In [1]:
import sys,re,json,os,csv,glob
import graphlab as gl
---------------------------------------------------------------------------
ImportError Traceback (most recent call last)
<ipython-input-1-0351ad49e384> in <module>()
1 import sys,re,json,os,csv,glob
----> 2 import graphlab as gl
ImportError: No module named graphlab
In [3]:
# Create data folder
# !mkdir -p ../data
In [4]:
# Grab (JSON) files from S3, takes a bunch of time
# %%capture capt # should work in IPython 3
!s3cmd sync s3://PLNY_Brasil/2014-08 ../data/
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328220.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328220.json [1 of 295]
4373 of 4373 100% in 0s 96.40 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328256.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328256.json [2 of 295]
12821 of 12821 100% in 0s 141.28 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328262.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328262.json [3 of 295]
4434 of 4434 100% in 0s 108.35 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328273.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328273.json [4 of 295]
1990 of 1990 100% in 0s 34.41 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328332.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328332.json [5 of 295]
9958 of 9958 100% in 0s 254.65 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328339.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328339.json [6 of 295]
1980 of 1980 100% in 0s 33.75 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328374.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328374.json [7 of 295]
12578 of 12578 100% in 0s 312.09 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328445.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328445.json [8 of 295]
15760 of 15760 100% in 0s 367.33 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328484.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328484.json [9 of 295]
4073 of 4073 100% in 0s 84.27 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328544.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328544.json [10 of 295]
8430 of 8430 100% in 0s 91.18 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328569.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328569.json [11 of 295]
4266 of 4266 100% in 0s 87.84 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328687.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328687.json [12 of 295]
18457 of 18457 100% in 0s 227.25 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328706.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328706.json [13 of 295]
6901 of 6901 100% in 0s 139.51 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328743.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328743.json [14 of 295]
19578 of 19578 100% in 0s 472.97 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328813.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328813.json [15 of 295]
18932 of 18932 100% in 0s 462.17 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328849.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328849.json [16 of 295]
14474 of 14474 100% in 0s 327.24 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328917.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328917.json [17 of 295]
29220 of 29220 100% in 0s 612.53 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328930.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328930.json [18 of 295]
2236 of 2236 100% in 0s 47.12 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328967.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328967.json [19 of 295]
13001 of 13001 100% in 0s 273.10 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329038.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329038.json [20 of 295]
32883 of 32883 100% in 0s 564.28 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329048.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329048.json [21 of 295]
11268 of 11268 100% in 0s 204.08 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329086.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329086.json [22 of 295]
4019 of 4019 100% in 0s 100.33 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329091.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329091.json [23 of 295]
6319 of 6319 100% in 0s 127.81 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329096.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329096.json [24 of 295]
2024 of 2024 100% in 0s 35.56 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329103.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329103.json [25 of 295]
1920 of 1920 100% in 0s 52.27 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329165.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329165.json [26 of 295]
27527 of 27527 100% in 0s 507.56 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329177.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329177.json [27 of 295]
2135 of 2135 100% in 0s 42.51 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329216.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329216.json [28 of 295]
31669 of 31669 100% in 0s 751.98 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329288.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329288.json [29 of 295]
33188 of 33188 100% in 0s 714.81 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329294.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329294.json [30 of 295]
2381 of 2381 100% in 0s 51.84 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329298.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329298.json [31 of 295]
1994 of 1994 100% in 0s 39.61 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329304.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329304.json [32 of 295]
1962 of 1962 100% in 0s 40.09 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329309.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329309.json [33 of 295]
2229 of 2229 100% in 0s 32.48 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329346.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329346.json [34 of 295]
16865 of 16865 100% in 0s 386.89 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329424.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329424.json [35 of 295]
32442 of 32442 100% in 0s 751.02 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329430.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329430.json [36 of 295]
2458 of 2458 100% in 0s 56.55 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329468.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329468.json [37 of 295]
21301 of 21301 100% in 0s 467.56 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329478.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329478.json [38 of 295]
4063 of 4063 100% in 0s 84.59 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329501.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329501.json [39 of 295]
4002 of 4002 100% in 0s 75.67 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329529.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329529.json [40 of 295]
14534 of 14534 100% in 0s 223.39 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329567.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329567.json [41 of 295]
8196 of 8196 100% in 0s 162.23 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329572.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329572.json [42 of 295]
15964 of 15964 100% in 0s 239.72 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329576.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329576.json [43 of 295]
2299 of 2299 100% in 0s 45.26 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329643.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329643.json [44 of 295]
26737 of 26737 100% in 0s 549.09 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329652.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329652.json [45 of 295]
10760 of 10760 100% in 0s 196.76 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329657.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329657.json [46 of 295]
2208 of 2208 100% in 0s 45.89 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329695.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329695.json [47 of 295]
12829 of 12829 100% in 0s 345.36 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329775.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329775.json [48 of 295]
29742 of 29742 100% in 0s 542.81 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329782.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329782.json [49 of 295]
4281 of 4281 100% in 0s 109.63 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329786.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329786.json [50 of 295]
