| 7 |
|
# |
| 8 |
|
# Interface |
| 9 |
|
# --------- |
| 10 |
< |
# new(ActiveConfig[,options]) : A new ActiveDoc object |
| 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 |
< |
# 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 |
< |
# checktag(tagname, hashref, param) : check for existence of param in |
| 19 |
< |
# hashref from a tag call |
| 20 |
< |
# includeparse(local_parsename, objparsename, activedoc) : copy the parse from |
| 21 |
< |
# one object to another |
| 22 |
< |
# currentparsename([name]) : get/set current parse name |
| 23 |
< |
# newdoc(file) : Return an new object of the appropriate type |
| 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 |
| 27 |
– |
# config([ActiveConfig]) : Set up/return Configuration for the document |
| 28 |
– |
# basequery([ActiveConfig]) : Set up/return UserQuery for the doc |
| 29 |
– |
# copydocconfig(ActiveDoc) : Copy the basic configuration from the ActiveDoc |
| 30 |
– |
# copydocquery(ActiveDoc) : Copy the basicquery from the ActiveDoc |
| 20 |
|
# userinterface() : Return the defaullt userinterface |
| 21 |
< |
# option(var) : return the value of the option var |
| 33 |
< |
# requestoption("message") : Ask the user to supply a value for an option |
| 34 |
< |
# if it dosnt already exist |
| 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::Parse; |
| 45 |
< |
use ActiveDoc::ActiveConfig; |
| 46 |
< |
use ActiveDoc::PreProcessedFile; |
| 47 |
< |
use ObjectUtilities::StorableObject; |
| 33 |
> |
use ActiveDoc::SimpleDoc; |
| 34 |
|
use URL::URLhandler; |
| 35 |
|
|
| 36 |
< |
@ISA = qw(ObjectUtilities::StorableObject); |
| 51 |
< |
|
| 52 |
< |
sub new { |
| 53 |
< |
my $class=shift; |
| 54 |
< |
$self={}; |
| 55 |
< |
bless $self, $class; |
| 56 |
< |
$self->config(shift); |
| 57 |
< |
|
| 58 |
< |
# have some override options been passed |
| 59 |
< |
if ( @_ ) { |
| 60 |
< |
$self->basequery(shift); |
| 61 |
< |
} |
| 62 |
< |
else { |
| 63 |
< |
# --- is there a starter document? |
| 64 |
< |
my $basedoc=$self->config()->basedoc(); |
| 65 |
< |
if ( defined $basedoc ) { |
| 66 |
< |
$self->copydocquery($basedoc); |
| 67 |
< |
} |
| 68 |
< |
else { |
| 69 |
< |
$self->error("Error : No base doc found"); |
| 70 |
< |
} |
| 71 |
< |
} |
| 72 |
< |
$self->_init2(); |
| 73 |
< |
} |
| 74 |
< |
|
| 75 |
< |
sub _init2 { |
| 76 |
< |
|
| 77 |
< |
my $self=shift; |
| 78 |
< |
# A URL handler per document |
| 79 |
< |
$self->{urlhandler}=URL::URLhandler->new($self->config()->cache()); |
| 80 |
< |
|
| 81 |
< |
# A default UserInterface |
| 82 |
< |
$self->{userinterface}=ActiveDoc::SimpleUserInterface->new(); |
| 83 |
< |
$self->init(@_); |
| 84 |
< |
return $self; |
| 85 |
< |
|
| 86 |
< |
} |
| 87 |
< |
|
| 88 |
< |
# ----- parse related routines -------------- |
| 89 |
< |
sub parse { |
| 90 |
< |
my $self=shift; |
| 91 |
< |
$parselabel=shift; |
| 92 |
< |
|
| 93 |
< |
my $file=$self->file(); |
| 94 |
< |
if ( $file ) { |
| 95 |
< |
if ( exists $self->{parsers}{$parselabel} ) { |
| 96 |
< |
$self->{currentparsename}=$parselabel; |
| 97 |
< |
$self->{currentparser}=$self->{parsers}{$parselabel}; |
| 98 |
< |
$self->{parsers}{$parselabel}->parse($file,@_); |
| 99 |
< |
delete $self->{currentparser}; |
| 100 |
< |
$self->{currentparsename}=""; |
| 101 |
< |
} |
| 102 |
< |
} |
| 103 |
< |
else { |
| 104 |
< |
print "Cannot parse - file not known\n"; |
| 105 |
< |
} |
| 106 |
< |
} |
| 107 |
< |
|
| 108 |
< |
sub currentparsename { |
| 109 |
< |
my $self=shift; |
| 110 |
< |
@_?$self->{currentparsename}=shift |
| 111 |
< |
:$self->{currentparsename}; |
| 112 |
< |
} |
| 113 |
< |
|
| 114 |
< |
sub newparse { |
| 115 |
< |
my $self=shift; |
| 116 |
< |
my $parselabel=shift; |
| 117 |
< |
|
| 118 |
< |
$self->{parsers}{$parselabel}=ActiveDoc::Parse->new(); |
| 119 |
< |
$self->{parsers}{$parselabel}->addignoretags(); |
| 120 |
< |
$self->{parsers}{$parselabel}->addgrouptags(); |
| 121 |
< |
} |
| 122 |
< |
|
| 123 |
< |
sub cleartags { |
| 124 |
< |
my $self=shift; |
| 125 |
< |
my $parselabel=shift; |
| 126 |
< |
|
| 127 |
< |
$self->{parsers}{$parselabel}->cleartags(); |
| 128 |
< |
} |
| 129 |
< |
|
| 130 |
< |
|
| 131 |
< |
sub includeparse { |
| 132 |
< |
my $self=shift; |
| 133 |
< |
my $parselabel=shift; |
| 134 |
< |
my $remoteparselabel=shift; |
| 135 |
< |
my $activedoc=shift; |
| 136 |
< |
|
| 137 |
< |
# Some error trapping |
| 138 |
< |
if ( ! exists $self->{parsers}{$parselabel} ) { |
| 139 |
< |
$self->error("Unknown local parse name specified"); |
| 140 |
< |
} |
| 141 |
< |
if ( ! exists $activedoc->{parsers}{$remoteparselabel} ) { |
| 142 |
< |
$self->error("Unknown parse name specified in remote obj $activedoc"); |
| 143 |
< |
} |
| 144 |
< |
|
| 145 |
< |
# |
| 146 |
< |
my $rp=$activedoc->{parsers}{$remoteparselabel}; |
| 147 |
< |
$self->{parsers}{$parselabel}->includeparse($rp); |
| 148 |
< |
} |
| 149 |
< |
|
| 150 |
< |
sub addtag { |
| 151 |
< |
my $self=shift; |
| 152 |
< |
my $parselabel=shift; |
| 153 |
< |
if ( $#_ != 6 ) { |
| 154 |
< |
$self->error("Incorrect addtags specification\n". |
| 155 |
< |
"called with :\n@_ \n"); |
| 156 |
< |
} |
| 157 |
< |
$self->{parsers}{$parselabel}->addtag(@_); |
| 158 |
< |
} |
| 36 |
> |
@ISA = qw(ActiveDoc::SimpleDoc); |
| 37 |
|
|
| 38 |
|
sub addurltags { |
| 39 |
|
my $self=shift; |
| 47 |
|
sub url { |
| 48 |
|
my $self=shift; |
| 49 |
|
# get file & preprocess |
| 50 |
< |
if ( @_ ) {$self->{File}=$self->getfile(shift)} |
| 51 |
< |
$self->{File}->url(); |
| 52 |
< |
} |
| 53 |
< |
|
| 54 |
< |
sub copydocconfig { |
| 55 |
< |
my $self=shift; |
| 56 |
< |
my $ActiveDoc=shift; |
| 57 |
< |
|
| 180 |
< |
$self->config($ActiveDoc->config()); |
| 181 |
< |
|
| 182 |
< |
} |
| 183 |
< |
|
| 184 |
< |
sub copydocquery { |
| 185 |
< |
my $self=shift; |
| 186 |
< |
my $ActiveDoc=shift; |
| 187 |
< |
|
| 188 |
< |
$self->basequery($ActiveDoc->basequery()); |
| 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 |
|
sub config { |
| 63 |
|
: $self->{ActiveConfig}; |
| 64 |
|
} |
| 65 |
|
|
| 66 |
< |
sub basequery { |
| 66 |
> |
sub getfile { |
| 67 |
|
my $self=shift; |
| 68 |
< |
@_ ? $self->{Query}=shift |
| 200 |
< |
: $self->{Query}; |
| 201 |
< |
} |
| 68 |
> |
my $origurl=shift; |
| 69 |
|
|
| 70 |
< |
sub option { |
| 71 |
< |
my $self=shift; |
| 72 |
< |
my $param=shift; |
| 73 |
< |
if ( defined $self->basequery()) { |
| 74 |
< |
return $self->basequery()->getparam($param); |
| 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 |
< |
return $undef; |
| 78 |
> |
$self->verbose("Attempting to get $origurl"); |
| 79 |
> |
($url, $file)=$self->{urlhandler}->get($origurl); |
| 80 |
|
} |
| 212 |
– |
} |
| 213 |
– |
|
| 214 |
– |
sub requestoption { |
| 215 |
– |
my $self=shift; |
| 216 |
– |
my $param=shift; |
| 217 |
– |
my $string=shift; |
| 218 |
– |
|
| 219 |
– |
my $par=undef; |
| 220 |
– |
if ( defined $self->basequery()) { |
| 221 |
– |
$par=$self->basequery()->getparam($param); |
| 222 |
– |
while ( ! defined $par ) { |
| 223 |
– |
$self->basequery()->querytype( $param, "basic"); |
| 224 |
– |
$self->basequery()->querymessage( $param, $string); |
| 225 |
– |
$self->userinterface()->askuser($self->basequery()); |
| 226 |
– |
$par=$self->basequery()->getparam($param); |
| 227 |
– |
} |
| 228 |
– |
} |
| 229 |
– |
return $par; |
| 230 |
– |
} |
| 231 |
– |
|
| 232 |
– |
sub getfile() { |
| 233 |
– |
my $self=shift; |
| 234 |
– |
my $origurl=shift; |
| 235 |
– |
|
| 236 |
– |
my $fileref; |
| 237 |
– |
my ($url, $file)=$self->{urlhandler}->get($origurl); |
| 81 |
|
# do we already have an appropriate object? |
| 82 |
|
($fileref)=$self->config()->find($url); |
| 83 |
|
#undef $fileref; |
| 84 |
|
if ( defined $fileref ) { |
| 85 |
< |
print "found $url in database ----\n"; |
| 85 |
> |
$self->verbose("Found $url in database"); |
| 86 |
|
$fileref->update(); |
| 87 |
|
} |
| 88 |
|
else { |
| 90 |
|
$self->parseerror("Unable to get $origurl"); |
| 91 |
|
} |
| 92 |
|
#-- set up a new preprocess file |
| 93 |
< |
print "Making a new file $url----\n"; |
| 93 |
> |
$self->verbose("Making a new preprocessed file $url"); |
| 94 |
|
$fileref=ActiveDoc::PreProcessedFile->new($self->config()); |
| 95 |
|
$fileref->url($url); |
| 96 |
|
$fileref->update(); |
| 107 |
|
|
| 108 |
|
# now parse it for the <DocType> tag |
| 109 |
|
my $tempdoc=ActiveDoc::ActiveDoc->new($self->config()); |
| 110 |
< |
$tempdoc->url($url); |
| 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, |
| 126 |
|
my $newobj=$tempdoc->{docobject}->new($self->config()); |
| 127 |
|
undef $tempdoc; |
| 128 |
|
$newobj->url($url); |
| 129 |
< |
$newobj->_initparse(); |
| 129 |
> |
$newobj->parent($self); |
| 130 |
|
return $newobj; |
| 131 |
|
} |
| 132 |
|
|
| 133 |
< |
sub _initparse { |
| 133 |
> |
sub parent { |
| 134 |
|
my $self=shift; |
| 135 |
|
|
| 136 |
< |
$self->parse("init"); |
| 136 |
> |
@_?$self->{parent}=shift |
| 137 |
> |
:$self->{parent}; |
| 138 |
|
} |
| 139 |
+ |
|
| 140 |
|
# -------- Error Handling and Error services -------------- |
| 141 |
|
|
| 142 |
|
sub error { |
| 150 |
|
my $self=shift; |
| 151 |
|
my $string=shift; |
| 152 |
|
|
| 153 |
< |
($line, $file)=$self->line(); |
| 154 |
< |
print "Parse Error in ".$file->url().", line ". |
| 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 |
< |
die; |
| 162 |
< |
} |
| 312 |
< |
|
| 313 |
< |
sub checktag { |
| 314 |
< |
my $self=shift; |
| 315 |
< |
my $tagname=shift; |
| 316 |
< |
my $hashref=shift; |
| 317 |
< |
my $param=shift; |
| 318 |
< |
|
| 319 |
< |
if ( ! exists $$hashref{$param} ) { |
| 320 |
< |
$self->parseerror("Incomplete Tag <$tagname> : $param required"); |
| 321 |
< |
} |
| 160 |
> |
print $string."\n"; |
| 161 |
> |
exit; |
| 162 |
> |
} |
| 163 |
|
} |
| 164 |
|
|
| 165 |
|
sub line { |
| 183 |
|
$self->{File}->file(); |
| 184 |
|
} |
| 185 |
|
|
| 186 |
+ |
sub ProcessFile { |
| 187 |
+ |
my $self=shift; |
| 188 |
+ |
|
| 189 |
+ |
return $self->{File}->ProcessedFile(); |
| 190 |
+ |
} |
| 191 |
+ |
|
| 192 |
|
# --------------- Initialisation Methods --------------------------- |
| 193 |
|
|
| 194 |
|
sub init { |
| 210 |
|
# Keep track of base tags |
| 211 |
|
push @{$self->{basestack}}, $$hashref{"type"}; |
| 212 |
|
# Set the base |
| 366 |
– |
print "BASE SET for ".$$hashref{"type"}."\n"; |
| 213 |
|
$self->{urlhandler}->setbase($$hashref{"type"},$hashref); |
| 368 |
– |
print "BASE SET for ".$$hashref{"type"}."\n"; |
| 369 |
– |
|
| 214 |
|
} |
| 215 |
|
|
| 216 |
|
sub Base_end { |
| 219 |
|
my $type; |
| 220 |
|
|
| 221 |
|
if ( $#{$self->{basestack}} == -1 ) { |
| 222 |
< |
print "Parse Error : unmatched </".$name."> on line ". |
| 379 |
< |
$self->line()."\n"; |
| 380 |
< |
die; |
| 222 |
> |
$self->parseerror("Parse Error : unmatched </$name>"); |
| 223 |
|
} |
| 224 |
|
else { |
| 225 |
|
$type = pop @{$self->{basestack}}; |
| 238 |
|
$self->{docobject}=$$hashref{'type'}; |
| 239 |
|
} |
| 240 |
|
} |
| 399 |
– |
|
| 400 |
– |
sub userinterface { |
| 401 |
– |
my $self=shift; |
| 402 |
– |
@_?$self->{userinterface}=shift |
| 403 |
– |
:$self->{userinterface} |
| 404 |
– |
} |
