pengjunjie/ragflow
1
0
Fork 1
mirror of https://github.com/infiniflow/ragflow.git synced 2025-12-08 20:42:30 +08:00
Code Issues Packages Projects Releases Wiki Activity

Fix: Merge main branch (#10377)

Browse Source 
### What problem does this PR solve?


### Type of change

- [x] Bug Fix (non-breaking change which fixes an issue)

---------

Signed-off-by: dependabot[bot] <support@github.com>
Signed-off-by: jinhai <haijin.chn@gmail.com>
Signed-off-by: Jin Hai <haijin.chn@gmail.com>
Co-authored-by: Lynn <lynn_inf@hotmail.com>
Co-authored-by: chanx <1243304602@qq.com>
Co-authored-by: balibabu <cike8899@users.noreply.github.com>
Co-authored-by: 纷繁下的无奈 <zhileihuang@126.com>
Co-authored-by: huangzl <huangzl@shinemo.com>
Co-authored-by: writinwaters <93570324+writinwaters@users.noreply.github.com>
Co-authored-by: Wilmer <33392318@qq.com>
Co-authored-by: Adrian Weidig <adrianweidig@gmx.net>
Co-authored-by: Zhichang Yu <yuzhichang@gmail.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Yongteng Lei <yongtengrey@outlook.com>
Co-authored-by: Liu An <asiro@qq.com>
Co-authored-by: buua436 <66937541+buua436@users.noreply.github.com>
Co-authored-by: BadwomanCraZY <511528396@qq.com>
Co-authored-by: cucusenok <31804608+cucusenok@users.noreply.github.com>
Co-authored-by: Russell Valentine <russ@coldstonelabs.org>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Billy Bao <newyorkupperbay@gmail.com>
Co-authored-by: Zhedong Cen <cenzhedong2@126.com>
Co-authored-by: TensorNull <129579691+TensorNull@users.noreply.github.com>
Co-authored-by: TensorNull <tensor.null@gmail.com>
Co-authored-by: Ajay <160579663+aybanda@users.noreply.github.com>
Co-authored-by: AB <aj@Ajays-MacBook-Air.local>
Co-authored-by: 天海蒼灆 <huangaoqin@tecpie.com>
Co-authored-by: He Wang <wanghechn@qq.com>
Co-authored-by: Atsushi Hatakeyama <atu729@icloud.com>
Co-authored-by: Jin Hai <haijin.chn@gmail.com>
Co-authored-by: Mohamed Mathari <155896313+melmathari@users.noreply.github.com>
Co-authored-by: Mohamed Mathari <nocodeventure@Mac-mini-van-Mohamed.fritz.box>
Co-authored-by: Stephen Hu <stephenhu@seismic.com>
Co-authored-by: Shaun Zhang <zhangwfjh@users.noreply.github.com>
Co-authored-by: zhimeng123 <60221886+zhimeng123@users.noreply.github.com>
Co-authored-by: mxc <mxc@example.com>
Co-authored-by: Dominik Novotný <50611433+SgtMarmite@users.noreply.github.com>
Co-authored-by: EVGENY M <168018528+rjohny55@users.noreply.github.com>
Co-authored-by: mcoder6425 <mcoder64@gmail.com>
Co-authored-by: TeslaZY <TeslaZY@outlook.com>
Co-authored-by: lemsn <lemsn@msn.com>
Co-authored-by: lemsn <lemsn@126.com>
Co-authored-by: Adrian Gora <47756404+adagora@users.noreply.github.com>
Co-authored-by: Womsxd <45663319+Womsxd@users.noreply.github.com>
Co-authored-by: FatMii <39074672+FatMii@users.noreply.github.com>
This commit is contained in:
Kevin Hu
2025-09-30 13:13:15 +08:00
committed by GitHub
parent 4d6ff672eb
commit 20b577a72c
201 changed files with 7929 additions and 1110 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
rag/llm/__init__.py
View File

@ -68,6 +68,7 @@ FACTORY_DEFAULT_BASE_URL = {
SupportedLiteLLMProvider.Lingyi_AI: "https://api.lingyiwanwu.com/v1",
SupportedLiteLLMProvider.GiteeAI: "https://ai.gitee.com/v1/",
SupportedLiteLLMProvider.AI_302: "https://api.302.ai/v1",
SupportedLiteLLMProvider.Anthropic: "https://api.anthropic.com/",
}

