source: main/trunk/greenstone2/perllib/plugins/GreenstoneSQLPlugin.pm@ 32643

###########################################################################
#
# GreenstoneSQLPlugin.pm -- reads into doc_obj from SQL db and docsql.xml
# Metadata and/or fulltext are stored in SQL db, the rest may be stored in
# the docsql .xml files.
# A component of the Greenstone digital library software
# from the New Zealand Digital Library Project at the 
# University of Waikato, New Zealand.
#
# Copyright (C) 2001 New Zealand Digital Library Project
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
#
###########################################################################

package GreenstoneSQLPlugin;


use strict;
no strict 'refs'; # allow filehandles to be variables and viceversa

use DBI;
use docprint; # for new unescape_text() subroutine
use GreenstoneXMLPlugin;
use gsmysql;

#use unicode;
#use Encode;

########################################################################################

# GreenstoneSQLPlugin inherits from GreenstoneXMLPlugin so that if meta or fulltext
# is still written out to doc.xml (docsql .xml), that will be processed as usual,
# whereas GreenstoneSQLPlugin will process all the rest (full text and/or meta, whichever
# is written out by GreenstoneSQLPlugout into the SQL db).


sub BEGIN {
    @GreenstoneSQLPlugin::ISA = ('GreenstoneXMLPlugin');
}

# This plugin must be in the document plugins pipeline IN PLACE OF GreenstoneXMLPlugin
# So we won't have a process exp conflict here.
# The structure of docsql.xml files is identical to doc.xml and the contents are similar except:
#   - since metadata and/or fulltxt are stored in mysql db instead, just XML comments indicating
#   this are left inside docsql.xml within the <Description> (for meta) and/or <Content> (for txt)
#   - the root element Archive now has a docoid attribute: <Archive docoid="OID">
sub get_default_process_exp {
    my $self = shift (@_);

    return q^(?i)docsql(-\d+)?\.xml$^; # regex based on this method in GreenstoneXMLPlugin
}

my $process_mode_list =
    [ { 'name' => "meta_only",
        'desc' => "{GreenstoneSQLPlug.process_mode.meta_only}" },      
      { 'name' => "text_only",
        'desc' => "{GreenstoneSQLPlug.process_mode.text_only}" },
      { 'name' => "all",
        'desc' => "{GreenstoneSQLPlug.process_mode.all}" } ];

my $rollback_on_cancel_list =
    [ { 'name' => "true",
        'desc' => "{GreenstoneSQLPlug.rollback_on_cancel}" },      
      { 'name' => "false",
        'desc' => "{GreenstoneSQLPlug.rollbacl_on_cancel}" } ];

# NOTE: If subclassing gsmysql for other supporting databases and if they have different required
# connection parameters, we can check how WordPlugin, upon detecting Word is installed,
# dynamically loads Word specific configuration options.
my $arguments =
    [ { 'name' => "process_exp",
    'desc' => "{BaseImporter.process_exp}",
    'type' => "regexp",
    'deft' => &get_default_process_exp(),
    'reqd' => "no" },
      { 'name' => "process_mode", 
    'desc' => "{GreenstoneSQLPlug.process_mode}",
    'type' => "enum",
    'list' => $process_mode_list,
    'deft' => "all",
    'reqd' => "no"},
      { 'name' => "rollback_on_cancel", 
    'desc' => "{GreenstoneSQLPlug.rollback_on_cancel}",
    'type' => "enum",
    'list' => $rollback_on_cancel_list,
    'deft' => "false", # better default than true
    'reqd' => "no",
    'hiddengli' => "no"},
      { 'name' => "db_driver", 
    'desc' => "{GreenstoneSQLPlug.db_driver}",
    'type' => "string", 
    'deft' => "mysql",
    'reqd' => "yes"},
      { 'name' => "db_client_user", 
    'desc' => "{GreenstoneSQLPlug.db_client_user}",
    'type' => "string", 
    'deft' => "root",
    'reqd' => "yes"},
      { 'name' => "db_client_pwd", 
    'desc' => "{GreenstoneSQLPlug.db_client_pwd}",
    'type' => "string",
    'deft' => "",
    'reqd' => "no"}, # pwd not required: can create mysql accounts without pwd
      { 'name' => "db_host", 
    'desc' => "{GreenstoneSQLPlug.db_host}",
    'type' => "string",
    'deft' => "127.0.0.1", # NOTE: make this int? No default for port, since it's not a required connection param
    'reqd' => "yes"},
      { 'name' => "db_port", 
    'desc' => "{GreenstoneSQLPlug.db_port}",
    'type' => "string", # NOTE: make this int? No default for port, since it's not a required connection param
    'reqd' => "no"}
    ];

