In [48]:

    

xml = ""

with open ("../demos/dataset/nvdcve-2.0-2016_mini.xml") as f:
    xml = f.read()
    
xml;


    

In [52]:

    

import xmldataset

# xmldataset declaration
profile = """
nvd
    entry.id
"""

# Print the output
print(xmldataset.parse_using_profile(xml, profile))


    

    




{}

In [18]:

    

import pandas as pd

header = pd.read_csv("../demos/dataset/cve_allitems.csv.gz", skiprows=2, nrows=0)
header


    

    
Out[18]:







  

    

      

      
Name
      
Status
      
Description
      
References
      
Phase
      
Votes
      
Comments
    
  
  

  

In [36]:

    

cve = pd.read_csv(
    "../demos/dataset/cve_allitems.csv.gz",
    skiprows=10,
    encoding='latin-1',
    names=header
    )
cve.head()


    

    




C:\dev\apps\Anaconda3\lib\site-packages\pandas\io\parsers.py:709: UserWarning: Duplicate names specified. This will raise an error in the future.
  return _read(filepath_or_buffer, kwds)
C:\dev\apps\Anaconda3\lib\site-packages\IPython\core\interactiveshell.py:2728: DtypeWarning: Columns (6) have mixed types. Specify dtype option on import or set low_memory=False.
  interactivity=interactivity, compiler=compiler, result=result)






    
Out[36]:







  

    

      

      
Name
      
Status
      
Description
      
References
      
Phase
      
Votes
      
Comments
    
  
  

    

      
0
      
CVE-1999-0001
      
Candidate
      
ip_input.c in BSD-derived TCP/IP implementatio...
      
CERT:CA-98-13-tcp-denial-of-service   |   BUGT...
      
Modified (20051217)
      
MODIFY(1) Frech  |     NOOP(2) Northcutt, W...
      
Christey> A Bugtraq posting indicates that th...
    
    

      
1
      
CVE-1999-0002
      
Entry
      
Buffer overflow in NFS mountd gives root acces...
      
SGI:19981006-01-I   |   URL:ftp://patches.sgi....
      
NaN
      
NaN
      
NaN
    
    

      
2
      
CVE-1999-0003
      
Entry
      
Execute commands as root via buffer overflow i...
      
NAI:NAI-29   |   CERT:CA-98.11.tooltalk   |   ...
      
NaN
      
NaN
      
NaN
    
    

      
3
      
CVE-1999-0004
      
Candidate
      
MIME buffer overflow in email clients, e.g. So...
      
CERT:CA-98.10.mime_buffer_overflows   |   XF:o...
      
Modified (19990621-01)
      
ACCEPT(8) Baker, Cole, Collins, Dik, Landfi...
      
Frech> Extremely minor, but I believe e-mail ...
    
    

      
4
      
CVE-1999-0005
      
Entry
      
Arbitrary command execution via IMAP buffer ov...
      
CERT:CA-98.09.imapd   |   SUN:00177   |   URL:...
      
NaN
      
NaN
      
NaN
    
  

In [39]:

    

cve.describe(include='all')


    

    
Out[39]:







  

    

      

      
Name
      
Status
      
Description
      
References
      
Phase
      
Votes
      
Comments
    
  
  

    

      
count
      
135983
      
135983
      
135983
      
109073
      
132930
      
132930
      
1855
    
    

      
unique
      
135983
      
2
      
108157
      
97379
      
4452
      
1289
      
1774
    
    

      
top
      
CVE-2016-7580
      
Candidate
      
** RESERVED ** This candidate has been reserve...
      
CONFIRM:https://source.android.com/security/bu...
      
Assigned (20160909)
      
None (candidate not yet proposed)
      
Cole> See comments for CVE-2000-0101  |   Fre...
    
    

      
freq
      
1
      
132930
      
20918
      
262
      
1322
      
130004
      
11
    
  

In [41]:

    

cve.info()


    

    




<class 'pandas.core.frame.DataFrame'>
RangeIndex: 135983 entries, 0 to 135982
Data columns (total 7 columns):
Name           135983 non-null object
Status         135983 non-null object
Description    135983 non-null object
References     109073 non-null object
Phase          132930 non-null object
Votes          132930 non-null object
Comments       1855 non-null object
dtypes: object(7)
memory usage: 7.3+ MB

In [ ]:

    




    

In [42]:

    

import xmldataset

# xmldataset declaration
profile = """
colleagues
    colleague
        title = dataset:colleagues
        phone = dataset:colleagues
        email = dataset:colleagues"""

# Print the output
print(xmldataset.parse_using_profile(xml, profile))


    

    




---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)
<ipython-input-42-d1071ce1b97a> in <module>()
     10 
     11 # Print the output
---> 12 print(xmldataset.parse_using_profile(xml, profile))

NameError: name 'xml' is not defined