class RenderWorker(QThread):
def __init__(self, parent, lrf_stream, logger, opts):
QThread.__init__(self, parent)
self.stream, self.logger, self.opts = lrf_stream, logger, opts
self.aborted = False
self.lrf = None
self.document = None
self.exception = None
def run(self):
try:
self.lrf = LRFDocument(self.stream)
self.lrf.parse()
self.stream.close()
self.stream = None
if self.aborted:
self.lrf = None
except Exception as err:
self.lrf, self.stream = None, None
self.exception = err
self.formatted_traceback = traceback.format_exc()
def abort(self):
if self.lrf is not None:
self.aborted = True
self.lrf.keep_parsing = False
def run(self):
try:
self.lrf = LRFDocument(self.stream)
self.lrf.parse()
self.stream.close()
self.stream = None
if self.aborted:
self.lrf = None
except Exception as err:
self.lrf, self.stream = None, None
self.exception = err
self.formatted_traceback = traceback.format_exc()
def process_file(lrfpath, opts, logger=None):
if logger is None:
level = logging.DEBUG if opts.verbose else logging.INFO
logger = logging.getLogger('lrf2html')
setup_cli_handlers(logger, level)
if opts.out is None:
opts.out = os.getcwdu()
else:
opts.out = os.path.abspath(opts.out)
if not os.path.isdir(opts.out):
raise ConversionError(opts.out + ' is not a directory')
if not os.path.exists(opts.out):
os.makedirs(opts.out)
document = LRFDocument(open(lrfpath, 'rb'))
LRFConverter(document, opts, logger)
def convert(self, stream, options, file_ext, log, accelerators):
from lxml import etree
from calibre.ebooks.lrf.input import (MediaType, Styles, TextBlock,
Canvas, ImageBlock, RuledLine)
self.log = log
self.log('Generating XML')
from calibre.ebooks.lrf.lrfparser import LRFDocument
d = LRFDocument(stream)
d.parse()
xml = d.to_xml(write_files=True)
if options.verbose > 2:
open(u'lrs.xml', 'wb').write(xml.encode('utf-8'))
parser = etree.XMLParser(no_network=True, huge_tree=True)
try:
doc = etree.fromstring(xml, parser=parser)
except:
self.log.warn('Failed to parse XML. Trying to recover')
parser = etree.XMLParser(no_network=True,
huge_tree=True,
recover=True)
doc = etree.fromstring(xml, parser=parser)
char_button_map = {}
for x in doc.xpath('//CharButton[@refobj]'):
ro = x.get('refobj')
jump_button = doc.xpath('//*[@objid="%s"]' % ro)
if jump_button:
jump_to = jump_button[0].xpath(
'descendant::JumpTo[@refpage and @refobj]')
if jump_to:
char_button_map[ro] = '%s.xhtml#%s' % (
jump_to[0].get('refpage'), jump_to[0].get('refobj'))
plot_map = {}
for x in doc.xpath('//Plot[@refobj]'):
ro = x.get('refobj')
image = doc.xpath('//Image[@objid="%s" and @refstream]' % ro)
if image:
imgstr = doc.xpath('//ImageStream[@objid="%s" and @file]' %
image[0].get('refstream'))
if imgstr:
plot_map[ro] = imgstr[0].get('file')
self.log('Converting XML to HTML...')
styledoc = etree.fromstring(P('templates/lrf.xsl', data=True))
media_type = MediaType()
styles = Styles()
text_block = TextBlock(styles, char_button_map, plot_map, log)
canvas = Canvas(doc, styles, text_block, log)
image_block = ImageBlock(canvas)
ruled_line = RuledLine()
extensions = {
('calibre', 'media-type'): media_type,
('calibre', 'text-block'): text_block,
('calibre', 'ruled-line'): ruled_line,
('calibre', 'styles'): styles,
('calibre', 'canvas'): canvas,
('calibre', 'image-block'): image_block,
}
transform = etree.XSLT(styledoc, extensions=extensions)
try:
result = transform(doc)
except RuntimeError:
sys.setrecursionlimit(5000)
result = transform(doc)
with open('content.opf', 'wb') as f:
f.write(result)
styles.write()
return os.path.abspath(u'content.opf')
def convert(self, stream, options, file_ext, log,
accelerators):
from lxml import etree
from calibre.ebooks.lrf.input import (MediaType, Styles, TextBlock,
Canvas, ImageBlock, RuledLine)
self.log = log
self.log('Generating XML')
from calibre.ebooks.lrf.lrfparser import LRFDocument
d = LRFDocument(stream)
d.parse()
xml = d.to_xml(write_files=True)
if options.verbose > 2:
open(u'lrs.xml', 'wb').write(xml.encode('utf-8'))
parser = etree.XMLParser(no_network=True, huge_tree=True)
try:
doc = etree.fromstring(xml, parser=parser)
except:
self.log.warn('Failed to parse XML. Trying to recover')
parser = etree.XMLParser(no_network=True, huge_tree=True,
recover=True)
doc = etree.fromstring(xml, parser=parser)
char_button_map = {}
for x in doc.xpath('//CharButton[@refobj]'):
ro = x.get('refobj')
jump_button = doc.xpath('//*[@objid="%s"]'%ro)
if jump_button:
jump_to = jump_button[0].xpath('descendant::JumpTo[@refpage and @refobj]')
if jump_to:
char_button_map[ro] = '%s.xhtml#%s'%(jump_to[0].get('refpage'),
jump_to[0].get('refobj'))
plot_map = {}
for x in doc.xpath('//Plot[@refobj]'):
ro = x.get('refobj')
image = doc.xpath('//Image[@objid="%s" and @refstream]'%ro)
if image:
imgstr = doc.xpath('//ImageStream[@objid="%s" and @file]'%
image[0].get('refstream'))
if imgstr:
plot_map[ro] = imgstr[0].get('file')
self.log('Converting XML to HTML...')
styledoc = etree.fromstring(P('templates/lrf.xsl', data=True))
media_type = MediaType()
styles = Styles()
text_block = TextBlock(styles, char_button_map, plot_map, log)
canvas = Canvas(doc, styles, text_block, log)
image_block = ImageBlock(canvas)
ruled_line = RuledLine()
extensions = {
('calibre', 'media-type') : media_type,
('calibre', 'text-block') : text_block,
('calibre', 'ruled-line') : ruled_line,
('calibre', 'styles') : styles,
('calibre', 'canvas') : canvas,
('calibre', 'image-block'): image_block,
}
transform = etree.XSLT(styledoc, extensions=extensions)
try:
result = transform(doc)
except RuntimeError:
sys.setrecursionlimit(5000)
result = transform(doc)
with open('content.opf', 'wb') as f:
f.write(result)
styles.write()
return os.path.abspath('content.opf')
