| 10 |
|
# new(ActiveConfig[,options]) : A new ActiveDoc object |
| 11 |
|
# url() : Return/set the docs url - essential |
| 12 |
|
# file() : Return the local filename of document |
| 13 |
+ |
# ProcessFile() : Return the filename of PreProcessed document |
| 14 |
|
# |
| 15 |
|
# parse(parselabel): Parse the document file for the given parse level |
| 16 |
+ |
# parent() : return the object ref of the calling parent |
| 17 |
|
# newparse(parselabel) : Create a new parse type |
| 18 |
|
# addtag(parselabel,tagname,start,obj,text,obj,end,obj) |
| 19 |
|
# : Add tags to the parse given by label |
| 21 |
|
# hashref from a tag call |
| 22 |
|
# includeparse(local_parsename, objparsename, activedoc) : copy the parse from |
| 23 |
|
# one object to another |
| 24 |
+ |
# currentparser() : return the current parser object |
| 25 |
|
# currentparsename([name]) : get/set current parse name |
| 23 |
– |
# newdoc(file) : Return an new object of the appropriate type |
| 26 |
|
# getfile(url) : get a processedfile object given a url |
| 27 |
|
# activatedoc(url) : Return the object ref for a doc described by the given url |
| 28 |
|
# -- any parse called "init" will also be run |
| 31 |
|
# copydocconfig(ActiveDoc) : Copy the basic configuration from the ActiveDoc |
| 32 |
|
# copydocquery(ActiveDoc) : Copy the basicquery from the ActiveDoc |
| 33 |
|
# userinterface() : Return the defaullt userinterface |
| 34 |
< |
# option(var) : return the value of the option var |
| 34 |
> |
# option(var) : return the value of the option var ( or undef ) |
| 35 |
|
# requestoption("message") : Ask the user to supply a value for an option |
| 36 |
|
# if it dosnt already exist |
| 37 |
+ |
# askuser(Query) : send a query object to the userinterface |
| 38 |
+ |
# verbose(string) : Print string in verbosity mode |
| 39 |
|
# |
| 40 |
|
# -- error methods -- |
| 41 |
|
# error(string) : Report an general error to the user |
| 42 |
|
# parseerror(string) : Report an error during parsing a file |
| 43 |
|
# line() : Return the current line number of the document |
| 44 |
|
# and the ProcessedFileObj it is in |
| 45 |
+ |
# |
| 46 |
+ |
# -- support for inheriting classes |
| 47 |
+ |
# _saveactivedoc(filehandle) |
| 48 |
+ |
# _restoreactivedoc(filehandle) |
| 49 |
|
|
| 50 |
|
package ActiveDoc::ActiveDoc; |
| 51 |
|
require 5.004; |
| 72 |
|
my $basedoc=$self->config()->basedoc(); |
| 73 |
|
if ( defined $basedoc ) { |
| 74 |
|
$self->copydocquery($basedoc); |
| 75 |
+ |
$self->verbose("Initialising from $basedoc"); |
| 76 |
|
} |
| 77 |
|
else { |
| 78 |
< |
$self->error("Error : No base doc found"); |
| 78 |
> |
$self->error("ActiveDoc Error : No base doc found"); |
| 79 |
|
} |
| 80 |
|
} |
| 81 |
+ |
$self->verbose("New ActiveDoc (".ref($self).") Created"); |
| 82 |
|
$self->_init2(); |
| 83 |
|
} |
| 84 |
|
|
| 95 |
|
|
| 96 |
|
} |
| 97 |
|
|
| 98 |
+ |
sub verbose { |
| 99 |
+ |
my $self=shift; |
| 100 |
+ |
my $string=shift; |
| 101 |
+ |
|
| 102 |
+ |
if ( $self->option('verbose_all') || |
| 103 |
+ |
$self->option('verbose_'.ref($self)) ) { |
| 104 |
+ |
print ">".ref($self)."($self) : \n->".$string."\n"; |
| 105 |
+ |
} |
| 106 |
+ |
} |
| 107 |
+ |
|
| 108 |
|
