byterat 0.4.1

Python library to export data from and interact with the Byterat Platform

Byterat Library Documentation

Welcome to the Byterat library documentation!

This Python library provides tools for interacting with the Byterat API, allowing you to access and analyze your battery data.

Key Features

• Data Retrieval: Easily fetch battery observation metrics, dataset cycle data, and metadata.
• Filtering: Refine your data retrieval using flexible filter options.
• Asynchronous Operations: Perform efficient data retrieval using asynchronous functions.
• Data Handling: Work with data in a structured format using pandas DataFrames.

Installation

pip install byterat

Usage

1. Initialization:

from byterat.sync.client import Client as ByteratClientSync
from byterat.async_.client import Client as ByteratClientAsync

# For synchronous operations
sync_client = ByteratClientSync(token="YOUR_API_TOKEN")

# For asynchronous operations
async_client = ByteratClientAsync(token="YOUR_API_TOKEN")

2. Data Retrieval:

• Get Battery Observation Metrics:

# Get all battery observation metrics
observation_data = sync_client.get_observation_metrics()

# Get observation metrics by dataset key
observation_data_by_key = sync_client.get_observation_metrics_by_dataset_key(dataset_key="your_battery_dataset_key")

# Get observation metrics by dataset key and cycle
observation_data_by_key_cycle = sync_client.get_observation_metrics_by_dataset_key_and_dataset_cycle(
    dataset_key="your_battery_dataset_key", dataset_cycle=1
)

# Get observation metrics by filename
observation_data_by_filename = sync_client.get_observation_metrics_by_filename(file_name="your_battery_data.csv")

• Get Battery Dataset Cycle Data:

# Get all battery dataset cycle data
cycle_data = sync_client.get_dataset_cycle_data()

# Get dataset cycle data by dataset key
cycle_data_by_key = sync_client.get_dataset_cycle_data_by_dataset_key(dataset_key="your_battery_dataset_key")

# Get dataset cycle data by dataset key and cycle
cycle_data_by_key_cycle = sync_client.get_dataset_cycle_data_by_dataset_key_and_dataset_cycle(
    dataset_key="your_battery_dataset_key", dataset_cycle=1
)

# Get dataset cycle data by filename
cycle_data_by_filename = sync_client.get_dataset_cycle_data_by_filename(file_name="your_battery_data.csv")

• Get Battery Metadata:

# Get all battery metadata
metadata = sync_client.get_metadata()

# Get metadata by dataset key
metadata_by_key = sync_client.get_metadata_by_dataset_key(dataset_key="your_battery_dataset_key")

3. Filtering:

from byterat.filter import Filter, FilterOperator, FilterGroup, FilterGroupType

# Create filters
filter1 = Filter(column="voltage", operator=FilterOperator.GT, value=3.5)
filter2 = Filter(column="temperature", operator=FilterOperator.LT, value=30)

# Create a filter group
filter_group = FilterGroup([filter1, filter2], mode=FilterGroupType.AND)

# Retrieve a chunk of filtered observation data -> Returns a ByteratData object
filtered_observation_data = sync_client.get_filtered_observation_data(filters=filter_group)

# Retrieve all filtered observation data (all chunks) -> Returns a DataFrame containing all data matching filters
all_filtered_observation_data = sync_client.get_all_filtered_observation_data(filters=filter_group)

# Retrieve a chunk of filtered dataset cycle data -> Returns a ByteratData object
filtered_cycle_data = sync_client.get_filtered_dataset_cycle_data(filters=filter_group)

# Retrieve all filtered dataset cycle data (all chunks) -> Returns a DataFrame containing all data matching filters
all_filtered_cycle_data = sync_client.get_all_filtered_dataset_cycle_data(filters=filter_group)

# Retrieve a chunk of filtered metadata -> Returns a ByteratData object
filtered_metadata = sync_client.get_filtered_metadata(filters=filter_group)

# Retrieve all filtered metadata (all chunks) -> Returns a DataFrame containing all data matching filters
all_filtered_metadata = sync_client.get_all_filtered_metadata(filters=filter_group)

4. Asynchronous Operations:

import asyncio

async def main():
    # Use async_client for asynchronous operations
    observation_data = await async_client.get_observation_metrics()
    #... (other asynchronous methods)

asyncio.run(main())

5. Data Handling:

The retrieved data is returned as a ByteratData object, which contains a pandas DataFrame (data) and a continuation token (continuation_token) for handling paginated results.

# Access the DataFrame
df = observation_data.data

# Use pandas functionalities for data analysis and manipulation
print(df.head())
#...

6. Handling Paginated Results

The Byterat API uses a continuation token for pagination. When you request data, and the result set is large, the API will only return a portion of the data along with a continuation token. This token acts as a pointer to the next chunk of data.

