Bad file descriptor with XMLRPC::Lite in Perl - perl

I have a call to an XMLRPC implemented in Java which I have verified that runs without exceptions and writes the output. The call in Perl goes like this:
my $result = XMLRPC::Lite
-> proxy($url)
-> call("someMethod",
SOAP::Data->type(string => $par1),
SOAP::Data->type(string => $par2),
# etc...
)
-> result;
But then I check for $result and it is not defined, I get Bad file descriptor error.
What could be happening? It was working before, I can't think of anything significant that may have changed...

OK, I found it, although I don't quite understand why it happened.
The XMLRPC app does this:
byte[] result = xServer.execute(request.getInputStream());
getLogger().log(new String(result));
response.setContentType("text/xml");
response.setContentLength(result.length);
OutputStream out = response.getOutputStream();
out.write(result);
out.flush();
getLogger().log("finished doPost");
I'm logging the result that is sent to the output and therefore I should be getting it in the Perl script's $result variable. The result is XML generated via the Jdom library.
While I got the error, what got logged was an XML cointaining an error message indicating a problem with Jdom (basically, the app wasn't fully recompiled to that version of the library).
Now that it works, the expected XML is logged and successfully assigned to $result in Perl.
However, since the byte array is an XML in both cases I don't quite understand how it makes any difference to the caller. It wasn't even looking for a given XML sctructure, the call resulted in an error.
Any insight on this will be appreciated. However the problem is solved.

Related

Reading an SB-Messaging Send Port properties using the Microsoft.BizTalk.ExplorerOM makes a breaking change

I am working on a PowerShell script making use of the Microsoft.BizTalk.ExplorerOM to dynamically update the SB-Messaging SAS key for BizTalk Receive Locations and Send Ports. This is to enable us to roll the SAS keys for our Service Bus queues, and update BizTalk with the new keys as painlessly as possible.
I have this working correctly for Receive Locations, but Send Ports are giving me a different issue.
As soon as I read the PrimaryTransport properties of the Send Port, it seems that some change is made under the covers, that then prevents SaveChanges from working, instead throwing an "Invalid or malformed XML data exception".
This is compared to the the ReceiveLocation, where I can read any of its properties, and then SaveChanges successfully.
Note that in both of this cases, no changes have been made by me. I am simply doing a Read, and then a Save.
Can anyone offer any advice as to what could be causing the issue, and any possible solutions to try?
Had this very same issue, when using Powershell to replace values in ServiceBus ReceiveLocations & SendPorts.
The problem is with the none valid xml symbols in the TransportTypeData, which are converted when the script reads them out in the PS cmd.
All none valid xml symbols (such as the one occuring for Namespace value, ) need to be converted to amp, and if I'm not mistaken even double amp:ed.
Here's an example article showing examples on what I mean by "double amp:ed":
How do I escape ampersands in XML so they are rendered as entities in HTML?
Hope this make sense, and if not, then let me know and I'll give it another go.
Just tried doing this from C#, seems to work ok:
var root = new Microsoft.BizTalk.ExplorerOM.BtsCatalogExplorer() { ConnectionString = "Data Source=(local);Initial Catalog=BizTalkMgmtDb;Integrated Security=SSPI;" };
var sendPort = root.SendPorts["xxxx.ServiceBusQueue"];
System.Diagnostics.Trace.TraceInformation(sendPort.PrimaryTransport.TransportTypeData);
sendPort .PrimaryTransport.TransportTypeData = sendPort.PrimaryTransport.TransportTypeData.Replace("RootManageSharedAccessKey", "MySharedAccessKey");
root.SaveChanges();

