Changeset - b92d9a0321e0
[Not reviewed]
beta
0 1 0
Marcin Kuzminski - 15 years ago 2010-12-03 21:23:20
marcin@python-works.com
code stats speed improvments
1 file changed with 4 insertions and 9 deletions:
0 comments (0 inline, 0 general)
rhodecode/lib/celerylib/tasks.py
Show inline comments
 
@@ -11,250 +11,245 @@ from pylons.i18n.translation import _
 
from rhodecode.lib.celerylib import run_task, locked_task, str2bool
 
from rhodecode.lib.helpers import person
 
from rhodecode.lib.smtp_mailer import SmtpMailer
 
from rhodecode.lib.utils import OrderedDict, add_cache
 
from rhodecode.model import init_model
 
from rhodecode.model import meta
 
from rhodecode.model.db import RhodeCodeUi
 

	
 
from vcs.backends import get_repo
 

	
 
from sqlalchemy import engine_from_config
 

	
 
add_cache(config)
 

	
 
try:
 
    import json
 
except ImportError:
 
    #python 2.5 compatibility
 
    import simplejson as json
 

	
 
__all__ = ['whoosh_index', 'get_commits_stats',
 
           'reset_user_password', 'send_email']
 

	
 
CELERY_ON = str2bool(config['app_conf'].get('use_celery'))
 

	
 
def get_session():
 
    if CELERY_ON:
 
        engine = engine_from_config(config, 'sqlalchemy.db1.')
 
        init_model(engine)
 
    sa = meta.Session()
 
    return sa
 

	
 
def get_repos_path():
 
    sa = get_session()
 
    q = sa.query(RhodeCodeUi).filter(RhodeCodeUi.ui_key == '/').one()
 
    return q.ui_value
 

	
 
@task
 
@locked_task
 
def whoosh_index(repo_location, full_index):
 
    log = whoosh_index.get_logger()
 
    from rhodecode.lib.indexers.daemon import WhooshIndexingDaemon
 
    index_location = config['index_dir']
 
    WhooshIndexingDaemon(index_location=index_location,
 
                         repo_location=repo_location, sa=get_session())\
 
                         .run(full_index=full_index)
 

	
 
@task
 
@locked_task
 
def get_commits_stats(repo_name, ts_min_y, ts_max_y):
 
    from rhodecode.model.db import Statistics, Repository
 
    log = get_commits_stats.get_logger()
 

	
 
    #for js data compatibilty
 
    author_key_cleaner = lambda k: person(k).replace('"', "")
 

	
 
    commits_by_day_author_aggregate = {}
 
    commits_by_day_aggregate = {}
 
    repos_path = get_repos_path()
 
    p = os.path.join(repos_path, repo_name)
 
    repo = get_repo(p)
 

	
 
    skip_date_limit = True
 
    parse_limit = 250 #limit for single task changeset parsing optimal for
 
    last_rev = 0
 
    last_cs = None
 
    timegetter = itemgetter('time')
 

	
 
    sa = get_session()
 

	
 
    dbrepo = sa.query(Repository)\
 
        .filter(Repository.repo_name == repo_name).scalar()
 
    cur_stats = sa.query(Statistics)\
 
        .filter(Statistics.repository == dbrepo).scalar()
 
    if cur_stats:
 
        last_rev = cur_stats.stat_on_revision
 
    if not repo.revisions:
 
        return True
 

	
 
    if last_rev == repo.revisions[-1] and len(repo.revisions) > 1:
 
        #pass silently without any work if we're not on first revision or 
 
        #current state of parsing revision(from db marker) is the last revision
 
        return True
 

	
 
    if cur_stats:
 
        commits_by_day_aggregate = OrderedDict(
 
                                       json.loads(
 
                                        cur_stats.commit_activity_combined))
 
        commits_by_day_author_aggregate = json.loads(cur_stats.commit_activity)
 

	
 
    log.debug('starting parsing %s', parse_limit)
 
    lmktime = mktime
 

	
 
    last_rev = last_rev + 1 if last_rev > 0 else last_rev
 
    for rev in repo.revisions[last_rev:last_rev + parse_limit]:
 
        last_cs = cs = repo.get_changeset(rev)
 
        k = '%s-%s-%s' % (cs.date.timetuple()[0], cs.date.timetuple()[1],
 
                          cs.date.timetuple()[2])
 
        timetupple = [int(x) for x in k.split('-')]
 
        timetupple.extend([0 for _ in xrange(6)])
 
        k = lmktime(timetupple)
 
        k = lmktime([cs.date.timetuple()[0], cs.date.timetuple()[1],
 
                      cs.date.timetuple()[2], 0, 0, 0, 0, 0, 0])
 

	
 
        if commits_by_day_author_aggregate.has_key(author_key_cleaner(cs.author)):
 
            try:
 
                l = [timegetter(x) for x in commits_by_day_author_aggregate\
 
                        [author_key_cleaner(cs.author)]['data']]
 
                time_pos = l.index(k)
 
            except ValueError:
 
                time_pos = False
 

	
 
            if time_pos >= 0 and time_pos is not False:
 

	
 
                datadict = commits_by_day_author_aggregate\
 
                    [author_key_cleaner(cs.author)]['data'][time_pos]
 

