script pour établir une bdd Ãà partir d'un tsv avec différent format - retour accueil
git clone git://bebou.netlib.re/laradb
Log | Files | Refs |
date_epoch.sh (2301B)
1 #! /bin/sh
2
3 # what this script does :
4 # - cleaning the date date so it is under the unix epoch format so that
5 # - it can be used as a value for bc in src/frise.sh
6
7 # @TODO: apply to both field, and use tsv utils rather than cut to use field header names instead of -f4,5 if
8 # if one day the start_date and end_date aren't anymore the 4th and 5th field of the tsv (however header names must stay persistent
9 # NaN values are defaulted to 1900 01 01 00:00
10
11 #start= `mktemp tmp/start.XXXXXXX`
12 #end= `mktemp tmp/end.XXXXXXX `
13 # | xargs date -j +%s # to transform every value into seconds since epoch to bc for the frise
14
15 clean(){
16 cut -f$1 -d" " src/db.tsv |
17 sed '1d;
18 s/-//g; # date 19701210
19 s/^\([0-9]*\)\(.*$\)/\1 #\2/g; # prepare for cleaning with comment
20 s/^\(.{2}\).*/\1/; # finale clean 1970 #comment
21 s/^\([0-9]*\)/\100000000/; #adding the zero we need for the date xargs command to work
22 # second part data cleaning for xargs date
23 s/[0-9]/X/13; #preparation to delete the non-needed zero
24 s/X.[^#]*/ /g; #delete the zero
25 s/^\([12]...\)0000/\10101/; # 197001010000 @TODO for date after 2000**
26 s/^\([12].....\)00/\101/; # 197009010000
27 s/#/ #/; # make the comments a new field #| once there was a pipe
28 s/^0\{8\}/190001010000/; # for NaN gives a date
29 # for debian date add :
30 s/0000 / /; # in format YYYYMMDD '
31 #cut -f1 |
32 #xargs -n1 date -j +%s
33 }
34 clean 5 > tmp/start
35 clean 6 > tmp/end
36 paste -d "\t" tmp/start tmp/end
37
38 cut -f1 tmp/start |
39 xargs -I XX -d'\n' -n1 date +%s --date="XX" |
40 tee tmp/start_unixepoch
41
42 cut -f1 tmp/end |
43 xargs -I XX -d'\n' -n1 date +%s --date="XX" |
44 tee tmp/end_unixepoch
45
46 paste -d "\t" tmp/start_unixepoch tmp/end_unixepoch | tee tmp/dateEpoch_final.tsv
47
48 # append the epoch time to the one tsv
49
50 #cat tmp/dateEpoch_final.tsv |
51 #sed '1s/^/start_unixepoch end_unixepoch @ /' |
52 # tr '@' '\n' | tee tmp/dateEpoch_final.tsv
53
54 cp --attributes-only tmp/dateEpoch_final.tsv tmp/dateEpoch_final.tsv.tmp
55 <tmp/dateEpoch_final.tsv sed '1s/^/start_unixepoch end_unixepoch @ /' |
56 tr '@' '\n' > tmp/dateEpoch_final.tsv.tmp &&
57 mv tmp/dateEpoch_final.tsv.tmp tmp/dateEpoch_final.tsv
58 cat tmp/dateEpoch_final.tsv
59
60 cp --attributes-only src/db.tsv tmp/db.tmp
61 paste -d "\t" src/db.tsv tmp/dateEpoch_final.tsv > tmp/db.tmp &&
62 mv tmp/db.tmp src/db.tsv
63 cut -f19- src/db.tsv
64
65
66