version 1.19, 2002/07/01 18:23:00
|
version 1.69, 2006/09/26 15:15:19
|
Line 26
|
Line 26
|
# |
# |
# http://www.lon-capa.org/ |
# http://www.lon-capa.org/ |
# |
# |
# YEAR=2001 |
|
# 04/14/2001, 04/16/2001 Scott Harrison |
|
# |
|
# YEAR=2002 |
|
# 05/11/2002 Scott Harrison |
|
# |
|
### |
### |
|
|
# This script goes through a LON-CAPA resource |
=pod |
# directory and gathers metadata. |
|
# The metadata is entered into a SQL database. |
=head1 NAME |
|
|
|
B<searchcat.pl> - put authoritative filesystem data into sql database. |
|
|
|
=head1 SYNOPSIS |
|
|
|
Ordinarily this script is to be called from a loncapa cron job |
|
(CVS source location: F<loncapa/loncom/cron/loncapa>; typical |
|
filesystem installation location: F</etc/cron.d/loncapa>). |
|
|
|
Here is the cron job entry. |
|
|
|
C<# Repopulate and refresh the metadata database used for the search catalog.> |
|
C<10 1 * * 7 www /home/httpd/perl/searchcat.pl> |
|
|
|
This script only allows itself to be run as the user C<www>. |
|
|
|
=head1 DESCRIPTION |
|
|
|
This script goes through a loncapa resource directory and gathers metadata. |
|
The metadata is entered into a SQL database. |
|
|
|
This script also does general database maintenance such as reformatting |
|
the C<loncapa:metadata> table if it is deprecated. |
|
|
|
This script evaluates dynamic metadata from the authors' |
|
F<nohist_resevaldata.db> database file in order to store it in MySQL. |
|
|
|
This script is playing an increasingly important role for a loncapa |
|
library server. The proper operation of this script is critical for a smooth |
|
and correct user experience. |
|
|
|
=cut |
|
|
|
use strict; |
|
use DBI; |
use lib '/home/httpd/lib/perl/'; |
use lib '/home/httpd/lib/perl/'; |
use LONCAPA::Configuration; |
use LONCAPA::lonmetadata; |
|
|
|
use Getopt::Long; |
use IO::File; |
use IO::File; |
use HTML::TokeParser; |
use HTML::TokeParser; |
use DBI; |
use GDBM_File; |
|
use POSIX qw(strftime mktime); |
|
|
|
use Apache::lonnet(); |
|
|
my @metalist; |
use File::Find; |
# ----------------- 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 $perlvarref=LONCAPA::Configuration::read_conf('loncapa_apache.conf', |
|
'loncapa.conf'); |
|
my %perlvar=%{$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 $perlvar{'lonRole'} eq 'library'; |
# Set up configuration options |
|
my ($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', |
|
); |
|
|
|
my %newnames; |
|
# new table names - append pid to have unique temporary tables |
|
foreach my $key (keys(%oldnames)) { |
|
$newnames{$key} = 'new'.$oldnames{$key}.$$; |
|
} |
|
|
|
# |
|
# Only run if machine is a library server |
|
exit if ($Apache::lonnet::perlvar{'lonRole'} ne 'library'); |
|
# |
|
# 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()."===="); |
|
|
|
|
|
if ($debug) { |
|
&log(0,'simulating') if ($simulate); |
|
&log(0,'only processing user '.$oneuser) if ($oneuser); |
|
&log(0,'verbosity level = '.$verbose); |
|
} |
|
# |
|
# Connect to database |
my $dbh; |
my $dbh; |
# ------------------------------------- Make sure that database can be accessed |
if (! ($dbh = DBI->connect("DBI:mysql:loncapa","www",$Apache::lonnet::perlvar{'lonSqlAccess'}, |
{ |
{ RaiseError =>0,PrintError=>0}))) { |
unless ( |
&log(0,"Cannot connect to database!"); |
$dbh = DBI->connect("DBI:mysql:loncapa","www",$perlvar{'lonSqlAccess'},{ RaiseError =>0,PrintError=>0}) |
die "MySQL Error: Cannot connect to database!\n"; |
) { |
} |
print "Cannot connect to database!\n"; |
