Message-ID: <A7F9F46A-121B-4E75-B13D-447AFDE42255@comcast.net>
Date: 2016-05-19T23:51:15Z
From: David Winsemius
Subject: Can I use PhantomJS or assume a firefox instalattion for usage with RSelenium in CRAN Machines?
In-Reply-To: <CANMhtdz9bFR+5QaLW8sXxt6upduiayW-xn2xDrHphSkPMfBmPQ@mail.gmail.com>
> On May 19, 2016, at 7:49 AM, Marcelo Perlin <marceloperlin at gmail.com> wrote:
>
> Hi Guys,
>
> First time posting here.
>
> I have a CRAN package called GetTDData that downloads and reads public data
> from a government website (
> http://www.tesouro.fazenda.gov.br/tesouro-direto-balanco-e-estatisticas).
>
> Recently (today), the website has changed its structure by removing
> permanent links of the files and creating a "random" html address for the
> files that really matter. This means that when I download the souce html
> code, I don't have the information for the actual links, but just a bunch
> of code.
>
> In the past I have dealed with this type of problem by forcing the
> renderization of the page using RSelenium with firefox or PhantomJS and
> then capturing the desired href locations.
>
> My question is, if integrate my code with RSelenium using firefox or
> PhantonJS, will it pass on all arquitectures (win, linux, solaris) of CRAN?
I dn't have a lot of experience at this but I can say that at least one person whose experience I trust recentlyreported in Rhelp that RSelenium tends to be a fragile interface. Nonetheless, he does use it on occasion and it clearly "works" on more than one platform. If your code is not confidential and the website has at least a guest login capacity, you could post it here and ask for trial runs on whatever platform(s) you may not have testing capacities for.
>
> --
> Marcelo Perlin
> Professor Adjunto | Escola de Administra??o
--
David Winsemius
Alameda, CA, USA