QAnything+Ollama搭建本地知识库_qanything ollama

1、Ollama安装使用

参考我的另一篇博客：ollama + fastgpt 搭建免费本地知识库-CSDN博客

2、QAnything

概述：

QAnything (Question and Answer based on Anything) 是致力于支持任意格式文件或数据库的本地知识库问答系统，可断网安装使用。

您的任何格式的本地文件都可以往里扔，即可获得准确、快速、靠谱的问答体验。

目前已支持格式: PDF(pdf)，Word(docx)，PPT(pptx)，XLS(xlsx)，Markdown(md)，电子邮件(eml)，TXT(txt)，图片(jpg，jpeg，png)，CSV(csv)，网页链接(html)，更多格式，敬请期待...

纯Python环境安装：

conda create -n qanything-python python=3.10
conda activate qanything-python
git clone -b qanything-python https://github.com/netease-youdao/QAnything.git
cd QAnything
pip install -e .

启动运行QAnything:

# 本地已经启动了ollama，这里做个运行测试
# 因为我的ollama里没有安装llama3模型，所以修改一下run_for_ollama_api_in_M1_mac.sh的参数。改成使用qwen2:latest模型
#!/bin/bash
#bash scripts/base_run.sh -s "M1mac" -w 4 -m 19530 -q 8777 -o -b 'http://localhost:11434/v1' -k 'ollama' -n 'llama3' -l '4096'
bash scripts/base_run.sh -s "M1mac" -w 4 -m 19530 -q 8777 -o -b 'http://localhost:11434/v1' -k 'ollama' -n 'qwen2:latest' -l '4096'
 
