Annotation of loncom/metadata_database/searchcat.pl, revision 1.48
1.1 harris41 1: #!/usr/bin/perl
2: # The LearningOnline Network
3: # searchcat.pl "Search Catalog" batch script
1.16 harris41 4: #
1.48 ! www 5: # $Id: searchcat.pl,v 1.47 2003/12/25 04:06:52 www 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;
68:
1.17 harris41 69: use lib '/home/httpd/lib/perl/';
70: use LONCAPA::Configuration;
71:
1.1 harris41 72: use IO::File;
73: use HTML::TokeParser;
1.6 harris41 74: use DBI;
1.21 www 75: use GDBM_File;
1.24 www 76: use POSIX qw(strftime mktime);
1.1 harris41 77:
1.45 www 78: require "find.pl";
79:
1.1 harris41 80: my @metalist;
1.21 www 81:
1.45 www 82: my $simplestatus='';
1.38 www 83: my %countext=();
1.36 www 84:
1.45 www 85: # ----------------------------------------------------- write out simple status
1.36 www 86: sub writesimple {
87: open(SMP,'>/home/httpd/html/lon-status/mysql.txt');
88: print SMP $simplestatus."\n";
89: close(SMP);
90: }
1.28 harris41 91:
1.38 www 92: sub writecount {
93: open(RSMP,'>/home/httpd/html/lon-status/rescount.txt');
94: foreach (keys %countext) {
95: print RSMP $_.'='.$countext{$_}.'&';
96: }
97: print RSMP 'time='.time."\n";
98: close(RSMP);
99: }
100:
1.45 www 101: # -------------------------------------- counts files with different extensions
1.38 www 102: sub count {
103: my $file=shift;
104: $file=~/\.(\w+)$/;
105: my $ext=lc($1);
106: if (defined($countext{$ext})) {
107: $countext{$ext}++;
108: } else {
109: $countext{$ext}=1;
110: }
111: }
1.31 harris41 112: # ----------------------------------------------------- Un-Escape Special Chars
1.28 harris41 113:
1.31 harris41 114: sub unescape {
115: my $str=shift;
1.21 www 116: $str =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack("C",hex($1))/eg;
1.31 harris41 117: return $str;
118: }
1.21 www 119:
1.31 harris41 120: # -------------------------------------------------------- Escape Special Chars
1.22 www 121:
1.31 harris41 122: sub escape {
123: my $str=shift;
1.22 www 124: $str =~ s/(\W)/"%".unpack('H2',$1)/eg;
1.31 harris41 125: return $str;
126: }
1.28 harris41 127:
1.31 harris41 128: # ------------------------------------------- Code to evaluate dynamic metadata
1.28 harris41 129:
1.31 harris41 130: sub dynamicmeta {
1.30 www 131: my $url=&declutter(shift);
132: $url=~s/\.meta$//;
1.48 ! www 133: my %returnhash=(
! 134: 'count' => 0,
! 135: 'course' => 0,
! 136: 'course_list' => '',
! 137: 'avetries' => 0,
! 138: 'avetries_list' => '',
! 139: 'stdno' => 0,
! 140: 'stdno_list' => '',
! 141: 'usage' => 0,
! 142: 'usage_list' => '',
! 143: 'goto' => 0,
! 144: 'goto_list' => '',
! 145: 'comefrom' => 0,
! 146: 'comefrom_list' => '',
! 147: 'difficulty' => 0,
! 148: 'difficulty_list' => ''
! 149: );
1.30 www 150: my ($adomain,$aauthor)=($url=~/^(\w+)\/(\w+)\//);
1.31 harris41 151: my $prodir=&propath($adomain,$aauthor);
1.45 www 152:
1.44 www 153: # Get metadata except counts
1.45 www 154: if (tie(my %evaldata,'GDBM_File',
1.44 www 155: $prodir.'/nohist_resevaldata.db',&GDBM_READER(),0640)) {
156: my %sum=();
157: my %cnt=();
158: my %concat=();
159: my %listitems=(
160: 'course' => 'add',
161: 'goto' => 'add',
162: 'comefrom' => 'add',
163: 'avetries' => 'avg',
164: 'stdno' => 'add',
165: 'difficulty' => 'avg',
166: 'clear' => 'avg',
167: 'technical' => 'avg',
168: 'helpful' => 'avg',
169: 'correct' => 'avg',
170: 'depth' => 'avg',
171: 'comments' => 'app',
172: 'usage' => 'cnt'
173: );
174:
175: my $regexp=$url;
176: $regexp=~s/(\W)/\\$1/g;
177: $regexp='___'.$regexp.'___([a-z]+)$';
178: while (my ($esckey,$value)=each %evaldata) {
1.45 www 179: my $key=&unescape($esckey);
1.44 www 180: if ($key=~/$regexp/) {
