version 1.44, 2003/12/23 15:47:26
|
version 1.84, 2016/01/31 21:25:49
|
Line 56 This script also does general database m
|
Line 56 This script also does general database m
|
the C<loncapa:metadata> table if it is deprecated. |
the C<loncapa:metadata> table if it is deprecated. |
|
|
This script evaluates dynamic metadata from the authors' |
This script evaluates dynamic metadata from the authors' |
F<nohist_resevaldata.db> database file in order to store it in MySQL, as |
F<nohist_resevaldata.db> database file in order to store it in MySQL. |
well as to compress the filesize (add up all "count"-type metadata). |
|
|
|
This script is playing an increasingly important role for a loncapa |
This script is playing an increasingly important role for a loncapa |
library server. The proper operation of this script is critical for a smooth |
library server. The proper operation of this script is critical for a smooth |
Line 65 and correct user experience.
|
Line 64 and correct user experience.
|
|
|
=cut |
=cut |
|
|
|
use strict; |
|
use DBI; |
use lib '/home/httpd/lib/perl/'; |
use lib '/home/httpd/lib/perl/'; |
use LONCAPA::Configuration; |
use LONCAPA::lonmetadata; |
|
use LONCAPA; |
|
use Getopt::Long; |
use IO::File; |
use IO::File; |
use HTML::TokeParser; |
use HTML::TokeParser; |
use DBI; |
|
use GDBM_File; |
use GDBM_File; |
use POSIX qw(strftime mktime); |
use POSIX qw(strftime mktime); |
|
use Mail::Send; |
|
use Apache::loncommon(); |
|
|
my @metalist; |
use Apache::lonnet(); |
|
|
$simplestatus=''; |
use File::Find; |
my %countext=(); |
|
|
# |
|
# Set up configuration options |
|
my ($simulate,$oneuser,$help,$verbose,$logfile,$debug); |
|
GetOptions ( |
|
'help' => \$help, |
|
'simulate' => \$simulate, |
|
'only=s' => \$oneuser, |
|
'verbose=s' => \$verbose, |
|
'debug' => \$debug, |
|
); |
|
|
|
if ($help) { |
|
print <<"ENDHELP"; |
|
$0 |
|
Rebuild and update the LON-CAPA metadata database. |
|
Options: |
|
-help Print this help |
|
-simulate Do not modify the database. |
|
-only=user Only compute for the given user. Implies -simulate |
|
-verbose=val Sets logging level, val must be a number |
|
-debug Turns on debugging output |
|
ENDHELP |
|
exit 0; |
|
} |
|
|
|
if (! defined($debug)) { |
|
$debug = 0; |
|
} |
|
|
|
if (! defined($verbose)) { |
|
$verbose = 0; |
|
} |
|
|
|
if (defined($oneuser)) { |
|
$simulate=1; |
|
} |
|
|
|
## |
|
## Use variables for table names so we can test this routine a little easier |
|
my %oldnames = ( |
|
'metadata' => 'metadata', |
|
'portfolio' => 'portfolio_metadata', |
|
'access' => 'portfolio_access', |
|
'addedfields' => 'portfolio_addedfields', |
|
'allusers' => 'allusers', |
|
); |
|
|
|
my %newnames; |
|
# new table names - append pid to have unique temporary tables |
|
foreach my $key (keys(%oldnames)) { |
|
$newnames{$key} = 'new'.$oldnames{$key}.$$; |
|
} |
|
|
sub writesimple { |
# |
open(SMP,'>/home/httpd/html/lon-status/mysql.txt'); |
# Only run if machine is a library server |
print SMP $simplestatus."\n"; |
exit if ($Apache::lonnet::perlvar{'lonRole'} ne 'library'); |
close(SMP); |
my $hostid = $Apache::lonnet::perlvar{'lonHostID'}; |
|
|
|
# |
|
# Make sure this process is running from user=www |
|
my $wwwid=getpwnam('www'); |
|
if ($wwwid!=$<) { |
|
my $emailto="$Apache::lonnet::perlvar{'lonAdmEMail'},$Apache::lonnet::perlvar{'lonSysEMail'}"; |
|
my $subj="LON: $Apache::lonnet::perlvar{'lonHostID'} User ID mismatch"; |
|
system("echo 'User ID mismatch. searchcat.pl must be run as user www.' |\ |
|
mail -s '$subj' $emailto > /dev/null"); |
|
exit 1; |
} |
} |
|
# |
|
# Let people know we are running |
|
open(LOG,'>>'.$Apache::lonnet::perlvar{'lonDaemons'}.'/logs/searchcat.log'); |
|
&log(0,'==== Searchcat Run '.localtime()."===="); |
|
|
sub writecount { |
|
open(RSMP,'>/home/httpd/html/lon-status/rescount.txt'); |
if ($debug) { |
foreach (keys %countext) { |
&log(0,'simulating') if ($simulate); |
print RSMP $_.'='.$countext{$_}.'&'; |
&log(0,'only processing user '.$oneuser) if ($oneuser); |
|
&log(0,'verbosity level = '.$verbose); |
|
} |
|
# |
|
# Connect to database |
|
my $dbh; |
|
if (! ($dbh = DBI->connect("DBI:mysql:loncapa","www",$Apache::lonnet::perlvar{'lonSqlAccess'}, |
|
{ RaiseError =>0,PrintError=>0}))) { |
|
&log(0,"Cannot connect to database!"); |
|
die "MySQL Error: Cannot connect to database!\n"; |
|
} |
|
# This can return an error and still be okay, so we do not bother checking. |
|
# (perhaps it should be more robust and check for specific errors) |
|
foreach my $key (keys(%newnames)) { |
|
if ($newnames{$key} ne '') { |
|
$dbh->do('DROP TABLE IF EXISTS '.$newnames{$key}); |
