Annotation of loncom/metadata_database/searchcat.pl, revision 1.81
1.1 harris41 1: #!/usr/bin/perl
2: # The LearningOnline Network
3: # searchcat.pl "Search Catalog" batch script
1.16 harris41 4: #
1.81 ! bisitz 5: # $Id: searchcat.pl,v 1.80 2013/08/12 16:52:00 raeburn Exp $
1.16 harris41 6: #
7: # Copyright Michigan State University Board of Trustees
8: #
1.29 albertel 9: # This file is part of the LearningOnline Network with CAPA (LON-CAPA).
1.16 harris41 10: #
1.29 albertel 11: # LON-CAPA is free software; you can redistribute it and/or modify
1.16 harris41 12: # it under the terms of the GNU General Public License as published by
13: # the Free Software Foundation; either version 2 of the License, or
14: # (at your option) any later version.
15: #
1.29 albertel 16: # LON-CAPA is distributed in the hope that it will be useful,
1.16 harris41 17: # but WITHOUT ANY WARRANTY; without even the implied warranty of
18: # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19: # GNU General Public License for more details.
20: #
21: # You should have received a copy of the GNU General Public License
1.29 albertel 22: # along with LON-CAPA; if not, write to the Free Software
1.16 harris41 23: # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
24: #
25: # /home/httpd/html/adm/gpl.txt
26: #
1.29 albertel 27: # http://www.lon-capa.org/
1.16 harris41 28: #
29: ###
1.33 matthew 30:
1.32 www 31: =pod
1.1 harris41 32:
1.32 www 33: =head1 NAME
34:
35: B<searchcat.pl> - put authoritative filesystem data into sql database.
36:
37: =head1 SYNOPSIS
38:
39: Ordinarily this script is to be called from a loncapa cron job
40: (CVS source location: F<loncapa/loncom/cron/loncapa>; typical
41: filesystem installation location: F</etc/cron.d/loncapa>).
42:
43: Here is the cron job entry.
44:
45: C<# Repopulate and refresh the metadata database used for the search catalog.>
46: C<10 1 * * 7 www /home/httpd/perl/searchcat.pl>
47:
48: This script only allows itself to be run as the user C<www>.
49:
50: =head1 DESCRIPTION
51:
52: This script goes through a loncapa resource directory and gathers metadata.
53: The metadata is entered into a SQL database.
54:
55: This script also does general database maintenance such as reformatting
56: the C<loncapa:metadata> table if it is deprecated.
57:
58: This script evaluates dynamic metadata from the authors'
1.48 www 59: F<nohist_resevaldata.db> database file in order to store it in MySQL.
1.32 www 60:
61: This script is playing an increasingly important role for a loncapa
62: library server. The proper operation of this script is critical for a smooth
63: and correct user experience.
64:
65: =cut
1.1 harris41 66:
1.45 www 67: use strict;
1.55 matthew 68: use DBI;
1.17 harris41 69: use lib '/home/httpd/lib/perl/';
1.55 matthew 70: use LONCAPA::lonmetadata;
1.76 albertel 71: use LONCAPA;
1.56 matthew 72: use Getopt::Long;
1.1 harris41 73: use IO::File;
74: use HTML::TokeParser;
1.21 www 75: use GDBM_File;
1.24 www 76: use POSIX qw(strftime mktime);
1.80 raeburn 77: use Mail::Send;
1.81 ! bisitz 78: use Apache::loncommon();
1.56 matthew 79:
1.63 matthew 80: use Apache::lonnet();
1.62 matthew 81:
1.55 matthew 82: use File::Find;
1.1 harris41 83:
1.56 matthew 84: #
85: # Set up configuration options
1.63 matthew 86: my ($simulate,$oneuser,$help,$verbose,$logfile,$debug);
1.56 matthew 87: GetOptions (
88: 'help' => \$help,
89: 'simulate' => \$simulate,
90: 'only=s' => \$oneuser,
91: 'verbose=s' => \$verbose,
92: 'debug' => \$debug,
93: );
94:
95: if ($help) {
96: print <<"ENDHELP";
97: $0
98: Rebuild and update the LON-CAPA metadata database.
99: Options:
100: -help Print this help
101: -simulate Do not modify the database.
