/bin/send-notifications.php

<?php # vim:ts=2:sw=2:et:
/* For licensing and copyright terms, see the file named LICENSE */
include dirname(__FILE__) . '/../inc/common.php';

$DEBUG = strlen(getenv('DEBUG_NOTIFY')) ? true : false;

$MAX_DIFF = 200 * 1024;
$USE_BATCHING = false;

if (!$DEBUG) {
  /* only allow one instance to run concurrently */
  $lockfp = fopen(dirname(__FILE__) . '/../var/.notifier.lock', 'w');
  if (!$lockfp) {
    exit(1);
  }
  if (!flock($lockfp, LOCK_EX|LOCK_NB)) {
    echo "Another instance is already running\n";
    exit(1);
  }
  /* "leak" $lockfp, so that the lock is held while we continue to run */
}

$db = MTrackDB::get();

// default to the last 10 minutes, but prefer the last recorded run time
$last = MTrackDB::unixtime(time() - 600);
foreach (MTrackDB::q('select last_run from last_notification')->fetchAll()
    as $row) {
  $last = $row[0];
}
$LATEST = strtotime($last);
if ($DEBUG) {
  $dtime = strtotime(getenv('DEBUG_TIME'));
  if ($dtime > 0) {
    $LATEST = $dtime;
    $last = MTrackDB::unixtime($LATEST);
    echo "Using $last as last time (specified via DEBUG_TIME var)\n";
  }
}

$by_object = array();

// Gather pertinent change information en-masse
$changes = MTrackDB::q("
  select * from changes where changedate > ? order by changedate asc", $last)
  ->fetchAll(PDO::FETCH_OBJ);
$cids = array();
$cs_by_cid = array();
foreach ($changes as $CS) {
  $cids[] = $CS->cid;
  $cs_by_cid[$CS->cid] = $CS;

  $t = strtotime($CS->changedate);
  if ($t > $LATEST) {
    $LATEST = $t;
  }

  list($object, $id) = explode(':', $CS->object, 3);
  switch ($object) {
    case 'ticket':
    case 'wiki':
    case 'milestone':
      $by_object[$object][$id][] = $CS->cid;
      break;
  }
}
$cidlist = join(',', $cids);
$change_audit = array();
$effort_audit = array();
if (count($cids)) {
  foreach (MTrackDB::q("select * from change_audit where cid in ($cidlist)")
      ->fetchAll(PDO::FETCH_ASSOC) as $citem) {
    $change_audit[$citem['cid']][] = $citem;
  }
  foreach (MTrackDB::q(
        "select * from effort where cid in ($cidlist)")
      ->fetchAll(PDO::FETCH_ASSOC) as $eff) {
    $effort_audit[$eff['cid']][] = $eff;
  }
}

$comp_by_id = array();
foreach (MTrackDB::q('select c.compid, c.name, p.name from components c left join components_by_project cbp on (c.compid = cbp.compid) left join projects p on (cbp.projid = p.projid) where deleted <> 1 order by c.name')
    ->fetchAll(PDO::FETCH_NUM) as $row) {
  if (strlen($row[2])) {
    $row[1] .= " ($row[2])";
  }
  $comp_by_id[$row[0]] = $row[1];
}

$milestone_by_id = array();
foreach (MTrackDB::q('select  mid, name from milestones')->fetchAll() as $m)
{
  $milestone_by_id[$m[0]] = $m[1];
}
$proj_by_comp = array();
foreach (MTrackDB::q('select projid, compid from components_by_project')
    ->fetchAll() as $r) {
  $proj_by_comp[$r['compid']] = $r['projid'];
}
$proj_by_id = array();
foreach (MTrackDB::q('select projid from projects')->fetchAll() as $r) {
  $proj_by_id[$r[0]] = MTrackProject::loadById($r[0]);
}

function get_user_info($name)
{
  $name = mtrack_canon_username($name);
  static $info = array();
  if (isset($info[$name])) {
    return $info[$name];
  }
  foreach (MTrackDB::q('select * from userinfo where userid = ?', $name)
      ->fetchAll(PDO::FETCH_ASSOC) as $data) {
    $info[$name] = $data;
    return $data;
  }
  // fake it
  $data = array(
    'userid' => $name,
    'email' => $name . '@' . MTrackConfig::get('core', 'default_email_domain'),
    'fullname' => $name,
  );
  $info[$name] = $data;
  return $data;
}


