#!/usr/bin/python
# -*- coding: utf-8 -*-
#
# Copyright 2005 Lars Wirzenius (liw@iki.fi)
# Copyright 2009-2018 Holger Levsen (holger@layer-acht.org)
# Copyright © 2011-2017 Andreas Beckmann (anbe@debian.org)
# Copyright 2013 David Steele (dsteele@gmail.com)
#
# This program is free software; you can redistribute it and/or modify it
# under the terms of the GNU General Public License as published by the
# Free Software Foundation; either version 2 of the License, or (at your
# option) any later version.
#
# This program is distributed in the hope that it will be useful, but
# WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
# Public License for more details.
#
# You should have received a copy of the GNU General Public License along with
# this program. If not, see <https://www.gnu.org/licenses/>


"""Create HTML reports of piuparts log files

Lars Wirzenius <liw@iki.fi>
"""


import os
import sys
import time
import logging
import shutil
import re
import string
import yaml
import hashlib
import pickle
import random
import fcntl
from urllib2 import HTTPError, URLError
from collections import deque

# if python-rpy2 ain't installed, we don't draw fancy graphs
try:
    from rpy2 import robjects
    from rpy2.robjects.packages import importr
except:
    pass

import piupartslib
from piupartslib.conf import MissingSection
from piupartslib.dwke import *
import piupartslib.pkgsummary as pkgsummary


CONFIG_FILE = "/etc/piuparts/piuparts.conf"
DISTRO_CONFIG_FILE = "/etc/piuparts/distros.conf"
KPR_DIRS = ('pass', 'bugged', 'affected', 'fail')
TPL_EXT = '.tpl'


PIUPARTS_VERSION = "0.84"

HTML_HEADER = """<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<!-- $content_md5 -->
 <html><head><meta http-equiv="Content-Type" content="text/html; charset=utf-8">
  <!-- Generated by piuparts-report $piuparts_version -->
  <title>
   $page_title
  </title>
  <link type="text/css" rel="stylesheet" href="$doc_root/style.css">
  <link rel="shortcut icon" href="$doc_root/favicon.ico">
 </head>

 <body>
 <div id="header">
   <h1 class="header">
    <a href="https://www.debian.org/">
     <img src="$doc_root/images/openlogo-nd-50.png" border="0" hspace="0" vspace="0" alt=""></a>
    <a href="https://www.debian.org/">
     <img src="$doc_root/images/debian.png" border="0" hspace="0" vspace="0" alt="Debian Project"></a>
    Quality Assurance
   </h1>
   <div id="obeytoyourfriend">Policy is your friend. Trust the Policy. Love the Policy. Obey the Policy.</div>
 </div>
 <hr>
<div id="main">
<table class="containertable">
 <tr class="containerrow" valign="top">
  <td class="containercell">
   <table class="lefttable">
    <tr class="titlerow">
     <td class="titlecell">
      General information
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="$doc_root/">About + News</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="https://wiki.debian.org/piuparts" target="_blank">Overview</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="https://wiki.debian.org/piuparts/FAQ" target="_blank">FAQ</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="https://anonscm.debian.org/cgit/piuparts/piuparts.git" target="_blank">Source</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      piuparts.d.o <a href="https://bugs.debian.org/src:piuparts.debian.org" target="_blank">bugs</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      piuparts <a href="https://bugs.debian.org/src:piuparts" target="_blank">bugs</a> / <a href="https://anonscm.debian.org/cgit/piuparts/piuparts.git/tree/TODO?h=develop" target="_blank">ToDo</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="mailto:piuparts-devel@lists.alioth.debian.org" target="_blank">Contact us</a>
     </td>
    </tr>
    <tr class="titlerow">
     <td class="titlecell">
      Documentation
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="$doc_root/doc/piuparts.1.html" target="_blank">piuparts manpage</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="$doc_root/doc/README.html" target="_blank">README</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="$doc_root/doc/README_server.html" target="_blank">README_server</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      piuparts.d.o configuration:<br>
      <a href="https://anonscm.debian.org/cgit/piuparts/piuparts.git/tree/instances/piuparts.conf-template.pejacevic?h=develop" target="_blank">piuparts.conf</a>,<br>
      <a href="https://anonscm.debian.org/cgit/piuparts/piuparts.git/tree/conf/distros.conf?h=develop" target="_blank">distros.conf</a><br> and
      <a href="https://anonscm.debian.org/cgit/piuparts/piuparts.git/tree/custom-scripts?h=develop" target="_blank">scripts</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="$doc_root/bug_howto.html">How to file bugs</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="https://www.debian.org/doc/debian-policy/" target="_blank">Debian policy</a>
     </td>
    </tr>
    <tr class="titlerow">
     <td class="alerttitlecell">
      Available reports
     </td>
    </tr>
    <tr>
     <td class="contentcell">
      <a href="https://bugs.debian.org/cgi-bin/pkgreport.cgi?tag=piuparts;users=debian-qa@lists.debian.org&amp;archive=both" target="_blank">Bugs filed</a>
     </td>
    </tr>
    $section_navigation
    <tr class="titlerow">
     <td class="titlecell">
      Other Debian QA efforts
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="https://wiki.debian.org/qa.debian.org" target="_blank">Debian QA Group</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="https://qa.debian.org/dose/" target="_blank">Dose tools (former: EDOS)</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="https://lintian.debian.org" target="_blank">Lintian</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="https://tracker.debian.org" target="_blank">Debian Package Tracker (former: PTS)</a>
     </td>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="https://udd.debian.org" target="_blank">Ultimate Debian Database</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="https://jenkins.debian.net" target="_blank">jenkins.debian.net</a>
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell">
      <a href="http://ci.debian.net" target="_blank">ci.debian.net</a>
     </td>
    </tr>
    <tr class="titlerow">
     <td class="titlecell">
      Last update
     </td>
    </tr>
    <tr class="normalrow">
     <td class="lastcell">
      $time
     </td>
    </tr>
   </table>
  </td>
  <td class="containercell">
"""


HTML_FOOTER = """
  </td>
 </tr>
</table>
</div>
 <hr>
 <div id="footer">
  <div>
   <a href="https://tracker.debian.org/pkg/piuparts" target="_blank">piuparts</a>
   is GPL2 <a href="https://packages.debian.org/changelogs/pool/main/p/piuparts/current/copyright" target="_blank">licenced</a>
   and was originally written by <a href="mailto:liw@iki.fi">Lars Wirzenius</a> and is now maintained by
   <a href="mailto:anbe@debian.org">Andreas Beckmann</a> and
   <a href="mailto:holger@layer-acht.org">Holger Levsen</a> and

   <a href="mailto:piuparts-devel@lists.alioth.debian.org">others</a> using
   <a href="https://anonscm.debian.org/cgit/piuparts/piuparts.git/" target="_blank">piuparts.git</a>.
   Ditto for this website.
   Weather icons are from the <a href="http://tango.freedesktop.org/Tango_Icon_Library" target="_blank">Tango Icon Library</a>.
   <a href="http://validator.w3.org/check?uri=referer">
    <img border="0" src="$doc_root/images/valid-html401.png" alt="Valid HTML 4.01!" height="15" width="80" align="middle">
   </a>
   <a href="http://jigsaw.w3.org/css-validator/check/referer">
    <img border="0" src="$doc_root/images/w3c-valid-css.png" alt="Valid CSS!"  height="15" width="80" align="middle">
   </a>
  </div>
 </div>
</body>
</html>
"""


LOG_LIST_BODY_TEMPLATE = """
   <table class="righttable">
    <tr class="titlerow">
     <td class="$title_style" colspan="2">
      $title in $section
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell2" colspan="2">
      $preface
      The list has $count packages, with $versioncount total versions.
     </td>
    </tr>
    $logrows
   </table>
"""


STATE_BODY_TEMPLATE = """
   <table class="righttable">
    <tr class="titlerow">
     <td class="alerttitlecell">
      Packages in state "$state" in $section $aside
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell2">
      <ul>
       $list
      </ul>
     </td>
    </tr>
   </table>
"""


SECTION_INDEX_BODY_TEMPLATE = """
   <table class="righttable">
    <tr class="titlerow">
     <td class="titlecell" colspan="3">
      $section statistics
     </td>
    </tr>
    <tr class="normalrow">
     <td class="contentcell2" colspan="3">
      $description
     </td>
    </tr>
    <tr class="titlerow">
     <td class="alerttitlecell" colspan="3">
      Binary packages per state
     </td>
    </tr>
    $tablerows
    <tr class="titlerow">
     <td class="titlecell" colspan="3">
      URL to Packages file
     </td>
    </tr>
     <tr class="normalrow">
     <td class="contentcell2" colspan="3">
      <code>$packagesurl</code>
     </td>
    </tr>
   </table>
"""