102: -only=user Only compute for the given user. Implies -simulate
103: -verbose=val Sets logging level, val must be a number
104: -debug Turns on debugging output
105: ENDHELP
106: exit 0;
107: }
108:
109: if (! defined($debug)) {
110: $debug = 0;
111: }
112:
113: if (! defined($verbose)) {
114: $verbose = 0;
115: }
116:
117: if (defined($oneuser)) {
118: $simulate=1;
119: }
120:
1.55 matthew 121: ##
122: ## Use variables for table names so we can test this routine a little easier
1.69 raeburn 123: my %oldnames = (
124: 'metadata' => 'metadata',
125: 'portfolio' => 'portfolio_metadata',
126: 'access' => 'portfolio_access',
127: 'addedfields' => 'portfolio_addedfields',
1.78 raeburn 128: 'allusers' => 'allusers',
1.69 raeburn 129: );
130:
131: my %newnames;
132: # new table names - append pid to have unique temporary tables
133: foreach my $key (keys(%oldnames)) {
134: $newnames{$key} = 'new'.$oldnames{$key}.$$;
135: }
1.45 www 136:
1.55 matthew 137: #
138: # Only run if machine is a library server
1.63 matthew 139: exit if ($Apache::lonnet::perlvar{'lonRole'} ne 'library');
1.78 raeburn 140: my $hostid = $Apache::lonnet::perlvar{'lonHostID'};
141:
1.55 matthew 142: #
143: # Make sure this process is running from user=www
144: my $wwwid=getpwnam('www');
145: if ($wwwid!=$<) {
1.63 matthew 146: my $emailto="$Apache::lonnet::perlvar{'lonAdmEMail'},$Apache::lonnet::perlvar{'lonSysEMail'}";
147: my $subj="LON: $Apache::lonnet::perlvar{'lonHostID'} User ID mismatch";
1.55 matthew 148: system("echo 'User ID mismatch. searchcat.pl must be run as user www.' |\
1.63 matthew 149: mail -s '$subj' $emailto > /dev/null");
1.55 matthew 150: exit 1;
151: }
152: #
153: # Let people know we are running
1.63 matthew 154: open(LOG,'>>'.$Apache::lonnet::perlvar{'lonDaemons'}.'/logs/searchcat.log');
1.56 matthew 155: &log(0,'==== Searchcat Run '.localtime()."====");
1.57 matthew 156:
157:
1.56 matthew 158: if ($debug) {
159: &log(0,'simulating') if ($simulate);
160: &log(0,'only processing user '.$oneuser) if ($oneuser);
161: &log(0,'verbosity level = '.$verbose);
162: }
1.55 matthew 163: #
164: # Connect to database
165: my $dbh;
1.63 matthew 166: if (! ($dbh = DBI->connect("DBI:mysql:loncapa","www",$Apache::lonnet::perlvar{'lonSqlAccess'},
1.55 matthew 167: { RaiseError =>0,PrintError=>0}))) {
1.56 matthew 168: &log(0,"Cannot connect to database!");
1.55 matthew 169: die "MySQL Error: Cannot connect to database!\n";
170: }
171: # This can return an error and still be okay, so we do not bother checking.
172: # (perhaps it should be more robust and check for specific errors)
1.69 raeburn 173: foreach my $key (keys(%newnames)) {
174: if ($newnames{$key} ne '') {
175: $dbh->do('DROP TABLE IF EXISTS '.$newnames{$key});
176: }
177: }
178:
1.55 matthew 179: #
1.77 raeburn 180: # Create the new metadata, portfolio and allusers tables
1.69 raeburn 181: foreach my $key (keys(%newnames)) {
182: if ($newnames{$key} ne '') {
183: my $request =
184: &LONCAPA::lonmetadata::create_metadata_storage($newnames{$key},$oldnames{$key});
185: $dbh->do($request);
186: if ($dbh->err) {
187: $dbh->disconnect();
188: &log(0,"MySQL Error Create: ".$dbh->errstr);
189: die $dbh->errstr;
190: }
191: }
1.55 matthew 192: }
1.69 raeburn 193:
1.55 matthew 194: #
195: # find out which users we need to examine
1.63 matthew 196: my @domains = sort(&Apache::lonnet::current_machine_domains());
197: &log(9,'domains ="'.join('","',@domains).'"');
1.62 matthew 198:
199: foreach my $dom (@domains) {
200: &log(9,'domain = '.$dom);