2174 of 2174 100% in 0s 44.12 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329822.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329822.json [51 of 295]
25008 of 25008 100% in 0s 509.23 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329894.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329894.json [52 of 295]
42053 of 42053 100% in 0s 838.16 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329901.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329901.json [53 of 295]
1981 of 1981 100% in 0s 36.52 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329940.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413329940.json [54 of 295]
29259 of 29259 100% in 0s 782.36 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330006.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330006.json [55 of 295]
19643 of 19643 100% in 0s 366.91 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330043.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330043.json [56 of 295]
6136 of 6136 100% in 0s 150.47 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330049.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330049.json [57 of 295]
8231 of 8231 100% in 0s 181.93 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330054.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330054.json [58 of 295]
2031 of 2031 100% in 0s 51.27 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330126.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330126.json [59 of 295]
51168 of 51168 100% in 0s 865.54 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330132.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330132.json [60 of 295]
14430 of 14430 100% in 0s 225.00 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330140.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330140.json [61 of 295]
2102 of 2102 100% in 0s 60.98 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330144.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330144.json [62 of 295]
4615 of 4615 100% in 0s 107.48 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330181.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330181.json [63 of 295]
12721 of 12721 100% in 0s 267.54 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330248.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330248.json [64 of 295]
20774 of 20774 100% in 0s 546.76 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330254.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330254.json [65 of 295]
2133 of 2133 100% in 0s 40.19 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330290.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330290.json [66 of 295]
28584 of 28584 100% in 0s 633.82 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330296.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330296.json [67 of 295]
2018 of 2018 100% in 0s 37.03 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330366.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330366.json [68 of 295]
35344 of 35344 100% in 0s 707.16 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330406.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330406.json [69 of 295]
8497 of 8497 100% in 0s 167.85 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330412.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330412.json [70 of 295]
1990 of 1990 100% in 0s 38.27 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330490.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330490.json [71 of 295]
32611 of 32611 100% in 0s 676.04 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330496.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330496.json [72 of 295]
4537 of 4537 100% in 0s 79.71 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330532.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330532.json [73 of 295]
29970 of 29970 100% in 0s 558.14 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330537.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330537.json [74 of 295]
4415 of 4415 100% in 0s 106.01 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330541.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330541.json [75 of 295]
2041 of 2041 100% in 0s 36.77 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330547.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330547.json [76 of 295]
6680 of 6680 100% in 0s 124.35 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330611.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330611.json [77 of 295]
39089 of 39089 100% in 0s 757.38 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330616.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330616.json [78 of 295]
21614 of 21614 100% in 0s 370.04 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330621.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330621.json [79 of 295]
1953 of 1953 100% in 0s 46.41 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330626.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330626.json [80 of 295]
4568 of 4568 100% in 0s 120.04 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330663.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330663.json [81 of 295]
21287 of 21287 100% in 0s 474.62 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330669.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330669.json [82 of 295]
2323 of 2323 100% in 0s 43.60 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330733.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330733.json [83 of 295]
52471 of 52471 100% in 0s 1198.77 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330738.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330738.json [84 of 295]
4519 of 4519 100% in 0s 105.96 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330743.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330743.json [85 of 295]
10153 of 10153 100% in 0s 351.52 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330779.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330779.json [86 of 295]
8530 of 8530 100% in 0s 205.94 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330849.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330849.json [87 of 295]
43961 of 43961 100% in 0s 1018.35 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330860.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330860.json [88 of 295]
6847 of 6847 100% in 0s 115.12 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330865.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330865.json [89 of 295]
2320 of 2320 100% in 0s 47.45 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330870.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330870.json [90 of 295]
2070 of 2070 100% in 0s 50.80 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330874.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330874.json [91 of 295]
19275 of 19275 100% in 0s 348.60 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330880.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330880.json [92 of 295]
4565 of 4565 100% in 0s 108.40 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330943.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330943.json [93 of 295]
16826 of 16826 100% in 0s 335.96 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330976.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413330976.json [94 of 295]
14901 of 14901 100% in 0s 247.73 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331014.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331014.json [95 of 295]
19607 of 19607 100% in 0s 438.89 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331019.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331019.json [96 of 295]
6471 of 6471 100% in 0s 159.79 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331024.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331024.json [97 of 295]
4417 of 4417 100% in 0s 48.33 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331035.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331035.json [98 of 295]
4567 of 4567 100% in 0s 111.54 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331060.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331060.json [99 of 295]
14709 of 14709 100% in 0s 302.59 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331087.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331087.json [100 of 295]
4143 of 4143 100% in 0s 93.70 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331129.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331129.json [101 of 295]
16764 of 16764 100% in 0s 456.96 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331213.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331213.json [102 of 295]