MAINTAINER_BODY_TEMPLATE = """
   <table class="righttable">
    <tr class="titlerow">
     <td class="titlecell" colspan="6">
      $maintainer
     </td>
    </tr>
    $distrolinks
    $rows
   </table>
"""


SOURCE_PACKAGE_BODY_TEMPLATE = """
   <table class="righttable">
    $rows
   </table>
"""

ANALYSIS_BODY_TEMPLATE = """
   <table class="righttable">
    $rows
   </table>
"""

PROB_TPL = \
    """<tr class="titlerow"><td class="titlecell">
$HEADER in $SECTION, sorted by reverse dependency count.
</td></tr><tr class="normalrow"><td class="contentcell2">
$HELPTEXT
<p>The commandline to find these logs is: <pre>
COMMAND='$COMMAND'
</pre></p>
</td></tr><tr class="titlerow"><td class="alerttitlecell">Please file bugs!</td></tr><tr class="normalrow"><td class="contentcell2" colspan="3">
<ul>
$PACKAGE_LIST</ul>
<p>Affected packages in $SECTION: $COUNT</p></td></tr>
"""

PKG_ERROR_TPL = \
    """<li>$RDEPS - <a href=\"$LOG\">$LOG</a>$ARCH
    (<a href=\"https://tracker.debian.org/pkg/$SPKG\" target=\"_blank\">PTS</a>)
    (<a href=\"https://bugs.debian.org/$PACKAGE?dist=unstable\" target=\"_blank\">BTS</a>)
$BUG</li>
"""


title_by_dir = {
    "pass": "PASSED piuparts logs",
    "fail": "Failed UNREPORTED piuparts logs",
    "bugged": "Failed REPORTED piuparts logs",
    "affected": "Failed AFFECTED piuparts logs",
    "reserved": "RESERVED packages",
    "untestable": "UNTESTABLE packages",
}


desc_by_dir = {
    "pass": "Log files for packages that have PASSED testing.",
    "fail": "Log files for packages that have FAILED testing. " +
            "Bugs have not yet been reported.",
    "bugged": "Log files for packages that have FAILED testing. " +
              "Bugs have been reported, but not yet fixed.",
    "affected": "Log files for packages that have dependencies FAILED testing. " +
              "Bugs have been reported, but not yet fixed.",
    "reserved": "Packages that are RESERVED for testing on a node in a " +
                "distributed piuparts network.",
    "untestable": "Log files for packages that have are UNTESTABLE with " +
                  "piuparts at the current time.",
}

state_by_dir = {
    "pass": "successfully-tested",
    "fail": "failed-testing",
    "bugged": "failed-testing",
    "affected": "failed-testing",
    "reserved": "waiting-to-be-tested",
    "untestable": "cannot-be-tested",
}

# better use XX_name.tpl and get the linktarget from the template
# (its a substring of the <title> of the that template
# maintaining this list is errorprone and tiresome
linktarget_by_template = [
    ("initdscript_lsb_header_issue.tpl", "but logfile contains update-rc.d issues"),
    ("command_not_found_issue.tpl", "but logfile contains 'command not found'"),
    ("debsums_mismatch_issue.tpl", "but logfile contains modified conffiles or other shipped files"),
    ("alternatives_after_purge_issue.tpl", "but logfile contains forgotten alternatives"),
    ("owned_files_after_purge_issue.tpl", "but logfile contains owned files existing after purge"),
    ("unowned_files_after_purge_issue.tpl", "but logfile contains unowned files after purge"),
    ("maintainer_script_issue.tpl", "but logfile contains maintainer script failures"),
    ("db_setup_issue.tpl", "but logfile contains failure to setup a database"),
    ("problems_and_no_force_issue.tpl", "but logfile reports that not enough force was used"),
    ("installs_over_symlink_issue.tpl", "but package installs something over existing symlinks"),
    ("broken_symlinks_issue.tpl", "but logfile contains 'broken symlinks'"),
    ("unknown_inadequate_issue.tpl", "but logfile contains unknown inadequate issues"),
    ("boring_obsolete_conffile_file_inadequate_issue.tpl", "...and logfile contains tag from adequate 'obsolete-conffile-file'"),
    ("boring_broken_symlink_file_inadequate_issue.tpl", "...and logfile contains tag from adequate 'broken-symlink-file'"),
    ("bin_or_sbin_binary_requires_usr_lib_library_inadequate_issue.tpl", "but adequate found a binary in /bin or /sbin that requires a /usr/lib library"),
    ("incompatible_licenses_inadequate_issue.tpl", "but adequate found a license incompatibility"),
    ("broken_binfmt_detector_inadequate_issue.tpl", "but adequate did not find the detector registered with update_binfmts"),
    ("broken_binfmt_interpreter_inadequate_issue.tpl", "but adequate did not find the interpreter registered with update_binfmts"),
    ("missing_alternative_inadequate_issue.tpl", "but adequate found a missing alternative"),
    ("missing_copyright_file_inadequate_issue.tpl", "but adequate couldn't find a copyright file"),
    ("missing_pkgconfig-dependency_issue.tpl", "but adequate found a missing pkgconfig dependency"),
    ("program_name_collision_inadequate_issue.tpl", "but adequate found a program name collision"),
    ("py_file_not_bytecompiled_inadequate_issue.tpl", "but adequate found a .py file that is not byte-compiled"),
    ("pyshared_file_not_bytecompiled_inadequate_issue.tpl", "but adequate found a .py file in /usr/share/pyshared that is not byte-compiled"),
    ("ldd_inadequate_issue.tpl", "but adequate failed to run ldd on a binary or library"),
    ("library_not_found_inadequate_issue.tpl", "but adequate couldn't find a required library"),
    ("undefined_symbol_inadequate_issue.tpl", "but adequate found an undefined symbol"),
    ("symbol-size-mismatch_inadequate_issue.tpl", "but adequate found that a symbol has changed size since the package was built"),
    ("missing-symbol-version-information_inadequate_issue.tpl", "but adequate found that a library is missing symbol version information"),
    ("unknown_inadequate_issue.tpl", "but an unknown adequate issue was found"),
    ("inadequate_exit_issue.tpl", "but adequate exited inadequately"),
    ("packages_have_been_kept_back_issue.tpl", "but logfile contains 'packages have been kept back'"),
    ("needs_rebuild_issue.tpl", "but logfile recommends to rebuild some packages"),
    ("module_build_error_issue.tpl", "but logfile contains dkms module build failures"),
    ("obsolete_conffiles_issue.tpl", "but logfile reports obsolete conffiles"),
    ("missing_md5sums_issue.tpl", "but logfile reports missing md5sums"),
    ("unowned_lib_symlink_issue.tpl", "but logfile reports unowned lib symlinks"),
    ("piuparts-depends-dummy_issue.tpl", "but logfile reports piuparts-depends-dummy.deb could not be installed"),
    ("used_exception_issue.tpl", "but package used a piuparts exception"),
    ("test_was_skipped_issue.tpl", "but package test was skipped"),