1.63 matthew 201: opendir(RESOURCES,"$Apache::lonnet::perlvar{'lonDocRoot'}/res/$dom");
1.62 matthew 202: my @homeusers =
203: grep {
1.63 matthew 204: &ishome("$Apache::lonnet::perlvar{'lonDocRoot'}/res/$dom/$_");
1.62 matthew 205: } grep {
206: !/^\.\.?$/;
207: } readdir(RESOURCES);
208: closedir RESOURCES;
209: &log(5,'users = '.$dom.':'.join(',',@homeusers));
210: #
211: if ($oneuser) {
212: @homeusers=($oneuser);
213: }
1.80 raeburn 214:
1.62 matthew 215: #
216: # Loop through the users
217: foreach my $user (@homeusers) {
218: &log(0,"=== User: ".$user);
219: &process_dynamic_metadata($user,$dom);
220: #
221: # Use File::Find to get the files we need to read/modify
222: find(
223: {preprocess => \&only_meta_files,
224: #wanted => \&print_filename,
225: #wanted => \&log_metadata,
226: wanted => \&process_meta_file,
1.66 albertel 227: no_chdir => 1,
1.63 matthew 228: }, join('/',($Apache::lonnet::perlvar{'lonDocRoot'},'res',$dom,$user)) );
1.62 matthew 229: }
1.77 raeburn 230: # Search for all users and public portfolio files
1.78 raeburn 231: my (%allusers,%portusers,%courses);
1.69 raeburn 232: if ($oneuser) {
233: %portusers = (
234: $oneuser => '',
235: );
1.77 raeburn 236: %allusers = (
237: $oneuser => '',
238: );
1.78 raeburn 239: %courses = &courseiddump($dom,'.',1,'.','.',$oneuser,undef,
240: undef,'.');
1.69 raeburn 241: } else {
1.78 raeburn 242: # get courseIDs for domain on current machine
243: %courses=&Apache::lonnet::courseiddump($dom,'.',1,'.','.','.',1,[$hostid],'.');
1.69 raeburn 244: my $dir = $Apache::lonnet::perlvar{lonUsersDir}.'/'.$dom;
1.77 raeburn 245: &descend_tree($dom,$dir,0,\%portusers,\%allusers);
1.69 raeburn 246: }
247: foreach my $uname (keys(%portusers)) {
248: my $urlstart = '/uploaded/'.$dom.'/'.$uname;
249: my $pathstart = &propath($dom,$uname).'/userfiles';
1.78 raeburn 250: my $is_course = '';
251: if (exists($courses{$dom.'_'.$uname})) {
252: $is_course = 1;
253: }
1.69 raeburn 254: my $curr_perm = &Apache::lonnet::get_portfile_permissions($dom,$uname);
255: my %access = &Apache::lonnet::get_access_controls($curr_perm);
1.75 raeburn 256: foreach my $file (keys(%access)) {
1.69 raeburn 257: my ($group,$url,$fullpath);
258: if ($is_course) {
259: ($group, my ($path)) = ($file =~ /^(\w+)(\/.+)$/);
1.72 raeburn 260: $fullpath = $pathstart.'/groups/'.$group.'/portfolio'.$path;
1.69 raeburn 261: $url = $urlstart.'/groups/'.$group.'/portfolio'.$path;
262: } else {
263: $fullpath = $pathstart.'/portfolio'.$file;
1.72 raeburn 264: $url = $urlstart.'/portfolio'.$file;
1.69 raeburn 265: }
266: if (ref($access{$file}) eq 'HASH') {
1.75 raeburn 267: my %portaccesslog =
268: &LONCAPA::lonmetadata::process_portfolio_access_data($dbh,
269: $simulate,\%newnames,$url,$fullpath,$access{$file});
270: &portfolio_logging(%portaccesslog);
1.69 raeburn 271: }
1.75 raeburn 272: my %portmetalog = &LONCAPA::lonmetadata::process_portfolio_metadata($dbh,$simulate,\%newnames,$url,$fullpath,$is_course,$dom,$uname,$group);
273: &portfolio_logging(%portmetalog);
1.69 raeburn 274: }
275: }
1.79 raeburn 276: my (%names_by_id,,%ids_by_name,%idstodelete,%idstoadd,%duplicates);
277: unless ($simulate || $oneuser) {
278: my $idshashref;
279: $idshashref = &tie_domain_hash($dom, "ids", &GDBM_WRCREAT());
280: if (ref($idshashref) eq 'HASH') {
281: %names_by_id = %{$idshashref};
282: while (my ($id,$uname) = each(%{$idshashref}) ) {
283: $id = &unescape($id);
284: $uname = &unescape($uname);
285: $names_by_id{$id} = $uname;
286: push(@{$ids_by_name{$uname}},$id);