40160 of 40160 100% in 0s 935.27 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331233.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331233.json [103 of 295]
8395 of 8395 100% in 0s 168.60 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331270.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331270.json [104 of 295]
50636 of 50636 100% in 0s 999.68 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331278.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331278.json [105 of 295]
1975 of 1975 100% in 0s 51.19 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331314.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331314.json [106 of 295]
19496 of 19496 100% in 0s 561.69 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331331.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331331.json [107 of 295]
1934 of 1934 100% in 0s 45.61 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331336.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331336.json [108 of 295]
10100 of 10100 100% in 0s 229.60 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331348.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331348.json [109 of 295]
4413 of 4413 100% in 0s 85.59 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331350.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331350.json [110 of 295]
7088 of 7088 100% in 0s 263.80 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331356.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331356.json [111 of 295]
2050 of 2050 100% in 0s 57.68 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331409.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331409.json [112 of 295]
21912 of 21912 100% in 0s 421.34 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331456.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331456.json [113 of 295]
28961 of 28961 100% in 0s 569.68 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331496.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331496.json [114 of 295]
11915 of 11915 100% in 0s 277.13 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331501.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331501.json [115 of 295]
2330 of 2330 100% in 0s 46.26 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331566.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331566.json [116 of 295]
19027 of 19027 100% in 0s 351.32 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331573.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331573.json [117 of 295]
12610 of 12610 100% in 0s 309.09 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331666.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331666.json [118 of 295]
28576 of 28576 100% in 0s 631.15 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331688.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331688.json [119 of 295]
10993 of 10993 100% in 0s 243.97 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331726.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331726.json [120 of 295]
12682 of 12682 100% in 0s 243.91 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331812.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331812.json [121 of 295]
47364 of 47364 100% in 0s 314.97 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331848.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331848.json [122 of 295]
22685 of 22685 100% in 0s 494.86 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331941.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331941.json [123 of 295]
59831 of 59831 100% in 0s 1442.61 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331979.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331979.json [124 of 295]
11422 of 11422 100% in 0s 230.14 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331985.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413331985.json [125 of 295]
2067 of 2067 100% in 0s 37.07 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332023.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332023.json [126 of 295]
27306 of 27306 100% in 0s 514.92 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332052.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332052.json [127 of 295]
15608 of 15608 100% in 0s 437.42 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332093.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332093.json [128 of 295]
10955 of 10955 100% in 0s 285.84 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332171.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332171.json [129 of 295]
52303 of 52303 100% in 0s 1057.04 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332207.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332207.json [130 of 295]
4081 of 4081 100% in 0s 105.02 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332213.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332213.json [131 of 295]
12891 of 12891 100% in 0s 207.57 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332218.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332218.json [132 of 295]
2019 of 2019 100% in 0s 48.01 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332223.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332223.json [133 of 295]
2277 of 2277 100% in 0s 53.28 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332289.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332289.json [134 of 295]
20857 of 20857 100% in 0s 485.20 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332294.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332294.json [135 of 295]
12662 of 12662 100% in 0s 262.69 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332300.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332300.json [136 of 295]
2331 of 2331 100% in 0s 61.75 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332304.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332304.json [137 of 295]
4024 of 4024 100% in 0s 91.41 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332341.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332341.json [138 of 295]
15702 of 15702 100% in 0s 342.79 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332346.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332346.json [139 of 295]
11217 of 11217 100% in 0s 304.16 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332363.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332363.json [140 of 295]
4460 of 4460 100% in 0s 71.55 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332405.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332405.json [141 of 295]
12687 of 12687 100% in 0s 290.75 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332412.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332412.json [142 of 295]
4141 of 4141 100% in 0s 98.04 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332418.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332418.json [143 of 295]
2518 of 2518 100% in 0s 59.39 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332454.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332454.json [144 of 295]
20739 of 20739 100% in 0s 409.18 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332459.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332459.json [145 of 295]
1950 of 1950 100% in 0s 45.70 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332531.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332531.json [146 of 295]
23068 of 23068 100% in 0s 703.89 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332576.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332576.json [147 of 295]
6381 of 6381 100% in 0s 141.09 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332641.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332641.json [148 of 295]
8388 of 8388 100% in 0s 166.54 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332652.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332652.json [149 of 295]
2198 of 2198 100% in 0s 48.43 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332658.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332658.json [150 of 295]
2070 of 2070 100% in 0s 44.52 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332665.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332665.json [151 of 295]
1998 of 1998 100% in 0s 49.68 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332703.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332703.json [152 of 295]
3899 of 3899 100% in 0s 78.98 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332765.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332765.json [153 of 295]
12935 of 12935 100% in 0s 273.61 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332804.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332804.json [154 of 295]
