require 'nokogiri'
require 'zip'
-require './author'
-require './cover'
+require 'author'
+require 'cover'
class Book
+ @@DC_NS_URL = 'http://purl.org/dc/elements/1.1/'
+
def initialize(fileName)
+ #puts 'InitBook(' + fileName + ')'
@author = nil
@cover = nil
+ @description = nil
@path = fileName
@series = nil
@title = nil
return false
end
+ #puts "Filename: " + fileName.to_s
lowerName = fileName.downcase()
if lowerName.end_with?(".epub")
return false
end
+ def author
+ return @author
+ end
+
+ def cover
+ return @cover
+ end
+
+ def description
+ @description
+ end
+
+ def heading
+ result = []
+
+ if nil != @title
+ result.push('<b>' + @title + '</b>')
+ else
+ result.push('<i>(Unknown title)</i>')
+ end
+ if nil != @author
+ result.push('<i>by ' + @author.reading_order + '</i>')
+ end
+
+ seriesInfo = []
+ if nil != @series
+ seriesInfo.push(@series.to_s)
+ end
+ if nil != @volume
+ seriesInfo.push(@volume.to_s)
+ end
+ if seriesInfo.length > 0
+ result.push(seriesInfo.join(' '))
+ end
+
+ return result.join('<br/>')
+ end
+
def inspect
data = []
if nil != @author
- data.push('author="' + @author.to_s + '"')
+ data.push('author="' + @author.inspect + '"')
end
if nil != @series
data.push('series="' + @series + '"')
return '(Book:' + data.join(',') + ')'
end
+ def path
+ @path
+ end
+
+ def series
+ @series
+ end
+
def to_s
return inspect()
end
+ def title
+ @title
+ end
+
+ def volume
+ @volume
+ end
+
protected
def isUpper?(c)
return /[[:upper:]]/.match(c)
end
protected
- def massageAuthor(input)
+ def massage_author(input)
if nil == input
return nil
end
- result = ""
+ reading_order = ""
input.each_char do |c|
- if isUpper?(c) and (result.length > 0)
- result += " "
- end
- result += c
+ if isUpper?(c) and (reading_order.length > 0)
+ reading_order += " "
+ end
+ reading_order += c
end
-
- return result
+
+ return reading_order
end
# Returns (series, volumeNo, titleText)
parts = fileName.split('/')
(@series, @volume, @title) = processTitle(parts[-1])
if parts.length > 1
- @author = massageAuthor(parts[-2])
+ grouping = parts[-2]
+ reading_order = massage_author(grouping)
+ sort_order = nil
+ @author = Author.new(grouping, reading_order, sort_order)
end
if fileName.downcase.end_with?(".epub")
protected
def scanEpub!(fileName)
- Zip::File.open(fileName) do |zipfile|
- contXml = zipfile.read('META-INF/container.xml')
- contDoc = Nokogiri::XML(contXml)
- opfPath = contDoc.css("container rootfiles rootfile")[0]['full-path']
+ #puts 'Scanning "' + fileName.to_s + '"...'
+ begin
+ Zip::File.open(fileName) do |zipfile|
+ entry = zipfile.find_entry('META-INF/container.xml')
+ if nil == entry
+ puts 'No META-INF/container.xml, skipping book ' + fileName
+ return
+ end
+ contXml = zipfile.read('META-INF/container.xml')
+ contDoc = Nokogiri::XML(contXml)
+ opfPath = contDoc.css("container rootfiles rootfile")[0]['full-path']
- scanOpf!(zipfile, opfPath)
+ scanOpf!(zipfile, opfPath)
+ end
+ rescue Zip::Error => exc
+ puts 'ERROR processing file "' + fileName + '":'
+ puts exc.message
+ puts exc.backtrace
end
end
#-------
# Author
- creator = opfDoc.css('dc|creator', 'dc' => 'http://purl.org/dc/elements/1.1/')
- if nil != creator
- roleNode = creator.attr('role')
- if nil != roleNode
- role = roleNode.value
+ grouping = @author.grouping
+ reading_order = @author.reading_order
+ sort_order = @author.sort_order
+
+ creators = opfDoc.css('dc|creator', 'dc' => @@DC_NS_URL)
+ if (creators.length > 0)
+ creator = creators[0]
+ if nil != creator
+ role = creator['opf:role']
if 'aut' == role
- name = creator.children[0].content
- parts = name.split(' ')
- if parts.length > 1
- surname = parts[-1]
- givenNames = parts[0..-2].join(' ')
- @author = Author.new(surname, givenNames)
- else
- @author = Author.new(name, '')
+ reading_order = creator.content
+
+ file_as = creator['opf:file-as']
+ if nil != file_as
+ sort_order = file_as
end
end
+
+ @author = Author.new(grouping, reading_order, sort_order)
+ end
+ end
+
+ #---------------------------------------
+ # Title
+
+ titles = opfDoc.css('dc|title', 'dc' => @@DC_NS_URL)
+ if titles.length > 0
+ title = titles[0]
+ if nil != title
+ @title = title.content
+ end
+ end
+
+ #---------------------------------------
+ # Description
+
+ descrNodes = opfDoc.css('dc|description', 'dc' => @@DC_NS_URL)
+ if (descrNodes.length > 0)
+ descrNode = descrNodes[0]
+ if nil != descrNode
+ @description = descrNode.content
end
end
@volume = content
elsif 'cover' == name
coverId = content
+ #puts 'File ' + @path + ' coverId ' + coverId
end
end
#---------------
# Load the cover
+ @cover = loadCover(zipfile, opfPath, opfDoc, coverId)
+ end
+
+ protected
+ def loadCover(zipfile, opfPath, opfDoc, coverId)
coverFile = nil
- if nil != coverId
- items = opfDoc.css('package manifest item')
- for i in items
- href = i['href']
- id = i['id']
- mimeType = i['media-type']
-
- if coverId == id
- entry = zipfile.find_entry(href)
+ if nil == coverId
+ coverId = "cover-image"
+ end
+
+ items = opfDoc.css('package manifest item')
+ for i in items
+ href = i['href']
+ id = i['id']
+ mimeType = i['media-type']
+
+ if coverId == id
+ entry = zipfile.find_entry(href)
+
+ if nil == entry
+ # Although the epub standard requires the path to be relative
+ # to the base of the epub (zip), some books encountered in the
+ # wild have been found to use a bath relative to the location
+ # of the opf file.
+ parts = opfPath.split('/')
+ opfBasePath = opfPath.split('/')[0..-2].join('/')
+ coverPath = opfBasePath + '/' + href
+ entry = zipfile.find_entry(coverPath)
+ end
+
+ if nil == entry
+ puts 'WARNING! Cover image "' + href + '" not found in file "' + @path + '".'
+ return nil
+ else
entry.get_input_stream() do |is|
- @cover = Cover.new(is, href, mimeType)
+ return Cover.new(is, href, mimeType)
end
end
end
end
+ return nil
end
end