Feature/visualisation

[schneegor]

No description provided.

[erget]

Might be easiest to just consume the CSV output by csv_observation_request. If you read the response into a StringIO buffer you could pass it to pandas using read_csv and you'd have the data as a DataFrame without having to manipulate the observations directly. There'd be no network bottleneck because there'd be no HTTP communication - we'd just use the responses.

The content would be accessible like this:

csv = csv_observation_request(request, station, start, end).content
# Continue processing

[erget]

Before the merge we need to add the following packages to install_requires in setup.py:

• matplotlib
• pySide

[erget]

It would be nice to build this so that png and csv are at the beginning, not at the end. Not sure how to best implement that - off the top of my head I'm thinking something like putting those up front and then passing them to a date_parser or something, somehow passing it the harvested request type so that it knows what function to call. I prefer that because that would be more consistent with line 31, where the pattern begins with the request type. Before the merge we'll need to find a DRY way of expressing that, or modify the API for requestion a station list.

I'm thinking something like this:

data_patterns = [
    url(r'^(?P<content_type>[a-z]+)/'  # Content type
        r'^(?P<station>[0-9]+)/'  # Station ID
        r'(?P<start>[0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2})'    # Start date
        r' - '
        r'(?P<end>[0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2})/', data_request),
        url(r'csv$', csv_observation_request)

And then in views.py something like:

DATA_TYPES_MAP = {"csv": csv_observation_request,
                  "png": png_observation_request}
def data_request(request, content_type, station, start, end):
    """Route a data request."""
    return DATA_TYPES_MAP[content_type](request, station, start, end)

Just a very rough prototype.