    ("dependency_error.tpl", "due to unsatisfied dependencies"),
    ("packages_have_been_kept_back_error.tpl", "...and logfile also contains 'packages have been kept back'"),
    ("command_not_found_error.tpl", "due to a 'command not found' error"),
    ("files_in_usr_local_error.tpl", "due to files in /usr/local"),
    ("overwrite_other_packages_files_error.tpl", "due to overwriting other packages files"),
    ("debsums_mismatch_error.tpl", "due to modifying conffiles or other shipped files"),
    ("alternatives_after_purge_error.tpl", "due to forgotten alternatives after purge"),
    ("owned_files_by_many_packages_error.tpl", "due to owned files by many packages"),
    ("owned_files_after_purge_error.tpl", "due to owned files existing after purge"),
    ("unowned_files_after_purge_error.tpl", "due to unowned files after purge"),
    ("modified_files_after_purge_error.tpl", "due to files having been modified after purge"),
    ("disappeared_files_after_purge_error.tpl", "due to files having disappeared after purge"),
    ("diversion_error.tpl", "due to diversions being modified after purge"),
    ("processes_running_error.tpl", "due to leaving processes running behind"),
    ("resource_violation_error.tpl", "due to resource violation"),
    ("conffile_prompt_error.tpl", "due to prompting due to modified conffiles"),
    ("db_setup_error.tpl", "due to failing to setup a database"),
    ("insserv_error.tpl", "due to a problem with insserv"),
    ("problems_and_no_force_error.tpl", "due to not enough force being used"),
    ("trigger_cycle_error.tpl", "due to dpkg encountered trigger problems"),
    ("immediate_configuration_error.tpl", "due to apt could not perform immediate configuration"),
    ("pre_depends_error.tpl", "due to a problem with pre-depends"),
    ("pre_installation_script_error.tpl", "due to pre-installation maintainer script failed"),
    ("post_installation_script_error.tpl", "due to post-installation maintainer script failed"),
    ("pre_removal_script_error.tpl", "due to pre-removal maintainer script failed"),
    ("post_removal_script_error.tpl", "due to post-removal maintainer script failed"),
    ("unknown_purge_error.tpl", "due to purge failed due to an unknown reason"),
    ("cron_error_after_removal_error.tpl", "due to errors from cronjob after removal"),
    ("logrotate_error_after_removal_error.tpl", "due to errors from logrotate after removal"),
    ("installs_over_symlink_error.tpl", "...and package installs something over existing symlinks"),
    ("broken_symlinks_error.tpl", "...and logfile also contains 'broken symlinks'"),
    ("obsolete_conffiles_error.tpl", "...and logfile reports obsolete conffiles"),
    ("missing_md5sums_error.tpl", "...and logfile reports missing md5sums"),
    ("unowned_lib_symlink_error.tpl", "...and logfile reports unowned lib symlinks"),
    ("piuparts-depends-dummy_error.tpl", "...and logfile reports piuparts-depends-dummy.deb could not be installed"),
    ("unclassified_failures.tpl", "due to unclassified failures"),
]


class Config(piupartslib.conf.Config):

    def __init__(self, section="report", defaults_section=None):
        self.section = section
        piupartslib.conf.Config.__init__(self, section,
            {
                "sections": "report",
                "output-directory": "html",
                "master-directory": ".",
                "depends-sections": None,
                "description": "",
                "proxy": None,
                "mirror": None,
                "distro": None,
                "area": None,
                "arch": None,
                "upgrade-test-distros": None,
                "max-reserved": 1,
                "doc-root": "/",
                "known-problem-directory": "/usr/share/piuparts/known_problems",
                "json-sections": "default",
                "precedence": 1,
                "web-host": "piuparts.debian.org",
            },
            defaults_section=defaults_section)


def setup_logging(log_level, log_file_name):
    logger = logging.getLogger()
    logger.setLevel(log_level)

    formatter = logging.Formatter(fmt="%(asctime)s %(message)s",
                                  datefmt="%H:%M:%S")

    handler = logging.StreamHandler(sys.stderr)
    handler.setFormatter(formatter)
    logger.addHandler(handler)

    if log_file_name:
        handler = logging.FileHandler(log_file_name)
        logger.addHandler(handler)


def html_protect(vstr):
    vstr = "&amp;".join(vstr.split("&"))
    vstr = "&lt;".join(vstr.split("<"))
    vstr = "&gt;".join(vstr.split(">"))
    vstr = "&#34;".join(vstr.split('"'))
    vstr = "&#39;".join(vstr.split("'"))
    return vstr


def is_bad_state(state):
    bad_states = [
        #"successfully-tested",
        "failed-testing",
        "cannot-be-tested",
        # "essential-required",  # obsolete
        #"waiting-to-be-tested",
        #"waiting-for-dependency-to-be-tested",
        "dependency-failed-testing",
        "dependency-cannot-be-tested",
        "dependency-does-not-exist",
        "circular-dependency",  # obsolete
        "unknown",
        "unknown-preferred-alternative",  # obsolete
        "no-dependency-from-alternatives-exists",  # obsolete
        "outdated",
        #"foreign:*",
        "does-not-exist",
    ]

    return(state in bad_states)


def emphasize_reason(reason):
    if is_bad_state(reason):
        reason = "<em>" + reason + "</em>"
    return reason


def source_subdir(source):
    if source[:3] == "lib":
        return source[:4]
    else:
        return source[:1]


def source_summary_url(web_host, doc_root, section, src_pkg):
    return("https://%s%s/%s/source/%s/%s.html" %
              (
                  web_host,
                  doc_root,
                  section,
                  source_subdir(src_pkg),
                  src_pkg,
              )
           )


def maintainer_subdir(maintainer):
    return maintainer.lower()[:1]


def find_files_with_suffix(vdir, suffix):
    pairs = []  # (mtime, name)
    for name in os.listdir(vdir):
        if name.endswith(suffix):
            try:
                if os.path.isfile(os.path.join(vdir, name)):
                    mtime = os.path.getmtime(os.path.join(vdir, name))
                    pairs.append((mtime, name))
            except OSError:
                pass
    # sort by mtime
    return [x[1] for x in sorted(pairs)]


def update_file(source, target):
    if os.path.exists(target):
        try:
            aa = os.stat(source)
            bb = os.stat(target)
        except OSError:
            pass
        else:
            if aa.st_size == bb.st_size and aa.st_mtime <= bb.st_mtime:
                return
        try:
            os.remove(target)
        except:
            pass
    try:
        os.link(source, target)
    except OSError:
        try:
            shutil.copyfile(source, target)
        except IOError as xxx_todo_changeme:
            (errno, strerror) = xxx_todo_changeme.args
            logging.error("failed to copy %s to %s: I/O error(%d): %s"
                           % (source, target, errno, strerror))


def copy_logs(logs_by_dir, output_dir):
    for vdir in logs_by_dir:
        fulldir = os.path.join(output_dir, vdir)
        if not os.path.exists(fulldir):
            os.makedirs(fulldir)
        for basename in logs_by_dir[vdir]:
            source = os.path.join(vdir, basename)
            target = os.path.join(fulldir, basename)
            update_file(source, target)


def remove_old_logs(logs_by_dir, output_dir):
    for vdir in logs_by_dir:
        fulldir = os.path.join(output_dir, vdir)

        # convert logs_by_dir array to a dict to avoid linear search
        logs_dict = {}
        for log in logs_by_dir[vdir]:
            logs_dict[log] = 1

        if os.path.exists(fulldir):
            for basename in os.listdir(fulldir):
                if not basename in logs_dict:
                    os.remove(os.path.join(fulldir, basename))


def create_file(filename, contents):
    with open(filename, "w") as f:
        f.write(contents)


def append_file(filename, contents):
    with open(filename, "a") as f:
        f.write(contents)


def read_file(filename):
    with open(filename, "r") as f:
        return f.read()


def readlines_file(filename):
    with open(filename, "r") as f:
        return f.readlines()


def fileage(filename):
    mtime = os.path.getmtime(filename)
    now = time.time()
    return now - mtime


def write_template_html(filename, body, mapping={}, defer_if_unmodified=False, md5cache=None):
    header = HTML_HEADER
    footer = HTML_FOOTER
    htmlpage = string.Template(header + body + footer)
    mapping = mapping.copy()
    mapping.update({
        "content_md5": "",
        "piuparts_version": "",
        "time": "",
    })
    content_md5 = hashlib.md5(htmlpage.safe_substitute(mapping)).hexdigest()

    if md5cache is not None:
        md5cache['new'][filename] = content_md5

    if defer_if_unmodified:
        if filename in md5cache['old'] and md5cache['old'][filename] == content_md5 and os.path.exists(filename):
            md5cache['unmodified'] += 1
            # defer updating the file between 1 and 4 weeks
            minage = 1 * 7 * 86400
            maxage = 4 * 7 * 86400
            if fileage(filename) < random.randint(minage, maxage):
                return
            md5cache['refreshed'] += 1

    if defer_if_unmodified and not filename in md5cache['old'] and os.path.exists(filename):
        # read the first 10 lines of the old file and look for a matching md5sum
        with open(filename, "r") as f:
            lineno = 0
            for x in range(10):
                line = f.readline()
                if not line:
                    break
                if line.startswith("<!-- ") and line[5:5 + len(content_md5)] == content_md5:
                    md5cache['unmodified'] += 1
                    # defer updating the file between 1 and 4 weeks
                    minage = 1 * 7 * 86400
                    maxage = 4 * 7 * 86400
                    if fileage(filename) < random.randint(minage, maxage):
                        return
                    md5cache['refreshed'] += 1
                    break

    mapping.update({
        "content_md5": content_md5,
        "piuparts_version": PIUPARTS_VERSION,
        "time": time.strftime("%Y-%m-%d %H:%M %Z"),
    })
    create_file(filename, htmlpage.safe_substitute(mapping))
    if md5cache is not None:
        md5cache['written'] += 1


def create_section_navigation(section_names, current_section, doc_root):
    tablerows = ""
    for section in section_names:
        tablerows += ("<tr class=\"normalrow\"><td class=\"contentcell\"><a href='%s/%s'>%s</a></td></tr>\n") % \
            (doc_root, html_protect(section), html_protect(section))
    tablerows += "<tr><td class=\"contentcell\"><a href=\"%s/%s/maintainer/\">by maintainer / uploader</a></td></tr>\n" \
                 % (doc_root, current_section)
    tablerows += "<tr><td class=\"contentcell\"><a href=\"%s/%s/source/\">by source package</a></td></tr>\n" \
                 % (doc_root, current_section)
    return tablerows;


def get_email_address(maintainer):
    email = "INVALID maintainer address: %s" % (maintainer)
    try:
        m = re.match(r"(.+)(<)(.+@.+)(>)", maintainer)
        email = m.group(3)
    except:
        pass
    return email


def package2id(package_name):
    # "+" is not a valid identifier char for id=... attributes
    return package_name.replace("+", "_")


# return order preserving list of the first occurrence of an element
def unique(stuff):
    # can't use set() because 'stuff' is a list of lists and list() is not hashable
    vlist = []
    previtem = stuff
    for item in stuff:
        if item != previtem:
            vlist.append(item)
            previtem = item
    return vlist


class Busy(Exception):

    def __init__(self):
        self.args = "section is locked by another process",


class Section:

    def __init__(self, section, master_directory, doc_root, packagedb_cache={}):
        self._config = Config(section=section, defaults_section="global")
        self._config.read(CONFIG_FILE)
        self._distro_config = piupartslib.conf.DistroConfig(
            DISTRO_CONFIG_FILE, self._config["mirror"])
        logging.debug("-------------------------------------------")
        logging.debug("Running section " + self._config.section)

        self._section_directory = os.path.abspath(os.path.join(master_directory,
                                                               self._config.section))
        if not os.path.exists(self._section_directory):
            logging.debug("Warning: %s did not exist, now created. Did you ever let the slave work?"
                           % self._section_directory)
            os.makedirs(self._section_directory)

        self._doc_root = doc_root

        logging.debug("Loading and parsing Packages file")
        self._packagedb_cache = packagedb_cache
        self._package_databases = {}
        self._load_package_database(section, master_directory)
        self._binary_db = self._package_databases[section]

        self._source_db = piupartslib.packagesdb.PackagesDB(prefix=self._section_directory)
        self._source_db.load_packages_urls(
            self._distro_config.get_sources_urls(
                self._config.get_distro(),
                    self._config.get_area()))
        if self._config.get_distro() != self._config.get_final_distro():
            # take version numbers (or None) from final distro
            self._source_db.load_alternate_versions_from_packages_urls(
                self._distro_config.get_sources_urls(
                    self._config.get_final_distro(),
                    self._config.get_area()))

        self._log_name_cache = {}
        self._md5cache = { 'old': {}, 'new': {}, 'written': 0, 'unmodified': 0, 'refreshed': 0 }

    def _load_package_database(self, section, master_directory):
        if section in self._package_databases:
            return
        elif section in self._packagedb_cache:
            self._package_databases[section] = self._packagedb_cache[section]
            return

        config = Config(section=section, defaults_section="global")
        config.read(CONFIG_FILE)
        if not config["depends-sections"]:
            # this is a base database eligible for caching
            # only cache the most recent base database
            self._packagedb_cache.clear()
        sectiondir = os.path.join(master_directory, section)
        db = piupartslib.packagesdb.PackagesDB(prefix=sectiondir)
        self._package_databases[section] = db
        if config["depends-sections"]:
            deps = config["depends-sections"].split()
            for dep in deps:
                self._load_package_database(dep, master_directory)
            db.set_dependency_databases([self._package_databases[dep] for dep in deps])
        else:
            # only cache the big base databases that don't have additional dependencies
            self._packagedb_cache[section] = db
        db.load_packages_urls(
            self._distro_config.get_packages_urls(
                config.get_distro(),
                    config.get_area(),
                    config.get_arch()))
        if config.get_distro() != config.get_final_distro():
            # take version numbers (or None) from final distro
            db.load_alternate_versions_from_packages_urls(
                self._distro_config.get_packages_urls(
                    config.get_final_distro(),
                    config.get_area(),
                    config.get_arch()))

    def _write_template_html(self, filename, body, mapping={}, defer_if_unmodified=False):
        mapping = mapping.copy()
        mapping.update({
            "section_navigation": self._section_navigation,
            "doc_root": self._doc_root,
            "section": html_protect(self._config.section),
        })
        write_template_html(filename, body, mapping, defer_if_unmodified=defer_if_unmodified, md5cache=self._md5cache)

    def write_log_list_page(self, filename, title, preface, logs):
        packages = {}
        for pathname, package, version in logs:
            packages[package] = packages.get(package, []) + [(pathname, version)]

        names = sorted(packages.keys())
        lines = []
        version_count = 0
        for package in names:
            versions = []
            for pathname, version in packages[package]:
                version_count += 1
                versions.append("<a href=\"%s\">%s</a>" %
                                (html_protect(pathname),
                                 html_protect(version)))
            line = "<tr class=\"normalrow\"><td class=\"contentcell2\">%s</td><td class=\"contentcell2\">%s</td></tr>" % \
                                (html_protect(package),
                                 ", ".join(versions))
            lines.append(line)

        if "FAIL" in preface:
            title_style = "alerttitlecell"
        else:
            title_style = "titlecell"

        self._write_template_html(
            filename,
                LOG_LIST_BODY_TEMPLATE,
                {
                    "page_title": html_protect(title + " in " + self._config.section),
                    "title": html_protect(title),
                    "title_style": title_style,
                    "preface": preface,
                    "count": len(packages),
                    "versioncount": version_count,
                    "logrows": "".join(lines),
                })

    def print_by_dir(self, output_directory, logs_by_dir):
        for vdir in logs_by_dir:
            vlist = []
            for basename in logs_by_dir[vdir]:
                assert basename.endswith(".log")
                assert "_" in basename
                package, version = basename[:-len(".log")].split("_")
                vlist.append((os.path.join(vdir, basename), package, version))
            self.write_log_list_page(os.path.join(output_directory, vdir + ".html"),
                                title_by_dir[vdir],
                                desc_by_dir[vdir], vlist)

    def find_links_to_logs(self, package_name, dirs, logs_by_dir):
        links = []
        for vdir in dirs:

            # avoid linear search against log file names by caching in a dict
            #
            # this cache was added to avoid a very expensive linear search
            # against the arrays in logs_by_dir. Note that the use of this cache
            # assumes that the contents of logs_by_dir is invarient across calls
            # to find_links_to_logs()
            #
            if vdir not in self._log_name_cache:
                self._log_name_cache[vdir] = {}

                for basename in logs_by_dir[vdir]:
                    if basename.endswith(".log"):
                        package, version = basename[:-len(".log")].split("_")

