File size: 9,119 Bytes
a8b3f00 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 |
import threading
from typing import Optional
from flask import Flask, current_app
from core.rag.data_post_processor.data_post_processor import DataPostProcessor
from core.rag.datasource.keyword.keyword_factory import Keyword
from core.rag.datasource.vdb.vector_factory import Vector
from core.rag.rerank.rerank_type import RerankMode
from core.rag.retrieval.retrieval_methods import RetrievalMethod
from extensions.ext_database import db
from models.dataset import Dataset
from services.external_knowledge_service import ExternalDatasetService
default_retrieval_model = {
"search_method": RetrievalMethod.SEMANTIC_SEARCH.value,
"reranking_enable": False,
"reranking_model": {"reranking_provider_name": "", "reranking_model_name": ""},
"top_k": 2,
"score_threshold_enabled": False,
}
class RetrievalService:
@classmethod
def retrieve(
cls,
retrieval_method: str,
dataset_id: str,
query: str,
top_k: int,
score_threshold: Optional[float] = 0.0,
reranking_model: Optional[dict] = None,
reranking_mode: Optional[str] = "reranking_model",
weights: Optional[dict] = None,
):
if not query:
return []
dataset = db.session.query(Dataset).filter(Dataset.id == dataset_id).first()
if not dataset:
return []
if not dataset or dataset.available_document_count == 0 or dataset.available_segment_count == 0:
return []
all_documents = []
threads = []
exceptions = []
# retrieval_model source with keyword
if retrieval_method == "keyword_search":
keyword_thread = threading.Thread(
target=RetrievalService.keyword_search,
kwargs={
"flask_app": current_app._get_current_object(),
"dataset_id": dataset_id,
"query": query,
"top_k": top_k,
"all_documents": all_documents,
"exceptions": exceptions,
},
)
threads.append(keyword_thread)
keyword_thread.start()
# retrieval_model source with semantic
if RetrievalMethod.is_support_semantic_search(retrieval_method):
embedding_thread = threading.Thread(
target=RetrievalService.embedding_search,
kwargs={
"flask_app": current_app._get_current_object(),
"dataset_id": dataset_id,
"query": query,
"top_k": top_k,
"score_threshold": score_threshold,
"reranking_model": reranking_model,
"all_documents": all_documents,
"retrieval_method": retrieval_method,
"exceptions": exceptions,
},
)
threads.append(embedding_thread)
embedding_thread.start()
# retrieval source with full text
if RetrievalMethod.is_support_fulltext_search(retrieval_method):
full_text_index_thread = threading.Thread(
target=RetrievalService.full_text_index_search,
kwargs={
"flask_app": current_app._get_current_object(),
"dataset_id": dataset_id,
"query": query,
"retrieval_method": retrieval_method,
"score_threshold": score_threshold,
"top_k": top_k,
"reranking_model": reranking_model,
"all_documents": all_documents,
"exceptions": exceptions,
},
)
threads.append(full_text_index_thread)
full_text_index_thread.start()
for thread in threads:
thread.join()
if exceptions:
exception_message = ";\n".join(exceptions)
raise Exception(exception_message)
if retrieval_method == RetrievalMethod.HYBRID_SEARCH.value:
data_post_processor = DataPostProcessor(
str(dataset.tenant_id), reranking_mode, reranking_model, weights, False
)
all_documents = data_post_processor.invoke(
query=query, documents=all_documents, score_threshold=score_threshold, top_n=top_k
)
return all_documents
@classmethod
def external_retrieve(cls, dataset_id: str, query: str, external_retrieval_model: Optional[dict] = None):
dataset = db.session.query(Dataset).filter(Dataset.id == dataset_id).first()
if not dataset:
return []
all_documents = ExternalDatasetService.fetch_external_knowledge_retrieval(
dataset.tenant_id, dataset_id, query, external_retrieval_model
)
return all_documents
@classmethod
def keyword_search(
cls, flask_app: Flask, dataset_id: str, query: str, top_k: int, all_documents: list, exceptions: list
):
with flask_app.app_context():
try:
dataset = db.session.query(Dataset).filter(Dataset.id == dataset_id).first()
keyword = Keyword(dataset=dataset)
documents = keyword.search(cls.escape_query_for_search(query), top_k=top_k)
all_documents.extend(documents)
except Exception as e:
exceptions.append(str(e))
@classmethod
def embedding_search(
cls,
flask_app: Flask,
dataset_id: str,
query: str,
top_k: int,
score_threshold: Optional[float],
reranking_model: Optional[dict],
all_documents: list,
retrieval_method: str,
exceptions: list,
):
with flask_app.app_context():
try:
dataset = db.session.query(Dataset).filter(Dataset.id == dataset_id).first()
vector = Vector(dataset=dataset)
documents = vector.search_by_vector(
cls.escape_query_for_search(query),
search_type="similarity_score_threshold",
top_k=top_k,
score_threshold=score_threshold,
filter={"group_id": [dataset.id]},
)
if documents:
if (
reranking_model
and reranking_model.get("reranking_model_name")
and reranking_model.get("reranking_provider_name")
and retrieval_method == RetrievalMethod.SEMANTIC_SEARCH.value
):
data_post_processor = DataPostProcessor(
str(dataset.tenant_id), RerankMode.RERANKING_MODEL.value, reranking_model, None, False
)
all_documents.extend(
data_post_processor.invoke(
query=query, documents=documents, score_threshold=score_threshold, top_n=len(documents)
)
)
else:
all_documents.extend(documents)
except Exception as e:
exceptions.append(str(e))
@classmethod
def full_text_index_search(
cls,
flask_app: Flask,
dataset_id: str,
query: str,
top_k: int,
score_threshold: Optional[float],
reranking_model: Optional[dict],
all_documents: list,
retrieval_method: str,
exceptions: list,
):
with flask_app.app_context():
try:
dataset = db.session.query(Dataset).filter(Dataset.id == dataset_id).first()
vector_processor = Vector(
dataset=dataset,
)
documents = vector_processor.search_by_full_text(cls.escape_query_for_search(query), top_k=top_k)
if documents:
if (
reranking_model
and reranking_model.get("reranking_model_name")
and reranking_model.get("reranking_provider_name")
and retrieval_method == RetrievalMethod.FULL_TEXT_SEARCH.value
):
data_post_processor = DataPostProcessor(
str(dataset.tenant_id), RerankMode.RERANKING_MODEL.value, reranking_model, None, False
)
all_documents.extend(
data_post_processor.invoke(
query=query, documents=documents, score_threshold=score_threshold, top_n=len(documents)
)
)
else:
all_documents.extend(documents)
except Exception as e:
exceptions.append(str(e))
@staticmethod
def escape_query_for_search(query: str) -> str:
return query.replace('"', '\\"')
|