1.45 www 181: my ($item,$purl,$cat)=split(/___/,$key);
1.44 www 182: if (defined($cnt{$cat})) { $cnt{$cat}++; } else { $cnt{$cat}=1; }
183: unless ($listitems{$cat} eq 'app') {
184: if (defined($sum{$cat})) {
1.45 www 185: $sum{$cat}+=$evaldata{$esckey};
1.44 www 186: $concat{$cat}.=','.$item;
187: } else {
1.45 www 188: $sum{$cat}=$evaldata{$esckey};
1.44 www 189: $concat{$cat}=$item;
190: }
191: } else {
192: if (defined($sum{$cat})) {
1.45 www 193: if ($evaldata{$esckey}=~/\w/) {
194: $sum{$cat}.='<hr>'.$evaldata{$esckey};
1.44 www 195: }
196: } else {
1.45 www 197: $sum{$cat}=''.$evaldata{$esckey};
1.44 www 198: }
199: }
1.35 www 200: }
1.44 www 201: }
202: untie(%evaldata);
1.45 www 203: # transfer gathered data to returnhash, calculate averages where applicable
204: while (my $cat=each(%cnt)) {
1.48 ! www 205: if ($cnt{$cat} eq 'nan') { next; }
! 206: if ($sum{$cat} eq 'nan') { next; }
1.45 www 207: if ($listitems{$cat} eq 'avg') {
1.48 ! www 208: if ($cnt{$cat}) {
! 209: $returnhash{$cat}=int(($sum{$cat}/$cnt{$cat})*100.0+0.5)/100.0;
! 210: } else {
! 211: $returnhash{$cat}='NULL';
! 212: }
1.45 www 213: } elsif ($listitems{$cat} eq 'cnt') {
214: $returnhash{$cat}=$cnt{$cat};
215: } else {
216: $returnhash{$cat}=$sum{$cat};
217: }
218: $returnhash{$cat.'_list'}=$concat{$cat};
1.44 www 219: }
220: }
221: # get count
1.45 www 222: if (tie(my %evaldata,'GDBM_File',
1.44 www 223: $prodir.'/nohist_accesscount.db',&GDBM_READER(),0640)) {
1.45 www 224: my $escurl=&escape($url);
225: if (! exists($evaldata{$escurl})) {
1.46 www 226: $returnhash{'count'}=0;
1.44 www 227: } else {
1.45 www 228: $returnhash{'count'}=$evaldata{$escurl};
1.44 www 229: }
230: untie %evaldata;
231: }
232: return %returnhash;
1.30 www 233: }
1.35 www 234:
1.31 harris41 235: # --------------- Read loncapa_apache.conf and loncapa.conf and get variables
236: my $perlvarref=LONCAPA::Configuration::read_conf('loncapa.conf');
237: my %perlvar=%{$perlvarref};
238: undef $perlvarref; # remove since sensitive and not needed
239: delete $perlvar{'lonReceipt'}; # remove since sensitive and not needed
1.28 harris41 240:
1.31 harris41 241: # ------------------------------------- Only run if machine is a library server
242: exit unless $perlvar{'lonRole'} eq 'library';
1.1 harris41 243:
1.31 harris41 244: # ----------------------------- Make sure this process is running from user=www
1.15 harris41 245:
1.31 harris41 246: my $wwwid=getpwnam('www');
247: if ($wwwid!=$<) {
1.45 www 248: my $emailto="$perlvar{'lonAdmEMail'},$perlvar{'lonSysEMail'}";
249: my $subj="LON: $perlvar{'lonHostID'} User ID mismatch";
1.33 matthew 250: system("echo 'User ID mismatch. searchcat.pl must be run as user www.' |\
1.31 harris41 251: mailto $emailto -s '$subj' > /dev/null");
1.33 matthew 252: exit 1;
1.31 harris41 253: }
1.1 harris41 254:
1.27 www 255:
1.31 harris41 256: # ---------------------------------------------------------- We are in business
1.27 www 257:
1.31 harris41 258: open(LOG,'>'.$perlvar{'lonDaemons'}.'/logs/searchcat.log');
259: print LOG '==== Searchcat Run '.localtime()."====\n\n";
1.36 www 260: $simplestatus='time='.time.'&';
1.31 harris41 261: my $dbh;
262: # ------------------------------------- Make sure that database can be accessed
263: {
264: unless (
265: $dbh = DBI->connect("DBI:mysql:loncapa","www",$perlvar{'lonSqlAccess'},{ RaiseError =>0,PrintError=>0})
266: ) {
267: print LOG "Cannot connect to database!\n";
1.36 www 268: $simplestatus.='mysql=defunct';
269: &writesimple();
1.31 harris41 270: exit;
271: }
1.36 www 272:
1.47 www 273: # Make temporary table
274: $dbh->do("DROP TABLE IF EXISTS newmetadata");
275: my $make_metadata_table = "CREATE TABLE IF NOT EXISTS newmetadata (".
