17 |
|
|
18 |
our $VERSION = '0.01'; |
our $VERSION = '0.01'; |
19 |
|
|
|
#my $LOOKUP_REGEX = '\[[^\[\]]+\]'; |
|
|
#my $LOOKUP_REGEX_SAVE = '\[([^\[\]]+)\]'; |
|
|
my $LOOKUP_REGEX = 'lookup{[^\{\}]+}'; |
|
|
my $LOOKUP_REGEX_SAVE = 'lookup{([^\{\}]+)}'; |
|
|
|
|
20 |
=head1 SYNOPSYS |
=head1 SYNOPSYS |
21 |
|
|
22 |
This module defines common functions, and is used as base for other, more |
This module defines common functions, and is used as base for other, more |
23 |
specific modules. |
specific modules. |
24 |
|
|
|
my $webpac = new WebPAC::Common( |
|
|
filter => { |
|
|
'filter_name_1' => sub { |
|
|
# filter code |
|
|
return length($_); |
|
|
}, ... |
|
|
}, |
|
|
} |
|
|
|
|
25 |
=head1 FUNCTIONS |
=head1 FUNCTIONS |
26 |
|
|
27 |
=head2 fill_in |
=head2 progress_bar |
|
|
|
|
Workhourse of all: takes record from in-memory structure of database and |
|
|
strings with placeholders and returns string or array of with substituted |
|
|
values from record. |
|
|
|
|
|
my $text = $webpac->fill_in($rec,'v250^a'); |
|
|
|
|
|
Optional argument is ordinal number for repeatable fields. By default, |
|
|
it's assume to be first repeatable field (fields are perl array, so first |
|
|
element is 0). |
|
|
Following example will read second value from repeatable field. |
|
|
|
|
|
my $text = $webpac->fill_in($rec,'Title: v250^a',1); |
|
28 |
|
|
29 |
This function B<does not> perform parsing of format to inteligenty skip |
Draw progress bar on STDERR. |
|
delimiters before fields which aren't used. |
|
30 |
|
|
31 |
This method will automatically decode UTF-8 string to local code page |
$webpac->progress_bar($current, $max); |
|
if needed. |
|
32 |
|
|
33 |
=cut |
=cut |
34 |
|
|
35 |
sub fill_in { |
sub progress_bar { |
36 |
my $self = shift; |
my $self = shift; |
37 |
|
|
38 |
|
my ($curr,$max) = @_; |
39 |
|
|
40 |
my $log = $self->_get_logger(); |
my $log = $self->_get_logger(); |
41 |
|
|
42 |
my $rec = shift || $log->logconfess("need data record"); |
$log->logconfess("no current value!") if (! $curr); |
43 |
my $format = shift || $log->logconfess("need format to parse"); |
$log->logconfess("no maximum value!") if (! $max); |
44 |
# iteration (for repeatable fields) |
|
45 |
my $i = shift || 0; |
if ($curr > $max) { |
46 |
|
$max = $curr; |
47 |
|
$log->debug("overflow to $curr"); |
48 |
|
} |
49 |
|
|
50 |
$log->logdie("infitite loop in format $format") if ($i > ($self->{'max_mfn'} || 9999)); |
$self->{'last_pcnt'} ||= 1; |
51 |
|
$self->{'start_t'} ||= time(); |
52 |
|
|
53 |
# FIXME remove for speedup? |
my $p = int($curr * 100 / $max) || 1; |
|
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
|
54 |
|
|
55 |
if (utf8::is_utf8($format)) { |
# reset on re-run |
56 |
$format = $self->_x($format); |
if ($p < $self->{'last_pcnt'}) { |
57 |
|
$self->{'last_pcnt'} = $p; |
58 |
|
$self->{'start_t'} = time(); |
59 |
} |
} |
60 |
|
|
61 |
my $found = 0; |
if ($p != $self->{'last_pcnt'}) { |
62 |
|
|
63 |