my $options = { 'name'     => "GreenstoneSQLPlugin",
        'desc'     => "{GreenstoneSQLPlugin.desc}",
        'abstract' => "no",
        'inherits' => "yes",
            'args'     => $arguments };


###### Methods called during buildcol and import #######

sub new {
    my ($class) = shift (@_);
    my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
    push(@$pluginlist, $class);

    push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
    push(@{$hashArgOptLists->{"OptList"}},$options);

    my $self = new GreenstoneXMLPlugin($pluginlist, $inputargs, $hashArgOptLists);

    
    #return bless $self, $class;
    $self = bless $self, $class;
    if ($self->{'info_only'}) {
    # If running pluginfo, we don't need to go further.
    return $self;
    }

    # do anything else that needs to be done here when not pluginfo
    
    return $self;
}

# Call init() not begin() because there can be multiple plugin passes and begin() called for
# each pass (one for doc level and another for section level indexing), whereas init() should
# be called before any and all passes.
# This way, we can connect to the SQL database once per buildcol run.
# Although now it doesn't matter, since gsmysql.pm uses the get_instance pattern to return a
# singleton db connection, regardless of the number of gsmysql objects instantiated and
# the number of connect() calls made on them.
sub init {
    my ($self) = shift (@_);
    
    $self->SUPER::init(@_); # super (GreenstoneXMLPlugin) will not yet be trying to read from doc.xml (docsql .xml) files in init().


    # create gsmysql object.
    # collection name will be used for naming tables (site name will be used for naming database)
    my $gs_sql = new gsmysql({
    'collection_name' => $ENV{'GSDLCOLLECTION'},
    'verbosity' => $self->{'verbosity'} || 0
               });
    
    # if autocommit is set, there's no rollback support
    my $autocommit = ($self->{'rollback_on_cancel'} eq "false") ? 1 : 0;

    # try connecting to the mysql db, die if that fails
    if(!$gs_sql->connect_to_db({
    'db_driver' => $self->{'db_driver'},
    'db_client_user' => $self->{'db_client_user'},
    'db_client_pwd' => $self->{'db_client_pwd'},
    'db_host' => $self->{'db_host'},
    'db_port' => $self->{'db_port'}, # undef by default, can leave as is
    'autocommit' => $autocommit
                   })
    )
    {
    # This is fatal for the plugout, let's terminate here
    # PrintError would already have displayed the warning message on connection fail    
    die("Could not connect to db. Can't proceed.\n");
    }
    
    my $db_name = $self->{'site'} || "greenstone2"; # one database per GS3 site, for GS2 the db is called greenstone2

    # Attempt to use the db, create it if it doesn't exist (but don't create the tables yet)
    # Bail if we can't use the database
    if(!$gs_sql->use_db($db_name)) {
    
    # This is fatal for the plugout, let's terminate here after disconnecting again
    # PrintError would already have displayed the warning message on load fail
    # And on die() perl will call gsmysql destroy which will ensure a disconnect() from db
    die("Could not use db $db_name. Can't proceed.\n");
    }
    
