fix_382

[ethanbsmith]

adds a lightweight session for use on yahoo quote requests
session is needed for both the cookies and the "crumb" field
the handle from the session must also be used for subsequent requests
fixes #382

[pstadler]

I thought I'd be a nice neighbor and let you know that certain regions (probably EU) seem to get redirected to a consent page during the initial call, thus failing to obtain a session. We're currently investigating possible solutions:
pstadler/ticker.sh#33 (comment)

[ethanbsmith]

thx @pstadler. it looks like the cookies and crumbs are no longer needed again for v7 (at least for now). r u seeing the redirects on calls to the v7 api w/out simulating a session?

[pstadler]

it‘s happening over at my project using something very similar to your fix. Going to remove it again if it‘s working again without crumb. Wondering for how long though, it‘s been flaky for a while.

[pstadler]

I gladly reverted the change on my end. Fingers crossed that it won't happen again.

[ethanbsmith]

im currently testing an update that caches the yahoo handle. i dont know what the error will ben when the handle becomes invalid, as this is probably driven by a timeout on the yahoo server side, so i cant force the error condition. will push once i have that part worked out

[pstadler]

my assumption is that you‘ll get a 401 Unauthorized if the session expires (if this is what you’re referring to). In this case I‘d simply try to obtain a new session and fetch the quotes again. See /~https://github.com/pstadler/ticker.sh/blob/54b73a3f57432b5efd102754c48f20ed246f8ed2/ticker.sh#L53

[joshuaulrich]

This seems backward. I expect to call .yahooSession(TRUE) when force.new is TRUE, else throw an error because we couldn't get a crumb.

[ethanbsmith]

its really a state machine. the force.new is never intended to be used directly, only in the recursive call. i can rewrite it as a loop to clean up the signature

[joshuaulrich]

Calling the function recursively is fine. The logic on this line (51) confused me, but seems correct once I walk through it.

Assume force.new is FALSE. Line 45 checks that we have an active session. Line 51 checks if we're able to get a crumb. If we can't get a crumb, then !force.new is TRUE and we call .yahooSession(TRUE). So that seems like it does what it should.

I think some comments would help me remember what this is doing.

# we were unable to get a crumb
if (force.new) {
    # we couldn't get a crumb with a new session
    stop("unable to get yahoo crumb")
} else {
    # we tried to re-use a session but couldn't get a crumb
    # try to get a crumb using a new session
    ses <- .yahooSession(TRUE)
}