ENH: Improved Yahoo finance api functionality

[nehalecky]

Added functionality of listing stock index components via query to yahoo finance
API. Expanded existing functionality of other yahoo finance features in
pandas.io.data.py, in detail:

• Created new function get_component_yahoo(), allowing for query of stock index
  components. Returns DataFrame containing list of component information for
  index represented by idx_sym from yahoo. Result includes component symbol
  (ticker), exchange, and full name.
• Refactored get_data_yahoo, now allows for multiple stock symbols with input
  of list-like object (tuple, list, Series), or DataFrame with stock symbols as index.
  Includes convenience options for adjusting price and adding return index.
• Refactored get_quote_yahoo. Can now accept a single symbol (str) along with
  list-like object (list, tuple, Series).
• Expanded test coverage to the new features.

EDIT: A few typos.

[ghost]

Hi,

You've changed the signature of get_data_yahoo so old code might break.
Is there no way to make your improvments backwards-compatible?

[nehalecky]

Hi @y-p,

Thanks for pointing this out! I'd be glad to fix this. Just to be sure, is the concern with the signature mostly positional or keys (or both)? Being that the function now allows for multiple stock symbols, I thought it would be beneficial to have a name that reflected such, but perhaps this isn't good practice.

The function signature was previously:

 get_data_yahoo(name=None, start=None, end=None, retry_count=3, pause=0)

This PR contains changes as:

get_data_yahoo(symbols=None, start=None, end=None, adjust_price=False,
                      ret_index=False, chunk=25, pause=0, **kwargs):

Would a change to the following be sufficient?

get_data_yahoo(name=None, start=None, end=None,  retry_count=3, pause=0, 
                       adjust_price=False, ret_index=False, chunk=25, pause=0, 
                       **kwargs):

When you can, let me know!

Thanks!

[ghost]

yep, keeping the old argument order and adding new ones at the end (all with defaults)
should do it, as long as the function body still behaves the same.

Some network related tests are marked as @slow so make sure you run the full
suite with test.sh (or manually use nose to run them)

a note on this very issue was just added a couple of days ago:
https://github.com/pydata/pandas/blob/master/CONTRIBUTING.md

Thanks for contributing.

[ghost]

Well, for back-compat, obviously both. admitadly it's rare to use named arguments
for the first arg in a call. but if you must change an arg name, detect the original's use
in **kwds and issue a deprecation warning and handle it so the call still succeeds.

there's a bunch of that in the tree, look around.

[nehalecky]

Ah, grand. That all makes sense and that's just the type of feedback I needed. I had spent some time looking around, but hadn't caught that treatment of maintaining compatibility with older args via passing in **kwds with warning of deprecation. That's great. Also, kudos to that well written contributing doc, it helps a lot.

I'll get these changes implemented, full tested, and pushed soon! Thanks again, @y-p!

[nehalecky]

I think with that commit, all should be good for backwards compatibility. Let me know of any other thoughts or suggestions you might have, thank you!

[nehalecky]

Sorry @y-p, are you referring to the previous default value in get_component_yahoo, '^DJI'?

[ghost]

yep.

[nehalecky]

While I found it convenient, looking around the tree it didn't seem like it was common practice. There is not a default equity to point towards, perhaps, similarly, there shouldn't be a default index? Also, I didn't want to seem particularly biased towards US markets. But, if you think it's a unique case where it's better with it, I'm certainly open to the idea....

[wesm]

Cherry-picked these commits. Thanks