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
# 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
#
# -- 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::ObjectBase;
use URL::URLhandler;

@ISA = qw(ObjectUtilities::ObjectBase);

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());

        $self->init(@_);
        return $self;
}

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

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

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

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

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->{UserQuery}=shift
           : $self->{UserQuery};
}

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->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());
        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;

        print keys %$hashref;
        print "-----------------------------------\n";
        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->{docobject}=$$hashref{'type'};
}
Revision:	1.9
Committed:	Mon Nov 29 17:32:14 1999 UTC (25 years, 5 months ago) by williamc
Content type:	text/plain
Branch:	MAIN
Changes since 1.8:	+38 -1 lines
Log Message:	activatdoc tested basic-level
#	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	1.6	# newdoc(file) : Return an new object of the appropriate type
21			# getfile(url) : get a processedfile object given a url
22	williamc	1.9	# activatedoc(url) : Return the object ref for a doc described by the given url
23	williamc	1.6	# config([ActiveConfig]) : Set up/return Configuration for the document
24			# basequery([ActiveConfig]) : Set up/return UserQuery for the doc
25			# copydocconfig(ActiveDoc) : Copy the basic configuration from the ActiveDoc
26			# copydocquery(ActiveDoc) : Copy the basicquery from the ActiveDoc
27			#
28			# -- error methods --
29			# error(string) : Report an general error to the user
30			# parseerror(string) : Report an error during parsing a file
31			# line() : Return the current line number of the document
32			# and the ProcessedFileObj it is in
33	williamc	1.2
34			package ActiveDoc::ActiveDoc;
35	williamc	1.6	require 5.004;
36			use ActiveDoc::Parse;
37			use ActiveDoc::ActiveConfig;
38			use ActiveDoc::PreProcessedFile;
39			use ObjectUtilities::ObjectBase;
40			use URL::URLhandler;
41
42			@ISA = qw(ObjectUtilities::ObjectBase);
43
44			sub new {
45			my $class=shift;
46			$self={};
47			bless $self, $class;
48			$self->config(shift);
49
50			# A URL handler per document
51			$self->{urlhandler}=URL::URLhandler->new($self->config()->cache());
52
53			$self->init(@_);
54			return $self;
55			}
56
57			# ----- parse related routines --------------
58			sub parse {
59			my $self=shift;
60			$parselabel=shift;
61
62			my $file=$self->file();
63			if ( $file ) {
64	williamc	1.7	$self->{currentparser}=$self->{parsers}{$parselabel};
65	williamc	1.6	$self->{parsers}{$parselabel}->parse($file,@_);
66	williamc	1.7	delete $self->{currentparser};
67	williamc	1.6	}
68			else {
69			print "Cannot parse - file not known\n";
70			}
71	williamc	1.1	}
72
73	williamc	1.6	sub newparse {
74			my $self=shift;
75			my $parselabel=shift;
76
77			$self->{parsers}{$parselabel}=ActiveDoc::Parse->new();
78			$self->{parsers}{$parselabel}->addignoretags();
79			$self->{parsers}{$parselabel}->addgrouptags();
80	williamc	1.2	}
81	williamc	1.6
82			sub addtag {
83			my $self=shift;
84			my $parselabel=shift;
85			if ( $#_ != 6 ) {
86			$self->error("Incorrect addtags specification\n".
87			"called with :\n@_ \n");
88	williamc	1.4	}
89	williamc	1.6	$self->{parsers}{$parselabel}->addtag(@_);
90			}
91	williamc	1.2
92	williamc	1.6	sub addurltags {
93			my $self=shift;
94			my $parselabel=shift;
95
96			$self->{parsers}{$parselabel}->
97			addtag("Base", \&Base_start, $self, "", $self,
98			\&Base_end, $self);
99	williamc	1.1	}
100
101	williamc	1.6	sub url {
102	williamc	1.2	my $self=shift;
103	williamc	1.8	# get file & preprocess
104			if ( @_ ) {$self->{File}=$self->getfile(shift)}
105			$self->{File}->url();
106	williamc	1.2	}
107
108	williamc	1.6	sub copydocconfig {
109	williamc	1.1	my $self=shift;
110	williamc	1.6	my $ActiveDoc=shift;
111
112			$self->config($ActiveDoc->config());
113
114	williamc	1.1	}
115
116	williamc	1.6	sub copydocquery {
117			my $self=shift;
118			my $ActiveDoc=shift;
119
120			$self->basequery($ActiveDoc->basequery());
121	williamc	1.1	}
122
123	williamc	1.6	sub config {
124	williamc	1.1	my $self=shift;
125	williamc	1.6	@_?$self->{ActiveConfig}=shift