                        self._log_name_cache[vdir][package] = version

            if vdir == "fail":
                style = " class=\"needs-bugging\""
            else:
                style = ""

            if package_name in self._log_name_cache[vdir]:
                basename = package_name \
                    + "_" \
                         + self._log_name_cache[vdir][package_name] \
                         + ".log"

                links.append("<a href=\"%s/%s\"%s>%s</a>" % (
                    self._doc_root,
                        os.path.join(self._config.section, vdir, basename),
                        style,
                        html_protect(self._log_name_cache[vdir][package_name]),
                ))

        return links

    def link_to_maintainer_summary(self, maintainer):
        email = get_email_address(maintainer)
        return "<a href=\"%s/%s/maintainer/%s/%s.html\">%s</a>" \
               % (self._doc_root, self._config.section, maintainer_subdir(email),
                  email, html_protect(maintainer))

    def link_to_uploaders(self, uploaders):
        link = ""
        for uploader in uploaders.split(","):
            link += self.link_to_maintainer_summary(uploader.strip()) + ", "
        return link[:-2]

    def link_to_source_summary(self, package_name):
        source_name = self._binary_db.get_source(package_name)
        link = "<a href=\"%s/%s/source/%s\">%s</a>" % (
            self._doc_root,
                self._config.section,
                source_subdir(source_name) + "/" + source_name + ".html",
                html_protect(package_name))
        return link

    def link_to_state_page(self, section, package_name, link_target):
        if self._binary_db.has_package(package_name):
            state = self._binary_db.get_package_state(package_name)
            link = "<a href=\"%s/%s/%s\">%s</a>" % (
                self._doc_root,
                section,
                "state-" + state + ".html" + "#" + package_name,
                link_target)
        else:
            if link_target == package_name:
                link = html_protect(package_name)
            else:
                link = "unknown-package"

        return link

    def links_to_logs(self, package_name, state, logs_by_dir):
        link = "N/A"
        dirs = ""

        if state == "successfully-tested":
            dirs = ["pass"]
        elif state == "failed-testing":
            dirs = ["fail", "bugged", "affected"]
        elif state == "cannot-be-tested":
            dirs = ["untestable"]

        if dirs != "":
            links = self.find_links_to_logs(package_name, dirs, logs_by_dir)
            link = ", ".join(links)

        if "/bugged/" in link or "/affected/" in link:
            link += " - <a href=\"https://bugs.debian.org/cgi-bin/pkgreport.cgi?package=" \
                    + package_name \
                    + "\" target=\"_blank\" class=\"bugged\">&nbsp;bug filed&nbsp;</a>\n"

        return link

    def write_counts_summary(self):
        logging.debug("Writing counts.txt")
        header = "date"
        current_day = "%s" % time.strftime("%Y%m%d")
        counts = current_day
        total = 0
        for state in self._binary_db.get_states():
            count = len(self._binary_db.get_pkg_names_in_state(state))
            header += ", %s" % state
            counts += ", %s" % count
            logging.debug("%s: %s" % (state, count))
            total += count
        logging.debug("total: %s" % total)
        logging.debug("source: %s" % len(self._source_db.get_all_packages()))
        header += "\n"
        counts += "\n"

        countsfile = os.path.join(self._section_directory, "counts.txt")
        if not os.path.isfile(countsfile):
            logging.debug("writing new file: %s" % countsfile)
            create_file(countsfile, header)
            last_line = ""
        else:
            last_line = readlines_file(countsfile)[-1]
        if not current_day in last_line:
            append_file(countsfile, counts)
            logging.debug("appending line: %s" % counts.strip())

        shutil.copy(countsfile, self._output_directory)

        return total

    def create_maintainer_summaries(self, maintainers, source_data):
        logging.debug("Writing %d maintainer summaries in %s" % (len(maintainers), self._config.section))
        maintainer_dir = os.path.join(self._output_directory, "maintainer")
        if not os.path.exists(maintainer_dir):
            os.mkdir(maintainer_dir)
        states = ["fail", "unknown", "pass"]
        for maintainer in sorted(maintainers.keys()):
            sources = maintainers[maintainer]
            maintainer_subdir_path = os.path.join(maintainer_dir, maintainer_subdir(maintainer))
            if not os.path.exists(maintainer_subdir_path):
                os.mkdir(maintainer_subdir_path)
            rows = ""
            package_rows = {}
            packages = {}
            for state in states:
                packages[state] = []
                package_rows[state] = ""
            for source in sorted(sources):
                (state, sourcerows, binaryrows) = source_data[source]
                packages[state].append(source)
                package_rows[state] += sourcerows + binaryrows

            for state in states:
                if len(packages[state]) > 0:
                    links = ""
                    for package in packages[state]:
                        links += "<a href=\"#%s\">%s</a> " % (package, package)
                else:
                    links = "&nbsp;"
                rows +=   "<tr class=\"normalrow\">" \
                        + "<td class=\"labelcell\">%s:</td>" % state \
                        + "<td class=\"contentcell2\">%s</td>" % len(packages[state]) \
                        + "<td class=\"contentcell2\" colspan=\"4\">%s</td>" % links \
                        + "</tr>\n"

            distrolinks = "<tr class=\"normalrow\">" \
                          + "<td class=\"labelcell\">other distributions: </td>" \
                          + "<td class=\"contentcell2\" colspan=\"5\">"

            for section in self._section_names:
                if section != self._config.section:
                    distrolinks += "<a href=\"" \
                                   + self._doc_root \
                                   + "/" \
                                   + section \
                                   + "/maintainer/" \
                                   + maintainer_subdir(maintainer) \
                                   + "/" \
                                   + maintainer \
                                   + ".html\">" \
                                   + html_protect(section) \
                                   + "</a> "
            distrolinks += "</td></tr>"

            self._write_template_html(
                os.path.join(maintainer_subdir_path, maintainer + ".html"),
                    MAINTAINER_BODY_TEMPLATE,
                    {
                        "page_title": html_protect("Status of "
                                          + maintainer
                                          + " packages in "
                                          + self._config.section),
                        "maintainer": html_protect(maintainer + " in " + self._config.section),
                        "distrolinks": distrolinks,
                        "rows": rows + "".join([package_rows[state] for state in states]),
                    },
                    defer_if_unmodified=True)

    def create_source_summary(self, source, logs_by_dir):
        source_version = self._source_db.get_test_versions(source)
        binaries = self._source_db.get_control_header(source, "Binary")
        maintainer = self._source_db.get_control_header(source, "Maintainer")
        uploaders = self._source_db.get_control_header(source, "Uploaders")

        success = True
        failed = False
        binaryrows = ""
        for binary in sorted([x.strip() for x in binaries.split(",") if x.strip()]):
            if not self._binary_db.has_package(binary):
                    # udebs or binary packages for other architectures
                    # The latter is a FIXME which needs parsing the Packages files from other archs too
                binaryrows +=   "<tr class=\"normalrow\">" \
                              + "<td class=\"labelcell\">Binary:</td>" \
                              + "<td class=\"contentcell2\">%s</td>" \
                                % binary \
                              + "<td class=\"contentcell2\" colspan=\"4\">unknown package</td>" \
                              + "</tr>\n"
                continue

            state = self._binary_db.get_package_state(binary, recurse=False)

            if not "waiting" in state and "dependency" in state:
                state_style = "lightalertlabelcell"
            elif state == "failed-testing":
                state_style = "lightlabelcell"
            else:
                state_style = "labelcell"

            binary_version = self._binary_db.get_test_versions(binary)
            binary_arch = self._binary_db.get_control_header(binary, "Architecture")
            archsuffix = ""
            if binary_arch == "all":
                archsuffix = ":all"
            binaryrows +=   "<tr class=\"normalrow\">" \
                          + "<td class=\"labelcell\">Binary:</td>" \
                          + "<td class=\"contentcell2\">%s%s</td>" \
                            % (binary, archsuffix) \
                          + "<td class=\"%s\">piuparts-result:</td>" \
                            % state_style \
                          + "<td class=\"contentcell2\">%s %s</td>" \
                            % (self.link_to_state_page(self._config.section, binary, state),
                               self.links_to_logs(binary, state, logs_by_dir)) \
                          + "<td class=\"labelcell\">Version:</td>" \
                          + "<td class=\"contentcell2\">%s</td>" \
                            % html_protect(binary_version) \
                          + "</tr>\n"

            if state not in ("successfully-tested", "essential-required"):
                success = False
            if state in ("failed-testing", "dependency-does-not-exist", "cannot-be-tested"):
                failed = True

        if binaryrows != "":
            source_state = "unknown"

            if success:
                source_state = "<img src=\"%s/images/sunny.png\" alt=\"success\">" % self._doc_root
            if failed:
                source_state = "<img src=\"%s/images/weather-severe-alert.png\" alt=\"failed\">" % self._doc_root

            sourcerows =    "<tr class=\"titlerow\">" \
                          + "<td class=\"titlecell\" colspan=\"6\" id=\"%s\">%s in %s</td>" \
                            % (package2id(source), source, self._config.section) \
                          + "</tr>\n"

            sourcerows +=   "<tr class=\"normalrow\">" \
                          + "<td class=\"labelcell\">Source:</td>" \
                          + "<td class=\"contentcell2\">" \
                            + "<a href=\"https://tracker.debian.org/%s\" target=\"_blank\">%s</a>" \
                              % (source, html_protect(source)) \
                            + "</td>" \
                          + "<td class=\"labelcell\">piuparts summary:</td>" \
                          + "<td class=\"contentcell2\">%s</td>" \
                            % source_state \
                          + "<td class=\"labelcell\">Version:</td>" \
                          + "<td class=\"contentcell2\">%s</td>" \
                            % html_protect(source_version) \
                          + "</tr>\n"

