memect-ppx 0.0.1__tar.gz

memect_ppx-0.0.1/LICENSE

@@ -0,0 +1,165 @@
+                   GNU LESSER GENERAL PUBLIC LICENSE
+                       Version 3, 29 June 2007
+
+ Copyright (C) 2007 Free Software Foundation, Inc. <http://fsf.org/>
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+
+  This version of the GNU Lesser General Public License incorporates
+the terms and conditions of version 3 of the GNU General Public
+License, supplemented by the additional permissions listed below.
+
+  0. Additional Definitions.
+
+  As used herein, "this License" refers to version 3 of the GNU Lesser
+General Public License, and the "GNU GPL" refers to version 3 of the GNU
+General Public License.
+
+  "The Library" refers to a covered work governed by this License,
+other than an Application or a Combined Work as defined below.
+
+  An "Application" is any work that makes use of an interface provided
+by the Library, but which is not otherwise based on the Library.
+Defining a subclass of a class defined by the Library is deemed a mode
+of using an interface provided by the Library.
+
+  A "Combined Work" is a work produced by combining or linking an
+Application with the Library.  The particular version of the Library
+with which the Combined Work was made is also called the "Linked
+Version".
+
+  The "Minimal Corresponding Source" for a Combined Work means the
+Corresponding Source for the Combined Work, excluding any source code
+for portions of the Combined Work that, considered in isolation, are
+based on the Application, and not on the Linked Version.
+
+  The "Corresponding Application Code" for a Combined Work means the
+object code and/or source code for the Application, including any data
+and utility programs needed for reproducing the Combined Work from the
+Application, but excluding the System Libraries of the Combined Work.
+
+  1. Exception to Section 3 of the GNU GPL.
+
+  You may convey a covered work under sections 3 and 4 of this License
+without being bound by section 3 of the GNU GPL.
+
+  2. Conveying Modified Versions.
+
+  If you modify a copy of the Library, and, in your modifications, a
+facility refers to a function or data to be supplied by an Application
+that uses the facility (other than as an argument passed when the
+facility is invoked), then you may convey a copy of the modified
+version:
+
+   a) under this License, provided that you make a good faith effort to
+   ensure that, in the event an Application does not supply the
+   function or data, the facility still operates, and performs
+   whatever part of its purpose remains meaningful, or
+
+   b) under the GNU GPL, with none of the additional permissions of
+   this License applicable to that copy.
+
+  3. Object Code Incorporating Material from Library Header Files.
+
+  The object code form of an Application may incorporate material from
+a header file that is part of the Library.  You may convey such object
+code under terms of your choice, provided that, if the incorporated
+material is not limited to numerical parameters, data structure
+layouts and accessors, or small macros, inline functions and templates
+(ten or fewer lines in length), you do both of the following:
+
+   a) Give prominent notice with each copy of the object code that the
+   Library is used in it and that the Library and its use are
+   covered by this License.
+
+   b) Accompany the object code with a copy of the GNU GPL and this license
+   document.
+
+  4. Combined Works.
+
+  You may convey a Combined Work under terms of your choice that,
+taken together, effectively do not restrict modification of the
+portions of the Library contained in the Combined Work and reverse
+engineering for debugging such modifications, if you also do each of
+the following:
+
+   a) Give prominent notice with each copy of the Combined Work that
+   the Library is used in it and that the Library and its use are
+   covered by this License.
+
+   b) Accompany the Combined Work with a copy of the GNU GPL and this license
+   document.
+
+   c) For a Combined Work that displays copyright notices during
+   execution, include the copyright notice for the Library among
+   these notices, as well as a reference directing the user to the
+   copies of the GNU GPL and this license document.
+
+   d) Do one of the following:
+
+       0) Convey the Minimal Corresponding Source under the terms of this
+       License, and the Corresponding Application Code in a form
+       suitable for, and under terms that permit, the user to
+       recombine or relink the Application with a modified version of
+       the Linked Version to produce a modified Combined Work, in the
+       manner specified by section 6 of the GNU GPL for conveying
+       Corresponding Source.
