CODEWITHSUNDEEP
pythondata-processing
Pete Houston
Pete Houston

Reputation: 15089

Combine data on single input

I am working on processing some data storage. However, after pre-processing, the data is like this, for example:

-1|news.cnet.com|Technology News - CNET News|-1|-1
-1|news.google.com|Google News|-1|-1
-1|www.bbc.co.uk|BBC News - Home|-1|-1
-1|www.cnn.com|CNN.com|-1|-1
-1|www.news.com.au|News.com.au|-1|-1
1|news.google.com|-1|2|5,156,672
2|www.cnn.com|-1|71|325,362
3|www.news.com.au|-1|569|74,584
4|www.bbc.co.uk|-1|49|442,302
5|news.cnet.com|-1|107|187,705

The format is like INDEX|URL|TITLE|RANK|SLI. The value -1 indicates that the column not having a specific value. There are possible of duplicate entries with the same URL, merging them all will complete the record.

Is there a neat trick and tip for quickly combine these records into one complete? I don't want to iterate and loop repetition for all lines to find the duplicate one and merge.

EDIT: The expecting output is like:

1|news.google.com|Google News|2|5,156,672
2|www.cnn.com|CNN.com|71|325,362
3|www.news.com.au|News.com.au|569|74,584
4|www.bbc.co.uk|BBC News - Home|49|442,302
5|news.cnet.com|Technology News - CNET News|107|187,705

EDIT 2: By using Panda, as root suggested below, I'm able to merge data columns:

from pandas import *

frame = read_csv(r'data.txt', sep='|', names=['index', 'url', 'title', 'rank', 'sli'])
mask = frame['index'].map(lambda x: x > 0)

frame1 = frame[mask].set_index('url')
frame2 = frame[~mask].set_index('url')

frame1.title = frame2.title
frame1.set_index('index')
print frame1

However, is there any quick way around w/o using any third-party libs?

Upvotes: 2

Views: 120

Answers (1)

root
root

Reputation: 80346

You can load the data into the pandas DataFrame and process it.

from pandas import *

In [360]: frame=read_csv(r'C:\Python26\test.csv',sep='|', names=['index', 'url', 'title','rank','sli'])

In [361]: print frame
   index              url                        title  rank        sli
0     -1    news.cnet.com  Technology News - CNET News    -1         -1
1     -1  news.google.com                  Google News    -1         -1
2     -1    www.bbc.co.uk              BBC News - Home    -1         -1
3     -1      www.cnn.com                      CNN.com    -1         -1
4     -1  www.news.com.au                  News.com.au    -1         -1
5      1  news.google.com                           -1     2  5,156,672
6      2      www.cnn.com                           -1    71    325,362
7      3  www.news.com.au                           -1   569     74,584
8      4    www.bbc.co.uk                           -1    49    442,302
9      5    news.cnet.com                           -1   107    187,705

In [362]: mask = frame['index'].map(lambda x: x>0)

In [363]: frame = frame[mask]

In [364]: print frame
   index              url title  rank        sli
5      1  news.google.com    -1     2  5,156,672
6      2      www.cnn.com    -1    71    325,362
7      3  www.news.com.au    -1   569     74,584
8      4    www.bbc.co.uk    -1    49    442,302
9      5    news.cnet.com    -1   107    187,705

if you have further duplicates, use:

df.drop_duplicates()

Also, notice that after you have dropped the dublicates from index you can "reindex":

In [372]: print frame.set_index('index')
                   url title  rank        sli
index                                        
1      news.google.com    -1     2  5,156,672
2          www.cnn.com    -1    71    325,362
3      www.news.com.au    -1   569     74,584
4        www.bbc.co.uk    -1    49    442,302
5        news.cnet.com    -1   107    187,705

Upvotes: 3

Related Questions