# This can return an error and still be okay, so we do not bother checking. |
exit; |
# (perhaps it should be more robust and check for specific errors) |
} |
foreach my $key (keys(%newnames)) { |
my $make_metadata_table = "CREATE TABLE IF NOT EXISTS metadata (". |
if ($newnames{$key} ne '') { |
"title TEXT, author TEXT, subject TEXT, url TEXT, keywords TEXT, ". |
$dbh->do('DROP TABLE IF EXISTS '.$newnames{$key}); |
"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), ". |
# Create the new metadata and portfolio tables |
"FULLTEXT idx_version (version), FULLTEXT idx_notes (notes), ". |
foreach my $key (keys(%newnames)) { |
"FULLTEXT idx_abstract (abstract), FULLTEXT idx_mime (mime), ". |
if ($newnames{$key} ne '') { |
"FULLTEXT idx_language (language), FULLTEXT idx_owner (owner), ". |
my $request = |
"FULLTEXT idx_copyright (copyright)) TYPE=MYISAM"; |
&LONCAPA::lonmetadata::create_metadata_storage($newnames{$key},$oldnames{$key}); |
# It would sure be nice to have some logging mechanism. |
$dbh->do($request); |
$dbh->do($make_metadata_table); |
if ($dbh->err) { |
} |
$dbh->disconnect(); |
|
&log(0,"MySQL Error Create: ".$dbh->errstr); |
# ------------------------------------------------------------- get .meta files |
die $dbh->errstr; |
opendir(RESOURCES,"$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}"); |
} |
my @homeusers=grep |
} |
{&ishome("$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$_")} |
} |
grep {!/^\.\.?$/} readdir(RESOURCES); |
|
closedir RESOURCES; |
# |
foreach my $user (@homeusers) { |
# find out which users we need to examine |
&find("$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$user"); |
my @domains = sort(&Apache::lonnet::current_machine_domains()); |
} |
&log(9,'domains ="'.join('","',@domains).'"'); |
|
|
# -- process each file to get metadata and put into search catalog SQL database |
foreach my $dom (@domains) { |
# Also, check to see if already there. |
&log(9,'domain = '.$dom); |
# I could just delete (without searching first), but this works for now. |
opendir(RESOURCES,"$Apache::lonnet::perlvar{'lonDocRoot'}/res/$dom"); |
foreach my $m (@metalist) { |
my @homeusers = |
my $ref=&metadata($m); |
grep { |
my $m2='/res/'.&declutter($m); |
&ishome("$Apache::lonnet::perlvar{'lonDocRoot'}/res/$dom/$_"); |
$m2=~s/\.meta$//; |
} grep { |
my $q2="select * from metadata where url like binary '$m2'"; |
!/^\.\.?$/; |
my $sth = $dbh->prepare($q2); |
} readdir(RESOURCES); |
$sth->execute(); |
closedir RESOURCES; |
my $r1=$sth->fetchall_arrayref; |
&log(5,'users = '.$dom.':'.join(',',@homeusers)); |
if (@$r1) { |
# |
$sth=$dbh->prepare("delete from metadata where url like binary '$m2'"); |
if ($oneuser) { |
$sth->execute(); |
@homeusers=($oneuser); |
} |
} |
$sth=$dbh->prepare('insert into metadata values ('. |
# |
'"'.delete($ref->{'title'}).'"'.','. |
# Loop through the users |
'"'.delete($ref->{'author'}).'"'.','. |
foreach my $user (@homeusers) { |
'"'.delete($ref->{'subject'}).'"'.','. |
&log(0,"=== User: ".$user); |
'"'.$m2.'"'.','. |
&process_dynamic_metadata($user,$dom); |
'"'.delete($ref->{'keywords'}).'"'.','. |
# |
'"'.'current'.'"'.','. |
# Use File::Find to get the files we need to read/modify |
'"'.delete($ref->{'notes'}).'"'.','. |
find( |
'"'.delete($ref->{'abstract'}).'"'.','. |
{preprocess => \&only_meta_files, |
'"'.delete($ref->{'mime'}).'"'.','. |
#wanted => \&print_filename, |
'"'.delete($ref->{'language'}).'"'.','. |
#wanted => \&log_metadata, |
'"'.sqltime(delete($ref->{'creationdate'})).'"'.','. |
wanted => \&process_meta_file, |
'"'.sqltime(delete($ref->{'lastrevisiondate'})).'"'.','. |
