version 1.50, 2003/12/26 15:13:48
|
version 1.59, 2004/06/22 14:16:08
|
Line 66 and correct user experience.
|
Line 66 and correct user experience.
|
|
|
use strict; |
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 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); |
|
|
require "find.pl"; |
use File::Find; |
|
|
my @metalist; |
|
|
|
my $simplestatus=''; |
|
my %countext=(); |
|
|
|
# ----------------------------------------------------- write out simple status |
|
sub writesimple { |
|
open(SMP,'>/home/httpd/html/lon-status/mysql.txt'); |
|
print SMP $simplestatus."\n"; |
|
close(SMP); |
|
} |
|
|
|
sub writecount { |
|
open(RSMP,'>/home/httpd/html/lon-status/rescount.txt'); |
|
foreach (keys %countext) { |
|
print RSMP $_.'='.$countext{$_}.'&'; |
|
} |
|
print RSMP 'time='.time."\n"; |
|
close(RSMP); |
|
} |
|
|
|
# -------------------------------------- counts files with different extensions |
|
sub count { |
|
my $file=shift; |
|
$file=~/\.(\w+)$/; |
|
my $ext=lc($1); |
|
if (defined($countext{$ext})) { |
|
$countext{$ext}++; |
|
} else { |
|
$countext{$ext}=1; |
|
} |
|
} |
|
# ----------------------------------------------------- Un-Escape Special Chars |
|
|
|
sub unescape { |
|
my $str=shift; |
|
$str =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack("C",hex($1))/eg; |
|
return $str; |
|
} |
|
|
|
# -------------------------------------------------------- Escape Special Chars |
|
|
|
sub escape { |
|
my $str=shift; |
|
$str =~ s/(\W)/"%".unpack('H2',$1)/eg; |
|
return $str; |
|
} |
|
|
|
# ------------------------------------------- Code to evaluate dynamic metadata |
# |
|
# Set up configuration options |
|
my ($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 $oldname = 'metadata'; |
|
my $newname = 'newmetadata'.$$; # append pid to have unique temporary table |
|
|
sub dynamicmeta { |
# |
my $url=&declutter(shift); |
# Read loncapa_apache.conf and loncapa.conf |
$url=~s/\.meta$//; |
|
my %returnhash=( |
|
'count' => 0, |
|
'course' => 0, |
|
'course_list' => '', |
|
'avetries' => 'NULL', |
|
'avetries_list' => '', |
|
'stdno' => 0, |
|
'stdno_list' => '', |
|
'usage' => 0, |
|
'usage_list' => '', |
|
'goto' => 0, |
|
'goto_list' => '', |
|
'comefrom' => 0, |
|
'comefrom_list' => '', |
|
'difficulty' => 'NULL', |
|
'difficulty_list' => '', |
|
'clear' => 'NULL', |
|
'technical' => 'NULL', |
|
'correct' => 'NULL', |
|
'helpful' => 'NULL', |
|
'depth' => 'NULL', |
|
'comments' => '' |
|
); |
|
my ($adomain,$aauthor)=($url=~/^(\w+)\/(\w+)\//); |
|
my $prodir=&propath($adomain,$aauthor); |
|
|
|
# Get metadata except counts |
|
if (tie(my %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) { |
|
my $key=&unescape($esckey); |
|
if ($key=~/$regexp/) { |
|
my ($item,$purl,$cat)=split(/___/,$key); |
|
if (defined($cnt{$cat})) { $cnt{$cat}++; } else { $cnt{$cat}=1; } |
|
unless ($listitems{$cat} eq 'app') { |
|
if (defined($sum{$cat})) { |
|
$sum{$cat}+=&unescape($evaldata{$esckey}); |
|
$concat{$cat}.=','.$item; |
|
} else { |
|
$sum{$cat}=&unescape($evaldata{$esckey}); |
|
$concat{$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); |
|
# transfer gathered data to returnhash, calculate averages where applicable |
|
while (my $cat=each(%cnt)) { |
|
if ($cnt{$cat} eq 'nan') { next; } |
|
if ($sum{$cat} eq 'nan') { next; } |
|
if ($listitems{$cat} eq 'avg') { |
|
if ($cnt{$cat}) { |
|
$returnhash{$cat}=int(($sum{$cat}/$cnt{$cat})*100.0+0.5)/100.0; |
|
} else { |
|
$returnhash{$cat}='NULL'; |
|
} |
|
} elsif ($listitems{$cat} eq 'cnt') { |
|
$returnhash{$cat}=$cnt{$cat}; |
|
} else { |
|
$returnhash{$cat}=$sum{$cat}; |
|
} |
|
$returnhash{$cat.'_list'}=$concat{$cat}; |
|
} |
|
} |
|
# get count |
|
