version 1.44, 2003/12/23 15:47:26
|
version 1.55, 2004/04/08 15:57:32
|
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::Configuration; |
|
use LONCAPA::lonmetadata; |
|
|
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 File::Find; |
|
|
my @metalist; |
## |
|
## Use variables for table names so we can test this routine a little easier |
|
my $oldname = 'metadata'; |
|
my $newname = 'newmetadata'; |
|
|
$simplestatus=''; |
# |
my %countext=(); |
# Read loncapa_apache.conf and loncapa.conf |
|
my $perlvarref=LONCAPA::Configuration::read_conf('loncapa.conf'); |
sub writesimple { |
my %perlvar=%{$perlvarref}; |
open(SMP,'>/home/httpd/html/lon-status/mysql.txt'); |
undef $perlvarref; |
print SMP $simplestatus."\n"; |
delete $perlvar{'lonReceipt'}; # remove since sensitive (really?) & not needed |
close(SMP); |
# |
|
# Only run if machine is a library server |
|
exit if ($perlvar{'lonRole'} ne 'library'); |
|
# |
|
# Make sure this process is running from user=www |
|
my $wwwid=getpwnam('www'); |
|
if ($wwwid!=$<) { |
|
my $emailto="$perlvar{'lonAdmEMail'},$perlvar{'lonSysEMail'}"; |
|
my $subj="LON: $perlvar{'lonHostID'} User ID mismatch"; |
|
system("echo 'User ID mismatch. searchcat.pl must be run as user www.' |\ |
|
mailto $emailto -s '$subj' > /dev/null"); |
|
exit 1; |
|
} |
|
# |
|
# Let people know we are running |
|
open(LOG,'>'.$perlvar{'lonDaemons'}.'/logs/searchcat.log'); |
|
print LOG '==== Searchcat Run '.localtime()."====\n"; |
|
# |
|
# Connect to database |
|
my $dbh; |
|
if (! ($dbh = DBI->connect("DBI:mysql:loncapa","www",$perlvar{'lonSqlAccess'}, |
|
{ RaiseError =>0,PrintError=>0}))) { |
|
print LOG "Cannot connect to database!\n"; |
|
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) |
|
$dbh->do('DROP TABLE IF EXISTS '.$newname); |
|
# |
|
# Create the new table |
|
my $request = &LONCAPA::lonmetadata::create_metadata_storage($newname); |
|
$dbh->do($request); |
|
if ($dbh->err) { |
|
$dbh->disconnect(); |
|
print LOG "\nMySQL Error Create: ".$dbh->errstr."\n"; |
|
die $dbh->errstr; |
} |
} |
|
# |
|
# find out which users we need to examine |
|
opendir(RESOURCES,"$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}"); |
|
my @homeusers = |
|
grep { |
|
&ishome("$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$_"); |
|
} grep { |
|
!/^\.\.?$/; |
|
} readdir(RESOURCES); |
|
closedir RESOURCES; |
|
# |
|
# Loop through the users |
|
foreach my $user (@homeusers) { |
|
print LOG "=== User: ".$user."\n"; |
|
my $prodir=&propath($perlvar{'lonDefDomain'},$user); |
|
# |
|
# Use File::Find to get the files we need to read/modify |
|
find( |
|
{preprocess => \&only_meta_files, |
|
# wanted => \&print_filename, |
|
# wanted => \&log_metadata, |
|
wanted => \&process_meta_file, |
|
}, |
|
"$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$user"); |
|
} |
|
# |
|
# Rename the table |
|
$dbh->do('DROP TABLE IF EXISTS '.$oldname); |
|
if (! $dbh->do('RENAME TABLE '.$newname.' TO '.$oldname)) { |
|
print LOG "MySQL Error Rename: ".$dbh->errstr."\n"; |
|
die $dbh->errstr; |
|
} |
|
if (! $dbh->disconnect) { |
|
print LOG "MySQL Error Disconnect: ".$dbh->errstr."\n"; |
|
die $dbh->errstr; |
|
} |
|
## |
|
## Finished! |
|
