1 |
williamc |
1.1 |
#
|
2 |
williamc |
1.6 |
# ActiveDoc.pm
|
3 |
|
|
#
|
4 |
|
|
# Originally Written by Christopher Williams
|
5 |
|
|
#
|
6 |
|
|
# Description
|
7 |
williamc |
1.1 |
#
|
8 |
|
|
# Interface
|
9 |
|
|
# ---------
|
10 |
williamc |
1.25.2.1 |
# new() : A new ActiveDoc object
|
11 |
|
|
# config([ActiveConfig]) : Set up/return Configuration for the document
|
12 |
williamc |
1.6 |
# url() : Return/set the docs url - essential
|
13 |
|
|
# file() : Return the local filename of document
|
14 |
williamc |
1.21 |
# ProcessFile() : Return the filename of PreProcessed document
|
15 |
williamc |
1.6 |
#
|
16 |
williamc |
1.21 |
# parent() : return the object ref of the calling parent
|
17 |
williamc |
1.6 |
# getfile(url) : get a processedfile object given a url
|
18 |
williamc |
1.9 |
# activatedoc(url) : Return the object ref for a doc described by the given url
|
19 |
williamc |
1.14 |
# -- any parse called "init" will also be run
|
20 |
williamc |
1.10 |
# userinterface() : Return the defaullt userinterface
|
21 |
williamc |
1.20 |
# option(var) : return the value of the option var ( or undef )
|
22 |
williamc |
1.6 |
#
|
23 |
williamc |
1.25.2.1 |
# addurltags(parse) : add the <base> tags to manage urls to parse
|
24 |
williamc |
1.6 |
# -- error methods --
|
25 |
|
|
# error(string) : Report an general error to the user
|
26 |
|
|
# parseerror(string) : Report an error during parsing a file
|
27 |
|
|
# line() : Return the current line number of the document
|
28 |
|
|
# and the ProcessedFileObj it is in
|
29 |
williamc |
1.23 |
#
|
30 |
williamc |
1.2 |
|
31 |
|
|
package ActiveDoc::ActiveDoc;
|
32 |
williamc |
1.6 |
require 5.004;
|
33 |
williamc |
1.25.2.1 |
use ActiveDoc::SimpleDoc;
|
34 |
williamc |
1.6 |
use URL::URLhandler;
|
35 |
|
|
|
36 |
williamc |
1.25.2.1 |
@ISA = qw(ActiveDoc::SimpleDoc);
|
37 |
williamc |
1.2 |
|
38 |
williamc |
1.6 |
sub addurltags {
|
39 |
|
|
my $self=shift;
|
40 |
|
|
my $parselabel=shift;
|
41 |
|
|
|
42 |
|
|
$self->{parsers}{$parselabel}->
|
43 |
|
|
addtag("Base", \&Base_start, $self, "", $self,
|
44 |
|
|
\&Base_end, $self);
|
45 |
williamc |
1.1 |
}
|
46 |
|
|
|
47 |
williamc |
1.6 |
sub url {
|
48 |
williamc |
1.2 |
my $self=shift;
|
49 |
williamc |
1.8 |
# get file & preprocess
|
50 |
williamc |
1.21 |
if ( @_ ) {
|
51 |
|
|
$self->{File}=$self->getfile(shift);
|
52 |
|
|
$self->verbose("url downloaded to $self->{File}");
|
53 |
|
|
}
|
54 |
williamc |
1.25 |
if ( defined $self->{File} ) {
|
55 |
|
|
return $self->{File}->url();
|
56 |
|
|
}
|
57 |
|
|
else { return "undefined"; }
|
58 |
williamc |
1.2 |
}
|
59 |
|
|
|
60 |
williamc |
1.6 |
sub config {
|
61 |
williamc |
1.1 |
my $self=shift;
|
62 |
williamc |
1.6 |
@_?$self->{ActiveConfig}=shift
|
63 |
|
|
: $self->{ActiveConfig};
|
64 |
|
|
}
|
65 |
williamc |
1.1 |
|
66 |
williamc |
1.22 |
sub getfile {
|
67 |
williamc |
1.3 |
my $self=shift;
|
68 |
williamc |
1.6 |
my $origurl=shift;
|
69 |
|
|
|
70 |
|
|
my $fileref;
|
71 |
williamc |
1.20 |
my ($url, $file);
|
72 |
williamc |
1.21 |
if ( (defined ($it=$self->option('url_update'))) &&
|
73 |
|
|
( $it eq "1" || $origurl=~/^$it/ )) {
|
74 |
|
|
$self->verbose("Forced download of $origurl");
|
75 |
|
|
($url, $file)=$self->{urlhandler}->download($origurl);
|
76 |
williamc |
1.20 |
}
|
77 |
|
|
else {
|
78 |
williamc |
1.21 |
$self->verbose("Attempting to get $origurl");
|
79 |
williamc |
1.20 |
($url, $file)=$self->{urlhandler}->get($origurl);