} |
} |
print RSMP 'time='.time."\n"; |
|
close(RSMP); |
|
} |
} |
|
|
sub count { |
# |
my $file=shift; |
# Create the new metadata, portfolio and allusers tables |
$file=~/\.(\w+)$/; |
foreach my $key (keys(%newnames)) { |
my $ext=lc($1); |
if ($newnames{$key} ne '') { |
if (defined($countext{$ext})) { |
my $request = |
$countext{$ext}++; |
&LONCAPA::lonmetadata::create_metadata_storage($newnames{$key},$oldnames{$key}); |
} else { |
$dbh->do($request); |
$countext{$ext}=1; |
if ($dbh->err) { |
|
$dbh->disconnect(); |
|
&log(0,"MySQL Error Create: ".$dbh->errstr); |
|
die $dbh->errstr; |
|
} |
} |
} |
} |
} |
# ----------------------------------------------------- Un-Escape Special Chars |
|
|
|
sub unescape { |
# |
my $str=shift; |
# find out which users we need to examine |
$str =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack("C",hex($1))/eg; |
my @domains = sort(&Apache::lonnet::current_machine_domains()); |
return $str; |
&log(9,'domains ="'.join('","',@domains).'"'); |
} |
|
|
foreach my $dom (@domains) { |
# -------------------------------------------------------- Escape Special Chars |
&log(9,'domain = '.$dom); |
|
opendir(RESOURCES,"$Apache::lonnet::perlvar{'lonDocRoot'}/res/$dom"); |
sub escape { |
my @homeusers = |
my $str=shift; |
grep { |
$str =~ s/(\W)/"%".unpack('H2',$1)/eg; |
&ishome("$Apache::lonnet::perlvar{'lonDocRoot'}/res/$dom/$_"); |
return $str; |
} grep { |
} |
!/^\.\.?$/; |
|
} readdir(RESOURCES); |
|
closedir RESOURCES; |
# ------------------------------------------- Code to evaluate dynamic metadata |
&log(5,'users = '.$dom.':'.join(',',@homeusers)); |
|
# |
sub dynamicmeta { |
if ($oneuser) { |
my $url=&declutter(shift); |
@homeusers=($oneuser); |
$url=~s/\.meta$//; |
} |
my %returnhash=(); |
|
my ($adomain,$aauthor)=($url=~/^(\w+)\/(\w+)\//); |
|
my $prodir=&propath($adomain,$aauthor); |
|
# Get metadata except counts |
|
if (tie(%evaldata,'GDBM_File', |
|
$prodir.'/nohist_resevaldata.db',&GDBM_READER(),0640)) { |
|
my %sum=(); |
|
my %cnt=(); |
|
my %concat=(); |
|
my %listitems=( |
|
'course' => 'add', |
|
'goto' => 'add', |
|
'comefrom' => 'add', |
|
'avetries' => 'avg', |
|
'stdno' => 'add', |
|
'difficulty' => 'avg', |
|
'clear' => 'avg', |
|
'technical' => 'avg', |
|
'helpful' => 'avg', |
|
'correct' => 'avg', |
|
'depth' => 'avg', |
|
'comments' => 'app', |
|
'usage' => 'cnt' |
|
); |
|
|
|
my $regexp=$url; |
|
$regexp=~s/(\W)/\\$1/g; |
|
$regexp='___'.$regexp.'___([a-z]+)$'; |
|
while (my ($esckey,$value)=each %evaldata) { |
|
$key=&unescape($esckey); |
|
if ($key=~/$regexp/) { |
|
my ($item,$purl,$cat)=split(/___/,$_); |
|
if (defined($cnt{$cat})) { $cnt{$cat}++; } else { $cnt{$cat}=1; } |
|
unless ($listitems{$cat} eq 'app') { |
|
if (defined($sum{$cat})) { |
|
$sum{$cat}+=$evaldata{$_}; |
|
$concat{$cat}.=','.$item; |
|
} else { |
|
$sum{$cat}=$evaldata{$_}; |
|
$concat{$cat}=$item; |
|
} |
|
} else { |
|
if (defined($sum{$cat})) { |
|
if ($evaldata{$_}) { |
|
$sum{$cat}.='<hr>'.$evaldata{$_}; |
|
} |
|
} else { |
|
$sum{$cat}=''.$evaldata{$_}; |
|
} |
|
} |
|
} |
|
} |
|
untie(%evaldata); |
|
} |
|
# construct the return hash for non-count data |
|
my %returnhash=(); |
|
while ($_=each(%cnt)) { |
|
if ($listitems{$_} eq 'avg') { |
|
$returnhash{$_}=int(($sum{$_}/$cnt{$_})*100.0+0.5)/100.0; |
|
} elsif ($listitems{$_} eq 'cnt') { |
|
$returnhash{$_}=$cnt{$_}; |
|
} else { |
|
$returnhash{$_}=$sum{$_}; |
|
} |
|
$returnhash{$_.'_list'}=$concat{$_}; |
|
} |
|
# get count |
|
if (tie(%evaldata,'GDBM_File', |
|
$prodir.'/nohist_accesscount.db',&GDBM_READER(),0640)) { |
|
if (! exists($evaldata{$uri})) { |
|
$returnhash{'count'}='Not Available'; |
|
} else { |
|
$returnhash{'count'}=$evaldata{$uri}; |
|
} |
|
untie %evaldata; |
|
} |
|
return %returnhash; |
|
} |
|
|
|
# ----------------- Code to enable 'find' subroutine listing of the .meta files |
|
require "find.pl"; |
|
sub wanted { |
|
(($dev,$ino,$mode,$nlink,$uid,$gid) = lstat($_)) && |
|
-f _ && |
|
/^.*\.meta$/ && !/^.+\.\d+\.[^\.]+\.meta$/ && |
|
push(@metalist,"$dir/$_"); |
|
} |
|
|
|
# --------------- Read loncapa_apache.conf and loncapa.conf and get variables |
|
my $perlvarref=LONCAPA::Configuration::read_conf('loncapa.conf'); |
|
my %perlvar=%{$perlvarref}; |
|
undef $perlvarref; # remove since sensitive and not needed |
|
delete $perlvar{'lonReceipt'}; # remove since sensitive and not needed |
|
|
|
# ------------------------------------- Only run if machine is a library server |
# |
exit unless $perlvar{'lonRole'} eq 'library'; |
# Loop through the users |
|
foreach my $user (@homeusers) { |
|
&log(0,"=== User: ".$user); |
|
&process_dynamic_metadata($user,$dom); |
|
# |
|
# Use File::Find to get the files we need to read/modify |
|
find( |
|
{preprocess => \&only_meta_files, |
|
#wanted => \&print_filename, |
|
#wanted => \&log_metadata, |
|
wanted => \&process_meta_file, |
|
no_chdir => 1, |
|
}, join('/',($Apache::lonnet::perlvar{'lonDocRoot'},'res',$dom,$user)) ); |
|
} |
|
# Search for all users and public portfolio files |
|
my (%allusers,%portusers,%courses); |
|
if ($oneuser) { |
|
%portusers = ( |
|
$oneuser => '', |
|
); |
|
%allusers = ( |
|
$oneuser => '', |
|
); |
|
%courses = &courseiddump($dom,'.',1,'.','.',$oneuser,undef, |
|
undef,'.'); |
|
} else { |
|
# get courseIDs for domain on current machine |
|
%courses=&Apache::lonnet::courseiddump($dom,'.',1,'.','.','.',1,[$hostid],'.'); |
|
my $dir = $Apache::lonnet::perlvar{lonUsersDir}.'/'.$dom; |
|
&descend_tree($dom,$dir,0,\%portusers,\%allusers); |
|
} |
|
foreach my $uname (keys(%portusers)) { |
|
my $urlstart = '/uploaded/'.$dom.'/'.$uname; |
|
my $pathstart = &propath($dom,$uname).'/userfiles'; |
|
my $is_course = ''; |
|
if (exists($courses{$dom.'_'.$uname})) { |
|
$is_course = 1; |
|
} |
|
my $curr_perm = &Apache::lonnet::get_portfile_permissions($dom,$uname); |
|
my %access = &Apache::lonnet::get_access_controls($curr_perm); |
|
foreach my $file (keys(%access)) { |
|
my ($group,$url,$fullpath); |
|
if ($is_course) { |
|
($group, my ($path)) = ($file =~ /^(\w+)(\/.+)$/); |
|
$fullpath = $pathstart.'/groups/'.$group.'/portfolio'.$path; |
|
$url = $urlstart.'/groups/'.$group.'/portfolio'.$path; |
|
} else { |
|
$fullpath = $pathstart.'/portfolio'.$file; |
|
$url = $urlstart.'/portfolio'.$file; |
|
} |
|
if (ref($access{$file}) eq 'HASH') { |
|
my %portaccesslog = |
|
&LONCAPA::lonmetadata::process_portfolio_access_data($dbh, |
|
$simulate,\%newnames,$url,$fullpath,$access{$file}); |
|
&portfolio_logging(%portaccesslog); |
|
} |
|
my %portmetalog = &LONCAPA::lonmetadata::process_portfolio_metadata($dbh,$simulate,\%newnames,$url,$fullpath,$is_course,$dom,$uname,$group); |
|
&portfolio_logging(%portmetalog); |
|
} |
|
} |
|
my %duplicates; |
|
my %names_by_id = ( |
|
id => {}, |
|
clickers => {}, |
|
); |
|
my %ids_by_name = ( |
|
id => {}, |
|
clickers => {}, |
|
); |
|
my %idstodelete = ( |
|
id => {}, |
|
clickers => {}, |
|
); |
|
my %idstoadd = ( |
|
id => {}, |
|
clickers => {}, |
|
); |
|
my %namespace = ( |
|
id => 'ids', |
|
clickers => 'clickers', |
|
); |
|
my %idtext = ( |
|
id => 'employee/student IDs', |
|
clickers => 'clicker IDs', |
|
); |
|
unless ($simulate || $oneuser) { |
|
foreach my $key ('id','clickers') { |
|
my $hashref = &tie_domain_hash($dom,$namespace{$key},&GDBM_WRCREAT()); |
|
if (ref($hashref) eq 'HASH') { |
|
while (my ($id,$unamestr) = each(%{$hashref}) ) { |
|
$id = &unescape($id); |
|
$unamestr = &unescape($unamestr); |
|
if ($key eq 'clickers') { |
|
my @unames = split(/,/,$unamestr); |
|
foreach my $uname (@unames) { |
|
push(@{$ids_by_name{$key}{$uname}},$id); |
|
} |
|
$names_by_id{$key}{$id} = $unamestr; |
|
} else { |
|
$names_by_id{$key}{$id} = $unamestr; |
|
push(@{$ids_by_name{$key}{$unamestr}},$id); |
|
} |
|
} |
|
&untie_domain_hash($hashref); |
|
} |
|
} |
|
} |
|
# Update allusers |
|
foreach my $uname (keys(%allusers)) { |
|
next if (exists($courses{$dom.'_'.$uname})); |
|
my %userdata = |
|
&Apache::lonnet::get('environment',['firstname','lastname', |
|
'middlename','generation','id','permanentemail','clickers'], |
|
$dom,$uname); |
|
unless ($simulate || $oneuser) { |
|
foreach my $key ('id','clickers') { |
|
my %addid = (); |
|
if ($userdata{$key} ne '') { |
|
my $idfromenv = $userdata{$key}; |
|
if ($key eq 'id') { |
|
$idfromenv=~tr/A-Z/a-z/; |
|
$addid{$idfromenv} = 1; |
|
} else { |
|
$idfromenv =~ s/^\s+//; |
|
$idfromenv =~ s/\s+$//; |
|
map { $addid{$_} = 1; } split(/,/,$idfromenv); |
|
} |
|
} |
|
if (ref($ids_by_name{$key}{$uname}) eq 'ARRAY') { |
|
if (scalar(@{$ids_by_name{$key}{$uname}}) > 1) { |
|
&log(0,"Multiple $idtext{$key} found in $namespace{$key}.db for $uname:$dom -- ". |
|
join(', ',@{$ids_by_name{$key}{$uname}})); |
|
} |
|
foreach my $id (@{$ids_by_name{$key}{$uname}}) { |
|
if ($addid{$id}) { |
|
delete($addid{$id}); |
|
} else { |
|
if ($key eq 'id') { |
|
$idstodelete{$key}{$id} = $uname; |
|
} else { |
|
$idstodelete{$key}{$id} .= $uname.','; |
|
} |
|
} |
|
} |
|
} |
|
if (keys(%addid)) { |
|
foreach my $id (keys(%addid)) { |
|
if ($key eq 'id') { |
|
if (exists($idstoadd{$key}{$id})) { |
|
push(@{$duplicates{$id}},$uname); |
|
} else { |
|
$idstoadd{$key}{$id} = $uname; |
|
} |
|
} else { |
|
$idstoadd{$key}{$id} .= $uname.','; |
|
} |
|
} |
|
} |
|
} |
|
} |
|
|
# ----------------------------- Make sure this process is running from user=www |
$userdata{'username'} = $uname; |
|
$userdata{'domain'} = $dom; |
|
my %alluserslog = |
|
&LONCAPA::lonmetadata::process_allusers_data($dbh,$simulate, |
|
\%newnames,$uname,$dom,\%userdata); |
|
foreach my $item (keys(%alluserslog)) { |
|
&log(0,$alluserslog{$item}); |
|
} |
|
} |
|
unless ($simulate || $oneuser) { |
|
foreach my $key ('id','clickers') { |
|
if (keys(%{$idstodelete{$key}}) > 0) { |
|
my %resulthash; |
|
if ($key eq 'id') { |
|
%resulthash = &Apache::lonnet::iddel($dom,$idstodelete{$key},$hostid,$namespace{$key}); |
|
} else { |
|
foreach my $delid (sort(keys(%{$idstodelete{$key}}))) { |
|
$idstodelete{$key}{$delid} =~ s/,$//; |
|
} |
|
%resulthash = &Apache::lonnet::iddel($dom,$idstodelete{$key},$hostid,$namespace{$key}); |
|
} |
|
if ($resulthash{$hostid} eq 'ok') { |
|
foreach my $id (sort(keys(%{$idstodelete{$key}}))) { |
|
&log(0,"Record deleted from $namespace{$key}.db for $dom -- $id => ".$idstodelete{$key}{$id}); |
|
} |
|
} else { |
|
&log(0,"Error: '$resulthash{$hostid}' occurred when attempting to delete records from $namespace{$key}.db for $dom"); |
|
} |
|
} |
|
if (keys(%{$idstoadd{$key}}) > 0) { |
|
my $idmessage = ''; |
|
my %newids; |
|
if ($key eq 'id') { |
|
foreach my $addid (sort(keys(%{$idstoadd{$key}}))) { |
|
if ((exists($names_by_id{$key}{$addid})) && ($names_by_id{$key}{$addid} ne $idstoadd{$key}{$addid}) && !($idstodelete{$key}{$addid})) { |
|
&log(0,"Two usernames associated with a single ID $addid in domain: $dom: $names_by_id{$key}{$addid} (current) and $idstoadd{$key}{$addid}\n"); |
|
$idmessage .= "$addid,$names_by_id{$key}{$addid},$idstoadd{$key}{$addid}\n"; |
|
} else { |
|
$newids{$addid} = $idstoadd{$key}{$addid}; |
|
} |
|
} |
|
} else { |
|
foreach my $addid (sort(keys(%{$idstoadd{$key}}))) { |
|
$idstoadd{$key}{$addid} =~ s/,$//; |
|
$newids{$addid} = $idstoadd{$key}{$addid}; |
|
} |
|
} |
|
if (keys(%newids) > 0) { |
|
my $putresult; |
|
if ($key eq 'clickers') { |
|
$putresult = &Apache::lonnet::updateclickers($dom,'add',\%newids,$hostid); |
|
} else { |
|
$putresult = &Apache::lonnet::put_dom($namespace{$key},\%newids,$dom,$hostid); |
|
} |
|
if ($putresult eq 'ok') { |
|
foreach my $id (sort(keys(%newids))) { |
|
&log(0,"Record added to $namespace{$key}.db for $dom -- $id => ".$newids{$id}); |
|
} |
|
} else { |
|
&log(0,"Error: '$putresult' occurred when attempting to add records to $namespace{$key}.db for $dom"); |
|
} |
|
} |
|
if ($idmessage) { |
|
my $to = &Apache::loncommon::build_recipient_list(undef,'idconflictsmail',$dom); |
|
if ($to ne '') { |
|
my $msg = new Mail::Send; |
|
$msg->to($to); |
|
$msg->subject('LON-CAPA studentIDs conflict'); |
|
my $lonhost = $Apache::lonnet::perlvar{'lonHostID'}; |
|
my $hostname = &Apache::lonnet::hostname($lonhost); |
|
my $replytoaddress = 'do-not-reply@'.$hostname; |
|
$msg->add('Reply-to',$replytoaddress); |
|
$msg->add('From','www@'.$hostname); |
|
$msg->add('Content-type','text/plain; charset=UTF-8'); |
|
if (my $fh = $msg->open()) { |
|
print $fh |
|
'The following IDs are used for more than one user in your domain:'."\n". |
|
'Each row contains: Student/Employee ID, Current username in ids.db file, '. |
|
'Additional username'."\n\n". |
|
$idmessage; |
|
$fh->close; |
|
} |
|
} |
|
} |
|
} |
|
} |
|
if (keys(%duplicates) > 0) { |
|
foreach my $id (sort(keys(%duplicates))) { |
|
if (ref($duplicates{$id}) eq 'ARRAY') { |
|
&log(0,"Duplicate IDs found for entries to add to ids.db in $dom -- $id => ".join(',',@{$duplicates{$id}})); |
|
} |
|
} |
|
} |
|
} |
|
} |
|
|
my $wwwid=getpwnam('www'); |
# |
if ($wwwid!=$<) { |
# Rename the tables |
$emailto="$perlvar{'lonAdmEMail'},$perlvar{'lonSysEMail'}"; |
if (! $simulate) { |
$subj="LON: $perlvar{'lonHostID'} User ID mismatch"; |
foreach my $key (keys(%oldnames)) { |
system("echo 'User ID mismatch. searchcat.pl must be run as user www.' |\ |
if (($oldnames{$key} ne '') && ($newnames{$key} ne '')) { |
mailto $emailto -s '$subj' > /dev/null"); |
$dbh->do('DROP TABLE IF EXISTS '.$oldnames{$key}); |
exit 1; |
if (! $dbh->do('RENAME TABLE '.$newnames{$key}.' TO '.$oldnames{$key})) { |
|
&log(0,"MySQL Error Rename: ".$dbh->errstr); |
|
die $dbh->errstr; |
|
} else { |
|
&log(1,"MySQL table rename successful for $key."); |
|
} |
|
} |
|
} |
} |
} |
|
if (! $dbh->disconnect) { |
|
&log(0,"MySQL Error Disconnect: ".$dbh->errstr); |
|
die $dbh->errstr; |
|
} |
|
## |
|
## Finished! |
|
&log(0,"==== Searchcat completed ".localtime()." ===="); |
|
close(LOG); |
|
|
|
&write_type_count(); |
|
&write_copyright_count(); |
|
|
# ---------------------------------------------------------- We are in business |
exit 0; |
|
|
open(LOG,'>'.$perlvar{'lonDaemons'}.'/logs/searchcat.log'); |
## |
print LOG '==== Searchcat Run '.localtime()."====\n\n"; |
## Status logging routine. Inputs: $level, $message |
$simplestatus='time='.time.'&'; |
## |
my $dbh; |
## $level 0 should be used for normal output and error messages |
# ------------------------------------- Make sure that database can be accessed |
## |
{ |
## $message does not need to end with \n. In the case of errors |
unless ( |
## the message should contain as much information as possible to |
$dbh = DBI->connect("DBI:mysql:loncapa","www",$perlvar{'lonSqlAccess'},{ RaiseError =>0,PrintError=>0}) |
## help in diagnosing the problem. |
) { |
## |
print LOG "Cannot connect to database!\n"; |
sub log { |
$simplestatus.='mysql=defunct'; |
my ($level,$message)=@_; |
&writesimple(); |
$level = 0 if (! defined($level)); |
exit; |
if ($verbose >= $level) { |
} |
print LOG $message.$/; |
|
} |
my $make_metadata_table = "CREATE TABLE IF NOT EXISTS metadata (". |
} |
"title TEXT, author TEXT, subject TEXT, url TEXT, keywords TEXT, ". |
|
"version TEXT, notes TEXT, abstract TEXT, mime TEXT, language TEXT, ". |
|
"creationdate DATETIME, lastrevisiondate DATETIME, owner TEXT, ". |
|
"copyright TEXT, FULLTEXT idx_title (title), ". |
|
"FULLTEXT idx_author (author), FULLTEXT idx_subject (subject), ". |
|
"FULLTEXT idx_url (url), FULLTEXT idx_keywords (keywords), ". |
|
"FULLTEXT idx_version (version), FULLTEXT idx_notes (notes), ". |
|
"FULLTEXT idx_abstract (abstract), FULLTEXT idx_mime (mime), ". |
|
"FULLTEXT idx_language (language), FULLTEXT idx_owner (owner), ". |
|
"FULLTEXT idx_copyright (copyright)) TYPE=MYISAM"; |
|
# It would sure be nice to have some logging mechanism. |
|
$dbh->do($make_metadata_table); |
|
} |
|
|
|
# ------------------------------------------------------------- get .meta files |
|
opendir(RESOURCES,"$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}"); |
|
my @homeusers = grep { |
|
&ishome("$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$_") |
|
} grep {!/^\.\.?$/} readdir(RESOURCES); |
|
closedir RESOURCES; |
|
|
|
# |
sub portfolio_logging { |
# Create the statement handlers we need |
my (%portlog) = @_; |
my $delete_sth = $dbh->prepare |
foreach my $key (keys(%portlog)) { |
("DELETE FROM metadata WHERE url LIKE BINARY ?"); |
if (ref($portlog{$key}) eq 'HASH') { |
|
foreach my $item (keys(%{$portlog{$key}})) { |
my $insert_sth = $dbh->prepare |
&log(0,$portlog{$key}{$item}); |
("INSERT INTO metadata VALUES (". |
} |
"?,". # title |
} |
"?,". # author |
} |
"?,". # subject |
} |
"?,". # m2??? |
|
"?,". # version |
|
"?,". # current |
|
"?,". # notes |
|
"?,". # abstract |
|
"?,". # mime |
|
"?,". # language |
|
"?,". # creationdate |
|
"?,". # revisiondate |
|
"?,". # owner |
|
"?)" # copyright |
|
); |
|
|
|
foreach my $user (@homeusers) { |
|
print LOG "\n=== User: ".$user."\n\n"; |
|
|
|
my $prodir=&propath($perlvar{'lonDefDomain'},$user); |
|
# Use find.pl |
|
undef @metalist; |
|
@metalist=(); |
|
&find("$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$user"); |
|
# -- process each file to get metadata and put into search catalog SQL |
|
# database. Also, check to see if already there. |
|
# I could just delete (without searching first), but this works for now. |
|
foreach my $m (@metalist) { |
|
print LOG "- ".$m."\n"; |
|
my $ref=&metadata($m); |
|
my $m2='/res/'.&declutter($m); |
|
$m2=~s/\.meta$//; |
|
&dynamicmeta($m2); |
|
if ($ref->{'obsolete'}) { print LOG "obsolete\n"; next; } |
|
if ($ref->{'copyright'} eq 'private') { print LOG "private\n"; next; } |
|
&count($m2); |
|
$delete_sth->execute($m2); |
|
$insert_sth->execute($ref->{'title'}, |
|
$ref->{'author'}, |
|
$ref->{'subject'}, |
|
$m2, |
|
$ref->{'keywords'}, |
|
'current', |
|
$ref->{'notes'}, |
|
$ref->{'abstract'}, |
|
$ref->{'mime'}, |
|
$ref->{'language'}, |
|
sqltime($ref->{'creationdate'}), |
|
sqltime($ref->{'lastrevisiondate'}), |
|
$ref->{'owner'}, |
|
$ref->{'copyright'}); |
|
# if ($dbh->err()) { |
|
# print STDERR "Error:".$dbh->errstr()."\n"; |
|
# } |
|
$ref = undef; |
|
} |
|
|
|
# --------------------------------------------------- Clean up database |
|
# Need to, perhaps, remove stale SQL database records. |
|
# ... not yet implemented |
|
|
|
} |
|
# --------------------------------------------------- Close database connection |
|
$dbh->disconnect; |
|
print LOG "\n==== Searchcat completed ".localtime()." ====\n"; |
|
close(LOG); |
|
&writesimple(); |
|
&writecount(); |
|
exit 0; |
|
|
|
|
sub descend_tree { |
|
my ($dom,$dir,$depth,$allportusers,$alldomusers) = @_; |
|
if (-d $dir) { |
|
opendir(DIR,$dir); |
|
my @contents = grep(!/^\./,readdir(DIR)); |
|
closedir(DIR); |
|
$depth ++; |
|
foreach my $item (@contents) { |
|
if (($depth < 4) && (length($item) == 1)) { |
|
&descend_tree($dom,$dir.'/'.$item,$depth,$allportusers,$alldomusers); |
|
} else { |
|
if (-e $dir.'/'.$item.'/file_permissions.db') { |
|
$$allportusers{$item} = ''; |
|
} |
|
if (-e $dir.'/'.$item.'/passwd') { |
|
$$alldomusers{$item} = ''; |
|
} |
|
} |
|
} |
|
} |
|
} |
|
|
|
######################################################## |
|
######################################################## |
|
### ### |
|
### File::Find support routines ### |
|
### ### |
|
######################################################## |
|
######################################################## |
|
## |
|
## &only_meta_files |
|
## |
|
## Called by File::Find. |
|
## Takes a list of files/directories in and returns a list of files/directories |
|
## to search. |
|
sub only_meta_files { |
|
my @PossibleFiles = @_; |
|
my @ChosenFiles; |
|
foreach my $file (@PossibleFiles) { |
|
if ( ($file =~ /\.meta$/ && # Ends in meta |
|
$file !~ /\.\d+\.[^\.]+\.meta$/ # is not for a prior version |
|
) || (-d $File::Find::dir."/".$file )) { # directories are okay |
|
# but we do not want /. or /.. |
|
push(@ChosenFiles,$file); |
|
} |
|
} |
|
return @ChosenFiles; |
|
} |
|
|
|
## |
|
## |
|
## Debugging routines, use these for 'wanted' in the File::Find call |
|
## |
|
sub print_filename { |
|
my ($file) = $_; |
|
my $fullfilename = $File::Find::name; |
|
if ($debug) { |
|
if (-d $file) { |
|
&log(5," Got directory ".$fullfilename); |
|
} else { |
|
&log(5," Got file ".$fullfilename); |
|
} |
|
} |
|
$_=$file; |
|
} |
|
|
# ============================================================================= |
sub log_metadata { |
|
my ($file) = $_; |
|
my $fullfilename = $File::Find::name; |
|
return if (-d $fullfilename); # No need to do anything here for directories |
|
if ($debug) { |
|
&log(6,$fullfilename); |
|
my $ref = &metadata($fullfilename); |
|
if (! defined($ref)) { |
|
&log(6," No data"); |
|
return; |
|
} |
|
while (my($key,$value) = each(%$ref)) { |
|
&log(6," ".$key." => ".$value); |
|
} |
|
&count_copyright($ref->{'copyright'}); |
|
} |
|
$_=$file; |
|
} |
|
|
# ---------------------------------------------------------------- Get metadata |
## |
# significantly altered from subroutine present in lonnet |
## process_meta_file |
|
## Called by File::Find. |
|
## Only input is the filename in $_. |
|
sub process_meta_file { |
|
my ($file) = $_; |
|
my $filename = $File::Find::name; # full filename |
|
return if (-d $filename); # No need to do anything here for directories |
|
# |
|
&log(3,$filename) if ($debug); |
|
# |
|
my $ref = &metadata($filename); |
|
# |
|
# $url is the original file url, not the metadata file |
|
my $target = $filename; |
|
$target =~ s/\.meta$//; |
|
my $url='/res/'.&declutter($target); |
|
&log(3," ".$url) if ($debug); |
|
# |
|
# Ignore some files based on their metadata |
|
if ($ref->{'obsolete'}) { |
|
&log(3,"obsolete") if ($debug); |
|
return; |
|
} |
|
&count_copyright($ref->{'copyright'}); |
|
if ($ref->{'copyright'} eq 'private') { |
|
&log(3,"private") if ($debug); |
|
return; |
|
} |
|
# |
|
# Find the dynamic metadata |
|
my %dyn; |
|
if ($url=~ m:/default$:) { |
|
$url=~ s:/default$:/:; |
|
&log(3,"Skipping dynamic data") if ($debug); |
|
} else { |
|
&log(3,"Retrieving dynamic data") if ($debug); |
|
%dyn=&get_dynamic_metadata($url); |
|
&count_type($url); |
|
} |
|
&LONCAPA::lonmetadata::getfiledates($ref,$target); |
|
# |
|
my %Data = ( |
|
%$ref, |
|
%dyn, |
|
'url'=>$url, |
|
'version'=>'current'); |
|
if (! $simulate) { |
|
my ($count,$err) = |
|
&LONCAPA::lonmetadata::store_metadata($dbh,$newnames{'metadata'}, |
|
'metadata',\%Data); |
|
if ($err) { |
|
&log(0,"MySQL Error Insert: ".$err); |
|
} |
|
if ($count < 1) { |
|
&log(0,"Unable to insert record into MySQL database for $url"); |
|
} |
|
} |
|
# |
|
# Reset $_ before leaving |
|
$_ = $file; |
|
} |
|
|
|
######################################################## |
|
######################################################## |
|
### ### |
|
### &metadata($uri) ### |
|
### Retrieve metadata for the given file ### |
|
### ### |
|
######################################################## |
|
######################################################## |
sub metadata { |
sub metadata { |
my ($uri,$what)=@_; |
my ($uri) = @_; |