13776 of 13776 100% in 0s 272.02 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332866.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332866.json [155 of 295]
31805 of 31805 100% in 0s 783.66 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332888.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332888.json [156 of 295]
6859 of 6859 100% in 0s 159.30 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332895.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332895.json [157 of 295]
6498 of 6498 100% in 0s 156.21 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332935.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332935.json [158 of 295]
19081 of 19081 100% in 0s 364.61 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332994.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413332994.json [159 of 295]
12199 of 12199 100% in 0s 325.58 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333022.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333022.json [160 of 295]
2278 of 2278 100% in 0s 52.12 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333033.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333033.json [161 of 295]
2129 of 2129 100% in 0s 34.86 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333067.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333067.json [162 of 295]
10623 of 10623 100% in 0s 179.53 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333076.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333076.json [163 of 295]
1991 of 1991 100% in 0s 33.31 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333138.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333138.json [164 of 295]
16537 of 16537 100% in 0s 335.80 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333179.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333179.json [165 of 295]
6317 of 6317 100% in 0s 171.80 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333246.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333246.json [166 of 295]
2300 of 2300 100% in 0s 60.28 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333254.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333254.json [167 of 295]
2173 of 2173 100% in 0s 48.00 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333291.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333291.json [168 of 295]
2096 of 2096 100% in 0s 54.70 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333353.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333353.json [169 of 295]
17157 of 17157 100% in 0s 396.08 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333371.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333371.json [170 of 295]
4206 of 4206 100% in 0s 82.22 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333407.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333407.json [171 of 295]
8901 of 8901 100% in 0s 166.02 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333485.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333485.json [172 of 295]
8014 of 8014 100% in 0s 232.85 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333524.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333524.json [173 of 295]
4092 of 4092 100% in 0s 83.07 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333529.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333529.json [174 of 295]
1940 of 1940 100% in 0s 34.41 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333606.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333606.json [175 of 295]
8580 of 8580 100% in 0s 198.16 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333615.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333615.json [176 of 295]
1913 of 1913 100% in 0s 43.20 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333623.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333623.json [177 of 295]
2130 of 2130 100% in 0s 46.89 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333661.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333661.json [178 of 295]
6516 of 6516 100% in 0s 120.89 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333669.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333669.json [179 of 295]
2311 of 2311 100% in 0s 45.77 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333731.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333731.json [180 of 295]
4675 of 4675 100% in 0s 111.89 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333741.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333741.json [181 of 295]
9548 of 9548 100% in 0s 252.42 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333779.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333779.json [182 of 295]
4305 of 4305 100% in 0s 93.70 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333860.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333860.json [183 of 295]
35640 of 35640 100% in 0s 619.95 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333863.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333863.json [184 of 295]
2069 of 2069 100% in 0s 27.25 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333901.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333901.json [185 of 295]
6264 of 6264 100% in 0s 134.58 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333968.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413333968.json [186 of 295]
37788 of 37788 100% in 0s 735.65 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334008.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334008.json [187 of 295]
8117 of 8117 100% in 0s 181.93 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334064.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334064.json [188 of 295]
18399 of 18399 100% in 0s 340.54 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334091.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334091.json [189 of 295]
9260 of 9260 100% in 0s 252.21 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334097.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334097.json [190 of 295]
6259 of 6259 100% in 0s 168.25 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334103.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334103.json [191 of 295]
1981 of 1981 100% in 0s 48.70 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334108.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334108.json [192 of 295]
2022 of 2022 100% in 0s 22.13 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334146.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334146.json [193 of 295]
19197 of 19197 100% in 0s 256.27 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334207.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334207.json [194 of 295]
16566 of 16566 100% in 0s 397.90 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334220.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334220.json [195 of 295]
4767 of 4767 100% in 0s 98.54 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334226.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334226.json [196 of 295]
4308 of 4308 100% in 0s 131.79 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334231.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334231.json [197 of 295]
8677 of 8677 100% in 0s 222.21 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334278.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334278.json [198 of 295]
8569 of 8569 100% in 0s 250.55 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334332.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334332.json [199 of 295]
12858 of 12858 100% in 0s 219.41 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334450.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334450.json [200 of 295]
14007 of 14007 100% in 0s 359.95 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334458.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334458.json [201 of 295]
4479 of 4479 100% in 0s 114.03 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334493.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334493.json [202 of 295]
6171 of 6171 100% in 0s 158.92 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334499.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334499.json [203 of 295]
3823 of 3823 100% in 0s 80.90 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334565.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334565.json [204 of 295]
4438 of 4438 100% in 0s 112.88 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334573.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334573.json [205 of 295]
1944 of 1944 100% in 0s 33.74 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334609.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334609.json [206 of 295]
14221 of 14221 100% in 0s 315.44 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334614.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334614.json [207 of 295]
