src/ActiveDoc/ActiveDoc.pm

#
# ActiveDoc.pm
#
# Originally Written by Christopher Williams
#
# Description
#
# Interface
# ---------
# new()         : A new ActiveDoc object
# url()         : Return/set the docs url - essential
# file()        : Return the local filename of document
#
# parse(parselabel): Parse the document file for the given parse level
# newparse(parselabel) : Create a new parse type
# addtag(parselabel,tagname,start,obj,text,obj,end,obj)
#                               : Add tags to the parse given by label
# checktag(tagname, hashref, param) : check for existence of param in
#                                       hashref from a tag call
# includeparse(local_parsename, objparsename, activedoc) : copy the parse from 
#                                                       one object to another
# currentparsename([name]) : get/set current parse name
# newdoc(file)  : Return an new object of the appropriate type
# getfile(url)  : get a processedfile object given a url
# activatedoc(url) : Return the object ref for a doc described by the given url
# config([ActiveConfig]) : Set up/return Configuration for the document
# basequery([ActiveConfig]) : Set up/return UserQuery for the doc
# copydocconfig(ActiveDoc) : Copy the basic configuration from the ActiveDoc
# copydocquery(ActiveDoc) : Copy the basicquery from the ActiveDoc
# userinterface()       : Return the defaullt userinterface
# options(var)          : return the value of the option var
#
# -- error methods --
# error(string)       : Report an general error to the user
# parseerror(string)  : Report an error during parsing a file
# line()              : Return the current line number of the document
#                       and the ProcessedFileObj it is in

package ActiveDoc::ActiveDoc;
require 5.004;
use ActiveDoc::Parse;
use ActiveDoc::ActiveConfig;
use ActiveDoc::PreProcessedFile;
use ObjectUtilities::StorableObject;
use URL::URLhandler;

@ISA = qw(ObjectUtilities::StorableObject);

sub new {
        my $class=shift;
        $self={};
        bless $self, $class;
        $self->config(shift);
        
        # A URL handler per document
        $self->{urlhandler}=URL::URLhandler->new($self->config()->cache());

        # A default UserInterface
        $self->{userinterface}=ActiveDoc::SimpleUserInterface->new();
        $self->init(@_);
        return $self;
}

# ----- parse related routines --------------
sub parse {
        my $self=shift;
        $parselabel=shift;

        my $file=$self->file();
        if ( $file ) {
          $self->{currentparsename}=$parselabel;
          $self->{currentparser}=$self->{parsers}{$parselabel};
          $self->{parsers}{$parselabel}->parse($file,@_);
          delete $self->{currentparser};
          $self->{currentparsename}="";
        }
        else {
          print "Cannot parse - file not known\n";
        }
}

sub currentparsename {
        my $self=shift;
        @_?$self->{currentparsename}=shift
          :$self->{currentparsename};
}

sub newparse {
        my $self=shift;
        my $parselabel=shift;

        $self->{parsers}{$parselabel}=ActiveDoc::Parse->new();
        $self->{parsers}{$parselabel}->addignoretags();
        $self->{parsers}{$parselabel}->addgrouptags();
}

sub cleartags {
        my $self=shift;
        my $parselabel=shift;

        $self->{parsers}{$parselabel}->cleartags();
}


sub includeparse {
        my $self=shift;
        my $parselabel=shift;
        my $remoteparselabel=shift;
        my $activedoc=shift;

        # Some error trapping
        if ( ! exists $self->{parsers}{$parselabel} ) {
          $self->error("Unknown local parse name specified");
        }
        if ( ! exists $activedoc->{parsers}{$remoteparselabel} ) {
          $self->error("Unknown parse name specified in remote obj $activedoc");
        }

        #
        my $rp=$activedoc->{parsers}{$remoteparselabel};
        $self->{parsers}{$parselabel}->includeparse($rp);
}

sub addtag {
        my $self=shift;
        my $parselabel=shift;
        if ( $#_ != 6 ) {
                $self->error("Incorrect addtags specification\n".
                                "called with :\n@_ \n");
        }
        $self->{parsers}{$parselabel}->addtag(@_);
}

sub addurltags {
        my $self=shift;
        my $parselabel=shift;
        
