A simple client wrapper for the doccano API.
doccano users have consistently been interested in a way to programmatically access a doccano instance.
-
In Issue #6, Hironsan provided a sample class to illustrate interaction with the doccano API.
-
In Issue #410, louisguitton asked for assistance in using the API endpoints.
-
In Issue #299, afparsons "figured out" how to upload files.
This project seeks to provide a temporary solution while the doccano/doccano team creates an official package to call APIs.
- investigate more secure alternatives to plaintext login
- improve docstrings
- a doccano instance to which you may connect
- Python 3
- Python requests library
cd <DESTINATION DIRECTORY>
git clone https://github.com/doccano/doccano_api_client.git
cd doccano_api_client
# to install normally
pip install ./
# to install with edit mode
pip install -e ./- Object instantiation takes care of session authorization.
- All methods return a
requests.models.Responseobject.
from doccano_api_client import DoccanoClient
# instantiate a client and log in to a Doccano instance
doccano_client = DoccanoClient(
'http://doccano.example.com',
'username',
'password'
)
# get basic information about the authorized user
r_me = doccano_client.get_me()
# print the details from the above query
print(r_me())
# get the label text from project 1, label 3
label_text = doccano_client.get_label_detail(1, 3)()['text']
# upload a json file to project 1. If file is in current directory, file_path is omittable
r_json_upload = doccano_client.post_doc_upload(1, 'json', 'file.json', '/path/to/file/without/filename/')This wrapper's methods are based on doccano url paths.
Key:
- ✔️ implemented
- ❌ not implemented
⚠️ currently broken or improperly implemented
Endpoint Names:
- ✔️
auth-token - ✔️
me - ✔️
user_list - ✔️
roles - ✔️
features - ✔️
project_list - ✔️
project_detail - ✔️
statistics - ✔️
label_list - ✔️
label_detail - ❌
label_upload - ✔️
doc_list - ✔️
doc_detail - ✔️
doc_uploader - ❌
cloud_uploader - ✔️
approve_labels - ✔️
annotation_list ⚠️ annotation_detail- ✔️
doc_downloader - ✔️
rolemapping_list ⚠️ rolemapping_detail
Why doesn't this package's source code use f-strings (PEP 498)?
F-strings are available for Python 3.6+. I wanted to be able to support slightly older versions of Python.
What kind of code formatting/style does this package use?
I've not yet prescribed a style guide. However:
- I generally like Python Black.
- I've provided Google-style docstrings throughout the source code.
Does this API client support <some_feature> in doccano (deleting, etc.)?
Probably not (yet). A few things to note:
- everything I know about the endpoints is listed in the README.md file.
- this was only supposed to be a temporary solution! I guess the doccano team hasn't yet made their own client?
- I based the API client code off of what I could find in doccano's source code, but I wasn't able to find much documentation to begin with.
- I originally created this API client to quickly solve a problem I had at work; the supported endpoints came in order of my own personal requirement!
Can I contribute?
Please do! See Contributing
- Fork the main project.
- Create an issue on doccano/doccano_api_client.
- Make a feature or bugfix branch on your fork referencing the newly-created issue.
- Commit to that branch.
- Submit a pull request when ready.
Once you've cloned the repository and created a virtual environement, just run the following command
make developThanks to Hironsan and louisguitton.