How to share session with IE

[zdp]

Hello!

I need to process some webpages of a forum which is powered by discuz!.
When I login, there are some options about how long to keep the
cookies: forever, month, week, et al. If I choose forever, I don't
need to login each time, and When I open the internet explorer I can
access any pages directly. Some urls of the pages like:

http://www.somesite.com/bbs/viewthread.php?tid=12345&extra=page=1

However, now I need to process some pages by a python program. When I
use urllib.urlopen(theurl), I can only get a page which told me I need
login. I think It's reasonable, becuase I wasn't in a loggined session
which as IE did.

So how can I do my job? I want to get the right webpage by the url. I
have search answers from the groups but didn't get clear answer. Should
I use win32com or urllib? Any reply or information is appreciate. Hope
I put it clear.

Dapu

 

[Bernard]

Hello Dapu,

You can do the same thing as IE on your forum using urllib2 and
cookielib. In short you need to code a small webcrawler. I can give you
my browser module if necessary.
You might not have the time to fiddle with the coding part or my
browser module so you can also use this particularly useful module :
http://wwwsearch.sourceforge.net/mechanize/
The documentation is pretty clear for an initiated python programmer.
If it's not your case, I'd recommend to read some ebooks on the python
language first to get use to it.

Bernard

 

[zdp]

It's exactly what I want. I'll try. Thanks!

Hello Dapu,

You can do the same thing as IE on your forum using urllib2 and
cookielib. In short you need to code a small webcrawler. I can give you
my browser module if necessary.
You might not have the time to fiddle with the coding part or my
browser module so you can also use this particularly useful module :
http://wwwsearch.sourceforge.net/mechanize/
The documentation is pretty clear for an initiated python programmer.
If it's not your case, I'd recommend to read some ebooks on the python
language first to get use to it.

Bernard

 

[John J. Lee]

zdp wrote: [...]

However, now I need to process some pages by a python program. When I
use urllib.urlopen(theurl), I can only get a page which told me I need
login. I think It's reasonable, becuase I wasn't in a loggined session
which as IE did.

So how can I do my job? I want to get the right webpage by the url. I
have search answers from the groups but didn't get clear answer. Should
I use win32com or urllib? Any reply or information is appreciate. Hope
I put it clear.

You can do the same thing as IE on your forum using urllib2 and
cookielib. In short you need to code a small webcrawler. I can give you
my browser module if necessary.
You might not have the time to fiddle with the coding part or my
browser module so you can also use this particularly useful module :
http://wwwsearch.sourceforge.net/mechanize/
The documentation is pretty clear for an initiated python programmer.
If it's not your case, I'd recommend to read some ebooks on the python
language first to get use to it.

In particular, if you're following the approach Bernard suggests, you
can either:

1. Log in every time your program runs, by going through the sequence
of clicks, pages, etc. that you would use in a browser to log in.

2. Once only (or once a month, or whatever), log in by hand using IE
with a "Remember me"-style feature (if the website offers that) --
where the webapp asks the browser to save the cookie rather than
just keeping it in memory until you close your browser. Then your
program can load the cookies from your real browser's cookie store
using this:

http://wwwsearch.sourceforge.net/mechanize/doc.html#browsers


There are other alternatives too, but they depend on knowing a little
bit more about how cookies and web apps work, and may or may not work
depending on what exactly the server does. I'm thinking specifically
here of saving *session* cookies (the kind that usually go away when
you close your browser) in a file -- but the server may not like them
when you send them back the next time, depending how much time has
elapsed since the last run. Of course, you can always detect the
"need to login" condition, and react accordingly.


John

 

[Cameron Walsh]

zdp wrote: [...]

However, now I need to process some pages by a python program. When I
use urllib.urlopen(theurl), I can only get a page which told me I need
login. I think It's reasonable, becuase I wasn't in a loggined session
which as IE did.

So how can I do my job? I want to get the right webpage by the url. I
have search answers from the groups but didn't get clear answer. Should
I use win32com or urllib? Any reply or information is appreciate. Hope
I put it clear.

You can do the same thing as IE on your forum using urllib2 and
cookielib. In short you need to code a small webcrawler. I can give you
my browser module if necessary.
You might not have the time to fiddle with the coding part or my
browser module so you can also use this particularly useful module :
http://wwwsearch.sourceforge.net/mechanize/
The documentation is pretty clear for an initiated python programmer.
If it's not your case, I'd recommend to read some ebooks on the python
language first to get use to it.

