| author | Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com> |
| Wed, 17 Jun 2015 17:28:35 +0200 | |
| changeset 1284 | dc1c1bed25e8 |
| parent 891 | 8628c590f608 |
| permissions | -rw-r--r-- |
|
9
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1 |
from optparse import OptionParser |
|
891
8628c590f608
Remove old script and correct obvious script errors
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
9
diff
changeset
|
2 |
from sqlite3 import connect, Row |
|
9
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
3 |
import re |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
4 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
5 |
if __name__ == "__main__" : |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
6 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
7 |
parser = OptionParser() |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
8 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
9 |
(options, args) = parser.parse_args() |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
10 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
11 |
conn = connect(args[0]) |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
12 |
conn.row_factory = Row |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
13 |
curs = conn.cursor() |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
14 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
15 |
names = {} |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
16 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
17 |
curs.execute("select tt.text as text from tweet_tweet as tt left join tweet_user as tu on tt.user = tu.rowid where tt.text like \"%ENMI09%\" order by tt.created_at_ts asc;") |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
18 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
19 |
regexp = re.compile("\@(\w+)") |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
20 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
21 |
for row in curs: |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
22 |
text = row["text"] |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
23 |
for m in regexp.finditer(text): |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
24 |
names[m.group(1)]=m.group(1) |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
25 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
26 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
27 |
print repr(names.keys()) |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
28 |
print repr(len(names.keys())) |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
29 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
30 |