File:
[LON-CAPA] /
loncom /
metadata_database /
searchcat.pl
Revision
1.59:
download - view:
text,
annotated -
select for diffs
Tue Jun 22 14:16:08 2004 UTC (19 years, 10 months ago) by
matthew
Branches:
MAIN
CVS tags:
version_1_2_X,
version_1_2_1,
version_1_2_0,
version_1_1_99_5,
version_1_1_99_4,
version_1_1_99_3,
version_1_1_99_2,
version_1_1_99_1,
HEAD
Append PID to temporary table name to prevent collisions when multiple
searchcat run simultaniously.
1: #!/usr/bin/perl
2: # The LearningOnline Network
3: # searchcat.pl "Search Catalog" batch script
4: #
5: # $Id: searchcat.pl,v 1.59 2004/06/22 14:16:08 matthew Exp $
6: #
7: # Copyright Michigan State University Board of Trustees
8: #
9: # This file is part of the LearningOnline Network with CAPA (LON-CAPA).
10: #
11: # LON-CAPA is free software; you can redistribute it and/or modify
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: #
16: # LON-CAPA is distributed in the hope that it will be useful,
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
22: # along with LON-CAPA; if not, write to the Free Software
23: # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
24: #
25: # /home/httpd/html/adm/gpl.txt
26: #
27: # http://www.lon-capa.org/
28: #
29: ###
30:
31: =pod
32:
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'
59: F<nohist_resevaldata.db> database file in order to store it in MySQL.
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
66:
67: use strict;
68:
69: use DBI;
70: use lib '/home/httpd/lib/perl/';
71: use LONCAPA::Configuration;
72: use LONCAPA::lonmetadata;
73:
74: use Getopt::Long;
75: use IO::File;
76: use HTML::TokeParser;
77: use GDBM_File;
78: use POSIX qw(strftime mktime);
79:
80: use File::Find;
81:
82: #
83: # Set up configuration options
84: my ($simulate,$oneuser,$help,$verbose,$logfile,$debug);
85: GetOptions (
86: 'help' => \$help,
87: 'simulate' => \$simulate,
88: 'only=s' => \$oneuser,
89: 'verbose=s' => \$verbose,
90: 'debug' => \$debug,
91: );
92:
93: if ($help) {
94: print <<"ENDHELP";
95: $0
96: Rebuild and update the LON-CAPA metadata database.
97: Options:
98: -help Print this help
99: -simulate Do not modify the database.
100: -only=user Only compute for the given user. Implies -simulate
101: -verbose=val Sets logging level, val must be a number
102: -debug Turns on debugging output
103: ENDHELP
104: exit 0;
105: }
106:
107: if (! defined($debug)) {
108: $debug = 0;
109: }
110:
111: if (! defined($verbose)) {
112: $verbose = 0;
113: }
114:
115: if (defined($oneuser)) {
116: $simulate=1;
117: }
118:
119: ##
120: ## Use variables for table names so we can test this routine a little easier
121: my $oldname = 'metadata';
122: my $newname = 'newmetadata'.$$; # append pid to have unique temporary table
123:
124: #
125: # Read loncapa_apache.conf and loncapa.conf
126: my $perlvarref=LONCAPA::Configuration::read_conf('loncapa.conf');
127: my %perlvar=%{$perlvarref};
128: undef $perlvarref;
129: delete $perlvar{'lonReceipt'}; # remove since sensitive (really?) & not needed
130: #
131: # Only run if machine is a library server
132: exit if ($perlvar{'lonRole'} ne 'library');
133: #
134: # Make sure this process is running from user=www
135: my $wwwid=getpwnam('www');
136: if ($wwwid!=$<) {
137: my $emailto="$perlvar{'lonAdmEMail'},$perlvar{'lonSysEMail'}";
138: my $subj="LON: $perlvar{'lonHostID'} User ID mismatch";
139: system("echo 'User ID mismatch. searchcat.pl must be run as user www.' |\
140: mailto $emailto -s '$subj' > /dev/null");
141: exit 1;
142: }
143: #
144: # Let people know we are running
145: open(LOG,'>'.$perlvar{'lonDaemons'}.'/logs/searchcat.log');
146: &log(0,'==== Searchcat Run '.localtime()."====");
147:
148:
149: if ($debug) {
150: &log(0,'simulating') if ($simulate);
151: &log(0,'only processing user '.$oneuser) if ($oneuser);
152: &log(0,'verbosity level = '.$verbose);
153: }
154: #
155: # Connect to database
156: my $dbh;
157: if (! ($dbh = DBI->connect("DBI:mysql:loncapa","www",$perlvar{'lonSqlAccess'},
158: { RaiseError =>0,PrintError=>0}))) {
159: &log(0,"Cannot connect to database!");
160: die "MySQL Error: Cannot connect to database!\n";
161: }
162: # This can return an error and still be okay, so we do not bother checking.