|
80 |
|
|
}
|
81 |
williamc |
1.6 |
# do we already have an appropriate object?
|
82 |
williamc |
1.7 |
($fileref)=$self->config()->find($url);
|
83 |
|
|
#undef $fileref;
|
84 |
williamc |
1.6 |
if ( defined $fileref ) {
|
85 |
williamc |
1.20 |
$self->verbose("Found $url in database");
|
86 |
williamc |
1.6 |
$fileref->update();
|
87 |
|
|
}
|
88 |
|
|
else {
|
89 |
|
|
if ( $file eq "" ) {
|
90 |
|
|
$self->parseerror("Unable to get $origurl");
|
91 |
|
|
}
|
92 |
|
|
#-- set up a new preprocess file
|
93 |
williamc |
1.21 |
$self->verbose("Making a new preprocessed file $url");
|
94 |
williamc |
1.6 |
$fileref=ActiveDoc::PreProcessedFile->new($self->config());
|
95 |
|
|
$fileref->url($url);
|
96 |
|
|
$fileref->update();
|
97 |
|
|
}
|
98 |
|
|
return $fileref;
|
99 |
williamc |
1.3 |
}
|
100 |
|
|
|
101 |
williamc |
1.9 |
sub activatedoc {
|
102 |
|
|
my $self=shift;
|
103 |
|
|
my $url=shift;
|
104 |
|
|
|
105 |
|
|
# first get a preprocessed copy of the file
|
106 |
williamc |
1.14 |
# my $fileob=$self->getfile($url);
|
107 |
williamc |
1.9 |
|
108 |
|
|
# now parse it for the <DocType> tag
|
109 |
williamc |
1.14 |
my $tempdoc=ActiveDoc::ActiveDoc->new($self->config());
|
110 |
williamc |
1.16 |
$tempdoc->{urlhandler}=$self->{urlhandler};
|
111 |
williamc |
1.17 |
my $fullurl=$tempdoc->url($url);
|
112 |
|
|
$url=$fullurl;
|
113 |
williamc |
1.14 |
$tempdoc->{doctypefound}=0;
|
114 |
|
|
$tempdoc->newparse("doctype");
|
115 |
|
|
$tempdoc->addtag("doctype","Doc", \&Doc_Start, $tempdoc,
|
116 |
|
|
"", $tempdoc, "", $tempdoc);
|
117 |
|
|
$tempdoc->parse("doctype");
|
118 |
williamc |
1.9 |
|
119 |
williamc |
1.14 |
if ( ! defined $tempdoc->{docobject} ) {
|
120 |
|
|
print "No <Doc type=> Specified in ".$url."\n";
|
121 |
williamc |
1.9 |
exit 1;
|
122 |
|
|
}
|
123 |
|
|
# Set up a new object of the specified type
|
124 |
williamc |
1.14 |
eval "require $tempdoc->{docobject}";
|
125 |
|
|
die $@ if $@;
|
126 |
|
|
my $newobj=$tempdoc->{docobject}->new($self->config());
|
127 |
|
|
undef $tempdoc;
|
128 |
williamc |
1.10 |
$newobj->url($url);
|
129 |
williamc |
1.21 |
$newobj->parent($self);
|
130 |
williamc |
1.9 |
return $newobj;
|
131 |
|
|
}
|
132 |
|
|
|
133 |
williamc |
1.21 |
sub parent {
|
134 |
|
|
my $self=shift;
|
135 |
|
|
|
136 |
|
|
@_?$self->{parent}=shift
|
137 |
|
|
:$self->{parent};
|
138 |
|
|
}
|
139 |
|
|
|
140 |
williamc |
1.6 |
# -------- Error Handling and Error services --------------
|
141 |
|
|
|
142 |
williamc |
1.3 |
sub error {
|
143 |
williamc |
1.6 |
my $self=shift;
|
144 |
|
|
my $string=shift;
|
145 |
|
|
|
146 |
|
|
die $string."\n";
|
147 |
|
|
}
|
148 |
|
|
|
149 |
|
|
sub parseerror {
|
150 |
|
|
my $self=shift;
|
151 |
|
|
my $string=shift;
|
152 |
|
|
|
153 |
williamc |
1.20 |
if ( $self->currentparsename() eq "" ) {
|
154 |
williamc |
1.18 |
$self->error($string);
|
155 |
|
|
}
|
156 |
|
|
else {
|
157 |
|
|
($line, $file)=$self->line();
|
158 |
|
|
print "Parse Error in ".$file->url().", line ".
