--- trunk/lib/WebPAC/Input.pm 2005/12/18 21:06:39 285 +++ trunk/lib/WebPAC/Input.pm 2008/08/02 23:46:41 1100 @@ -7,42 +7,47 @@ use WebPAC::Common; use base qw/WebPAC::Common/; -use Text::Iconv; +use Data::Dump qw/dump/; +use Encode qw/decode from_to/; =head1 NAME -WebPAC::Input - core module for input file format - -=head1 VERSION - -Version 0.02 +WebPAC::Input - read different file formats into WebPAC =cut -our $VERSION = '0.02'; +our $VERSION = '0.19'; =head1 SYNOPSIS -This module is used as base class for all database specific modules -(basically, files which have one handle, fixed size while indexing and some -kind of numeric idefinirier which goes from 1 to filesize). +This module implements input as database which have fixed and known +I while indexing and single unique numeric identifier for database +position ranging from 1 to I. + +Simply, something that is indexed by unmber from 1 .. I. + +Examples of such databases are CDS/ISIS files, MARC files, lines in +text file, and so on. + +Specific file formats are implemented using low-level interface modules, +located in C namespace which export C, +C and optional C functions. Perhaps a little code snippet. - use WebPAC::Input; + use WebPAC::Input; + + my $db = WebPAC::Input->new( + module => 'WebPAC::Input::ISIS', + ); + + $db->open( path => '/path/to/database' ); + print "database size: ",$db->size,"\n"; + while (my $rec = $db->fetch) { + # do something with $rec + } + - my $db = WebPAC::Input->new( - format => 'NULL', - config => $config, - lookup => $lookup_obj, - low_mem => 1, - ); - - $db->open('/path/to/database'); - print "database size: ",$db->size,"\n"; - while (my $row = $db->fetch) { - ... - } =head1 FUNCTIONS @@ -51,16 +56,21 @@ Create new input database object. my $db = new WebPAC::Input( - format => 'NULL' - code_page => 'ISO-8859-2', - low_mem => 1, + module => 'WebPAC::Input::MARC', + recode => 'char pairs', + no_progress_bar => 1, + input_config => { + mapping => [ 'foo', 'bar', 'baz' ], + }, ); -Optional parametar C specify application code page (which will be -used internally). This should probably be your terminal encoding, and by -default, it C. +C is low-level file format module. See L and +L. + +C is optional string constisting of character or words pairs that +should be replaced in input stream. -Default is not to use C options (see L below). +C disables progress bar output on C This function will also call low-level C if it exists with same parametars. @@ -74,69 +84,81 @@ my $log = $self->_get_logger; - # check if required subclasses are implemented - foreach my $subclass (qw/open_db fetch_rec/) { - $log->logdie("missing implementation of $subclass") unless ($self->SUPER::can($subclass)); - } - - if ($self->can('init')) { - $log->debug("calling init"); - $self->init(@_); - } + $log->logconfess("code_page argument is not suppored any more.") if $self->{code_page}; + $log->logconfess("encoding argument is not suppored any more.") if $self->{encoding}; + $log->logconfess("lookup argument is not suppored any more. rewrite call to lookup_ref") if $self->{lookup}; + $log->logconfess("low_mem argument is not suppored any more. rewrite it to load_row and save_row") if $self->{low_mem}; + + $log->logconfess("specify low-level file format module") unless ($self->{module}); + my $module_path = $self->{module}; + $module_path =~ s#::#/#g; + $module_path .