	
 
                datadict["commits"] += 1
 
                datadict["added"] += len(cs.added)
 
                datadict["changed"] += len(cs.changed)
 
                datadict["removed"] += len(cs.removed)
 

	
 
            else:
 
                if k >= ts_min_y and k <= ts_max_y or skip_date_limit:
 

	
 
                    datadict = {"time":k,
 
                                "commits":1,
 
                                "added":len(cs.added),
 
                                "changed":len(cs.changed),
 
                                "removed":len(cs.removed),
 
                               }
 
                    commits_by_day_author_aggregate\
 
                        [author_key_cleaner(cs.author)]['data'].append(datadict)
 

	
 
        else:
 
            if k >= ts_min_y and k <= ts_max_y or skip_date_limit:
 
                commits_by_day_author_aggregate[author_key_cleaner(cs.author)] = {
 
                                    "label":author_key_cleaner(cs.author),
 
                                    "data":[{"time":k,
 
                                             "commits":1,
 
                                             "added":len(cs.added),
 
                                             "changed":len(cs.changed),
 
                                             "removed":len(cs.removed),
 
                                             }],
 
                                    "schema":["commits"],
 
                                    }
 

	
 
        #gather all data by day
 
        if commits_by_day_aggregate.has_key(k):
 
            commits_by_day_aggregate[k] += 1
 
        else:
 
            commits_by_day_aggregate[k] = 1
 

	
 
    overview_data = []
 
    for k, v in commits_by_day_aggregate.items():
 
        overview_data.append([k, v])
 
    overview_data = sorted(overview_data, key=itemgetter(0))
 
    overview_data = sorted(commits_by_day_aggregate.items(), key=itemgetter(0))
 
    if not commits_by_day_author_aggregate:
 
        commits_by_day_author_aggregate[author_key_cleaner(repo.contact)] = {
 
            "label":author_key_cleaner(repo.contact),
 
            "data":[0, 1],
 
            "schema":["commits"],
 
        }
 

	
 
    stats = cur_stats if cur_stats else Statistics()
 
    stats.commit_activity = json.dumps(commits_by_day_author_aggregate)
 
    stats.commit_activity_combined = json.dumps(overview_data)
 

	
 
    log.debug('last revison %s', last_rev)
 
    leftovers = len(repo.revisions[last_rev:])
 
    log.debug('revisions to parse %s', leftovers)
 

	
 
    if last_rev == 0 or leftovers < parse_limit:
 
        stats.languages = json.dumps(__get_codes_stats(repo_name))
 

	
 
    stats.repository = dbrepo
 
    stats.stat_on_revision = last_cs.revision
 

	
 
    try:
 
        sa.add(stats)
 
        sa.commit()
 
    except:
 
        log.error(traceback.format_exc())
 
        sa.rollback()
 
        return False
 
    if len(repo.revisions) > 1:
 
        run_task(get_commits_stats, repo_name, ts_min_y, ts_max_y)
 

	
 
    return True
 

	
 
@task
 
def reset_user_password(user_email):
 
    log = reset_user_password.get_logger()
 
    from rhodecode.lib import auth
 
    from rhodecode.model.db import User
 

	
 
    try:
 
        try:
 
            sa = get_session()
 
            user = sa.query(User).filter(User.email == user_email).scalar()
 
            new_passwd = auth.PasswordGenerator().gen_password(8,
 
                             auth.PasswordGenerator.ALPHABETS_BIG_SMALL)
 
            if user:
 
                user.password = auth.get_crypt_password(new_passwd)
 
                sa.add(user)
 
                sa.commit()
 
                log.info('change password for %s', user_email)
 
            if new_passwd is None:
 
                raise Exception('unable to generate new password')
 

	
 
        except:
 
            log.error(traceback.format_exc())
 
            sa.rollback()
 

	
 
        run_task(send_email, user_email,
 
                 "Your new rhodecode password",
 
                 'Your new rhodecode password:%s' % (new_passwd))
 
        log.info('send new password mail to %s', user_email)
 

	
 

	
 
    except:
 
        log.error('Failed to update user password')
 
        log.error(traceback.format_exc())
 

	
 
    return True
 

	
 
@task
 
def send_email(recipients, subject, body):
 
    """
 
    Sends an email with defined parameters from the .ini files.
 
    
 
    
 
    :param recipients: list of recipients, it this is empty the defined email
 
        address from field 'email_to' is used instead
 
    :param subject: subject of the mail
 
    :param body: body of the mail
 
    """
 
    log = send_email.get_logger()
 
    email_config = config
 

	
 
    if not recipients:
 
        recipients = [email_config.get('email_to')]
 

	
 
    mail_from = email_config.get('app_email_from')
 
    user = email_config.get('smtp_username')
 
    passwd = email_config.get('smtp_password')
 
    mail_server = email_config.get('smtp_server')
 
    mail_port = email_config.get('smtp_port')
 
    tls = str2bool(email_config.get('smtp_use_tls'))
 
    ssl = str2bool(email_config.get('smtp_use_ssl'))
 

	
 
    try:
 
        m = SmtpMailer(mail_from, user, passwd, mail_server,
0 comments (0 inline, 0 general)