        $self->{parsers}{$parselabel}->
                addtag("Base", \&Base_start, $self, "", $self,
                        \&Base_end, $self);
}

sub url {
        my $self=shift;
        # get file & preprocess
        if ( @_  ) {$self->{File}=$self->getfile(shift)} 
        $self->{File}->url();
}

sub copydocconfig {
        my $self=shift;
        my $ActiveDoc=shift;
        
        $self->config($ActiveDoc->config());

}

sub copydocquery {
        my $self=shift;
        my $ActiveDoc=shift;

         $self->basequery($ActiveDoc->basequery());
}

sub config {
        my $self=shift;
        @_?$self->{ActiveConfig}=shift
           : $self->{ActiveConfig};
}

sub basequery {
        my $self=shift;
        @_ ? $self->{Query}=shift
           : $self->{Query};
}

sub options {
        my $self=shift;
        my $param=shift;
        $self->basequery()->getparam('option_'.$param);
}

sub getfile() {
        my $self=shift;
        my $origurl=shift; 

        my $fileref;
        my ($url, $file)=$self->{urlhandler}->get($origurl);
        # do we already have an appropriate object?
        ($fileref)=$self->config()->find($url);
        #undef $fileref;
        if (  defined $fileref ) {
         print "found $url in database ----\n";
         $fileref->update();
        }
        else {
         if ( $file eq "" ) {
           $self->parseerror("Unable to get $origurl");
         }
         #-- set up a new preprocess file
         print "Making a new file $url----\n";
         $fileref=ActiveDoc::PreProcessedFile->new($self->config());
         $fileref->url($url);
         $fileref->update();
        }
        return $fileref;
}

sub activatedoc {
        my $self=shift;
        my $url=shift;

        # first get a preprocessed copy of the file 
        my $fileob=$self->getfile($url);

        # now parse it for the <DocType> tag
        $self->{doctypefound}=0;
        $self->newparse("doctype");
        $self->addtag("doctype","Doc", \&Doc_Start, $self,
                                          "", $self, "", $self);
        $self->parse("doctype");

        if ( ! defined $self->{docobject} ) {
          print "No <Doc type=> Specified in ".$fileob->url()."\n";
          exit 1;
        }
        # Set up a new object of the specified type
        my $newobj=$self->{docobject}->new($self->config());
        $newobj->url($url);
        return $newobj;
}

# -------- Error Handling and Error services --------------

sub error {
        my $self=shift;
        my $string=shift;

        die $string."\n";
}

sub parseerror {
        my $self=shift;
        my $string=shift;

        ($line, $file)=$self->line();
        print "Parse Error in ".$file->url().", line ".
                                        $line."\n";
        print $string."\n";
        die;
}

sub checktag {
        my $self=shift;
        my $tagname=shift;
        my $hashref=shift;
        my $param=shift;

        if ( ! exists $$hashref{$param} ) {
          $self->parseerror("Incomplete Tag <$tagname> : $param required");
        }
}

sub line {
        my $self=shift;

        my ($line, $fileobj)=
                $self->{File}->realline($self->{currentparser}->line());
        return ($line, $fileobj);
}

sub tagstartline {
        my $self=shift;
        my ($line, $fileobj)=$self->{File}->line(
                $self->{currentparser}->tagstartline());
        return ($line, $fileobj);
}

sub file {
        my $self=shift;

        $self->{File}->file();
}

# --------------- Initialisation Methods ---------------------------

sub init {
        # Dummy Routine - override for derived classes
}

# ------------------- Tag Routines -----------------------------------
#
# Base - for setting url bases
#
sub Base_start {
        my $self=shift;
        my $name=shift;
        my $hashref=shift;

        $self->checktag($name, $hashref, 'type' );
        $self->checktag($name, $hashref, 'base' );
       
        # Keep track of base tags
        push @{$self->{basestack}}, $$hashref{"type"};
        # Set the base
        $self->{urlhandler}->setbase($$hashref{"type"},$hashref);

