Changeset 28177 for main/trunk/greenstone2/perllib/util.pm

Timestamp:

2013-08-28T21:02:52+12:00 (11 years ago)

Author:

ak19

Message:

Added new subroutines to convert from absolute paths to Greenstone locations into paths containing a placeholder for the longest Greenstone prefix matched (path to collection folder, collect folder, site folder or else to GSDLHOME/GSDL3HOME folder). The reverse function has also been added. These haven't been used yet where they need to be, but have been tested to work when called from a testing script.

File:

• main/trunk/greenstone2/perllib/util.pm (modified) (2 diffs)

main/trunk/greenstone2/perllib/util.pm

@@ -1231 +1231 @@
     
     elsif (defined $site && $site) { # site non-empty, so get default collect dir for GS3
-    if (defined $ENV{'GSDL3HOME'}) {
-        return &FileUtils::filenameConcatenate($ENV{'GSDL3HOME'},"sites",$site,"collect");
-    } else {
-        return &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'},"web","sites",$site,"collect");
-    }
+    return &FileUtils::filenameConcatenate(&util::get_site_dir($site),"collect");
     }
     
@@ -1241 +1237 @@
     return &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"collect");
     }
+}
+
+# returns path to current GS3 site.
+# Returns "" if $site parameter is empty string, undefined or otherwise false
+sub get_site_dir {
+    my ($site) = @_;
+    
+    if (defined $ENV{'GSDL3HOME'}) {
+    return &FileUtils::filenameConcatenate($ENV{'GSDL3HOME'},"sites",$site); # web folder
+    } 
+    elsif (defined $ENV{'GSDL3SRCHOME'}) {
+    return &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'},"web","sites",$site);
+    }
+    else { # site has no meaning, return "" (covers the case of this subroutine having been called for a GS2 situation)
+    return "";
+    }
+}
+
+# returns path to GSDL3HOME if GS3, or GSDLHOME if GS2
+sub get_gs_home {
+    my ($site) = @_;
+
+    if (defined $site && $site) { # not undefined or empty string, so GS3 case
+    if (defined $ENV{'GSDL3HOME'}) {
+        return $ENV{'GSDL3HOME'};
+    } else {
+        return &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'},"web");
+    }
+    } else {
+    return $ENV{'GSDLHOME'};
+    }
+}
+
+sub is_abs_path {
+    my ($path) = @_;
+
+    return 1 if($path =~ m@^/@); # full paths begin with forward slash on linux/mac
+    return 1 if($path =~ m@^([a-zA-Z]\:|\\)@); # full paths begin with drive letter colon for Win or \ for volume, http://stackoverflow.com/questions/13011013/get-only-volume-name-from-filepath
+
+    return 0;
+
+    # File::Spec loads the module appropriate for the OS, but abs paths in archiveinf-doc.xml may be generated on another OS to this one
+    #return File::Spec->file_name_is_absolute( $path );    
+}
+
+# returns an array of the FULL PATHS to the specific collection given, to the general collect_dir,
+# to the site directory if gs3 and to the greenstone home folder (GSDLHOME or GSDL3HOME). 
+# These elements are returned in this ORDER, in order to allow matching from longest to shortest path.
+sub get_common_gs_paths {
+
+    my ($site, $qualified_collection_name) = @_;
+    
+    my $collect_dir = &util::get_working_collect_dir($site);
+    my $collection_path = "";
+    if($qualified_collection_name ne "") {
+    $collection_path = &util::resolve_collection_dir($collect_dir, $qualified_collection_name, $site);
+    }
+    
+    my $site_dir = &util::get_site_dir($site); # returns "" for GS2
+    my $gs_home = &util::get_gs_home(); # GSDLHOME or GSDL3HOME folder
+    
+    return ($collection_path, $collect_dir, $site_dir, $gs_home); # list in this order: from longest to shortest path
+    #my @gs_paths = ($collection_path, $collect_dir, $site_dir, $gs_home); # list in this order: from longest to shortest path
+    #return @gs_paths;
+}
+
+# This subroutine is for improving portability of Greenstone collections from one OS to another,