if (tie(my %evaldata,'GDBM_File', |
|
$prodir.'/nohist_accesscount.db',&GDBM_READER(),0640)) { |
|
my $escurl=&escape($url); |
|
if (! exists($evaldata{$escurl})) { |
|
$returnhash{'count'}=0; |
|
} else { |
|
$returnhash{'count'}=$evaldata{$escurl}; |
|
} |
|
untie %evaldata; |
|
} |
|
return %returnhash; |
|
} |
|
|
|
# --------------- Read loncapa_apache.conf and loncapa.conf and get variables |
|
my $perlvarref=LONCAPA::Configuration::read_conf('loncapa.conf'); |
my $perlvarref=LONCAPA::Configuration::read_conf('loncapa.conf'); |
my %perlvar=%{$perlvarref}; |
my %perlvar=%{$perlvarref}; |
undef $perlvarref; |
undef $perlvarref; |
delete $perlvar{'lonReceipt'}; # remove since sensitive and not needed |
delete $perlvar{'lonReceipt'}; # remove since sensitive (really?) & not needed |
|
# |
# ------------------------------------- Only run if machine is a library server |
# Only run if machine is a library server |
exit unless $perlvar{'lonRole'} eq 'library'; |
exit if ($perlvar{'lonRole'} ne 'library'); |
|
# |
# ----------------------------- Make sure this process is running from user=www |
# Make sure this process is running from user=www |
|
|
my $wwwid=getpwnam('www'); |
my $wwwid=getpwnam('www'); |
if ($wwwid!=$<) { |
if ($wwwid!=$<) { |
my $emailto="$perlvar{'lonAdmEMail'},$perlvar{'lonSysEMail'}"; |
my $emailto="$perlvar{'lonAdmEMail'},$perlvar{'lonSysEMail'}"; |
Line 257 if ($wwwid!=$<) {
|
Line 140 if ($wwwid!=$<) {
|
mailto $emailto -s '$subj' > /dev/null"); |
mailto $emailto -s '$subj' > /dev/null"); |
exit 1; |
exit 1; |
} |
} |
|
# |
|
# Let people know we are running |
|
open(LOG,'>'.$perlvar{'lonDaemons'}.'/logs/searchcat.log'); |
|
&log(0,'==== Searchcat Run '.localtime()."===="); |
|
|
|
|
# ---------------------------------------------------------- We are in business |
if ($debug) { |
|
&log(0,'simulating') if ($simulate); |
open(LOG,'>'.$perlvar{'lonDaemons'}.'/logs/searchcat.log'); |
&log(0,'only processing user '.$oneuser) if ($oneuser); |
print LOG '==== Searchcat Run '.localtime()."====\n\n"; |
&log(0,'verbosity level = '.$verbose); |
$simplestatus='time='.time.'&'; |
} |
|
# |
|
# Connect to database |
my $dbh; |
my $dbh; |
# ------------------------------------- Make sure that database can be accessed |
if (! ($dbh = DBI->connect("DBI:mysql:loncapa","www",$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 LOG "Cannot connect to database!\n"; |
# This can return an error and still be okay, so we do not bother checking. |
$simplestatus.='mysql=defunct'; |
# (perhaps it should be more robust and check for specific errors) |
&writesimple(); |
$dbh->do('DROP TABLE IF EXISTS '.$newname); |
exit; |
# |
} |
# Create the new table |
|
my $request = &LONCAPA::lonmetadata::create_metadata_storage($newname); |
# Make temporary table |
$dbh->do($request); |
$dbh->do("DROP TABLE IF EXISTS newmetadata"); |
if ($dbh->err) { |
my $make_metadata_table = "CREATE TABLE IF NOT EXISTS newmetadata (". |
$dbh->disconnect(); |
"title TEXT, author TEXT, subject TEXT, url TEXT, keywords TEXT, ". |
&log(0,"MySQL Error Create: ".$dbh->errstr); |
"version TEXT, notes TEXT, abstract TEXT, mime TEXT, language TEXT, ". |
die $dbh->errstr; |
"creationdate DATETIME, lastrevisiondate DATETIME, owner TEXT, ". |
|
"copyright TEXT, dependencies TEXT, ". |
|
"count INTEGER UNSIGNED, ". |
|
"course INTEGER UNSIGNED, course_list TEXT, ". |
|
"goto INTEGER UNSIGNED, goto_list TEXT, ". |
|
"comefrom INTEGER UNSIGNED, comefrom_list TEXT, ". |
|
"sequsage INTEGER UNSIGNED, sequsage_list TEXT, ". |
|
"stdno INTEGER UNSIGNED, stdno_list TEXT, ". |
|
"avetries FLOAT, avetries_list TEXT, ". |
|
"difficulty FLOAT, difficulty_list TEXT, ". |
|