287: }
288: &untie_domain_hash($idshashref);
289: }
290: }
1.77 raeburn 291: # Update allusers
292: foreach my $uname (keys(%allusers)) {
1.78 raeburn 293: next if (exists($courses{$dom.'_'.$uname}));
1.77 raeburn 294: my %userdata =
295: &Apache::lonnet::get('environment',['firstname','lastname',
296: 'middlename','generation','id','permanentemail'],$dom,$uname);
1.79 raeburn 297: unless ($simulate || $oneuser) {
298: my $addid;
299: if ($userdata{'id'} ne '') {
300: $addid = $userdata{'id'};
301: $addid=~tr/A-Z/a-z/;
302: }
303: if (exists($ids_by_name{$uname})) {
304: if (ref($ids_by_name{$uname}) eq 'ARRAY') {
305: if (scalar(@{$ids_by_name{$uname}}) > 1) {
306: &log(0,"Multiple employee/student IDs found in ids.db for $uname:$dom -- ".join(', ',@{$ids_by_name{$uname}}));
307: }
308: foreach my $id (@{$ids_by_name{$uname}}) {
309: if ($id eq $userdata{'id'}) {
310: undef($addid);
311: } else {
312: $idstodelete{$id} = $uname;
313: }
314: }
315: }
316: }
317: if ($addid ne '') {
318: if (exists($idstoadd{$addid})) {
319: push(@{$duplicates{$addid}},$uname);
320: } else {
321: $idstoadd{$addid} = $uname;
322: }
323: }
324: }
325:
1.77 raeburn 326: $userdata{'username'} = $uname;
327: $userdata{'domain'} = $dom;
328: my %alluserslog =
329: &LONCAPA::lonmetadata::process_allusers_data($dbh,$simulate,
330: \%newnames,$uname,$dom,\%userdata);
331: foreach my $item (keys(%alluserslog)) {
332: &log(0,$alluserslog{$item});
333: }
334: }
1.79 raeburn 335: unless ($simulate || $oneuser) {
336: if (keys(%idstodelete) > 0) {
337: my %resulthash = &Apache::lonnet::iddel($dom,\%idstodelete,$hostid);
338: if ($resulthash{$hostid} eq 'ok') {
339: foreach my $id (sort(keys(%idstodelete))) {
340: &log(0,"Record deleted from ids.db for $dom -- $id => ".$idstodelete{$id});
341: }
342: } else {
343: &log(0,"Error: '$resulthash{$hostid}' occurred when attempting to delete records from ids.db for $dom");
344: }
345: }
346: if (keys(%idstoadd) > 0) {
1.80 raeburn 347: my $idmessage = '';
348: my %newids;
349: foreach my $addid (sort(keys(%idstoadd))) {
350: if ((exists($names_by_id{$addid})) && ($names_by_id{$addid} ne $idstoadd{$addid}) && !($idstodelete{$addid})) {
351: &log(0,"Two usernames associated with a single ID $addid in domain: $dom: $names_by_id{$addid} (current) and $idstoadd{$addid}\n");
352: $idmessage .= "$addid,$names_by_id{$addid},$idstoadd{$addid}\n";
353: } else {
354: $newids{$addid} = $idstoadd{$addid};
355: }
356: }
357: if (keys(%newids) > 0) {
358: my $putresult = &Apache::lonnet::put_dom('ids',\%idstoadd,$dom,$hostid);
359: if ($putresult eq 'ok') {
360: foreach my $id (sort(keys(%idstoadd))) {
361: &log(0,"Record added to ids.db for $dom -- $id => ".$idstoadd{$id});
362: }
363: } else {
364: &log(0,"Error: '$putresult' occurred when attempting to add records to ids.db for $dom");
365: }
366: }
367: if ($idmessage) {
368: my $to = &Apache::loncommon::build_recipient_list(undef,'idconflictsmail',$dom);
369: if ($to ne '') {
370: my $msg = new Mail::Send;
371: $msg->to($to);
372: $msg->subject('LON-CAPA studentIDs conflict');
373: my $lonhost = $Apache::lonnet::perlvar{'lonHostID'};
374: my $hostname = &Apache::lonnet::hostname($lonhost);
375: my $replytoaddress = 'do-not-reply@'.$hostname;
376: $msg->add('Reply-to',$replytoaddress);
377: $msg->add('From',"www@$hostname");
378: $msg->add('Content-type','text/plain; charset=UTF-8');
379: if (my $fh = $msg->open()) {
380: print $fh
381: 'The following IDs are used for more than one user in your domain:'."\n".