no_chdir => 1, |
'"'.delete($ref->{'owner'}).'"'.','. |
}, join('/',($Apache::lonnet::perlvar{'lonDocRoot'},'res',$dom,$user)) ); |
'"'.delete($ref->{'copyright'}).'"'.')'); |
} |
$sth->execute(); |
# Search for public portfolio files |
} |
my %portusers; |
|
if ($oneuser) { |
# ----------------------------------------------------------- Clean up database |
%portusers = ( |
# Need to, perhaps, remove stale SQL database records. |
$oneuser => '', |
# ... not yet implemented |
); |
|
} else { |
|
my $dir = $Apache::lonnet::perlvar{lonUsersDir}.'/'.$dom; |
|
&descend_tree($dir,0,\%portusers); |
|
} |
|
foreach my $uname (keys(%portusers)) { |
|
my $urlstart = '/uploaded/'.$dom.'/'.$uname; |
|
my $pathstart = &propath($dom,$uname).'/userfiles'; |
|
my $is_course = &check_for_course($dom,$uname); |
|
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') { |
|
&process_portfolio_access_data($url,$access{$file}); |
|
} |
|
&process_portfolio_metadata($url,$fullpath,$is_course,$dom, |
|
$uname,$group); |
|
} |
|
} |
|
} |
|
|
|
# |
|
# Rename the tables |
|
if (! $simulate) { |
|
foreach my $key (keys(%oldnames)) { |
|
if (($oldnames{$key} ne '') && ($newnames{$key} ne '')) { |
|
$dbh->do('DROP TABLE IF EXISTS '.$oldnames{$key}); |
|
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(); |
|
|
|
exit 0; |
|
|
|
## |
|
## Status logging routine. Inputs: $level, $message |
|
## |
|
## $level 0 should be used for normal output and error messages |
|
## |
|
## $message does not need to end with \n. In the case of errors |
|
## the message should contain as much information as possible to |
|
## help in diagnosing the problem. |
|
## |
|
sub log { |
|
my ($level,$message)=@_; |
|
$level = 0 if (! defined($level)); |
|
if ($verbose >= $level) { |
|
print LOG $message.$/; |
|
} |
|
} |
|
|
|
sub descend_tree { |
|
my ($dir,$depth,$alldomusers) = @_; |
|
if (-d $dir) { |
|
opendir(DIR,$dir); |
|
my @contents = grep(!/^\./,readdir(DIR)); |
|
closedir(DIR); |
|
$depth ++; |
|
foreach my $item (@contents) { |
|
if ($depth < 4) { |
|
&descend_tree($dir.'/'.$item,$depth,$alldomusers); |
|
} else { |
|
if (-e $dir.'/'.$item.'/file_permissions.db') { |
|
|
|
$$alldomusers{$item} = ''; |
|
} |
|
} |
|
} |
|
} |
|
} |
|
|
|
sub check_for_course { |
|
my ($cdom,$cnum) = @_; |
|
my %courses = &Apache::lonnet::courseiddump($cdom,'.',1,'.','.',$cnum,undef, |
|
undef,'.'); |
|
if (exists($courses{$cdom.'_'.$cnum})) { |
|
return 1; |
|
} |
|
return 0; |
|
} |
|
|
# --------------------------------------------------- Close database connection |
|
$dbh->disconnect; |
|
|
|
# ---------------------------------------------------------------- Get metadata |
sub process_portfolio_access_data { |
# significantly altered from subroutine present in lonnet |
my ($url,$access_hash) = @_; |
|
foreach my $key (keys(%{$access_hash})) { |
|
my $acc_data; |
|
$acc_data->{url} = $url; |
|
$acc_data->{keynum} = $key; |
|
my ($num,$scope,$end,$start) = |
|
($key =~ /^([^:]+):([a-z]+)_(\d*)_?(\d*)$/); |
|
$acc_data->{scope} = $scope; |
|
if ($end != 0) { |
|
$acc_data->{end} = &sqltime($end); |
|
} |
|
$acc_data->{start} = &sqltime($start); |
|
if (! $simulate) { |
|
my ($count,$err) = |
|
&LONCAPA::lonmetadata::store_metadata($dbh, |
|
$newnames{'access'}, |
|
'portfolio_access',$acc_data); |
|
if ($err) { |
|
&log(0,"MySQL Error Insert: ".$err); |
|
} |
|
if ($count < 1) { |
|
&log(0,"Unable to insert record into MySQL database for $url"); |
|
} |
|
} |
|
} |
|
} |
|
|
|
sub process_portfolio_metadata { |
|