print LOG "==== Searchcat completed ".localtime()." ====\n"; |
|
close(LOG); |
|
|
|
&write_type_count(); |
|
&write_copyright_count(); |
|
|
|
exit 0; |
|
|
sub writecount { |
######################################################## |
open(RSMP,'>/home/httpd/html/lon-status/rescount.txt'); |
######################################################## |
foreach (keys %countext) { |
### ### |
print RSMP $_.'='.$countext{$_}.'&'; |
### 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 $file )) { # directories are okay |
|
# but we do not want /. or /.. |
|
push(@ChosenFiles,$file); |
|
} |
} |
} |
print RSMP 'time='.time."\n"; |
return @ChosenFiles; |
close(RSMP); |
|
} |
} |
|
|
sub count { |
## |
my $file=shift; |
## |
$file=~/\.(\w+)$/; |
## Debugging routines, use these for 'wanted' in the File::Find call |
my $ext=lc($1); |
## |
if (defined($countext{$ext})) { |
sub print_filename { |
$countext{$ext}++; |
my ($file) = $_; |
|
my $fullfilename = $File::Find::name; |
|
if (-d $file) { |
|
print LOG " Got directory ".$fullfilename."\n"; |
} else { |
} else { |
$countext{$ext}=1; |
print LOG " Got file ".$fullfilename."\n"; |
} |
} |
|
$_=$file; |
} |
} |
# ----------------------------------------------------- Un-Escape Special Chars |
|
|
|
sub unescape { |
sub log_metadata { |
my $str=shift; |
my ($file) = $_; |
$str =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack("C",hex($1))/eg; |
my $fullfilename = $File::Find::name; |
return $str; |
return if (-d $fullfilename); # No need to do anything here for directories |
|
print LOG $fullfilename."\n"; |
|
my $ref=&metadata($fullfilename); |
|
if (! defined($ref)) { |
|
print LOG " No data\n"; |
|
return; |
|
} |
|
while (my($key,$value) = each(%$ref)) { |
|
print LOG " ".$key." => ".$value."\n"; |
|
} |
|
&count_copyright($ref->{'copyright'}); |
|
$_=$file; |
} |
} |
|
|
# -------------------------------------------------------- Escape Special Chars |
|
|
|
sub escape { |
## |
my $str=shift; |
## process_meta_file |
$str =~ s/(\W)/"%".unpack('H2',$1)/eg; |
## Called by File::Find. |
return $str; |
## Only input is the filename in $_. |
|
sub process_meta_file { |
|
my ($file) = $_; |
|
my $filename = $File::Find::name; |
|
return if (-d $filename); # No need to do anything here for directories |
|
# |
|
print LOG $filename."\n"; |
|
# |
|
my $ref=&metadata($filename); |
|
# |
|
# $url is the original file url, not the metadata file |
|
my $url='/res/'.&declutter($filename); |
|
$url=~s/\.meta$//; |
|
print LOG " ".$url."\n"; |
|
# |
|
# Ignore some files based on their metadata |
|
if ($ref->{'obsolete'}) { |
|
print LOG "obsolete\n"; |
|
return; |
|
} |
|
&count_copyright($ref->{'copyright'}); |
|
if ($ref->{'copyright'} eq 'private') { |
|
print LOG "private\n"; |
|
return; |
|
} |
|
# |
|
# Find the dynamic metadata |
|
my %dyn; |
|
if ($url=~ m:/default$:) { |
|
$url=~ s:/default$:/:; |
|
} else { |
|
# %dyn=&dynamicmeta($url); |
|
&count_type($url); |
|
} |
|
# |
|
$ref->{'creationdate'} = &sqltime($ref->{'creationdate'}); |
|
$ref->{'lastrevisiondate'} = &sqltime($ref->{'lastrevisiondate'}); |
|
my %Data = ( |
|
%$ref, |
|
%dyn, |
|
'url'=>$url, |
|
'version'=>'current'); |
|
my ($count,$err) = &LONCAPA::lonmetadata::store_metadata($dbh,$newname, |