            sourcerows += "<tr class=\"normalrow\">" \
                          + "<td class=\"labelcell\">Maintainer:</td>" \
                          + "<td class=\"contentcell2\" colspan=\"5\">%s</td>" \
                            % self.link_to_maintainer_summary(maintainer) \
                          + "</tr>\n"

            if uploaders:
                sourcerows += "<tr class=\"normalrow\">" \
                              + "<td class=\"labelcell\">Uploaders:</td>" \
                              + "<td class=\"contentcell2\" colspan=\"5\">%s</td>" \
                                % self.link_to_uploaders(uploaders) \
                              + "</tr>\n"

            source_summary_page_path = os.path.join(self._output_directory,
                                                    "source",
                                                     source_subdir(source))

            if not os.path.exists(source_summary_page_path):
                os.makedirs(source_summary_page_path)

            self._write_template_html(
                    os.path.join(source_summary_page_path, (source + ".html")),
                    SOURCE_PACKAGE_BODY_TEMPLATE,
                    {
                        "page_title": html_protect("Status of source package " + source + " in " + self._config.section),
                        "rows": sourcerows + binaryrows,
                    },
                    defer_if_unmodified=True)

            # return parsable values
            if success:
                source_state = "pass"
            if failed:
                source_state = "fail"
        else:
            source_state = "udeb"
            sourcerows = ""

        return sourcerows, binaryrows, source_state, maintainer, uploaders

    def create_package_summaries(self, logs_by_dir):
        src_names = sorted(self._source_db.get_all_package_names())
        logging.debug("Writing %d source summaries in %s" % (len(src_names), self._config.section))

        maintainers = {}
        source_binary_rows = {}
        sources = ""
        sources_data = []
        for source in src_names:
            (sourcerows, binaryrows, source_state, maintainer, uploaders) = \
                               self.create_source_summary(source, logs_by_dir)

            if source_state != "udeb":
                source_version = self._source_db.get_test_versions(source)
                source_data = {};
                source_data["source"] = source
                source_data["state"] = source_state
                source_data["version"] = source_version
                sources_data.append(source_data)

                sources += "%s: %s\n" % (source, source_state)
                source_binary_rows[source] = (source_state, sourcerows, binaryrows)
                for maint in [maintainer] + uploaders.split(","):
                    if maint.strip():
                        email = get_email_address(maint.strip())
                        if not "INVALID" in email:
                            if not email in maintainers:
                                maintainers[email] = []
                            maintainers[email].append(source)

        create_file(os.path.join(self._output_directory, "sources.txt"), sources)
        create_file(os.path.join(self._output_directory, "sources.yaml"),
            yaml.dump(sources_data, default_flow_style=False))

        self.create_maintainer_summaries(maintainers, source_binary_rows)

    def make_section_stats_graph(self):
        countsfile = os.path.join(self._section_directory, "counts.txt")
        pngfile = os.path.join(self._output_directory, "states.png")
        grdevices = importr('grDevices')
        grdevices.png(file=pngfile, width=1600, height=900, pointsize=10, res=100, antialias="none")
        r = robjects.r
        states = self._binary_db.get_states()
        active_states = self._binary_db.get_active_states()
        r('nstate = %d' % len(states))
        r('snames = c(%s)' % ','.join(['"%s"' % s if s in active_states else '"(%s)"' % s for s in states]))
        r('cnames = c("date",snames)')
        r('t <- (read.table("' + countsfile + '",sep=",",header=0,skip=1,col.names=cnames,row.names=1,fill=TRUE))')
        # here we define how many days we wants stats for (163=half a year)
        # r('v <- t[(nrow(t)-163):nrow(t),0:nstate]')
        # make graph since day 1
        r('v <- t[0:nrow(t),0:nstate]')
        # thanks to http://tango.freedesktop.org/Generic_Icon_Theme_Guidelines for those nice colors
        r('palsize = 14')
        r('palette(c("#4e9a06", "#ef2929", "#d3d7cf", "#5c3566", "#c4a000", \
                     "#fce94f", "#a40000", "#888a85", "#2e3436", "#729fcf", \
                     "#3465a4", "#204a87", "#555753", "#ce5c00"))')
        r('barplot(t(v),col = 1:palsize, \
          main="Binary packages per state in ' + self._config.section + '", \
          xlab="", ylab="Number of binary packages", space=0, border=NA)')
        r('legend(x="bottom",legend=snames, ncol=2,fill=1:palsize,xjust=0.5,yjust=0,bty="n")')
        grdevices.dev_off()

        stats_html = "<tr class=\"normalrow\"> " \
                     + "<td class=\"contentcell2\" colspan=\"3\">" \
                     + "<a href=\"%s\">" \
                       % "states.png" \
                     + "<img src=\"%s\" height=\"450\" width=\"800\" alt=\"Binary package states in %s\">" \
                       % ("states.png", self._config.section) \
                     + "</a>" \
                     + "<br>(<a href=\"counts.txt\">Source</a>)\n" \
                     + "</td></tr>\n"

        return stats_html

    def create_and_link_to_analysises(self, state):
        link = "<ul>\n"
        for template, linktarget in linktarget_by_template:
            # successful logs only have issues and failed logs only have errors
            if (state == "failed-testing" and template[-9:] != "issue.tpl") \
                    or (state == "successfully-tested" and template[-9:] == "issue.tpl"):
                substats = ""

                tpl = os.path.join(self._output_directory, template)
                if (os.path.exists(tpl)):
                    logging.debug("Writing analysis page for %s" % template)
                    rows = read_file(tpl)
                    os.unlink(tpl)

                    self._write_template_html(
                            os.path.join(self._output_directory, template[:-len(".tpl")] + ".html"),
                            ANALYSIS_BODY_TEMPLATE,
                            {
                                "page_title": html_protect("Packages in state " + state + " " + linktarget),
                                "rows": rows,
                            })
                    if state == "failed-testing":
                        count_bugged = string.count(rows, '"bugged/')
                        count_affected = string.count(rows, '"affected/')
                        count_failed = string.count(rows, '"fail/')
                        sep = ": "
                        if count_bugged > 0:
                            substats += sep + "%s bugged" % count_bugged
                            sep = ", "
                        if count_affected > 0:
                            substats += sep + "%s affected" % count_affected
                            sep = ", "
                        if count_failed > 0:
                            substats += sep + "<span class=\"needs-bugging\">%s failed</span>" % count_failed
                    else:
                        count_passed = string.count(rows, '"pass/')
                        if count_passed > 0:
                            substats += ": %s passed" % count_passed
                    link += "<li><a href=%s>%s</a>%s</li>\n" % \
                             (
                                 template[:-len(".tpl")] + ".html",
                                 linktarget,
                                 substats,
                             )

        link += "</ul>"
        if link == "<ul>\n</ul>":
            link = ""
        return link

    def write_section_index_page(self, dirs, total_packages):
        tablerows = ""
        for state in self._binary_db.get_active_states():
            dir_link = ""
            analysis = ""
            for vdir in dirs:
                if vdir in ("pass", "fail", "bugged", "affected", "untestable") and state_by_dir[vdir] == state:
                    dir_link += "<a href='%s.html'>%s</a> logs<br>" % (vdir, html_protect(vdir))
            if state in ("successfully-tested", "failed-testing"):
                analysis = self.create_and_link_to_analysises(state)
            tablerows += ("<tr class=\"normalrow\"><td class=\"contentcell2\"><a href='state-%s.html'>%s</a>%s</td>" +
                          "<td class=\"contentcell2\">%d</td><td class=\"contentcell2\">%s</td></tr>\n") % \
                          (html_protect(state), html_protect(state), analysis, len(self._binary_db.get_pkg_names_in_state(state)),
                          dir_link)
        try:
            tablerows += self.make_section_stats_graph();
        except:
            logging.debug("Error generating the graph images, probably python-rpy2 is not installed, disabling graphs.")

