read_df 未找到 #74
Comments
|
很抱歉，由于人手和时间的紧缺，文档中存在许多不足。 import os
from pathlib import Path
from typing import NamedTuple, cast
import pandas as pd
import concurrent.futures
class FileEncoding(NamedTuple):
"""File encoding as the NamedTuple."""
encoding: str | None
"""The encoding of the file."""
confidence: float
"""The confidence of the encoding."""
language: str | None
"""The language of the file."""
def detect_file_encodings(
file_path: str | Path, timeout: int = 5
) -> list[FileEncoding]:
"""Try to detect the file encoding.
Returns a list of `FileEncoding` tuples with the detected encodings ordered
by confidence.
Args:
file_path: The path to the file to detect the encoding for.
timeout: The timeout in seconds for the encoding detection.
"""
import chardet
file_path = str(file_path)
def read_and_detect(file_path: str) -> list[dict]:
with open(file_path, "rb") as f:
rawdata = f.read()
return cast(list[dict], chardet.detect_all(rawdata))
with concurrent.futures.ThreadPoolExecutor() as executor:
future = executor.submit(read_and_detect, file_path)
try:
encodings = future.result(timeout=timeout)
except concurrent.futures.TimeoutError:
raise TimeoutError(
f"Timeout reached while detecting encoding for {file_path}"
)
if all(encoding["encoding"] is None for encoding in encodings):
raise RuntimeError(f"Could not detect encoding for {file_path}")
return [FileEncoding(**enc) for enc in encodings if enc["encoding"] is not None]
def path_from_uri(uri: str) -> Path:
"""Return a new path from the given 'file' URI.
This is implemented in Python 3.13.
See <https://github.com/python/cpython/pull/107640>
and <https://github.com/python/cpython/pull/107640/files#diff-fa525485738fc33d05b06c159172ff1f319c26e88d8c6bb39f7dbaae4dc4105c>
TODO: remove when we migrate to Python 3.13"""
if not uri.startswith("file:"):
raise ValueError(f"URI does not start with 'file:': {uri!r}")
path = uri[5:]
if path[:3] == "///":
# Remove empty authority
path = path[2:]
elif path[:12] == "//localhost/":
# Remove 'localhost' authority
path = path[11:]
if path[:3] == "///" or (path[:1] == "/" and path[2:3] in ":|"):
# Remove slash before DOS device/UNC path
path = path[1:]
if path[1:2] == "|":
# Replace bar with colon in DOS drive
path = path[:1] + ":" + path[2:]
from urllib.parse import unquote_to_bytes
path = Path(os.fsdecode(unquote_to_bytes(path)))
if not path.is_absolute():
raise ValueError(f"URI is not absolute: {uri!r}")
return path
def file_extention(file: str) -> str:
path = Path(file)
return path.suffix
def read_df(uri: str, autodetect_encoding: bool = True, **kwargs) -> pd.DataFrame:
"""A simple wrapper to read different file formats into DataFrame."""
try:
return _read_df(uri, **kwargs)
except UnicodeDecodeError as e:
if autodetect_encoding:
detected_encodings = detect_file_encodings(path_from_uri(uri), timeout=30)
for encoding in detected_encodings:
try:
return _read_df(uri, encoding=encoding.encoding, **kwargs)
except UnicodeDecodeError:
continue
# Either we ran out of detected encoding, or autodetect_encoding is False,
# we should raise encoding error
raise ValueError(f"不支持的文件编码{e.encoding}，请转换成 utf-8 后重试")
def _read_df(uri: str, encoding: str = "utf-8", **kwargs) -> pd.DataFrame:
"""A simple wrapper to read different file formats into DataFrame."""
ext = file_extention(uri).lower()
if ext == ".csv":
df = pd.read_csv(uri, encoding=encoding, **kwargs)
elif ext == ".tsv":
df = pd.read_csv(uri, sep="\t", encoding=encoding, **kwargs)
elif ext in [".xls", ".xlsx", ".xlsm", ".xlsb", ".odf", ".ods", ".odt"]:
# read_excel does not support 'encoding' arg, also it seems that it does not need it.
df = pd.read_excel(uri, **kwargs)
else:
raise ValueError(
f"TableGPT 目前支持 csv、tsv 以及 xlsx 文件，您上传的文件格式 {ext} 暂不支持。"
)
return df |
|
加了这个文件之后好像还是跑不通
|
|
@jianpugh 你好，请问你使用的vllm版本是多少？ |
v0.4.0 |
嗯嗯，请尝试升级vllm试试 |
感谢您的使用与反馈, 经测试, 请保证您的vllm版本 |
|
升级之后确实可以了，感谢指导！ |
|
不好意思，再打扰一下，还有两个问题
|
|
1.event_stream执行的时候会有很多事件流，详情请参考 https://python.langchain.com/docs/how_to/streaming/#event-reference 你可以通过 human_message = HumanMessage(content="How many men survived?")
response = await agent.ainvoke(
input={
# After using checkpoint, you only need to add new messages here.
"messages": [human_message],
"parent_id": "some-parent-id2",
"date": date.today(), # noqa: DTZ011
},
config={
"configurable": {"thread_id": "some-thread-id"},
},
)
print(response["messages"][-1])2.你分析的是一个excel文件，看日志好像是pandas读取excel的相关依赖缺失了，请通过 |
感谢您的解答~ |
请问文件名是什么 |
|
你好，我直接运行examples/quick_start.py文件，报错ModuleNotFoundError: No module named 'tablegpt'。 我该如何正确跑通一次demo呢 |
|
@edc3000 你有先安装 tablegpt 吗？ |
|
@edwardzjl 感谢，安装了tablegpt已经解决！ 但我在尝试跑chat on tablular data的部分，发现日志有这样的报错：ToolMessage(content="Error: NoSuchKernel('python3')\n Please fix your mistakes.", name='python', id='d8cee44a-6895-4172-a928-f555c073e34a', tool_call_id='7e8dcc69-3228-46cf-9998-75331fba2082', status='error'） 这是我缺乏安装什么包导致吗，还是原本生成的代码无法执行的问题 |
|
本地运行应该通过 |
|
@vegetablest 是的，我就是 |
|
你好，我新建了一个全新的python venv，然后执行chat on tablular data没有复现你的问题。 11425 python -m venv venv
11426 source ./venv/bin/activate
11427 pip install langchain-openai
11428 pip install "tablegpt-agent[local]"你可以先尝试运行 |
|
@edc3000 你的问题解决了吗，如果还有问题我想我们应该开一个新的issue，因为你的问题已经与该主题无关了。 |
|
@edwardzjl read_df的讨论已经结束，建议关闭该Issue。 |
跑官方示例 调用tools name 'read_df' is not defined\n```"}
The text was updated successfully, but these errors were encountered: