|
1 ''' |
|
2 Created on 27 juin 2012 |
|
3 |
|
4 @author: gerard |
|
5 ''' |
|
6 import MySQLdb |
|
7 import logging |
|
8 import warnings |
|
9 |
|
10 logger = logging.getLogger('document') |
|
11 |
|
12 class MediapartReader() : |
|
13 def __init__(self): |
|
14 self.conn = MySQLdb.connect (host = 'localhost', |
|
15 user = 'mediapart', |
|
16 passwd = 'mediapart', |
|
17 db = 'www_mediapart_fr') |
|
18 |
|
19 warnings.filterwarnings('ignore', category=MySQLdb.Warning) |
|
20 |
|
21 def __del__(self): |
|
22 self.conn.close () |
|
23 |
|
24 def get_date(self, document_id): |
|
25 |
|
26 self.conn.query('SELECT changed FROM node where nid=\''+document_id+'\'') |
|
27 r = self.conn.use_result() |
|
28 row = r.fetch_row() |
|
29 if row: |
|
30 date = (row)[0][0] |
|
31 else: |
|
32 date = '0' |
|
33 |
|
34 #changed_date = datetime.datetime.fromtimestamp(int(change_date_unix[0])).strftime('%Y-%m-%d %H:%M:%S') |
|
35 return date |
|
36 |
|
37 def get_title(self, document_id): |
|
38 self.conn.query('SELECT title FROM node where nid=\''+document_id+'\'') |
|
39 r = self.conn.use_result() |
|
40 row = r.fetch_row() |
|
41 if row: |
|
42 title = (row)[0][0].decode("windows-1252").encode("utf8") |
|
43 else: |
|
44 title = '' |
|
45 return title |
|
46 |
|
47 '''def get_category(self, document_id): |
|
48 self.conn.query('SELECT dst FROM url_alias where src="node/'+document_id+'"') |
|
49 r = self.conn.use_result() |
|
50 row = r.fetch_row() |
|
51 if row: |
|
52 dst = (row)[0][0] |
|
53 fields = dst.split('/') |
|
54 category = fields[0] |
|
55 else: |
|
56 category = '' |
|
57 return category''' |
|
58 |
|
59 |
|
60 def get_category(self, document_id): |
|
61 cursor =self.conn.cursor() |
|
62 cursor.execute('SELECT tid FROM term_node where nid=\''+document_id+'\'') |
|
63 tids = cursor.fetchall() |
|
64 category = '' |
|
65 for tid in tids: |
|
66 if str(tid[0]) == '27': |
|
67 category = 'France' |
|
68 return category |
|
69 elif str(tid[0]) == '28': |
|
70 category = 'International' |
|
71 return category |
|
72 elif str(tid[0]) == '29': |
|
73 category = 'Economie' |
|
74 return category |
|
75 elif str(tid[0]) == '30': |
|
76 category = 'Culture' |
|
77 return category |
|
78 else: |
|
79 category = '' |
|
80 return category |
|
81 |
|
82 def get_tags(self, document_id): |
|
83 tags=[] |
|
84 cursor = self.conn.cursor() |
|
85 cursor.execute('SELECT tid FROM term_node where nid=\''+document_id+'\'') |
|
86 tids = cursor.fetchall() |
|
87 for tid in tids: |
|
88 if str(tid[0]) != '27' and str(tid[0]) != '28' and str(tid[0]) != '29' and str(tid[0]) != '30': |
|
89 cursor.execute('SELECT name FROM term_data where tid=' + str(tid[0])) |
|
90 names = cursor.fetchall() |
|
91 for name in names: |
|
92 tags.append(str(name).decode("windows-1252").encode("utf8")) |
|
93 return tags |
|
94 |
|
95 def get_url(self, document_id): |
|
96 self.conn.query('SELECT dst FROM url_alias where src="node/'+document_id+'"') |
|
97 r = self.conn.use_result() |
|
98 row = r.fetch_row() |
|
99 |
|
100 if row: |
|
101 print 'row!!!!!!!!!!!!!!!!!!!!!!!!' |
|
102 print row |
|
103 print (row)[0][0] |
|
104 url = 'http://www.mediapart.fr/'+(row)[0][0] |
|
105 else: |
|
106 url = '' |
|
107 return url |
|
108 |
|
109 def get_author(self, document_id): |
|
110 self.conn.query('SELECT u.uid, u.name FROM users u, node n where u.uid=n.uid AND n.nid=\''+document_id+'\'') |
|
111 r = self.conn.use_result() |
|
112 row = r.fetch_row() |
|
113 if row: |
|
114 name = (row)[0][1].decode("windows-1252").encode("utf8") |
|
115 uid = (row)[0][0] |
|
116 else: |
|
117 name = '' |
|
118 uid='' |
|
119 return {'id':uid, 'name':name} |
|
120 |