2124 of 2124 100% in 0s 36.13 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334619.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334619.json [208 of 295]
1977 of 1977 100% in 0s 48.94 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334680.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334680.json [209 of 295]
36103 of 36103 100% in 0s 708.38 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334720.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334720.json [210 of 295]
6529 of 6529 100% in 0s 129.50 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334726.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334726.json [211 of 295]
10044 of 10044 100% in 0s 198.24 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334822.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334822.json [212 of 295]
38789 of 38789 100% in 0s 780.82 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334825.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334825.json [213 of 295]
4051 of 4051 100% in 0s 87.02 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334861.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334861.json [214 of 295]
23586 of 23586 100% in 0s 610.44 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334930.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334930.json [215 of 295]
34223 of 34223 100% in 0s 613.77 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334969.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413334969.json [216 of 295]
15007 of 15007 100% in 0s 363.61 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335049.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335049.json [217 of 295]
33912 of 33912 100% in 0s 899.31 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335054.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335054.json [218 of 295]
1981 of 1981 100% in 0s 40.27 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335091.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335091.json [219 of 295]
10862 of 10862 100% in 0s 247.35 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335171.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335171.json [220 of 295]
21670 of 21670 100% in 0s 434.73 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335176.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335176.json [221 of 295]
2032 of 2032 100% in 0s 32.64 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335181.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335181.json [222 of 295]
1990 of 1990 100% in 0s 75.02 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335219.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335219.json [223 of 295]
7639 of 7639 100% in 0s 198.75 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335281.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335281.json [224 of 295]
1826 of 1826 100% in 0s 35.53 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335325.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335325.json [225 of 295]
9839 of 9839 100% in 0s 284.32 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335409.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335409.json [226 of 295]
32663 of 32663 100% in 0s 929.98 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335447.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335447.json [227 of 295]
20782 of 20782 100% in 0s 480.31 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335535.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335535.json [228 of 295]
26629 of 26629 100% in 0s 671.02 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335571.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335571.json [229 of 295]
10670 of 10670 100% in 0s 268.04 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335652.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335652.json [230 of 295]
44583 of 44583 100% in 0s 1085.87 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335690.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335690.json [231 of 295]
3911 of 3911 100% in 0s 89.20 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335695.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335695.json [232 of 295]
4848 of 4848 100% in 0s 90.98 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335777.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335777.json [233 of 295]
18351 of 18351 100% in 0s 556.45 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335783.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335783.json [234 of 295]
2286 of 2286 100% in 0s 67.26 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335787.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335787.json [235 of 295]
1972 of 1972 100% in 0s 32.49 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335824.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335824.json [236 of 295]
6288 of 6288 100% in 0s 107.75 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335842.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335842.json [237 of 295]
17677 of 17677 100% in 0s 368.50 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335885.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335885.json [238 of 295]
10170 of 10170 100% in 0s 199.93 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335930.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335930.json [239 of 295]
9358 of 9358 100% in 0s 250.74 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335987.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413335987.json [240 of 295]
21030 of 21030 100% in 0s 578.28 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336042.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336042.json [241 of 295]
18955 of 18955 100% in 0s 331.24 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336101.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336101.json [242 of 295]
27102 of 27102 100% in 0s 730.97 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336131.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336131.json [243 of 295]
6521 of 6521 100% in 0s 169.50 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336139.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336139.json [244 of 295]
4042 of 4042 100% in 0s 77.94 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336176.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336176.json [245 of 295]
9026 of 9026 100% in 0s 231.53 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336253.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336253.json [246 of 295]
35984 of 35984 100% in 0s 761.21 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336288.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336288.json [247 of 295]
10479 of 10479 100% in 0s 179.69 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336380.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336380.json [248 of 295]
54270 of 54270 100% in 0s 1333.08 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336385.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336385.json [249 of 295]
2008 of 2008 100% in 0s 53.86 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336390.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336390.json [250 of 295]
1928 of 1928 100% in 0s 47.07 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336426.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336426.json [251 of 295]
33753 of 33753 100% in 0s 856.93 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336443.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336443.json [252 of 295]
8556 of 8556 100% in 0s 194.60 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336487.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336487.json [253 of 295]
56651 of 56651 100% in 0s 1108.06 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336526.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336526.json [254 of 295]
22141 of 22141 100% in 0s 658.95 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336531.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336531.json [255 of 295]
3972 of 3972 100% in 0s 91.35 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336618.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336618.json [256 of 295]
33238 of 33238 100% in 0s 720.13 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336654.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336654.json [257 of 295]
17142 of 17142 100% in 0s 458.25 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336733.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336733.json [258 of 295]
10553 of 10553 100% in 0s 271.11 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336741.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336741.json [259 of 295]