}

sub Base_end {
        my $self=shift;
        my $name=shift;
        my $type;

        if ( $#{$self->{basestack}} == -1 ) {
                print "Parse Error : unmatched </".$name."> on line ".
                        $self->line()."\n";
                die;
        }
        else {
          $type = pop @{$self->{basestack}};
          $self->{urlhandler}->unsetbase($type);
        }
}

sub Doc_Start {
        my $self=shift;
        my $name=shift;
        my $hashref=shift;
        
        $self->checktag($name, $hashref, "type");
        $self->{doctypefound}++;
        if ( $self->{doctypefound} == 1 ) { # only take first doctype
           $self->{docobject}=$$hashref{'type'};
        }
}

sub userinterface {
        my $self=shift;
        @_?$self->{userinterface}=shift
          :$self->{userinterface}
}
Revision:	1.13
Committed:	Fri Jan 14 18:49:57 2000 UTC (25 years, 3 months ago) by williamc
Content type:	text/plain
Branch:	MAIN
Changes since 1.12:	+8 -0 lines
Log Message:	New interfaces
#	User	Rev	Content
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	1.6	# 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	williamc	1.9	# checktag(tagname, hashref, param) : check for existence of param in
19			# hashref from a tag call
20	williamc	1.12	# includeparse(local_parsename, objparsename, activedoc) : copy the parse from
21			# one object to another
22			# currentparsename([name]) : get/set current parse name
23	williamc	1.6	# newdoc(file) : Return an new object of the appropriate type
24			# getfile(url) : get a processedfile object given a url
25	williamc	1.9	# activatedoc(url) : Return the object ref for a doc described by the given url
26	williamc	1.6	# config([ActiveConfig]) : Set up/return Configuration for the document
27			# basequery([ActiveConfig]) : Set up/return UserQuery for the doc
28			# copydocconfig(ActiveDoc) : Copy the basic configuration from the ActiveDoc
29			# copydocquery(ActiveDoc) : Copy the basicquery from the ActiveDoc
30	williamc	1.10	# userinterface() : Return the defaullt userinterface
31			# options(var) : return the value of the option var
32	williamc	1.6	#
33			# -- error methods --
34			# error(string) : Report an general error to the user
35			# parseerror(string) : Report an error during parsing a file
36			# line() : Return the current line number of the document
37			# and the ProcessedFileObj it is in
38	williamc	1.2
39			package ActiveDoc::ActiveDoc;
40	williamc	1.6	require 5.004;
41			use ActiveDoc::Parse;
42			use ActiveDoc::ActiveConfig;
43			use ActiveDoc::PreProcessedFile;
44	williamc	1.10	use ObjectUtilities::StorableObject;
45	williamc	1.6	use URL::URLhandler;
46
47	williamc	1.10	@ISA = qw(ObjectUtilities::StorableObject);
48	williamc	1.6
49			sub new {
50			my $class=shift;
51			$self={};
52			bless $self, $class;
53			$self->config(shift);
54
55			# A URL handler per document
56			$self->{urlhandler}=URL::URLhandler->new($self->config()->cache());
57
58	williamc	1.10	# A default UserInterface
59	williamc	1.11	$self->{userinterface}=ActiveDoc::SimpleUserInterface->new();
60	williamc	1.6	$self->init(@_);
61			return $self;
62			}
63
64			# ----- parse related routines --------------
65			sub parse {
66			my $self=shift;
67			$parselabel=shift;
68
69			my $file=$self->file();
70			if ( $file ) {
71	williamc	1.12	$self->{currentparsename}=$parselabel;
72	williamc	1.7	$self->{currentparser}=$self->{parsers}{$parselabel};
73	williamc	1.6	$self->{parsers}{$parselabel}->parse($file,@_);
74	williamc	1.7	delete $self->{currentparser};
75	williamc	1.12	$self->{currentparsename}="";
76	williamc	1.6	}
77			else {
