~launchpad-pqm/launchpad/devel

#!/usr/bin/python -S

#

# Copyright 2009 Canonical Ltd.  This software is licensed under the

# GNU Affero General Public License version 3 (see the file LICENSE).

"""

Process LinkChecker .csv results for the staging server, stuff them into

a database and generate a report suitable for spamming developers with.

"""

__metaclass__ = type

# pylint: disable-msg=W0403

import _pythonpath

import csv, re, sys

from StringIO import StringIO

from optparse import OptionParser

from sqlobject import StringCol, IntCol, BoolCol, FloatCol, DatabaseIndex

from canonical.database.datetimecol import UtcDateTimeCol

from canonical.database.constants import UTC_NOW

from canonical.launchpad.scripts import db_options, logger_options, logger

from canonical.lp import initZopeless

from canonical.database.sqlbase import SQLBase

from canonical.config import config

from canonical.launchpad.mail import simple_sendmail

class CheckedLink(SQLBase):

    _table = 'CheckedLink'

    urlname = StringCol(notNull=True)

    recursionlevel = IntCol(notNull=True)

    parentname = StringCol(notNull=True)

    baseref = StringCol(notNull=True)

    result = StringCol(notNull=True)

    resultcode = IntCol(notNull=True)

    warningstring = StringCol(notNull=True)

    infostring = StringCol(notNull=True)

    valid = BoolCol(notNull=True)

    url = StringCol(notNull=True, unique=True, alternateID=True)

    line = IntCol(notNull=True)

    col = IntCol(notNull=True)

    name = StringCol(notNull=True)

    dltime = FloatCol()

    dlsize = IntCol()

    checktime = FloatCol(notNull=True)

    brokensince = UtcDateTimeCol(notNull=False, default=UTC_NOW)

    #cached = BoolCol(notNull=True)

    resultcode_index = DatabaseIndex('resultcode')

    recursionlevel_index = DatabaseIndex('recursionlevel')

def main(csvfile, log):

    # Where we store broken links

    broken = set()

    # Suck in the csv file, updating the database and adding to the broken set

    reader = csv.DictReader(

            (line.replace('\0','') for line in csvfile

                if not line.startswith('#'))

            )

    for row in reader:

        # Get the result code

        if row['valid']:

            row['resultcode'] = 200

            row['result'] = '200 Ok'

        else:

            m = re.search('^(\d+)', row['result'] or '')

            if m is None:

                if row['result'] == 'URL is empty':

                    continue

                elif 'The read operation timed out' in row['result']:

                    row['result'] = '601 %s' % row['result']

                    row['resultcode'] = 601

                else:

                    row['result'] = '602 %s' % row['result']

                    row['resultcode'] = 602

            else:

                row['resultcode'] = int(m.group(1))

        # Cast input and nuke crap (to avoid confusing SQLObject)

        row['recursionlevel'] = int(row['recursionlevel'])

        row['valid'] = row['valid'] in ('True', 'true')

        row['line'] = int(row['line'])

        row['col'] = int(row['column']) # Renamed - column is a SQL keyword

        del row['column']

        row['dltime'] = float(row['dltime'])

        row['dlsize'] = int(row['dlsize'])

        row['checktime'] = float(row['checktime'])

        del row['cached']

        if row['resultcode'] < 400:

            row['brokensince'] = None

        try:

            link = CheckedLink.byUrl(row['url'])

            link.set(**row)

        except LookupError:

            link = CheckedLink(**row)

        broken.add(link)

    total = len(broken)

    # Delete any entries that were not spidered

    # XXX StuartBishop 2005-07-04: Only if older than a threshold.

    for link in CheckedLink.select():

        if link in broken:

            continue

        link.destroySelf()

    new_broken_links = CheckedLink.select("""

        resultcode in (404, 500, 601)

        AND brokensince > CURRENT_TIMESTAMP AT TIME ZONE 'UTC'

            - '1 day 12 hours'::interval

        """, orderBy=["recursionlevel", "parentname", "url"])

    rep = report("New Arrivals", new_broken_links, total, brokensince=False)

    old_broken_links = CheckedLink.select("""

        resultcode in (404, 500, 601)

        AND brokensince <= CURRENT_TIMESTAMP AT TIME ZONE 'UTC'

            - '1 day 12 hours'::interval

        AND brokensince >

            CURRENT_TIMESTAMP AT TIME ZONE 'UTC' - '14 days'::interval

        """, orderBy=["recursionlevel", "parentname", "url"])

    rep += report("Old Favorites", old_broken_links, total, brokensince=True)

    antique_broken_links = CheckedLink.select("""

        resultcode in (404, 500, 601)

        AND brokensince <=

            CURRENT_TIMESTAMP AT TIME ZONE 'UTC' - '14 days'::interval

        """, orderBy=["brokensince", "recursionlevel", "parentname", "url"])

    rep += report(

            "Hall of Shame", antique_broken_links, total, brokensince=True

            )

    if not options.email:

        # Print to stdout in system encoding - might raise UnicodeError on

        # some systems. Tough.

        print rep

    else:

        # Override this setting - we are only here if email explicitly

        # requested on the command line.

        send_email_data = """

            [zopeless]

            send_email: True

            """

        config.push('send_email_data', send_email_data)

        simple_sendmail(

                "noreply@canonical.com", [options.email], options.subject,

                rep, {'Keywords': 'LinkChecker', 'X-Fnord': 'Fnord'}

                )

        config.pop('send_email_data')

def report(title, links, total, brokensince=True):

    out = StringIO()

    heading = "%s (%d/%d)" % (title, links.count(), total)

    print >> out, heading

    print >> out, "=" * len(heading)

    def print_row(title, value):

        print >> out, "%-7s: %s" % (title, str(value))

    for link in links:

        print_row("Link", link.url)

        print_row("Parent", link.parentname)

        print_row("Result", link.result)

        if link.warningstring:

            print_row("Warning", link.warningstring)

        if brokensince:

            print_row("Since", link.since.strftime('%A %d %B %Y'))

        print >> out

    print >> out

    return out.getvalue()

if __name__ == '__main__':

    parser = OptionParser("Usage: %prog [OPTIONS] [input.csv]")

    db_options(parser)

    logger_options(parser)

    parser.add_option(

            "-c", "--create", action="store_true", dest="create",

            default=False, help="Create the database tables"

            )

    parser.add_option(

            "-s", "--subject", dest="subject", help="Email using SUBJECT",

            metavar="SUBJECT", default="LinkChecker report"

            )

    parser.add_option(

            "-t", "--to", dest="email", help="Email to ADDRESS",

            metavar="ADDRESS", default=None

            )

    options, args = parser.parse_args()

    log = logger(options)

    if len(args) == 0 or args[0] == '-':

        log.debug("Reading from stdin")

        csvfile = sys.stdin

    else:

        csvfile = open(args[0], 'rb')

    ztm = initZopeless()

    if options.create:

        # Create the table if it doesn't exist. Unfortunately, this is broken

        # so we only create the table if requested on the command line

        CheckedLink.createTable(ifNotExists=True)

    main(csvfile, log)

    ztm.commit()