+
+       1) Use a suitable shared library mechanism for linking with the
+       Library.  A suitable mechanism is one that (a) uses at run time
+       a copy of the Library already present on the user's computer
+       system, and (b) will operate properly with a modified version
+       of the Library that is interface-compatible with the Linked
+       Version.
+
+   e) Provide Installation Information, but only if you would otherwise
+   be required to provide such information under section 6 of the
+   GNU GPL, and only to the extent that such information is
+   necessary to install and execute a modified version of the
+   Combined Work produced by recombining or relinking the
+   Application with a modified version of the Linked Version. (If
+   you use option 4d0, the Installation Information must accompany
+   the Minimal Corresponding Source and Corresponding Application
+   Code. If you use option 4d1, you must provide the Installation
+   Information in the manner specified by section 6 of the GNU GPL
+   for conveying Corresponding Source.)
+
+  5. Combined Libraries.
+
+  You may place library facilities that are a work based on the
+Library side by side in a single library together with other library
+facilities that are not Applications and are not covered by this
+License, and convey such a combined library under terms of your
+choice, if you do both of the following:
+
+   a) Accompany the combined library with a copy of the same work based
+   on the Library, uncombined with any other library facilities,
+   conveyed under the terms of this License.
+
+   b) Give prominent notice with the combined library that part of it
+   is a work based on the Library, and explaining where to find the
+   accompanying uncombined form of the same work.
+
+  6. Revised Versions of the GNU Lesser General Public License.
+
+  The Free Software Foundation may publish revised and/or new versions
+of the GNU Lesser General Public License from time to time. Such new
+versions will be similar in spirit to the present version, but may
+differ in detail to address new problems or concerns.
+
+  Each version is given a distinguishing version number. If the
+Library as you received it specifies that a certain numbered version
+of the GNU Lesser General Public License "or any later version"
+applies to it, you have the option of following the terms and
+conditions either of that published version or of any later version
+published by the Free Software Foundation. If the Library as you
+received it does not specify a version number of the GNU Lesser
+General Public License, you may choose any version of the GNU Lesser
+General Public License ever published by the Free Software Foundation.
+
+  If the Library as you received it specifies that a proxy can decide
+whether future versions of the GNU Lesser General Public License shall
+apply, that proxy's public statement of acceptance of any version is
+permanent authorization for you to choose that version for the
