def print_model(model):
printable_model = dict(model, type=model['type'].__name__)
print fmt % printable_model
if dump:
from hwp5.binmodel import model_to_json
print model_to_json(model, sort_keys=True, indent=2)
def print_log(fmt, *args):
print fmt % args
from hwp5.bintype import log_events
list(log_events(model['binevents'], print_log))
def print_model(model):
printable_model = dict(model, type=model['type'].__name__)
print fmt % printable_model
if dump:
from hwp5.binmodel import model_to_json
print model_to_json(model, sort_keys=True, indent=2)
def print_log(fmt, *args):
print fmt % args
from hwp5.bintype import log_events
list(log_events(model['binevents'], print_log))
def print_models_json(models):
from hwp5.binmodel import model_to_json
from hwp5.utils import generate_json_array
jsonobjects = (model_to_json(model, sort_keys=True, indent=2)
for model in models)
for s in generate_json_array(jsonobjects):
sys.stdout.write(s)
def print_models_json(models):
from hwp5.binmodel import model_to_json
from hwp5.utils import generate_json_array
jsonobjects = (model_to_json(model, sort_keys=True, indent=2)
for model in models)
for s in generate_json_array(jsonobjects):
sys.stdout.write(s)
def test_model_to_json_with_unparsed(self):
model = dict(type=RecordModel, content=[], payload='\x00\x01\x02\x03',
unparsed='\xff\xfe\xfd\xfc')
json_string = model_to_json(model)
jsonobject = json.loads(json_string)
self.assertEquals(['ff fe fd fc'], jsonobject['unparsed'])
def test_model_to_json(self):
from hwp5.binmodel import model_to_json
model = self.hwp5file.docinfo.model(0)
json = model_to_json(model)
simplejson = importjson()
jsonobject = simplejson.loads(json)
self.assertEquals('DocumentProperties', jsonobject['type'])
def test_model_to_json_with_unparsed(self):
model = dict(type=RecordModel, content=[], payload=b'\x00\x01\x02\x03',
unparsed=b'\xff\xfe\xfd\xfc')
json_string = model_to_json(model)
jsonobject = json.loads(json_string)
self.assertEqual(['ff fe fd fc'], jsonobject['unparsed'])
def test_model_to_json(self):
from hwp5.binmodel import model_to_json
model = self.hwp5file.docinfo.model(0)
json = model_to_json(model)
simplejson = importjson()
jsonobject = simplejson.loads(json)
self.assertEquals('DocumentProperties', jsonobject['type'])
def test_model_to_json_with_controlchar(self):
model = self.hwp5file.bodytext.section(0).model(1)
json_string = model_to_json(model)
jsonobject = json.loads(json_string)
self.assertEqual('ParaText', jsonobject['type'])
self.assertEqual([[0, 8],
dict(code=2, param='\x00' * 8, chid='secd')],
jsonobject['content']['chunks'][0])
def test_model_to_json_with_controlchar(self):
model = self.hwp5file.bodytext.section(0).model(1)
json_string = model_to_json(model)
jsonobject = json.loads(json_string)
self.assertEqual('ParaText', jsonobject['type'])
self.assertEqual(
[[0, 8], dict(code=2, param='\x00' * 8, chid='secd')],
jsonobject['content']['chunks'][0])
def test_model_to_json_with_unparsed(self):
from hwp5.binmodel import model_to_json
model = dict(type=RecordModel, content=[], payload='\x00\x01\x02\x03',
unparsed='\xff\xfe\xfd\xfc')
json = model_to_json(model)
simplejson = importjson()
jsonobject = simplejson.loads(json)
self.assertEquals(['ff fe fd fc'], jsonobject['unparsed'])
def test_model_to_json_with_controlchar(self):
from hwp5.binmodel import model_to_json
model = self.hwp5file.bodytext.section(0).model(1)
json = model_to_json(model)
simplejson = importjson()
jsonobject = simplejson.loads(json)
self.assertEquals('ParaText', jsonobject['type'])
self.assertEquals(
[[0, 8], dict(code=2, param='\x00' * 8, chid='secd')],
jsonobject['content']['chunks'][0])
def test_model_to_json_with_controlchar(self):
from hwp5.binmodel import model_to_json
model = self.hwp5file.bodytext.section(0).model(1)
json = model_to_json(model)
simplejson = importjson()
jsonobject = simplejson.loads(json)
self.assertEquals('ParaText', jsonobject['type'])
self.assertEquals([[0, 8],
dict(code=2, param='\x00' * 8, chid='secd')],
jsonobject['content']['chunks'][0])
def test_model_to_json_with_unparsed(self):
from hwp5.binmodel import model_to_json
model = dict(type=RecordModel,
content=[],
payload='\x00\x01\x02\x03',
unparsed='\xff\xfe\xfd\xfc')
json = model_to_json(model)
simplejson = importjson()
jsonobject = simplejson.loads(json)
self.assertEquals(['ff fe fd fc'], jsonobject['unparsed'])
def main(args):
filenames = args['<hwp5files>']
from hwp5.dataio import ParseError
from hwp5.binmodel import Hwp5File
conditions = []
if args['--model']:
def with_model_name(model):
return args['--model'] == model['type'].__name__
conditions.append(with_model_name)
if args['--tag']:
tag = args['--tag']
try:
tag = int(tag)
except ValueError:
pass
else:
from hwp5.tagids import tagnames
tag = tagnames[tag]
def with_tag(model):
return model['tagname'] == tag
conditions.append(with_tag)
if args['--incomplete']:
def with_incomplete(model):
return 'unparsed' in model
conditions.append(with_incomplete)
def flat_models(hwp5file, **kwargs):
for model in hwp5file.docinfo.models(**kwargs):
model['stream'] = 'DocInfo'
yield model
for section in hwp5file.bodytext:
for model in hwp5file.bodytext[section].models(**kwargs):
model['stream'] = 'BodyText/'+section
yield model
for filename in filenames:
try:
hwp5file = Hwp5File(filename)
def with_filename(models):
for model in models:
model['filename'] = filename
yield model
models = flat_models(hwp5file)
models = with_filename(models)
for model in models:
if all(condition(model) for condition in conditions):
print '%s:%s(%s): %s' % (model['filename'],
model['stream'],
model['seqno'],
model['type'].__name__)
if args['--dump']:
from hwp5.binmodel import model_to_json
print model_to_json(model, sort_keys=True, indent=2)
def print_log(fmt, *args):
print fmt % args
from hwp5.bintype import log_events
list(log_events(model['binevents'], print_log))
except ParseError, e:
from hwp5.proc import logger
logger.error('---- On processing %s:', filename)
e.print_to_logger(logger)
def test_model_to_json_should_not_modify_input(self):
from hwp5.binmodel import model_to_json
model = self.hwp5file.docinfo.model(0)
model_to_json(model, indent=2, sort_keys=True)
self.assertFalse(isinstance(model['type'], basestring))
def test_model_to_json(self):
model = self.hwp5file.docinfo.model(0)
json_string = model_to_json(model)
jsonobject = json.loads(json_string)
self.assertEqual('DocumentProperties', jsonobject['type'])
def test_model_to_json(self):
model = self.hwp5file.docinfo.model(0)
json_string = model_to_json(model)
jsonobject = json.loads(json_string)
self.assertEqual('DocumentProperties', jsonobject['type'])
def test_model_to_json_should_not_modify_input(self):
model = self.hwp5file.docinfo.model(0)
model_to_json(model, indent=2, sort_keys=True)
self.assertFalse(isinstance(model['type'], basestring))
