Sha256: c4a8f72b5943f084500cb228e92008eb6d758a0f94b799aefd35a4289f9c107e
Contents?: true
Size: 1.04 KB
Versions: 53
Compression:
Stored size: 1.04 KB
Contents
require 'pdf-reader'
require 'ndr_support/safe_file'
require_relative 'registry'
module NdrImport
# This is one of a collection of file handlers that deal with individual formats of data.
# They can be instantiated directly or via the factory method Registry.tables
module File
# This class is a PDF file handler that returns a single table.
class Pdf < Base
private
def rows(&block)
return enum_for(:rows) unless block
reader = PDF::Reader.new(SafeFile.safepath_to_string(@filename))
reader.pages.each do |page|
process_page(page, &block)
end
rescue NoMethodError
raise "Failed to read #{SafeFile.basename(@filename)} as a PDF"
end
def process_page(page, &block)
page.text.split("\n").each do |line|
block.call(line)
end
rescue => e
raise "Invalid format on page #{page.number} of #{SafeFile.basename(@filename)} " \
"[#{e.class}: #{e.message}]"
end
end
Registry.register(Pdf, 'pdf')
end
end
Version data entries
53 entries across 53 versions & 1 rubygems