my %metacache; |
my %metacache=(); |
$uri=&declutter($uri); |
$uri=&declutter($uri); |
my $filename=$uri; |
my $filename=$uri; |
$uri=~s/\.meta$//; |
$uri=~s/\.meta$//; |
$uri=''; |
$uri=''; |
unless ($metacache{$uri.'keys'}) { |
if ($filename !~ /\.meta$/) { |
unless ($filename=~/\.meta$/) { $filename.='.meta'; } |
$filename.='.meta'; |
my $metastring=&getfile($perlvar{'lonDocRoot'}.'/res/'.$filename); |
} |
my $parser=HTML::TokeParser->new(\$metastring); |
my $metastring = |
my $token; |
&LONCAPA::lonmetadata::getfile($Apache::lonnet::perlvar{'lonDocRoot'}.'/res/'.$filename); |
while ($token=$parser->get_token) { |
return undef if (! defined($metastring)); |
if ($token->[0] eq 'S') { |
my $parser=HTML::TokeParser->new(\$metastring); |
my $entry=$token->[1]; |
my $token; |
my $unikey=$entry; |
while ($token=$parser->get_token) { |
if (defined($token->[2]->{'part'})) { |
if ($token->[0] eq 'S') { |
$unikey.='_'.$token->[2]->{'part'}; |
my $entry=$token->[1]; |
} |
my $unikey=$entry; |
if (defined($token->[2]->{'name'})) { |
if (defined($token->[2]->{'part'})) { |
$unikey.='_'.$token->[2]->{'name'}; |
$unikey.='_'.$token->[2]->{'part'}; |
} |
|
if ($metacache{$uri.'keys'}) { |
|
$metacache{$uri.'keys'}.=','.$unikey; |
|
} else { |
|
$metacache{$uri.'keys'}=$unikey; |
|
} |
|
map { |
|
$metacache{$uri.''.$unikey.'.'.$_}=$token->[2]->{$_}; |
|
} @{$token->[3]}; |
|
unless ( |
|
$metacache{$uri.''.$unikey}=$parser->get_text('/'.$entry) |
|
) { $metacache{$uri.''.$unikey}= |
|
$metacache{$uri.''.$unikey.'.default'}; |
|
} |
|
} |
} |
} |
if (defined($token->[2]->{'name'})) { |
|
$unikey.='_'.$token->[2]->{'name'}; |
|
} |
|
if ($metacache{$uri.'keys'}) { |
|
$metacache{$uri.'keys'}.=','.$unikey; |
|
} else { |
|
$metacache{$uri.'keys'}=$unikey; |
|
} |
|
foreach ( @{$token->[3]}) { |
|
$metacache{$uri.''.$unikey.'.'.$_}=$token->[2]->{$_}; |
|
} |
|
if (! ($metacache{$uri.''.$unikey}=$parser->get_text('/'.$entry))){ |
|
$metacache{$uri.''.$unikey} = |
|
$metacache{$uri.''.$unikey.'.default'}; |
|
} |
|
} # End of ($token->[0] eq 'S') |
} |
} |
return \%metacache; |
return \%metacache; |
} |
} |
|
|
# ------------------------------------------------------------ Serves up a file |
######################################################## |
# returns either the contents of the file or a -1 |
######################################################## |
sub getfile { |
### ### |
|
### Dynamic Metadata ### |
|
### ### |
|
######################################################## |
|
######################################################## |
|
## |
|
## Dynamic metadata description (incomplete) |
|
## |
|
## For a full description of all fields, |
|
## see LONCAPA::lonmetadata |
|
## |
|
## Field Type |
|
##----------------------------------------------------------- |
|
## count integer |
|
## course integer |
|
## course_list comma separated list of course ids |
|
## avetries real |
|
## avetries_list comma separated list of real numbers |
|
## stdno real |
|
## stdno_list comma separated list of real numbers |
|
## usage integer |
|
## usage_list comma separated list of resources |
|
## goto scalar |
|
## goto_list comma separated list of resources |
|
## comefrom scalar |
|
## comefrom_list comma separated list of resources |
|
## difficulty real |
|
## difficulty_list comma separated list of real numbers |
|
## sequsage scalar |
|
## sequsage_list comma separated list of resources |
|
## clear real |
|
## technical real |
|
## correct real |
|
## helpful real |
|
## depth real |
|
## comments html of all the comments made |
|
## |
|
{ |
|
|
|
my %DynamicData; |
|
my %Counts; |
|
|
|
sub process_dynamic_metadata { |
|
my ($user,$dom) = @_; |
|
undef(%DynamicData); |
|
undef(%Counts); |
|
# |
|
my $prodir = &propath($dom,$user); |
|
# |
|
# Read in the dynamic metadata |
|
my %evaldata; |
|
if (! tie(%evaldata,'GDBM_File', |
|
$prodir.'/nohist_resevaldata.db',&GDBM_READER(),0640)) { |
|
return 0; |
|
} |
|
# |
|
%DynamicData = &LONCAPA::lonmetadata::process_reseval_data(\%evaldata); |
|
untie(%evaldata); |
|
$DynamicData{'domain'} = $dom; |
|
#print('user = '.$user.' domain = '.$dom.$/); |
|
# |
|
# Read in the access count data |
|
&log(7,'Reading access count data') if ($debug); |
|
my %countdata; |
|
if (! tie(%countdata,'GDBM_File', |
|
$prodir.'/nohist_accesscount.db',&GDBM_READER(),0640)) { |
|
return 0; |
|
} |
|
while (my ($key,$count) = each(%countdata)) { |
|
next if ($key !~ /^$dom/); |
|
$key = &unescape($key); |
|
&log(8,' Count '.$key.' = '.$count) if ($debug); |
|
$Counts{$key}=$count; |
|
} |
|
untie(%countdata); |
|
if ($debug) { |
|
