| 1 |
#
|
| 2 |
# ActiveDoc.pm
|
| 3 |
#
|
| 4 |
# Originally Written by Christopher Williams
|
| 5 |
#
|
| 6 |
# Description
|
| 7 |
#
|
| 8 |
# Interface
|
| 9 |
# ---------
|
| 10 |
# new() : A new ActiveDoc object
|
| 11 |
# url() : Return/set the docs url - essential
|
| 12 |
# file() : Return the local filename of document
|
| 13 |
#
|
| 14 |
# parse(parselabel): Parse the document file for the given parse level
|
| 15 |
# newparse(parselabel) : Create a new parse type
|
| 16 |
# addtag(parselabel,tagname,start,obj,text,obj,end,obj)
|
| 17 |
# : Add tags to the parse given by label
|
| 18 |
# newdoc(file) : Return an new object of the appropriate type
|
| 19 |
# getfile(url) : get a processedfile object given a url
|
| 20 |
# config([ActiveConfig]) : Set up/return Configuration for the document
|
| 21 |
# basequery([ActiveConfig]) : Set up/return UserQuery for the doc
|
| 22 |
# copydocconfig(ActiveDoc) : Copy the basic configuration from the ActiveDoc
|
| 23 |
# copydocquery(ActiveDoc) : Copy the basicquery from the ActiveDoc
|
| 24 |
#
|
| 25 |
# -- error methods --
|
| 26 |
# error(string) : Report an general error to the user
|
| 27 |
# parseerror(string) : Report an error during parsing a file
|
| 28 |
# line() : Return the current line number of the document
|
| 29 |
# and the ProcessedFileObj it is in
|
| 30 |
|
| 31 |
package ActiveDoc::ActiveDoc;
|
| 32 |
require 5.004;
|
| 33 |
use ActiveDoc::Parse;
|
| 34 |
use ActiveDoc::ActiveConfig;
|
| 35 |
use ActiveDoc::PreProcessedFile;
|
| 36 |
use ObjectUtilities::ObjectBase;
|
| 37 |
use URL::URLhandler;
|
| 38 |
|
| 39 |
@ISA = qw(ObjectUtilities::ObjectBase);
|
| 40 |
|
| 41 |
sub new {
|
| 42 |
my $class=shift;
|
| 43 |
$self={};
|
| 44 |
bless $self, $class;
|
| 45 |
$self->config(shift);
|
| 46 |
|
| 47 |
# A URL handler per document
|
| 48 |
$self->{urlhandler}=URL::URLhandler->new($self->config()->cache());
|
| 49 |
|
| 50 |
$self->init(@_);
|
| 51 |
return $self;
|
| 52 |
}
|
| 53 |
|
| 54 |
# ----- parse related routines --------------
|
| 55 |
sub parse {
|
| 56 |
my $self=shift;
|
| 57 |
$parselabel=shift;
|
| 58 |
|
| 59 |
my $file=$self->file();
|
| 60 |
print "Parse called on file $file\n";
|
| 61 |
if ( $file ) {
|
| 62 |
$self->{parsers}{$parselabel}->parse($file,@_);
|
| 63 |
}
|
| 64 |
else {
|
| 65 |
print "Cannot parse - file not known\n";
|
| 66 |
}
|
| 67 |
}
|
| 68 |
|
| 69 |
sub newparse {
|
| 70 |
my $self=shift;
|
| 71 |
my $parselabel=shift;
|
| 72 |
|
| 73 |
$self->{parsers}{$parselabel}=ActiveDoc::Parse->new();
|
| 74 |
$self->{parsers}{$parselabel}->addignoretags();
|
| 75 |
$self->{parsers}{$parselabel}->addgrouptags();
|
| 76 |
}
|
| 77 |
|
| 78 |
sub addtag {
|
| 79 |
my $self=shift;
|
| 80 |
my $parselabel=shift;
|
| 81 |
if ( $#_ != 6 ) {
|
| 82 |
$self->error("Incorrect addtags specification\n".
|
| 83 |
"called with :\n@_ \n");
|
| 84 |
}
|
| 85 |
$self->{parsers}{$parselabel}->addtag(@_);
|
| 86 |
}
|
| 87 |
|
| 88 |
sub addurltags {
|
| 89 |
my $self=shift;
|
| 90 |
my $parselabel=shift;
|
| 91 |
|
| 92 |
$self->{parsers}{$parselabel}->
|
| 93 |
addtag("Base", \&Base_start, $self, "", $self,
|
| 94 |
\&Base_end, $self);
|
| 95 |
}
|
| 96 |
|
| 97 |
sub url {
|
| 98 |
my $self=shift;
|
| 99 |
@_ ?$self->{File}=$self->getfile(shift)
|
| 100 |
: $self->{File};
|
| 101 |
}
|
| 102 |
|
| 103 |
sub copydocconfig {
|
| 104 |
my $self=shift;
|
| 105 |
my $ActiveDoc=shift;
|
| 106 |
|
| 107 |
$self->config($ActiveDoc->config());
|
| 108 |
|
| 109 |
}
|
| 110 |
|
| 111 |
sub copydocquery {
|
| 112 |
my $self=shift;
|
| 113 |
my $ActiveDoc=shift;
|
| 114 |
|
| 115 |
$self->basequery($ActiveDoc->basequery());
|
| 116 |
}
|
| 117 |
|
| 118 |
sub config {
|
| 119 |
my $self=shift;
|
| 120 |
@_?$self->{ActiveConfig}=shift
|
| 121 |
: $self->{ActiveConfig};
|
| 122 |
}
|
| 123 |
|
| 124 |
sub basequery {
|
| 125 |
my $self=shift;
|
| 126 |
@_ ? $self->{UserQuery}=shift
|
| 127 |
: $self->{UserQuery};
|
| 128 |
}
|
| 129 |
|
| 130 |
sub getfile() {
|
| 131 |
my $self=shift;
|
| 132 |
my $origurl=shift;
|
| 133 |
|
| 134 |
my $fileref;
|
| 135 |
print "GETFILE called\n";
|
| 136 |
my ($url, $file)=$self->{urlhandler}->get($origurl);
