diff gpp/legacy/management/commands/import_old_users.py @ 290:64c188a9d31f

Adding a legacy app to contain management commands to convert the old data to the new database format. This first commit has the import_old_users command.
author Brian Neal <bgneal@gmail.com>
date Fri, 24 Dec 2010 05:28:58 +0000
parents
children a6357f2bcbbc
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/gpp/legacy/management/commands/import_old_users.py	Fri Dec 24 05:28:58 2010 +0000
@@ -0,0 +1,163 @@
+"""
+import_old_users.py - For importing users from SG101 1.0 as csv files.
+"""
+from __future__ import with_statement
+import csv
+import optparse
+import re
+import sys
+from datetime import datetime
+
+import postmarkup
+
+from django.core.management.base import LabelCommand, CommandError
+from django.contrib.auth.models import User
+
+import bio.models
+from legacy.phpbb import unphpbb
+from legacy.html2md import MarkdownWriter
+
+TIME_ZONES = {
+    '-5': 'US/Eastern',
+    '-6': 'US/Central',
+    '-7': 'US/Mountain',
+    '-8': 'US/Pacific',
+}
+USERNAME_RE = re.compile(r'^[\w.@+-]+$')
+USERNAME_LEN = (1, 30)      # min & max length values
+
+
+def _valid_username(username):
+    """
+    Return true if the username is valid.
+    """
+    return (USERNAME_LEN[0] <= len(username) <= USERNAME_LEN[1] and
+            USERNAME_RE.match(username))
+
+
+def _break_name(name):
+    """
+    Break name into a first and last name.
+    Return a 2-tuple of first_name, last_name.
+    """
+    parts = name.split()
+    n = len(parts)
+    if n == 0:
+        t = '', ''
+    elif n == 1:
+        t = parts[0], ''
+    else:
+        t = ' '.join(parts[:-1]), parts[-1]
+    return t[0][:USERNAME_LEN[1]], t[1][:USERNAME_LEN[1]]
+
+
+class Command(LabelCommand):
+    args = '<filename filename ...>'
+    help = 'Imports users from the old database in CSV format'
+    option_list = LabelCommand.option_list + (
+        optparse.make_option("-s", "--super-user",
+            help="Make the user with this name a superuser"),
+        optparse.make_option("-a", "--anon-user",
+            help="Make the user with this name the anonymous user "
+                "[default: Anonymous]"),
+        optparse.make_option("-p", "--progress", action="store_true",
+            help="Output a . after every 20 users to show progress"),
+    )
+    bb_parser = postmarkup.create(use_pygments=False, annotate_links=False)
+    md_writer = MarkdownWriter()
+
+    def handle_label(self, filename, **options):
+        """
+        Process each line in the CSV file given by filename by
+        creating a new user and profile.
+
+        """
+        self.superuser = options.get('super_user')
+        self.anonymous = options.get('anon_user')
+        if self.anonymous is None:
+            self.anonymous = 'Anonymous'
+        self.show_progress = options.get('progress')
+
+        if self.superuser == self.anonymous:
+            raise CommandError("super-user name should not match anon-user")
+
+        try:
+            with open(filename, "rb") as f:
+                self.reader = csv.DictReader(f)
+                num_rows = 0
+                try:
+                    for row in self.reader:
+                        self.process_row(row)
+                        num_rows += 1
+                        if num_rows % 20 == 0:
+                            sys.stdout.write('.')
+                            sys.stdout.flush()
+                except csv.Error, e:
+                    raise CommandError("CSV error: %s %s %s" % (
+                        filename, self.reader.line_num, e))
+
+                print
+
+        except IOError:
+            raise CommandError("Could not open file: %s" % filename)
+
+    def process_row(self, row):
+        """
+        Process one row from the CSV file: create a user and user profile for
+        the row and save it in the database.
+
+        """
+        row = dict((k, v if v != 'NULL' else '') for k, v in row.iteritems())
+
+        if not _valid_username(row['username']):
+            print "Skipping import of %s; invalid username" % row['username']
+            return
+
+        n = User.objects.filter(username=row['username']).count()
+        if n > 0:
+            print "Skipping import of %s; user already exists" % row['username']
+            return
+
+        first_name, last_name = _break_name(row['name'])
+        is_superuser = self.superuser == row['username']
+        is_anonymous = self.anonymous == row['username']
+
+        u = User(id=int(row['user_id']),
+                username=row['username'],
+                first_name=first_name,
+                last_name=last_name,
+                email=row['user_email'],
+                password=row['user_password'] if row['user_password'] else None,
+                is_staff=is_superuser,
+                is_active=True if not is_anonymous else False,
+                is_superuser=is_superuser,
+                last_login=datetime.fromtimestamp(int(row['user_lastvisit'])),
+                date_joined=datetime.strptime(row['user_regdate'], "%b %d, %Y"))
+
+        if is_anonymous:
+            u.set_unusable_password()
+
+        u.save()
+
+        p = u.get_profile()
+        p.location = row['user_from']
+        p.occupation = row['user_occ']
+        p.interests = row['user_interests']
+        p.profile_text = u''
+        p.hide_email = True if row['user_viewemail'] != '1' else False
+        p.signature = self.to_markdown(row['user_sig']) if row['user_sig'] else u''
+        p.time_zone = TIME_ZONES.get(row['user_timezone'], 'US/Pacific')
+        p.use_24_time = False
+        p.forum_post_count = int(row['user_posts'])
+        p.status = bio.models.STA_ACTIVE if p.forum_post_count > 10 else bio.models.STA_STRANGER
+        p.status_date = datetime.now()
+        p.update_date = p.status_date
+        p.save()
+
+    def to_html(self, s):
+        return self.bb_parser.render_to_html(unphpbb(s), cosmetic_replace=False)
+
+    def to_markdown(self, s):
+        self.md_writer.reset()
+        self.md_writer.feed(self.to_html(s))
+        return self.md_writer.markdown()