Revision: 826
Author: hussami
Date: 2013-06-20 15:29:48 +0200 (Thu, 20 Jun 2013)
Log Message:
-----------
ohanal.fix
Modified Paths:
--------------
trunk/misc/osiris_xml/ohanalyze.py
Modified: trunk/misc/osiris_xml/ohanalyze.py
===================================================================
--- trunk/misc/osiris_xml/ohanalyze.py 2013-06-20 12:15:35 UTC (rev 825)
+++ trunk/misc/osiris_xml/ohanalyze.py 2013-06-20 13:29:48 UTC (rev 826)
@@ -5,6 +5,7 @@
import sys
import MySQLdb
import re
+sys.path.append("../../web2py/applications/helyesiras_webdev/modules")
from egybekulon2_humor import StemmingAnalysis
from egybekulon2_humor import HumorAna
import itertools
@@ -22,6 +23,7 @@
def connect(self, server, user, pwd, dbs):
self.connection = MySQLdb.connect(host=server, user=user, passwd=pwd, \
+# db=dbs)
db=dbs, charset='utf8')
def disconnect(self):
@@ -56,8 +58,9 @@
self.data = {}
results = self.mysqlhandler.fetchall()
for row in results:
- self.data[row[0].encode("utf8")] = 0
-# self.data[row[0]] = 0
+# print row[0]
+# self.data[row[0].encode("utf8")] = 0
+ self.data[row[0]] = 0
# if c == 300:
# print row[0].encode("utf8")
# print self.data
@@ -313,9 +316,10 @@
if where == 0:
self.output("use dbdict", where)
self.output("drop table incidences", where)
- self.output("create table incidences(term varchar(100) " +\
-# self.output("create table incidences(term varchar(100) collate utf8_bin " +\
- "not null, dict_id int, " + "idx int)", where)
+# self.output("create table incidences(term varchar(100) " +\
+ self.output("create table incidences(term varchar(100) collate utf8_bin " +\
+ "not null, dict_id int, idx int) engine=MyISAM default charset=utf8 " +\
+ "collate=utf8_bin", where)
self.output("create index incidence_index on incidences(term)", where)
query = "select id, actual, norm from ohdict";