+# to be used to convert absolute paths going into db files into paths with placeholders instead.
+# This sub works with util::get_common_gs_paths and takes a path to a greenstone file and, if it's
+# an absolute path, then it will replace the longest matching greenstone-path prefix of the given 
+# path with a placeholder to match.
+# The Greenstone-path prefixes that can be matched are the following common Greenstone paths: 
+# the path to the current (specific) collection, the path to the general GS collect directory, 
+# the path to the site directory if GS3, else the path to the GSDLHOME/GSDL3HOME folder.
+# The longest matching prefix will be replaced with the equivalent placeholder: 
+# @THISCOLLECTPATH@, else @COLLECTHOME@, else @SITEHOME@, else @GSDLHOME@.
+sub abspath_to_placeholders {
+    my $path = shift(@_); # path to convert from absolute to one with placeholders
+
+    my ($collection_path, $collect_dir, $site_dir, $gs_home) = @_;
+
+    return $path unless is_abs_path($path); # path is relative
+    
+    # now we know we're dealing with absolute paths and have to replace gs prefixes with placeholders
+
+    my @gs_paths = ($collection_path, $collect_dir, $site_dir, $gs_home); # list in this order: from longest to shortest path
+
+    my %placeholder_map = ($gs_home => '@GSDLHOME@', # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
+               $site_dir => '@SITEHOME@',
+               $collect_dir => '@COLLECTHOME@',
+               $collection_path => '@THISCOLLECTPATH@'
+    );
+
+
+    # The sequence of elements in @gs_paths matters    
+    # Need to loop starting from the *longest* matching path (the path to the specific collection) 
+    # to the shortest matching path (the path to gsdlhome/gsdl3home folder):
+
+    foreach my $gs_path (@gs_paths) { 
+    my $re_path =  &util::filename_to_regex($gs_path); # escape for regex
+
+    next if($re_path eq ""); # site can be empty
+
+    if($path =~ m/^$re_path/) { # case sensitive or not for OS?
+
+        my $placeholder = $placeholder_map{$gs_path}; # get the placeholder to replace the matched path with
+
+        $path =~ s/^$re_path/$placeholder/; #case sensitive or not?
+        #$path =~ s/^[\\\/]//; # remove gs_path's trailing separator left behind at the start of the path
+        last; # done
+    }
+    }
+    
+    return $path;
+}
+
+# Function that does the reverse of the util::abspath_to_placeholders subroutine
+# Once again, call this with the values returned from util::get_common_gs_paths
+sub placeholders_to_abspath {
+    my $path = shift(@_); # path that can contain placeholders to convert to resolved absolute path
+
+    my ($collection_path, $collect_dir, $site_dir, $gs_home) = @_;
+
+    return $path if($path !~ m/@/); # path contains no placeholders
+    
+    # replace placeholders with gs prefixes 
+
+    my @placeholders = ('@THISCOLLECTPATH@', '@COLLECTHOME@', '@SITEHOME@', '@GSDLHOME@'); # order of paths not crucial in this case, 
+                       # but listed here from longest to shortest once placeholders are have been resolved
+
+    my %placeholder_to_gspath_map = ('@GSDLHOME@' => $gs_home, # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
+               '@SITEHOME@' => $site_dir,
+               '@COLLECTHOME@' => $collect_dir,
+               '@THISCOLLECTPATH@' => $collection_path
+    );
+
+
+    foreach my $placeholder (@placeholders) { 
+    my $gs_path = $placeholder_to_gspath_map{$placeholder};
+
+    next if($gs_path eq ""); # no value for sitehome for GS2
+
+    #my $sep = ($ENV{'GSDLOS'} eq "windows") ? "\\" : "/";
+    #$gs_path = $gs_path . "$sep" if ($gs_path !~ m/$sep$/); # add in any trailing separator to the path prefix
+
+    if($path =~ m/^$placeholder/) {
+        $path =~ s/^$placeholder/$gs_path/;
+        last; # done
+    }
+    }
+    
+    return $path;
 }