2288 of 2288 100% in 0s 67.84 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336752.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336752.json [260 of 295]
18549 of 18549 100% in 0s 439.73 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336795.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336795.json [261 of 295]
9040 of 9040 100% in 0s 196.86 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336847.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336847.json [262 of 295]
38229 of 38229 100% in 0s 921.30 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336885.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336885.json [263 of 295]
20913 of 20913 100% in 0s 428.99 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336891.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336891.json [264 of 295]
4387 of 4387 100% in 0s 74.99 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336898.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336898.json [265 of 295]
6753 of 6753 100% in 0s 107.08 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336972.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336972.json [266 of 295]
20503 of 20503 100% in 0s 490.70 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336980.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336980.json [267 of 295]
8306 of 8306 100% in 0s 113.68 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336985.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336985.json [268 of 295]
1906 of 1906 100% in 0s 46.94 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336990.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413336990.json [269 of 295]
2200 of 2200 100% in 0s 55.93 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337026.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337026.json [270 of 295]
12656 of 12656 100% in 0s 302.94 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337093.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337093.json [271 of 295]
12637 of 12637 100% in 0s 286.78 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337097.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337097.json [272 of 295]
1891 of 1891 100% in 0s 33.21 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337103.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337103.json [273 of 295]
2552 of 2552 100% in 0s 64.86 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337108.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337108.json [274 of 295]
11685 of 11685 100% in 0s 308.30 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337144.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337144.json [275 of 295]
7942 of 7942 100% in 0s 172.69 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337153.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337153.json [276 of 295]
19833 of 19833 100% in 0s 492.97 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337188.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337188.json [277 of 295]
6581 of 6581 100% in 0s 114.35 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337212.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337212.json [278 of 295]
17009 of 17009 100% in 0s 369.74 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337248.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337248.json [279 of 295]
4416 of 4416 100% in 0s 110.50 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337253.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337253.json [280 of 295]
4077 of 4077 100% in 0s 82.45 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337331.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337331.json [281 of 295]
15385 of 15385 100% in 0s 435.24 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337338.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337338.json [282 of 295]
12705 of 12705 100% in 0s 279.85 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337386.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337386.json [283 of 295]
12524 of 12524 100% in 0s 309.04 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337440.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337440.json [284 of 295]
11477 of 11477 100% in 0s 292.44 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337459.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337459.json [285 of 295]
2439 of 2439 100% in 0s 50.49 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337490.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337490.json [286 of 295]
15310 of 15310 100% in 0s 410.83 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337498.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337498.json [287 of 295]
4234 of 4234 100% in 0s 123.09 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337559.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337559.json [288 of 295]
13340 of 13340 100% in 0s 249.77 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337616.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337616.json [289 of 295]
11185 of 11185 100% in 0s 224.24 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337626.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337626.json [290 of 295]
4392 of 4392 100% in 0s 148.03 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337658.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337658.json [291 of 295]
4195 of 4195 100% in 0s 76.51 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337680.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337680.json [292 of 295]
16872 of 16872 100% in 0s 377.58 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337731.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337731.json [293 of 295]
13363 of 13363 100% in 0s 214.13 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337794.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337794.json [294 of 295]
6634 of 6634 100% in 0s 156.94 kB/s done
s3://PLNY_Brasil/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337857.json -> ../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413337857.json [295 of 295]
2243 of 2243 100% in 0s 51.07 kB/s done
Done. Downloaded 3768192 bytes in 13.9 seconds, 265.31 kB/s
In [5]:
files=glob.glob('../data/2014-08/*json')
files.sort()
print files[0]
../data/2014-08/DataSift-0caa7377dc9c1434844fcd67cb83f477-1413328220.json
In [6]:
# Find Number of JSON Files
print('We have a total of %d files' % len(files))
We have a total of 295 files
In [7]:
tweets=[]
for file in files:
# Cycle through files
fileString=open(file,'r').read().decode('utf-8')
# Read file as one long string and convert to unicode
fileTweets=[json.loads(line) for line in fileString.split('\n')]
# Split into lines and load as JSON
tweets.extend(fileTweets)
# Add list of tweets from file to global list
print('We have %d tweets' % len(tweets))
We have 1748 tweets
In [8]:
intLinebreakError=0
tweetLinebreakError=0
for tweet in tweets:
try:
tweet['interaction']['content'] = tweet['interaction']['content'].replace('\n', ' ').replace('\r', '')
except:
intLinebreakError+=1
tweet['interaction']['content']= 'NaN'
try:
tweet['twitter']['text'] = tweet['twitter']['text'].replace('\n', ' ').replace('\r', '')
except:
tweetLinebreakError+=1
tweet['twitter']['text']='NaN'
print('Failed removing line breaks in %d interaction content' % intLinebreakError)
print('Failed removing line breaks in %d tweets' % tweetLinebreakError)
Failed removing line breaks in 0 interaction content
Failed removing line breaks in 0 tweets
In [9]:
geoError=0
for tweet in tweets:
try:
tweet['geolocated']=geo.geoLocate(tweet['twitter']['retweet']['user']['location'])[0][3]
except:
try:
tweet['geolocated']=geo.geoLocate(tweet['twitter']['user']['location'])[0][3]
except:
geoError+=1
tweet['geolocated']=None
print('Couldn\'t geolocate %d tweets' % geoError)
print('Geolocated %d tweets' % (len(tweets) - (geoError)))
print('Managed to geolocate %d percent' % (100.0*(1.0-(float(geoError)/len(tweets)))))
Couldn't geolocate 893 tweets
Geolocated 855 tweets
Managed to geolocate 48 percent
In [10]:
# Testing that it worked
tweets[3]['geolocated']
Out[10]:
u'BR'
In [12]:
import gender
g=gender.Gender()
g.gender(tweets[0]['interaction']['author']['name']) #Testing that it works
Out[12]:
{u'JR': {'gender': 'm',
'probability': 1.0,
'volume_female': 0.0,
'volume_male': 2816.0}}
In [13]:
# Gender of tweeter or retweeter
genderError=0
for tweet in tweets:
try:
tweet['gender']=g.gender(tweet['interaction']['author']['name'])
except:
genderError+=1
tweet['gender']=None
print('Couldn\'t add gender probability for %d tweets' % genderError)
print('Managed to add gender to %d p.c.' % (100.0*(1.0-(float(genderError)/len(tweets)))))
Couldn't add gender probability for 0 tweets
Managed to add gender to 100 p.c.
