def __init__(self, fpath, lang, regex, any_ext=False):
"""
:param fpath:
:param lang:
:param regex:
:param any_ext:
"""
# Check lang input
if lang not in LANGUAGE_EXTENSIONS:
raise ValueError('Unrecognized input language "{:s}"'.format(lang))
# Check file extension
if not any_ext:
ext = getext(fpath)
if ext not in LANGUAGE_EXTENSIONS[lang]:
raise FileExistsError('File extension is not supported')
# Check file content
self.content = readfile(fpath)
if self.content.strip() == '':
raise FileExistsError('File is empty')
# Define self properties
self.lang = lang.lower()
self.regex = regex
# Parse file
self.nodes = OrderedDict()
self.founds = find_matches(self.content, self.regex['main'])
if self.founds:
self.parse()
def parse(fpath):
"""
Routes what parser to use depending on the file extension of a file
:param fpath: file path of file
:return: none
"""
# Check existence
if not isfile(fpath):
raise FileNotFoundError('File is not existed')
# Get file extension
ext = getext(fpath)
# Get module
parser = None
for lang, exts in LANGUAGE_EXTENSIONS.items():
if ext in exts:
parser = getattr(sys.modules[__name__], lang)
break
# If extension is not found
if parser is None:
raise FileExistsError('File extension is not supported')
# Parse and return
return parser(fpath)
def __parse(self):
assert isxml(self.xml_file)
with open(self.xml_file, 'rb') as fp:
tree = ElementTree.parse(fp)
pages = tree.findall('page')
for page in pages:
self.pages.append(
TXTSEP.join(getext(text).strip() for text in page.findall('text')).strip()
)
rmfile(self.xml_file)
def update_gui(self):
"""
Update label text, with info about paths and file
size and number.
"""
text = self.lng['label_file'] + str(self.set.set_file) + '\n' + \
self.lng['label_dir'] + str(self.set.set_dir) + '\n' + \
self.lng['label_dirout'] + str(self.set.set_dirout)
# A directory is selected, so calculate the size
# and the number of the files it contains. Skip
# the calculation if user selected the same folder.
if self.set.set_file != self.set.NOP:
ext = helpers.getext(self.set.set_file)
# See if the extension is described in the language file.
try:
ext = self.lng['ext_%s' % ext]
except:
pass
text = text + '\n' + self.lng['label_type'] + ext
elif self.set.set_dir != self.set.NOP:
# Calculate file number and size:
# TODO: Add an option to override this.
#if self.set.set_dir != self.set.previous_folder:
if 1: # overrided for time being
# Inform user that this migh take time:
self.master.update_status('label_pleasewait', 0)
# Pass the extension list, depending on the
# conversion mode:
if self.set.set_convmode == 'tolat':
extensions = self.set.set_extensions
self.master.tocyr.EXT = self.set.extensions
else:
extensions = self.set.set_extensions_tocyr
self.master.tocyr.EXT = self.set.extensions_tocyr
# Calculate
self.master.tocyr.RECURSIVE = self.set.set_recursive
self.filecount, self.filesize = \
self.master.tocyr.calculatedirsize(self.set.set_dir)
self.set.previous_folder = self.set.set_dir
if not self.filecount:
self.master.update_status('label_nosupportedfiles', 0)
else:
self.master.update_status('label_ok', 0)
# The number of files
text = text + '\n' + self.lng['label_number'] % self.filecount
# The size of file(s)
text = text + '\n' + self.lng['label_size'] % \
'%0.2f' % self.filesize
# The list of recognised extensions
text = text + '\n' + self.lng['options_extensions'] + \
extensions.replace(",", ", ")
# Conversion mode
text = text + '\n' + self.lng['label_conv%s' % \
self.set.set_convmode]
self.label_selection.configure(text=text, justify='left')
def check_allowed_extensions(self):
"""
Check if file extension is allowed in the
conversion mode.
"""
# If file is slected, but the converson mode cannot be
# applied to it, reset the path.
file_selected = self.set.set_file != self.set.NOP
if file_selected and (self.set.set_convmode == 'tocyr'):
if helpers.getext(self.set.set_file) not in \
self.tocyr.EXTENSIONS:
self.set.set_file = self.set.NOP
messagebox.showwarning('',
self.lng['msg_extension_not_supported'])