+Library.

memect_ppx-0.0.1/PKG-INFO

@@ -0,0 +1,301 @@
+Metadata-Version: 2.4
+Name: memect-ppx
+Version: 0.0.1
+Summary: memect-ppx 是一个 PDF / 图片文档解析工具，将输入文件转换为结构化的 Markdown、JSON。支持本地模型（默认）和多种 LLM 后端（DeepSeek、PaddleOCR、GLM等），适用于高精度文档理解场景。
+Requires-Python: >=3.12
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: anyio>=4.12.1
+Requires-Dist: beautifulsoup4>=4.14.3
+Requires-Dist: concurrent-log-handler>=0.9.29
+Requires-Dist: fastapi>=0.135.1
+Requires-Dist: filetype>=1.2.0
+Requires-Dist: fonttools>=4.62.1
+Requires-Dist: freetype-py>=2.5.1
+Requires-Dist: httpx[brotli,http2,socks]>=0.28.1
+Requires-Dist: openai>=2.29.0
+Requires-Dist: orjson>=3.11.7
+Requires-Dist: pdf-oxide>=0.3.17
+Requires-Dist: pillow>=12.1.1
+Requires-Dist: pptx-ea-font>=0.0.4
+Requires-Dist: pydantic>=2.12.5
+Requires-Dist: pymupdf>=1.27.2
+Requires-Dist: pypdf>=6.9.1
+Requires-Dist: python-multipart>=0.0.22
+Requires-Dist: python-pptx>=1.0.2
+Requires-Dist: rapid-layout>=1.2.1
+Requires-Dist: rapidocr>=3.8.1
+Requires-Dist: rich>=14.3.3
+Requires-Dist: shapely>=2.1.2
+Requires-Dist: table-cls>=1.2.0
+Requires-Dist: typer>=0.24.1
+Requires-Dist: uvicorn[standard]>=0.42.0
+Requires-Dist: openvino==2025.4.1; sys_platform == "darwin" and platform_machine == "x86_64"
+Requires-Dist: openvino; sys_platform != "darwin" or platform_machine != "x86_64"
+Provides-Extra: cuda
+Requires-Dist: nvidia-cuda-runtime-cu12; extra == "cuda"
+Requires-Dist: nvidia-cudnn-cu12; extra == "cuda"
+Requires-Dist: nvidia-cublas-cu12; extra == "cuda"
+Requires-Dist: nvidia-cufft-cu12; extra == "cuda"
+Requires-Dist: nvidia-curand-cu12; extra == "cuda"
+Requires-Dist: nvidia-cuda-nvrtc-cu12; extra == "cuda"
+Requires-Dist: nvidia-nvjpeg-cu12; extra == "cuda"
+Dynamic: license-file
+
+# 说明
+
+## 系统要求
+
+| 平台 | 要求 |
+|------|------|
+| Linux | Ubuntu **>= 20.04**（推荐 22.04 / 24.04），Python >= 3.12 |
+| macOS (Apple Silicon) | Python >= 3.12 |
+| macOS (Intel) | Python 3.12 ~ 3.13（受 OpenVINO 限制） |
+
+> GPU 加速需要 NVIDIA 驱动 + CUDA 12.x，并安装 `onnxruntime-gpu` 及 `memect-ppx[cuda]`。
+
+## 安装
+
+```bash
+#>=3.12
+$uv venv -p 3.12
+$source .venv/bin/activate
+
+#如果是在已经存在的环境中，建议先删除
+#$uv pip uninstall opencv-python opencv-contrib-python opencv-contrib-headless opencv-contrib-python-headless
+#$uv pip uninstall onnxruntime onnxruntime-gpu
+
+#or opencv-contrib-python-headless
+$uv pip install opencv-contrib-python --no-config
+
+#cpu版本
+$uv pip install memect-ppx
+$uv pip install onnxruntime --no-config
+
+#gpu版本
+#安装依赖的cuda库，如果系统中已经全局安装，可以不安装，需要和onnxruntime-gpu的一致
+#如果是其他版本，请根据onnxruntime-gpu的要求安装几个
+$uv pip install memect-ppx[cuda]
+#这个必须安装
+$uv pip install onnxruntime-gpu --no-config
+
+
+#安装方法二
+$git clone 
+$cd ppx
+$uv venv -p 3.12
+$uv sync --no-install-project
+#如果需要使用gpu，如果系统中已经全局安装，可以不安装，或者安装另外的版本
+$uv sync --extra cuda
+
+#这两个必须手动安装
+#or opencv-contrib-python-headless
+$uv pip install opencv-contrib-python --no-config
+#or onnxruntime-gpu 
+$uv pip install onnxruntime --no-config