In [15]:
# Testing that it worked
tweets[0]['gender'].values()[0]['gender']
Out[15]:
'm'
In [16]:
topics=['Campaign','Discrimination','Prevention','Testing']
emptyTopics=[0 for t in topics]
header=['id','time','content','type','datasift_lang','twitter_lang','twitter_location','UNGP_location',
'datasift_gender','UNGP_gender','gender_prob','followers','friends','topic','subtopic']
header.extend(topics)
header.extend(['interaction_hashtags','twitter_mentions','normalised_links','links_domain','user_description',
'user_screen_name'])
In [17]:
with open('../data/2014-08.json','wb') as f: f.write(json.dumps(tweets))
In [18]:
outFile=csv.writer(open('../data/2014-08.tsv','wb'),delimiter='\t')
outFile.writerow(header)
In [19]:
nIdError=0
nDateError=0
nContentError=0
nTypeError=0
nLanguageError=0
nTwitterLanguageError=0
nLocationError=0
nUngpLocationError=0
nGenderError=0
nUngpGenderError=0
nUngpGenderProbError=0
nFollowersError=0
nFriendsError=0
nTopicKeyError=0
nTopicValueError=0
nTopicError=0
nTopicLengthError=0
nTagsError=0
nMentionsError=0
nLinksError=0
nDomainsError=0
nDescriptionError=0
nScreenNameError=0
documents=[]
for tweet in tweets:
outList=[]
try:
outList.append(tweet['interaction']['id'])
documents.append(tweet['interaction']['id'])
except:
outList.append('NaN')
documents.append('NaN')
nIdError+=1
try:
outList.append(tweet['interaction']['created_at'])
documents.append(tweet['interaction']['created_at'])
except:
outList.append('NaN')
documents.append('NaN')
nDateError+=1
try:
outList.append(tweet['interaction']['content'].encode('utf-8').replace('\n',' '))
documents.append(tweet['interaction']['content'].encode('utf-8').replace('\n',' '))
except:
outList.append('NaN')
documents.append('NaN')
nContentError+=1
try:
outList.append(tweet['interaction']['type'].encode('utf-8'))
documents.append(tweet['interaction']['type'].encode('utf-8'))
except:
outList.append('NaN')
documents.append('NaN')
nTypeError+=1
try:
outList.append(tweet['language']['tag'].encode('utf-8'))
documents.append(tweet['language']['tag'].encode('utf-8'))
except:
outList.append('NaN')
documents.append('NaN')
nLanguageError+=1
try:
outList.append(tweet['twitter']['lang'].encode('utf-8'))
documents.append(tweet['twitter']['lang'].encode('utf-8'))
except:
outList.append('NaN')
documents.append('NaN')
nTwitterLanguageError+=1
try:
outList.append(tweet['twitter']['user']['location'].encode('utf-8'))
documents.append(tweet['twitter']['user']['location'].encode('utf-8'))
except:
outList.append('NaN')
documents.append('NaN')
nLocationError+=1
try:
outList.append(tweet['geolocated'].encode('utf-8'))
documents.append(tweet['geolocated'].encode('utf-8'))
except:
outList.append('NaN')
documents.append('NaN')
nUngpLocationError+=1
try:
outList.append(tweet['demographic']['gender'].encode('utf-8'))
documents.append(tweet['demographic']['gender'].encode('utf-8'))
except:
outList.append('NaN')
documents.append('NaN')
nGenderError+=1
try:
outList.append(tweet['gender'].values()[0]['gender'].encode('utf-8'))
documents.append(tweet['gender'].values()[0]['gender'].encode('utf-8'))
except:
outList.append('NaN')
documents.append('NaN')
nUngpGenderError+=1
try:
outList.append(tweet['gender'].values()[0]['probability'])
documents.append(tweet['gender'].values()[0]['probability'])
except:
outList.append('NaN')
documents.append('NaN')
nUngpGenderProbError+=1
try:
outList.append(tweet['twitter']['user']['followers_count'])
documents.append(tweet['twitter']['user']['followers_count'])
except:
outList.append('NaN')
documents.append('NaN')
nFollowersError+=1
try:
outList.append(tweet['twitter']['user']['friends_count'])
documents.append(tweet['twitter']['user']['friends_count'])
except:
outList.append('NaN')
documents.append('NaN')
nFriendsError+=1
try:
outList.append(tweet['interaction']['tag_tree']['topic'].keys()[0].encode('utf-8'))