= '.pm'; + $log->debug("require low-level module $self->{module} from $module_path"); - $self->{'code_page'} ||= 'ISO-8859-2'; + require $module_path; - # running with low_mem flag? well, use DBM::Deep then. - if ($self->{'low_mem'}) { - $log->info("running with low_mem which impacts performance (<32 Mb memory usage)"); + $self ? return $self : return undef; +} - my $db_file = "data.db"; +=head2 open - if (-e $db_file) { - unlink $db_file or $log->logdie("can't remove '$db_file' from last run"); - $log->debug("removed '$db_file' from last run"); - } +This function will read whole database in memory and produce lookups. - require DBM::Deep; + my $store; # simple in-memory hash - my $db = new DBM::Deep $db_file; + $input->open( + path => '/path/to/database/file', + input_encoding => 'cp852', + strict_encoding => 0, + limit => 500, + offset => 6000, + stats => 1, + lookup_coderef => sub { + my $rec = shift; + # store lookups + }, + modify_records => { + 900 => { '^a' => { ' : ' => '^b' } }, + 901 => { '*' => { '^b' => ' ; ' } }, + }, + modify_file => 'conf/modify/mapping.map', + save_row => sub { + my $a = shift; + $store->{ $a->{id} } = $a->{row}; + }, + load_row => sub { + my $a = shift; + return defined($store->{ $a->{id} }) && + $store->{ $a->{id} }; + }, - $log->logdie("DBM::Deep error: $!") unless ($db); + ); - if ($db->error()) { - $log->logdie("can't open '$db_file' under low_mem: ",$db->error()); - } else { - $log->debug("using file '$db_file' for DBM::Deep"); - } +By default, C is assumed to be C. - $self->{'db'} = $db; - } +C is optional parametar to position at some offset before reading from database. - $self ? return $self : return undef; -} +C is optional parametar to read just C records from database -=head2 open +C create optional report about usage of fields and subfields -This function will read whole database in memory and produce lookups. +C is closure to called to save data into lookups - $isis->open( - path => '/path/to/database/file', - code_page => '852', - limit_mfn => 500, - start_mfn => 6000, - lookup => $lookup_obj, - ); +C specify mapping from subfields to delimiters or from +delimiters to subfields, as well as oprations on fields (if subfield is +defined as C<*>. -By default, C is assumed to be C<852>. +C is alternative for C above which preserves order and offers +(hopefully) simplier sintax than YAML or perl (see L). This option +overrides C if both exists for same input. -If optional parametar C is set, this will be first MFN to read -from database (so you can skip beginning of your database if you need to). +C and C are low-level implementation of store engine. Calling convention +is documented in example above. -If optional parametar C is set, it will read just 500 records -from database in example above. +C should really default to 1, but it doesn't for now. -Returns size of database, regardless of C and C -parametars, see also C<$isis->size>. +Returns size of database, regardless of C and C +parametars, see also C. =cut @@ -145,87 +167,227 @@ my $arg = {@_}; my $log = $self->_get_logger(); + $log->debug( "arguments: ",dump( $arg )); + + $log->logconfess("encoding argument is not suppored any more.") if $self->{encoding}; + $log->logconfess("code_page argument is not suppored any more.") if $self->{code_page}; + $log->logconfess("lookup argument is not suppored any more. rewrite call to lookup_coderef") if ($arg->{lookup}); + $log->logconfess("lookup_coderef must be CODE, not ",ref($arg->{lookup_coderef})) + if ($arg->{lookup_coderef} && ref($arg->{lookup_coderef}) ne 'CODE'); + + $log->debug( $arg->{lookup_coderef} ? '' : 'not ', "using lookup_coderef"); $log->logcroak("need path") if (! $arg->{'path'}); - my $code_page = $arg->{'code_page'} || '852'; + my $input_encoding = $arg->{'input_encoding'} || $self->{'input_encoding'} || 'cp852'; # store data in object - $self->{'code_page'} = $code_page; - foreach my $v (qw/path start_mfn limit_mfn/) { + foreach my $v (qw/path offset limit/) { $self->{$v} = $arg->{$v} if ($arg->{$v}); } - # create