|
\%Data); |
|
if ($err) { |
|
print LOG "\nMySQL Error Insert: ".$err."\n"; |
|
die $err; |
|
} |
|
if ($count < 1) { |
|
print LOG "Unable to insert record into MySQL database for $url\n"; |
|
die "Unable to insert record into MySQl database for $url"; |
|
} else { |
|
print LOG "Count = ".$count."\n"; |
|
} |
|
# |
|
# Reset $_ before leaving |
|
$_ = $file; |
} |
} |
|
|
|
######################################################## |
|
######################################################## |
|
### ### |
|
### &metadata($uri) ### |
|
### Retrieve metadata for the given file ### |
|
### ### |
|
######################################################## |
|
######################################################## |
|
sub metadata { |
|
my ($uri)=@_; |
|
my %metacache=(); |
|
$uri=&declutter($uri); |
|
my $filename=$uri; |
|
$uri=~s/\.meta$//; |
|
$uri=''; |
|
if ($filename !~ /\.meta$/) { |
|
$filename.='.meta'; |
|
} |
|
my $metastring=&getfile($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; |
|
} |
|
|
# ------------------------------------------- Code to evaluate dynamic metadata |
## |
|
## &getfile($filename) |
|
## Slurps up an entire file into a scalar. |
|
## Returns undef if the file does not exist |
|
sub getfile { |
|
my $file = shift(); |
|
if (! -e $file ) { |
|
return undef; |
|
} |
|
my $fh=IO::File->new($file); |
|
my $contents = ''; |
|
while (<$fh>) { |
|
$contents .= $_; |
|
} |
|
return $contents; |
|
} |
|
|
|
######################################################## |
|
######################################################## |
|
### ### |
|
### Dynamic Metadata ### |
|
### ### |
|
######################################################## |
|
######################################################## |
sub dynamicmeta { |
sub dynamicmeta { |
my $url=&declutter(shift); |
my $url = &declutter(shift()); |
$url=~s/\.meta$//; |
$url =~ s/\.meta$//; |
my %returnhash=(); |
my %data = ('count' => 0, |
my ($adomain,$aauthor)=($url=~/^(\w+)\/(\w+)\//); |
'course' => 0, |
my $prodir=&propath($adomain,$aauthor); |
'course_list' => '', |
# Get metadata except counts |
'avetries' => 'NULL', |
if (tie(%evaldata,'GDBM_File', |
'avetries_list' => '', |
$prodir.'/nohist_resevaldata.db',&GDBM_READER(),0640)) { |
'stdno' => 0, |
my %sum=(); |
'stdno_list' => '', |
my %cnt=(); |
'usage' => 0, |
my %concat=(); |
'usage_list' => '', |
my %listitems=( |
'goto' => 0, |
'course' => 'add', |
'goto_list' => '', |
'goto' => 'add', |
'comefrom' => 0, |
'comefrom' => 'add', |
'comefrom_list' => '', |
'avetries' => 'avg', |
'difficulty' => 'NULL', |
'stdno' => 'add', |
'difficulty_list' => '', |
'difficulty' => 'avg', |
'sequsage' => '0', |
'clear' => 'avg', |
'sequsage_list' => '', |
'technical' => 'avg', |
'clear' => 'NULL', |
'helpful' => 'avg', |
'technical' => 'NULL', |
'correct' => 'avg', |
'correct' => 'NULL', |
'depth' => 'avg', |
'helpful' => 'NULL', |
'comments' => 'app', |
'depth' => 'NULL', |
'usage' => 'cnt' |
'comments' => '', |
); |
); |
|
my ($dom,$auth)=($url=~/^(\w+)\/(\w+)\//); |
my $regexp=$url; |
my $prodir=&propath($dom,$auth); |
$regexp=~s/(\W)/\\$1/g; |
# |
$regexp='___'.$regexp.'___([a-z]+)$'; |
# Get metadata except counts |
while (my ($esckey,$value)=each %evaldata) { |