    
    # store db handle now that we're connected
    $self->{'gs_sql'} = $gs_sql;    
}


# This method also runs on import.pl if gs_sql has a value.
# Call deinit() not end() because there can be multiple plugin passes:
# one for doc level and another for section level indexing
# and deinit() should be called before all passes
# This way, we can close the SQL database once per buildcol run.
# Again, this doesn't matter because we gsmysql the ensures the connection
# is a singleton connection instance, which connects once and disconnects once per perl process.
sub deinit {
    my ($self) = shift (@_);
    
    if($self->{'gs_sql'}) {

    # Important to call finished():
    # it will disconnect from db if this is the last gsmysql instance,
    # and it will commit to db before disconnecting if rollbback_on_cancel turned on
    $self->{'gs_sql'}->finished();

    # Clear gs_sql (setting key to undef has a different meaning from deleting:
    # undef makes key still exist but its value is unded whereas delete deletes the key)
    # So all future use has to make the connection again
    delete $self->{'gs_sql'};
    }

    $self->SUPER::deinit(@_);
}



###### Methods only called during import.pl #####

# This is called once if removeold is set with import.pl. Most plugins will do
# nothing but if a plugin does any stuff outside of creating doc obj, then 
# it may need to clear something.
# In the case of GreenstoneSQL plugs: this is the first time we have a chance
# to purge the tables of the current collection from the current site's database
sub remove_all {
    my $self = shift (@_);
    my ($pluginfo, $base_dir, $processor, $maxdocs) = @_;

    $self->SUPER::remove_all(@_);
    
    print STDERR "   Building with removeold option set, so deleting current collection's tables if they exist\n" if($self->{'verbosity'});
    
    # if we're in here, we'd already have run 'use database <site>;' during sub init()
    # so we can go ahead and delete the collection's tables
    my $gs_sql = $self->{'gs_sql'};
    $gs_sql->delete_collection_tables(); # will delete them if they exist

    # and recreate tables? No. Tables' existence is ensured in GreenstoneSQLPlugout::begin()
    my $proc_mode = $self->{'process_mode'};
    if($proc_mode ne "text_only") {
    $gs_sql->ensure_meta_table_exists();
    }
    if($proc_mode ne "meta_only") {
    $gs_sql->ensure_fulltxt_table_exists();
    }

}

# This is called during import.pl per document for docs that have been deleted from the 
# collection. Most plugins will do nothing
# but if a plugin does any stuff outside of creating doc obj, then it may need
# to clear something.
# In the case of GreenstoneSQL plugs: Remove the doc(s) denoted by oids from GS SQL db.
# This takes care of incremental deletes (docs marked D by ArchivesInfPlugin when building
# incrementally) as well as cases of "Non-icremental Delete", see ArchivesInfPlugin.pm.
# As well as cases involving reindexing, which are implemented here as delete followed by add.
sub remove_one {
    my $self = shift (@_);
    
    my ($file, $oids, $archivedir) = @_;

    my $rv = $self->SUPER::remove_one(@_);
    
    print STDERR "@@@ IN SQLPLUG::REMOVE_ONE: $file\n";
    
    #return undef unless $self->can_process_this_file($file); # NO, DON'T DO THIS (inherited remove_one behaviour) HERE: 
           # WE DON'T CARE IF IT'S AN IMAGE FILE THAT WAS DELETED.
           # WE CARE ABOUT REMOVING THE DOC_OID OF THAT IMAGE FILE FROM THE SQL DB
           # SO DON'T RETURN IF CAN'T_PROCESS_THIS_FILE
    
    
    my $gs_sql = $self->{'gs_sql'} || return 0; # couldn't make the connection or no db etc
    
    my $proc_mode = $self->{'process_mode'};
    foreach my $oid (@$oids) {  
    if($proc_mode eq "all" || $proc_mode eq "meta_only") {
        print STDERR "@@@@@@@@ Deleting $oid from meta table\n" if $self->{'verbosity'} > 2;
        $gs_sql->delete_recs_from_metatable_with_docid($oid);
    }
    if($proc_mode eq "all" || $proc_mode eq "text_only") {
        print STDERR "@@@@@@@@ Deleting $oid from fulltxt table\n" if $self->{'verbosity'} > 2;
        $gs_sql->delete_recs_from_texttable_with_docid($oid);
    }
    }
    return $rv;
}

##### Methods called only during buildcol #####

sub xml_start_tag {
    my $self = shift(@_);
    my ($expat, $element) = @_;

    my $outhandle = $self->{'outhandle'};
    
    $self->{'element'} = $element;
    if ($element eq "Archive") { # docsql.xml files contain a OID attribute on Archive element
    # the element's attributes are in %_ as per ReadXMLFile::xml_start_tag() (while $_
    # contains the tag)

    # Don't access %_{'docoid'} directly: keep getting a warning message to 
    # use $_{'docoid'} for scalar contexts, but %_ is the element's attr hashmap 
    # whereas $_ has the tag info. So we don't want to do $_{'docoid'}.
    my %attr_hash = %_; # right way, see OAIPlugin.pm
    $self->{'doc_oid'} = $attr_hash{'docoid'};
    print $outhandle "Extracted OID from docsql.xml: ".$self->{'doc_oid'}."\n"
        if $self->{'verbosity'} > 2;

    }
    else { # let superclass GreenstoneXMLPlugin continue to process <Section> and <Metadata> elements
    $self->SUPER::xml_start_tag(@_);
    }
}

# There are multiple passes processing the document (see buildcol's mode parameter description):
# - compressing the text which may be a dummy pass for lucene/solr, wherein they still want the
# docobj for different purposes,
# - the pass(es) for indexing, e.g. doc/didx and section/sidx level passes
# - and an infodb pass for processing the classifiers. This pass too needs the docobj
# Since all passes need the doc_obj, all are read in from docsql + SQL db into the docobj in memory

# We only ever get here or do any parsing of the docsql.xml file during the buildcol.pl phase.
# At the end of superclass GreenstoneXMLPlugin.pm's close_document() method,
# the doc_obj in memory is processed (indexed) and then made undef.
# So we have to work with doc_obj before superclass close_document() is finished.
sub close_document {
    my $self = shift(@_);

    my $gs_sql = $self->{'gs_sql'};
    
    my $outhandle = $self->{'outhandle'};
    my $doc_obj = $self->{'doc_obj'};

    my $oid = $self->{'doc_oid'}; # we stored current doc's OID during sub xml_start_tag()
    my $proc_mode = $self->{'process_mode'};
    
    # For now, we have access to doc_obj (until just before super::close_document() terminates)

    # OID parsed of docsql.xml file does need to be set on $doc_obj, as noticed in this case:
    # when a doc in import is renamed, and you do incremental import, it is marked for reindexing 
    # (reindexing is implemented by this plugin as a delete followed by add into the sql db).
    # In that case, UNLESS you set the OID at this stage, the old deleted doc id (for the old doc
    # name) continues to exist in the index at the end of incremental rebuilding if you were to
    # browse the rebuilt collection by files/titles. So unless you set the OID here, the deleted
    # doc oids will still be listed in the index.
    
    print STDERR "   GreenstoneSQLPlugin processing doc $oid (reading into docobj from SQL db)\n"
        if $self->{'verbosity'};
    
    my $build_proc_mode = $self->{'processor'}->get_mode(); # can be "text" as per basebuildproc or infodb
    if($build_proc_mode =~ m/(delete)$/) {
    # build_proc_mode could be "(infodb|text)(delete|reindex)"
    # "...delete" or "...reindex" as per ArchivesInfPlugin
    # But reindex is implemented as delete for GreenstoneSQLPlugs, so that's all we see here?
    print STDERR "   DOC $oid WAS MARKED FOR DELETION. Won't attempt to retrieve from SQL db.\n" if $self->{'verbosity'};
    $self->{'doc_obj'}->set_OID($oid); # oid is all we care about for a doc marked for deletion
    $self->SUPER::close_document(@_); # at the end of this method, doc will have been deleted
    return;  # oid of doc marked for deletion is not in the SQL db, don't bother looking it up
    }