+
+
+
+#默认解析
+$./app parse a.pdf
+
+#使用大模型解析，如果模型的base_url,model,api_key 和默认设置不一样，可以在命令行中指定，或者在./conf/settings.py中
+$./app parse a.pdf --backend deepseek --deepseek '{"base_url":"http://127.0.0.1:4000/v1","model":"deepseek-ocr-2","api_key":""}'
+$./app parse a.pdf --backend paddle   --paddle  '{"base_url":"http://127.0.0.1:4001/v1","model":"paddleocr-vl","api_key":""}'
+$./app parse a.pdf --backend glm      --glm  '{"base_url":"http://127.0.0.1:4002/v1","model":"glmocr","api_key":""}'
+
+
+#如果经常使用，可以写到配置文件中
+$mkdir conf
+#可以为json文件或者py文件: settings={}
+#参考src/memect/conf/settings.custom.py 语法
+$vi conf/settings.py
+$vi conf/log.py
+
+#如果在配置文件中写好了路径和模型等，就不需要在命令行再指定
+$ppx parse a.pdf --backend deepseek
+```
+
+```bash
+$git clone 
+#mac intel，且需要使用onnxruntime等，只能够使用3.12<=python<=3.13
+#linux或者mac m，使用>=3.12
+$uv venv -p 3.12
+#uv的其它操作，都是对默认的环境：.venv，即使激活了一个环境，也需要"--active"参数
+#而pip操作，为了和pip兼容，会安装到激活环境，或者通过"-p",UV_PYTHON=xxx等来设置
+
+#目前还是使用requirements.txt安装依赖更加方便，虽然uv支持了很多的冲突处理，但是
+#还不够，还无法快速的解决多个同名包的冲突
+
+$source .venv/bin/activate
+$uv pip install requirements.txt
+#因为opencv，onnxruntime，torch可能被第三方库指定了不同的版本，如：headless/contrib，cpu/gpu
+#所以这里统一先清除，再安装
+$uv pip uninstall -r requirements.uninstall.txt
+#安装cpu或者cuda一个即可
+$uv pip install -r requirements.cpu.txt
+$uv pip install -r requirements.cuda.txt
+$uv pip install -r requirements.opencv.txt
+
+$./app start
+
+#解析pdf，pdf+ocr（自动）
+./app parse a.pdf
+
+#解析pdf，总是使用ocr
+./app parse a.pdf --ocr yes
+
+#解析pdf，不使用ocr
+./app parse a.pdf --ocr no
+
+#解析图片
+./app parse 1.png
+
+#指定使用哪个llm：deepseek,paddle,glm，对于文本，没有每个字符的坐标
+./app parse 1.pdf --backend deepseek
+
+#常用配置可以通过命令行参数或者环境变量，或者本地文件，如：
+#本地文件，目录为：./conf/settings.py, ./conf/log.py
+#常用参数  
+$./app parse 1.pdf --set xx.xx.xx=1 
+$./app parse 1.pdf --backend deepseek --url http://127.0.0.1:/1111/v1
+$./app parse 1.pdf --backend glm --url http://127.0.0.1:1111/v1
+$./app parse 1.pdf --backend paddle --url http://127.0.0.1:1111/v1
+
+#解析目录下的所有pdf或者图片，如果没有
+$./app parse dir1 
+
+#输出到这个目录
+$./app parse dir1 -o dir2 
+
+#表示dir1目录下的所有图片作为一个整体解析，也就是连续的多个页面
+$./app parse dir1 --images 
+
+
+```
+
+## Docker Compose
+
+```bash
+#
+$cd x2x
+#日常操作就3个：build，up，push，其他用户：pull，up
+#注意：这里是自动获得日期，如：20260324，也可以手动指定一个
+#重新构建所有的镜像
+$TAG=$(date +%Y%m%d) docker compose build
+#apiserver,deepseek,paddle,glm
+$TAG=$(date +%Y%m%d) docker compose build apiserver
+
+#注意切换TAG为指定日前
+#推送到hub
+$TAG=20260324 docker compose push apiserver deepseek
+
+#其他用户需要先pull
+$TAG=20260324 docker compose pull apiserver deepseek
+
+#启动apiserver
+#注意：需要指定具体的版本（日期），否则会自动build
+$TAG=20260324 docker compose up apiserver
+#启动apiserver+deepseek
+$TAG=20260324 docker compose up apiserver deepseek
+
+#可以通过".env"修改设置，如果使用其它的文件名，需要--env-file ./.env.my
+$cp .env.sample .env
+```
+
+## Docker Build
+
+如果想单独操作单个docker的，通过"-t"可以指定名字，也可以使用和docker compose中的一致，
+这里就使用简单的名字了，因为主要是本地测试，通过tag也可以方便修改
+
+```bash
+#docker>=23 如果是<23，使用docker buildx build 
+$docker build --target apiserver -t x2x-apiserver .