Text::Iconv object - $self->{iconv} = Text::Iconv->new($code_page,$self->{'code_page'}); + if ($arg->{load_row} || $arg->{save_row}) { + $log->logconfess("save_row and load_row must be defined in pair and be CODE") unless ( + ref($arg->{load_row}) eq 'CODE' && + ref($arg->{save_row}) eq 'CODE' + ); + $self->{load_row} = $arg->{load_row}; + $self->{save_row} = $arg->{save_row}; + $log->debug("using load_row and save_row instead of in-memory hash"); + } + + my $filter_ref; + my $recode_regex; + my $recode_map; + + if ($self->{recode}) { + my @r = split(/\s/, $self->{recode}); + if ($#r % 2 != 1) { + $log->logwarn("recode needs even number of elements (some number of valid pairs)"); + } else { + while (@r) { + my $from = shift @r; + my $to = shift @r; + $recode_map->{$from} = $to; + } + + $recode_regex = join '|' => keys %{ $recode_map }; + + $log->debug("using recode regex: $recode_regex"); + } + + } + + my $rec_regex; + if (my $p = $arg->{modify_file}) { + $log->debug("using modify_file $p"); + $rec_regex = $self->modify_file_regexps( $p ); + } elsif (my $h = $arg->{modify_records}) { + $log->debug("using modify_records ", sub { dump( $h ) }); + $rec_regex = $self->modify_record_regexps(%{ $h }); + } + $log->debug("rec_regex: ", sub { dump($rec_regex) }) if ($rec_regex); + + my $class = $self->{module} || $log->logconfess("can't get low-level module name!"); - my ($db, $size) = $self->open_db( + my $ll_db = $class->new( path => $arg->{path}, + input_config => $arg->{input_config} || $self->{input_config}, +# filter => sub { +# my ($l,$f_nr) = @_; +# return unless defined($l); +# $l = decode($input_encoding, $l); +# $l =~ s/($recode_regex)/$recode_map->{$1}/g if ($recode_regex && $recode_map); +# return $l; +# }, + %{ $arg }, ); - unless ($db) { + unless (defined($ll_db)) { $log->logwarn("can't open database $arg->{path}, skipping..."); return; } + my $size = $ll_db->size; + unless ($size) { $log->logwarn("no records in database $arg->{path}, skipping..."); return; } - my $startmfn = 1; - my $maxmfn = $size; + my $from_rec = 1; + my $to_rec = $size; - if (my $s = $self->{start_mfn}) { - $log->info("skipping to MFN $s"); - $startmfn = $s; + if (my $s = $self->{offset}) { + $log->debug("skipping to MFN $s"); + $from_rec = $s; } else { - $self->{start_mfn} = $startmfn; + $self->{offset} = $from_rec; } - if ($self->{limit_mfn}) { - $log->info("limiting to ",$self->{limit_mfn}," records"); - $maxmfn = $startmfn + $self->{limit_mfn} - 1; - $maxmfn = $size if ($maxmfn > $size); + if ($self->{limit}) { + $log->debug("limiting to ",$self->{limit}," records"); + $to_rec = $from_rec + $self->{limit} - 1; + $to_rec = $size if ($to_rec > $size); } # store size for later - $self->{size} = ($maxmfn - $startmfn) ? ($maxmfn - $startmfn + 1) : 0; + $self->{size} = ($to_rec - $from_rec) ? ($to_rec - $from_rec + 1) : 0; + + my $strict_encoding = $arg->{strict_encoding} || $self->{strict_encoding}; ## FIXME should be 1 really - $log->info("processing $self->{size} records in $code_page, convert to $self->{code_page}"); + $log->info("processing $self->{size}/$size records [$from_rec-$to_rec]", + " encoding $input_encoding ", $strict_encoding ? ' [strict]' : '', + $self->{stats} ? ' [stats]' : '', + ); # read database - for (my $mfn = $startmfn; $mfn <= $maxmfn; $mfn++) { + for (my $pos = $from_rec; $pos <= $to_rec; $pos++) { + + $log->debug("position: $pos\n"); - $log->debug("mfn: $mfn\n"); + my $rec = $ll_db->fetch_rec($pos, sub { + my ($l,$f_nr,$debug) = @_; +# return unless defined($l); +# return $l unless ($rec_regex && $f_nr); + + return unless ( defined($l) && defined($f_nr) ); + + warn "-=> $f_nr ## |$l|\n" if ($debug); + $log->debug("-=> $f_nr ## $l"); + + # codepage conversion and recode_regex +# $l = decode($input_encoding, $l, 1); + from_to( $l, $input_encoding, 'utf-8', 1 ); + $l =~ s/($recode_regex)/$recode_map->{$1}/g if ($recode_regex && $recode_map); + + # apply regexps + if ($rec_regex && defined($rec_regex->{$f_nr})) { + $log->logconfess("regexps->{$f_nr} must be ARRAY") if (ref($rec_regex->{$f_nr}) ne 'ARRAY'); + my $c = 0; + foreach my $r (@{ $rec_regex->{$f_nr} }) { + my $old_l = $l; + $log->logconfess("expected regex in ", dump( $r )) unless defined($r->{regex}); + eval '$l =~ ' . $r->{regex}; + if ($old_l ne $l) { + my $d = "|$old_l| -> |$l| "; # . $r->{regex}; + $d .= ' +' . $r->{line} . ' ' . $r->{file} if defined($r->{line}); + $d .= ' ' . $r->{debug} if defined($r->{debug}); + $log->debug("MODIFY $d"); + warn "*** $d\n" if ($debug); + + } + $log->error("error applying regex: $r") if ($@); + } + } + + $log->debug("<=- $f_nr ## |$l|"); + warn "<=- $f_nr ## $l\n" if ($debug); + return $l; + }); - my $rec = $self->fetch_rec( $db, $mfn ); + $log->debug(sub { dump($rec) }); if (! $rec) { - $log->warn("record $mfn empty? skipping..."); + $log->warn("record $pos empty? skipping..."); next; } # store - if ($self->{'low_mem'}) { - $self->{'db'}->put($mfn, $rec); + if ($self->{save_row}) { + $self->{save_row}->({ + id => $pos, + row => $rec, + }); } else { - $self->{'data'}->{$mfn} = $rec; + $self->{data}->{$pos} = $rec; } # create lookup - $self->{'lookup'}->add( $rec ) if ($rec && $self->{'lookup'}); + $arg->{'lookup_coderef'}->( $rec ) if ($rec && $arg->{'lookup_coderef'}); - $self->progress_bar($mfn,$maxmfn); + # update counters for statistics + if ($self->{stats}) { - } + # fetch clean record with regexpes applied for statistics + my $rec = $ll_db->fetch_rec($pos); + + foreach my $fld (keys %{ $rec }) { + $self->{_stats}->{fld}->{ $fld }++; + + #$log->logdie("invalid record fild $fld, not ARRAY") + next unless (ref($rec->{ $fld }) eq 'ARRAY'); + + foreach my $row (@{ $rec->{$fld} }) { + + if (ref($row) eq 'HASH') { + + foreach my $sf (keys %{ $row }) { + next if ($sf eq 'subfields'); + $self->{_stats}->{sf}->{ $fld }->{ $sf }->{count}++; + $self->{_stats}->{sf}->{ $fld }->{ $sf }->{repeatable}++ + if (ref($row->{$sf}) eq 'ARRAY'); + } + + } else { + $self->{_stats}->{repeatable}->{ $fld }++; + } + } + } + } + + $self->progress_bar($pos,$to_rec) unless ($self->{no_progress_bar}); - $self->{'current_mfn'} = -1; - $self->{'last_pcnt'} = 0; + } - $log->debug("max mfn: $maxmfn"); + $self->{pos} = -1; + $self->{last_pcnt} = 0; # store max mfn and return it. - $self->{'max_mfn'} = $maxmfn; + $self->{max_pos} = $to_rec; + $log->debug("max_pos: $to_rec"); + + # save for dump + $self->{ll_db} = $ll_db; return $size; } @@ -246,30 +408,30 @@ my $log = $self->_get_logger(); - $log->logconfess("it seems that you didn't load database!") unless ($self->{'current_mfn'}); + $log->logconfess("it seems that you didn't load database!") unless ($self->{pos}); - if ($self->{'current_mfn'} == -1) { - $self->{'current_mfn'} = $self->{'start_mfn'}; + if ($self->{pos} == -1) { + $self->{pos} = $self->{offset}; } else { - $self->{'current_mfn'}++; + $self->{pos}++; } - my $mfn = $self->{'current_mfn'}; + my $mfn = $self->{pos}; - if ($mfn > $self->{'max_mfn'}) { - $self->{'current_mfn'} = $self->{'max_mfn'}; + if ($mfn > $self->{max_pos}) { + $self->{pos} = $self->{max_pos}; $log->debug("at EOF"); return; } - $self->progress_bar($mfn,$self->{'max_mfn'}); + $self->progress_bar($mfn,$self->{max_pos}) unless ($self->{no_progress_bar}); my $rec; - if ($self->{'low_mem'}) { - $rec = $self->{'db'}->get($mfn); + if ($self->{load_row}) { + $rec = $self->{load_row}->({ id => $mfn }); } else { - $rec = $self->{'data'}->{$mfn}; + $rec = $self->{data}->{$mfn}; } $rec ||= 0E0; @@ -287,7 +449,7 @@ sub pos { my $self = shift; - return $self->{'current_mfn'}; + return $self->{pos}; } @@ -301,13 +463,13 @@ foreach my $mfn ( 1 ... $isis->size ) { ... } -because it takes into account C and C. +because it takes into account C and C. =cut sub size { my $self = shift; - return $self->{'size'}; + return $self->{size}; } =head2 seek @@ -322,52 +484,247 @@ sub seek { my $self = shift; - my $pos = shift || return; + my $pos = shift; my $log = $self->_get_logger(); + $log->logconfess("called without pos") unless defined($pos); + if ($pos < 1) { $log->warn("seek before first record"); $pos = 1; - } elsif ($pos > $self->{'max_mfn'}) { + } elsif ($pos > $self->{max_pos}) { $log->warn("seek beyond last record"); - $pos = $self->{'max_mfn'}; + $pos = $self->{max_pos}; } - return $self->{'current_mfn'} = (($pos - 1) || -1); + return $self->{pos} = (($pos - 1) || -1); } +=head2 stats + +Dump statistics about field and subfield usage + + print $input->stats; + +=cut + +sub stats { + my $self = shift; + + my $log = $self->_get_logger(); + + my $s = $self->{_stats}; + if (! $s) { + $log->warn("called stats, but there is no statistics collected"); + return; + } -=head1 MEMORY USAGE + my $max_fld = 0; -C options is double-edged sword. If enabled, WebPAC -will run on memory constraint machines (which doesn't have enough -physical RAM to create memory structure for whole source database). + my $out = join("\n", + map { + my $f = $_; + die "no field in ", dump( $s->{fld} ) unless defined( $f ); + my $v = $s->{fld}->{$f} || die "no s->{fld}->{$f}"; + $max_fld = $v if ($v > $max_fld); + + my $o = sprintf("%4s %d ~", $f, $v); + + if (defined($s->{sf}->{$f})) { + my @subfields = keys %{ $s->{sf}->{$f} }; + map { + $o .= sprintf(" %s:%d%s", $_, + $s->{sf}->{$f}->{$_}->{count}, + $s->{sf}->{$f}->{$_}->{repeatable} ? '*' : '', + ); + } ( + # first indicators and other special subfields + sort( grep { length($_) > 1 } @subfields ), + # then subfileds (single char) + sort( grep { length($_) == 1 } @subfields ), + ); + } + + if (my $v_r = $s->{repeatable}->{$f}) { + $o .= " ($v_r)" if ($v_r != $v); + } + + $o; + } sort { + if ( $a =~ m/^\d+$/ && $b =~ m/^\d+$/ ) { + $a <=> $b + } else { + $a cmp $b + } + } keys %{ $s->{fld} } + ); + + $log->debug( sub { dump($s) } ); -If your machine has 512Mb or more of RAM and database is around 10000 records, -memory shouldn't be an issue. If you don't have enough physical RAM, you -might consider using virtual memory (if your operating system is handling it -well, like on FreeBSD or Linux) instead of dropping to L to handle -parsed structure of ISIS database (this is what C option does). + return $out; +} -Hitting swap at end of reading source database is probably o.k. However, -hitting swap before 90% will dramatically decrease performance and you will -be better off with C and using rest of availble memory for -operating system disk cache (Linux is particuallary good about this). -However, every access to database record will require disk access, so -generation phase will be slower 10-100 times. +=head2 dump_ascii -Parsed structures are essential - you just have option to trade RAM memory -(which is fast) for disk space (which is slow). Be sure to have planty of -disk space if you are using C and thus L. +Display humanly readable dump of record -However, when WebPAC is running on desktop machines (or laptops :-), it's -highly undesireable for system to start swapping. Using C option can -reduce WecPAC memory usage to around 64Mb for same database with lookup -fields and sorted indexes which stay in RAM. Performance will suffer, but -memory usage will