Explore a store in a notebook

chronozarr.view("my_store")

Install chronozarr[notebook]. For a local store, view starts a server on 127.0.0.1 that answers byte-range requests. It then shows the hosted viewer in an iframe. The viewer loads from the internet. The store is read from the machine that runs the kernel.

This works in local Jupyter and VS Code sessions. The browser must reach 127.0.0.1 on the machine that runs the kernel. Chrome asks for permission before a public page reaches a local address. Allow it, or use the "open in a tab" link under the frame. Safari can block an https page from loading http://127.0.0.1. Use Chrome or Firefox, or a self-hosted viewer.

If the frame stays empty, run chronozarr.diagnose_view("my_store"). It prints how many requests the browser has sent to the server and what to try next.

Remote notebooks

The server listens on 127.0.0.1 of the machine that runs the kernel, and it never listens on another address. On JupyterHub, over SSH or in a container, your browser runs elsewhere. Pick the route that matches your setup.

Setup	Route	Status
Local Jupyter or VS Code	Default, nothing to set	Works
JupyterHub with `jupyter-server-proxy`	`viewer_dir="viewer"` (see below)	Works (checked with a stand-in proxy only)
Another proxy or tunnel that you control	`base_url="https://.../"`	Works
SSH port forward	`port=8765`, then `ssh -L 8765:127.0.0.1:8765 host`	Works with the same port number on both ends
VS Code Remote, dev containers	Port forwarding of the same port number	Documented only

The proxy behind base_url must map that address to http://127.0.0.1:<port>/, remove its own prefix and pass Range headers. The stand-in proxy is described in Measurement log.

JupyterHub and jupyter-server-proxy

A hub login cookie belongs to the hub origin. The hosted viewer on chronozarr.org cannot send it, so the hub rejects its requests to a proxied store. A viewer served from the hub origin can. Serve a self-hosted viewer next to the store:

Write a viewer folder once, where Node.js is available. See Self-host the viewer.
```
npx --package chronozarr chronozarr-viewer viewer
```
Pass the folder to view or player.
```
chronozarr.view("my_store", viewer_dir="viewer")
```
When the kernel runs under JupyterHub, view sends the browser to /user/<name>/proxy/<port>/ on the hub. The notebook login protects the viewer and the store. Install jupyter-server-proxy where the Jupyter server runs, and restart the server.
On a Jupyter server that is not under JupyterHub, pass the route yourself. For a server at the root of its host, the route is /proxy/<port>. Otherwise put the server base URL in front.
```
chronozarr.view("my_store", viewer_dir="viewer", port=8765, base_url="/proxy/8765")
```

base_url is an absolute http(s) URL or a path on the notebook origin that starts with /. It is the address at which the browser reaches the root of the store server. The store is at <base_url>/<store name>.

Do not expose the server with a public base_url unless you want anyone with the link to read the store.

chronozarr.player shows a viewer that Python can control. See examples/anywidget.

view and player take the initial view: view("my_store", product="ndvi"), player(url, product="band", band="B8", range=[0, 4000]). See Band roles and the first view for what these mean.