+#后面这些是可选择的，使用其它方式部署模型也可以，如果是使用docker compose，使用这些会更好
+$docker build --target deepseek -t x2x-deepseek  .
+$docker build --target paddle -t x2x-paddle  .
+$docker build --target glm -t x2x-glm .
+
+$docker build --target llm -t x2x-llm .
+
+```
+
+## Docker Run
+
+```bash
+#使用docker run 启动单个服务，可以把"--it"替换为"-d"
+#--gpus "device=0"  => 容器仅仅可以看到0这个gpu，也就是nvidia-smi只看到0
+#-e NVIDIA_VISIBLE_DEVICES=0 => 在容器限制，仅仅在docker下可用
+#-e CUDA_VISIBLE_DEVICES=0   => 在应用层限制，实际使用上两张并没有什么区别，个人建议还是使用这个
+
+
+#启动apiserver，有内部模型，可以指定使用gpu或者cpu
+#默认的配置都是使用127.0.0.1
+#方案1，使用host网络，不需要修改任何配置
+#--network host
+#方案2，创建网络，对于这种，建议使用docker compose完成了
+
+$docker run --gpus all -it --rm -p 9527:9527 x2x-apiserver
+
+#使用cpu
+$docker run -it --rm -p 9527:9527 x2x-apiserver
+
+#启动大模型，可以通过环境变量控制gpu的映射
+#常用的命令参数，如：控制显存，同时启动多个实体
+#--gpu-memory-utilization 0.5 -dp 2
+$docker run --gpus all -it --rm -p 4000:4000 x2x-deepseek
+$docker run --gpus all -it --rm -p 4001:4001 x2x-paddle
+$docker run --gpus all -it --rm -p 4002:4002 x2x-glm
+#可以把模型下载在外部，然后启动
+$docker run --gpus all -it --rm -p 4003:4003 -v ./hub:/apps/llm/hub x2x-llm vllm serve ...
+
+```
+
+## 启动模型
+
+```bash
+
+#常用环境变量，可以附加在命令前面
+$export CUDA_VISIBLE_DEVICES=0
+
+#需要大概20G
+$vllm serve ./hub/deepseek-ai/DeepSeek-OCR-2 --served-model-name deepseek-ocr-2 \
+--logits-processors vllm.model_executor.models.deepseek_ocr:NGramPerReqLogitsProcessor \
+--mm-processor-cache-gb 0 \
+--no-enable-prefix-caching \
+--port 4000 \
+--gpu-memory-utilization 0.8 
+
+
+#需要10G
+$vllm serve ./hub/PaddlePaddle/PaddleOCR-VL \
+  --served-model-name paddleocr-vl \
+  --trust-remote-code \
+  --max-num-batched-tokens 16384 \
+  --no-enable-prefix-caching \
+  --mm-processor-cache-gb 0 \
+  --gpu-memory-utilization 0.5\
+  --port 4001
+
+#需要10G
+#也可以启动1.5版本，模型名字，端口号等都一样，配置等就都不需要改变
+$vllm serve ./hub/PaddlePaddle/PaddleOCR-VL-1.5 \
+  --served-model-name paddleocr-vl \
+  --trust-remote-code \
+  --max-num-batched-tokens 16384 \
+  --no-enable-prefix-caching \
+  --mm-processor-cache-gb 0 \
+  --gpu-memory-utilization 0.5\
+  --port 4001
+
+#https://modelscope.cn/models/ZhipuAI/GLM-OCR
+#需要手动更新为这个，以后vllm的依赖大于这个就不需要手动了
+$uv pip install "transformers>=5.3.0"
+$vllm serve ./hub/ZhipuAI/GLM-OCR \
+--served-model-name glmocr \
+--max-num-batched-tokens 16384 \
+--max-model-len 16384 \
+--speculative-config '{"method": "mtp", "num_speculative_tokens": 1}' \
+--gpu-memory-utilization 0.5 \
+--port 4002
+```