my %evaldata; |
$key=&unescape($esckey); |
if (! tie(%evaldata,'GDBM_File', |
if ($key=~/$regexp/) { |
$prodir.'/nohist_resevaldata.db',&GDBM_READER(),0640)) { |
my ($item,$purl,$cat)=split(/___/,$_); |
return (undef); |
if (defined($cnt{$cat})) { $cnt{$cat}++; } else { $cnt{$cat}=1; } |
} |
unless ($listitems{$cat} eq 'app') { |
my %sum=(); |
if (defined($sum{$cat})) { |
my %count=(); |
$sum{$cat}+=$evaldata{$_}; |
my %concat=(); |
$concat{$cat}.=','.$item; |
my %listitems=( |
} else { |
'course' => 'add', |
$sum{$cat}=$evaldata{$_}; |
'goto' => 'add', |
$concat{$cat}=$item; |
'comefrom' => 'add', |
} |
'avetries' => 'average', |
} else { |
'stdno' => 'add', |
if (defined($sum{$cat})) { |
'difficulty' => 'average', |
if ($evaldata{$_}) { |
'clear' => 'average', |
$sum{$cat}.='<hr>'.$evaldata{$_}; |
'technical' => 'average', |
} |
'helpful' => 'average', |
} else { |
'correct' => 'average', |
$sum{$cat}=''.$evaldata{$_}; |
'depth' => 'average', |
|
'comments' => 'append', |
|
'usage' => 'count' |
|
); |
|
# |
|
my $regexp=$url; |
|
$regexp=~s/(\W)/\\$1/g; |
|
$regexp='___'.$regexp.'___([a-z]+)$'; |
|
while (my ($esckey,$value)=each %evaldata) { |
|
my $key=&unescape($esckey); |
|
if ($key=~/$regexp/) { |
|
my ($item,$purl,$cat)=split(/___/,$key); |
|
$count{$cat}++; |
|
if ($listitems{$cat} ne 'append') { |
|
if (defined($sum{$cat})) { |
|
$sum{$cat}+=&unescape($value); |
|
$concat{$cat}.=','.$item; |
|
} else { |
|
$sum{$cat}=&unescape($value); |
|
$concat{$cat}=$item; |
|
} |
|
} else { |
|
if (defined($sum{$cat})) { |
|
if ($evaldata{$esckey}=~/\w/) { |
|
$sum{$cat}.='<hr />'.&unescape($evaldata{$esckey}); |
|
} |
|
} else { |
|
$sum{$cat}=''.&unescape($evaldata{$esckey}); |
} |
} |
} |
} |
} |
} |
} |
|
untie(%evaldata); |
|
} |
} |
# construct the return hash for non-count data |
untie(%evaldata); |
my %returnhash=(); |
# transfer gathered data to returnhash, calculate averages where applicable |
while ($_=each(%cnt)) { |
my %returnhash; |
if ($listitems{$_} eq 'avg') { |
while (my $cat=each(%count)) { |
$returnhash{$_}=int(($sum{$_}/$cnt{$_})*100.0+0.5)/100.0; |
if ($count{$cat} eq 'nan') { next; } |
} elsif ($listitems{$_} eq 'cnt') { |
if ($sum{$cat} eq 'nan') { next; } |
$returnhash{$_}=$cnt{$_}; |
if ($listitems{$cat} eq 'average') { |
} else { |
if ($count{$cat}) { |
$returnhash{$_}=$sum{$_}; |
$returnhash{$cat}=int(($sum{$cat}/$count{$cat})*100.0+0.5)/100.0; |
} |
} else { |
$returnhash{$_.'_list'}=$concat{$_}; |
$returnhash{$cat}='NULL'; |
|
} |
|
} elsif ($listitems{$cat} eq 'count') { |
|
$returnhash{$cat}=$count{$cat}; |
|
} else { |
|
$returnhash{$cat}=$sum{$cat}; |
|
} |
|
$returnhash{$cat.'_list'}=$concat{$cat}; |
} |
} |
# get count |
# |
if (tie(%evaldata,'GDBM_File', |
# get count |
|
if (tie(my %evaldata,'GDBM_File', |
$prodir.'/nohist_accesscount.db',&GDBM_READER(),0640)) { |
$prodir.'/nohist_accesscount.db',&GDBM_READER(),0640)) { |
if (! exists($evaldata{$uri})) { |
my $escurl=&escape($url); |
$returnhash{'count'}='Not Available'; |
if (! exists($evaldata{$escurl})) { |
|
$returnhash{'count'}=0; |
} else { |
} else { |
$returnhash{'count'}=$evaldata{$uri}; |
$returnhash{'count'}=$evaldata{$escurl}; |
} |
} |