&log(7,scalar(keys(%Counts)). |
|
" Counts read for ".$user."@".$dom); |
|
&log(7,scalar(keys(%DynamicData)). |
|
" Dynamic metadata read for ".$user."@".$dom); |
|
} |
|
# |
|
return 1; |
|
} |
|
|
|
sub get_dynamic_metadata { |
|
my ($url) = @_; |
|
$url =~ s:^/res/::; |
|
my %data = &LONCAPA::lonmetadata::process_dynamic_metadata($url, |
|
\%DynamicData); |
|
# find the count |
|
$data{'count'} = $Counts{$url}; |
|
# |
|
# Log the dynamic metadata |
|
if ($debug) { |
|
while (my($k,$v)=each(%data)) { |
|
&log(8," ".$k." => ".$v); |
|
} |
|
} |
|
return %data; |
|
} |
|
|
|
} # End of %DynamicData and %Counts scope |
|
|
|
######################################################## |
|
######################################################## |
|
### ### |
|
### Counts ### |
|
### ### |
|
######################################################## |
|
######################################################## |
|
{ |
|
|
|
my %countext; |
|
|
|
sub count_type { |
my $file=shift; |
my $file=shift; |
if (! -e $file ) { return -1; }; |
$file=~/\.(\w+)$/; |
my $fh=IO::File->new($file); |
my $ext=lc($1); |
my $a=''; |
$countext{$ext}++; |
while (<$fh>) { $a .=$_; } |
|
return $a; |
|
} |
} |
|
|
# ------------------------------------------------------------- Declutters URLs |
sub write_type_count { |
sub declutter { |
open(RESCOUNT,'>/home/httpd/html/lon-status/rescount.txt'); |
my $thisfn=shift; |
while (my ($extension,$count) = each(%countext)) { |
$thisfn=~s/^$perlvar{'lonDocRoot'}//; |
print RESCOUNT $extension.'='.$count.'&'; |
$thisfn=~s/^\///; |
} |
$thisfn=~s/^res\///; |
print RESCOUNT 'time='.time."\n"; |
return $thisfn; |
close(RESCOUNT); |
} |
} |
|
|
# --------------------------------------- Is this the home server of an author? |
} # end of scope for %countext |
# (copied from lond, modification of the return value) |
|
|
{ |
|
|
|
my %copyrights; |
|
|
|
sub count_copyright { |
|
$copyrights{@_[0]}++; |
|
} |
|
|
|
sub write_copyright_count { |
|
open(COPYCOUNT,'>/home/httpd/html/lon-status/copyrightcount.txt'); |
|
while (my ($copyright,$count) = each(%copyrights)) { |
|
print COPYCOUNT $copyright.'='.$count.'&'; |
|
} |
|
print COPYCOUNT 'time='.time."\n"; |
|
close(COPYCOUNT); |
|
} |
|
|
|
} # end of scope for %copyrights |
|
|
|
######################################################## |
|
######################################################## |
|
### ### |
|
### Miscellanous Utility Routines ### |
|
### ### |
|
######################################################## |
|
######################################################## |
|
## |
|
## &ishome($username) |
|
## Returns 1 if $username is a LON-CAPA author, 0 otherwise |
|
## (copied from lond, modification of the return value) |
sub ishome { |
sub ishome { |
my $author=shift; |
my $author=shift; |
$author=~s/\/home\/httpd\/html\/res\/([^\/]*)\/([^\/]*).*/$1\/$2/; |
$author=~s{/home/httpd/html/res/([^/]*)/([^/]*).*}{$1/$2}; |
my ($udom,$uname)=split(/\//,$author); |
my ($udom,$uname)=split(/\//,$author); |
my $proname=propath($udom,$uname); |
my $proname=propath($udom,$uname); |
if (-e $proname) { |
if (-e $proname) { |
Line 432 sub ishome {
|
Line 902 sub ishome {
|
} |
} |
} |
} |
|
|
# -------------------------------------------- Return path to profile directory |
## |
# (copied from lond) |
## &declutter($filename) |
sub propath { |
## Given a filename, returns a url for the filename. |
my ($udom,$uname)=@_; |
sub declutter { |
$udom=~s/\W//g; |
my $thisfn=shift; |
$uname=~s/\W//g; |
$thisfn=~s/^$Apache::lonnet::perlvar{'lonDocRoot'}//; |
my $subdir=$uname.'__'; |
$thisfn=~s/^\///; |
$subdir =~ s/(.)(.)(.).*/$1\/$2\/$3/; |
$thisfn=~s/^res\///; |
my $proname="$perlvar{'lonUsersDir'}/$udom/$subdir/$uname"; |
return $thisfn; |
return $proname; |
|
} |
|
|
|
# ---------------------------- convert 'time' format into a datetime sql format |
|
sub sqltime { |
|
my ($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst) = |
|
localtime(&unsqltime(@_[0])); |
|
$mon++; $year+=1900; |
|
return "$year-$mon-$mday $hour:$min:$sec"; |
|
} |
|
|
|
sub maketime { |
|
my %th=@_; |
|
return POSIX::mktime(($th{'seconds'},$th{'minutes'},$th{'hours'}, |
|
$th{'day'},$th{'month'}-1, |
|
$th{'year'}-1900,0,0,$th{'dlsav'})); |
|
} |
|
|
|
|
|
######################################### |
|
# |
|
# Retro-fixing of un-backward-compatible time format |
|
|
|
sub unsqltime { |
|
my $timestamp=shift; |
|
if ($timestamp=~/^(\d+)\-(\d+)\-(\d+)\s+(\d+)\:(\d+)\:(\d+)$/) { |
|
$timestamp=&maketime('year'=>$1,'month'=>$2,'day'=>$3, |
|
'hours'=>$4,'minutes'=>$5,'seconds'=>$6); |
|
} |
|
return $timestamp; |
|
} |
} |
|
|