my $eval_code; |
my $t = time(); |
64 |
# remove eval{...} from beginning |
my $rate = ($curr / ($t - $self->{'start_t'} || 1)); |
65 |
$eval_code = $1 if ($format =~ s/^eval{([^}]+)}//s); |
my $eta = ($max-$curr) / ($rate || 1); |
66 |
|
printf STDERR ("%5d [%-38s] %-5d %0.1f/s %s\r",$curr,"=" x ($p/3)."$p%>", $max, $rate, $self->fmt_time($eta)); |
67 |
my $filter_name; |
$self->{'last_pcnt'} = $p; |
68 |
# remove filter{...} from beginning |
$self->{'last_curr'} = $curr; |
|
$filter_name = $1 if ($format =~ s/^filter{([^}]+)}//s); |
|
|
|
|
|
# do actual replacement of placeholders |
|
|
# repeatable fields |
|
|
$format =~ s/v(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,$i,\$found)/ges; |
|
|
# non-repeatable fields |
|
|
$format =~ s/s(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,0,\$found)/ges; |
|
|
|
|
|
if ($found) { |
|
|
$log->debug("format: $format"); |
|
|
if ($eval_code) { |
|
|
my $eval = $self->fill_in($rec,$eval_code,$i); |
|
|
return if (! $self->_eval($eval)); |
|
|
} |
|
|
if ($filter_name && $self->{'filter'}->{$filter_name}) { |
|
|
$log->debug("filter '$filter_name' for $format"); |
|
|
$format = $self->{'filter'}->{$filter_name}->($format); |
|
|
return unless(defined($format)); |
|
|
$log->debug("filter result: $format"); |
|
|
} |
|
|
# do we have lookups? |
|
|
if ($format =~ /$LOOKUP_REGEX/o) { |
|
|
$log->debug("format '$format' has lookup"); |
|
|
return $self->lookup($format); |
|
|
} else { |
|
|
return $format; |
|
|
} |
|
|
} else { |
|
|
return; |
|
69 |
} |
} |
70 |
|
print STDERR "\n" if ($p == 100); |
71 |
} |
} |
72 |
|
|
73 |
|
=head2 fmt_time |
74 |
|
|
75 |
=head2 get_data |
Format time (in seconds) for display. |
|
|
|
|
Returns value from record. |
|
76 |
|
|
77 |
my $text = $self->get_data(\$rec,$f,$sf,$i,\$found); |
print $webpac->fmt_time(time()); |
78 |
|
|
79 |
Arguments are: |
This method is called by L<progress_bar> to display remaining time. |
|
record reference C<$rec>, |
|
|
field C<$f>, |
|
|
optional subfiled C<$sf>, |
|
|
index for repeatable values C<$i>. |
|
|
|
|
|
Optinal variable C<$found> will be incremeted if there |
|
|
is field. |
|
|
|
|
|
Returns value or empty string. |
|
80 |
|
|
81 |
=cut |
=cut |
82 |
|
|
83 |
sub get_data { |
sub fmt_time { |
84 |
my $self = shift; |
my $self = shift; |
85 |
|
|
86 |
my ($rec,$f,$sf,$i,$found) = @_; |
my $t = shift || 0; |
87 |
|
my $out = ""; |
88 |
|
|
89 |
if ($$rec->{$f}) { |
my ($ss,$mm,$hh) = gmtime($t); |
90 |
return '' if (! $$rec->{$f}->[$i]); |
$out .= "${hh}h" if ($hh); |
91 |
no strict 'refs'; |
$out .= sprintf("%02d:%02d", $mm,$ss); |
92 |
if ($sf && $$rec->{$f}->[$i]->{$sf}) { |
$out .= " " if ($hh == 0); |
93 |
$$found++ if (defined($$found)); |
return $out; |
|
return $$rec->{$f}->[$i]->{$sf}; |
|
|
} elsif ($$rec->{$f}->[$i]) { |
|
|
$$found++ if (defined($$found)); |
|
|
# it still might have subfield, just |
|
|
# not specified, so we'll dump all |
|
|
if ($$rec->{$f}->[$i] =~ /HASH/o) { |
