Changeset 22338

Timestamp:

2010-07-02T12:57:33+12:00 (14 years ago)

Author:

ak19

Message:

Image URLs are adjusted to refer to their location in Fedora.

Location:

main/trunk/greenstone2

Files:

• bin/script/g2f-import.pl (modified) (2 diffs)
• perllib/plugouts/FedoraMETSPlugout.pm (modified) (2 diffs)

main/trunk/greenstone2/bin/script/g2f-import.pl

@@ -15 +15 @@
     $ENV{'FEDORA_PROTOCOL'} = "http" if (!defined $ENV{'FEDORA_PROTOCOL'});
     $ENV{'FEDORA_PID_NAMESPACE'} = "greenstone" if (!defined $ENV{'FEDORA_PID_NAMESPACE'});
+    $ENV{'FEDORA_PREFIX'} = "/fedora" if (!defined $ENV{'FEDORA_PREFIX'});
 
     unshift (@INC, "$ENV{'GSDLHOME'}/perllib/");
@@ -187 +188 @@
     }
 
-    # if GS3, and if Fedora uses Greenstone's tomcat, then we do not need to write out the file gsdl.xml into Fedora's tomcat
+    # if GS3, and if Fedora uses Greenstone 3's tomcat, then we do not need to write out the file gsdl.xml into Fedora's tomcat
     my $localfedora = &util::filename_cat($ENV{'GSDL3SRCHOME'}, "packages", "tomcat", "conf", "Catalina", "localhost", "fedora.xml");
     unless($ENV{'GSDL3SRCHOME'} && -e $localfedora) {

main/trunk/greenstone2/perllib/plugouts/FedoraMETSPlugout.pm

@@ -131 +131 @@
 
     $self->output_xml_header($outhandler);
+
+    ## change links to be Fedora cognant:
+    my $txt = $section_ptr->{'text'};
+    $section_ptr->{'text'} = $self->adjust_links($doc_obj, \$txt);
+
     $self->output_txt_section($outhandler,$doc_obj, $section);
     $self->output_xml_footer($outhandler);
@@ -148 +153 @@
     $self->saveas_doctxt_section($doc_obj, $working_dir, "$section.$subsection");
     }
-
-
+}
+
+
+#sub adjust_text_before_saving()
+sub adjust_links()
+{
+    my $self = shift(@_);
+    my ($doc_obj, $textref) = @_;
+
+    ## change links to be Fedora cognant:
+    # 1. retrieve txt  $$textref ???
+    # 2. change it:
+    # /$ENV{'FEDORA_PREFIX'}/objects/$greenstone-docobj-hash-xxx/datastreams/FG<orig-img-name>/content
+    # (Note that the first image is always "url", instead of FG<orig-img-name>)
+    # 3. only replace it back in doc_obj if we didn't get a ref in the first place
+
+    my $OID = $doc_obj->get_OID();
+    my $fnamespace = $self->{'fedora_namespace'};
+    my $replace_img_name = 1;
+    if($OID ne "collection" && defined $fnamespace) {
+    my $fed_id = "$fnamespace:".$ENV{'GSDLCOLLECTION'}."-$OID"; #oid_namespace:collection-OID
+    my $fedora_url_prefix = $ENV{'FEDORA_PREFIX'}."/objects/$fed_id/datastreams/";
+    my $fedora_url_suffix = "/content";
+
+    $$textref =~ s/(<(?:img|embed|table|tr|td|link|script)[^>]*?(?:src|background|href)\s*=\s*)((?:[\"][^\"]+[\"])|(?:[\'][^\']+[\'])|(?:[^\s\/>]+))([^>]*>)/$self->replace_rel_link($1, $2, $3, $fedora_url_prefix, $fedora_url_suffix, $replace_img_name--)/isge;
+#   print STDERR "*** all text after: $$textref\n\n";
+    }
+
+    return $$textref;
+}
+
+# replace relative link with the prefix and suffix given
+sub replace_rel_link
+{
+    my $self = shift (@_);
+    my ($front, $link, $back, $url_prefix, $url_suffix, $replace_img_name) = @_;
+
+    # only process relative links. Return if absolute link
+    if($link =~ m/^http/) {
+    return "$front$link$back";
+    }
+
+    # remove quotes from link at start and end if necessary
+    if ($link=~/^[\"\']/) {
+    $link=~s/^[\"\']//;
+    $link=~s/[\"\']$//;
+    $front.='"';
+    $back="\"$back";
+    }
+
+    # remove any _httpdocimg/ that greenstone may have prefixed to the image
+    $link =~ s/^_httpdocimg_(?:\/|\\)//;
+
+    # the datastream of the first image of each section is always called "url"
+    if($replace_img_name) {
+    return "$front$url_prefix"."url"."$url_suffix$back";
+    }
+    # else
+    return "$front$url_prefix"."FG"."$link$url_suffix$back";
 }