# ----- parse related routines -------------- |
| 109 |
|
sub parse { |
| 110 |
|
my $self=shift; |
| 111 |
|
$parselabel=shift; |
| 112 |
|
|
| 113 |
< |
my $file=$self->file(); |
| 113 |
> |
my $file=$self->ProcessFile(); |
| 114 |
|
if ( $file ) { |
| 115 |
|
if ( exists $self->{parsers}{$parselabel} ) { |
| 116 |
+ |
$self->verbose("Parsing $parselabel in file $file"); |
| 117 |
|
$self->{currentparsename}=$parselabel; |
| 118 |
|
$self->{currentparser}=$self->{parsers}{$parselabel}; |
| 119 |
|
$self->{parsers}{$parselabel}->parse($file,@_); |
| 120 |
|
delete $self->{currentparser}; |
| 121 |
|
$self->{currentparsename}=""; |
| 122 |
+ |
$self->verbose("Parse $parselabel Complete"); |
| 123 |
|
} |
| 124 |
|
} |
| 125 |
|
else { |
| 133 |
|
:$self->{currentparsename}; |
| 134 |
|
} |
| 135 |
|
|
| 136 |
+ |
sub currentparser { |
| 137 |
+ |
my $self=shift; |
| 138 |
+ |
return $self->{currentparser}; |
| 139 |
+ |
} |
| 140 |
+ |
|
| 141 |
+ |
|
| 142 |
|
sub newparse { |
| 143 |
|
my $self=shift; |
| 144 |
|
my $parselabel=shift; |
| 197 |
|
sub url { |
| 198 |
|
my $self=shift; |
| 199 |
|
# get file & preprocess |
| 200 |
< |
if ( @_ ) {$self->{File}=$self->getfile(shift)} |
| 200 |
> |
if ( @_ ) { |
| 201 |
> |
$self->{File}=$self->getfile(shift); |
| 202 |
> |
$self->verbose("url downloaded to $self->{File}"); |
| 203 |
> |
} |
| 204 |
|
$self->{File}->url(); |
| 205 |
|
} |
| 206 |
|
|
| 216 |
|
my $self=shift; |
| 217 |
|
my $ActiveDoc=shift; |
| 218 |
|
|
| 219 |
< |
$self->basequery($ActiveDoc->basequery()); |
| 219 |
> |
if ( defined $ActiveDoc->basequery() ) { |
| 220 |
> |
$self->basequery($ActiveDoc->basequery()); |
| 221 |
> |
} |
| 222 |
> |
else { |
| 223 |
> |
$self->error("Cannot copy basequery - undefined"); |
| 224 |
> |
} |
| 225 |
|
} |
| 226 |
|
|
| 227 |
|
sub config { |
| 232 |
|
|
| 233 |
|
sub basequery { |
| 234 |
|
my $self=shift; |
| 235 |
< |
@_ ? $self->{Query}=shift |
| 236 |
< |
: $self->{Query}; |
| 235 |
> |
@_?$self->{Query}=shift |
| 236 |
> |
:$self->{Query}; |
| 237 |
|
} |
| 238 |
|
|
| 239 |
|
sub option { |
| 265 |
|
return $par; |
| 266 |
|
} |
| 267 |
|
|
| 268 |
< |
sub getfile() { |
| 268 |
> |
sub askuser { |
| 269 |
> |
my $self=shift; |
| 270 |
> |
return $self->userinterface()->askuser(@_); |
| 271 |
> |
} |
| 272 |
> |
|
| 273 |
> |
sub getfile { |
| 274 |
|
my $self=shift; |
| 275 |
|
my $origurl=shift; |
| 276 |
|
|
| 277 |
|
my $fileref; |
| 278 |
< |
my ($url, $file)=$self->{urlhandler}->get($origurl); |
| 278 |
> |
my ($url, $file); |
| 279 |
> |
if ( (defined ($it=$self->option('url_update'))) && |
| 280 |
> |
( $it eq "1" || $origurl=~/^$it/ )) { |
| 281 |
> |
$self->verbose("Forced download of $origurl"); |
| 282 |
> |
($url, $file)=$self->{urlhandler}->download($origurl); |
| 283 |
> |
} |
| 284 |
> |
else { |
| 285 |
> |
$self->verbose("Attempting to get $origurl"); |
| 286 |
> |
($url, $file)=$self->{urlhandler}->get($origurl); |
| 287 |
> |
} |
| 288 |
|
# do we already have an appropriate object? |
| 289 |
|
($fileref)=$self->config()->find($url); |
| 290 |
|
#undef $fileref; |
| 291 |
|
if ( defined $fileref ) { |
| 292 |
< |
print "found $url in database ----\n"; |
| 292 |
> |
$self->verbose("Found $url in database"); |