106
rag/llm/chat_model.py
View File

@ -36,7 +36,7 @@ from zhipuai import ZhipuAI
from rag.llm import FACTORY_DEFAULT_BASE_URL, LITELLM_PROVIDER_PREFIX, SupportedLiteLLMProvider
from rag.nlp import is_chinese, is_english
from rag.utils import num_tokens_from_string
from rag.utils import num_tokens_from_string, total_token_count_from_response
# Error message constants
@ -143,9 +143,10 @@ class Base(ABC):
logging.info("[HISTORY]" + json.dumps(history, ensure_ascii=False, indent=2))
if self.model_name.lower().find("qwen3") >= 0:
kwargs["extra_body"] = {"enable_thinking": False}
response = self.client.chat.completions.create(model=self.model_name, messages=history, **gen_conf, **kwargs)
if any([not response.choices, not response.choices[0].message, not response.choices[0].message.content]):
if not response.choices or not response.choices[0].message or not response.choices[0].message.content:
return "", 0
ans = response.choices[0].message.content.strip()
if response.choices[0].finish_reason == "length":
@ -155,10 +156,12 @@ class Base(ABC):
def _chat_streamly(self, history, gen_conf, **kwargs):
logging.info("[HISTORY STREAMLY]" + json.dumps(history, ensure_ascii=False, indent=4))
reasoning_start = False
if kwargs.get("stop") or "stop" in gen_conf:
response = self.client.chat.completions.create(model=self.model_name, messages=history, stream=True, **gen_conf, stop=kwargs.get("stop"))
else:
response = self.client.chat.completions.create(model=self.model_name, messages=history, stream=True, **gen_conf)
for resp in response:
if not resp.choices:
continue
@ -190,21 +193,30 @@ class Base(ABC):
return ans + LENGTH_NOTIFICATION_CN
return ans + LENGTH_NOTIFICATION_EN
def _exceptions(self, e, attempt):
@property
def _retryable_errors(self) -> set[str]:
return {
LLMErrorCode.ERROR_RATE_LIMIT,
LLMErrorCode.ERROR_SERVER,
}
def _should_retry(self, error_code: str) -> bool:
return error_code in self._retryable_errors
def _exceptions(self, e, attempt) -> str | None:
logging.exception("OpenAI chat_with_tools")
# Classify the error
error_code = self._classify_error(e)
if attempt == self.max_retries:
error_code = LLMErrorCode.ERROR_MAX_RETRIES
# Check if it's a rate limit error or server error and not the last attempt
should_retry = error_code == LLMErrorCode.ERROR_RATE_LIMIT or error_code == LLMErrorCode.ERROR_SERVER
if not should_retry:
return f"{ERROR_PREFIX}: {error_code} - {str(e)}"
if self._should_retry(error_code):
delay = self._get_delay()
logging.warning(f"Error: {error_code}. Retrying in {delay:.2f} seconds... (Attempt {attempt + 1}/{self.max_retries})")
time.sleep(delay)
return None
delay = self._get_delay()
logging.warning(f"Error: {error_code}. Retrying in {delay:.2f} seconds... (Attempt {attempt + 1}/{self.max_retries})")
time.sleep(delay)
return f"{ERROR_PREFIX}: {error_code} - {str(e)}"
def _verbose_tool_use(self, name, args, res):
return "<tool_call>" + json.dumps({"name": name, "args": args, "result": res}, ensure_ascii=False, indent=2) + "</tool_call>"
@ -445,15 +457,7 @@ class Base(ABC):
yield total_tokens
def total_token_count(self, resp):
try:
return resp.usage.total_tokens
except Exception:
pass
try:
return resp["usage"]["total_tokens"]
except Exception:
pass
return 0
return total_token_count_from_response(resp)
def _calculate_dynamic_ctx(self, history):
"""Calculate dynamic context window size"""
@ -541,6 +545,14 @@ class AzureChat(Base):
self.client = AzureOpenAI(api_key=api_key, azure_endpoint=base_url, api_version=api_version)
self.model_name = model_name
@property
def _retryable_errors(self) -> set[str]:
return {
LLMErrorCode.ERROR_RATE_LIMIT,
LLMErrorCode.ERROR_SERVER,
LLMErrorCode.ERROR_QUOTA,
}
class BaiChuanChat(Base):
_FACTORY_NAME = "BaiChuan"
@ -629,6 +641,10 @@ class ZhipuChat(Base):
def _clean_conf(self, gen_conf):
if "max_tokens" in gen_conf:
del gen_conf["max_tokens"]