78			print "Cannot parse - file not known\n";
79			}
80	williamc	1.1	}
81
82	williamc	1.12	sub currentparsename {
83			my $self=shift;
84			@_?$self->{currentparsename}=shift
85			:$self->{currentparsename};
86			}
87
88	williamc	1.6	sub newparse {
89			my $self=shift;
90			my $parselabel=shift;
91
92			$self->{parsers}{$parselabel}=ActiveDoc::Parse->new();
93			$self->{parsers}{$parselabel}->addignoretags();
94			$self->{parsers}{$parselabel}->addgrouptags();
95	williamc	1.12	}
96
97	williamc	1.13	sub cleartags {
98			my $self=shift;
99			my $parselabel=shift;
100
101			$self->{parsers}{$parselabel}->cleartags();
102			}
103
104
105	williamc	1.12	sub includeparse {
106			my $self=shift;
107			my $parselabel=shift;
108			my $remoteparselabel=shift;
109			my $activedoc=shift;
110
111			# Some error trapping
112			if ( ! exists $self->{parsers}{$parselabel} ) {
113			$self->error("Unknown local parse name specified");
114			}
115			if ( ! exists $activedoc->{parsers}{$remoteparselabel} ) {
116			$self->error("Unknown parse name specified in remote obj $activedoc");
117			}
118
119			#
120			my $rp=$activedoc->{parsers}{$remoteparselabel};
121			$self->{parsers}{$parselabel}->includeparse($rp);
122	williamc	1.2	}
123	williamc	1.6
124			sub addtag {
125			my $self=shift;
126			my $parselabel=shift;
127			if ( $#_ != 6 ) {
128			$self->error("Incorrect addtags specification\n".
129			"called with :\n@_ \n");
130	williamc	1.4	}
131	williamc	1.6	$self->{parsers}{$parselabel}->addtag(@_);
132			}
133	williamc	1.2
134	williamc	1.6	sub addurltags {
135			my $self=shift;
136			my $parselabel=shift;
137
138			$self->{parsers}{$parselabel}->
139			addtag("Base", \&Base_start, $self, "", $self,
140			\&Base_end, $self);
141	williamc	1.1	}
142
143	williamc	1.6	sub url {
144	williamc	1.2	my $self=shift;
145	williamc	1.8	# get file & preprocess
146			if ( @_ ) {$self->{File}=$self->getfile(shift)}
147			$self->{File}->url();
148	williamc	1.2	}
149
150	williamc	1.6	sub copydocconfig {
151	williamc	1.1	my $self=shift;
152	williamc	1.6	my $ActiveDoc=shift;
153
154			$self->config($ActiveDoc->config());
155
156	williamc	1.1	}
157
158	williamc	1.6	sub copydocquery {
159			my $self=shift;
160			my $ActiveDoc=shift;
161
162			$self->basequery($ActiveDoc->basequery());
163	williamc	1.1	}
164
165	williamc	1.6	sub config {
166	williamc	1.1	my $self=shift;
167	williamc	1.6	@_?$self->{ActiveConfig}=shift
168			: $self->{ActiveConfig};
169			}
170	williamc	1.1
171	williamc	1.6	sub basequery {
172			my $self=shift;
173	williamc	1.10	@_ ? $self->{Query}=shift
174			: $self->{Query};
175			}
176
177			sub options {
178			my $self=shift;
179			my $param=shift;
180			$self->basequery()->getparam('option_'.$param);
181	williamc	1.2	}
182
183	williamc	1.6	sub getfile() {
184	williamc	1.3	my $self=shift;
185	williamc	1.6	my $origurl=shift;
186
187			my $fileref;
188			my ($url, $file)=$self->{urlhandler}->get($origurl);
189			# do we already have an appropriate object?
190	williamc	1.7	($fileref)=$self->config()->find($url);
191			#undef $fileref;
192	williamc	1.6	if ( defined $fileref ) {
193			print "found $url in database ----\n";
194			$fileref->update();
195			}
196			else {
197			if ( $file eq "" ) {
198			$self->parseerror("Unable to get $origurl");
199			}
200			#-- set up a new preprocess file
201			print "Making a new file $url----\n";
202			$fileref=ActiveDoc::PreProcessedFile->new($self->config());
203			$fileref->url($url);
204			$fileref->update();
205			}
206			return $fileref;
207	williamc	1.3	}
208
209	williamc	1.9	sub activatedoc {
210			my $self=shift;
211			my $url=shift;
212
213			# first get a preprocessed copy of the file
