使用 LangChain 的 UnstructedURLLoader 加载数据因“TP_NUM_C_BUFS 太小：50”而停止

Question

我正在尝试复制 LangChain 文档中提供的代码（URL - LangChain 0.0.167），以便能够将 HTML 文件从 URL 列表加载到文档格式中，然后可以由复杂的自然语言处理模型进行处理以执行下游任务。但是，我遇到了一个问题，代码url_data = url_loader.load()挂起半个多小时而没有加载任何 HTML 文件。

我还遇到了堆栈跟踪，并且无法解释该错误的错误消息TP_NUM_C_BUFS too small: 50。此错误之前已在 LangChain 存储库中报告为已解决的问题（链接）。该问题的作者报告说，执行之前TP_NUM_C_BUFS too small: 50在 Windows 命令提示符下导致错误的脚本解决了该问题。但是，在 Windows 命令提示符下执行我的脚本并没有解决问题。

有人能够找出这个问题的根源并提供解决方案吗？

微量元素

from langchain.document_loaders import UnstructuredURLLoader
import session_info

session_info.show()

urls = [
    "https://www.understandingwar.org/backgrounder/russian-offensive-campaign-assessment-february-8-2023",
    "https://www.understandingwar.org/backgrounder/russian-offensive-campaign-assessment-february-9-2023",
]

print(urls)

loader = UnstructuredURLLoader(urls=urls)

print(loader)

data = loader.load()

print(data)

MWE执行结果

D:\path>C:/Python310/python.exe d:/path/src/langchain-url-mwe.py
-----
langchain           0.0.157
session_info        1.0.0
-----
Python 3.10.8 (tags/v3.10.8:aaaf517, Oct 11 2022, 16:50:30) [MSC v.1933 64 bit (AMD64)]
Windows-10-10.0.19045-SP0
-----
Session information updated at 2023-05-14 21:22
['https://www.understandingwar.org/backgrounder/russian-offensive-campaign-assessment-february-8-2023', 'https://www.understandingwar.org/backgrounder/russian-offensive-campaign-assessment-february-9-2023']
<langchain.document_loaders.url.UnstructuredURLLoader object at 0x0000022A495A7C40>
      0 [main] python (12524) C:\Python310\python.exe: *** fatal error - Internal error: TP_NUM_C_BUFS too small: 50
Stack trace:
Frame        Function    Args
045573E7140  001800629AE (001802AEB6B, 00180274E41, 00180318880, 045573E4A70)
045573E7140  0018004846A (00000000020, 045573E5C94, 00000000001, 00040400000)
045573E7140  001800484A2 (00000000032, 000000010A1, 00180318880, 045573E5B70)
045573E7140  00180168194 (00061948FE1, 00000000000, 045573E5CA0, 045571C3000)
045573E7140  00180106358 (00000000000, 008000010A1, 00180318880, 22A4BD5DBC0)
045573E7140  0018005ADC3 (045573E7140, 00000000000, 0018024EBA0, 008000003E0)
045573E7140  0018014E0B4 (7FFD54886255, 00000000016, 008000003E0, 004B394F3A6)
045573E7140  00180198EEB (7FFD54886255, 00000000016, 008000003E0, 004B394F3A6)
045573E7140  004B393212C (00000000020, 22A487BBAF8, 00000000000, 22A6C7F4120)
045573E7140  004B3936867 (00000000000, 00000000001, 00000000000, 045573E7170)
045573E7140  7FFD1B0F4461 (0000000000A, 045573E73A0, 22A6C7F3D60, D6D941F662B1)
045573E7170  7FFD1B0F418D (00000000000, 045573E7390, 00000000000, 00000000000)
045573E73E0  7FFD1B0F4042 (22A487D3B10, 045573E7278, 7FFD1BD4B4D8, 045573E7380)
045573E73E0  7FFD1B1032B5 (004B39319A0, 045573E7390, 045573E73A0, 22A00000000)
045573E73E0  7FFD1B102EC8 (22A6DC70940, 22A6DB2F030, 00000000000, 04500001101)
00000000000  7FFD1B102A8C (7FFD1B102940, 00000000000, 00000000000, 00000000000)
045573E7610  7FFD1BD33457 (00000000001, 22A487D3B10, 22A6DBB9BA8, 00000000001)
045573E7970  7FFD1BD2F616 (00000000000, 045573E7C40, 00000000000, 00000000000)
22A6DC85D10  7FFD1BD49BAD (22A6DC7FC00, 90DC8CB43D62805D, 00000000000, 00000000000)
22A6DC85D10  7FFD1BDDCDB6 (7FFD1C1017F8, 22A6DC85D10, 22A6DBCF0E0, 00000000000)
7FFD1C1017F8  7FFD1BDDCC3F (22A6DC70918, 7FFD1C0F32A0, 00000000000, 22A6DC70918)
22A6DC7FC00  7FFD1BDDCB3F (00000000002, 22A49634BD0, 00000000000, 22A6DC70900)
00000000002  7FFD1BDA0FDD (22A487D3B10, 00000000002, 172B7BEF138823D3, 00000000001)
22A487D3B10  7FFD1BD593C8 (22A487D3B10, 22A6DC7FE00, 00000000000, 045573E7FE0)
22A6DC7FE00  7FFD1BD5919F (22A496350D0, 22A6DC70900, 22A487D3B10, 22A6DBAD7E0)
22A496350D0  7FFD1BD592E3 (045573E7F50, 00000000001, 22A48828EA0, 22A6DC7FE00)
045573E7F50  7FFD1BD345CE (00000000002, 22A496015B0, 22A6C709708, 00000000003)
045573E8300  7FFD1BD2D787 (22A6C709700, 8000000000000003, 22A6C709700, 00000000002)
045573E8300  7FFD1BD30018 (00000000002, 22A487D3B10, 22A6C70AE30, 00000000002)
045573E8660  7FFD1BD2F616 (00000000001, 22A49602180, 22A6C70A470, 00000000001)
045573E8A10  7FFD1BD2D787 (22A6C70A468, 8000000000000001, 22A6DC83EB0, 22A4962C7F0)
045573E8A10  7FFD1BD2ECE4 (00000000002, 22A49603380, 22A6DBE81D0, 00000000002)
End of stack trace (more stack frames may be present)
  23855 [main] python (12524) C:\Python310\python.exe: *** fatal error - Internal error: TP_NUM_C_BUFS too small: 50

Answer 1

事实上，通过 Python 安装 libmagic 解决了我的问题。

pip install python-magic python-magic-bin

可能需要安装以下软件包才能使用url_loader.load().

pip install tabulate pdf2image pytesseract