163: # (perhaps it should be more robust and check for specific errors)
164: $dbh->do('DROP TABLE IF EXISTS '.$newname);
165: #
166: # Create the new table
167: my $request = &LONCAPA::lonmetadata::create_metadata_storage($newname);
168: $dbh->do($request);
169: if ($dbh->err) {
170: $dbh->disconnect();
171: &log(0,"MySQL Error Create: ".$dbh->errstr);
172: die $dbh->errstr;
173: }
174: #
175: # find out which users we need to examine
176: my $dom = $perlvar{'lonDefDomain'};
177: opendir(RESOURCES,"$perlvar{'lonDocRoot'}/res/$dom");
178: my @homeusers =
179: grep {
180: &ishome("$perlvar{'lonDocRoot'}/res/$dom/$_");
181: } grep {
182: !/^\.\.?$/;
183: } readdir(RESOURCES);
184: closedir RESOURCES;
185: #
186: if ($oneuser) {
187: @homeusers=($oneuser);
188: }
189: #
190: # Loop through the users
191: foreach my $user (@homeusers) {
192: &log(0,"=== User: ".$user);
193: &process_dynamic_metadata($user,$dom);
194: #
195: # Use File::Find to get the files we need to read/modify
196: find(
197: {preprocess => \&only_meta_files,
198: # wanted => \&print_filename,
199: # wanted => \&log_metadata,
200: wanted => \&process_meta_file,
201: },
202: "$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$user");
203: }
204: #
205: # Rename the table
206: if (! $simulate) {
207: $dbh->do('DROP TABLE IF EXISTS '.$oldname);
208: if (! $dbh->do('RENAME TABLE '.$newname.' TO '.$oldname)) {
209: &log(0,"MySQL Error Rename: ".$dbh->errstr);
210: die $dbh->errstr;
211: } else {
212: &log(1,"MySQL table rename successful.");
213: }
214: }
215:
216: if (! $dbh->disconnect) {
217: &log(0,"MySQL Error Disconnect: ".$dbh->errstr);
218: die $dbh->errstr;
219: }
220: ##
221: ## Finished!
222: &log(0,"==== Searchcat completed ".localtime()." ====");
223: close(LOG);
224:
225: &write_type_count();
226: &write_copyright_count();
227:
228: exit 0;
229:
230: ##
231: ## Status logging routine. Inputs: $level, $message
232: ##
233: ## $level 0 should be used for normal output and error messages
234: ##
235: ## $message does not need to end with \n. In the case of errors
236: ## the message should contain as much information as possible to
237: ## help in diagnosing the problem.
238: ##
239: sub log {
240: my ($level,$message)=@_;
241: $level = 0 if (! defined($level));
242: if ($verbose >= $level) {
243: print LOG $message.$/;
244: }
245: }
246:
247: ########################################################
248: ########################################################
249: ### ###
250: ### File::Find support routines ###
251: ### ###
252: ########################################################
253: ########################################################
254: ##
255: ## &only_meta_files
256: ##
257: ## Called by File::Find.
258: ## Takes a list of files/directories in and returns a list of files/directories
259: ## to search.
260: sub only_meta_files {
261: my @PossibleFiles = @_;
262: my @ChosenFiles;
263: foreach my $file (@PossibleFiles) {
264: if ( ($file =~ /\.meta$/ && # Ends in meta
265: $file !~ /\.\d+\.[^\.]+\.meta$/ # is not for a prior version
266: ) || (-d $file )) { # directories are okay
267: # but we do not want /. or /..