214			my $fileob=$self->getfile($url);
215
216			# now parse it for the <DocType> tag
217	williamc	1.10	$self->{doctypefound}=0;
218	williamc	1.9	$self->newparse("doctype");
219			$self->addtag("doctype","Doc", \&Doc_Start, $self,
220			"", $self, "", $self);
221			$self->parse("doctype");
222
223			if ( ! defined $self->{docobject} ) {
224			print "No <Doc type=> Specified in ".$fileob->url()."\n";
225			exit 1;
226			}
227			# Set up a new object of the specified type
228			my $newobj=$self->{docobject}->new($self->config());
229	williamc	1.10	$newobj->url($url);
230	williamc	1.9	return $newobj;
231			}
232
233	williamc	1.6	# -------- Error Handling and Error services --------------
234
235	williamc	1.3	sub error {
236	williamc	1.6	my $self=shift;
237			my $string=shift;
238
239			die $string."\n";
240			}
241
242			sub parseerror {
243			my $self=shift;
244			my $string=shift;
245
246			($line, $file)=$self->line();
247			print "Parse Error in ".$file->url().", line ".
248			$line."\n";
249			print $string."\n";
250			die;
251			}
252
253			sub checktag {
254			my $self=shift;
255			my $tagname=shift;
256			my $hashref=shift;
257			my $param=shift;
258	williamc	1.3
259	williamc	1.6	if ( ! exists $$hashref{$param} ) {
260			$self->parseerror("Incomplete Tag <$tagname> : $param required");
261			}
262			}
263	williamc	1.3
264	williamc	1.6	sub line {
265	williamc	1.7	my $self=shift;
266	williamc	1.9
267	williamc	1.6	my ($line, $fileobj)=
268	williamc	1.9	$self->{File}->realline($self->{currentparser}->line());
269	williamc	1.6	return ($line, $fileobj);
270	williamc	1.7	}
271
272			sub tagstartline {
273			my $self=shift;
274	williamc	1.8	my ($line, $fileobj)=$self->{File}->line(
275	williamc	1.7	$self->{currentparser}->tagstartline());
276			return ($line, $fileobj);
277	williamc	1.3	}
278	williamc	1.6
279			sub file {
280	williamc	1.2	my $self=shift;
281
282	williamc	1.8	$self->{File}->file();
283	williamc	1.2	}
284
285	williamc	1.6	# --------------- Initialisation Methods ---------------------------
286	williamc	1.2
287	williamc	1.6	sub init {
288			# Dummy Routine - override for derived classes
289	williamc	1.1	}
290
291	williamc	1.6	# ------------------- Tag Routines -----------------------------------
292	williamc	1.1	#
293	williamc	1.6	# Base - for setting url bases
294	williamc	1.1	#
295	williamc	1.6	sub Base_start {
296			my $self=shift;
297			my $name=shift;
298			my $hashref=shift;
299	williamc	1.2
300	williamc	1.6	$self->checktag($name, $hashref, 'type' );
301			$self->checktag($name, $hashref, 'base' );
302
303			# Keep track of base tags
304			push @{$self->{basestack}}, $$hashref{"type"};
305			# Set the base
306			$self->{urlhandler}->setbase($$hashref{"type"},$hashref);
307	williamc	1.2
308	williamc	1.1	}
309
310	williamc	1.6	sub Base_end {
311			my $self=shift;
312	williamc	1.1	my $name=shift;
313	williamc	1.6	my $type;
314	williamc	1.1
315	williamc	1.6	if ( $#{$self->{basestack}} == -1 ) {
316			print "Parse Error : unmatched </".$name."> on line ".
317			$self->line()."\n";
318			die;
319			}
320			else {
321			$type = pop @{$self->{basestack}};
322			$self->{urlhandler}->unsetbase($type);
323			}
324	williamc	1.9	}
325
326			sub Doc_Start {
327			my $self=shift;
328			my $name=shift;
329			my $hashref=shift;
330
331			$self->checktag($name, $hashref, "type");
332	williamc	1.10	$self->{doctypefound}++;
333			if ( $self->{doctypefound} == 1 ) { # only take first doctype
334			$self->{docobject}=$$hashref{'type'};
335			}
336			}
337
338			sub userinterface {
339			my $self=shift;
340			@_?$self->{userinterface}=shift
341			:$self->{userinterface}
342	williamc	1.1	}