Expected response code 200, got 400. Unable to convert document

In the last 2 weeks or so, I've suddenly started getting reports of users getting an error in our application saying "Expected response code 200, got 400. Unable to convert document." This is code that has been in place for years without any issue. We are using Zend Framework (GData) in conjunction with Google Docs (AuthSub).
We are logging the issue to a text file when it happens. When it gets logged, the user often tries multiple times (sometimes separated by a few seconds, other times separated by longer times) and it continues to fail. The code in question just creates a new Google document in the user's account and gives it a title (no body content).
Originally, I used this code:
// Create new document
$data = new Zend_Gdata_Docs_DocumentListEntry();
$data->setCategory(
array(new Zend_Gdata_App_Extension_Category(
"http://schemas.google.com/docs/2007#document",
"http://schemas.google.com/g/2005#kind"
)));
$data->setTitle(new Zend_Gdata_App_Extension_Title($title, null));
// Add document to your list
$test = $sharedocs->insertDocument($data, Zend_Gdata_Docs::DOCUMENTS_LIST_FEED_URI);
To experiment and see if there was an issue with that particular function, I tried creating a blank word doc and changing the code to:
$test = $sharedocs->uploadFile('/mypath/empty.doc', $title, null, Zend_Gdata_Docs::DOCUMENTS_LIST_FEED_URI);
However, I'm still seeing the "Unable to convert document" errors. They are relatively infrequent, and I am not able to reproduce the issue on my own computers here. The $title variable does not contain anything unusual (special characters, etc.).
This code was all working fine before -- is there a known issue with the Google Docs API right now? What else can I try?
NOTE: Please see my follow-up comments below, where I have identified the reproducible scenario in which this error occurs.
I had exactly the same problem, but I noticed that I could use the api to save a presentation if not a document... so, it is a terrible hack, but I try to save the document (works if the account has already been accessed)... if that fails, I save and delete a presentation and retry to save the document, which then works. Horrible, horrible, horrible hack

Strange problem getting $ENV{'QUERY_STRING'}

I'm having a strange problem here. I'm moving a (working) site to a new apache server to which I don't have direct access (I have to go through two people to get stuff done).
The site uses a perl script called adframe to parse html templates. The URLs with which it's called look like /cgi-bin/adframe/index.html?x=something with adframe being the script. The missing suffix never caused any real problems. But on this new Ubuntu server $ENV{'QUERY_STRING'} is always empty. $ENV{'REQUEST_METHOD'} shows up correctly as GET, but the query_string shows nothing ...
Regular *.cgi scripts show the query_string without problems.
From the logs I gathered that the server seems to be running fastcgi, mod_fcgid and the server doesn't even accept .pl as an extension for scripts. I don't have that much experience with server software, but I figured it might be a problem with the server not accepting adframe as a cgi script and thus not passing the query_string correctly ... Can anyone give me a few hints to where I could point the administrator or maybe something I could do in .htaccess myself? Anyway to make sure, adframe is recognized as a cgi script!? (if that's the problem ...)
Any help is appreciated!
thomas
EDIT: I found more details: The server seems to be running a VARNISH cache ... thats's the main difference to my usual configurations ...
Also, the way the script works is, if you call /cgi-bin/adframe/somedir/somefile.html?x=something, $ENV{PATH_INFO} tells which template to parse and $ENV{QUERY_STRING} is, well, the query string. Now the query string is empty, but if I call /cgi-bin/adframe?x=something (without any PATH_INFO), the query string shows up!
Does anyone have an idea what's going on here?
thanks!
Got it. The VARNISH cache strips all the query strings off static content (*.html etc) ... phew
Just ran into the same problem. I am complete newbie in perl scripting.
I tried following:
#values = split (/&/, $ENV{'QUERY_STRING'});
but it didn`t work
this worked:
#values = split (/&/, "$ENV{'QUERY_STRING'}");
just in case if other newbies have ran into the same problem.

xmlCtxtGetLastError - Iphone

I have been programming with NSXMLParser for quite a while and lately, I came out with this error. The strangiest thing is that it only happens in debug mode. Once I load the App in Simulator and run it from Simulator (without Xcode involved), it runs fine.
The code is very straight foward, it is a simple XML parsing whose contents were loaded from the web in a separated thread.
Does anybody have alredy encoutered that error??
Thanks in advance.
EDIT:
In time I realized that this error occur when you have a bad formated XML Document. In my case, I was extracting some of the contents of an HTML Page and parsing the resulting string, but this string was not allways well formed. A little modification an voilá...
Ensure that your documents are well formed when parsing or you may have the same mistake...
PS: The parser error method did not catch this error.
In time I realized that this error occur when you have a bad formated XML Document. In my case, I was extracting some of the contents of an HTML Page and parsing the resulting string, but this string was not allways well formed. A little modification an voilá...
Ensure that your documents are well formed when parsing or you may have the same mistake... PS: The parser error method did not catch this error.

