Skip to content
PYGWALKER
API Reference
Notebook APIs

PyGWalker Notebook APIs

In notebooks, start with pygwalker.walk(df) or create a reusable pygwalker.Walker. The current default Jupyter rendering path is anywidget, so new code should omit env or use env="JupyterAnywidget".

import pygwalker as pyg
 
walker = pyg.walk(df, spec_path="./gw_config.json", computation="kernel")

Quick decision table

NeedUseWhy
Explore a DataFrame oncepyg.walk(df)Smallest notebook call.
Reuse the same configurationpyg.Walker(df, ...).show()One object can render in notebooks, Streamlit, webserver, and HTML.
Render saved charts onlypyg.render(df, spec_path=...)Opens the renderer view instead of the full explorer.
Show the table/profiling viewpyg.table(df)Starts in the data/table mode.
Save chart state in a local filespec_path="./gw_config.json"Local chart state has a dedicated argument.

pygwalker.walk

walk opens the full Graphic Walker explorer.

import pygwalker as pyg
 
walker = pyg.walk(
    df,
    spec_path="./gw_config.json",
    computation="browser",
)

Signature:

pyg.walk(
    dataset,
    gid=None,
    *,
    env="JupyterAnywidget",
    field_specs=None,
    theme_key="g2",
    appearance="media",
    spec="",
    spec_path=None,
    computation=None,
    use_kernel_calc=None,
    kernel_computation=None,
    cloud_computation=False,
    show_cloud_tool=True,
    kanaries_api_key="",
    default_tab="vis",
    **kwargs,
)

dataset can be a pandas DataFrame, polars DataFrame, pyarrow Table, database Connector, connector-style string, or reusable pygwalker.Walker.

Anywidget is the notebook default

New notebook code should use the anywidget path:

pyg.walk(df)
pyg.walk(df, env="JupyterAnywidget")
pyg.Walker(df).show()
pyg.Walker(df).show(env="jupyter-anywidget")

Legacy Jupyter transport aliases still resolve to anywidget and emit deprecation warnings:

Legacy valueCurrent behaviorStatus
env="Jupyter"Treated as JupyterAnywidgetDeprecated, scheduled for removal in PyGWalker 0.7.0.
env="JupyterWidget"Treated as JupyterAnywidgetDeprecated, scheduled for removal in PyGWalker 0.7.0.
Walker.show("jupyter-inline")Treated as jupyter-anywidgetDeprecated, scheduled for removal in PyGWalker 0.7.0.
Walker.show("jupyter-widget")Treated as jupyter-anywidgetDeprecated, scheduled for removal in PyGWalker 0.7.0.

Walker.show(env="auto") detects the current environment. In notebooks it resolves to jupyter-anywidget; outside notebooks it starts the webserver adapter.

Reuse pygwalker.Walker

Use Walker when chart state and computation mode should be defined once.

import pygwalker as pyg
 
walker = pyg.Walker(
    df,
    spec_path="./gw_config.json",
    spec_io_mode="rw",
    computation="kernel",
)
 
walker.show()

Constructor:

pyg.Walker(
    dataset,
    gid=None,
    *,
    field_specs=None,
    theme_key="g2",
    appearance="media",
    spec="",
    spec_path=None,
    spec_io_mode="rw",
    computation=None,
    use_kernel_calc=None,
    kernel_computation=None,
    cloud_computation=False,
    show_cloud_tool=True,
    kanaries_api_key="",
    default_tab="vis",
    **kwargs,
)

Useful methods:

MethodReturnsUse
walker.corePygWalkerAccess the compatibility object used by older APIs.
walker.show(env="auto", ...)WalkerDisplay in notebook or webserver mode.
walker.to_html()strExport static iframe HTML. Browser computation only.
walker.to_html_without_iframe()strExport static HTML without the iframe wrapper. Browser computation only.
walker.to_streamlit(**kwargs)StreamlitRendererReuse the constructor options in Streamlit.

When a Walker is passed to another adapter, do not pass construction options again. This is rejected:

walker = pyg.Walker(df, spec_path="./gw_config.json", computation="browser")
 
# Raises ValueError: spec_path belongs on the original Walker.
pyg.walk(walker, spec_path="./other.json")

Computation in notebooks

Use computation for new code.

pyg.walk(df, computation="browser")
pyg.walk(df, computation="kernel")
pyg.walk(df, computation="cloud", kanaries_api_key="...")

kernel_computation, cloud_computation, and use_kernel_calc are legacy flags and are scheduled for removal in PyGWalker 0.7.0. If computation is set to "browser", "kernel", or "cloud", do not also set enabled legacy flags.

Notebook convert/static HTML output does not support live kernel or cloud computation. Use computation="browser" when the output must be static, or run PyGWalker in a live backend.

pygwalker.render

render shows saved charts without the full drag-and-drop explorer.

import pygwalker as pyg
 
pyg.render(df, spec_path="./gw_config.json", computation="browser")

Signature:

pyg.render(
    dataset,
    spec="",
    *,
    theme_key="g2",
    appearance="media",
    spec_path=None,
    computation=None,
    kernel_computation=None,
    kanaries_api_key="",
    **kwargs,
)

pygwalker.table

table opens PyGWalker's data table/profiling mode.

import pygwalker as pyg
 
pyg.table(df, spec_path="./gw_config.json")

Signature:

pyg.table(
    dataset,
    *,
    theme_key="g2",
    appearance="media",
    spec_path=None,
    computation=None,
    kernel_computation=None,
    kanaries_api_key="",
    **kwargs,
)

Related Guides

Dataset InputsComputation Modes