|
| 137 |
# do we already have an appropriate object?
|
| 138 |
#my ($fileref)=$self->config()->find("__preprocessed",$url);
|
| 139 |
undef $fileref;
|
| 140 |
if ( defined $fileref ) {
|
| 141 |
print "found $url in database ----\n";
|
| 142 |
$fileref->update();
|
| 143 |
}
|
| 144 |
else {
|
| 145 |
if ( $file eq "" ) {
|
| 146 |
$self->parseerror("Unable to get $origurl");
|
| 147 |
}
|
| 148 |
#-- set up a new preprocess file
|
| 149 |
print "Making a new file $url----\n";
|
| 150 |
$fileref=ActiveDoc::PreProcessedFile->new($self->config());
|
| 151 |
$fileref->url($url);
|
| 152 |
$fileref->update();
|
| 153 |
$self->config()->store($fileref,"__preprocessed",$url);
|
| 154 |
}
|
| 155 |
print "---------- returning".$fileref."\n";
|
| 156 |
return $fileref;
|
| 157 |
}
|
| 158 |
|
| 159 |
# -------- Error Handling and Error services --------------
|
| 160 |
|
| 161 |
sub error {
|
| 162 |
my $self=shift;
|
| 163 |
my $string=shift;
|
| 164 |
|
| 165 |
die $string."\n";
|
| 166 |
}
|
| 167 |
|
| 168 |
sub parseerror {
|
| 169 |
my $self=shift;
|
| 170 |
my $string=shift;
|
| 171 |
|
| 172 |
($line, $file)=$self->line();
|
| 173 |
print "Parse Error in ".$file->url().", line ".
|
| 174 |
$line."\n";
|
| 175 |
print $string."\n";
|
| 176 |
die;
|
| 177 |
}
|
| 178 |
|
| 179 |
sub checktag {
|
| 180 |
my $self=shift;
|
| 181 |
my $tagname=shift;
|
| 182 |
my $hashref=shift;
|
| 183 |
my $param=shift;
|
| 184 |
|
| 185 |
if ( ! exists $$hashref{$param} ) {
|
| 186 |
$self->parseerror("Incomplete Tag <$tagname> : $param required");
|
| 187 |
}
|
| 188 |
}
|
| 189 |
|
| 190 |
sub line {
|
| 191 |
$self=shift;
|
| 192 |
my ($line, $fileobj)=
|
| 193 |
$self->{Processedfile}->line($self->{switch}->line());
|
| 194 |
return ($line, $fileobj);
|
| 195 |
}
|
| 196 |
|
| 197 |
sub file {
|
| 198 |
my $self=shift;
|
| 199 |
|
| 200 |
$self->{PPf}->file();
|
| 201 |
}
|
| 202 |
|
| 203 |
# --------------- Initialisation Methods ---------------------------
|
| 204 |
|
| 205 |
sub preprocess_init {
|
| 206 |
my $self=shift;
|
| 207 |
$self->{PPfile}=PreProcessedFile->new($self->config());
|
| 208 |
}
|
| 209 |
|
| 210 |
sub init {
|
| 211 |
# Dummy Routine - override for derived classes
|
| 212 |
}
|
| 213 |
|
| 214 |
# ------------------- Tag Routines -----------------------------------
|
| 215 |
#
|
| 216 |
# Base - for setting url bases
|
| 217 |
#
|
| 218 |
sub Base_start {
|
| 219 |
my $self=shift;
|
| 220 |
my $name=shift;
|
| 221 |
my $hashref=shift;
|
| 222 |
|
| 223 |
$self->checktag($name, $hashref, 'type' );
|
| 224 |
$self->checktag($name, $hashref, 'base' );
|
| 225 |
|
| 226 |
# Keep track of base tags
|
| 227 |
push @{$self->{basestack}}, $$hashref{"type"};
|
| 228 |
# Set the base
|
| 229 |
$self->{urlhandler}->setbase($$hashref{"type"},$hashref);
|
| 230 |
|
| 231 |
}
|
| 232 |
|
| 233 |
sub Base_end {
|
| 234 |
my $self=shift;
|
| 235 |
my $name=shift;
|
| 236 |
my $type;
|
| 237 |
|
| 238 |
if ( $#{$self->{basestack}} == -1 ) {
|
| 239 |
print "Parse Error : unmatched </".$name."> on line ".
|
| 240 |
$self->line()."\n";
|
| 241 |
die;
|
| 242 |
}
|
| 243 |
else {
|
| 244 |
$type = pop @{$self->{basestack}};
|
| 245 |
$self->{urlhandler}->unsetbase($type);
|
| 246 |
}
|
| 247 |
}
|