382: 'Each row contains: Student/Employee ID, Current username in ids.db file, '.
383: 'Additional username'."\n\n".
384: $idmessage;
385: $fh->close;
386: }
1.79 raeburn 387: }
388: }
389: }
390: if (keys(%duplicates) > 0) {
391: foreach my $id (sort(keys(%duplicates))) {
392: &log(0,"Duplicate IDs found for entries to add to ids.db in $dom -- $id => $idstodelete{$id}");
393: }
394: }
395: }
1.55 matthew 396: }
1.69 raeburn 397:
1.55 matthew 398: #
1.69 raeburn 399: # Rename the tables
1.56 matthew 400: if (! $simulate) {
1.69 raeburn 401: foreach my $key (keys(%oldnames)) {
402: if (($oldnames{$key} ne '') && ($newnames{$key} ne '')) {
403: $dbh->do('DROP TABLE IF EXISTS '.$oldnames{$key});
404: if (! $dbh->do('RENAME TABLE '.$newnames{$key}.' TO '.$oldnames{$key})) {
405: &log(0,"MySQL Error Rename: ".$dbh->errstr);
406: die $dbh->errstr;
407: } else {
408: &log(1,"MySQL table rename successful for $key.");
409: }
410: }
1.56 matthew 411: }
1.55 matthew 412: }
413: if (! $dbh->disconnect) {
1.56 matthew 414: &log(0,"MySQL Error Disconnect: ".$dbh->errstr);
1.55 matthew 415: die $dbh->errstr;
416: }
417: ##
418: ## Finished!
1.56 matthew 419: &log(0,"==== Searchcat completed ".localtime()." ====");
1.55 matthew 420: close(LOG);
1.21 www 421:
1.55 matthew 422: &write_type_count();
423: &write_copyright_count();
1.36 www 424:
1.55 matthew 425: exit 0;
1.28 harris41 426:
1.56 matthew 427: ##
428: ## Status logging routine. Inputs: $level, $message
429: ##
430: ## $level 0 should be used for normal output and error messages
431: ##
432: ## $message does not need to end with \n. In the case of errors
433: ## the message should contain as much information as possible to
434: ## help in diagnosing the problem.
435: ##
436: sub log {
437: my ($level,$message)=@_;
438: $level = 0 if (! defined($level));
439: if ($verbose >= $level) {
440: print LOG $message.$/;
441: }
442: }
443:
1.75 raeburn 444: sub portfolio_logging {
445: my (%portlog) = @_;
446: foreach my $key (keys(%portlog)) {
447: if (ref($portlog{$key}) eq 'HASH') {
448: foreach my $item (keys(%{$portlog{$key}})) {
449: &log(0,$portlog{$key}{$item});
450: }
451: }
452: }
453: }
454:
1.69 raeburn 455: sub descend_tree {
1.77 raeburn 456: my ($dom,$dir,$depth,$allportusers,$alldomusers) = @_;
1.69 raeburn 457: if (-d $dir) {
458: opendir(DIR,$dir);
459: my @contents = grep(!/^\./,readdir(DIR));
460: closedir(DIR);
461: $depth ++;
462: foreach my $item (@contents) {
463: if ($depth < 4) {
1.77 raeburn 464: &descend_tree($dom,$dir.'/'.$item,$depth,$allportusers,$alldomusers);
1.69 raeburn 465: } else {
466: if (-e $dir.'/'.$item.'/file_permissions.db') {
1.78 raeburn 467: $$allportusers{$item} = '';
1.77 raeburn 468: }
1.78 raeburn 469: if (-e $dir.'/'.$item.'/passwd') {
1.69 raeburn 470: $$alldomusers{$item} = '';
471: }
472: }
473: }
474: }
475: }
476:
1.55 matthew 477: ########################################################
478: ########################################################
479: ### ###
480: ### File::Find support routines ###
481: ### ###
482: ########################################################
483: ########################################################
484: ##
485: ## &only_meta_files
486: ##
487: ## Called by File::Find.
488: ## Takes a list of files/directories in and returns a list of files/directories
489: ## to search.
490: sub only_meta_files {
491: my @PossibleFiles = @_;
492: my @ChosenFiles;
493: foreach my $file (@PossibleFiles) {
494: if ( ($file =~ /\.meta$/ && # Ends in meta
495: $file !~ /\.\d+\.[^\.]+\.meta$/ # is not for a prior version
1.67 albertel 496: ) || (-d $File::Find::dir."/".$file )) { # directories are okay
1.55 matthew 497: # but we do not want /. or /..
