def invokeCommand(self, tmpdir, fullname):
if os.name=='posix':
cmd = 'cd "%s" && unzip "%s" 2>error_log 1>/dev/null' % (
tmpdir, fullname)
p = Popen(cmd, shell = True)
sts = os.waitpid(p.pid, 0)
cmd = ('cd "%s" && xsltproc --novalid %s content.xml >"%s.html" '
'2>"%s.log-xsltproc"') % (
tmpdir, XSL_STYLESHEET, sansext(fullname), sansext(fullname))
LOG(self.__name__, DEBUG, "cmd = %s" % cmd)
p = Popen(cmd, shell = True)
sts = os.waitpid(p.pid, 0)
try:
htmlfile = open(os.path.join(tmpdir, "%s.html" % sansext(fullname)),
'r')
html = htmlfile.read()
htmlfile.close()
except:
try:
return open(os.path.join(tmpdir, 'error_log'), 'r').read()
except:
return ''
return html
def convert(self):
"""Convert the document to HTML.
"""
if not os.name == 'posix':
return
name = self.name()
cmd = 'cd "%s" && unzip "%s" 2>unzip_error.log 1>/dev/null' % (
self.tmpdir, name)
os.system(cmd)
cmd = 'cd "%s" && %s --novalid "%s" content.xml '
cmd += '> "%s.html" 2> "error.log"'
cmd = cmd % (
self.tmpdir, self.binary, XSL_STYLESHEET, sansext(name))
os.system(cmd)
try:
htmlfile = open(os.path.join(
self.tmpdir, "%s.html" % sansext(name)), 'r')
html = htmlfile.read()
htmlfile.close()
except:
try:
return open(os.path.join(self.tmpdir, 'unzip_error.log'),
'r').read()
except:
return ''
return html
def convert(self, data, cache, **kwargs):
base_name = sansext(kwargs.get("filename") or 'unknown.xml')
dtds = self.config['dtds']
tmpdir, fullname = self.initialize_tmpdir(data, filename=base_name)
try:
try:
doctype = get_doctype(data)
except DTException:
try:
doctype = get_dtd(data)
except DTException:
log('Unable to get doctype nor dtd in %s' % data)
doctype = None
if doctype and doctype in dtds:
data = self.invokeCommand(fullname, dtds[doctype])
elif self.config['default_transform']:
data = self.invokeCommand(fullname,
self.config['default_transform'])
cache.setData(data)
path, images = self.subObjects(tmpdir)
objects = {}
if images:
self.fixImages(path, images, objects)
cache.setSubObjects(objects)
return cache
finally:
self.cleanDir(tmpdir)
def convert(self, data, cache, **kwargs):
base_name = sansext(kwargs.get("filename") or 'unknown.xml')
dtds = self.config['dtds']
tmpdir, fullname = self.initialize_tmpdir(data, filename=base_name)
try:
try:
doctype = get_doctype(data)
except DTException:
try:
doctype = get_dtd(data)
except DTException:
log('Unable to get doctype nor dtd in %s' % data)
doctype = None
if doctype and dtds.has_key(doctype):
data = self.invokeCommand(fullname, dtds[doctype])
elif self.config['default_transform']:
data = self.invokeCommand(fullname, self.config['default_transform'])
cache.setData(data)
path, images = self.subObjects(tmpdir)
objects = {}
if images:
self.fixImages(path, images, objects)
cache.setSubObjects(objects)
return cache
finally:
self.cleanDir(tmpdir)
def invokeCommand(self, tmpdir, fullname):
cmd = 'cd "%s" && unzip %s 2>error_log 1>/dev/null' % (
tmpdir, fullname)
os.system(cmd)
cmd = ('cd "%s" && xsltproc --novalid %s content.xml >"%s.html" '
'2>"%s.log-xsltproc"') % (
tmpdir, XSL_STYLESHEET, sansext(fullname), sansext(fullname))
os.system(cmd)
try:
htmlfile = open(os.path.join(tmpdir, "%s.html" % sansext(fullname)),
'r')
html = htmlfile.read()
htmlfile.close()
except:
try:
return open(os.path.join(tmpdir, 'error_log'), 'r').read()
except:
return ''
return html
def invokeCommand(self, tmpdir, fullname):
# FIXME: windows users...
htmlfile = "%s/%s.html" % (tmpdir, sansext(fullname))
cmd = 'cd "%s" && %s -o %s "%s" 2>error_log 1>/dev/null' % (tmpdir, self.binary, htmlfile, fullname)
os.system(cmd)
try:
html = open(htmlfile).read()
except:
try:
return open("%s/error_log" % tmpdir, "r").read()
except:
return ""
return html
def invokeCommand(self, tmpdir, fullname):
# FIXME: windows users...
htmlfile = "%s/%s.html" % (tmpdir, sansext(fullname))
cmd = 'cd "%s" && %s -o %s "%s" 2>error_log 1>/dev/null' % (
tmpdir, self.binary, htmlfile, fullname)
os.system(cmd)
try:
html = open(htmlfile).read()
except:
try:
return open("%s/error_log" % tmpdir, 'r').read()
except:
return ''
return html
def invokeCommand(self, tmpdir, fullname):
# FIXME: windows users...
textfile = "%s/%s.txt" % (tmpdir, sansext(fullname))
cmd = 'cd "%s" && %s -enc UTF-8 "%s" "%s" 2>error_log 1>/dev/null' % (
tmpdir, self.binary, fullname, textfile)
os.system(cmd)
try:
text = open(textfile).read()
except:
try:
return open("%s/error_log" % tmpdir, 'r').read()
except:
return ''
return text
def invokeCommand(self, tmpdir, fullname):
cmd = 'cd "%s" && unzip %s 2>error_log 1>/dev/null' % (tmpdir,
fullname)
p = Popen(cmd, shell=True)
sts = os.waitpid(p.pid, 0)
cmd = ('cd "%s" && xsltproc --novalid %s content.xml >"%s.html" '
'2>"%s.log-xsltproc"') % (tmpdir, XSL_STYLESHEET,
sansext(fullname), sansext(fullname))
p = Popen(cmd, shell=True)
sts = os.waitpid(p.pid, 0)
try:
htmlfile = open(
os.path.join(tmpdir, "%s.html" % sansext(fullname)), 'r')
html = htmlfile.read()
htmlfile.close()
except:
try:
return open(os.path.join(tmpdir, 'error_log'), 'r').read()
except:
return ''
return html
def invokeCommand(self, tmpdir, fullname):
# FIXME: windows users...
xmlfile = "%s/%s.xml" % (tmpdir, sansext(fullname))
cmd = 'cd "%s" && %s -o %s "%s" 2>error_log 1>/dev/null' % (
tmpdir, self.binary, xmlfile, fullname)
if six.PY2:
os.system(cmd)
else:
subprocess.run(cmd, shell=True)
try:
xml = open(xmlfile).read()
except:
try:
return open("%s/error_log" % tmpdir, 'r').read()
except:
return ''
return xml
def invokeCommand(self, tmpdir, fullname):
# FIXME: windows users...
htmlfile = "%s/%s.html" % (tmpdir, sansext(fullname))
cmd = 'cd "%s" && %s -o %s "%s" 2>error_log 1>/dev/null' % (
tmpdir, self.binary, htmlfile, fullname)
if six.PY2:
os.system(cmd)
else:
subprocess.run(cmd, shell=True)
try:
html = open(htmlfile).read()
except:
try:
return open("%s/error_log" % tmpdir, 'r').read()
except:
return ''
return html
def invokeCommand(self, tmpdir, fullname):
if os.name=='posix':
cmd = 'cd "%s" && %s %s "%s" 2>error_log 1>/dev/null' % (
tmpdir, self.binary, self.binaryArgs, fullname)
else:
cmd = 'cd "%s" && %s %s "%s"' % (
tmpdir, self.binary, self.binaryArgs, fullname)
os.system(cmd)
try:
htmlfilename = os.path.join(tmpdir, sansext(fullname) + '.html')
htmlfile = open(htmlfilename, 'r')
html = htmlfile.read()
htmlfile.close()
except:
try:
return open("%s/error_log" % tmpdir, 'r').read()
except:
return "transform failed while running %s (maybe this pdf file doesn't support transform)" % cmd
return html
def invokeCommand(self, tmpdir, fullname):
if os.name == 'posix':
cmd = 'cd "%s" && %s %s "%s" 2>error_log 1>/dev/null' % (
tmpdir, self.binary, self.binaryArgs, fullname)
else:
cmd = 'cd "%s" && %s %s "%s"' % (
tmpdir, self.binary, self.binaryArgs, fullname)
if six.PY2:
os.system(cmd)
else:
subprocess.run(cmd, shell=True)
try:
htmlfilename = os.path.join(tmpdir, sansext(fullname) + '.html')
with open(htmlfilename, 'rb') as htmlfile:
html = htmlfile.read()
except:
try:
with open("%s/error_log" % tmpdir, 'r') as fd:
error_log = fd.read()
return error_log
except:
return ("transform failed while running %s (maybe this pdf "
"file doesn't support transform)" % cmd)
return html