After making a request to the Byterat API (e.g., get_observation_metrics), the response will include:

• data: A pandas DataFrame containing the current chunk of data.
• continuation_token: A string that can be used to fetch the next chunk of data. If this is None, then there is no more data to retrieve.

To get the next chunk of data, simply pass this continuation_token back to the same API function as an argument.

from byterat.sync.client import Client as ByteratClientSync

# Initialize the client
client = ByteratClientSync(token="YOUR_API_TOKEN")

# Fetch the first batch of data
response = client.get_observation_metrics()

# Process the first batch
if not response.data.empty:
    print(response.data.head())

    # Keep fetching and processing data until no more is available
    while response.continuation_token is not None:  # Correct termination condition
        # Fetch the next batch using the token
        response = client.get_observation_metrics(continuation_token=response.continuation_token)

        # Process the next batch
        if not response.data.empty:
            print(response.data.head())

Important Considerations

• Efficiency: Retrieving data in chunks can be more efficient than trying to download a massive dataset all at once.
• Large Datasets: For very large datasets, consider using asynchronous operations to avoid blocking your application while waiting for data.
• State: The continuation token represents a specific point in the data. If the underlying data changes significantly, the token may become invalid.

Class and Function Explanations

• Client (sync and async): This class provides the main interface for interacting with the Byterat API. It handles authentication, data retrieval, and filtering.
• get_observation_metrics: Retrieves battery observation metrics data. You can filter by dataset key, dataset cycle, or filename.
• get_dataset_cycle_data: Retrieves battery dataset cycle data. You can filter by dataset key, dataset cycle, or filename.
• get_metadata: Retrieves battery metadata. You can filter by dataset key.
• get_filtered_observation_data: Retrieves a single chunk of observation data filtered by the provided filter group.
• get_all_filtered_observation_data: Retrieves all observation data that matches the filter by retrieving data in chunks until no continuation token is returned.
• get_filtered_dataset_cycle_data: Retrieves a single chunk of dataset cycle data filtered by the provided filter group.
• get_all_filtered_dataset_cycle_data: Retrieves all dataset cycle data that matches the filter by retrieving data in chunks until no continuation token is returned.
• get_filtered_metadata: Retrieves a single chunk of metadata filtered by the provided filter group.
• get_all_filtered_metadata: Retrieves all metadata that matches the filter by retrieving data in chunks until no continuation token is returned.
• Filter: Represents a single filter condition.
• FilterOperator: Defines the available filter operators (e.g., equals, greater than, contains).
• FilterGroup: Combines multiple filters using logical operators (AND, OR). FilterGroups can be nested within each other to create complex filter conditions.
• ByteratData: A container class that holds the retrieved data as a pandas DataFrame and a continuation token for pagination.

Building Complex Filters with FilterGroup

The FilterGroup class allows you to create complex filter expressions by combining multiple Filter objects or even other FilterGroup objects. This recursive structure enables you to build sophisticated filtering logic.

Example:

from byterat.filter import Filter, FilterOperator, FilterGroup, FilterGroupType

# Create individual filters
filter1 = Filter(column="cycle_count", operator=FilterOperator.GT, value=100)
filter2 = Filter(column="state_of_charge", operator=FilterOperator.EQ, value=1.0)
filter3 = Filter(column="manufacturer", operator=FilterOperator.EQ, value="Tesla")

# Create a nested FilterGroup
battery_filter = FilterGroup([filter2, filter3], mode=FilterGroupType.AND)

# Combine with the cycle count filter
combined_filter = FilterGroup([filter1, battery_filter], mode=FilterGroupType.AND)

# Use the combined filter to retrieve data
filtered_data = sync_client.get_all_filtered_observation_data(filters=combined_filter)

In this example, battery_filter combines the state of charge and manufacturer filters with an AND condition. Then, combined_filter further combines the battery_filter with the cycle count filter, again using an AND condition. This results in a filter that selects data where the cycle count is greater than 100 AND the state of charge is 1.0 AND the manufacturer is Tesla.

By nesting and chaining FilterGroup objects, you can express arbitrarily complex filter conditions to precisely target the data you need.

Additional Notes

• Error Handling: The library may raise exceptions for invalid input or API errors. Make sure to handle exceptions appropriately in your code.
• Asynchronous vs. Synchronous: Choose the appropriate client (ByteratClientSync or ByteratClientAsync) based on your application's needs.
• Filtering: The library offers a variety of filter operators for precise data selection.
• Pagination: Use the continuation token to retrieve subsequent pages of data when dealing with large datasets.