documents.append(tweet['interaction']['tag_tree']['topic'].keys()[0].encode('utf-8'))
except:
outList.append('NaN')
documents.append('NaN')
nTopicKeyError+=1
try:
outList.append(tweet['interaction']['tag_tree']['topic'].values()[0][0].encode('utf-8'))
documents.append(tweet['interaction']['tag_tree']['topic'].values()[0][0].encode('utf-8'))
except:
outList.append('NaN')
documents.append('NaN')
nTopicValueError+=1
try:
tweetTopics=tweet['interaction']['tag_tree']['topic']
binaryTopics=[0 for e in emptyTopics]
for t in tweetTopics:
binaryTopics[topics.index(t)]=1
outList.extend(binaryTopics)
documents.extend(binaryTopics)
except:
outList.extend(emptyTopics)
nTopicError+=1
try:
tweetTags=','.join([h.lower() for h in tweet['interaction']['hashtags']])
outList.append(tweetTags.decode('utf-8'))
documents.append(tweetTags.decode('utf-8'))
except:
nTagsError+=1
outList.append('NaN')
documents.append('NaN')
try:
tweetMentions=','.join([m.lower() for m in tweet['twitter']['mentions']])
outList.append(tweetMentions.decode('utf-8'))
documents.append(tweetMentions.decode('utf-8'))
except:
nMentionsError+=1
outList.append('NaN')
documents.append('NaN')
try:
tweetLinks=','.join(tweet['links']['normalized_url'])
outList.append(tweetLinks.decode('utf-8'))
documents.append(tweetLinks.decode('utf-8'))
except:
nLinksError+=1
outList.append('NaN')
documents.append('NaN')
try:
tweetDomain=','.join(tweet['links']['domain'])
outList.append(tweetDomain.decode('utf-8'))
documents.append(tweetDomain.decode('utf-8'))
except:
nDomainsError+=1
outList.append('NaN')
documents.append('NaN')
try:
outList.append(tweet['twitter']['user']['description'].encode('utf-8'))
documents.append(tweet['twitter']['user']['description'].encode('utf-8'))
except:
nDescriptionError+=1
outList.append('NaN')
documents.append('NaN')
try:
outList.append(tweet['twitter']['user']['screen_name'].encode('utf-8'))
documents.append(tweet['twitter']['user']['screen_name'].encode('utf-8'))
except:
nScreenNameError+=1
outList.append('NaN')
documents.append('NaN')
outFile.writerow(outList)
print "%d ID errors." % nIdError
print "%d Date errors." % nDateError
print "%d Content errors." % nContentError
print "%d Type errors." % nTypeError
print "%d DataSift language errors." % nLanguageError
print "%d Twitter language errors." % nTwitterLanguageError
print "%d Twitter Location errors." % nLocationError
print "%d UNGP Location errors." % nUngpLocationError
print "%d Gender errors." % nGenderError
print "%d UNGP gender errors." % nUngpGenderError
print "%d UNGP gender probability errors." % nUngpGenderProbError
print "%d Follower errors." % nFollowersError
print "%d Friends errors." % nFriendsError
print "%d Topic Key errors." % nTopicKeyError
print "%d Topic Value errors." % nTopicValueError
print "%d Topic errors." % nTopicError
print "%d Topic lengtherrors." % nTopicLengthError
print "%d Interaction hashtag errors." % nTagsError
print "%d Interaction mention errors." % nMentionsError
print "%d Interaction link errors." % nLinksError
print "%d Domain errors." % nDomainsError
print "%d Description errors." % nDescriptionError
print "%d Screen name errors." % nScreenNameError
0 ID errors.
0 Date errors.
0 Content errors.
0 Type errors.
87 DataSift language errors.
0 Twitter language errors.
569 Twitter Location errors.
893 UNGP Location errors.
837 Gender errors.
719 UNGP gender errors.
719 UNGP gender probability errors.
0 Follower errors.
0 Friends errors.
0 Topic Key errors.
0 Topic Value errors.
0 Topic errors.
0 Topic lengtherrors.
1668 Interaction hashtag errors.
1245 Interaction mention errors.
1725 Interaction link errors.
1748 Domain errors.
189 Description errors.
0 Screen name errors.
In [1]:
from IPython.core.display import HTML
styles = open("../css/custom.css", "r").read()
HTML(styles)
Out[1]:
In [ ]:
Content source: renecnielsen/twitter-diy
Similar notebooks: