Annotation of nsdl/harvestsmete/matti.pl, revision 1.1
1.1 ! www 1: #!/usr/local/bin/perl -w
! 2:
! 3: #
! 4: # cstc.pl - Use OAI MHP to harvest metadata from CSTS in oai_ims format
! 5: #
! 6: # Written by Andy Dong <adong@smete.org> 11/01/2001
! 7: #
! 8:
! 9: use strict;
! 10: use Getopt::Std;
! 11: use Time::HiRes qw(usleep ualarm gettimeofday tv_interval);
! 12:
! 13: use HTTP::Request;
! 14: use LWP::UserAgent;
! 15:
! 16: use XML::Element;
! 17: use XML::Parser;
! 18: use XML::TreeBuilder;
! 19:
! 20: use DBI;
! 21: use DBD::ODBC;
! 22:
! 23: require OAIvocabulary_v2;
! 24: require OAIcataloging_v2;
! 25:
! 26: # -u flag specifies [u]pdate database; otherwise output to STDOUT
! 27:
! 28: my $usage = << "EOT";
! 29: Usage: cstc.pl -u
! 30:
! 31: -u (U)pdate the database
! 32:
! 33: Without -u it simply prints to STDOUT
! 34: EOT
! 35:
! 36: my %args;
! 37: getopts('u', \%args) || die $usage;
! 38:
! 39: my $inserted = 0;
! 40: my $updated = 0;
! 41:
! 42: my $useDatabase = 1 if ($args{'u'});
! 43: my $dbh;
! 44: # Database Configuration
! 45: if ( $useDatabase ) {
! 46: print "Updating the database\n";
! 47: my $DBI_DSN='dbi:ODBC:mel.odbc';
! 48: my $DBI_USER='autocataloger';
! 49: my $DBI_PWD='regolatacotua';
! 50: $dbh= DBI->connect($DBI_DSN, $DBI_USER, $DBI_PWD, { RaiseError => 1, AutoCommit => 0 }) || die "Unable to connect to database $DBI_DSN as $DBI_USER\n";;
! 51: }
! 52:
! 53: my $content;
! 54: my @cstc;
! 55:
! 56: # All possible LOM record variables
! 57: my $logeneralDescription = "";
! 58: my $logeneralTitle1 = "";
! 59: my $logeneralCreation_Date = "";
! 60: my $logeneralPub_Year = "";
! 61: my $logeneralPub_Month = "";
! 62: my $logeneralPub_Day = "";
! 63: my $logeneralLanguage1 = "";
! 64: my $logeneralKeywords;
! 65: my $publisher = "";
! 66: my $personLastname = "";
! 67: my $personFirstname = "";
! 68: my $personEmail = "";
! 69: my $personCompany = "";
! 70: my $platformVersion = "";
! 71: my $platformType = "";
! 72: my $platformOS = "";
! 73: my @pedagogyEndUserRole = ();
! 74: my $pedagogyLContext = "";
! 75: my $platformFormat = "";
! 76: my $platformLocation_URL = "";
! 77: my $role = "";
! 78: my $pedagogy_description = "Unknown";
! 79:
! 80: #my $url = 'http://www.cstc.org/cgi-bin/OAI/CSTC.pl?verb=ListRecords&metadataPrefix=ims1_1';
! 81:
! 82: #my $ua = new LWP::UserAgent;
! 83: #my $request = HTTP::Request->new('GET', $url);
! 84: #my $response = $ua->request( $request );
! 85:
! 86: #if ( $response->is_success ) {
! 87: # $content = $response->content;
! 88: #} else {
! 89: # warn 'OAI request failed: ' . $response->message;
! 90: # exit 1;
! 91: #}
! 92:
! 93: my $tree = XML::TreeBuilder->new();
! 94: $tree->parse_file('matti.xml');
! 95:
! 96: my $t0 = [gettimeofday];
! 97:
! 98: my @records = $tree->find_by_tag_name('record');
! 99: foreach my $record (@records){
! 100: # Extract information from <metametadata> tag
! 101: my $header = $record->find_by_tag_name('metametadata');
! 102: next if ! $header;
! 103: $logeneralCreation_Date = $header->find_by_tag_name('datetime')->as_text;
! 104: ($logeneralPub_Year, $logeneralPub_Month, $logeneralPub_Day) = ($logeneralCreation_Date =~ /^(\d{4})-(\d{2})-(\d{2})$/);
! 105: # Extract information from <general> tag
! 106: my $general = $record->find_by_tag_name('general');
! 107: my $title = $general->find_by_tag_name('title');
! 108: $logeneralTitle1= $title->find_by_tag_name('langstring')->as_text;
! 109: my $description = $general->find_by_tag_name('description');
