Python 2 error - to_ipaddress

[siyer32]

Not sure if this is supported in Python 2.7.13, I got this error. Works fine in Python 3.6.5

appdata['sa'] = cypd.to_ipaddress(appdata['sa'])

63 '%r does not appear to be an IPv4 or IPv6 address. ' 164 'Did you pass in a bytes (str in Python 2) instead of' --> 165 ' a unicode object?' % address) 166 167 raise ValueError('%r does not appear to be an IPv4 or IPv6 address' %

AddressValueError: '10.44.129.135' does not appear to be an IPv4 or IPv6 address. Did you pass in a bytes (str in Python 2) instead of a unicode object?

[siyer32]

Here is Python 3.6.5 output: appdata['sa'] = cypd.to_ipaddress(appdata['sa']) appdata.dtypes sa ip da object sp int64 dp int64 ipkt int64 ibyt int64 Application Label object

[TomAugspurger]

We either need to make a better error message here, or break with Python 2's ipaddress module.

In Python2, it expects unicode object when parsing a string IP Address like '192.168.1.1'. https://cyberpandas.readthedocs.io/en/latest/usage.html#parsing

In [13]: import pandas as pd

In [14]: from cyberpandas import to_ipaddress

In [15]: df = pd.DataFrame({"addr": ['192.168.1.1', '192.168.1.2']})

In [16]: to_ipaddress(df.addr)
---------------------------------------------------------------------------
AddressValueError                         Traceback (most recent call last)
<ipython-input-16-1f1c4ac488eb> in <module>()
----> 1 to_ipaddress(df.addr)

/Users/taugspurger/sandbox/cyberpandas/cyberpandas/parser.py in to_ipaddress(values)
     40         values = [values]
     41
---> 42     return IPArray(_to_ip_array(values))
     43
     44

/Users/taugspurger/sandbox/cyberpandas/cyberpandas/parser.py in _to_ip_array(values)
     59     elif not (isinstance(values, np.ndarray) and
     60               values.dtype == IPType._record_type):
---> 61         values = _to_int_pairs(values)
     62     return np.atleast_1d(np.asarray(values, dtype=IPType._record_type))
     63

/Users/taugspurger/sandbox/cyberpandas/cyberpandas/parser.py in _to_int_pairs(values)
     79         pass
     80     else:
---> 81         values = [ipaddress.ip_address(v)._ip for v in values]
     82         values = [unpack(pack(v)) for v in values]
     83     return values

/Users/taugspurger/miniconda3/envs/py27-ipaddr/lib/python2.7/site-packages/ipaddress.pyc in ip_address(address)
    163             '%r does not appear to be an IPv4 or IPv6 address. '
    164             'Did you pass in a bytes (str in Python 2) instead of'
--> 165             ' a unicode object?' % address)
    166
    167     raise ValueError('%r does not appear to be an IPv4 or IPv6 address' %

AddressValueError: '192.168.1.1' does not appear to be an IPv4 or IPv6 address. Did you pass in a bytes (str in Python 2) instead of a unicode object?

In [17]: to_ipaddress(df.addr.astype(unicode))
Out[17]: IPArray([u'192.168.1.1', u'192.168.1.2'])

So in literal code it should be u'192.168.1.1' instead of '192.168.1.1'. The current way is pretty unfriendly :/

[seibert]

By definition, IP address strings have to be ASCII (unlike hostnames), so I don't see a problem with to_ipaddress silently decoding Python 2 str to unicode assuming it is ASCII. Does that seem reasonable?

[TomAugspurger]

Does that seem reasonable?

Yeah, I think so.

[siyer32]

Does this mean, the ip address passed have to be strings ? Most data (like the one I tested) that are captured from the devices are not strings.

[seibert]

There are other input methods describe in the docs. Python integers, or raw address in byte form (see IPArray.from_bytes)

[TomAugspurger]

To clarify things, let's use Python 3's terminology. "string" is a unicode string, and "bytes" is a bytestring.

The valid options are

• A string in decimal-dot notation, consisting of four decimal integers in the inclusive range 0–255, separated by dots (e.g. 192.168.0.1). Each integer represents an octet (byte) in the address. Leading zeroes are tolerated only for values less than 8 (as there is no ambiguity between the decimal and octal interpretations of such strings).
• An integer that fits into 32 bits.
• An integer packed into a bytes object of length 4 (most significant octet first).

Most data (like the one I tested) that are captured from the devices are not strings.

What does the raw data look like for you? If performance is a concern, the absolute fastest was is https://cyberpandas.readthedocs.io/en/latest/api.html#cyberpandas.IPArray.from_bytes