def pdfrw_object(fn):
'''hachoir doesn't do pdf'''
if sys.platform == 'win32':
newfn = os.path.join(CACHEDIR, 'Workaround', os.path.basename(fn))
if os.path.isfile(newfn):
newfn = auto_increment_fn(newfn)
shutil.copyfile(fn, newfn)
fn = newfn
fakeobj = get_fake_metadata_object(fn)
try:
reader = PdfReader(fn)
except Exception, exc:
traceback.print_exc(file=open(os.path.join(SYSDIR, 'dbgop'), 'w'))
return None
def get_real_metadata_object(fn):
'''get object with metadata gleaned from internal file metadata'''
#TODO: workaround for file locking bug
if sys.platform == 'win32':
newfn = os.path.join(CACHEDIR, 'Workaround', os.path.basename(fn))
if os.path.isfile(newfn):
newfn = auto_increment_fn(newfn)
shutil.copyfile(fn, newfn)
fn = newfn
obj = get_fake_metadata_object(fn)
d = get_real_metadata(fn)
if not d: return obj
obj.author = d['author']
obj.title = d['title']
obj.date = d['creation_date']
obj.FileData_DateCreated = d['creation_date']
obj.FileData_DateModified = d['modification_date']
obj.FileData_FileType = determine_file_type(fn)
obj.FileData_FileName = os.path.basename(fn)
return obj
def get_odf_metadata_object(fn):
# if sys.platform == 'win32':
# newfn = os.path.join(CACHEDIR, 'Workaround', os.path.basename(fn))
# if os.path.isfile(newfn):
# newfn = auto_increment_fn(newfn)
# shutil.copyfile(fn, newfn)
# fn = newfn
obj = get_fake_metadata_object(fn)
try:
d = get_odf_metadata(fn)
except:
traceback.print_exc()
print 'Parsing oxml document %s failed' % fn
return obj
obj.author = d['author']
obj.title = d['title']
obj.date = d['creation_date']
obj.FileData_DateCreated = d['creation_date']
obj.FileData_DateModified = d['modification_date']
if d['description']: obj.BibData_Annote = d['description']
obj.FileData_FileType = 'odf_doc'
obj.FileData_FileName = os.path.basename(fn)
return obj
def get_oxml_metadata_object(fn):
# if sys.platform == 'win32':
# newfn = os.path.join(CACHEDIR, 'Workaround', os.path.basename(fn))
# if os.path.isfile(newfn):
# newfn = auto_increment_fn(newfn)
# shutil.copyfile(fn, newfn)
# fn = newfn
obj = get_fake_metadata_object(fn)
try:
d = get_oxml_metadata(fn)
except:
traceback.print_exc()
print 'Parsing oxml document %s failed' % fn
return obj
obj.author = d['author']
obj.title = d['title']
obj.date = d['creation_date']
obj.FileData_DateCreated = d['creation_date']
obj.FileData_DateModified = d['modification_date']
if d['description']: obj.BibData_Annote = d['description']
obj.FileData_FileType = 'oxml_doc'
obj.FileData_FileName = os.path.basename(fn)
return obj