! 110: $logeneralDescription = $description->find_by_tag_name('langstring')->as_text;
! 111: $logeneralDescription =~ s/\n/ /g;
! 112: $logeneralLanguage1 = $general->find_by_tag_name('language')->as_text;
! 113: my $keywordsElement = $general->find_by_tag_name('keywords');
! 114: my $logeneralKeywords = $keywordsElement->find_by_tag_name('langstring')->as_text;
! 115: # Extract information from <lifecycle> tag
! 116: my $lifecycle = $record->find_by_tag_name('lifecycle');
! 117: my $version = $lifecycle->find_by_tag_name('version');
! 118: $platformVersion = $version->find_by_tag_name('langstring')->as_text;
! 119: my @contributeElement = $lifecycle->find_by_tag_name('contribute');
! 120: foreach my $contribute (@contributeElement) {
! 121: # We will only take the author information
! 122: my $roletype = $contribute->find_by_tag_name('role');
! 123: my $langstring = $roletype->find_by_tag_name('langstring')->as_text;
! 124: if ( $langstring eq "author" ) {
! 125: $role = 'Author';
! 126: my $centity = $contribute->find_by_tag_name('centity');
! 127: my $entity = $centity->find_by_tag_name('vcard')->as_text;
! 128: ($personLastname, $personFirstname, $personEmail, $personCompany) = OAIv_parseVcard_matti($entity);
! 129: }
! 130: }
! 131: # Extract information from <technical> tag
! 132: my $technical = $record->find_by_tag_name('technical');
! 133: my $platform = $technical->find_by_tag_name('format');
! 134: $platformFormat = $technical->find_by_tag_name('langstring')->as_text;
! 135: # All MATTI are 1-Generate Automatically from MIME Type
! 136: $platformFormat = 1;
! 137: $platformLocation_URL = $technical->find_by_tag_name('location')->as_text;
! 138: $platformLocation_URL =~ tr/ //d;
! 139: $platformLocation_URL =~ s/^\n(.*)$/$1/;
! 140: chomp($platformLocation_URL);
! 141: # ALL MATTI are Java applets
! 142: $platformType = 64;
! 143: $platformOS = '';
! 144: # Special Java applet tag to go in version.installation_note
! 145: my $installation_note = $technical->find_by_tag_name('applettag')->as_XML();
! 146: # Extract information from <educational> tag
! 147: my $educational = $record->find_by_tag_name('educational');
! 148: # Use Learner (end_user_type = 2)
! 149: my $pedagogyEndUserType = '2';
! 150: # Learning context must be mapped to grade levels
! 151: my $learningcontextElement = $educational->find_by_tag_name('learningcontext');
! 152: my $langstring = $learningcontextElement->find_by_tag_name('langstring')->as_text;
! 153: my @learningcontext = ($langstring =~ /^(.*),\s(.*)$/);
! 154: $pedagogyLContext = OAIv_findLContext(@learningcontext);
! 155:
! 156: my $difficulty = $educational->find_by_tag_name('difficulty')->as_text;
! 157: my ($difficulty_id) = ($difficulty =~ /^(\d{1})-\w+$/);
! 158: my @pedagogy_description_element = $educational->find_by_tag_name('description');
! 159: foreach my $p_d_e (@pedagogy_description_element) {
! 160: $pedagogy_description = $p_d_e->find_by_tag_name('langstring')->as_text;
! 161: }
! 162: my $interactivity_level = $educational->find_by_tag_name('interactivitylevel')->as_text;
! 163: my ($interactivity_level_id) = ($interactivity_level =~ /^(\d{1})-\w+$/);
! 164: # Java Applets
! 165: my $resource_type_id = 100;
! 166:
! 167:
! 168: if ( $useDatabase ) {
! 169: # Some specific configuration information for MATTI
! 170: # Logo
! 171: my $image = "http://www.smete.org/images/affiliation/matti.gif";
! 172: my $submitter_key = '{710FE693-46E9-4002-BA94-1BE2E6218CD6}'; # Andy Dong