268: push(@ChosenFiles,$file);
269: }
270: }
271: return @ChosenFiles;
272: }
273:
274: ##
275: ##
276: ## Debugging routines, use these for 'wanted' in the File::Find call
277: ##
278: sub print_filename {
279: my ($file) = $_;
280: my $fullfilename = $File::Find::name;
281: if ($debug) {
282: if (-d $file) {
283: &log(5," Got directory ".$fullfilename);
284: } else {
285: &log(5," Got file ".$fullfilename);
286: }
287: }
288: $_=$file;
289: }
290:
291: sub log_metadata {
292: my ($file) = $_;
293: my $fullfilename = $File::Find::name;
294: return if (-d $fullfilename); # No need to do anything here for directories
295: if ($debug) {
296: &log(6,$fullfilename);
297: my $ref=&metadata($fullfilename);
298: if (! defined($ref)) {
299: &log(6," No data");
300: return;
301: }
302: while (my($key,$value) = each(%$ref)) {
303: &log(6," ".$key." => ".$value);
304: }
305: &count_copyright($ref->{'copyright'});
306: }
307: $_=$file;
308: }
309:
310: ##
311: ## process_meta_file
312: ## Called by File::Find.
313: ## Only input is the filename in $_.
314: sub process_meta_file {
315: my ($file) = $_;
316: my $filename = $File::Find::name; # full filename
317: return if (-d $filename); # No need to do anything here for directories
318: #
319: &log(3,$filename) if ($debug);
320: #
321: my $ref=&metadata($filename);
322: #
323: # $url is the original file url, not the metadata file
324: my $url='/res/'.&declutter($filename);
325: $url=~s/\.meta$//;
326: &log(3," ".$url) if ($debug);
327: #
328: # Ignore some files based on their metadata
329: if ($ref->{'obsolete'}) {
330: &log(3,"obsolete") if ($debug);
331: return;
332: }
333: &count_copyright($ref->{'copyright'});
334: if ($ref->{'copyright'} eq 'private') {
335: &log(3,"private") if ($debug);
336: return;
337: }
338: #
339: # Find the dynamic metadata
340: my %dyn;
341: if ($url=~ m:/default$:) {
342: $url=~ s:/default$:/:;
343: &log(3,"Skipping dynamic data") if ($debug);
344: } else {
345: &log(3,"Retrieving dynamic data") if ($debug);
346: %dyn=&get_dynamic_metadata($url);
347: &count_type($url);
348: }
349: #
350: $ref->{'creationdate'} = &sqltime($ref->{'creationdate'});
351: $ref->{'lastrevisiondate'} = &sqltime($ref->{'lastrevisiondate'});
352: my %Data = (
353: %$ref,
354: %dyn,
355: 'url'=>$url,
356: 'version'=>'current');
357: if (! $simulate) {
358: my ($count,$err) = &LONCAPA::lonmetadata::store_metadata($dbh,$newname,
359: \%Data);
360: if ($err) {
361: &log(0,"MySQL Error Insert: ".$err);
362: }
363: if ($count < 1) {
364: &log(0,"Unable to insert record into MySQL database for $url");
365: }
366: }
367: #
368: # Reset $_ before leaving
369: $_ = $file;
370: }
371:
372: ########################################################
373: ########################################################
374: ### ###
375: ### &metadata($uri) ###
376: ### Retrieve metadata for the given file ###
377: ### ###
378: ########################################################
379: ########################################################
380: sub metadata {
381: my ($uri)=@_;
382: my %metacache=();
383: $uri=&declutter($uri);
384: my $filename=$uri;
385: $uri=~s/\.meta$//;
386: $uri='';
387: if ($filename !~ /\.meta$/) {
388: $filename.='.meta';
389: }
390: my $metastring=&getfile($perlvar{'lonDocRoot'}.'/res/'.$filename);
391: return undef if (! defined($metastring));
392: my $parser=HTML::TokeParser->new(\$metastring);
393: my $token;
394: while ($token=$parser->get_token) {
395: if ($token->[0] eq 'S') {
396: my $entry=$token->[1];
397: my $unikey=$entry;
398: if (defined($token->[2]->{'part'})) {
399: $unikey.='_'.$token->[2]->{'part'};
400: }
401: if (defined($token->[2]->{'name'})) {
402: $unikey.='_'.$token->[2]->{'name'};
403: }
404: if ($metacache{$uri.'keys'}) {
405: $metacache{$uri.'keys'}.=','.$unikey;
406: } else {
407: $metacache{$uri.'keys'}=$unikey;
408: }
409: foreach ( @{$token->[3]}) {
410: $metacache{$uri.''.$unikey.'.'.$_}=$token->[2]->{$_};
411: }
412: if (! ($metacache{$uri.''.$unikey}=$parser->get_text('/'.$entry))){
413: $metacache{$uri.''.$unikey} =
414: $metacache{$uri.''.$unikey.'.default'};
415: }
416: } # End of ($token->[0] eq 'S')
417: }
418: return \%metacache;
419: }
420:
421: ##
422: ## &getfile($filename)
423: ## Slurps up an entire file into a scalar.
