- Added document types to guess_doc

nigel.stanger / Digital_Repository

Browse code - Added document types to guess_doc_type. master
1 parent 179ed63 commit 5e71f1286be41133d06b877599dce4423c45c7c2 nstanger authored on 17 Jul 2008

Patch

Showing 1 changed file

Ignore Space Show notes View Repositories/otago_eprints3/otago_eprints/cfg/cfg.d/document_upload.pl
###################################################################### # # Document file upload information # ###################################################################### # AT LEAST one of the following formats will be required. If you do # not want this requirement, then make the list empty (Although this # means users will be able to submit eprints with ZERO documents. # # Available formats are configured elsewhere. See the docs. $c->{required_formats} = [ # "text/html", # "application/pdf", # "application/postscript", # "text/plain", ]; # if you want to make this depend on the values in the eprint then # you can make it a function pointer instead. The function should # return a list as above. # This example requires all normal formats for all eprints except # for those of type book where a document is optional. # # $c->{required_formats} = sub { # my( $eprint ) = @_; # # if( $eprint->get_value( 'type' ) eq "book" ) # { # return []; # } # return [ # 'text/html', # 'application/pdf', # 'application/postscript', # 'text/plain', # ]; # }; # This sets the minimum amount of free space allowed on a disk before EPrints # starts using the next available disk to store EPrints. Specified in kilobytes. $c->{diskspace_error_threshold} = 641024; # If ever the amount of free space drops below this threshold, the # repository administrator is sent a warning email. In kilobytes. $c->{diskspace_warn_threshold} = 5121024; # make a very loose stab at the file format # By default this just looks at the filename suffix, but there's no reason # It can't be much more clever. # It must return a legal document format id. $c->{guess_doc_type} = sub { my( $session, $filename ) = @_; my @formats = $session->get_repository->get_types( "document" ); if( $filename=~m/\.([^.]+)$/ ) { my $suffix = "\L$1"; return "text/html" if $suffix eq "htm"; return "text/html" if $suffix eq "html"; return "application/pdf" if $suffix eq "pdf"; return "application/postscript" if $suffix eq "ps"; return "text/plain" if $suffix eq "txt"; return "application/vnd.ms-powerpoint" if $suffix eq "ppt"; return "application/msword" if $suffix eq "doc"; return "image/jpeg" if $suffix eq "jpg"; return "image/jpeg" if $suffix eq "jpeg"; return "image/png" if $suffix eq "png"; return "image/gif" if $suffix eq "gif"; return "image/bmp" if $suffix eq "bmp"; return "image/tiff" if $suffix eq "tiff"; return "image/tiff" if $suffix eq "tif"; return "video/mpeg" if $suffix eq "mpg"; return "video/mpeg" if $suffix eq "mpeg"; return "video/quicktime" if $suffix eq "mov"; return "video/x-msvideo" if $suffix eq "avi"; ## NJS 2008-07-17: Added various missing document types. # MS Excel is inexplicably missing above. return "application/application/vnd.ms-excel" if $suffix eq "xls"; # May as well map CSV to Excel too. return "application/application/vnd.ms-excel" if $suffix eq "csv"; # Office XML document types. return "application/application/vnd.ms-excel" if $suffix eq "xlsx"; return "application/vnd.ms-powerpoint" if $suffix eq "pptx"; return "application/msword" if $suffix eq "docx"; # LaTeX return "application/x-latex" if $suffix eq "tex"; # Audio return "audio/mpeg" if $suffix eq "mp3"; return "audio/x-wav" if $suffix eq "wav"; # Compressed archives return "application/zip" if $suffix eq "zip"; return "application/x-gzip" if $suffix eq "gz"; return "application/x-gzip" if $suffix eq "tgz"; ## NJS end } return "other"; }; # This subroutine is called every time that the files in a document are # modified (added or removed). # You can place hooks for things like automatic data extraction here. $c->{on_files_modified} = sub { my( $session, $document ) = @_; # do your stuff }; ###################################################################### # # Document file upload information # ###################################################################### # AT LEAST one of the following formats will be required. If you do # not want this requirement, then make the list empty (Although this # means users will be able to submit eprints with ZERO documents. # # Available formats are configured elsewhere. See the docs. $c->{required_formats} = [ # "text/html", # "application/pdf", # "application/postscript", # "text/plain", ]; # if you want to make this depend on the values in the eprint then # you can make it a function pointer instead. The function should # return a list as above. # This example requires all normal formats for all eprints except # for those of type book where a document is optional. # # $c->{required_formats} = sub { # my( $eprint ) = @_; # # if( $eprint->get_value( 'type' ) eq "book" ) # { # return []; # } # return [ # 'text/html', # 'application/pdf', # 'application/postscript', # 'text/plain', # ]; # }; # This sets the minimum amount of free space allowed on a disk before EPrints # starts using the next available disk to store EPrints. Specified in kilobytes. $c->{diskspace_error_threshold} = 641024; # If ever the amount of free space drops below this threshold, the # repository administrator is sent a warning email. In kilobytes. $c->{diskspace_warn_threshold} = 5121024; # make a very loose stab at the file format # By default this just looks at the filename suffix, but there's no reason # It can't be much more clever. # It must return a legal document format id. $c->{guess_doc_type} = sub { my( $session, $filename ) = @_; my @formats = $session->get_repository->get_types( "document" ); if( $filename=~m/\.([^.]+)$/ ) { my $suffix = "\L$1"; return "text/html" if $suffix eq "htm"; return "text/html" if $suffix eq "html"; return "application/pdf" if $suffix eq "pdf"; return "application/postscript" if $suffix eq "ps"; return "text/plain" if $suffix eq "txt"; return "application/vnd.ms-powerpoint" if $suffix eq "ppt"; return "application/msword" if $suffix eq "doc"; return "image/jpeg" if $suffix eq "jpg"; return "image/jpeg" if $suffix eq "jpeg"; return "image/png" if $suffix eq "png"; return "image/gif" if $suffix eq "gif"; return "image/bmp" if $suffix eq "bmp"; return "image/tiff" if $suffix eq "tiff"; return "image/tiff" if $suffix eq "tif"; return "video/mpeg" if $suffix eq "mpg"; return "video/mpeg" if $suffix eq "mpeg"; return "video/quicktime" if $suffix eq "mov"; return "video/x-msvideo" if $suffix eq "avi"; } return "other"; }; # This subroutine is called every time that the files in a document are # modified (added or removed). # You can place hooks for things like automatic data extraction here. $c->{on_files_modified} = sub { my( $session, $document ) = @_; # do your stuff };