"clear FLOAT, technical FLOAT, correct FLOAT, helpful FLOAT, depth FLOAT, ". |
|
"comments TEXT, ". |
|
"FULLTEXT idx_title (title), ". |
|
"FULLTEXT idx_author (author), FULLTEXT idx_subject (subject), ". |
|
"FULLTEXT idx_url (url), FULLTEXT idx_keywords (keywords), ". |
|
"FULLTEXT idx_version (version), FULLTEXT idx_notes (notes), ". |
|
"FULLTEXT idx_abstract (abstract), FULLTEXT idx_mime (mime), ". |
|
"FULLTEXT idx_language (language), FULLTEXT idx_owner (owner), ". |
|
"FULLTEXT idx_copyright (copyright)) ". |
|
"TYPE=MyISAM"; |
|
# It would sure be nice to have some logging mechanism. |
|
unless ($dbh->do($make_metadata_table)) { |
|
print LOG "\nMySQL Error Create: ".$dbh->errstr."\n"; |
|
die $dbh->errstr; |
|
} |
|
} |
} |
|
# |
# ------------------------------------------------------------- get .meta files |
# find out which users we need to examine |
opendir(RESOURCES,"$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}"); |
my $dom = $perlvar{'lonDefDomain'}; |
my @homeusers = grep { |
opendir(RESOURCES,"$perlvar{'lonDocRoot'}/res/$dom"); |
&ishome("$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$_") |
my @homeusers = |
} grep {!/^\.\.?$/} readdir(RESOURCES); |
grep { |
|
&ishome("$perlvar{'lonDocRoot'}/res/$dom/$_"); |
|
} grep { |
|
!/^\.\.?$/; |
|
} readdir(RESOURCES); |
closedir RESOURCES; |
closedir RESOURCES; |
|
|
# |
# |
# Create the statement handlers we need |
if ($oneuser) { |
|
@homeusers=($oneuser); |
my $insert_sth = $dbh->prepare |
} |
("INSERT INTO newmetadata VALUES (". |
# |
"?,". # title |
# Loop through the users |
"?,". # author |
|
"?,". # subject |
|
"?,". # declutter url |
|
"?,". # version |
|
"?,". # current |
|
"?,". # notes |
|
"?,". # abstract |
|
"?,". # mime |
|
"?,". # language |
|
"?,". # creationdate |
|
"?,". # revisiondate |
|
"?,". # owner |
|
"?,". # copyright |
|
"?,". # dependencies |
|
"?,". # count |
|
"?,". # course |
|
"?,". # course_list |
|
"?,". # goto |
|
"?,". # goto_list |
|
"?,". # comefrom |
|
"?,". # comefrom_list |
|
"?,". # usage |
|
"?,". # usage_list |
|
"?,". # stdno |
|
"?,". # stdno_list |
|
"?,". # avetries |
|
"?,". # avetries_list |
|
"?,". # difficulty |
|
"?,". # difficulty_list |
|
"?,". # clear |
|
"?,". # technical |
|
"?,". # correct |
|
"?,". # helpful |
|
"?,". # depth |
|
"?". # comments |
|
")" |
|
); |
|
|
|
foreach my $user (@homeusers) { |
foreach my $user (@homeusers) { |
print LOG "\n=== User: ".$user."\n\n"; |
&log(0,"=== User: ".$user); |
|
&process_dynamic_metadata($user,$dom); |
my $prodir=&propath($perlvar{'lonDefDomain'},$user); |
# |
# Use find.pl |
# Use File::Find to get the files we need to read/modify |
undef @metalist; |
find( |
@metalist=(); |
{preprocess => \&only_meta_files, |
&find("$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$user"); |
# wanted => \&print_filename, |
# -- process each file to get metadata and put into search catalog SQL |
# wanted => \&log_metadata, |
# database. Also, check to see if already there. |
wanted => \&process_meta_file, |
# I could just delete (without searching first), but this works for now. |
}, |
foreach my $m (@metalist) { |
"$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$user"); |
print LOG "- ".$m."\n"; |
} |
my $ref=&metadata($m); |
# |
my $m2='/res/'.&declutter($m); |
# Rename the table |
$m2=~s/\.meta$//; |
if (! $simulate) { |
if ($ref->{'obsolete'}) { print LOG "obsolete\n"; next; } |
$dbh->do('DROP TABLE IF EXISTS '.$oldname); |
if ($ref->{'copyright'} eq 'private') { print LOG "private\n"; next; } |
if (! $dbh->do('RENAME TABLE '.$newname.' TO '.$oldname)) { |
my %dyn=&dynamicmeta($m2); |
&log(0,"MySQL Error Rename: ".$dbh->errstr); |
&count($m2); |
die $dbh->errstr; |
unless ($insert_sth->execute( |
