Official client for Siftrics' Hydra API, which is a text recognition documents-to-database service
Project description
This repository contains the official Hydra API Python client. The Hydra API is a text recognition service.
Quickstart
- Install the package.
pip install hydra-api
or
poetry add hydra-api
etc.
- Create a new data source on siftrics.com.
- Grab an API key from the page of your newly created data source.
- Create a client, passing your API key into the constructor.
- Use the client to processes documents, passing in the id of a data source and the filepaths of the documents.
import hydra_api
client = hydra_api.Client('xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx')
rows = client.recognize('my_data_source_id', ['invoice.pdf', 'receipt_1.png'])
rows
looks like this:
[
{
"Error": "",
"FileIndex": 0,
"RecognizedText": { ... }
},
...
]
FileIndex
is the index of this file in the original request's "files" array.
RecognizedText
is a dictionary mapping labels to values. Labels are the titles of the bounding boxes drawn during the creation of the data source. Values are the recognized text inside those bounding boxes.
Using Base64 Strings Instead of File Paths
There is another function, client.recognizeBase64(dataSourceId, base64Files, doFaster=False)
, which accepts base64 strings (file contents) instead of file paths. Because it is not trivial to infer MIME type from the contents of a file, you must specify the MIME type associated to each base64 file string: base64Files
must be a list of dict
objects containing two fields: "mimeType"
and ``"base64File"`. Example:
base64Files = [
{
'mimeType': 'image/png',
'base64File': '...',
},
{
'mimeType': 'application/pdf',
'base64File': '...',
},
]
rows = client.recognizeBase64('Helm-Test-Againe', base64Files, doFaster=True)
Faster Results
client.recognize(dataSourceId, files, doFaster=False)
has a parameter, doFaster
, which defaults to False
, but if it's set to True
then Siftrics processes the documents faster at the risk of lower text recognition accuracy. Experimentally, doFaster=true seems not to affect accuracy when all the documents to be processed have been rotated no more than 45 degrees.
Exporting JPGs instead of PNGs
client.recognize(dataSourceId, files, returnJpgs=True, jpgQuality=85)
has a parameter, returnJpgs
, which defaults to False
, but if it's set to True
then Siftrics returns cropped images in JPG format instead of PNG format. jpgQuality
defaults to 85 and must be an integer between 1 and 100 inclusive.
Official API Documentation
Here is the official documentation for the Hydra API.
Apache V2 License
This code is licensed under Apache V2.0. The full text of the license can be found in the "LICENSE" file.
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for hydra_api-1.1.0-py2.py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 63b41143ec27852d32cb87e61b7d2391ebc373f84c59a00de9199b5274a63ddb |
|
MD5 | 9ae2e238aa25bfbb60cabb01272a0c59 |
|
BLAKE2b-256 | 9c2df762ff7b10163d2b512389cb72b644d8ecae31cbe434937c87cd3555acc2 |