# 启动运行
bash scripts/run_for_ollama_api_in_M1_mac.sh

报错了

报错信息如下：

(base) ➜  QAnything git:(qanything-python) ✗ bash scripts/run_for_ollama_api_in_M1_mac.sh
即将启动后端服务，启动成功后请复制[http://0.0.0.0:8777/qanything/]到浏览器进行测试。
运行qanything-server的命令是：
CUDA_VISIBLE_DEVICES=0 python3 -m qanything_kernel.qanything_server.sanic_api --host 0.0.0.0 --port 8777 --model_size 7B  --use_openai_api --openai_api_base http://localhost:11434/v1 --openai_api_key ollama --openai_api_model_name llama3 --openai_api_context_length 4096 --workers 4
LOCAL DATA PATH: /Users/renlibin/PythonProjects/QAnything/QANY_DB/content
LOCAL_RERANK_REPO: maidalun/bce-reranker-base_v1
LOCAL_EMBED_REPO: maidalun/bce-embedding-base_v1
2024-06-28 17:34:14,614 - modelscope - INFO - PyTorch version 2.1.2 Found.
2024-06-28 17:34:14,614 - modelscope - INFO - Loading ast index from /Users/renlibin/.cache/modelscope/ast_indexer
2024-06-28 17:34:14,614 - modelscope - INFO - No valid ast index found from /Users/renlibin/.cache/modelscope/ast_indexer, generating ast index from prebuilt!
2024-06-28 17:34:14,691 - modelscope - INFO - Loading done! Current index file version is 1.13.0, with md5 f0dca17f6b67762919fb8486c6f931f5 and a total number of 972 components indexed
use_cpu: False
use_openai_api: True
The server is starting on port: 8777
onnxruntime 1.18.0 已经安装。
2024-06-28 17:34:15,469 使用openai api llama3 无需下载大模型
Traceback (most recent call last):
  File "/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/qanything_server/sanic_api.py", line 164, in <module>
    from .handler import *
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/qanything_server/handler.py", line 1, in <module>
    from qanything_kernel.core.local_file import LocalFile
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/core/local_file.py", line 5, in <module>
    from qanything_kernel.utils.loader.my_recursive_url_loader import MyRecursiveUrlLoader
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/__init__.py", line 4, in <module>
    from .self_pdf_loader import PdfLoader
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/self_pdf_loader.py", line 3, in <module>
    from qanything_kernel.utils.loader.pdf_to_markdown.core.parser import PdfParser
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/pdf_to_markdown/core/parser/__init__.py", line 3, in <module>
    from .pdf_parser import HuParser as PdfParser, PlainParser
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/pdf_to_markdown/core/parser/pdf_parser.py", line 18, in <module>
    from qanything_kernel.utils.loader.pdf_to_markdown.core.nlp import huqie
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/pdf_to_markdown/core/nlp/__init__.py", line 4, in <module>
    from . import huqie
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/pdf_to_markdown/core/nlp/huqie.py", line 4, in <module>
    import datrie
ImportError: dlopen(/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/site-packages/datrie.cpython-310-darwin.so, 0x0002): tried: '/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/site-packages/datrie.cpython-310-darwin.so' (mach-o file, but is an incompatible architecture (have 'x86_64', need 'arm64e' or 'arm64')), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/site-packages/datrie.cpython-310-darwin.so' (no such file), '/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/site-packages/datrie.cpython-310-darwin.so' (mach-o file, but is an incompatible architecture (have 'x86_64', need 'arm64e' or 'arm64'))
(base) ➜  QAnything git:(qanything-python) ✗ bash scripts/run_for_ollama_api_in_M1_mac.sh
即将启动后端服务，启动成功后请复制[http://0.0.0.0:8777/qanything/]到浏览器进行测试。
运行qanything-server的命令是：
CUDA_VISIBLE_DEVICES=0 python3 -m qanything_kernel.qanything_server.sanic_api --host 0.0.0.0 --port 8777 --model_size 7B  --use_openai_api --openai_api_base http://localhost:11434/v1 --openai_api_key ollama --openai_api_model_name qwen2 --openai_api_context_length 4096 --workers 4
LOCAL DATA PATH: /Users/renlibin/PythonProjects/QAnything/QANY_DB/content
LOCAL_RERANK_REPO: maidalun/bce-reranker-base_v1
LOCAL_EMBED_REPO: maidalun/bce-embedding-base_v1
2024-06-28 17:38:17,124 - modelscope - INFO - PyTorch version 2.1.2 Found.
2024-06-28 17:38:17,124 - modelscope - INFO - Loading ast index from /Users/renlibin/.cache/modelscope/ast_indexer
2024-06-28 17:38:17,150 - modelscope - INFO - Loading done! Current index file version is 1.13.0, with md5 f0dca17f6b67762919fb8486c6f931f5 and a total number of 972 components indexed
use_cpu: False
use_openai_api: True
The server is starting on port: 8777
onnxruntime 1.18.0 已经安装。
2024-06-28 17:38:17,274 使用openai api qwen2 无需下载大模型
Traceback (most recent call last):
  File "/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/qanything_server/sanic_api.py", line 164, in <module>
    from .handler import *
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/qanything_server/handler.py", line 1, in <module>
    from qanything_kernel.core.local_file import LocalFile
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/core/local_file.py", line 5, in <module>
    from qanything_kernel.utils.loader.my_recursive_url_loader import MyRecursiveUrlLoader
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/__init__.py", line 4, in <module>
    from .self_pdf_loader import PdfLoader
