def convert(self,
doc,
encoding,
mimetype,
logError=False,
raiseException=False):
"""Convert WinWord document to raw text"""
tmp_name = self.saveFile(doc)
err = TmpFile('')
if sys.platform == 'win32':
result = (self.execute(
'%s -c utf-8 --nographics -x "%s" "%s" 2> "%s"' %
(self.depends_on, wvConf_file, tmp_name, str(err))), 'utf-8')
else:
result = (self.execute(
'%s -c utf-8 --nographics -x "%s" "%s" 2> "%s"' %
(self.depends_on, wvConf_file, tmp_name, str(err))), 'utf-8')
try:
errors = open(str(err), 'r+').read()
except OSError:
errors = ""
if errors:
if logError:
LOG.warn('Converter %s experienced an error %s' %
(self.content_description, errors))
if raiseException:
raise ConversionError(errors)
return result
def convert(self, doc, encoding, mimetype,
logError=False, raiseException=False):
"""Convert WinWord document to raw text"""
tmp_name = self.saveFile(doc)
err = TmpFile('')
if sys.platform == 'win32':
result = (self.execute(
'%s -c utf-8 --nographics -x "%s" "%s" 2> "%s"' % (
self.depends_on, wvConf_file, tmp_name, str(err))), 'utf-8')
else:
result = (self.execute(
'%s -c utf-8 --nographics -x "%s" "%s" 2> "%s"' % (
self.depends_on, wvConf_file, tmp_name, str(err))), 'utf-8')
try:
errors = open(str(err), 'r+').read()
except OSError:
errors = ""
if errors:
if logError:
LOG.warn('Converter %s experienced an error %s' % (
self.content_description, errors)
)
if raiseException:
raise ConversionError(errors)
return result
def convert(self, doc, encoding, mimetype,
logError=False, raiseException=False):
"""Convert PowerPoint document to raw text"""
tmp_name = self.saveFile(doc)
err = TmpFile('')
if sys.platform == 'win32':
html = self.execute('ppthtml "%s" 2> "%s"' % (
tmp_name, str(err)))
else:
html = self.execute('ppthtml "%s" 2> "%s"' % (
tmp_name, str(err)))
try:
errors = open(str(err), 'r+').read()
except OSError:
errors = ""
if errors:
if logError:
LOG.warn('Converter %s experienced an error %s' % (
self.content_description, errors)
)
if raiseException:
raise ConversionError(errors)
return html2text(html,
ignore_tags=('img',),
indent_width=4,
page_width=80), 'iso-8859-15'
def convert(self, doc, encoding, mimetype,
logError=False, raiseException=False):
"""Convert Excel document to raw text"""
tmp_name = self.saveFile(doc)
err = TmpFile('')
if sys.platform == 'win32':
result = (self.execute('xls2csv /d 8859-1 /q 0 "%s" 2> %s' % (
tmp_name, str(err))), 'iso-8859-15')
else:
result = (self.execute('xls2csv -d 8859-1 -q 0 "%s" 2> %s' % (
tmp_name, str(err))), 'iso-8859-15')
try:
errors = open(str(err), 'r+').read()
except OSError:
errors = ""
if errors:
if logError:
LOG.warn('Converter %s experienced an error %s' % (
self.content_description, errors)
)
if raiseException:
raise ConversionError(errors)
return result
def convert(self,
doc,
encoding,
mimetype,
logError=False,
raiseException=False):
"""Convert Excel document to raw text"""
tmp_name = self.saveFile(doc)
err = TmpFile('')
if sys.platform == 'win32':
result = (self.execute('xls2csv /d 8859-1 /q 0 "%s" 2> %s' %
(tmp_name, str(err))), 'iso-8859-15')
else:
result = (self.execute('xls2csv -d 8859-1 -q 0 "%s" 2> %s' %
(tmp_name, str(err))), 'iso-8859-15')
try:
errors = open(str(err), 'r+').read()
except OSError:
errors = ""
if errors:
if logError:
LOG.warn('Converter %s experienced an error %s' %
(self.content_description, errors))
if raiseException:
raise ConversionError(errors)
return result
def _c(self, text):
if isinstance(text, unicode):
return text
try:
return unicode(text, self.encoding)
except UnicodeDecodeError:
LOG.warn('Content from %s could not be converted to unicode using the site encoding %s' %
(self.context.absolute_url(1), self.encoding))
raise
def _c(self, text):
if isinstance(text, str):
return text
try:
return str(text, self.encoding)
except UnicodeDecodeError:
LOG.warn(
'Content from %s could not be converted to unicode using the site encoding %s'
% (self.context.absolute_url(1), self.encoding))
raise
