Open mstfldmr opened 4 months ago
ShilinHe
commented
4 months ago By default, the code will be executed in the container mode. Therefore, please make sure the package install environment is the docker. Or, you can try with the local mode first as your packages are usually installed there.
mstfldmr
commented
4 months ago I added RUN pip install --no-cache-dir --no-warn-script-location yfinance easyocr to ~/taskweaver/docker/ces_container/Dockerfile
and then, I built the image ~/taskweaver/scripts$ source build_executor.sh
~/taskweaver/scripts$ source build_executor.sh
The script directory is: /home/mstfldmr/taskweaver/scripts
Found module files from /home/mstfldmr/taskweaver/scripts/../taskweaver
Dockerfile path: /home/mstfldmr/taskweaver/scripts/../docker/ces_container/Dockerfile
Context path: /home/mstfldmr/taskweaver/scripts/../
[+] Building 24.6s (16/16) FINISHED docker:default
=> [internal] load build definition from Dockerfile 0.0s
=> => transferring dockerfile: 762B 0.0s
=> [internal] load metadata for docker.io/library/python:3.10-slim 0.6s
=> [internal] load .dockerignore 0.0s
=> => transferring context: 2B 0.0s
=> [ 1/11] FROM docker.io/library/python:3.10-slim@sha256:38ced5e623fb8a6b256826a9f1bbcf5247a05123d6b6cb3847c9cd6ba46ef9ac 0.0s
=> [internal] load build context 0.0s
=> => transferring context: 3.10kB 0.0s
=> CACHED [ 2/11] WORKDIR /app 0.0s
=> CACHED [ 3/11] COPY requirements.txt . 0.0s
=> CACHED [ 4/11] RUN pip install --no-cache-dir --no-warn-script-location -r requirements.txt 0.0s
=> [ 5/11] RUN pip install --no-cache-dir --no-warn-script-location --user yfinance 11.2s
=> [ 6/11] COPY taskweaver/ces /app/taskweaver/ces 0.1s
=> [ 7/11] COPY taskweaver/plugin /app/taskweaver/plugin 0.1s
=> [ 8/11] COPY taskweaver/module /app/taskweaver/module 0.1s
=> [ 9/11] COPY taskweaver/__init__.py /app/taskweaver/__init__.py 0.1s
=> [10/11] COPY docker/ces_container/entrypoint.sh /app/entrypoint.sh 0.1s
=> [11/11] RUN chmod +x /app/entrypoint.sh 0.4s
=> exporting to image 11.4s
=> => exporting layers 11.4s
=> => writing image sha256:ca18227dfb7f802b32aa36fca28b777e1f435332fa089c1d1efbba3867f4bb36 0.0s
=> => naming to docker.io/taskweavercontainers/taskweaver-executor:0.2 0.0s
2 warnings found (use docker --debug to expand):
- UndefinedVar: Usage of undefined variable '$PYTHONPATH' (line 21)
- LegacyKeyValueFormat: "ENV key=value" should be used instead of legacy "ENV key value" format (line 21)I still get the same error. Should I register the newly built image somewhere?
=========================================================
_____ _ _ __
|_ _|_ _ ___| | _ | | / /__ ____ __ _____ _____
| |/ _` / __| |/ /| | /| / / _ \/ __ `/ | / / _ \/ ___/
| | (_| \__ \ < | |/ |/ / __/ /_/ /| |/ / __/ /
|_|\__,_|___/_|\_\|__/|__/\___/\__,_/ |___/\___/_/
=========================================================
--- new session starts ---
TaskWeaver ▶ I am TaskWeaver, an AI assistant. To get started, could you please enter your request?
Human ▶ convert image to text
╭───< Planner >
├─► [init_plan] 1. Ask the user for the image path
├─► [plan] 1. Ask the user for the image path
├─► [current_plan_step] 1. Ask the user for the image path
├──● Could you please provide the path to the image you want to convert to text?
├─► [board]
│ I have drawn up a plan:
│ 1. Ask the user for the image path
│
│ Please proceed with this step of this plan: Could you please provide the path to the image you want to convert to text?
╰──● sending message to User
TaskWeaver ▶ Could you please provide the path to the image you want to convert to text?
Human ▶ https://www.google.com/images/branding/googlelogo/1x/googlelogo_color_272x92dp.png
╭───< Planner >
├─► [init_plan] 1. Convert the image at the provided URL to text using OCR
├─► [plan] 1. Instruct CodeInterpreter to convert the image at the provided URL to text using OCR
├─► [current_plan_step] 1. Instruct CodeInterpreter to convert the image at the provided URL to text using OCR
├──● Please convert the image to text from the following URL: [googlelogo_color_272x92dp.png](https://www.google.com/images/branding/googlelogo/1x/googlelogo_color_272x92dp.png)
├─► [board]
│ I have drawn up a plan:
│ 1. Instruct CodeInterpreter to convert the image at the provided URL to text using OCR
│
│ Please proceed with this step of this plan: Please convert the image to text from the following URL:
│ [googlelogo_color_272x92dp.png](https://www.google.com/images/branding/googlelogo/1x/googlelogo_color_272x92dp.png)
╰──● sending message to CodeInterpreter
╭───< CodeInterpreter >
├─► [thought] ProgramApe will use the image2text plugin function to convert the image from the provided URL to text using OCR. Since the plugin functions can be directly called without importing,
│ there's no need to import or install any modules.