really be minimal. It might be also more confortable to -run WebPAC reniced on those machines. +=cut +sub dump_ascii { + my $self = shift; + + return unless $self->{ll_db}; + + if ($self->{ll_db}->can('dump_ascii')) { + return $self->{ll_db}->dump_ascii( $self->{pos} ); + } else { + return dump( $self->{ll_db}->fetch_rec( $self->{pos} ) ); + } +} + +=head2 _get_regex + +Helper function called which create regexps to be execute on code. + + _get_regex( 900, 'regex:[0-9]+' ,'numbers' ); + _get_regex( 900, '^b', ' : ^b' ); + +It supports perl regexps with C prefix to from value and has +additional logic to skip empty subfields. + +=cut + +sub _get_regex { + my ($sf,$from,$to) = @_; + + # protect / + $from =~ s!/!\\/!gs; + $to =~ s!/!\\/!gs; + + if ($from =~ m/^regex:(.+)$/) { + $from = $1; + } else { + $from = '\Q' . $from . '\E'; + } + if ($sf =~ /^\^/) { + my $need_subfield_data = '*'; # no + # if from is also subfield, require some data in between + # to correctly skip empty subfields + $need_subfield_data = '+' if ($from =~ m/^\\Q\^/); + return + 's/\Q'. $sf .'\E([^\^]' . $need_subfield_data . '?)'. $from .'([^\^]*?)/'. $sf .'$1'. $to .'$2/'; + } else { + return + 's/'. $from .'/'. $to .'/g'; + } +} + + +=head2 modify_record_regexps + +Generate hash with regexpes to be applied using L. + + my $regexpes = $input->modify_record_regexps( + 900 => { '^a' => { ' : ' => '^b' } }, + 901 => { '*' => { '^b' => ' ; ' } }, + ); + +=cut + +sub modify_record_regexps { + my $self = shift; + my $modify_record = {@_}; + + my $regexpes; + + my $log = $self->_get_logger(); + + foreach my $f (keys %$modify_record) { + $log->debug("field: $f"); + + foreach my $sf (keys %{ $modify_record->{$f} }) { + $log->debug("subfield: $sf"); + + foreach my $from (keys %{ $modify_record->{$f}->{$sf} }) { + my $to = $modify_record->{$f}->{$sf}->{$from}; + #die "no field?" unless defined($to); + my $d = "|$from| -> |$to|"; + $log->debug("transform: $d"); + + my $regex = _get_regex($sf,$from,$to); + push @{ $regexpes->{$f} }, { regex => $regex, debug => $d }; + $log->debug("regex: $regex"); + } + } + } + + return $regexpes; +} + +=head2 modify_file_regexps + +Generate hash with regexpes to be applied using L from +pseudo hash/yaml format for regex mappings. + +It should be obvious: + + 200 + '^a' + ' : ' => '^e' + ' = ' => '^d' + +In field I<200> find C<'^a'> and then C<' : '>, and replace it with C<'^e'>. +In field I<200> find C<'^a'> and then C<' = '>, and replace it with C<'^d'>. + + my $regexpes = $input->modify_file_regexps( 'conf/modify/common.pl' ); + +On undef path it will just return. + +=cut + +sub modify_file_regexps { + my $self = shift; + + my $modify_path = shift || return; + + my $log = $self->_get_logger(); + + my $regexpes; + + CORE::open(my $fh, $modify_path) || $log->logdie("can't open modify file $modify_path: $!"); + + my ($f,$sf); + + while(<$fh>) { + chomp; + next if (/^#/ || /^\s*$/); + + if (/^\s*(\d+)\s*$/) { + $f = $1; + $log->debug("field: $f"); + next; + } elsif (/^\s*'([^']*)'\s*$/) { + $sf = $1; + $log->die("can't define subfiled before field in: $_") unless ($f); + $log->debug("subfield: $sf"); + } elsif (/^\s*'([^']*)'\s*=>\s*'([^']*)'\s*$/) { + my ($from,$to) = ($1, $2); + + $log->debug("transform: |$from| -> |$to|"); + + my $regex = _get_regex($sf,$from,$to); + push @{ $regexpes->{$f} }, { + regex => $regex, + file => $modify_path, + line => $., + }; + $log->debug("regex: $regex"); + } + } + + return $regexpes; +} =head1 AUTHOR @@ -375,7 +732,7 @@ =head1 COPYRIGHT & LICENSE -Copyright 2005 Dobrica Pavlinusic, All Rights Reserved. +Copyright 2005-2006 Dobrica Pavlinusic, All Rights Reserved. This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.