Re: [pysnmp-users] pysnmp / pyasn1 decode problem with multi-line traps

[Tom Kr. <to...@we...>]

Some good news to the beginning of the new week.
After I changed the method from .getVarBindsList() to .getVarBinds(), I can read the data with the stupid line feed.
 
Now I tried to search and replace the line feed with this lines
            for oid, val in varBinds:
                print("DEBUG_oid: %s" % oid)
                print("DEUBG_val: %s" % val)
                new_val = "".join(val)
                val = re.sub(r"[\r|\s]", "", new_val)
                print("DEBUG_oid_2: %s" % oid)
                print("DEBUG_val_2: %s" % val)
 
Each time there is a trap with this characters I get this error message. 
    return _compile(pattern, flags).sub(repl, string, count)
;TypeError: expected string or buffer
 
How can I find out which type the val variable have?
And how can I change this to string?
 
Tom



> Am 12.03.2015 um 16:18 schrieb Ilya Etingof <il...@gl...>:
> 
> 
> Right, in 'val' you are getting the whole SNMP PDU value structure, not just a leaf scalar. That's the reason it does not match OctetString class instance check.
> 
> Use .getVarBinds() method instead of .getVarBindsList(). The former returns a list of tuples of terminal values, while the latter returns original var-bindings list from SNMP PDU which always holds a record-like structure for value.
> 
> -ilya
> 
>> On 03/12/2015 05:29 PM, Tom Kr. wrote:
>> Hmm, I'm confused.
>> With this changes I also get hex values back and I still enter the else part for each value.
>>  import sys,os
>>  sys.path.append('/usr/share/pyshared/')
>>  from pysnmp.carrier.asynsock.dispatch import AsynsockDispatcher
>> from pysnmp.carrier.asynsock.dgram import udp, udp6
>> from pyasn1.codec.ber import decoder
>> from pysnmp.proto import api
>> #from pysnmp.proto.rfc1902 import OctetString
>> from pyasn1.type.univ import OctetString
>>  def pySnmpCheck(oval):
>>     if isinstance(oval, OctetString):
>>         print("DEBUG: OctetString")
>>         val = oval.asOctets()
>>     else:
>>         print("DEBUG: else")
>>         val = oval
>>     return val
>>  If this information are useful, this is the header I've got.
>>  # python snmp-transports.py
>> Notification message from (1, 3, 6, 1, 6, 1, 1):('xxx.xxx.xxx.xxx', 52032):
>> Enterprise: 1.3.6.1.4.1.18494.2
>> Agent Address: xxx.xxx.xxx.xxx
>> Generic Trap: 'enterpriseSpecific'
>> Specific Trap: 1
>> Uptime: 1
>> Var-binds:
>> DEBUG: else
>> 1.3.6.1.4.1.18494.2.1.9 = ObjectSyntax:
>> simple=SimpleSyntax:
>> string=0x4f70657261746f722058585858585858585858203a20585858585858585858582c2058585858585858202c204c6f63616c6c792061757468656e74696361746564202d207375636365737366756c6c79207369676e6564206f6e20746f20746865207465726d696e616c20275858582e5858582e582e584058272061742031343a3238207573696e672027585858585858585820585858205858585858585858272e0a4f70657261746f722050726f66696c65733a2058585858585858585858585858
>>  Tom
>>    
>> 
>> 
>>> Am 12.03.2015 um 14:39 schrieb Ilya Etingof <il...@gl...>:
>>> 
>>> 
>>> Oh, if you are using SNMPv1 you should check against the base class:
>>> 
>>> from pyasn1.type.univ import OctetString
>>> 
>>> if isinstance(oval, OctetString):
>>>    ...
>>> 
>>> Perhaps this should be improved in pysnmp.
>>> 
>>> -ilya
>>> 
>>>> On 03/12/2015 04:29 PM, Tom Kr. wrote:
>>>> I made the following changes to the script.
>>>>  ...
>>>> from pysnmp.proto.rfc1902 import OctetString
>>>>  def pySnmpCheck(oval):
>>>>     if isinstance(oval, OctetString):
>>>>         print("DEBUG: OctetString")
>>>>         val = oval.asOctets()
>>>>     else:
>>>>         print("DEBUG: else")
>>>>         val = oval
>>>>     return val
>>>>  ...
>>>>             for oid, val in varBinds:
>>>>                 val = pySnmpCheck(val)
>>>>                 print('%s = %s' % (oid.prettyPrint(), val.prettyPrint()))
>>>> ...
>>>>  Then I start grabbing the snmp messages again.
>>>> Unfortunately every value of the message go through the else part.
>>>> But how could this happen?
>>>>  Is there a possibility to check which type of string the values have?
>>>>  Tom
>>>> 
>>>> 
>>>> 
>>>>> Am 11.03.2015 um 14:17 schrieb Ilya Etingof <il...@gl...>:
>>>>> 
>>>>> 
>>>>> That's right, .asOctets() is only defined for OctetString type and its
>>>>> subclasses. Therefore, to deal with octets of OctetString() you should
>>>>> do additional check:
>>>>> 
>>>>> from pysnmp.proto.rfc1902 import OctetString
>>>>> 
>>>>> ...
>>>>> 
>>>>> if isinstance(val, OctetString):
>>>>>     val = val.asOctets()
>>>>> 
>>>>> Keep in mind that octet string is not always the same as printable
>>>>> string. The former is not immune to your terminal potentially eating
>>>>> non-printable characters or garbling output in a funny way. Also, text
>>>>> encoding is not defined by OctetString. To be on a safe side default
>>>>> OctetString printout is done in hex whenever a non-printable is seen in
>>>>> its contents.
>>>>> 
>>>>> -ilya
>>>>> 
>>>>>>> On 03/11/2015 03:18 PM, Craig Small wrote:
>>>>>>> On Wed, Mar 11, 2015 at 07:02:44AM +0100, Tom wrote:
>>>>>>>    ;AttributeError: ObjectSyntax instance has no attribute 'asOctets'
>>>>>> Are you sure each row is returning something asOctets makes sense for?
>>>>>> An Integer doesn't make sense.
>>>>>> 
>>>>>>>    Does it only works if a hex value is send by snmp?
>>>>>> I believe it works for certain types of values, such as an OctetString
>>>>>> 
>>>>>>>    How can I check if the value is hex or not?
>>>>>> Not sure, but the types are found in pyasn1.type.univ so you could
>>>>>> compare it to, say, univ.OctetString
>>>>>> 
>>>>>> ObjectSyntax seems to be the superset of all types, from what I can tell
>>>>>> it means pysnmp wasn't sure what the type should be.
>>>>>> 
>>>>>> It's been an interesting and educational discussion anyhow.
>>>>>> 
>>>>>>  - Craig
>