ai_pipeline/tests/pipeline/core/test_nodes.py

159 lines
4.3 KiB
Python

import asyncio
from platform import node
import pytest
import json
from src.pipeline.core.pocket_flow import AsyncFlow
from src.pipeline.utils import logger
from src.pipeline.core import llm, es, nodes
from src.pipeline.core.skills import load_skills
from src.pipeline import utils
@pytest.fixture(scope="session")
async def init_llm():
await llm.init_client()
yield
await llm.close_client()
@pytest.fixture(scope="session")
async def init_es():
await es.init_client()
yield
await es.close_client()
@pytest.mark.asyncio
async def test_embedding(init_llm, init_es):
logger.debug("file to es")
shared = {
"files": [ "./files/西游记.txt"],
"documents": [], # [{text, file_name, file_type, uuid, embedding}]
"index": "test_kb_1",
}
readNode = nodes.ReadDocumentsNode()
chunkNode = nodes.ChunkDocumentsNode()
embeddingNode = nodes.EmbeddingDocumentsNode()
writeToESNode = nodes.WriteDocumentsToESNode()
readNode >> chunkNode >> embeddingNode >> writeToESNode
flow = AsyncFlow(readNode)
await flow.run_async(shared)
logger.debug(json.dumps([{**x, "embedding":x["embedding"][:4]} for x in shared["documents"]], indent=4, ensure_ascii=False))
@pytest.mark.asyncio
async def test_search(init_llm, init_es):
await llm.init_client()
await es.init_client()
logger.debug("search from es")
shared = {
"text": "哪里盛产矿石",
"index": "test_kb",
"top_k": 10,
"results": [], # [{es_id, doc_id, title, type, created_at, score, content}]
}
embeddingNode = nodes.EmbeddingNode()
searchNode = nodes.SearchNode()
embeddingNode >> searchNode
flow = AsyncFlow(embeddingNode)
await flow.run_async(shared)
logger.debug(json.dumps({**shared, "embedding": shared["embedding"][:4]}, indent=4, ensure_ascii=False))
request = llm.client.stream_chat(
messages=[
{"role": "system", "content": utils.rag_system_prompt()},
{"role": "user", "content": utils.rag_user_prompt(shared["text"], shared["results"])},
]
)
async for chunk in request:
logger.debug(chunk)
await llm.close_client()
await es.close_client()
@pytest.mark.asyncio
async def test_rerank(init_llm, init_es):
await llm.init_client()
await es.init_client()
logger.debug("search from es")
shared = {
"text": "山海经中描述了哪里盛产矿石",
"index": "test_kb",
"top_k": 10,
"search_web": False,
"rerank_top_k": 5,
"top_n": 3,
"results": [], # [{es_id, doc_id, title, type, created_at, score, content}]
}
embeddingNode = nodes.EmbeddingNode()
searchNode = nodes.SearchNode()
rerankNode = nodes.RerankNode()
embeddingNode >> searchNode >> rerankNode
# embeddingNode >> searchNode
flow = AsyncFlow(embeddingNode)
await flow.run_async(shared)
logger.debug(json.dumps({**shared, "embedding": shared["embedding"][:4]}, indent=4, ensure_ascii=False))
res = await llm.client.chat(
messages=[
{"role": "system", "content": utils.rag_system_prompt()},
{"role": "user", "content": utils.rag_user_prompt(shared["text"], shared["results"][:shared["top_n"]])},
]
)
logger.debug(res)
await llm.close_client()
await es.close_client()
@pytest.mark.asyncio
async def test_agent(init_llm, init_es):
await llm.init_client()
await es.init_client()
shared = {
"query": "山海经中描述了哪里盛产矿石",
"skills": await load_skills("./src/skills"),
}
decideNode = nodes.DecideSkillNode()
refineNode = nodes.RefineSkillNode()
disambiguateNode = nodes.DisambiguateSkillNode()
runNode = nodes.ExecuteSkillNode()
decideNode >> refineNode
refineNode >> runNode
disambiguateNode >> runNode
refineNode - "disambiguate" >> disambiguateNode
flow = AsyncFlow(decideNode)
await flow.run_async(shared)
res = await llm.client.chat(
messages=[
{"role": "system", "content": utils.rag_system_prompt()},
{"role": "user", "content": utils.rag_user_prompt(shared["query"], shared["results"])},
]
)
logger.debug(res)
await llm.close_client()
await es.close_client()