1.31 harris41 276: "title TEXT, author TEXT, subject TEXT, url TEXT, keywords TEXT, ".
277: "version TEXT, notes TEXT, abstract TEXT, mime TEXT, language TEXT, ".
278: "creationdate DATETIME, lastrevisiondate DATETIME, owner TEXT, ".
1.47 www 279: "copyright TEXT, ".
280: "count INTEGER UNSIGNED, ".
281: "course INTEGER UNSIGNED, course_list TEXT, ".
282: "goto INTEGER UNSIGNED, goto_list TEXT, ".
283: "comefrom INTEGER UNSIGNED, comefrom_list TEXT, ".
1.48 ! www 284: "sequsage INTEGER UNSIGNED, sequsage_list TEXT, ".
1.47 www 285: "stdno INTEGER UNSIGNED, stdno_list TEXT, ".
286: "avetries FLOAT, avetries_list TEXT, ".
287: "difficulty FLOAT, difficulty_list TEXT, ".
288: "FULLTEXT idx_title (title), ".
1.31 harris41 289: "FULLTEXT idx_author (author), FULLTEXT idx_subject (subject), ".
290: "FULLTEXT idx_url (url), FULLTEXT idx_keywords (keywords), ".
291: "FULLTEXT idx_version (version), FULLTEXT idx_notes (notes), ".
292: "FULLTEXT idx_abstract (abstract), FULLTEXT idx_mime (mime), ".
293: "FULLTEXT idx_language (language), FULLTEXT idx_owner (owner), ".
1.47 www 294: "FULLTEXT idx_copyright (copyright)) ".
295: "TYPE=MyISAM";
1.31 harris41 296: # It would sure be nice to have some logging mechanism.
1.47 www 297: unless ($dbh->do($make_metadata_table)) {
298: print LOG "\nMySQL Error Create: ".$dbh->errstr."\n";
299: die $dbh->errstr;
300: }
1.31 harris41 301: }
1.27 www 302:
1.31 harris41 303: # ------------------------------------------------------------- get .meta files
304: opendir(RESOURCES,"$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}");
1.33 matthew 305: my @homeusers = grep {
306: &ishome("$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$_")
307: } grep {!/^\.\.?$/} readdir(RESOURCES);
1.31 harris41 308: closedir RESOURCES;
1.34 matthew 309:
310: #
311: # Create the statement handlers we need
312:
313: my $insert_sth = $dbh->prepare
1.47 www 314: ("INSERT INTO newmetadata VALUES (".
1.34 matthew 315: "?,". # title
316: "?,". # author
317: "?,". # subject
1.47 www 318: "?,". # declutter url
1.34 matthew 319: "?,". # version
320: "?,". # current
321: "?,". # notes
322: "?,". # abstract
323: "?,". # mime
324: "?,". # language
325: "?,". # creationdate
326: "?,". # revisiondate
327: "?,". # owner
1.47 www 328: "?,". # copyright
329: "?,". # count
330: "?,". # course
331: "?,". # course_list
332: "?,". # goto
333: "?,". # goto_list
334: "?,". # comefrom
335: "?,". # comefrom_list
336: "?,". # usage
337: "?,". # usage_list
338: "?,". # stdno
339: "?,". # stdno_list
340: "?,". # avetries
341: "?,". # avetries_list
342: "?,". # difficulty
1.48 ! www 343: "?". # difficulty_list
! 344: ")"
1.34 matthew 345: );
346:
1.31 harris41 347: foreach my $user (@homeusers) {
348: print LOG "\n=== User: ".$user."\n\n";
1.44 www 349:
1.31 harris41 350: my $prodir=&propath($perlvar{'lonDefDomain'},$user);
1.33 matthew 351: # Use find.pl
1.31 harris41 352: undef @metalist;
353: @metalist=();
354: &find("$perlvar{'lonDocRoot'}/res/$perlvar{'lonDefDomain'}/$user");
1.33 matthew 355: # -- process each file to get metadata and put into search catalog SQL
356: # database. Also, check to see if already there.