gen_conf = self._clean_conf_plealty(gen_conf)
return gen_conf
def _clean_conf_plealty(self, gen_conf):
if "presence_penalty" in gen_conf:
del gen_conf["presence_penalty"]
if "frequency_penalty" in gen_conf:
@ -636,22 +652,14 @@ class ZhipuChat(Base):
return gen_conf
def chat_with_tools(self, system: str, history: list, gen_conf: dict):
if "presence_penalty" in gen_conf:
del gen_conf["presence_penalty"]
if "frequency_penalty" in gen_conf:
del gen_conf["frequency_penalty"]
gen_conf = self._clean_conf_plealty(gen_conf)
return super().chat_with_tools(system, history, gen_conf)
def chat_streamly(self, system, history, gen_conf={}, **kwargs):
if system and history and history[0].get("role") != "system":
history.insert(0, {"role": "system", "content": system})
if "max_tokens" in gen_conf:
del gen_conf["max_tokens"]
if "presence_penalty" in gen_conf:
del gen_conf["presence_penalty"]
if "frequency_penalty" in gen_conf:
del gen_conf["frequency_penalty"]
gen_conf = self._clean_conf(gen_conf)
ans = ""
tk_count = 0
try:
@ -677,11 +685,7 @@ class ZhipuChat(Base):
yield tk_count
def chat_streamly_with_tools(self, system: str, history: list, gen_conf: dict):
if "presence_penalty" in gen_conf:
del gen_conf["presence_penalty"]
if "frequency_penalty" in gen_conf:
del gen_conf["frequency_penalty"]
gen_conf = self._clean_conf_plealty(gen_conf)
return super().chat_streamly_with_tools(system, history, gen_conf)
@ -858,6 +862,7 @@ class MistralChat(Base):
return gen_conf
def _chat(self, history, gen_conf={}, **kwargs):
gen_conf = self._clean_conf(gen_conf)
response = self.client.chat(model=self.model_name, messages=history, **gen_conf)
ans = response.choices[0].message.content
if response.choices[0].finish_reason == "length":
@ -870,9 +875,7 @@ class MistralChat(Base):
def chat_streamly(self, system, history, gen_conf={}, **kwargs):
if system and history and history[0].get("role") != "system":
history.insert(0, {"role": "system", "content": system})
for k in list(gen_conf.keys()):
if k not in ["temperature", "top_p", "max_tokens"]:
del gen_conf[k]
gen_conf = self._clean_conf(gen_conf)
ans = ""
total_tokens = 0
try:
@ -1302,10 +1305,6 @@ class LiteLLMBase(ABC):
"302.AI",
]
import litellm
litellm._turn_on_debug()
def __init__(self, key, model_name, base_url=None, **kwargs):
self.timeout = int(os.environ.get("LM_TIMEOUT_SECONDS", 600))
self.provider = kwargs.get("provider", "")
@ -1429,21 +1428,30 @@ class LiteLLMBase(ABC):
return ans + LENGTH_NOTIFICATION_CN
return ans + LENGTH_NOTIFICATION_EN
def _exceptions(self, e, attempt):
@property
def _retryable_errors(self) -> set[str]:
return {
LLMErrorCode.ERROR_RATE_LIMIT,
LLMErrorCode.ERROR_SERVER,
}
def _should_retry(self, error_code: str) -> bool:
return error_code in self._retryable_errors
def _exceptions(self, e, attempt) -> str | None:
logging.exception("OpenAI chat_with_tools")
# Classify the error
error_code = self._classify_error(e)
if attempt == self.max_retries:
error_code = LLMErrorCode.ERROR_MAX_RETRIES
# Check if it's a rate limit error or server error and not the last attempt
should_retry = error_code == LLMErrorCode.ERROR_RATE_LIMIT or error_code == LLMErrorCode.ERROR_SERVER
if not should_retry:
return f"{ERROR_PREFIX}: {error_code} - {str(e)}"
if self._should_retry(error_code):
delay = self._get_delay()
logging.warning(f"Error: {error_code}. Retrying in {delay:.2f} seconds... (Attempt {attempt + 1}/{self.max_retries})")
time.sleep(delay)
return None
delay = self._get_delay()
logging.warning(f"Error: {error_code}. Retrying in {delay:.2f} seconds... (Attempt {attempt + 1}/{self.max_retries})")
time.sleep(delay)
return f"{ERROR_PREFIX}: {error_code} - {str(e)}"
def _verbose_tool_use(self, name, args, res):
return "<tool_call>" + json.dumps({"name": name, "args": args, "result": res}, ensure_ascii=False, indent=2) + "</tool_call>"