// Need to assess changesets from the various repos to catch changes
// that don't reference tickets

// TODO: For mercurial and presumably other DVCS, we need to track changegroups
// independently, because the timestamp of the push is not stored in the repo.
// This means that it is possible to sit on a commit for a couple of days
// before pushing it, and this script will not see it because the log date in
// the repo will always be far in the past even though the push just happened.

$repo_changes_by_ticket = array();
$repos_by_name = array();

foreach (MTrackDB::q('select shortname from repos order by shortname')
      ->fetchAll(PDO::FETCH_COLUMN, 0) as $reponame) {
  $repo = MTrackRepo::loadByName($reponame);
  $repos_by_name[$reponame] = $repo;
  $checker = new MTrackCommitChecker($repo);

  foreach ($repo->history(null, $last) as $e) {
    // the SCM may give us information on an item older than the date
    // we requested, so we need to filter here too
    $t = strtotime($e->ctime);
    if ($t <= strtotime($last)) {
echo "[$e->rev] exclude $e->ctime $last\n";
      continue;
    }
echo "[$e->rev] include $e->ctime $last\n";
    if ($t > $LATEST) {
      $LATEST = $t;
echo "  update latest to $t\n";
    }

    $pid = $repo->projectFromPath($e->files);
    if ($pid > 1) {
      $proj = $proj_by_id[$pid];
      $e->changelog = $proj->adjust_links($e->changelog, true);
    }
    $actions = $checker->parseCommitMessage($e->changelog);
    $tickets = array();
    foreach ($actions as $act) {
      $tkt = $act[1];
      $tickets[$tkt] = $tkt;
      $repo_changes_by_ticket[$tkt][$reponame][$e->rev] = $e->rev;
    }
    if (count($tickets) == 0) {
      // This changeset is not represented by a change in a ticket
      $by_object['changeset'][$reponame][$e->rev] = $e;
    }
  }
}

$batch_fields = array(
  'owner' => true,
  'status' => true,
  'priority' => true,
  '@milestones' => true,
);

function major_contributor($list)
{
  $major = null;
  $count = 0;
  foreach ($list as $user => $input) {
    if ($input > $count) {
      $major = $user;
      $count = $input;
    }
  }
  return $major;
}

if (isset($by_object['ticket'])) {
  // If tickets were subject to roadmap/pri change only, say as part of
  // a mass update, then we recognize those and send out a batch notification
  // separately

  $batch_tkt = array();
  $tkt_by_tid = array();

  foreach ($by_object['ticket'] as $tid => $cslist) {
    $T = MTrackIssue::loadById($tid);

    $fields = array();
    $oid = "ticket:$tid";
    $is_batch = true;
    $old_values = array();
    $comments = array();
    $field_changers = array();
    $contributors = array();
    $projects = array(); // defines who gets notified

    foreach ($cslist as $cid) {
      $contributors[$cs_by_cid[$cid]->who]++;
      if (isset($change_audit[$cid])) foreach ($change_audit[$cid] as $C) {
        // fieldname is of the form: "ticket:id:fieldname"
        $field = substr($C['fieldname'], strlen($oid)+1);
        if (!isset($batch_fields[$field])) {
          $is_batch = false;
        }
        if ($field == '@comment') {
          $comments[] = "Comment by " . 
            $cs_by_cid[$cid]->who . ":\n" . $C['value'];
        } elseif ($field != 'spent') {
          $field_changers[$field] = $cs_by_cid[$cid]->who;
          if (!isset($old_values[$field])) {
            $old_values[$field] = $C['oldvalue'];
          }
        }
        $contributors[$cs_by_cid[$cid]->who]++;
      }
    }
    if ($is_batch && $USE_BATCHING) {
      $batch_tkt[$tid] = $cslist;
      $tkt_by_tid[$tid] = $T;
      continue;
    }
    /* changes were not suitable for batching, so now we represent them
     * as history on that ticket */
    $plain = '';
    $plain .= MTrackConfig::get('core', 'weburl')
            . "ticket.php/$T->nsident\n\n";
    $plain .=  "#$T->nsident: $T->summary ($T->status $T->classification)\n";
    $owner = strlen($T->owner) == 0 ? 'nobody' : $T->owner;
    $plain .= "Responsible: $owner ($T->priority / $T->severity)\n";
    $plain .= "Milestone: " . 
                join(', ', $T->getMilestones()) . "\n";
    $plain .= "Component: " . 
                join(', ', $T->getComponents()) . "\n";
    $plain .= "\n";

    foreach ($T->getComponents() as $compid => $comp) {
      $projects[$proj_by_comp[$compid]]++;
    }

    // Display changed fields grouped by the person that last changed them
    $who_changed = array();
    foreach ($field_changers as $field => $who) {
      $who_changed[$who][] = $field;
    }

    foreach ($who_changed as $who => $fieldlist) {
      $plain .= "Changes by $who:\n";
      foreach ($fieldlist as $field) {
        $old = $old_values[$field];

        if (!strlen($old) && $field == 'nsident') {
          continue;
        }

        $value = null;
        switch ($field) {
          case '@components':
            $old = array();
            foreach (preg_split("/\s*,\s*/", $old_values[$field]) as $id) {
              $old[] = $comp_by_id[$id];
              $projects[$proj_by_comp[$id]]++;
            }
            $value = array();
            foreach ($T->getComponents() as $id => $compname) {
              $value[$id] = $comp_by_id[$id];
            }
            $field = 'Component';
            break;
          case '@milestones':
            $old = array();
            foreach (preg_split("/\s*,\s*/", $old_values[$field]) as $id) {
              $old[] = $milestone_by_id[$id];
            }
            $value = $T->getMilestones();
            $field = 'Milestone';
            break;
          case '@keywords':
            $field = 'Keywords';
            $value = $T->getKeywords();
            break;
          default:
            $value = $T->$field;
        }
        if (is_array($value)) {
          $value = join(', ', $value);
        }
        if (is_array($old)) {
          $old = join(', ', $old);
        }
        if ($value == $old) {
          continue;
        }
        if ($field == 'description') {
          $lines = count(explode("\n", $old));
          $diff = mtrack_diff_strings($old, $value);
          $diff_add = 0;
          $diff_rem = 0;
          foreach (explode("\n", $diff) as $line) {
            if ($line[0] == '-') {
              $diff_rem++;
            } else if ($line[0] == '+') {
              $diff_add++;
            }
          }
          if (abs($diff_add - $diff_rem) > $lines / 2) {
            $plain .= "Description changed to:\n" . $value . "\n\n";
          } else {
            $plain .= "Description changed:\n" . $diff . "\n\n";
          }
        } else {
          $plain .= "$field $old -> $value\n";
        }
      }
    }
    foreach ($comments as $comment) {
// parse the description to see if it matches the "(In
// [changeset:repo,id], [changeset:repo,id]...)" form.  This is needed because
// there is a possibility that we will opt to exclude a changeset at the top of
// this script based on date.  We want to pluck out all the changesets that are
// referenced in such a way and add them to the $repo_changes_by_ticket array
// for this ticket
// FIXME: need to handle the multiple changeset case for DVCS like hg.
      if (preg_match_all(
          "/\(In \[changeset:(([^,]+),([a-zA-Z0-9]+))\]\)/sm",
          $comment, $CSM)) {
        // $CSM[2] => repo
        // $CSM[3] => changeset
        foreach ($CSM[2] as $csm => $csm_repo) {
          $csm_rev = $CSM[3][$csm];
          if (!isset($repo_changes_by_ticket[$T->nsident]) ||
              !in_array($csm_rev,
                $repo_changes_by_ticket[$T->nsident][$csm_repo])) {
            $repo_changes_by_ticket[$T->nsident][$csm_repo][] = $csm_rev;
          }
        }
      }
      $plain .= "\n" . $comment . "\n";
    }

    /* bundle up the changesets along with the ticket changes */
    if (isset($repo_changes_by_ticket[$T->nsident])) {
      foreach ($repo_changes_by_ticket[$T->nsident] as $reponame => $revlist) {
        $plain .= "\nChanges in $reponame:\n";

        $repo = $repos_by_name[$reponame];
        /* TODO: 'compress' a run of changes so that we show the net change */

        // all affected files in this repo for this notification
        $files = array();
        $ent_by_rev = array();

        foreach ($revlist as $rev) {
          list($ent) = $repo->history(null, 1, 'rev', $rev);
          $ent_by_rev[$rev] = $ent;
          $projects[$repo->projectFromPath($ent->files)]++;
          foreach ($ent->files as $file) {
            $files[$file] = "$file->status $file->name";
          }
        }

        $plain .= "  Affected files:\n   " . join("\n   ", $files) . "\n";

        $too_big = false;
        foreach ($ent_by_rev as $ent) {
          $plain .= "\n[$ent->rev] by $ent->changeby\n";
          $plain .= MTrackConfig::get('core', 'weburl')
            . "changeset.php/$reponame/$ent->rev\n\n";

          if (strlen($plain) < $MAX_DIFF) {
            foreach ($ent->files as $file) {
              $diff = stream_get_contents($repo->diff($file, $ent->rev));
              if (strlen($plain) + strlen($diff) < $MAX_DIFF) {
                $plain .= $diff . "\n";
              } else {
                $too_big = true;
              }
            }
          } else {
            $too_big = true;
          }
        }
        if ($too_big) {
          $plain .= "  ** Diff exceeds configured limit\n";
        }
      }
    }
    $plain .= "\n" . MTrackConfig::get('core', 'weburl')
            . "ticket.php/$T->nsident\n";

    $to = preg_split("/(\s+|\s*[,;]\s*)/", $T->cc);
    $fromuser = major_contributor($contributors);
    $U = get_user_info($fromuser);
    $from = "From: <$U[email]> \"$U[fullname]\"\n";
    $to[$U['email']] = $U['email'];
    $repto = array("<$U[email]> \"$U[fullname]\"");
    foreach ($contributors as $c => $count) {
      if ($c !== $fromuser) {
        $U = get_user_info($c);
        if (strlen($U['email'])) {
          $repto[] = "<$U[email]> \"$U[fullname]\"";
          $to[$U['email']] = $U['email'];
        }
      }
    }
    if (count($repto) > 1) {
      $repto = "Reply-To: " . join(", ", $repto) . "\n";
    } else {
      $repto = '';
    }
    $pnames = array();
    if (count($projects) == 0) {
      $projects[1] = 1; // force in a default project
    }
    foreach ($projects as $pid => $count) {
      $P = $proj_by_id[$pid];
      if (is_object($P)) {
        $pnames[$P->shortname] = $P->shortname;
        $to[$P->notifyemail] = $P->notifyemail;
      }
    }
    natsort($pnames);

    // compute base message-id
    $mid = $T->tid . '@' . php_uname('n');
    // if one of the changesets we saw in this run matches the created cid
    // then we are the initial message
    $is_initial = false;
    foreach ($by_object['ticket'][$T->tid] as $cid) {
      if ($cid == $T->created) {
        $is_initial = true;
      }
    }
    if ($is_initial) {
      $mid = "Message-ID: <$mid>\n";
    } else {
      $mid = "Message-ID: <$T->updated.$mid>\n" .
             "In-Reply-To: <$mid>\n" .
             "References: <$mid>\n";
    }

    $phdrs = "X-mtrack-project-list: " . join(' ', $pnames) . "\n";
    foreach ($pnames as $p) {
      $phdrs .= "X-mtrack-project-$p: $p\n";
      $phdrs .= "X-mtrack-project: $p\n";
    }

    foreach ($to as $_ => $recip) {
      if (!strlen($recip)) {
        unset($to[$_]);
      }
    }