├─► [reply_type] python
├─► [reply_content]
│ image_url = "https://www.google.com/images/branding/googlelogo/1x/googlelogo_color_272x92dp.png"
│ # Since the plugin function image2text can be called directly, we will use it to convert the image to text.
│ ocr_result = image2text(image_url)
│ ocr_result
├─► [verification] NONE
├─► [code_error] No code verification is performed.
Plugin image2text failed to load: Plugin image2text failed to register: failed to load plugin image2text Please install easyocr with `pip install easyocr`.
Plugin stock_price failed to load: Plugin stock_price failed to register: failed to load plugin stock_price No module named 'yfinance'
├─► [execution_status] FAILUREThe new image should be in your local env, which can be checked by docker images. You don't need to register it anywhere.
mstfldmr
commented
4 months ago OK, I see the newly built image
taskweavercontainers/taskweaver-executor 0.2 e1d9aff32d19 30 minutes ago 6.08GB
<none> <none> a5708e2ac27a 47 minutes ago 6.08GB
<none> <none> ca18227dfb7f 52 minutes ago 748MB
<none> <none> b1c0abd6ac88 55 minutes ago 5.93GB
taskweavercontainers/taskweaver-executor latest a9f5c9f0d958 3 months ago 716MB
hello-world latest d2c94e258dcb 15 months ago 13.3kBbut the plugin still can't find the package.
ShilinHe
commented
4 months ago can you try to use the tag "latest" instead of 0.2 for your image?
mstfldmr
commented
4 months ago I deleted all containers and images, then built it with "latest" tag. Now I have only 1 image:
~/taskweaver$ docker images
REPOSITORY TAG IMAGE ID CREATED SIZE
taskweavercontainers/taskweaver-executor latest 8de32ae5ad19 6 minutes ago 6.08GBWhen I start Taskweaver, it downloads the old image again and removes the name of the image I built:
~/taskweaver$ docker images
REPOSITORY TAG IMAGE ID CREATED SIZE
<none> <none> 8de32ae5ad19 8 minutes ago 6.08GB
taskweavercontainers/taskweaver-executor latest a9f5c9f0d958 3 months ago 716MBand Taskweaver gives the same error.
mstfldmr
commented
4 months ago The lines 148-159 in taskweaver/ces/environment.py are responsible of downloading the old image from the repo if it's not the same as the local image. Thus, I commented out these lines
try:
local_image = self.docker_client.images.get(self.image_name)
registry_image = self.docker_client.images.get_registry_data(self.image_name)
if local_image.id != registry_image.id:
logger.info(f"Local image {local_image.id} does not match registry image {registry_image.id}.")
raise docker.errors.ImageNotFound("Local image is outdated.")
except docker.errors.ImageNotFound:
logger.info("Pulling image from docker.io.")
try:
self.docker_client.images.pull(self.image_name)
except docker.errors.DockerException as e:
raise docker.errors.DockerException(f"Failed to pull image: {e}. ")This prevented Taskweaver from downloading the old image, now it's using the image I built and the plugin works well.
ShilinHe
commented
4 months ago okay. @liqul Can you take a look?
liqul
commented
4 months ago => [ 5/11] RUN pip install --no-cache-dir --no-warn-script-location --user yfinance
I see this log from your previous reply. Did you miss adding easyocr?
mstfldmr
commented
4 months ago @liqul no, I later split that line to multiple lines. I confirm that it is redownloading the old docker image instead of the newly built one.
liqul
commented
3 months ago So you are saying that, after commenting out the lines in environment.py, the code still downloads the old image instead of loading your newly built one? This sounds weird and can you check if you are running the modified code? for instance, did you somehow installed TaskWeaver as a package? I'm pretty sure that the code you commented out is the only place determining which image is going to be loaded.
mstfldmr
commented
3 months ago No, it's the other way around. It was always re-downloading the image instead of using the one I built. I commented out these lines, now it doesn't download anything.
Describe the bug The plugin can't find a package which is already installed in the virtual environment.
To Reproduce Steps to reproduce the behavior:
pip install easyocr`Plugin image2text failed to load: Plugin image2text failed to register: failed to load plugin image2text Please install easyocr withpip install easyocr.Expected behavior use the easyocr package installed in the virtual environment.
Environment Information (please complete the following information):