        tablerows += "<tr class=\"normalrow\"> <td class=\"labelcell2\">Total</td> <td class=\"labelcell2\" colspan=\"2\">%d</td></tr>\n" % total_packages
        vendor = "Debian"
        if len(self._config.get_distros()) > 1:
            description = "%s %s: package installation in %s" % (
                vendor,
                self._config.get_area(),
                self._config.get_start_distro())
            for distro in self._config.get_distros()[1:]:
                description += ", dist-upgrade to %s" % distro
            description += ", removal, and purge test."
        else:
            description = "%s %s / %s: package installation, removal, and purge test." % (
                vendor,
                self._config.get_distro(),
                self._config.get_area())
        if self._config["description"].startswith("+"):
            description = description + " " + self._config["description"][1:]
        elif self._config["description"].endswith("+"):
            description = self._config["description"][:-1] + " " + description
        elif self._config["description"]:
            description = self._config["description"]

        self._write_template_html(
                os.path.join(self._output_directory, "index.html"),
                SECTION_INDEX_BODY_TEMPLATE,
                {
                    "page_title": html_protect(self._config.section + " statistics"),
                    "description": html_protect(description),
                    "tablerows": tablerows,
                    "packagesurl": "<br>".join([html_protect(url) for url in self._binary_db.get_urls()]),
                })

    def _show_providers(self, dep):
        providers = self._binary_db.get_providers(dep)
        vlist = ""
        if providers:
            vlist += "\n<ul>\n"
            for provider in providers:
                vlist += "<li>provider %s is %s</li>\n" % \
                          (self.link_to_state_page(self._config.section, provider, provider),
                          emphasize_reason(html_protect(self._binary_db.get_package_state(provider))))
            vlist += "</ul>\n"
        return vlist

    def write_state_pages(self):
        for state in self._binary_db.get_active_states():
            logging.debug("Writing page for %s" % state)
            vlist = ""
            if state in self._binary_db.get_error_states():
                with_counts = True
                aside = " (reverse deps, blocked pkgs)"
                sort_key = lambda x: (-self._binary_db.block_count(x["Package"]), x["Package"])
            else:
                with_counts = False
                aside = ""
                sort_key = lambda x: x["Package"]

            names = self._binary_db.get_pkg_names_in_state(state)
            packages = [self._binary_db.get_package(name) for name in names]
            packages.sort(key=sort_key)

            for package in packages:
                vlist += "<li id=\"%s\">%s" % (
                                         package2id(package["Package"]),
                                         self.link_to_source_summary(package["Package"]))
                if package["Architecture"] == "all":
                    vlist += ":all"
                if with_counts:
                    vlist += " (%d, %d)" % (self._binary_db.rrdep_count(package["Package"]),
                                            self._binary_db.block_count(package["Package"]))
                vlist += " (%s)" % html_protect(package["Maintainer"])
                all_deps = unique(package.all_dependencies())
                if all_deps:
                    vlist += "\n<ul>\n"
                    for alternatives in all_deps:
                        dep = alternatives[0]
                        vlist += "<li>dependency %s is %s" % \
                                  (self.link_to_state_page(self._config.section, dep, dep),
                                  emphasize_reason(html_protect(self._binary_db.get_package_state(dep, resolve_virtual=False))))
                        vlist += self._show_providers(dep)
                        if len(alternatives) > 1:
                            vlist += "\n<ul>\n"
                            for dep in alternatives[1:]:
                                vlist += "<li>alternative dependency %s is %s" % \
                                          (self.link_to_state_page(self._config.section, dep, dep),
                                          emphasize_reason(html_protect(self._binary_db.get_package_state(dep, resolve_virtual=False))))
                                vlist += self._show_providers(dep)
                                vlist += "</li>\n"
                            vlist += "</ul>\n"
                        vlist += "</li>\n"
                    vlist += "</ul>\n"
                vlist += "</li>\n"

            self._write_template_html(
                    os.path.join(self._output_directory, "state-%s.html" % state),
                    STATE_BODY_TEMPLATE,
                    {
                                        "page_title": html_protect("Packages in state " + state + " in " + self._config.section),
                                        "state": html_protect(state),
                                        "list": vlist,
                                        "aside": aside,
                    })

    def archive_logfile(self, vdir, log):
        archivedir = os.path.join("archive", vdir)
        if not os.path.exists(archivedir):
            os.makedirs(archivedir)
        try:
            os.rename(os.path.join(vdir, log), os.path.join("archive", vdir, log))
        except OSError:
            logging.debug("OSError while archiving %s/%s" % (vdir, log))

    def cleanup_removed_packages(self, logs_by_dir):
        vdirs = logs_by_dir.keys()
        vdirs.remove("reserved")
        for vdir in vdirs:
            for log in sorted(logs_by_dir[vdir]):
                if log.endswith(".log"):
                    package, version = log[:-len(".log")].split("_")
                    if not self._binary_db.has_package(package):
                        logging.debug("Archiving %s/%s, package was removed" % (vdir, log))
                        self.archive_logfile(vdir, log)
                        logs_by_dir[vdir].remove(log)
                    else:
                        current = self._binary_db.get_test_versions(package)
                        state = self._binary_db.get_package_state(package, resolve_virtual=False, recurse=False)
                        if version != current:
                            logging.debug("Archiving %s/%s, package is outdated (%s)" %
                                          (vdir, log, current))
                            self.archive_logfile(vdir, log)
                            logs_by_dir[vdir].remove(log)
                        elif state == "outdated":
                            logging.debug("Archiving %s/%s, package is superseded" %
                                          (vdir, log))
                            self.archive_logfile(vdir, log)
                            logs_by_dir[vdir].remove(log)

    def generate_html(self):
        md5cachefile = os.path.join(self._output_directory, '.md5cache')
        try:
            with open(md5cachefile, "r") as f:
                self._md5cache['old'] = pickle.load(f)
        except IOError:
            pass

        logging.debug("Finding log files")
        dirs = ["pass", "fail", "bugged", "affected", "reserved", "untestable"]
        logs_by_dir = {}
        for vdir in dirs:
            logs_by_dir[vdir] = find_files_with_suffix(vdir, ".log")

        logging.debug("Archiving logs of obsolete packages")
        self.cleanup_removed_packages(logs_by_dir)

        logging.debug("Copying log files")
        copy_logs(logs_by_dir, self._output_directory)

        logging.debug("Removing old log files")
        remove_old_logs(logs_by_dir, self._output_directory)

        logging.debug("Writing per-dir HTML pages")
        self.print_by_dir(self._output_directory, logs_by_dir)

        total_packages = self.write_counts_summary()

        self.create_package_summaries(logs_by_dir)

        logging.debug("generate known problem *.tpl")
        dwke_process_section(self._config.section, '.', self._output_directory, self._problem_list, self._binary_db)

        logging.debug("Writing section index page")
        self.write_section_index_page(dirs, total_packages)

        logging.debug("Writing stats pages for %s" % self._config.section)
        self.write_state_pages()

        logging.debug("Wrote %d out of %d html files, refreshed %d out of %d unmodified files" % ( \
                self._md5cache['written'], len(self._md5cache['new']),
                self._md5cache['refreshed'], self._md5cache['unmodified']))
        with open(md5cachefile, "w") as f:
            pickle.dump(self._md5cache['new'], f)


    def generate_summary(self, web_host):
        summary_path = os.path.join(self._output_directory, "summary.json")

        if os.path.isfile(summary_path):
            os.unlink(summary_path)

        reporting_sections = self._config['json-sections'].split()
        if not reporting_sections or reporting_sections[0] == 'default':
            reporting_sections = [self._config.get_std_distro()]

        if reporting_sections[0] == 'none':
            logging.debug("Skipping summary")
        else:
            logging.debug("Generating summary")

            summary = pkgsummary.new_summary()

            for reporting_section in reporting_sections:
                for binpkg in self._binary_db.get_all_packages():
                    pkgname = binpkg["Package"]
                    state = self._binary_db.get_package_state(pkgname)
                    flag = pkgsummary.get_flag(state)
                    block_cnt = 0
                    if flag == 'F':
                        block_cnt = self._binary_db.block_count(pkgname)
                    srcpkg = self._binary_db.get_source(pkgname)
                    url = source_summary_url(
                              web_host, self._doc_root,
                              self._config.section, srcpkg)

                    pkgsummary.add_summary(
                                   summary, reporting_section, srcpkg,
                                   flag, block_cnt, url)

            pkgsummary.write_summary(summary, summary_path)

    def generate_output(self, output_directory, section_names, problem_list, web_host):
        # skip output generation for disabled sections
        if int(self._config["max-reserved"]) == 0:
            return

        self._section_names = section_names
        self._section_navigation = create_section_navigation(
                self._section_names, self._config.section, self._doc_root)
        self._output_directory = os.path.abspath(os.path.join(output_directory, self._config.section))
        if not os.path.exists(self._output_directory):
            os.makedirs(self._output_directory)
        self._problem_list = problem_list

        oldcwd = os.getcwd()
        os.chdir(self._section_directory)
        self.generate_html()
        os.chdir(oldcwd)

        self.generate_summary(web_host)


def sections_by_precedence(sections):
    precedence = {}
    count = 0
    for section in sections:
        config = Config(section=section, defaults_section="global")
        config.read(CONFIG_FILE)
        precedence[section] = (config["precedence"], count)
        count += 1

    return sorted(sections, key=lambda x: precedence[x])


def generate_global_summary(dir, sections):
    json_name = "summary.json"

    logging.debug("Generating global summary")

    summary = pkgsummary.new_summary()

    for section in sections_by_precedence(sections):
        sec_path = os.path.join(dir, section, json_name)
        if os.path.isfile(sec_path):
            sec_summ = pkgsummary.read_summary(sec_path)
            summary = pkgsummary.merge_summary(summary, sec_summ)

    summary_path = os.path.join(dir, json_name)
    pkgsummary.write_summary(summary, summary_path)

