Mercurial > public > sg101

comparison gpp/legacy/management/commands/import_old_news.py @ 291:a6357f2bcbbc

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Added a legacy command to import old news stories.
author Brian Neal <bgneal@gmail.com>
date Fri, 24 Dec 2010 20:45:33 +0000
parents
children 2367c4795c92
comparison
equal deleted inserted replaced
290:64c188a9d31f 291:a6357f2bcbbc
1 """
2 import_old_news.py - For importing news stories from SG101 1.0 as csv files.
3 """
4 from __future__ import with_statement
5 import csv
6 import optparse
7 import sys
8 from datetime import datetime
9
10 from django.core.management.base import LabelCommand, CommandError
11 from django.contrib.auth.models import User
12
13 from news.models import Category, Story
14 from legacy.phpbb import unescape
15
16 KNOWN_USERNAME_CHANGES = {
17 'cavefishbutchdelux': 'butchdelux',
18 'Findicator1': 'WaveOhhh',
19 }
20
21 class Command(LabelCommand):
22 args = '<filename filename ...>'
23 help = 'Imports news stories from the old database in CSV format'
24 option_list = LabelCommand.option_list + (
25 optparse.make_option("-p", "--progress", action="store_true",
26 help="Output a . after every 20 stories to show progress"),
27 )
28
29 def handle_label(self, filename, **options):
30 """
31 Process each line in the CSV file given by filename by
32 creating a new story.
33
34 """
35 self.show_progress = options.get('progress')
36 self.users = {}
37
38 # Create a mapping from the old database's topics to our
39 # Categories.
40 self.topics = {}
41 try:
42 self.topics[2] = Category.objects.get(slug='site-news')
43 self.topics[3] = Category.objects.get(slug='bands')
44 self.topics[4] = Category.objects.get(slug='show-announcements')
45 self.topics[5] = Category.objects.get(slug='show-reports')
46 self.topics[6] = Category.objects.get(slug='gear')
47 self.topics[7] = Category.objects.get(slug='reviews')
48 self.topics[8] = Category.objects.get(slug='surf-scene-news')
49 self.topics[9] = Category.objects.get(slug='articles')
50 self.topics[10] = Category.objects.get(slug='interviews')
51 self.topics[11] = Category.objects.get(slug='tablature')
52 self.topics[12] = Category.objects.get(slug='featured-videos')
53 except Category.DoesNotExist:
54 sys.exit("Category does not exist; check topic mapping.")
55
56 try:
57 with open(filename, "rb") as f:
58 self.reader = csv.DictReader(f)
59 num_rows = 0
60 try:
61 for row in self.reader:
62 self.process_row(row)
63 num_rows += 1
64 if self.show_progress and num_rows % 20 == 0:
65 sys.stdout.write('.')
66 sys.stdout.flush()
67 except csv.Error, e:
68 raise CommandError("CSV error: %s %s %s" % (
69 filename, self.reader.line_num, e))
70
71 print
72
73 except IOError:
74 raise CommandError("Could not open file: %s" % filename)
75
76 def process_row(self, row):
77 """
78 Process one row from the CSV file: create a Story object for
79 the row and save it in the database.
80
81 """
82 row = dict((k, v if v != 'NULL' else '') for k, v in row.iteritems())
83
84 try:
85 submitter = self._get_user(row['informant'])
86 except User.DoesNotExist:
87 print "Could not find user %s for story %s; skipping." % (
88 row['informant'], row['sid'])
89 return
90
91 story = Story(id=int(row['sid']),
92 title=unescape(row['title']),
93 submitter=submitter,
94 category=self.topics[int(row['topic'])],
95 short_text=row['hometext'],
96 long_text=row['bodytext'],
97 date_submitted=datetime.strptime(row['time'], "%Y-%m-%d %H:%M:%S"),
98 allow_comments=True)
99
100 story.save()
101
102 def _get_user(self, username):
103 """
104 Returns the user object with the given username.
105 Throws User.DoesNotExist if not found.
106
107 """
108 try:
109 return self.users[username]
110 except KeyError:
111 pass
112
113 try:
114 user = User.objects.get(username=username)
115 except User.DoesNotExist:
116 try:
117 user = User.objects.get(username=KNOWN_USERNAME_CHANGES[username])
118 except KeyError:
119 raise User.DoesNotExist
120
121 self.users[username] = user
122 return user