untie %evaldata; |
untie %evaldata; |
} |
} |
return %returnhash; |
return %returnhash; |
} |
} |
|
|
# ----------------- Code to enable 'find' subroutine listing of the .meta files |
|
require "find.pl"; |
|
sub wanted { |
|
(($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 |
### Counts ### |
delete $perlvar{'lonReceipt'}; # remove since sensitive and not needed |
### ### |
|
######################################################## |
# ------------------------------------- Only run if machine is a library server |
######################################################## |
exit unless $perlvar{'lonRole'} eq 'library'; |
{ |
|
|
# ----------------------------- Make sure this process is running from user=www |
my %countext; |
|
|
my $wwwid=getpwnam('www'); |
sub count_type { |
if ($wwwid!=$<) { |
my $file=shift; |
$emailto="$perlvar{'lonAdmEMail'},$perlvar{'lonSysEMail'}"; |
$file=~/\.(\w+)$/; |
$subj="LON: $perlvar{'lonHostID'} User ID mismatch"; |
my $ext=lc($1); |
system("echo 'User ID mismatch. searchcat.pl must be run as user www.' |\ |
$countext{$ext}++; |
mailto $emailto -s '$subj' > /dev/null"); |
|
exit 1; |
|
} |
} |
|
|
|
sub write_type_count { |
# ---------------------------------------------------------- We are in business |
open(RESCOUNT,'>/home/httpd/html/lon-status/rescount.txt'); |
|
while (my ($extension,$count) = each(%countext)) { |
open(LOG,'>'.$perlvar{'lonDaemons'}.'/logs/searchcat.log'); |
print RESCOUNT $extension.'='.$count.'&'; |
print LOG '==== Searchcat Run '.localtime()."====\n\n"; |
} |
$simplestatus='time='.time.'&'; |
print RESCOUNT 'time='.time."\n"; |
my $dbh; |
close(RESCOUNT); |
# ------------------------------------- Make sure that database can be accessed |
|
{ |
|
unless ( |
|
$dbh = DBI->connect("DBI:mysql:loncapa","www",$perlvar{'lonSqlAccess'},{ RaiseError =>0,PrintError=>0}) |
|
) { |
|
print LOG "Cannot connect to database!\n"; |
|
$simplestatus.='mysql=defunct'; |
|
&writesimple(); |
|
exit; |
|
} |
|
|
|
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 |
} # end of scope for %countext |
opendir(RESOURCES,"$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}"); |
|
my @homeusers = grep { |
|
&ishome("$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$_") |
|
} grep {!/^\.\.?$/} readdir(RESOURCES); |
|
closedir RESOURCES; |
|
|
|
# |
|
# Create the statement handlers we need |
|
my $delete_sth = $dbh->prepare |
|
("DELETE FROM metadata WHERE url LIKE BINARY ?"); |
|
|
|
my $insert_sth = $dbh->prepare |
|
("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; |
|
|
|
|
|
|
{ |
|
|
# ============================================================================= |
my %copyrights; |
|
|
# ---------------------------------------------------------------- Get metadata |
sub count_copyright { |
# significantly altered from subroutine present in lonnet |
$copyrights{@_[0]}++; |
sub metadata { |
|
my ($uri,$what)=@_; |
|
my %metacache; |
|
$uri=&declutter($uri); |
|
my $filename=$uri; |
|
$uri=~s/\.meta$//; |
|
$uri=''; |
|
unless ($metacache{$uri.'keys'}) { |
|
unless ($filename=~/\.meta$/) { $filename.='.meta'; } |
|
my $metastring=&getfile($perlvar{'lonDocRoot'}.'/res/'.$filename); |
|
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; |
