| 8 |
|
# Interface |
| 9 |
|
# --------- |
| 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 |
< |
# 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 |
< |
# config([ActiveConfig]) : Set up/return Configuration for the document |
| 20 |
< |
# basequery([ActiveConfig]) : Set up/return UserQuery for the doc |
| 21 |
< |
# copydocconfig(ActiveDoc) : Copy the basic configuration from the ActiveDoc |
| 26 |
< |
# copydocquery(ActiveDoc) : Copy the basicquery from the ActiveDoc |
| 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::Parse; |
| 37 |
< |
use ActiveDoc::ActiveConfig; |
| 38 |
< |
use ActiveDoc::PreProcessedFile; |
| 39 |
< |
use ObjectUtilities::ObjectBase; |
| 33 |
> |
use ActiveDoc::SimpleDoc; |
| 34 |
|
use URL::URLhandler; |
| 35 |
|
|
| 36 |
< |
@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 |
< |
$self->{currentparser}=$self->{parsers}{$parselabel}; |
| 65 |
< |
$self->{parsers}{$parselabel}->parse($file,@_); |
| 66 |
< |
delete $self->{currentparser}; |
| 67 |
< |
} |
| 68 |
< |
else { |
| 69 |
< |
print "Cannot parse - file not known\n"; |
| 70 |
< |
} |
| 71 |
< |
} |
| 72 |
< |
|
| 73 |
< |
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 |
< |
} |
| 81 |
< |
|
| 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 |
< |
} |
| 89 |
< |
$self->{parsers}{$parselabel}->addtag(@_); |
| 90 |
< |
} |
| 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 |
< |
|
| 112 |
< |
$self->config($ActiveDoc->config()); |
| 113 |
< |
|
| 114 |
< |
} |
| 115 |
< |
|
| 116 |
< |
sub copydocquery { |
| 117 |
< |
my $self=shift; |
| 118 |
< |
my $ActiveDoc=shift; |
| 119 |
< |
|
| 120 |
< |
$self->basequery($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 { |
| 130 |
< |
my $self=shift; |
| 131 |
< |
@_ ? $self->{UserQuery}=shift |
| 132 |
< |
: $self->{UserQuery}; |
| 133 |
< |
} |
| 134 |
< |
|
| 135 |
< |
sub getfile() { |
| 66 |
> |
sub getfile { |
| 67 |
|
my $self=shift; |
| 68 |
|
my $origurl=shift; |
| 69 |
|
|
| 70 |
|
my $fileref; |
| 71 |
< |
my ($url, $file)=$self->{urlhandler}->get($origurl); |
| 71 |
> |
my ($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 |
< |
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(); |
| 103 |
|
my $url=shift; |
| 104 |
|
|
| 105 |
|
# first get a preprocessed copy of the file |
| 106 |
< |
my $fileob=$self->getfile($url); |
| 106 |
> |
# my $fileob=$self->getfile($url); |
| 107 |
|
|
| 108 |
|
# now parse it for the <DocType> tag |
| 109 |
< |
$self->newparse("doctype"); |
| 110 |
< |
$self->addtag("doctype","Doc", \&Doc_Start, $self, |
| 111 |
< |
"", $self, "", $self); |
| 112 |
< |
$self->parse("doctype"); |
| 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 $self->{docobject} ) { |
| 120 |
< |
print "No <Doc type=> Specified in ".$fileob->url()."\n"; |
| 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 |
< |
my $newobj=$self->{docobject}->new($self->config()); |
| 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 parent { |
| 134 |
+ |
my $self=shift; |
| 135 |
+ |
|
| 136 |
+ |
@_?$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 |
< |
} |
| 202 |
< |
|
| 203 |
< |
sub checktag { |
| 204 |
< |
my $self=shift; |
| 205 |
< |
my $tagname=shift; |
| 206 |
< |
my $hashref=shift; |
| 207 |
< |
my $param=shift; |
| 208 |
< |
|
| 209 |
< |
print keys %$hashref; |
| 210 |
< |
print "-----------------------------------\n"; |
| 211 |
< |
if ( ! exists $$hashref{$param} ) { |
| 212 |
< |
$self->parseerror("Incomplete Tag <$tagname> : $param required"); |
| 213 |
< |
} |
| 160 |
> |
print $string."\n"; |
| 161 |
> |
exit; |
| 162 |
> |
} |
| 163 |
|
} |
| 164 |
|
|
| 165 |
|
sub line { |
| 183 |
|
$self->{File}->file(); |
| 184 |
|
} |
| 185 |
|
|
| 186 |
+ |
sub ProcessFile { |
| 187 |
+ |
my $self=shift; |
| 188 |
+ |
|
| 189 |
+ |
return $self->{File}->ProcessedFile(); |
| 190 |
+ |
} |
| 191 |
+ |
|
| 192 |
|
# --------------- Initialisation Methods --------------------------- |
| 193 |
|
|
| 194 |
|
sub init { |
| 211 |
|
push @{$self->{basestack}}, $$hashref{"type"}; |
| 212 |
|
# Set the base |
| 213 |
|
$self->{urlhandler}->setbase($$hashref{"type"},$hashref); |
| 259 |
– |
|
| 214 |
|
} |
| 215 |
|
|
| 216 |
|
sub Base_end { |
| 219 |
|
my $type; |
| 220 |
|
|
| 221 |
|
if ( $#{$self->{basestack}} == -1 ) { |
| 222 |
< |
print "Parse Error : unmatched </".$name."> on line ". |
| 269 |
< |
$self->line()."\n"; |
| 270 |
< |
die; |
| 222 |
> |
$self->parseerror("Parse Error : unmatched </$name>"); |
| 223 |
|
} |
| 224 |
|
else { |
| 225 |
|
$type = pop @{$self->{basestack}}; |
| 233 |
|
my $hashref=shift; |
| 234 |
|
|
| 235 |
|
$self->checktag($name, $hashref, "type"); |
| 236 |
< |
$self->{docobject}=$$hashref{'type'}; |
| 236 |
> |
$self->{doctypefound}++; |
| 237 |
> |
if ( $self->{doctypefound} == 1 ) { # only take first doctype |
| 238 |
> |
$self->{docobject}=$$hashref{'type'}; |
| 239 |
> |
} |
| 240 |
|
} |