126			: $self->{ActiveConfig};
127			}
128	williamc	1.1
129	williamc	1.6	sub basequery {
130			my $self=shift;
131			@_ ? $self->{UserQuery}=shift
132			: $self->{UserQuery};
133	williamc	1.2	}
134
135	williamc	1.6	sub getfile() {
136	williamc	1.3	my $self=shift;
137	williamc	1.6	my $origurl=shift;
138
139			my $fileref;
140			my ($url, $file)=$self->{urlhandler}->get($origurl);
141			# do we already have an appropriate object?
142	williamc	1.7	($fileref)=$self->config()->find($url);
143			#undef $fileref;
144	williamc	1.6	if ( defined $fileref ) {
145			print "found $url in database ----\n";
146			$fileref->update();
147			}
148			else {
149			if ( $file eq "" ) {
150			$self->parseerror("Unable to get $origurl");
151			}
152			#-- set up a new preprocess file
153			print "Making a new file $url----\n";
154			$fileref=ActiveDoc::PreProcessedFile->new($self->config());
155			$fileref->url($url);
156			$fileref->update();
157			}
158			return $fileref;
159	williamc	1.3	}
160
161	williamc	1.9	sub activatedoc {
162			my $self=shift;
163			my $url=shift;
164
165			# first get a preprocessed copy of the file
166			my $fileob=$self->getfile($url);
167
168			# now parse it for the <DocType> tag
169			$self->newparse("doctype");
170			$self->addtag("doctype","Doc", \&Doc_Start, $self,
171			"", $self, "", $self);
172			$self->parse("doctype");
173
174			if ( ! defined $self->{docobject} ) {
175			print "No <Doc type=> Specified in ".$fileob->url()."\n";
176			exit 1;
177			}
178			# Set up a new object of the specified type
179			my $newobj=$self->{docobject}->new($self->config());
180			return $newobj;
181			}
182
183	williamc	1.6	# -------- Error Handling and Error services --------------
184
185	williamc	1.3	sub error {
186	williamc	1.6	my $self=shift;
187			my $string=shift;
188
189			die $string."\n";
190			}
191
192			sub parseerror {
193			my $self=shift;
194			my $string=shift;
195
196			($line, $file)=$self->line();
197			print "Parse Error in ".$file->url().", line ".
198			$line."\n";
199			print $string."\n";
200			die;
201			}
202
203			sub checktag {
204			my $self=shift;
205			my $tagname=shift;
206			my $hashref=shift;
207			my $param=shift;
208	williamc	1.3
209	williamc	1.9	print keys %$hashref;
210			print "-----------------------------------\n";
211	williamc	1.6	if ( ! exists $$hashref{$param} ) {
212			$self->parseerror("Incomplete Tag <$tagname> : $param required");
213			}
214			}
215	williamc	1.3
216	williamc	1.6	sub line {
217	williamc	1.7	my $self=shift;
218	williamc	1.9
219	williamc	1.6	my ($line, $fileobj)=
220	williamc	1.9	$self->{File}->realline($self->{currentparser}->line());
221	williamc	1.6	return ($line, $fileobj);
222	williamc	1.7	}
223
224			sub tagstartline {
225			my $self=shift;
226	williamc	1.8	my ($line, $fileobj)=$self->{File}->line(
227	williamc	1.7	$self->{currentparser}->tagstartline());
228			return ($line, $fileobj);
229	williamc	1.3	}
230	williamc	1.6
231			sub file {
232	williamc	1.2	my $self=shift;
233
234	williamc	1.8	$self->{File}->file();
235	williamc	1.2	}
236
237	williamc	1.6	# --------------- Initialisation Methods ---------------------------
238	williamc	1.2
239	williamc	1.6	sub init {
240			# Dummy Routine - override for derived classes
241	williamc	1.1	}
242
243	williamc	1.6	# ------------------- Tag Routines -----------------------------------
244	williamc	1.1	#
245	williamc	1.6	# Base - for setting url bases
246	williamc	1.1	#
247	williamc	1.6	sub Base_start {
248			my $self=shift;
249			my $name=shift;
250			my $hashref=shift;
251	williamc	1.2
252	williamc	1.6	$self->checktag($name, $hashref, 'type' );
253			$self->checktag($name, $hashref, 'base' );
254
255			# Keep track of base tags
256			push @{$self->{basestack}}, $$hashref{"type"};
257			# Set the base
258			$self->{urlhandler}->setbase($$hashref{"type"},$hashref);
259	williamc	1.2
260	williamc	1.1	}
261
262	williamc	1.6	sub Base_end {
263			my $self=shift;
264	williamc	1.1	my $name=shift;
265	williamc	1.6	my $type;
266	williamc	1.1
267	williamc	1.6	if ( $#{$self->{basestack}} == -1 ) {
268			print "Parse Error : unmatched </".$name."> on line ".
269			$self->line()."\n";
270			die;
271			}
272			else {
273			$type = pop @{$self->{basestack}};
274			$self->{urlhandler}->unsetbase($type);
275			}
276	williamc	1.9	}
277
278			sub Doc_Start {
279			my $self=shift;
280			my $name=shift;
281			my $hashref=shift;
282
283			$self->checktag($name, $hashref, "type");
284			$self->{docobject}=$$hashref{'type'};
285	williamc	1.1	}