    # else, doc denoted by oid was not marked for deletion, look up its oid in db and read it into doc obj
    
    if($proc_mode eq "all" || $proc_mode eq "meta_only") {
    # read in meta for the collection (i.e. select * from <col>_metadata table
    
    my $records = $gs_sql->select_from_metatable_matching_docid($oid, $outhandle);
    
    print $outhandle "----------SQL DB contains meta-----------\n" if $self->{'verbosity'} > 2;

    foreach my $row (@$records) {
        my ($primary_key, $did, $sid, $metaname, $metaval) = @$row;
        
        # get rid of the artificial "root" introduced in section id when saving to sql db
        $sid =~ s@^root@@;
        $sid = $doc_obj->get_top_section() unless $sid;
        print $outhandle "### did: $did, sid: |$sid|, meta: $metaname, val: $metaval\n"
        if $self->{'verbosity'} > 2;
        
        # We're only dealing with utf8 data where docobj is concerned
        # Data stored unescaped in db: escaping only for html/xml files, not for txt files or db
        $doc_obj->add_utf8_metadata($sid, $metaname, $metaval);
    }
    
    # UTF8 debugging, e.g. if we have a with macron in dc.Title
    ##binmode(STDERR, ":utf8"); # not necessary when calling utf8decomp to print wide characters as unicode codept, avoiding 'wide character' warnings.
    #my $dcTitle = $doc_obj->get_metadata_element($doc_obj->get_top_section(), "dc.Title"); 
    #print STDERR "###### dc.Title: ".&unicode::utf8decomp($dcTitle)."\n";
    #print STDERR "###### DECODED dc.Title: ".&unicode::utf8decomp(&Encode::decode_utf8($dcTitle))."\n"; # decoding was needed for perl 5.18
    
    print $outhandle "----------FIN READING DOC's META FROM SQL DB------------\n"
        if $self->{'verbosity'} > 2;
    }
    
    if($proc_mode eq "all" || $proc_mode eq "text_only") {
    # read in fulltxt for the collection (i.e. select * from <col>_fulltxt table
    
    my $fulltxt_table = $gs_sql->get_fulltext_table_name();
    
    
    my $records = $gs_sql->select_from_texttable_matching_docid($oid, $outhandle);
    
    
    print $outhandle "----------\nSQL DB contains txt entries for-----------\n"
        if $self->{'verbosity'} > 2;

    foreach my $row (@$records) {
        my ($primary_key, $did, $sid, $text) = @$row;
        
        # get rid of the artificial "root" introduced in section id when saving to sql db
        #$sid =~ s@^root@@;
        $sid = $doc_obj->get_top_section() if ($sid eq "root");
        print $outhandle "### did: $did, sid: |$sid|, fulltext: <TXT>\n"
        if $self->{'verbosity'} > 2;

        # We're only dealing with utf8 data where docobj is concerned
        # Data stored unescaped in db: escaping is only for html/xml files, not for txt files or db
        $doc_obj->add_utf8_textref($sid, \$text);
    }   
    print $outhandle "----------FIN READING DOC's TXT FROM SQL DB------------\n"
        if $self->{'verbosity'} > 2;
    }
    

    # setting OID here instead of before reading from SQL db into docobj, will prevent duplicate values for Identifier
    # since doc::set_OID() calls doc::set_metadata_element() for metadata that can't occur more than once
    $self->{'doc_obj'}->set_OID($oid); # may only be necessary if doc was marked for deletion so that SUPER::close_document knows
                                       # the oid of marked doc to remove from index
    
    # done reading into docobj from SQL db
    
    # don't forget to clean up on close() in superclass
    # It will get the doc_obj indexed then make it undef
    $self->SUPER::close_document(@_);
}


1;