! 173: my $collection = 'National Library of Virtual Manipulatives for Interactive Mathematics';
! 174: my $collection_reg_key = '{8D09C011-23B6-4F23-A690-6C74EFF5E4C7}';
! 175: my $publisher = 'MATTI Associates LLC';
! 176: my $publisher_reg_key = '{120E9D92-6F22-4FEC-A9C0-F95C571174BA}';
! 177: # Determine if this author already exists in the database (person and entity tables)
! 178: my $author_reg_key;
! 179: if ( ! ($author_reg_key = OAIc_personexists($dbh,$personEmail)) ) {
! 180: printf("Inserting person email=%s\n",$personEmail);
! 181: my $success = OAIc_insert_person($dbh,$publisher_reg_key,$submitter_key,$personLastname,$personFirstname,$personEmail,$personCompany);
! 182: $author_reg_key = OAIc_personexists($dbh,$personEmail);
! 183: }
! 184: if ( my $general_key = OAIc_loexists($dbh,$logeneralTitle1) ) {
! 185: my $success = OAIc_update_matti($dbh, $general_key, $installation_note);
! 186: # my $success = OAIc_update_lo($dbh, $general_key, $logeneralTitle1, $logeneralLanguage1, $logeneralDescription, $image, $logeneralPub_Month, $logeneralPub_Year, $logeneralKeywords, $submitter_key, join(" ", $personFirstname, $personLastname), $publisher, $collection, $platformFormat, $platformType, $platformOS, $platformLocation_URL, $pedagogyLContext, $pedagogyEndUserType, $author_reg_key, $publisher_reg_key, $collection_reg_key);
! 187: $updated = $updated + 1;
! 188: } else {
! 189: printf("Inserting new record for %s\n",$logeneralTitle1);
! 190: my $success = OAIc_insert_lo($dbh, $logeneralTitle1, $logeneralLanguage1, $logeneralDescription, $image, $logeneralPub_Month, $logeneralPub_Year, $logeneralKeywords, $submitter_key, join(" ", $personFirstname, $personLastname), $publisher, $collection, $platformFormat, $platformType, $platformOS, $platformLocation_URL, $pedagogyLContext, $pedagogyEndUserType, $author_reg_key, $publisher_reg_key, $collection_reg_key, $difficulty_id, $interactivity_level_id, $pedagogy_description, $resource_type_id);
! 191: $inserted = $inserted + 1;
! 192: }
! 193: } else {
! 194: # Print Results
! 195: printf("Title: %s\tDescription: %s\tKeywords: %s\n", $logeneralTitle1,$logeneralDescription, $logeneralKeywords);
! 196: printf("Creation Date: %s\tPublication Year: %4d\tPublication Month: %02d\n", $logeneralCreation_Date, $logeneralPub_Year, $logeneralPub_Month);
! 197: printf("Role: %s\n", $role);
! 198: printf("Firstname: %s\tLastname: %s\tEmail: %s\tOrganization: %s\n", $personFirstname, $personLastname, $personEmail, $personCompany);
! 199: printf("Language: %s\n", $logeneralLanguage1);
! 200: printf("Format: %s\tURL: %s\tPlatform: %s\tOS: %s\n", $platformFormat, $platformLocation_URL,$platformType,$platformOS);
! 201: printf("IntendedEndUserRole: %s\tLearningContext: %s\n", $pedagogyEndUserType, $pedagogyLContext);
! 202: printf("Pedagogy Description: %s\tDifficulty id: %d\tInteractivity Level id: %d\n", $pedagogy_description, $difficulty_id, $interactivity_level_id);
! 203: printf("Installation Note: %s\n", $installation_note);
! 204: }
! 205: } # end for loop
! 206: $tree->delete;
! 207:
! 208: if ( $useDatabase ) {
! 209: $dbh->commit;
! 210: $dbh->disconnect();
! 211: }
! 212:
! 213: printf("Inserted %d records and Updated %d records in %f seconds.\n", $inserted, $updated, tv_interval($t0));
! 214:
! 215: exit 0;
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/icons/back.gif){kind=icon}
Return to matti.pl CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [LON-CAPA] / nsdl / harvestsmete