2
rag/llm/cv_model.py
View File

@ -25,7 +25,7 @@ from openai import OpenAI
from openai.lib.azure import AzureOpenAI
from zhipuai import ZhipuAI
from rag.nlp import is_english
from rag.prompts import vision_llm_describe_prompt
from rag.prompts.generator import vision_llm_describe_prompt
from rag.utils import num_tokens_from_string

18
rag/llm/embedding_model.py
View File

@ -33,7 +33,7 @@ from zhipuai import ZhipuAI
from api import settings
from api.utils.file_utils import get_home_cache_dir
from api.utils.log_utils import log_exception
from rag.utils import num_tokens_from_string, truncate
from rag.utils import num_tokens_from_string, truncate, total_token_count_from_response
class Base(ABC):
@ -52,15 +52,7 @@ class Base(ABC):
raise NotImplementedError("Please implement encode method!")
def total_token_count(self, resp):
try:
return resp.usage.total_tokens
except Exception:
pass
try:
return resp["usage"]["total_tokens"]
except Exception:
pass
return 0
return total_token_count_from_response(resp)
class DefaultEmbedding(Base):
@ -497,7 +489,6 @@ class MistralEmbed(Base):
def encode_queries(self, text):
import time
import random
retry_max = 5
while retry_max > 0:
try:
@ -662,7 +653,7 @@ class OpenAI_APIEmbed(OpenAIEmbed):
def __init__(self, key, model_name, base_url):
if not base_url:
raise ValueError("url cannot be None")
#base_url = urljoin(base_url, "v1")
base_url = urljoin(base_url, "v1")
self.client = OpenAI(api_key=key, base_url=base_url)
self.model_name = model_name.split("___")[0]
@ -945,6 +936,7 @@ class GiteeEmbed(SILICONFLOWEmbed):
base_url = "https://ai.gitee.com/v1/embeddings"
super().__init__(key, model_name, base_url)
class DeepInfraEmbed(OpenAIEmbed):
_FACTORY_NAME = "DeepInfra"
@ -963,7 +955,7 @@ class Ai302Embed(Base):
super().__init__(key, model_name, base_url)
class CometEmbed(OpenAIEmbed):
class CometAPIEmbed(OpenAIEmbed):
_FACTORY_NAME = "CometAPI"
def __init__(self, key, model_name, base_url="https://api.cometapi.com/v1"):

17
rag/llm/rerank_model.py
View File

@ -30,7 +30,7 @@ from yarl import URL
from api import settings
from api.utils.file_utils import get_home_cache_dir
from api.utils.log_utils import log_exception
from rag.utils import num_tokens_from_string, truncate
from rag.utils import num_tokens_from_string, truncate, total_token_count_from_response
class Base(ABC):
def __init__(self, key, model_name, **kwargs):
@ -44,18 +44,7 @@ class Base(ABC):
raise NotImplementedError("Please implement encode method!")
def total_token_count(self, resp):
if hasattr(resp, "usage") and hasattr(resp.usage, "total_tokens"):
try:
return resp.usage.total_tokens
except Exception:
pass
if 'usage' in resp and 'total_tokens' in resp['usage']:
try:
return resp["usage"]["total_tokens"]
except Exception:
pass
return 0
return total_token_count_from_response(resp)
class DefaultRerank(Base):
@ -365,7 +354,7 @@ class OpenAI_APIRerank(Base):
max_rank = np.max(rank)
# Avoid division by zero if all ranks are identical
if np.isclose(min_rank, max_rank, atol=1e-3):
if not np.isclose(min_rank, max_rank, atol=1e-3):
rank = (rank - min_rank) / (max_rank - min_rank)
else:
rank = np.zeros_like(rank)

2
rag/llm/sequence2txt_model.py
View File

@ -236,7 +236,7 @@ class DeepInfraSeq2txt(Base):
self.model_name = model_name
class CometSeq2txt(Base):
class CometAPISeq2txt(Base):
_FACTORY_NAME = "CometAPI"
def __init__(self, key, model_name="whisper-1", base_url="https://api.cometapi.com/v1", **kwargs):