How do I get the text-form verification code when doing auto site access in Perl?

I'm playing around with Win32::IE:Mechanize to try to access some authentication-required sites automatically. So far I've achieved moderate success, for example, I can automatically log in to my yahoo mailbox. But I find many sites are using some kind of image verification mechanism, which is possibly called CAPTCHA. I can do nothing to them. But one of the sites I'm trying to auto access is using a plain-text verification code. It is comnposed of four digits, selectable and copyable. But they're not in the source file which can be fetched using
$mech->content;
I searched for the keyword that appears on the webpage but not in the source file through all the files in the Temporary Internet Files but still can't find it.
Any idea what's going on? I was suspecting that the verification code was somehow hidden in some cookie file but I can't seem to find it :(
The following is the code that completes all the fields requirements except for the verification code:
use warnings;
use Win32::IE::Mechanize;
my $url = "http://www.zjsmap.com/smap/smap_login.jsp";
my $eccode = "myeccode";
my $username = "myaccountname";
my $password = "mypassword";
my $verify = "I can't figure out how to let the script get the code yet"
my $mech = Win32::IE::Mechanize->new(visible=>1);
$mech->get($url);
sleep(1); #avoids undefined value error
$mech->form_name("BaseForm");
$mech->field(ECCODE => $eccode);
$mech->field(MEMBERACCOUNT => $username);
$mech->field(PASSWORD => $password);
$mech->field(verify => $verify);
$mech->click();
Like always any suggestions/comments would be greatly appreciated :)
UPDATE
I've figured out a not-so-smart way to solve this problem. Please comment on my own asnwer posted below. Thanks like always :)
This is the reason why they are there. To stop program like yours to do automated stuff ;-)
A CAPTCHA or Captcha is a type of
challenge-response test used in
computing to ensure that the response
is not generated by a computer.
This appears to be an irrelevant number. The page uses it in 3 places: generating it; displaying it on the form next to the input field for it; and checking for the input value being equal to the random number chosen. That is, it is a client-only check. Still, if you disable javascript it looks like, I'm guessing, important cookies don't get set. If you can execute JavaScript in the context of the page (you should be able to with a get method call and a javascript URI), you could change the value of random_number to f.e. 42 and fill that in on the form.
The code is inserted by JavaScript – disable JS, reload the page and see it disappear. You have to hunt through the JS code to get an idea where it comes from and how to replicate it.
Thanks to james2vegas, zoul and Shoban.
I've finally figured out on my own a not-so-smart but at-least-workable way to solve the problem I described here. I'd like to share it here. I think the approach suggested by #james2vegas is probably much better...but anyway I'm learning along the way.
My approach is this:
Although the verification code is not in the source file but since it is still selectable and copyable, I can let my script copy everything in the login page and then extract the verification code.
To do this, I use the sendkeys functions in the Win32::Guitest module to do "Select All" and "Copy" to the login page.
Then I use Win32:Clipboard to get the clipboard content and then Regexp to extract the code. Something like this:
$verify = Win32::Clipboard::GetText();
$verify =~ s/.* (\d{4}).*/$1/msg;
A few thoughts:
The random number is generated by something like this in Perl
my $random_number = int(rand(8999)) + 1000; #var random_number = rand(1000,10000);
And then it checks if $verify == $random_number. I don't know how to catch the value of one-session-only $random_number. I think it is stored somewhere in the memory. If I can capture the value directly then I wouldn't have gone to so much trouble of using this and that extra module.