In particular, if you're following the approach Bernard suggests, you
can either:

1. Log in every time your program runs, by going through the sequence
of clicks, pages, etc. that you would use in a browser to log in.

2. Once only (or once a month, or whatever), log in by hand using IE
with a "Remember me"-style feature (if the website offers that) --
where the webapp asks the browser to save the cookie rather than
just keeping it in memory until you close your browser. Then your
program can load the cookies from your real browser's cookie store
using this:

http://wwwsearch.sourceforge.net/mechanize/doc.html#browsers


There are other alternatives too, but they depend on knowing a little
bit more about how cookies and web apps work, and may or may not work
depending on what exactly the server does. I'm thinking specifically
here of saving *session* cookies (the kind that usually go away when
you close your browser) in a file -- but the server may not like them
when you send them back the next time, depending how much time has
elapsed since the last run. Of course, you can always detect the
"need to login" condition, and react accordingly.


John

Another option instead of making your program run through a series of
clicks and text inputs, which is difficult to program, is to browse the
html source until you find the name of the script that processes the
login, and use python to request the page with the necessary form fields
encoded in the request. Request something like
http://www.targetsite.com/login.cgi?username=pyuser&password="fhqwhgads"
This format is not guaranteed to work, since the login script or server
might only support one of GET and POST. If this is the case, creating
the request is slightly more involved and to be honest I haven't looked
into how to do it.

Thereafter, you will have to pass the environment to every page request
so the server can read the cookie. Which brings me to question whether
or not it is possible to do this manually once, export the environment
variable to a file, and reload this file each time the program is run.
Or to generate the cookie in the environment yourself. Quite frankly
any server application that allows the client to control whether or not
they have logged in sucks, but I've seen a fair few that do.[citation
required]

Cameron.

 

[Cameron Walsh]

I just thought, your original question was whether or not it was
possible to share your browser session with IE. Unless you do this
explicitly, you may require a different login for your Python program
and for your IE user. If the Python program does not get the same
cookie as used by IE, or vice-versa, and tries to login, you may find
this resets the login in the other browser.

Oh and at risk of starting a flame war <a href="www.getfirefox.com"
reason="obligatory pro open source link">get a real browser.</a>

 

[zdp]

I found some similar topics in the newsgroup and get some ideas from
them.
http://groups.google.com/group/comp.lang.python/browse_thread/thread/2fe0be6c386adce4
http://groups.google.com/group/comp.lang.python/browse_thread/thread/a51cec8747f64619

According to all you suggestions, there are at least two ways to get my
result.

1. Use the cookie of IE, so I don't need to code to logon. That means I
must use ClientCookie. I found some example in the docs and the
newsgroup. Below is some code based on the docs of ClientCookie. But
the page I get is still the page told me must login ( I CAN get the
right page in IE).

import ClientCookie, urllib2

url_string="http://www.targetsite.com/bbs/viewthread.php?tid=12345"
#the page I want to get

cj = ClientCookie.MSIECookieJar(delayload=True)
cj.load_from_registry()
print cj #I want to know what I get

opener =
ClientCookie.build_opener(ClientCookie.HTTPCookieProcessor(cj))
ClientCookie.install_opener(opener)
f = ClientCookie.urlopen(url_string)
print f.read() # NOT the right page html


2. Logon myself by python. First, I access the login page and submit
the form of username and password. The form has many fields other than
username and passwd, so the dict "data" has all the fields even if it's
hide. Then, if the login succeed, I can get my page use the opener with
CookieJar.

import urllib2, cookielib

url_string="http://www.targetsite.com/bbs/viewthread.php?tid=12345"
#the page I want to get
url_login="http://www.targetsite.com/bbs/logging.php?action=login"
#the login page

headers = {'User-agent' : 'Mozilla/4.0 (compatible; MSIE 5.5;
Windows NT)'}
cj = cookielib.CookieJar()
opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(cj))

urllib2.install_opener(opener)
data = {
'formhash': '3bd8bc0a',
"referer" : "index.php",
"loginfield": "username",
'username': 'myname',
'password': 'mypass',
"questionid": 0,
"answer":"",
"cookietime" : "315360000",
"loginmode":"",
"styleid":""
}
req=urllib2.Request(url_login, urllib.urlencode(data), headers)
f = opener.open(req)
print req.get_data()
print req.header_items()
print f.info()
print f.read()

## if login succeed, I can get my page
f=opener.open( url_string)


However, both ways didn't work for me. I don't know what's wrong. If
it's because the server page check the header or the submit of the form
is wrong?