357: # I could just delete (without searching first), but this works for now.
358: foreach my $m (@metalist) {
359: print LOG "- ".$m."\n";
360: my $ref=&metadata($m);
1.41 www 361: my $m2='/res/'.&declutter($m);
362: $m2=~s/\.meta$//;
1.40 www 363: if ($ref->{'obsolete'}) { print LOG "obsolete\n"; next; }
364: if ($ref->{'copyright'} eq 'private') { print LOG "private\n"; next; }
1.47 www 365: my %dyn=&dynamicmeta($m2);
1.38 www 366: &count($m2);
1.47 www 367: unless ($insert_sth->execute(
368: $ref->{'title'},
1.34 matthew 369: $ref->{'author'},
370: $ref->{'subject'},
371: $m2,
372: $ref->{'keywords'},
373: 'current',
374: $ref->{'notes'},
375: $ref->{'abstract'},
376: $ref->{'mime'},
377: $ref->{'language'},
378: sqltime($ref->{'creationdate'}),
379: sqltime($ref->{'lastrevisiondate'}),
380: $ref->{'owner'},
1.47 www 381: $ref->{'copyright'},
382: $dyn{'count'},
383: $dyn{'course'},
384: $dyn{'course_list'},
385: $dyn{'goto'},
386: $dyn{'goto_list'},
387: $dyn{'comefrom'},
388: $dyn{'comefrom_list'},
389: $dyn{'usage'},
390: $dyn{'usage_list'},
391: $dyn{'stdno'},
392: $dyn{'stdno_list'},
393: $dyn{'avetries'},
394: $dyn{'avetries_list'},
395: $dyn{'difficulty'},
396: $dyn{'difficulty_list'}
397: )) {
398: print LOG "\nMySQL Error Insert: ".$dbh->errstr."\n";
399: die $dbh->errstr;
400: }
1.34 matthew 401: $ref = undef;
1.31 harris41 402: }
403: }
404: # --------------------------------------------------- Close database connection
1.47 www 405: $dbh->do("DROP TABLE IF EXISTS metadata");
406: unless ($dbh->do("RENAME TABLE newmetadata TO metadata")) {
407: print LOG "\nMySQL Error Rename: ".$dbh->errstr."\n";
408: die $dbh->errstr;
409: }
410: unless ($dbh->disconnect) {
411: print LOG "\nMySQL Error Disconnect: ".$dbh->errstr."\n";
412: die $dbh->errstr;
413: }
1.31 harris41 414: print LOG "\n==== Searchcat completed ".localtime()." ====\n";
415: close(LOG);
1.36 www 416: &writesimple();
1.38 www 417: &writecount();
1.31 harris41 418: exit 0;
1.33 matthew 419:
420:
421:
1.31 harris41 422: # =============================================================================
1.1 harris41 423:
1.31 harris41 424: # ---------------------------------------------------------------- Get metadata
425: # significantly altered from subroutine present in lonnet
426: sub metadata {
427: my ($uri,$what)=@_;
1.47 www 428: my %metacache=();
1.31 harris41 429: $uri=&declutter($uri);
430: my $filename=$uri;
431: $uri=~s/\.meta$//;
432: $uri='';
433: unless ($metacache{$uri.'keys'}) {
434: unless ($filename=~/\.meta$/) { $filename.='.meta'; }
435: my $metastring=&getfile($perlvar{'lonDocRoot'}.'/res/'.$filename);
436: my $parser=HTML::TokeParser->new(\$metastring);
437: my $token;
438: while ($token=$parser->get_token) {
1.33 matthew 439: if ($token->[0] eq 'S') {
440: my $entry=$token->[1];
441: my $unikey=$entry;
442: if (defined($token->[2]->{'part'})) {
443: $unikey.='_'.$token->[2]->{'part'};
444: }
445: if (defined($token->[2]->{'name'})) {
446: $unikey.='_'.$token->[2]->{'name'};
447: }
448: if ($metacache{$uri.'keys'}) {
449: $metacache{$uri.'keys'}.=','.$unikey;
450: } else {
451: $metacache{$uri.'keys'}=$unikey;
452: }
453: map {
454: $metacache{$uri.''.$unikey.'.'.$_}=$token->[2]->{$_};
