
xb[:, 0] += np.arange(nb) 
print(xb)
print(xb.shape)

ids = [1, 2, 3, 4]
datas = [{"id": 50}, {"id": 120}, {"id": 130}, {"id": 4.40}]
vectors = xb.tolist()

向量添加

for i in range(10):
    x.vector_add(index_name, datas, vectors, ids,"binary")

向量删除

x.vector_delete(index_name,ids)

获取自增

x.vector_get_auto_increment_id(index_name)

向量序列自增

x.vector_update_auto_increment_id(index_name,16)

向量插入

x.vector_upsert(index_name, datas, vectors, ids,"binary")

向量搜索

x.vector_search(index_name, vectors[0],value_type="binary")
# return 
    # error RuntimeError

# vector_search with pre_filter or post_filter
x.vector_search(index_name, vectors[0], 10, {"meta_expr": {"id": 1}},value_type="binary")

向量获取

x.vector_get(index_name, [1, 2, 6])

向量扫描

x.vector_scan(index_name, 20, 60,is_reverse=True,end_id=0)

添加向量计数

x.vector_count(index_name)

向量度量

x.vector_metrics(index_name)

获取最大索引行数

x.get_max_index_row(index_name)

删除索引

x.delete_index(index_name)

7.Vector_diskann_index

导入

import numpy as np
import os

from dingodb import SDKVectorDingoDB, SDKClient
from dingodb.common.vector_rep import ScalarType, ScalarColumn, ScalarSchema

addrs = "172.30.14.11:22001,172.30.14.11:22002,172.30.14.11:22003"
sdk_client = SDKClient(addrs)
x = SDKVectorDingoDB(sdk_client)
print(x)

index_name = "test_index_grpc"

删除索引

x.delete_index(index_name)

创建索引

# help(x.create_index)
x.create_index(index_name, 6, "diskann", "euclidean", 3, index_config={"valueType": "float","searchListSize": 100,"maxDegree": 64}, operand=[5,10,15,20])

创建训练数据和查询数据

d = 6                           # dimension
nb = 4                      # database size
np.random.seed(1234)             # make reproducible
xb = np.random.random((nb, d)).astype('float32')
print(xb)
print(xb.shape)
xb[:, 0] += np.arange(nb) / 1000.
print(xb)
print(xb.shape)

ids = [1, 2, 3, 4]
datas = [{"a1": "b1"}, {"a2": "b2"}, {"a3": "b3"}, {"a4": "b4"}]
vectors = xb.tolist()

向量添加

x.vector_import_add(index_name, datas, vectors, ids)
x.vector_import_add(index_name, datas, vectors, ids)
x.vector_import_add(index_name, datas, vectors, ids)
x.vector_import_add(index_name, datas, vectors, ids)

按索引构建向量

x.vector_build_by_index(index_name)

转为字典输出

[r.to_dict() for r in x.vector_status_by_index(index_name)]

内存向量计数

x.vector_count_memory(index_name)

向量搜索

x.vector_search(index_name, vectors[0])

转为字典输出

[r.to_dict() for r in x.vector_load_by_index(index_name)]

[r.to_dict() for r in x.vector_reset_by_index(index_name)]

id = [1,80001]
[r.to_dict() for r in x.vector_status_by_region(index_name,id)]

[r.to_dict() for r in x.vector_build_by_region(index_name,id)]

[r.to_dict() for r in x.vector_load_by_region(index_name,id)]

[r.to_dict() for r in x.vector_reset_by_region(index_name,id)]

删除

x.vector_import_delete(index_name,ids)

x.delete_index(index_name)