I didn't study Mechanize module yet. I want a solution as simple as
possible for distribution reason.

John J. Lee 写é“：

zdp wrote: [...]

However, now I need to process some pages by a python program. When I
use urllib.urlopen(theurl), I can only get a page which told me I need
login. I think It's reasonable, becuase I wasn't in a loggined session
which as IE did.

So how can I do my job? I want to get the right webpage by the url. I
have search answers from the groups but didn't get clear answer. Should
I use win32com or urllib? Any reply or information is appreciate. Hope
I put it clear.

You can do the same thing as IE on your forum using urllib2 and
cookielib. In short you need to code a small webcrawler. I can give you
my browser module if necessary.
You might not have the time to fiddle with the coding part or my
browser module so you can also use this particularly useful module :
http://wwwsearch.sourceforge.net/mechanize/
The documentation is pretty clear for an initiated python programmer.
If it's not your case, I'd recommend to read some ebooks on the python
language first to get use to it.

In particular, if you're following the approach Bernard suggests, you
can either:

1. Log in every time your program runs, by going through the sequence
of clicks, pages, etc. that you would use in a browser to log in.

2. Once only (or once a month, or whatever), log in by hand using IE
with a "Remember me"-style feature (if the website offers that) --
where the webapp asks the browser to save the cookie rather than
just keeping it in memory until you close your browser. Then your
program can load the cookies from your real browser's cookie store
using this:

http://wwwsearch.sourceforge.net/mechanize/doc.html#browsers


There are other alternatives too, but they depend on knowing a little
bit more about how cookies and web apps work, and may or may not work
depending on what exactly the server does. I'm thinking specifically
here of saving *session* cookies (the kind that usually go away when
you close your browser) in a file -- but the server may not like them
when you send them back the next time, depending how much time has
elapsed since the last run. Of course, you can always detect the
"need to login" condition, and react accordingly.


John

 

[John J. Lee]

Another option instead of making your program run through a series of
clicks and text inputs, which is difficult to program, is to browse
the html source until you find the name of the script that processes
the login, and use python to request the page with the necessary form
fields encoded in the request. Request something like
http://www.targetsite.com/login.cgi?username=pyuser&password="fhqwhgads"
This format is not guaranteed to work, since the login script or
server might only support one of GET and POST. If this is the case,
creating the request is slightly more involved and to be honest I
haven't looked into how to do it.

Absolutely, that's often a great way to do things, since it's very
simple, and is not in conflict with handling cookies (where that's
required).

(But of course if you need to handle cookies, you still need to
arrange to actually handle the cookies somewhere.)

Thereafter, you will have to pass the environment to every page
request so the server can read the cookie. Which brings me to
question whether or not it is possible to do this manually once,
export the environment variable to a file, and reload this file each
time the program is run. Or to generate the cookie in the environment
yourself.

[...]

Standard library module cookielib (or mechanize, which is not part of
the stdlib, and does some more stuff automatically and provides some
extra features for page navigation and form handling) does all this
automatically:

http://docs.python.org/lib/cookielib-examples.html

import cookielib, urllib2
cj = cookielib.CookieJar()
opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(cj))
r = opener.open("http://example.com/")


For loading and saving (including Firefox support):

http://docs.python.org/lib/file-cookie-jar-classes.html

http://docs.python.org/lib/cookie-jar-objects.html


For loading IE cookies, use mechanize.

http://wwwsearch.sourceforge.net/mechanize/


John

 

[John J. Lee]

1. Use the cookie of IE, so I don't need to code to logon. That means I
must use ClientCookie. I found some example in the docs and the
newsgroup. Below is some code based on the docs of ClientCookie. But
the page I get is still the page told me must login ( I CAN get the
right page in IE).

Try mechanize (same website as ClientCookie -- though right now, that
part of sourceforge seems to be down for me). It supports more
browser features automatically.

[...]

However, both ways didn't work for me. I don't know what's wrong. If
it's because the server page check the header or the submit of the form
is wrong?

Changing the HTTP headers you send may solve your problem, yes.
Either that, or the response body ;-)

I didn't study Mechanize module yet. I want a solution as simple as
possible for distribution reason.

OK, then you should compare the HTTP requests that a real browser
sends with the HTTP requests that your Python script sends. The
following pages give some help with that (from memory, since the site
is down right now):

http://wwwsearch.sf.net/mechanize/doc.html#debugging

http://wwwsearch.sf.net/bits/GeneralFAQ.html


John