424: ## Returns undef if the file does not exist
425: sub getfile {
426: my $file = shift();
427: if (! -e $file ) {
428: return undef;
429: }
430: my $fh=IO::File->new($file);
431: my $contents = '';
432: while (<$fh>) {
433: $contents .= $_;
434: }
435: return $contents;
436: }
437:
438: ########################################################
439: ########################################################
440: ### ###
441: ### Dynamic Metadata ###
442: ### ###
443: ########################################################
444: ########################################################
445: ##
446: ## Dynamic metadata description (incomplete)
447: ##
448: ## For a full description of all fields,
449: ## see LONCAPA::lonmetadata
450: ##
451: ## Field Type
452: ##-----------------------------------------------------------
453: ## count integer
454: ## course integer
455: ## course_list comma separated list of course ids
456: ## avetries real
457: ## avetries_list comma separated list of real numbers
458: ## stdno real
459: ## stdno_list comma separated list of real numbers
460: ## usage integer
461: ## usage_list comma separated list of resources
462: ## goto scalar
463: ## goto_list comma separated list of resources
464: ## comefrom scalar
465: ## comefrom_list comma separated list of resources
466: ## difficulty real
467: ## difficulty_list comma separated list of real numbers
468: ## sequsage scalar
469: ## sequsage_list comma separated list of resources
470: ## clear real
471: ## technical real
472: ## correct real
473: ## helpful real
474: ## depth real
475: ## comments html of all the comments made
476: ##
477: {
478:
479: my %DynamicData;
480: my %Counts;
481:
482: sub process_dynamic_metadata {
483: my ($user,$dom) = @_;
484: undef(%DynamicData);
485: undef(%Counts);
486: #
487: my $prodir = &propath($dom,$user);
488: #
489: # Read in the dynamic metadata
490: my %evaldata;
491: if (! tie(%evaldata,'GDBM_File',
492: $prodir.'/nohist_resevaldata.db',&GDBM_READER(),0640)) {
493: return 0;
494: }
495: #
496: %DynamicData = &LONCAPA::lonmetadata::process_reseval_data(\%evaldata);
497: untie(%evaldata);
498: #
499: # Read in the access count data
500: &log(7,'Reading access count data') if ($debug);
501: my %countdata;
502: if (! tie(%countdata,'GDBM_File',
503: $prodir.'/nohist_accesscount.db',&GDBM_READER(),0640)) {
504: return 0;
505: }
506: while (my ($key,$count) = each(%countdata)) {
507: next if ($key !~ /^$dom/);
508: $key = &unescape($key);
509: &log(8,' Count '.$key.' = '.$count) if ($debug);
510: $Counts{$key}=$count;
511: }
512: untie(%countdata);
513: if ($debug) {
514: &log(7,scalar(keys(%Counts)).
515: " Counts read for ".$user."@".$dom);
516: &log(7,scalar(keys(%DynamicData)).