498: push(@ChosenFiles,$file);
499: }
1.38 www 500: }
1.55 matthew 501: return @ChosenFiles;
1.38 www 502: }
503:
1.55 matthew 504: ##
505: ##
506: ## Debugging routines, use these for 'wanted' in the File::Find call
507: ##
508: sub print_filename {
509: my ($file) = $_;
510: my $fullfilename = $File::Find::name;
1.56 matthew 511: if ($debug) {
512: if (-d $file) {
513: &log(5," Got directory ".$fullfilename);
514: } else {
515: &log(5," Got file ".$fullfilename);
516: }
1.38 www 517: }
1.55 matthew 518: $_=$file;
1.38 www 519: }
1.28 harris41 520:
1.55 matthew 521: sub log_metadata {
522: my ($file) = $_;
523: my $fullfilename = $File::Find::name;
524: return if (-d $fullfilename); # No need to do anything here for directories
1.56 matthew 525: if ($debug) {
526: &log(6,$fullfilename);
1.69 raeburn 527: my $ref = &metadata($fullfilename);
1.56 matthew 528: if (! defined($ref)) {
529: &log(6," No data");
530: return;
531: }
532: while (my($key,$value) = each(%$ref)) {
533: &log(6," ".$key." => ".$value);
534: }
535: &count_copyright($ref->{'copyright'});
1.55 matthew 536: }
537: $_=$file;
1.31 harris41 538: }
1.21 www 539:
1.55 matthew 540: ##
541: ## process_meta_file
542: ## Called by File::Find.
543: ## Only input is the filename in $_.
544: sub process_meta_file {
545: my ($file) = $_;
1.56 matthew 546: my $filename = $File::Find::name; # full filename
1.55 matthew 547: return if (-d $filename); # No need to do anything here for directories
548: #
1.56 matthew 549: &log(3,$filename) if ($debug);
1.55 matthew 550: #
1.69 raeburn 551: my $ref = &metadata($filename);
1.55 matthew 552: #
553: # $url is the original file url, not the metadata file
1.61 matthew 554: my $target = $filename;
555: $target =~ s/\.meta$//;
556: my $url='/res/'.&declutter($target);
1.56 matthew 557: &log(3," ".$url) if ($debug);
1.55 matthew 558: #
559: # Ignore some files based on their metadata
560: if ($ref->{'obsolete'}) {
1.56 matthew 561: &log(3,"obsolete") if ($debug);
1.55 matthew 562: return;
563: }
564: &count_copyright($ref->{'copyright'});
565: if ($ref->{'copyright'} eq 'private') {
1.56 matthew 566: &log(3,"private") if ($debug);
1.55 matthew 567: return;
568: }
569: #
570: # Find the dynamic metadata
571: my %dyn;
572: if ($url=~ m:/default$:) {
573: $url=~ s:/default$:/:;
1.56 matthew 574: &log(3,"Skipping dynamic data") if ($debug);
1.55 matthew 575: } else {
1.56 matthew 576: &log(3,"Retrieving dynamic data") if ($debug);
577: %dyn=&get_dynamic_metadata($url);
1.55 matthew 578: &count_type($url);
579: }
1.75 raeburn 580: &LONCAPA::lonmetadata::getfiledates($ref,$target);
1.55 matthew 581: #
582: my %Data = (
583: %$ref,
584: %dyn,
585: 'url'=>$url,
586: 'version'=>'current');
1.56 matthew 587: if (! $simulate) {
1.69 raeburn 588: my ($count,$err) =
589: &LONCAPA::lonmetadata::store_metadata($dbh,$newnames{'metadata'},
590: 'metadata',\%Data);
1.56 matthew 591: if ($err) {
592: &log(0,"MySQL Error Insert: ".$err);
593: }
594: if ($count < 1) {
595: &log(0,"Unable to insert record into MySQL database for $url");
596: }
1.55 matthew 597: }
598: #
599: # Reset $_ before leaving
600: $_ = $file;
601: }
602:
603: ########################################################
604: ########################################################
605: ### ###
606: ### &metadata($uri) ###
607: ### Retrieve metadata for the given file ###
608: ### ###
609: ########################################################
610: ########################################################
611: sub metadata {
1.69 raeburn 612: my ($uri) = @_;
1.55 matthew 613: my %metacache=();
614: $uri=&declutter($uri);
615: my $filename=$uri;
616: $uri=~s/\.meta$//;
617: $uri='';
618: if ($filename !~ /\.meta$/) {
619: $filename.='.meta';
620: }
1.75 raeburn 621: my $metastring =
622: &LONCAPA::lonmetadata::getfile($Apache::lonnet::perlvar{'lonDocRoot'}.'/res/'.$filename);