|
159 |
williamc |
1.6 |
$line."\n";
|
160 |
williamc |
1.18 |
print $string."\n";
|
161 |
williamc |
1.20 |
exit;
|
162 |
williamc |
1.18 |
}
|
163 |
williamc |
1.6 |
}
|
164 |
|
|
|
165 |
|
|
sub line {
|
166 |
williamc |
1.7 |
my $self=shift;
|
167 |
williamc |
1.9 |
|
168 |
williamc |
1.6 |
my ($line, $fileobj)=
|
169 |
williamc |
1.9 |
$self->{File}->realline($self->{currentparser}->line());
|
170 |
williamc |
1.6 |
return ($line, $fileobj);
|
171 |
williamc |
1.7 |
}
|
172 |
|
|
|
173 |
|
|
sub tagstartline {
|
174 |
|
|
my $self=shift;
|
175 |
williamc |
1.8 |
my ($line, $fileobj)=$self->{File}->line(
|
176 |
williamc |
1.7 |
$self->{currentparser}->tagstartline());
|
177 |
|
|
return ($line, $fileobj);
|
178 |
williamc |
1.3 |
}
|
179 |
williamc |
1.6 |
|
180 |
|
|
sub file {
|
181 |
williamc |
1.2 |
my $self=shift;
|
182 |
|
|
|
183 |
williamc |
1.8 |
$self->{File}->file();
|
184 |
williamc |
1.21 |
}
|
185 |
|
|
|
186 |
|
|
sub ProcessFile {
|
187 |
|
|
my $self=shift;
|
188 |
|
|
|
189 |
|
|
return $self->{File}->ProcessedFile();
|
190 |
williamc |
1.2 |
}
|
191 |
|
|
|
192 |
williamc |
1.6 |
# --------------- Initialisation Methods ---------------------------
|
193 |
williamc |
1.2 |
|
194 |
williamc |
1.6 |
sub init {
|
195 |
|
|
# Dummy Routine - override for derived classes
|
196 |
williamc |
1.1 |
}
|
197 |
|
|
|
198 |
williamc |
1.6 |
# ------------------- Tag Routines -----------------------------------
|
199 |
williamc |
1.1 |
#
|
200 |
williamc |
1.6 |
# Base - for setting url bases
|
201 |
williamc |
1.1 |
#
|
202 |
williamc |
1.6 |
sub Base_start {
|
203 |
|
|
my $self=shift;
|
204 |
|
|
my $name=shift;
|
205 |
|
|
my $hashref=shift;
|
206 |
williamc |
1.2 |
|
207 |
williamc |
1.6 |
$self->checktag($name, $hashref, 'type' );
|
208 |
|
|
$self->checktag($name, $hashref, 'base' );
|
209 |
|
|
|
210 |
|
|
# Keep track of base tags
|
211 |
|
|
push @{$self->{basestack}}, $$hashref{"type"};
|
212 |
|
|
# Set the base
|
213 |
|
|
$self->{urlhandler}->setbase($$hashref{"type"},$hashref);
|
214 |
williamc |
1.1 |
}
|
215 |
|
|
|
216 |
williamc |
1.6 |
sub Base_end {
|
217 |
|
|
my $self=shift;
|
218 |
williamc |
1.1 |
my $name=shift;
|
219 |
williamc |
1.6 |
my $type;
|
220 |
williamc |
1.1 |
|
221 |
williamc |
1.6 |
if ( $#{$self->{basestack}} == -1 ) {
|
222 |
williamc |
1.19 |
$self->parseerror("Parse Error : unmatched </$name>");
|
223 |
williamc |
1.6 |
}
|
224 |
|
|
else {
|
225 |
|
|
$type = pop @{$self->{basestack}};
|
226 |
|
|
$self->{urlhandler}->unsetbase($type);
|
227 |
|
|
}
|
228 |
williamc |
1.9 |
}
|
229 |
|
|
|
230 |
|
|
sub Doc_Start {
|
231 |
|
|
my $self=shift;
|
232 |
|
|
my $name=shift;
|
233 |
|
|
my $hashref=shift;
|
234 |
|
|
|
235 |
|
|
$self->checktag($name, $hashref, "type");
|
236 |
williamc |
1.10 |
$self->{doctypefound}++;
|
237 |
|
|
if ( $self->{doctypefound} == 1 ) { # only take first doctype
|
238 |
|
|
$self->{docobject}=$$hashref{'type'};
|
239 |
|
|
}
|
240 |
williamc |
1.1 |
}
|