|
|
my $out; |
|
|
foreach my $k (keys %{$$rec->{$f}->[$i]}) { |
|
|
$out .= $$rec->{$f}->[$i]->{$k}." "; |
|
|
} |
|
|
return $out; |
|
|
} else { |
|
|
return $$rec->{$f}->[$i]; |
|
|
} |
|
|
} |
|
|
} else { |
|
|
return ''; |
|
|
} |
|
94 |
} |
} |
95 |
|
|
96 |
|
# |
97 |
|
# |
98 |
|
# |
99 |
|
|
100 |
=head1 INTERNAL METHODS |
=head1 INTERNAL METHODS |
101 |
|
|
128 |
return $ret || undef; |
return $ret || undef; |
129 |
} |
} |
130 |
|
|
|
=head2 _sort_by_order |
|
|
|
|
|
Sort xml tags data structure accoding to C<order=""> attribute. |
|
|
|
|
|
=cut |
|
|
|
|
|
sub _sort_by_order { |
|
|
my $self = shift; |
|
|
|
|
|
my $va = $self->{'import_xml'}->{'indexer'}->{$a}->{'order'} || |
|
|
$self->{'import_xml'}->{'indexer'}->{$a}; |
|
|
my $vb = $self->{'import_xml'}->{'indexer'}->{$b}->{'order'} || |
|
|
$self->{'import_xml'}->{'indexer'}->{$b}; |
|
|
|
|
|
return $va <=> $vb; |
|
|
} |
|
|
|
|
|
=head2 _x |
|
|
|
|
|
Convert string from UTF-8 to code page defined in C<import_xml>. |
|
|
|
|
|
my $text = $webpac->_x('utf8 text'); |
|
|
|
|
|
Default application code page is C<ISO-8859-2>. You will probably want to |
|
|
change that when creating new instance of object based on this one. |
|
|
|
|
|
=cut |
|
|
|
|
|
sub _x { |
|
|
my $self = shift; |
|
|
my $utf8 = shift || return; |
|
|
|
|
|
# create UTF-8 convertor for import_xml files |
|
|
$self->{'utf2cp'} ||= Text::Iconv->new('UTF-8' ,$self->{'code_page'} || 'ISO-8859-2'); |
|
|
|
|
|
return $self->{'utf2cp'}->convert($utf8) || |
|
|
$self->_get_logger()->logwarn("can't convert '$utf8'"); |
|
|
} |
|
|
|
|
131 |
=head2 _init_logger |
=head2 _init_logger |
132 |
|
|
133 |
This function will init C<Log::Log4perl> using provided configuration file. |
This function will init C<Log::Log4perl> using provided configuration file. |
135 |
$webpac->_init_logger('/path/to/log.conf'); |
$webpac->_init_logger('/path/to/log.conf'); |
136 |
|
|
137 |
If no path to configuration file is given, dummy empty configuration |
If no path to configuration file is given, dummy empty configuration |
138 |
will be create. |
will be created. If any mode which inherits from this one is called |
139 |
|
with C<debug> flag, it will turn logging to debug level. |
140 |
|
|
141 |
=cut |
=cut |
142 |
|
|
147 |
Log::Log4perl->init($file); |
Log::Log4perl->init($file); |
148 |
} else { |
} else { |
149 |
my $conf = q( ); |
my $conf = q( ); |
150 |
|
if ($self->{'debug'}) { |
151 |
|
$conf = << '_log4perl_'; |
152 |
|
|
153 |
|
log4perl.rootLogger=INFO, SCREEN |
154 |
|
|
155 |
|
log4perl.logger.WebPAC.=DEBUG |
156 |
|
|
157 |
|
log4perl.appender.SCREEN=Log::Log4perl::Appender::Screen |
158 |
|
log4perl.appender.SCREEN.layout=PatternLayout |
159 |
|
log4perl.appender.SCREEN.layout.ConversionPattern=%d %p> %F{1}:%L %M - %m%n |
160 |
|
|
161 |
|
_log4perl_ |
162 |
|
} |
163 |
Log::Log4perl->init( \$conf ); |
Log::Log4perl->init( \$conf ); |
164 |
} |
} |
165 |
} |
} |