1.55 matthew 623: return undef if (! defined($metastring));
624: my $parser=HTML::TokeParser->new(\$metastring);
625: my $token;
626: while ($token=$parser->get_token) {
627: if ($token->[0] eq 'S') {
628: my $entry=$token->[1];
629: my $unikey=$entry;
630: if (defined($token->[2]->{'part'})) {
631: $unikey.='_'.$token->[2]->{'part'};
632: }
633: if (defined($token->[2]->{'name'})) {
634: $unikey.='_'.$token->[2]->{'name'};
635: }
636: if ($metacache{$uri.'keys'}) {
637: $metacache{$uri.'keys'}.=','.$unikey;
638: } else {
639: $metacache{$uri.'keys'}=$unikey;
640: }
641: foreach ( @{$token->[3]}) {
642: $metacache{$uri.''.$unikey.'.'.$_}=$token->[2]->{$_};
1.69 raeburn 643: }
1.55 matthew 644: if (! ($metacache{$uri.''.$unikey}=$parser->get_text('/'.$entry))){
645: $metacache{$uri.''.$unikey} =
646: $metacache{$uri.''.$unikey.'.default'};
647: }
648: } # End of ($token->[0] eq 'S')
649: }
650: return \%metacache;
1.31 harris41 651: }
1.28 harris41 652:
1.55 matthew 653: ########################################################
654: ########################################################
655: ### ###
656: ### Dynamic Metadata ###
657: ### ###
658: ########################################################
659: ########################################################
1.56 matthew 660: ##
1.58 www 661: ## Dynamic metadata description (incomplete)
662: ##
663: ## For a full description of all fields,
664: ## see LONCAPA::lonmetadata
1.56 matthew 665: ##
666: ## Field Type
667: ##-----------------------------------------------------------
668: ## count integer
669: ## course integer
1.58 www 670: ## course_list comma separated list of course ids
1.56 matthew 671: ## avetries real
1.58 www 672: ## avetries_list comma separated list of real numbers
1.56 matthew 673: ## stdno real
1.58 www 674: ## stdno_list comma separated list of real numbers
1.56 matthew 675: ## usage integer
1.58 www 676: ## usage_list comma separated list of resources
1.56 matthew 677: ## goto scalar
1.58 www 678: ## goto_list comma separated list of resources
1.56 matthew 679: ## comefrom scalar
1.58 www 680: ## comefrom_list comma separated list of resources
1.56 matthew 681: ## difficulty real
1.58 www 682: ## difficulty_list comma separated list of real numbers
1.56 matthew 683: ## sequsage scalar
1.58 www 684: ## sequsage_list comma separated list of resources
1.56 matthew 685: ## clear real
686: ## technical real
687: ## correct real
688: ## helpful real
689: ## depth real
690: ## comments html of all the comments made
691: ##
692: {
693:
694: my %DynamicData;
695: my %Counts;
696:
697: sub process_dynamic_metadata {
698: my ($user,$dom) = @_;
699: undef(%DynamicData);
700: undef(%Counts);
701: #
702: my $prodir = &propath($dom,$user);
1.55 matthew 703: #
1.56 matthew 704: # Read in the dynamic metadata
1.55 matthew 705: my %evaldata;
706: if (! tie(%evaldata,'GDBM_File',
707: $prodir.'/nohist_resevaldata.db',&GDBM_READER(),0640)) {
1.56 matthew 708: return 0;
1.55 matthew 709: }
1.56 matthew 710: #
1.57 matthew 711: %DynamicData = &LONCAPA::lonmetadata::process_reseval_data(\%evaldata);
1.55 matthew 712: untie(%evaldata);
1.62 matthew 713: $DynamicData{'domain'} = $dom;
1.64 albertel 714: #print('user = '.$user.' domain = '.$dom.$/);
1.56 matthew 715: #
716: # Read in the access count data
717: &log(7,'Reading access count data') if ($debug);
718: my %countdata;
719: if (! tie(%countdata,'GDBM_File',
720: $prodir.'/nohist_accesscount.db',&GDBM_READER(),0640)) {
721: return 0;
722: }
723: while (my ($key,$count) = each(%countdata)) {
724: next if ($key !~ /^$dom/);
725: $key = &unescape($key);
726: &log(8,' Count '.$key.' = '.$count) if ($debug);
727: $Counts{$key}=$count;
728: }
729: untie(%countdata);
730: if ($debug) {
731: &log(7,scalar(keys(%Counts)).
