1 |
|
# |
2 |
< |
# The base functionality for the ActiveDocument - inherits from Basetags |
2 |
> |
# ActiveDoc.pm |
3 |
> |
# |
4 |
> |
# Originally Written by Christopher Williams |
5 |
> |
# |
6 |
> |
# Description |
7 |
|
# |
4 |
– |
# Inherits from BaseTags |
5 |
– |
# -------- |
8 |
|
# Interface |
9 |
|
# --------- |
10 |
< |
# new(filename, ObjectStoreCont): create a new object based on a file and |
11 |
< |
# associate with the given ObjectStoreCont |
12 |
< |
# parse() : parse the input file |
13 |
< |
# include(url) : Activate include file mechanism, returns the object ref if OK |
14 |
< |
# treenode() : return the associated TreeNode object reference |
15 |
< |
# getincludeObjectStore : Return a pointer to the ObectStore that contains all |
16 |
< |
# included objects |
17 |
< |
# find(string) : find the object reference related to string in the associated |
18 |
< |
# tree. Mechanism for getting object references |
10 |
> |
# new() : A new ActiveDoc object |
11 |
> |
# config([ActiveConfig]) : Set up/return Configuration for the document |
12 |
> |
# url() : Return/set the docs url - essential |
13 |
> |
# file() : Return the local filename of document |
14 |
> |
# ProcessFile() : Return the filename of PreProcessed document |
15 |
> |
# |
16 |
> |
# parent() : return the object ref of the calling parent |
17 |
> |
# getfile(url) : get a processedfile object given a url |
18 |
> |
# activatedoc(url) : Return the object ref for a doc described by the given url |
19 |
> |
# -- any parse called "init" will also be run |
20 |
> |
# userinterface() : Return the defaullt userinterface |
21 |
> |
# option(var) : return the value of the option var ( or undef ) |
22 |
> |
# |
23 |
> |
# addurltags(parse) : add the <base> tags to manage urls to parse |
24 |
> |
# -- 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 |
> |
# |
30 |
> |
|
31 |
> |
package ActiveDoc::ActiveDoc; |
32 |
> |
require 5.004; |
33 |
> |
use ActiveDoc::SimpleDoc; |
34 |
> |
use URL::URLhandler; |
35 |
|
|
36 |
+ |
@ISA = qw(ActiveDoc::SimpleDoc); |
37 |
|
|
38 |
< |
package ActiveDoc; |
39 |
< |
use BaseTags; |
40 |
< |
use DOChandler; |
41 |
< |
use ObjectStoreCont; |
38 |
> |
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 |
> |
} |
46 |
|
|
47 |
< |
@ISA = qw (BaseTags); |
47 |
> |
sub url { |
48 |
> |
my $self=shift; |
49 |
> |
# get file & preprocess |
50 |
> |
if ( @_ ) { |
51 |
> |
$self->{File}=$self->getfile(shift); |
52 |
> |
$self->verbose("url downloaded to $self->{File}"); |
53 |
> |
} |
54 |
> |
if ( defined $self->{File} ) { |
55 |
> |
return $self->{File}->url(); |
56 |
> |
} |
57 |
> |
else { return "undefined"; } |
58 |
> |
} |
59 |
|
|
60 |
< |
# Initialise |
27 |
< |
sub _init { |
60 |
> |
sub config { |
61 |
|
my $self=shift; |
62 |
< |
my $OC=shift; |
62 |
> |
@_?$self->{ActiveConfig}=shift |
63 |
> |
: $self->{ActiveConfig}; |
64 |
> |
} |
65 |
> |
|
66 |
> |
sub getfile { |
67 |
> |
my $self=shift; |
68 |
> |
my $origurl=shift; |
69 |
|
|
70 |
< |
$self->_addurl(); |
71 |
< |
$self->{OC}=$OC; |
72 |
< |
$self->{treenode)=TreeNode->new(); |
73 |
< |
$self->{includeOS}=$self->{OC}->newStore(); |
74 |
< |
$self->{dochandler}=DOChandler->new($self->{includeOS}); |
75 |
< |
# Add the minimal functionality tag - feel free to override |
76 |
< |
$self->{tags}->addtag("Include", \&Include_Start, "", ""); |
70 |
> |
my $fileref; |
71 |
> |
my ($url, $file); |
72 |
> |
if ( (defined ($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 |
> |
} |
77 |
> |
else { |
78 |
> |
$self->verbose("Attempting to get $origurl"); |
79 |
> |
($url, $file)=$self->{urlhandler}->get($origurl); |
80 |
> |
} |
81 |
> |
# do we already have an appropriate object? |
82 |
> |
($fileref)=$self->config()->find($url); |
83 |
> |
#undef $fileref; |
84 |
> |
if ( defined $fileref ) { |
85 |
> |
$self->verbose("Found $url in database"); |
86 |
> |
$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 |
> |
$self->verbose("Making a new preprocessed file $url"); |
94 |
> |
$fileref=ActiveDoc::PreProcessedFile->new($self->config()); |
95 |
> |
$fileref->url($url); |
96 |
> |
$fileref->update(); |
97 |
> |
} |
98 |
> |
return $fileref; |
99 |
|
} |
100 |
|
|
101 |
< |
# |
102 |
< |
# Include mechanism |
103 |
< |
# |
104 |
< |
sub include { |
105 |
< |
my $self=shift; |
106 |
< |
my $url=shift; |
46 |
< |
my $obj; |
101 |
> |
sub activatedoc { |
102 |
> |
my $self=shift; |
103 |
> |
my $url=shift; |
104 |
> |
|
105 |
> |
# first get a preprocessed copy of the file |
106 |
> |
# my $fileob=$self->getfile($url); |
107 |
|
|
108 |
< |
$obj=$self->{dochandler}->newdoc($url); |
109 |
< |
# Now Extend our tree |
110 |
< |
$self->{treenode}->grow($obj->treenode()); |
111 |
< |
return $obj; |
108 |
> |
# now parse it for the <DocType> tag |
109 |
> |
my $tempdoc=ActiveDoc::ActiveDoc->new($self->config()); |
110 |
> |
$tempdoc->{urlhandler}=$self->{urlhandler}; |
111 |
> |
my $fullurl=$tempdoc->url($url); |
112 |
> |
$url=$fullurl; |
113 |
> |
$tempdoc->{doctypefound}=0; |
114 |
> |
$tempdoc->newparse("doctype"); |
115 |
> |
$tempdoc->addtag("doctype","Doc", \&Doc_Start, $tempdoc, |
116 |
> |
"", $tempdoc, "", $tempdoc); |
117 |
> |
$tempdoc->parse("doctype"); |
118 |
> |
|
119 |
> |
if ( ! defined $tempdoc->{docobject} ) { |
120 |
> |
print "No <Doc type=> Specified in ".$url."\n"; |
121 |
> |
exit 1; |
122 |
> |
} |
123 |
> |
# Set up a new object of the specified type |
124 |
> |
eval "require $tempdoc->{docobject}"; |
125 |
> |
die $@ if $@; |
126 |
> |
my $newobj=$tempdoc->{docobject}->new($self->config()); |
127 |
> |
undef $tempdoc; |
128 |
> |
$newobj->url($url); |
129 |
> |
$newobj->parent($self); |
130 |
> |
return $newobj; |
131 |
|
} |
132 |
|
|
133 |
< |
sub treenode { |
133 |
> |
sub parent { |
134 |
|
my $self=shift; |
135 |
< |
return $self->treenode; |
135 |
> |
|
136 |
> |
@_?$self->{parent}=shift |
137 |
> |
:$self->{parent}; |
138 |
> |
} |
139 |
> |
|
140 |
> |
# -------- Error Handling and Error services -------------- |
141 |
> |
|
142 |
> |
sub error { |
143 |
> |
my $self=shift; |
144 |
> |
my $string=shift; |
145 |
> |
|
146 |
> |
die $string."\n"; |
147 |
|
} |
148 |
|
|
149 |
< |
sub getincludeObjectStore { |
149 |
> |
sub parseerror { |
150 |
|
my $self=shift; |
151 |
< |
return $self->{includeOS}; |
151 |
> |
my $string=shift; |
152 |
> |
|
153 |
> |
if ( $self->currentparsename() eq "" ) { |
154 |
> |
$self->error($string); |
155 |
> |
} |
156 |
> |
else { |
157 |
> |
($line, $file)=$self->line(); |
158 |
> |
print "Parse Error in ".$file->url().", line ". |
159 |
> |
$line."\n"; |
160 |
> |
print $string."\n"; |
161 |
> |
exit; |
162 |
> |
} |
163 |
|
} |
164 |
|
|
165 |
< |
sub find($) { |
165 |
> |
sub line { |
166 |
|
my $self=shift; |
66 |
– |
my $string=shift; |
167 |
|
|
168 |
< |
$self->{treenode}->find($string); |
168 |
> |
my ($line, $fileobj)= |
169 |
> |
$self->{File}->realline($self->{currentparser}->line()); |
170 |
> |
return ($line, $fileobj); |
171 |
|
} |
172 |
|
|
173 |
< |
# ------------------------ Tag Routines ------------------------------ |
174 |
< |
# |
175 |
< |
# A default Include tag |
176 |
< |
# |
177 |
< |
sub Include_Start { |
178 |
< |
my $returnval; |
179 |
< |
# Just call the basic - this is only a default wrapper for the |
180 |
< |
# <INCLUDE> tag assuming with no futher processing of the DOCObjref |
181 |
< |
$returnval=_includetag(@_) |
182 |
< |
# dont return anything if its a basic tag |
173 |
> |
sub tagstartline { |
174 |
> |
my $self=shift; |
175 |
> |
my ($line, $fileobj)=$self->{File}->line( |
176 |
> |
$self->{currentparser}->tagstartline()); |
177 |
> |
return ($line, $fileobj); |
178 |
> |
} |
179 |
> |
|
180 |
> |
sub file { |
181 |
> |
my $self=shift; |
182 |
> |
|
183 |
> |
$self->{File}->file(); |
184 |
> |
} |
185 |
> |
|
186 |
> |
sub ProcessFile { |
187 |
> |
my $self=shift; |
188 |
> |
|
189 |
> |
return $self->{File}->ProcessedFile(); |
190 |
|
} |
191 |
|
|
192 |
< |
# ----------------------- Support Routines --------------------------- |
192 |
> |
# --------------- Initialisation Methods --------------------------- |
193 |
> |
|
194 |
> |
sub init { |
195 |
> |
# Dummy Routine - override for derived classes |
196 |
> |
} |
197 |
> |
|
198 |
> |
# ------------------- Tag Routines ----------------------------------- |
199 |
|
# |
200 |
< |
# the real workings of the include tag returns the ref |
200 |
> |
# Base - for setting url bases |
201 |
|
# |
202 |
< |
sub _includetag { |
202 |
> |
sub Base_start { |
203 |
|
my $self=shift; |
204 |
|
my $name=shift; |
205 |
|
my $hashref=shift; |
206 |
|
|
207 |
< |
$self->{switch}->checkparam( $name, "ref"); |
208 |
< |
$url=$$hashref{'ref'}; |
209 |
< |
return $self->include($url); |
207 |
> |
$self->checktag($name, $hashref, 'type' ); |
208 |
> |
$self->checktag($name, $hashref, 'base' ); |
209 |
> |
|
210 |
> |
# Keep track of base tags |
211 |
> |
push @{$self->{basestack}}, $$hashref{"type"}; |
212 |
> |
# Set the base |
213 |
> |
$self->{urlhandler}->setbase($$hashref{"type"},$hashref); |
214 |
> |
} |
215 |
> |
|
216 |
> |
sub Base_end { |
217 |
> |
my $self=shift; |
218 |
> |
my $name=shift; |
219 |
> |
my $type; |
220 |
> |
|
221 |
> |
if ( $#{$self->{basestack}} == -1 ) { |
222 |
> |
$self->parseerror("Parse Error : unmatched </$name>"); |
223 |
> |
} |
224 |
> |
else { |
225 |
> |
$type = pop @{$self->{basestack}}; |
226 |
> |
$self->{urlhandler}->unsetbase($type); |
227 |
> |
} |
228 |
|
} |
229 |
|
|
230 |
+ |
sub Doc_Start { |
231 |
+ |
my $self=shift; |
232 |
+ |
my $name=shift; |
233 |
+ |
my $hashref=shift; |
234 |
+ |
|
235 |
+ |
$self->checktag($name, $hashref, "type"); |
236 |
+ |
$self->{doctypefound}++; |
237 |
+ |
if ( $self->{doctypefound} == 1 ) { # only take first doctype |
238 |
+ |
$self->{docobject}=$$hashref{'type'}; |
239 |
+ |
} |
240 |
+ |
} |