} else { |
$ref->{'title'}, |
&log(1,"MySQL table rename successful."); |
$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'}, |
|
$ref->{'dependencies'}, |
|
$dyn{'count'}, |
|
$dyn{'course'}, |
|
$dyn{'course_list'}, |
|
$dyn{'goto'}, |
|
$dyn{'goto_list'}, |
|
$dyn{'comefrom'}, |
|
$dyn{'comefrom_list'}, |
|
$dyn{'usage'}, |
|
$dyn{'usage_list'}, |
|
$dyn{'stdno'}, |
|
$dyn{'stdno_list'}, |
|
$dyn{'avetries'}, |
|
$dyn{'avetries_list'}, |
|
$dyn{'difficulty'}, |
|
$dyn{'difficulty_list'}, |
|
$dyn{'clear'}, |
|
$dyn{'technical'}, |
|
$dyn{'correct'}, |
|
$dyn{'helpful'}, |
|
$dyn{'depth'}, |
|
$dyn{'comments'} |
|
)) { |
|
print LOG "\nMySQL Error Insert: ".$dbh->errstr."\n"; |
|
die $dbh->errstr; |
|
} |
|
$ref = undef; |
|
} |
|
} |
|
# --------------------------------------------------- Close database connection |
|
$dbh->do("DROP TABLE IF EXISTS metadata"); |
|
unless ($dbh->do("RENAME TABLE newmetadata TO metadata")) { |
|
print LOG "\nMySQL Error Rename: ".$dbh->errstr."\n"; |
|
die $dbh->errstr; |
|
} |
} |
unless ($dbh->disconnect) { |
|
print LOG "\nMySQL Error Disconnect: ".$dbh->errstr."\n"; |
if (! $dbh->disconnect) { |
|
&log(0,"MySQL Error Disconnect: ".$dbh->errstr); |
die $dbh->errstr; |
die $dbh->errstr; |
} |
} |
print LOG "\n==== Searchcat completed ".localtime()." ====\n"; |
## |
|
## Finished! |
|
&log(0,"==== Searchcat completed ".localtime()." ===="); |
close(LOG); |
close(LOG); |
&writesimple(); |
|
&writecount(); |
&write_type_count(); |
|
&write_copyright_count(); |
|
|
exit 0; |
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 $message.$/; |
|
} |
|
} |
|
|
|
######################################################## |
|
######################################################## |
|
### ### |
|
### 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); |
|
} |
|
} |
|
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 $url='/res/'.&declutter($filename); |
|
$url=~s/\.meta$//; |
|
&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); |
|
} |
|
# |
|
$ref->{'creationdate'} = &sqltime($ref->{'creationdate'}); |
|
$ref->{'lastrevisiondate'} = &sqltime($ref->{'lastrevisiondate'}); |
|
my %Data = ( |
|
%$ref, |
|
%dyn, |
|
'url'=>$url, |
|
'version'=>'current'); |
|
if (! $simulate) { |
|
my ($count,$err) = &LONCAPA::lonmetadata::store_metadata($dbh,$newname, |
|
\%Data); |
|
if ($err) { |
|
&log(0,"MySQL Error Insert: ".$err); |
|
} |
|
if ($count < 1) { |
|
&log(0,"Unable to insert record into MySQL database for $url"); |
|
} |
|
} |
|
# |
|
# Reset $_ before leaving |
|
$_ = $file; |
|
} |
|
|
# ---------------------------------------------------------------- Get metadata |
######################################################## |
# significantly altered from subroutine present in lonnet |
######################################################## |
|
### ### |
|
### &metadata($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 $metastring=&getfile($perlvar{'lonDocRoot'}.'/res/'.$filename); |
my $token; |
return undef if (! defined($metastring)); |
while ($token=$parser->get_token) { |
my $parser=HTML::TokeParser->new(\$metastring); |
if ($token->[0] eq 'S') { |
my $token; |
my $entry=$token->[1]; |
while ($token=$parser->get_token) { |
my $unikey=$entry; |
if ($token->[0] eq 'S') { |
if (defined($token->[2]->{'part'})) { |
my $entry=$token->[1]; |
$unikey.='_'.$token->[2]->{'part'}; |
my $unikey=$entry; |
} |
if (defined($token->[2]->{'part'})) { |
if (defined($token->[2]->{'name'})) { |
$unikey.='_'.$token->[2]->{'part'}; |
$unikey.='_'.$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.''.$unikey.'.default'}; |
|
} |
|
} |
} |
} |
if (defined($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 |
