zhanghongyu_0626
70aa861983
- Added new modules for advisor compliance, KYC sessions, and script templates, enhancing the advisor agent's capabilities. - Implemented a comprehensive API structure under the `/api/advisor-agent` prefix, ensuring clear organization and access to new features. - Established database models and repositories for compliance rules and KYC sessions, facilitating robust data management. - Integrated exception handling and response models to improve error management and user feedback. - Updated settings to include new configurations for compliance and KYC features, ensuring flexibility and adaptability. This update significantly expands the advisor agent's functionality, providing essential tools for compliance and customer interaction while maintaining a structured API design.
66 lines
2.1 KiB
Python
66 lines
2.1 KiB
Python
"""向量化：Ollama bge-m3，1024 维（客服 KB 灌库 / fin_* 检索用）。
|
||
|
||
薄封装 `app.service.embedding`，与 T-21 共用同一 Ollama 配置。
|
||
"""
|
||
|
||
from __future__ import annotations
|
||
|
||
import httpx
|
||
|
||
from app.config.settings import settings
|
||
from app.service import embedding
|
||
|
||
|
||
class EmbeddingError(RuntimeError):
|
||
pass
|
||
|
||
|
||
class OllamaEmbeddingTool:
|
||
"""顾问话术向量：直连 Ollama（与 Embedder 共用 settings）。"""
|
||
|
||
def __init__(
|
||
self,
|
||
*,
|
||
base_url: str | None = None,
|
||
model: str | None = None,
|
||
expected_dim: int | None = None,
|
||
timeout_seconds: float | None = None,
|
||
) -> None:
|
||
self.base_url = (base_url or settings.ollama_base_url).rstrip("/")
|
||
self.model = model or settings.embed_model
|
||
self.expected_dim = expected_dim or settings.embed_dim
|
||
self.timeout_seconds = timeout_seconds or settings.embed_timeout_seconds
|
||
|
||
def embed_text(self, text: str) -> list[float]:
|
||
try:
|
||
response = httpx.post(
|
||
f"{self.base_url}/api/embeddings",
|
||
json={"model": self.model, "prompt": text},
|
||
timeout=self.timeout_seconds,
|
||
)
|
||
response.raise_for_status()
|
||
body = response.json()
|
||
except httpx.HTTPError as exc:
|
||
raise EmbeddingError(f"Ollama embedding request failed: {exc}") from exc
|
||
|
||
vector = body.get("embedding")
|
||
if not isinstance(vector, list):
|
||
raise EmbeddingError("Ollama embedding response missing embedding list")
|
||
if len(vector) != self.expected_dim:
|
||
raise EmbeddingError(f"Embedding dimension must be {self.expected_dim}, got {len(vector)}")
|
||
return [float(value) for value in vector]
|
||
|
||
|
||
class Embedder:
|
||
"""批量/单条 embedding（build_collections / test_search 调用面）。"""
|
||
|
||
def embed(self, text: str) -> list[float]:
|
||
return embedding.embed_text(text)
|
||
|
||
def embed_batch(self, texts: list[str]) -> list[list[float]]:
|
||
return embedding.embed_texts(texts)
|
||
|
||
|
||
def get_embedder() -> Embedder:
|
||
return Embedder()
|