| 293 |
|
$fileref->update(); |
| 294 |
|
} |
| 295 |
|
else { |
| 297 |
|
$self->parseerror("Unable to get $origurl"); |
| 298 |
|
} |
| 299 |
|
#-- set up a new preprocess file |
| 300 |
< |
print "Making a new file $url----\n"; |
| 300 |
> |
$self->verbose("Making a new preprocessed file $url"); |
| 301 |
|
$fileref=ActiveDoc::PreProcessedFile->new($self->config()); |
| 302 |
|
$fileref->url($url); |
| 303 |
|
$fileref->update(); |
| 315 |
|
# now parse it for the <DocType> tag |
| 316 |
|
my $tempdoc=ActiveDoc::ActiveDoc->new($self->config()); |
| 317 |
|
$tempdoc->{urlhandler}=$self->{urlhandler}; |
| 318 |
< |
$tempdoc->url($url); |
| 318 |
> |
my $fullurl=$tempdoc->url($url); |
| 319 |
> |
$url=$fullurl; |
| 320 |
|
$tempdoc->{doctypefound}=0; |
| 321 |
|
$tempdoc->newparse("doctype"); |
| 322 |
|
$tempdoc->addtag("doctype","Doc", \&Doc_Start, $tempdoc, |
| 333 |
|
my $newobj=$tempdoc->{docobject}->new($self->config()); |
| 334 |
|
undef $tempdoc; |
| 335 |
|
$newobj->url($url); |
| 336 |
+ |
$newobj->parent($self); |
| 337 |
|
$newobj->_initparse(); |
| 338 |
|
return $newobj; |
| 339 |
|
} |
| 340 |
|
|
| 341 |
+ |
sub parent { |
| 342 |
+ |
my $self=shift; |
| 343 |
+ |
|
| 344 |
+ |
@_?$self->{parent}=shift |
| 345 |
+ |
:$self->{parent}; |
| 346 |
+ |
} |
| 347 |
+ |
|
| 348 |
|
sub _initparse { |
| 349 |
|
my $self=shift; |
| 350 |
|
|
| 363 |
|
my $self=shift; |
| 364 |
|
my $string=shift; |
| 365 |
|
|
| 366 |
< |
($line, $file)=$self->line(); |
| 367 |
< |
print "Parse Error in ".$file->url().", line ". |
| 366 |
> |
if ( $self->currentparsename() eq "" ) { |
| 367 |
> |
$self->error($string); |
| 368 |
> |
} |
| 369 |
> |
else { |
| 370 |
> |
($line, $file)=$self->line(); |
| 371 |
> |
print "Parse Error in ".$file->url().", line ". |
| 372 |
|
$line."\n"; |
| 373 |
< |
print $string."\n"; |
| 374 |
< |
die; |
| 373 |
> |
print $string."\n"; |
| 374 |
> |
exit; |
| 375 |
> |
} |
| 376 |
|
} |
| 377 |
|
|
| 378 |
|
sub checktag { |
| 407 |
|
$self->{File}->file(); |
| 408 |
|
} |
| 409 |
|
|
| 410 |
+ |
sub ProcessFile { |
| 411 |
+ |
my $self=shift; |
| 412 |
+ |
|
| 413 |
+ |
return $self->{File}->ProcessedFile(); |
| 414 |
+ |
} |
| 415 |
+ |
|
| 416 |
|
# --------------- Initialisation Methods --------------------------- |
| 417 |
|
|
| 418 |
|
sub init { |
| 443 |
|
my $type; |
| 444 |
|
|
| 445 |
|
if ( $#{$self->{basestack}} == -1 ) { |
| 446 |
< |
print "Parse Error : unmatched </".$name."> on line ". |
| 377 |
< |
$self->line()."\n"; |
| 378 |
< |
die; |
| 446 |
> |
$self->parseerror("Parse Error : unmatched </$name>"); |
| 447 |
|
} |
| 448 |
|
else { |
| 449 |
|
$type = pop @{$self->{basestack}}; |
| 468 |
|
@_?$self->{userinterface}=shift |
| 469 |
|
:$self->{userinterface} |
| 470 |
|
} |
| 471 |
+ |
|
| 472 |
+ |
sub _saveactivedoc { |
| 473 |
+ |
my $self=shift; |
| 474 |
+ |
my $fh=shift; |
| 475 |
+ |
print "Storing $self\n"; |
| 476 |
+ |
print $fh $self->url()."\n"; |
| 477 |
+ |
} |
| 478 |
+ |
|
| 479 |
+ |
sub _restoreactivedoc { |
| 480 |
+ |
my $self=shift; |
| 481 |
+ |
my $fh=shift; |
| 482 |
+ |
|
| 483 |
+ |
my $url=<$fh>; |
| 484 |
+ |
chomp $url; |
| 485 |
+ |
$self->url($url); |
| 486 |
+ |
} |