    $mail = $from . $repto . $mid .
      "To: " . join(', ', $to) . "\n" .
      'Subject: =?UTF-8?B?' . base64_encode('[' . join(' ', $pnames) . "] #$T->nsident $T->summary ($T->status $T->classification)") . '?=' . "\n" .
      "MIME-Version: 1.0\n" .
      "Content-Type: text/plain; charset=utf-8\n" .
      "X-mtrack-ticket: $T->nsident\n" .
      $phdrs .
      "\n" .
      $plain
      ;

    $reciplist = array();
    foreach ($to as $recip) {
      $reciplist[] = escapeshellarg($recip);
    }
    $reciplist = join(' ', $reciplist);
    if ($DEBUG) {
      echo "would mail: $reciplist\n$mail\n";
    } else {
      $pipe = popen("/usr/sbin/sendmail $reciplist", 'w');
      fwrite($pipe, $mail);
      pclose($pipe);
    }
  }

  if (count($batch_tkt) > 0) { // FIXME: implement batched ticket overview
    $plain = '';
    echo "\nBatched ticket notifications\n";
    foreach ($batch_tkt as $tid => $cslist) {
      $T = $tkt_by_tid[$tid];
      echo "  $T->nsident";
    }
    echo "\n";
  }
}

if (isset($by_object['changeset'])) 
foreach ($by_object['changeset'] as $reponame => $revlist) {
  $repo = $repos_by_name[$reponame];
  foreach ($revlist as $rev => $ent) {
    $plain = "Changes in $reponame [$ent->rev] by $ent->changeby:\n";
    $plain .= MTrackConfig::get('core', 'weburl')
      . "changeset.php/$reponame/$ent->rev\n\n";
    $plain .= "\n" . $ent->changelog . "\n\n";
    $flist = array();
    foreach ($ent->files as $file) {
      $flist[] = "$file->status $file->name";
    }
    $plain .= "  Affected files:\n   " . join("\n   ", $flist) . "\n\n";

    if (strlen($plain) < $MAX_DIFF) {
      foreach ($ent->files as $file) {
        $diff = stream_get_contents($repo->diff($file, $ent->rev));
        if (strlen($plain) + strlen($diff) < $MAX_DIFF) {
          $plain .= $diff . "\n";
        } else {
          $too_big = true;
        }
      }
    } else {
      $too_big = true;
    }
    if ($too_big) {
      $plain .= "  ** Diff exceeds configured limit\n";
    }

    $U = get_user_info($ent->changeby);
    $from = "From: <$U[email]> \"$U[fullname]\"\n";
    $pid = $repo->projectFromPath($ent->files);
    if ($pid === null) $pid = 1;
    $P = $proj_by_id[$pid];

    $mail = $from . 
      "To: $P->notifyemail\n" .
      'Subject: =?UTF-8?B?' . base64_encode("[$P->shortname] $reponame commit [$ent->rev]") . '?=' . "\n" .
      "MIME-Version: 1.0\n" .
      "Content-Type: text/plain; charset=utf-8\n" .
      "\n" .
      $plain
      ;

    if ($DEBUG) {
      echo "Would mail: $P->notifyemail\n";
      echo $mail . "\n";
    } else {
      $pipe = popen("/usr/sbin/sendmail " .
                escapeshellarg($P->notifyemail), 'w');
      fwrite($pipe, $mail);
      pclose($pipe);
    }
  }
}

if (isset($by_object['wiki'])) {
  foreach ($by_object['wiki'] as $pagename => $cslist) {
    //echo "Wiki $pagename\n";
    // TODO: implement wiki page change notification
  }
}

if (isset($by_object['milestone'])) {
  foreach ($by_object['milestone'] as $mname => $cslist) {
    //echo "Milestone $mname\n";
    // TODO: implement milestone change notification
  }
}

if (!$DEBUG) {
  // Now we are done, update the last run time
  $last_change = end($changes);
  $db->beginTransaction();
  $db->exec("delete from last_notification");
  $t = MTrackDB::unixtime($LATEST);
  echo "updating last run to $t $LATEST\n";
  $db->exec("insert into last_notification (last_run) values ('$t')");
  $db->commit();
}