## &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 ) { |
|
return undef; |
|
} |
my $fh=IO::File->new($file); |
my $fh=IO::File->new($file); |
my $a=''; |
my $contents = ''; |
while (<$fh>) { $a .=$_; } |
while (<$fh>) { |
return $a; |
$contents .= $_; |
|
} |
|
return $contents; |
} |
} |
|
|
# ------------------------------------------------------------- Declutters URLs |
######################################################## |
sub declutter { |
######################################################## |
my $thisfn=shift; |
### ### |
$thisfn=~s/^$perlvar{'lonDocRoot'}//; |
### Dynamic Metadata ### |
$thisfn=~s/^\///; |
### ### |
$thisfn=~s/^res\///; |
######################################################## |
return $thisfn; |
######################################################## |
|
## |
|
## Dynamic metadata description (incomplete) |
|
## |
|
## For a full description of all fields, |
|
## see LONCAPA::lonmetadata |
|
## |
|
## Field Type |
|
##----------------------------------------------------------- |
|
## count integer |
|
## course integer |
|
## course_list comma separated list of course ids |
|
## avetries real |
|
## avetries_list comma separated list of real numbers |
|
## stdno real |
|
## stdno_list comma separated list of real numbers |
|
## usage integer |
|
## usage_list comma separated list of resources |
|
## goto scalar |
|
## goto_list comma separated list of resources |
|
## comefrom scalar |
|
## comefrom_list comma separated list of resources |
|
## difficulty real |
|
## difficulty_list comma separated list of real numbers |
|
## sequsage scalar |
|
## sequsage_list comma separated list of resources |
|
## clear real |
|
## technical real |
|
## correct real |
|
## helpful real |
|
## depth real |
|
## comments html of all the comments made |
|
## |
|
{ |
|
|
|
my %DynamicData; |
|
my %Counts; |
|
|
|
sub process_dynamic_metadata { |
|
my ($user,$dom) = @_; |
|
undef(%DynamicData); |
|
undef(%Counts); |
|
# |
|
my $prodir = &propath($dom,$user); |
|
# |
|
# Read in the dynamic metadata |
|
my %evaldata; |
|
if (! tie(%evaldata,'GDBM_File', |
|
$prodir.'/nohist_resevaldata.db',&GDBM_READER(),0640)) { |
|
return 0; |
|
} |
|
# |
|
%DynamicData = &LONCAPA::lonmetadata::process_reseval_data(\%evaldata); |
|
untie(%evaldata); |
|
# |
|
# Read in the access count data |
|
&log(7,'Reading access count data') if ($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) = @_; |
|
$url =~ s:^/res/::; |
|
if (! exists($DynamicData{$url})) { |
|
&log(7,' No dynamic data for '.$url) if ($debug); |
|
return (); |
|
} |
|
my %data = &LONCAPA::lonmetadata::process_dynamic_metadata($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\/$2/; |
Line 520 sub ishome {
|
Line 614 sub ishome {
|
} |
} |
} |
} |
|
|
# -------------------------------------------- Return path to profile directory |
## |
# (copied from lond) |
## &propath($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 532 sub propath {
|
Line 628 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; |
|
$mysqltime = sprintf('%04d-%02d-%02d %02d:%02d:%02d', |
|
@TimeData[5,4,3,2,1,0]); |
|
} elsif (! defined($time) || $time == 0) { |
|
$mysqltime = 0; |
|
} else { |
|
&log(0," sqltime:Unable to decode time ".$time); |
|
$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; |
|
} |
} |
|
|
# ----------------- Code to enable 'find' subroutine listing of the .meta files |
sub escape { |
|
my $str=shift; |
no strict "vars"; |
$str =~ s/(\W)/"%".unpack('H2',$1)/eg; |
|
return $str; |
sub wanted { |
|
(($dev,$ino,$mode,$nlink,$uid,$gid) = lstat($_)) && |
|
-f _ && |
|
/^.*\.meta$/ && !/^.+\.\d+\.[^\.]+\.meta$/ && |
|
push(@metalist,"$dir/$_"); |
|
} |
} |