Ignore Space Show notes View

Repositories/otago_eprints3/otago_eprints/cfg/cfg.d/document_upload.pl

######################################################################
#
#  Document file upload information
#
######################################################################

# AT LEAST one of the following formats will be required. If you do
# not want this requirement, then make the list empty (Although this
# means users will be able to submit eprints with ZERO documents.
#
# Available formats are configured elsewhere. See the docs.

$c->{required_formats} = 
[
#	"text/html",
# 	"application/pdf",
#	"application/postscript",
#	"text/plain",
];

# if you want to make this depend on the values in the eprint then
# you can make it a function pointer instead. The function should
# return a list as above.

# This example requires all normal formats for all eprints except
# for those of type book where a document is optional.
#
# $c->{required_formats} = sub {
# 	my( $eprint ) = @_;
# 
# 	if( $eprint->get_value( 'type' ) eq "book" )
# 	{
# 		return [];
# 	}
# 	return [
#		'text/html',
#		'application/pdf',
#		'application/postscript',
#		'text/plain',
#	];
# };

# This sets the minimum amount of free space allowed on a disk before EPrints
# starts using the next available disk to store EPrints. Specified in kilobytes.
$c->{diskspace_error_threshold} = 64*1024;

# If ever the amount of free space drops below this threshold, the
# repository administrator is sent a warning email. In kilobytes.
$c->{diskspace_warn_threshold} = 512*1024;

# make a very loose stab at the file format
# By default this just looks at the filename suffix, but there's no reason
# It can't be much more clever.
# It must return a legal document format id.
$c->{guess_doc_type} = sub
{
	my( $session, $filename ) = @_;

my @formats = $session->get_repository->get_types( "document" );

if( $filename=~m/\.([^.]+)$/ )
	{
		my $suffix = "\L$1";

return "text/html" if $suffix eq "htm";
		return "text/html" if $suffix eq "html";
		return "application/pdf" if $suffix eq "pdf";
		return "application/postscript" if $suffix eq "ps";
		return "text/plain" if $suffix eq "txt";
		return "application/vnd.ms-powerpoint" if $suffix eq "ppt";
		return "application/msword" if $suffix eq "doc";
		return "image/jpeg" if $suffix eq "jpg";
		return "image/jpeg" if $suffix eq "jpeg";
		return "image/png" if $suffix eq "png";
		return "image/gif" if $suffix eq "gif";
		return "image/bmp" if $suffix eq "bmp";
		return "image/tiff" if $suffix eq "tiff";
		return "image/tiff" if $suffix eq "tif";
		return "video/mpeg" if $suffix eq "mpg";
		return "video/mpeg" if $suffix eq "mpeg";
		return "video/quicktime" if $suffix eq "mov";
		return "video/x-msvideo" if $suffix eq "avi";
		
		## NJS 2008-07-17: Added various missing document types.
		
		# MS Excel is inexplicably missing above.
		return "application/application/vnd.ms-excel" if $suffix eq "xls";
		
		# May as well map CSV to Excel too.
		return "application/application/vnd.ms-excel" if $suffix eq "csv";
		
		# Office XML document types.
		return "application/application/vnd.ms-excel" if $suffix eq "xlsx";
		return "application/vnd.ms-powerpoint" if $suffix eq "pptx";
		return "application/msword" if $suffix eq "docx";
		
		# LaTeX
		return "application/x-latex" if $suffix eq "tex";
		
		# Audio
		return "audio/mpeg" if $suffix eq "mp3";
		return "audio/x-wav" if $suffix eq "wav";
		
		# Compressed archives
		return "application/zip" if $suffix eq "zip";
		return "application/x-gzip" if $suffix eq "gz";
		return "application/x-gzip" if $suffix eq "tgz";
		
		## NJS end
	}

return "other";
};

# This subroutine is called every time that the files in a document are
# modified (added or removed).
# You can place hooks for things like automatic data extraction here.

$c->{on_files_modified} = sub
{
	my( $session, $document ) = @_;

# do your stuff
};