517: " Dynamic metadata read for ".$user."@".$dom);
518: }
519: #
520: return 1;
521: }
522:
523: sub get_dynamic_metadata {
524: my ($url) = @_;
525: $url =~ s:^/res/::;
526: if (! exists($DynamicData{$url})) {
527: &log(7,' No dynamic data for '.$url) if ($debug);
528: return ();
529: }
530: my %data = &LONCAPA::lonmetadata::process_dynamic_metadata($url,
531: \%DynamicData);
532: # find the count
533: $data{'count'} = $Counts{$url};
534: #
535: # Log the dynamic metadata
536: if ($debug) {
537: while (my($k,$v)=each(%data)) {
538: &log(8," ".$k." => ".$v);
539: }
540: }
541: return %data;
542: }
543:
544: } # End of %DynamicData and %Counts scope
545:
546: ########################################################
547: ########################################################
548: ### ###
549: ### Counts ###
550: ### ###
551: ########################################################
552: ########################################################
553: {
554:
555: my %countext;
556:
557: sub count_type {
558: my $file=shift;
559: $file=~/\.(\w+)$/;
560: my $ext=lc($1);
561: $countext{$ext}++;
562: }
563:
564: sub write_type_count {
565: open(RESCOUNT,'>/home/httpd/html/lon-status/rescount.txt');
566: while (my ($extension,$count) = each(%countext)) {
567: print RESCOUNT $extension.'='.$count.'&';
568: }
569: print RESCOUNT 'time='.time."\n";
570: close(RESCOUNT);
571: }
572:
573: } # end of scope for %countext
574:
575: {
576:
577: my %copyrights;
578:
579: sub count_copyright {
580: $copyrights{@_[0]}++;
581: }
582:
583: sub write_copyright_count {
584: open(COPYCOUNT,'>/home/httpd/html/lon-status/copyrightcount.txt');
585: while (my ($copyright,$count) = each(%copyrights)) {
586: print COPYCOUNT $copyright.'='.$count.'&';
587: }
588: print COPYCOUNT 'time='.time."\n";
589: close(COPYCOUNT);
590: }
591:
592: } # end of scope for %copyrights
593:
594: ########################################################
595: ########################################################
596: ### ###
597: ### Miscellanous Utility Routines ###
598: ### ###
599: ########################################################
600: ########################################################
601: ##
602: ## &ishome($username)
603: ## Returns 1 if $username is a LON-CAPA author, 0 otherwise
604: ## (copied from lond, modification of the return value)
605: sub ishome {
606: my $author=shift;
607: $author=~s/\/home\/httpd\/html\/res\/([^\/]*)\/([^\/]*).*/$1\/$2/;
608: my ($udom,$uname)=split(/\//,$author);
609: my $proname=propath($udom,$uname);
610: if (-e $proname) {
611: return 1;
612: } else {
613: return 0;
614: }
615: }
616:
617: ##
618: ## &propath($udom,$uname)
619: ## Returns the path to the users LON-CAPA directory
620: ## (copied from lond)
621: sub propath {
622: my ($udom,$uname)=@_;
623: $udom=~s/\W//g;
624: $uname=~s/\W//g;
625: my $subdir=$uname.'__';
626: $subdir =~ s/(.)(.)(.).*/$1\/$2\/$3/;
627: my $proname="$perlvar{'lonUsersDir'}/$udom/$subdir/$uname";
628: return $proname;
629: }
630:
631: ##
632: ## &sqltime($timestamp)
633: ##
634: ## Convert perl $timestamp to MySQL time. MySQL expects YYYY-MM-DD HH:MM:SS
635: ##
636: sub sqltime {
637: my ($time) = @_;
638: my $mysqltime;
639: if ($time =~
640: /(\d+)-(\d+)-(\d+) # YYYY-MM-DD
641: \s # a space
642: (\d+):(\d+):(\d+) # HH:MM::SS
643: /x ) {
644: # Some of the .meta files have the time in mysql
645: # format already, so just make sure they are 0 padded and
646: # pass them back.
647: $mysqltime = sprintf('%04d-%02d-%02d %02d:%02d:%02d',
648: $1,$2,$3,$4,$5,$6);
649: } elsif ($time =~ /^\d+$/) {
650: my @TimeData = gmtime($time);
651: # Alter the month to be 1-12 instead of 0-11
652: $TimeData[4]++;
653: # Alter the year to be from 0 instead of from 1900
654: $TimeData[5]+=1900;
655: $mysqltime = sprintf('%04d-%02d-%02d %02d:%02d:%02d',
656: @TimeData[5,4,3,2,1,0]);
657: } elsif (! defined($time) || $time == 0) {
658: $mysqltime = 0;
659: } else {
660: &log(0," sqltime:Unable to decode time ".$time);
661: $mysqltime = 0;
662: }
663: return $mysqltime;
664: }
665:
666: ##
667: ## &declutter($filename)
668: ## Given a filename, returns a url for the filename.
669: sub declutter {
670: my $thisfn=shift;
671: $thisfn=~s/^$perlvar{'lonDocRoot'}//;
672: $thisfn=~s/^\///;
673: $thisfn=~s/^res\///;
674: return $thisfn;
675: }
676:
677: ##
678: ## Escape / Unescape special characters
679: sub unescape {
680: my $str=shift;
681: $str =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack("C",hex($1))/eg;
682: return $str;
683: }
684:
685: sub escape {
686: my $str=shift;
687: $str =~ s/(\W)/"%".unpack('H2',$1)/eg;
688: return $str;
689: }
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