732: " Counts read for ".$user."@".$dom);
733: &log(7,scalar(keys(%DynamicData)).
734: " Dynamic metadata read for ".$user."@".$dom);
735: }
736: #
737: return 1;
738: }
739:
740: sub get_dynamic_metadata {
741: my ($url) = @_;
742: $url =~ s:^/res/::;
1.57 matthew 743: my %data = &LONCAPA::lonmetadata::process_dynamic_metadata($url,
744: \%DynamicData);
1.56 matthew 745: # find the count
746: $data{'count'} = $Counts{$url};
747: #
748: # Log the dynamic metadata
749: if ($debug) {
750: while (my($k,$v)=each(%data)) {
751: &log(8," ".$k." => ".$v);
752: }
1.44 www 753: }
1.56 matthew 754: return %data;
1.30 www 755: }
1.28 harris41 756:
1.56 matthew 757: } # End of %DynamicData and %Counts scope
758:
1.55 matthew 759: ########################################################
760: ########################################################
761: ### ###
762: ### Counts ###
763: ### ###
764: ########################################################
765: ########################################################
766: {
1.1 harris41 767:
1.55 matthew 768: my %countext;
1.15 harris41 769:
1.55 matthew 770: sub count_type {
771: my $file=shift;
772: $file=~/\.(\w+)$/;
773: my $ext=lc($1);
774: $countext{$ext}++;
1.31 harris41 775: }
1.1 harris41 776:
1.55 matthew 777: sub write_type_count {
778: open(RESCOUNT,'>/home/httpd/html/lon-status/rescount.txt');
779: while (my ($extension,$count) = each(%countext)) {
780: print RESCOUNT $extension.'='.$count.'&';
1.47 www 781: }
1.55 matthew 782: print RESCOUNT 'time='.time."\n";
783: close(RESCOUNT);
1.31 harris41 784: }
1.27 www 785:
1.55 matthew 786: } # end of scope for %countext
1.34 matthew 787:
1.55 matthew 788: {
1.34 matthew 789:
1.55 matthew 790: my %copyrights;
1.44 www 791:
1.55 matthew 792: sub count_copyright {
793: $copyrights{@_[0]}++;
1.31 harris41 794: }
1.33 matthew 795:
1.55 matthew 796: sub write_copyright_count {
797: open(COPYCOUNT,'>/home/httpd/html/lon-status/copyrightcount.txt');
798: while (my ($copyright,$count) = each(%copyrights)) {
799: print COPYCOUNT $copyright.'='.$count.'&';
1.31 harris41 800: }
1.55 matthew 801: print COPYCOUNT 'time='.time."\n";
802: close(COPYCOUNT);
1.31 harris41 803: }
1.28 harris41 804:
1.55 matthew 805: } # end of scope for %copyrights
1.28 harris41 806:
1.55 matthew 807: ########################################################
808: ########################################################
809: ### ###
810: ### Miscellanous Utility Routines ###
811: ### ###
812: ########################################################
813: ########################################################
814: ##
815: ## &ishome($username)
816: ## Returns 1 if $username is a LON-CAPA author, 0 otherwise
817: ## (copied from lond, modification of the return value)
1.31 harris41 818: sub ishome {
819: my $author=shift;
1.76 albertel 820: $author=~s{/home/httpd/html/res/([^/]*)/([^/]*).*}{$1/$2};
1.31 harris41 821: my ($udom,$uname)=split(/\//,$author);
822: my $proname=propath($udom,$uname);
823: if (-e $proname) {
824: return 1;
825: } else {
826: return 0;
827: }
828: }
1.28 harris41 829:
1.55 matthew 830: ##
831: ## &declutter($filename)
832: ## Given a filename, returns a url for the filename.
833: sub declutter {
834: my $thisfn=shift;
1.63 matthew 835: $thisfn=~s/^$Apache::lonnet::perlvar{'lonDocRoot'}//;
1.55 matthew 836: $thisfn=~s/^\///;
837: $thisfn=~s/^res\///;
838: return $thisfn;
1.31 harris41 839: }
1.28 harris41 840:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/icons/back.gif){kind=icon}
Return to searchcat.pl CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [LON-CAPA] / loncom / metadata_database