Search API feature request: follows:username

[Amir Michail]

Hi,

I would like to be able to issue queries like this:

web 2.0 follows:techcrunch

This would identify tweets containing web 2.0 posted by people
following techcrunch.

Amir

[Matt Sanford]

Hi Amir,

This is something we've talked about internally for some time.
Please open a ticket at http://code.google.com/p/twitter-api/issues so
we can track your request.

Thanks;
— Matt Sanford (@mzsanford)

[Amir Michail]

On Dec 2, 1:33 pm, Matt Sanford <m...@twitter.com> wrote:
> Hi Amir,
>
>      This is something we've talked about internally for some time.  
> Please open a ticket athttp://code.google.com/p/twitter-api/issuesso  
> we can track your request.

Done:

http://code.google.com/p/twitter-api/issues/detail?id=172

Amir

[Amir Michail]

On Dec 2, 1:38 pm, Amir Michail <amich...@gmail.com> wrote:
> On Dec 2, 1:33 pm, Matt Sanford <m...@twitter.com> wrote:
>
> > Hi Amir,
>
> >      This is something we've talked about internally for some time.  
> > Please open a ticket athttp://code.google.com/p/twitter-api/issuesso 
> > we can track your request.
>
> Done:
>
> http://code.google.com/p/twitter-api/issues/detail?id=172
>
> Amir

So it has been categorized as "WontFix" because it is not an API
issue.

Could you please work on it? It would be really helpful for a service
I am building: http://b4utweet.com.

[Matthew]

You could implement this in the following way.

1. get all the follows of "techcrunch"
2. search for the keyword you want, saying "from: username1 OR
username2 OR ... usernameN"

an example

http://search.twitter.com/search?q=github+from%3Alebreeze+OR+matthewrudy

shows all the tweets about github from myself and my friend Levent.

Although with 32000 usernames I imagine it wont be fast,
and you'd have to iterate over 320 requests to /statuses/followers.xml
in order to grab all their usernames.

[Chad Etzel]

Not to mention that the query for the search API is limited to 140 characters (or it was the last time I checked). So you'd need to split up your query and then thread the results by timestamp afterward (if time is of concern for your app...).

Not really an elegant solution/implementation, but I could imagine that the database query needed to the equivalent on the back-end servers would not exactly be trivial either.

-Chad

[Amir Michail]

On Dec 4, 12:44 pm, Matthew <matthewrudyjac...@gmail.com> wrote:
> You could implement this in the following way.
>
> 1. get all the follows of "techcrunch"
> 2. search for the keyword you want, saying "from: username1 OR
> username2 OR ... usernameN"

Unfortunately that sort of query can be very slow even with only a few
from:'s.

Amir

[Matthew]

On Dec 4, 8:32 pm, "Chad Etzel" <jazzyc...@gmail.com> wrote:
> Not really an elegant solution/implementation, but I could imagine that the
> database query needed to the equivalent on the back-end servers would not
> exactly be trivial either.

Yeah.
I imagine the search functionality is actually done using lucene, and
some sort of facets?

Or at least it'd make sense given the nature of the search,
and the fact that the query language looks like lucene syntax.

Not that I've worked with such big fulltext searching,
but our search of 30,000 jobs is incredibly slow once you start adding
criteria (especially range queries).

With twitters one billion messages I'm sure the game changes quite a
lot.