|
} |
|
map { |
|
$metacache{$uri.''.$unikey.'.'.$_}=$token->[2]->{$_}; |
|
} @{$token->[3]}; |
|
unless ( |
|
$metacache{$uri.''.$unikey}=$parser->get_text('/'.$entry) |
|
) { $metacache{$uri.''.$unikey}= |
|
$metacache{$uri.''.$unikey.'.default'}; |
|
} |
|
} |
|
} |
|
} |
|
return \%metacache; |
|
} |
} |
|
|
# ------------------------------------------------------------ Serves up a file |
sub write_copyright_count { |
# returns either the contents of the file or a -1 |
open(COPYCOUNT,'>/home/httpd/html/lon-status/copyrightcount.txt'); |
sub getfile { |
while (my ($copyright,$count) = each(%copyrights)) { |
my $file=shift; |
print COPYCOUNT $copyright.'='.$count.'&'; |
if (! -e $file ) { return -1; }; |
} |
my $fh=IO::File->new($file); |
print COPYCOUNT 'time='.time."\n"; |
my $a=''; |
close(COPYCOUNT); |
while (<$fh>) { $a .=$_; } |
|
return $a; |
|
} |
} |
|
|
# ------------------------------------------------------------- Declutters URLs |
} # end of scope for %copyrights |
sub declutter { |
|
my $thisfn=shift; |
|
$thisfn=~s/^$perlvar{'lonDocRoot'}//; |
|
$thisfn=~s/^\///; |
|
$thisfn=~s/^res\///; |
|
return $thisfn; |
|
} |
|
|
|
# --------------------------------------- Is this the home server of an author? |
######################################################## |
# (copied from lond, modification of the return value) |
######################################################## |
|
### ### |
|
### Miscellanous Utility Routines ### |
|
### ### |
|
######################################################## |
|
######################################################## |
|
## |
|
## &ishome($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\/$2/; |
Line 432 sub ishome {
|
Line 553 sub ishome {
|
} |
} |
} |
} |
|
|
# -------------------------------------------- Return path to profile directory |
## |
# (copied from lond) |
## &propath($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; |
Line 444 sub propath {
|
Line 567 sub propath {
|
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(&unsqltime(@_[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; |
|
$mysqltime = sprintf('%04d-%02d-%02d %02d:%02d:%02d', |
|
@TimeData[5,4,3,2,1,0]); |
|
} else { |
|
print LOG " Unable to decode time ".$time."\n"; |
|
$mysqltime = 0; |
|
} |
|
return $mysqltime; |
} |
} |
|
|
sub maketime { |
## |
my %th=@_; |
## &declutter($filename) |
return POSIX::mktime(($th{'seconds'},$th{'minutes'},$th{'hours'}, |
## Given a filename, returns a url for the filename. |
$th{'day'},$th{'month'}-1, |
sub declutter { |
$th{'year'}-1900,0,0,$th{'dlsav'})); |
my $thisfn=shift; |
|
$thisfn=~s/^$perlvar{'lonDocRoot'}//; |
|
$thisfn=~s/^\///; |
|
$thisfn=~s/^res\///; |
|
return $thisfn; |
} |
} |
|
|
|
## |
######################################### |
## Escape / Unescape special characters |
# |
sub unescape { |
# Retro-fixing of un-backward-compatible time format |
my $str=shift; |
|
$str =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack("C",hex($1))/eg; |
sub unsqltime { |
return $str; |
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; |
|
} |
} |
|
|
|
sub escape { |
|
my $str=shift; |
|
$str =~ s/(\W)/"%".unpack('H2',$1)/eg; |
|
return $str; |
|
} |