my ($url,$fullpath,$is_course,$dom,$uname,$group) = @_; |
|
my ($ref,$crs,$addedfields) = &portfolio_metadata($fullpath,$dom,$uname, |
|
$group); |
|
&getfiledates($ref,$fullpath); |
|
if ($is_course) { |
|
$ref->{'groupname'} = $group; |
|
} |
|
my %Data; |
|
if (ref($ref) eq 'HASH') { |
|
%Data = %{$ref}; |
|
} |
|
%Data = ( |
|
%Data, |
|
'url'=>$url, |
|
'version'=>'current', |
|
); |
|
if (! $simulate) { |
|
my ($count,$err) = |
|
&LONCAPA::lonmetadata::store_metadata($dbh, |
|
$newnames{'portfolio'}, |
|
'portfolio_metadata',\%Data); |
|
if ($err) { |
|
&log(0,"MySQL Error Insert: ".$err); |
|
} |
|
if ($count < 1) { |
|
&log(0,"Unable to insert record into MySQL portfolio_metadata database table for $url"); |
|
} |
|
if (ref($addedfields) eq 'HASH') { |
|
if (keys(%{$addedfields}) > 0) { |
|
foreach my $key (keys(%{$addedfields})) { |
|
my $added_data = { |
|
'url' => $url, |
|
'field' => $key, |
|
'value' => $addedfields->{$key}, |
|
'courserestricted' => $crs, |
|
}; |
|
($count,$err) = &LONCAPA::lonmetadata::store_metadata($dbh, |
|
$newnames{'addedfields'}, |
|
'portfolio_addedfields', |
|
$added_data); |
|
if ($err) { |
|
&log(0,"MySQL Error Insert: ".$err); |
|
} |
|
if ($count < 1) { |
|
&log(0,"Unable to insert record into MySQL portfolio_addedfields database table for url = $url and field = $key"); |
|
} |
|
} |
|
} |
|
} |
|
} |
|
return; |
|
} |
|
|
|
######################################################## |
|
######################################################## |
|
### ### |
|
### 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; |
|
} |
|
|
|
## |
|
## 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); |
|
} |
|
&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 $metastring=&getfile($Apache::lonnet::perlvar{'lonDocRoot'}.'/res/'.$filename); |
|
return undef if (! defined($metastring)); |
|
my $parser=HTML::TokeParser->new(\$metastring); |
|
my $token; |
|
while ($token=$parser->get_token) { |
|
if ($token->[0] eq 'S') { |
|
my $entry=$token->[1]; |
|
my $unikey=$entry; |
|
if (defined($token->[2]->{'part'})) { |
|
$unikey.='_'.$token->[2]->{'part'}; |
|
} |
|
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; |
|
} |
|
|
|
############################################################### |
|
############################################################### |
|
### ### |
|
### &portfolio_metadata($filepath,$dom,$uname,$group) ### |
|
### Retrieve metadata for the given file ### |
|
### Returns array - ### |
|
### contains reference to metadatahash and ### |
|
### optional reference to addedfields hash ### |
|
### ### |
|
############################################################### |
|
############################################################### |
|
sub portfolio_metadata { |
|
my ($fullpath,$dom,$uname,$group)=@_; |
|
my ($mime) = ( $fullpath=~/\.(\w+)$/ ); |
|
my %metacache=(); |
|
if ($fullpath !~ /\.meta$/) { |
|
$fullpath .= '.meta'; |
|
} |
|
my (@standard_fields,%addedfields); |
|
my $colsref = |
|
$LONCAPA::lonmetadata::Portfolio_metadata_table_description; |
|
if (ref($colsref) eq 'ARRAY') { |
|
my @columns = @{$colsref}; |
|
foreach my $coldata (@columns) { |
|
push(@standard_fields,$coldata->{'name'}); |
|
} |
|
} |
|
my $metastring=&getfile($fullpath); |
|
if (! defined($metastring)) { |
|
$metacache{'keys'}= 'owner,domain,mime'; |
|
$metacache{'owner'} = $uname.':'.$dom; |
|
$metacache{'domain'} = $dom; |
|
$metacache{'mime'} = $mime; |
|
if (defined($group)) { |
|
$metacache{'keys'} .= ',courserestricted'; |
|
$metacache{'courserestricted'} = 'course.'.$dom.'_'.$uname; |
|
} |
|
} else { |
my $parser=HTML::TokeParser->new(\$metastring); |
my $parser=HTML::TokeParser->new(\$metastring); |
my $token; |
my $token; |
while ($token=$parser->get_token) { |
while ($token=$parser->get_token) { |
if ($token->[0] eq 'S') { |
if ($token->[0] eq 'S') { |
my $entry=$token->[1]; |
my $entry=$token->[1]; |
my $unikey=$entry; |
if ($metacache{'keys'}) { |
if (defined($token->[2]->{'part'})) { |
$metacache{'keys'}.=','.$entry; |
$unikey.='_'.$token->[2]->{'part'}; |
} else { |
} |
$metacache{'keys'}=$entry; |
if (defined($token->[2]->{'name'})) { |
} |
$unikey.='_'.$token->[2]->{'name'}; |
my $value = $parser->get_text('/'.$entry); |
} |
if (!grep(/^\Q$entry\E$/,@standard_fields)) { |
if ($metacache{$uri.'keys'}) { |
my $clean_value = lc($value); |
$metacache{$uri.'keys'}.=','.$unikey; |
$clean_value =~ s/\s/_/g; |
} else { |
if ($clean_value ne $entry) { |
$metacache{$uri.'keys'}=$unikey; |
if (defined($addedfields{$entry})) { |
} |
$addedfields{$entry} .=','.$value; |
map { |
} else { |
$metacache{$uri.''.$unikey.'.'.$_}=$token->[2]->{$_}; |
$addedfields{$entry} = $value; |
} @{$token->[3]}; |
} |
unless ( |
} |
$metacache{$uri.''.$unikey}=$parser->get_text('/'.$entry) |
} else { |
) { $metacache{$uri.''.$unikey}= |
$metacache{$entry} = $value; |
$metacache{$uri.''.$unikey.'.default'}; |
} |
} |
} |
} |
} # End of ($token->[0] eq 'S') |
} |
|
} |
} |
return \%metacache; |
if (keys(%addedfields) > 0) { |
|
foreach my $key (sort keys(%addedfields)) { |
|
$metacache{'addedfieldnames'} .= $key.','; |
|
$metacache{'addedfieldvalues'} .= $addedfields{$key}.'&&&'; |
|
} |
|
$metacache{'addedfieldnames'} =~ s/,$//; |
|
$metacache{'addedfieldvalues'} =~ s/\&\&\&$//; |
|
if ($metacache{'keys'}) { |
|
$metacache{'keys'}.=',addedfieldnames'; |
|
} else { |
|
$metacache{'keys'}='addedfieldnames'; |
|
} |
|
$metacache{'keys'}.=',addedfieldvalues'; |
|
} |
|
return (\%metacache,$metacache{'courserestricted'},\%addedfields); |
} |
} |
|
|
# ------------------------------------------------------------ Serves up a file |
## |
# returns either the contents of the file or a -1 |
## &getfile($filename) |
|
## Slurps up an entire file into a scalar. |
|
## Returns undef if the file does not exist |
sub getfile { |
sub getfile { |
my $file=shift; |
my $file = shift(); |
if (! -e $file ) { return -1; }; |
if (! -e $file ) { |
my $fh=IO::File->new($file); |
return undef; |
my $a=''; |
} |
while (<$fh>) { $a .=$_; } |
my $fh=IO::File->new($file); |
return $a |
my $contents = ''; |
|
while (<$fh>) { |
|
$contents .= $_; |
|
} |
|
return $contents; |
} |
} |
|
|
# ------------------------------------------------------------- Declutters URLs |
## |
sub declutter { |
## &getfiledates() |
my $thisfn=shift; |
## Converts creationdate and modifieddates to SQL format |
$thisfn=~s/^$perlvar{'lonDocRoot'}//; |
## Applies stat() to file to retrieve dates if missing |
$thisfn=~s/^\///; |
sub getfiledates { |
$thisfn=~s/^res\///; |
my ($ref,$target) = @_; |
return $thisfn; |
if (! defined($ref->{'creationdate'}) || |
|
$ref->{'creationdate'} =~ /^\s*$/) { |
|
$ref->{'creationdate'} = (stat($target))[9]; |
|
} |
|
if (! defined($ref->{'lastrevisiondate'}) || |
|
$ref->{'lastrevisiondate'} =~ /^\s*$/) { |
|
$ref->{'lastrevisiondate'} = (stat($target))[9]; |
|
} |
|
$ref->{'creationdate'} = &sqltime($ref->{'creationdate'}); |
|
$ref->{'lastrevisiondate'} = &sqltime($ref->{'lastrevisiondate'}); |
|
} |
|
|
|
######################################################## |
|
######################################################## |
|
### ### |
|
### 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; |
|
$file=~/\.(\w+)$/; |
|
my $ext=lc($1); |
|
$countext{$ext}++; |
|
} |
|
|
|
sub write_type_count { |
|
open(RESCOUNT,'>/home/httpd/html/lon-status/rescount.txt'); |
|
while (my ($extension,$count) = each(%countext)) { |
|
print RESCOUNT $extension.'='.$count.'&'; |
|
} |
|
print RESCOUNT 'time='.time."\n"; |
|
close(RESCOUNT); |
|
} |
|
|
|
} # end of scope for %countext |
|
|
|
{ |
|
|
|
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); |
} |
} |
|
|
# --------------------------------------- Is this the home server of an author? |
} # end of scope for %copyrights |
# (copied from lond, modification of the return value) |
|
|
######################################################## |
|
######################################################## |
|
### ### |
|
### 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\/$2/; |
Line 216 sub ishome {
|
Line 884 sub ishome {
|
} |
} |
} |
} |
|
|
# -------------------------------------------- Return path to profile directory |
## |
# (copied from lond) |
## &propath($udom,$uname) |
|
## Returns the path to the users LON-CAPA directory |
|
## (copied from lond) |
sub propath { |
sub propath { |
my ($udom,$uname)=@_; |
my ($udom,$uname)=@_; |
$udom=~s/\W//g; |
$udom=~s/\W//g; |
$uname=~s/\W//g; |
$uname=~s/\W//g; |
my $subdir=$uname.'__'; |
my $subdir=$uname.'__'; |
$subdir =~ s/(.)(.)(.).*/$1\/$2\/$3/; |
$subdir =~ s/(.)(.)(.).*/$1\/$2\/$3/; |
my $proname="$perlvar{'lonUsersDir'}/$udom/$subdir/$uname"; |
my $proname="$Apache::lonnet::perlvar{'lonUsersDir'}/$udom/$subdir/$uname"; |
return $proname; |
return $proname; |
} |
} |
|
|
# ---------------------------- convert 'time' format into a datetime sql format |
## |
|
## &sqltime($timestamp) |
|
## |
|
## Convert perl $timestamp to MySQL time. MySQL expects YYYY-MM-DD HH:MM:SS |
|
## |
sub sqltime { |
sub sqltime { |
my ($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst) = |
my ($time) = @_; |
localtime(@_[0]); |
my $mysqltime; |
$mon++; $year+=1900; |
if ($time =~ |
return "$year-$mon-$mday $hour:$min:$sec"; |
/(\d+)-(\d+)-(\d+) # YYYY-MM-DD |
|
\s # a space |
|
(\d+):(\d+):(\d+) # HH:MM::SS |
|
/x ) { |
|
# Some of the .meta files have the time in mysql |
|
# format already, so just make sure they are 0 padded and |
|
# pass them back. |
|
$mysqltime = sprintf('%04d-%02d-%02d %02d:%02d:%02d', |
|
$1,$2,$3,$4,$5,$6); |
|
} elsif ($time =~ /^\d+$/) { |
|
my @TimeData = gmtime($time); |
|
# Alter the month to be 1-12 instead of 0-11 |
|
$TimeData[4]++; |
|
# Alter the year to be from 0 instead of from 1900 |
|
$TimeData[5]+=1900; |
|
$mysqltime = sprintf('%04d-%02d-%02d %02d:%02d:%02d', |
|
@TimeData[5,4,3,2,1,0]); |
|
} elsif (! defined($time) || $time == 0) { |
|
$mysqltime = 0; |
|
} else { |
|
&log(0," sqltime:Unable to decode time ".$time); |
|
$mysqltime = 0; |
|
} |
|
return $mysqltime; |
|
} |
|
|
|
## |
|
## &declutter($filename) |
|
## Given a filename, returns a url for the filename. |
|
sub declutter { |
|
my $thisfn=shift; |
|
$thisfn=~s/^$Apache::lonnet::perlvar{'lonDocRoot'}//; |
|
$thisfn=~s/^\///; |
|
$thisfn=~s/^res\///; |
|
return $thisfn; |
|
} |
|
|
|
## |
|
## Escape / Unescape special characters |
|
sub unescape { |
|
my $str=shift; |
|
$str =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack("C",hex($1))/eg; |
|
return $str; |
|
} |
|
|
|
sub escape { |
|
my $str=shift; |
|
$str =~ s/(\W)/"%".unpack('H2',$1)/eg; |
|
return $str; |
} |
} |