455: } @{$token->[3]};
456: unless (
457: $metacache{$uri.''.$unikey}=$parser->get_text('/'.$entry)
458: ) { $metacache{$uri.''.$unikey}=
459: $metacache{$uri.''.$unikey.'.default'};
460: }
461: }
462: }
1.31 harris41 463: }
464: return \%metacache;
465: }
1.28 harris41 466:
1.31 harris41 467: # ------------------------------------------------------------ Serves up a file
468: # returns either the contents of the file or a -1
469: sub getfile {
1.33 matthew 470: my $file=shift;
471: if (! -e $file ) { return -1; };
472: my $fh=IO::File->new($file);
473: my $a='';
474: while (<$fh>) { $a .=$_; }
475: return $a;
1.31 harris41 476: }
1.28 harris41 477:
1.31 harris41 478: # ------------------------------------------------------------- Declutters URLs
479: sub declutter {
480: my $thisfn=shift;
481: $thisfn=~s/^$perlvar{'lonDocRoot'}//;
482: $thisfn=~s/^\///;
483: $thisfn=~s/^res\///;
484: return $thisfn;
485: }
1.28 harris41 486:
1.31 harris41 487: # --------------------------------------- Is this the home server of an author?
488: # (copied from lond, modification of the return value)
489: sub ishome {
490: my $author=shift;
491: $author=~s/\/home\/httpd\/html\/res\/([^\/]*)\/([^\/]*).*/$1\/$2/;
492: my ($udom,$uname)=split(/\//,$author);
493: my $proname=propath($udom,$uname);
494: if (-e $proname) {
495: return 1;
496: } else {
497: return 0;
498: }
499: }
1.28 harris41 500:
1.31 harris41 501: # -------------------------------------------- Return path to profile directory
502: # (copied from lond)
503: sub propath {
504: my ($udom,$uname)=@_;
505: $udom=~s/\W//g;
506: $uname=~s/\W//g;
507: my $subdir=$uname.'__';
508: $subdir =~ s/(.)(.)(.).*/$1\/$2\/$3/;
509: my $proname="$perlvar{'lonUsersDir'}/$udom/$subdir/$uname";
510: return $proname;
511: }
1.28 harris41 512:
1.31 harris41 513: # ---------------------------- convert 'time' format into a datetime sql format
514: sub sqltime {
1.13 harris41 515: my ($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst) =
1.31 harris41 516: localtime(&unsqltime(@_[0]));
517: $mon++; $year+=1900;
518: return "$year-$mon-$mday $hour:$min:$sec";
519: }
1.28 harris41 520:
1.31 harris41 521: sub maketime {
522: my %th=@_;
1.33 matthew 523: return POSIX::mktime(($th{'seconds'},$th{'minutes'},$th{'hours'},
524: $th{'day'},$th{'month'}-1,
525: $th{'year'}-1900,0,0,$th{'dlsav'}));
1.31 harris41 526: }
1.28 harris41 527:
528:
1.31 harris41 529: #########################################
530: #
531: # Retro-fixing of un-backward-compatible time format
1.28 harris41 532:
1.31 harris41 533: sub unsqltime {
534: my $timestamp=shift;
535: if ($timestamp=~/^(\d+)\-(\d+)\-(\d+)\s+(\d+)\:(\d+)\:(\d+)$/) {
1.33 matthew 536: $timestamp=&maketime('year'=>$1,'month'=>$2,'day'=>$3,
537: 'hours'=>$4,'minutes'=>$5,'seconds'=>$6);
1.31 harris41 538: }
539: return $timestamp;
540: }
1.28 harris41 541:
1.45 www 542: # ----------------- Code to enable 'find' subroutine listing of the .meta files
543:
544: no strict "vars";
545:
546: sub wanted {
547: (($dev,$ino,$mode,$nlink,$uid,$gid) = lstat($_)) &&
548: -f _ &&
549: /^.*\.meta$/ && !/^.+\.\d+\.[^\.]+\.meta$/ &&
550: push(@metalist,"$dir/$_");
551: }
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