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/self_pdf_loader.py", line 3, in <module>
    from qanything_kernel.utils.loader.pdf_to_markdown.core.parser import PdfParser
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/pdf_to_markdown/core/parser/__init__.py", line 3, in <module>
    from .pdf_parser import HuParser as PdfParser, PlainParser
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/pdf_to_markdown/core/parser/pdf_parser.py", line 18, in <module>
    from qanything_kernel.utils.loader.pdf_to_markdown.core.nlp import huqie
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/pdf_to_markdown/core/nlp/__init__.py", line 4, in <module>
    from . import huqie
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/pdf_to_markdown/core/nlp/huqie.py", line 4, in <module>
    import datrie
ImportError: dlopen(/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/site-packages/datrie.cpython-310-darwin.so, 0x0002): tried: '/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/site-packages/datrie.cpython-310-darwin.so' (mach-o file, but is an incompatible architecture (have 'x86_64', need 'arm64e' or 'arm64')), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/site-packages/datrie.cpython-310-darwin.so' (no such file), '/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/site-packages/datrie.cpython-310-darwin.so' (mach-o file, but is an incompatible architecture (have 'x86_64', need 'arm64e' or 'arm64'))
(base) ➜  QAnything git:(qanything-python) ✗ clear
(base) ➜  QAnything git:(qanything-python) ✗ bash scripts/run_for_ollama_api_in_M1_mac.sh
即将启动后端服务，启动成功后请复制[http://0.0.0.0:8777/qanything/]到浏览器进行测试。
运行qanything-server的命令是：
CUDA_VISIBLE_DEVICES=0 python3 -m qanything_kernel.qanything_server.sanic_api --host 0.0.0.0 --port 8777 --model_size 7B  --use_openai_api --openai_api_base http://localhost:11434/v1 --openai_api_key ollama --openai_api_model_name qwen2:latest --openai_api_context_length 4096 --workers 4
LOCAL DATA PATH: /Users/renlibin/PythonProjects/QAnything/QANY_DB/content
LOCAL_RERANK_REPO: maidalun/bce-reranker-base_v1
LOCAL_EMBED_REPO: maidalun/bce-embedding-base_v1
2024-06-28 17:38:56,804 - modelscope - INFO - PyTorch version 2.1.2 Found.
2024-06-28 17:38:56,804 - modelscope - INFO - Loading ast index from /Users/renlibin/.cache/modelscope/ast_indexer
2024-06-28 17:38:56,830 - modelscope - INFO - Loading done! Current index file version is 1.13.0, with md5 f0dca17f6b67762919fb8486c6f931f5 and a total number of 972 components indexed
use_cpu: False
use_openai_api: True
The server is starting on port: 8777
onnxruntime 1.18.0 已经安装。
2024-06-28 17:38:56,951 使用openai api qwen2:latest 无需下载大模型
Traceback (most recent call last):
  File "/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/qanything_server/sanic_api.py", line 164, in <module>
    from .handler import *
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/qanything_server/handler.py", line 1, in <module>
    from qanything_kernel.core.local_file import LocalFile
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/core/local_file.py", line 5, in <module>
    from qanything_kernel.utils.loader.my_recursive_url_loader import MyRecursiveUrlLoader
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/__init__.py", line 4, in <module>
    from .self_pdf_loader import PdfLoader
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/self_pdf_loader.py", line 3, in <module>
    from qanything_kernel.utils.loader.pdf_to_markdown.core.parser import PdfParser
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/pdf_to_markdown/core/parser/__init__.py", line 3, in <module>
    from .pdf_parser import HuParser as PdfParser, PlainParser
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/pdf_to_markdown/core/parser/pdf_parser.py", line 18, in <module>
    from qanything_kernel.utils.loader.pdf_to_markdown.core.nlp import huqie
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/pdf_to_markdown/core/nlp/__init__.py", line 4, in <module>
    from . import huqie
  File "/Users/renlibin/PythonProjects/QAnything/qanything_kernel/utils/loader/pdf_to_markdown/core/nlp/huqie.py", line 4, in <module>
    import datrie
ImportError: dlopen(/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/site-packages/datrie.cpython-310-darwin.so, 0x0002): tried: '/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/site-packages/datrie.cpython-310-darwin.so' (mach-o file, but is an incompatible architecture (have 'x86_64', need 'arm64e' or 'arm64')), '/System/Volumes/Preboot/Cryptexes/OS/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/site-packages/datrie.cpython-310-darwin.so' (no such file), '/opt/homebrew/Caskroom/miniconda/base/lib/python3.10/site-packages/datrie.cpython-310-darwin.so' (mach-o file, but is an incompatible architecture (have 'x86_64', need 'arm64e' or 'arm64'))

由报错信息可知，启动用到的库datrie下载了个'x86_64'版本的，而我的电脑是Mac M1 arm架构的，需要的是'arm64'的库。

解决方案：

# 卸载datrie库
pip uninstall datrie
 
# 重新安装arm架构的datrie库
ARCHFLAGS="-arch arm64" pip install datrie  --compile --no-cache-dir

解决问题参考：

macos - M1 mac: mach-o file, but is an incompatible architecture (have 'x86_64', need 'arm64e') - python - Ask Different

重新运行，问题解决！

点击“返回对话”。问问题测试，给出了回答和信息来源，搞定。

参考：

README_zh.md · 网易有道/QAnything - Gitee.com

QAnything/README_zh.md at qanything-python · netease-youdao/QAnything · GitHub