# START detect_well_known_errors


def get_bug_text(logpath):
    bugpath = replace_ext(logpath, BUG_EXT)

    if os.path.exists(bugpath):
        return read_file(bugpath)
    return ""


def populate_tpl(tmpl, vals):

    for key in vals:
        tmpl = re.sub("\$%s" % key, str(vals[key]), tmpl)

    return tmpl


def update_tpl(basedir, section, problem, failures, logdict, ftpl, ptpl, pkgsdb):

    pkg_text = ""
    bugged_section = False
    for failure in failures:
        bin_pkg = get_pkg(failure.pkgspec)
        src_pkg = pkgsdb.get_source(bin_pkg)
        rdep_cnt = pkgsdb.rrdep_count(bin_pkg)
        bin_arch = ""
        if pkgsdb.get_control_header(bin_pkg, "Architecture") == "all":
            bin_arch = " [all]"

        if bugged_section is False and failure.where != 'fail':
            bugged_section = True
            pkg_text += "</ul><ul>\n"

        log = os.path.join(failure.where, failure.pkgspec + LOG_EXT)

        pkg_text += populate_tpl(ftpl, {
                                'LOG': log,
                                'PACKAGE': bin_pkg,
                                'ARCH': bin_arch,
                                'BUG': get_bug_text(log),
                                'RDEPS': rdep_cnt,
                                'SPKG': src_pkg,
                                   })

    if len(pkg_text):
        return populate_tpl(ptpl, {
                                'HEADER': problem.HEADER,
                                'SECTION': section,
                                'HELPTEXT': problem.HELPTEXT,
                                'COMMAND': problem.get_command(),
                                'PACKAGE_LIST': pkg_text,
                                'COUNT': len(failures),
        })
    return ""


def update_html(section, html_dir, logdict, problem_list, failures, pkgsdb):
    for problem in problem_list:
        tpl_text = update_tpl(html_dir, section, problem,
                   failures.filtered(problem.name),
                   logdict,
                   PKG_ERROR_TPL, PROB_TPL, pkgsdb)
        if len(tpl_text):
            with open(os.path.join(html_dir, problem.name[:-5] + TPL_EXT), 'w') as pf:
                pf.write(tpl_text)


def dwke_process_section(section, sectiondir, htmldir, problem_list, pkgsdb):
    workdirs = [os.path.join(sectiondir, x) for x in KPR_DIRS]

    logdict = get_file_dict(workdirs, LOG_EXT)

    failures = FailureManager(logdict)
    failures.sort_by_bugged_and_rdeps(pkgsdb)

    for prob in problem_list:
        pcount = len(failures.filtered(prob.name))
        if pcount:
            logging.info("%7d %s" % (pcount, prob.name))

    update_html(section, htmldir, logdict, problem_list, failures, pkgsdb)

# END detect_well_known_errors


def make_bts_stats_graph(master_dir, out_dir):
    countsfile = os.path.join(master_dir, "bts_stats.txt")
    pngfile = os.path.join(out_dir, "images", "bts_stats.png")
    grdevices = importr('grDevices')
    grdevices.png(file=pngfile, width=1600, height=900, pointsize=10, res=100)
    r = robjects.r
    r('t <- (read.table("' + countsfile + '",sep=",",header=1,row.names=1))')
    # make graph since day 1
    r('v <- t[c(4, 2, 3, 1)]')  # reorder columns
    # tango colors again:
    r('palette(c("#a40000", "#ef2929", "#4e9a06", "#8ae234"))')
    r('barplot(t(v),col = 1:4, \
        main="Bugs with usertag=piuparts and user=debian-qa@lists.debian.org", \
        xlab="", ylab="Total number of RC and non-RC bugs submitted and closed", space=0, border=NA)')
    r('legend("right", legend=rev(colnames(v)), fill=rev(1:4), inset=0.05, bty="n")')
    grdevices.dev_off()


def main():
    setup_logging(logging.DEBUG, None)
    global_config = Config(section="global")
    global_config.read(CONFIG_FILE)
    if global_config["proxy"]:
        os.environ["http_proxy"] = global_config["proxy"]
    section_names = global_config["sections"].split()
    process_section_names = section_names
    if len(sys.argv) > 1:
        process_section_names = sys.argv[1:]
    master_directory = global_config["master-directory"]
    output_directory = global_config["output-directory"]
    web_host = global_config["web-host"]

    doc_root = global_config["doc-root"].strip()
    if not doc_root.startswith("/"):
        doc_root = "/" + doc_root
    if doc_root.endswith("/"):
        doc_root = doc_root[:-1]
    problem_list = create_problem_list(global_config['known-problem-directory'])

    if not os.path.exists(master_directory):
        logging.debug("Warning: %s does not exist!?! Creating it for you now." % master_directory)
        os.makedirs(master_directory)
        return

    with open(os.path.join(master_directory, "report.lock"), "we") as lock:
        try:
            fcntl.flock(lock, fcntl.LOCK_EX | fcntl.LOCK_NB)
        except IOError:
            if sys.stdout.isatty():
                sys.exit("another piuparts-report process is already running")
            else:
                sys.exit(0)

        packagedb_cache = {}
        create_file(os.path.join(output_directory, "sections.yaml"),
            yaml.dump(section_names, default_flow_style=False))
        todo = deque([(s, 0) for s in process_section_names])
        while len(todo):
            (section_name, next_try) = todo.popleft()
            now = time.time()
            if (now < next_try):
                logging.info("Sleeping while section is busy")
                time.sleep(max(30, next_try - now) + 30)
            try:
                section_directory = os.path.join(master_directory, section_name)
                if not os.path.exists(section_directory):
                    raise MissingSection("", section_name)
                with open(os.path.join(section_directory, "master.lock"), "we") as lock:
                    try:
                        fcntl.flock(lock, fcntl.LOCK_EX | fcntl.LOCK_NB)
                    except IOError:
                        raise Busy()

                    section = Section(section_name, master_directory, doc_root, packagedb_cache=packagedb_cache)
                    section.generate_output(output_directory, section_names, problem_list, web_host)
            except Busy:
                logging.info("Section is busy")
                todo.append((section_name, time.time() + 300))
            except MissingSection as e:
                logging.error("Configuration Error in section '%s': %s" % (section_name, e))
            except (HTTPError, URLError) as e:
                logging.error("Failed to fetch Packages files for section '%s' : %s" % (section_name, e))

        # static pages
        logging.debug("Writing static pages")
        for page in ("index", "bug_howto"):
            tpl = os.path.join(output_directory, page + ".tpl")
            INDEX_BODY = read_file(tpl)
            write_template_html(
                    os.path.join(output_directory, page + ".html"),
                    INDEX_BODY,
                    {
                                   "page_title": "About piuparts.debian.org and News",
                                   "section_navigation": create_section_navigation(section_names, "sid", doc_root),
                                   "doc_root": doc_root,
                    })
        # create graph about bugs filed
        try:
            make_bts_stats_graph(master_directory, output_directory);
        except:
            logging.debug("Error generating the graph images, probably python-rpy2 is not installed, disabling graphs.")

        generate_global_summary(output_directory, section_names)

        logging.debug("Done")


if __name__ == "__main__":
    main()

# vi:set et ts=4 sw=4 :
