Mechanoid Web Browser - Recording Capability
Seymour
(http://online.wsj.com/public/us) and automatically download various
financial pages on stocks and mutual funds that I am interested in
tracking. I have a subscription to this site and am trying to figure
out how to use python, which I have been trying to learn for the past
year, to automatically login and capture a few different pages.
I have mastered capturing web pages on non-password sites, but am
struggling otherwise and have been trying to learn how to program the
Mechanoid module (http://cheeseshop.python.org/pypi/mechanoid) to get
past the password protected site hurdle.
My questions are:
1. Is there an easier way to grab these pages from a password protected
site, or is the use of Mechanoid a reasonable approach?
2. Is there an easy way of recording a web surfing session in Firefox
to see what the browser sends to the site? I am thinking that this
might help me better understand the Mechanoid commands, and more easily
program it. I do a fair amount of VBA Programming in Microsoft Excel
and have always found the Macro Recording feature a very useful
starting point which has greatly helped me get up to speed.
Thanks for your help/insights.
Seymour
John J. Lee
> I am trying to find a way to sign onto my Wall Street Journal account
> (http://online.wsj.com/public/us) and automatically download various
> financial pages on stocks and mutual funds that I am interested in
> tracking. I have a subscription to this site and am trying to figure
[...]
> My questions are:
> 1. Is there an easier way to grab these pages from a password protected
> site, or is the use of Mechanoid a reasonable approach?
This is the first time I heard of anybody using mechanoid. As the
author of mechanize, of which mechnoid is a fork, I was always in the
dark about why the author decided to fork it (he hasn't emailed
me...).
I don't know if there's any activity on the mechanoid project, but I'm
certainly still working on mechanize, and there's an active mailing list:
http://wwwsearch.sourceforge.net/
https://lists.sourceforge.net/lists/listinfo/wwwsearch-general
> 2. Is there an easy way of recording a web surfing session in Firefox
> to see what the browser sends to the site? I am thinking that this
> might help me better understand the Mechanoid commands, and more easily
> program it. I do a fair amount of VBA Programming in Microsoft Excel
> and have always found the Macro Recording feature a very useful
> starting point which has greatly helped me get up to speed.
With Firefox, you can use the Livehttpheaders extension:
http://livehttpheaders.mozdev.org/
The mechanize docs explain how to turn on display of HTTP headers that
it sends.
Going further, certainly there's at least one HTTP-based recorder for
twill, which actually watches your browser traffic and generates twill
code for you (twill is a simple language for functional testing and
scraping built on top of mechanize):
http://darcs.idyll.org/%7Et/projects/scotch/doc/
That's not an entirely reliable process, but some people might find it
helpful.
I think there may be one for zope.testbrowser too (or ZopeTestBrowser
(sp?), the standalone version that works without Zope) -- I'm not
sure. (zope.testbrowser is also built on mechanize.) Despite the
name, I'm told this can be used for scraping as well as testing.
I would imagine that it would be fairly easy to modify or extend
Selenium IDE to emit mechanize or twill or zope.testbrowser (etc.)
code (perhaps without any coding, I used too many Firefox Selenium
plugins and now forget which had which features). Personally I would
avoid using Selenium itself to actually automate tasks, though, since
unlike mechanize &c., Selenium drags in an entire browser, which
brings with it some inflexibility (though not as bad as in the past).
It does have advantages though: most obviously, it knows JavaScript.
John
John J. Lee
[...]
> struggling otherwise and have been trying to learn how to program the
> Mechanoid module (http://cheeseshop.python.org/pypi/mechanoid) to get
> past the password protected site hurdle.
>
> My questions are:
> 1. Is there an easier way to grab these pages from a password protected
> site, or is the use of Mechanoid a reasonable approach?
Again, can't speak for mechanoid, but it should be straightforward
with mechanize (simplifiying one of the examples from the URL below):
http://wwwsearch.sourceforge.net/mechanize/
br = Browser()
br.add_password("http://example.com/protected/", "joe", "password")
br.set_debug_http(True) # Print HTTP headers.
br.open("http://www.example.com/protected/blah.html")
print br.response().read()
John
John J. Lee
> I am trying to find a way to sign onto my Wall Street Journal account
> (http://online.wsj.com/public/us) and automatically download various
> financial pages on stocks and mutual funds that I am interested in
> tracking. I have a subscription to this site and am trying to figure
> out how to use python, which I have been trying to learn for the past
> year, to automatically login and capture a few different pages.
[...]
Just to add: It's quite possible that site has an "no scraping"
condition in their terms of use. It seems standard legal boilerplate
on commercial sites these days. Not a good thing on the whole, I tend
to think, but you should be aware of it.
John
Seymour
Lots of great leads in your post that I am busy looking at. I did try
one program, MaxQ, that records web surfing. It seems to work great.
I have looked at all of your leads and plan to give them all a try.
BTW, I am not sure how I came accross Mechanoid before Mechanize, but I
did and started to study that. Somehow I had the notion that
Mechanize was a Pearl script.
Thanks again,
Seymour
John J. Lee
> Somehow I had the notion that Mechanize was a Pearl script.
mechanize the Python module started as a port of Andy Lester's Perl
module WWW::Mechanize (in turn based on Gisle Aas' libwww-perl), and
on some very high level has "the same" conceptual interface, but most
of the details (internal structure, features and bugs ;-) are
different to LWP and WWW::Mechanize due to the integration with
urllib2, httplib and friends, and with my own code. Most parts of the
code are no longer recognisable as having originated in LWP (and of
course, lots of it *didn't* originate there).
John
John J. Lee
[...]
> one program, MaxQ, that records web surfing. It seems to work great.
There are lots of such programs about (ISTR twill used to use MaxQ for
its recording feature, but I think Titus got rid of it in favour of
his own code, for some reason). How useful they are depends on the
details of what you're doing: the information that goes across HTTP is
on a fairly low level, so e.g., most obviously, you may need to be
sending a session ID that varies per-request. Those programs usually
have some way of dealing with that specific problem, but you may run
into other problems that have the same origin. Don't let me put you
off it gets your job done, but it's good to be a bit wary: All current
web-scraping approaches using free software suck in one way or
another.
John
alex...@hotmail.com
It works very well with the password protected sites. SWEA is .Net API,
but you can use IronPython to access it.