1 # -*- ruby -*-

     2 #encoding: utf-8

     3 

     4 require 'pdf-reader'

     5 

     6 require 'strelka'

     7 require 'thingfish' unless defined?( Thingfish )

     8 require 'thingfish/processor' unless defined?( Thingfish::Processor )

     9 

    10 

    11 # Attach PDF metadata to the Adobe Portable Document Format.

    12 class Thingfish::Processor::PDF < Thingfish::Processor

    13 	extend Loggability

    14 

    15 	# Package version

    16 	VERSION = '0.1.0'

    17 

    18 	# Version control revision

    19 	REVISION = %q$Revision$

    20 

    21 

    22 	# Loggability API -- log to the :thingfish logger

    23 	log_to :thingfish

    24 

    25 	# The list of handled types

    26 	handled_types 'application/pdf'

    27 

    28 

    29 	### Synchronous processor API -- extract metadata from uploaded PDFs

    30 	###

    31 	def on_request( request )

    32 		reader   = ::PDF::Reader.new( request.body )

    33 		metadata = self.extract_pdf_metadata( reader )

    34 

    35 		request.add_metadata( metadata )

    36 	end

    37 

    38 

    39 	#########

    40 	protected

    41 	#########

    42 

    43 	### Normalize metadata from the PDFReader object and return it as a hash.

    44 	###

    45 	def extract_pdf_metadata( reader )

    46 		self.log.debug "Extracting PDF metadata..."

    47 

    48 		pdf_metadata = {

    49 			'pdf:version'   => reader.pdf_version,

    50 			'pdf:pagecount' => reader.page_count,

    51 		}.reject {|_,v| v.nil? }

    52 

    53 		reader.info.each_pair do |key, val|

    54 			pdf_metadata[ "pdf:#{key}" ] = val unless val.is_a?( ::PDF::Reader::Reference )

    55 		end

    56 

    57 		self.log.debug "  raw PDF metadata: %p" % [ pdf_metadata ]

    58 		return pdf_metadata

    59 	end

    60 

    61 end # class Thingfish::Processor::PDF

    62