def find_command(self, name):
config_name = '%s_bin' % name
config_name = config_name.replace('-', '_').upper()
command_path = self.manager.get_env(config_name, find_executable(name))
if command_path is None:
raise SystemException('Cannot find binary: %s' % name)
return command_path
def dump_table(self, file_path, table_name, csv_path):
mdb_export = self.find_command('mdb-export')
if mdb_export is None:
raise SystemException('mdb-tools is not available')
args = [mdb_export, '-b', 'strip', file_path, table_name]
with open(csv_path, 'w') as fh:
subprocess.call(args, stdout=fh)
return csv_path
def get_convert():
"""Find the best available method to convert documents to the
PDF format."""
if not hasattr(settings, '_ingestors_convert'):
if ServiceDocumentConverter.is_available():
settings._ingestors_convert = ServiceDocumentConverter()
elif LocalDocumentConverter.is_available():
settings._ingestors_convert = LocalDocumentConverter()
else:
raise SystemException("Document conversion is not available")
return settings._ingestors_convert
def get_ocr():
"""Find the best available method to perform OCR."""
if not hasattr(settings, '_ingestors_ocr'):
if GoogleOCRService.is_available():
settings._ingestors_ocr = GoogleOCRService()
elif ServiceOCRService.is_available():
settings._ingestors_ocr = ServiceOCRService()
elif LocalOCRService.is_available():
settings._ingestors_ocr = LocalOCRService()
else:
raise SystemException("OCR is not available")
return settings._ingestors_ocr
def get_tables(self, local_path):
mdb_tables = self.find_command('mdb-tables')
if mdb_tables is None:
raise SystemException('mdb-tools is not available')
try:
output = subprocess.check_output([mdb_tables, local_path])
return [
t.strip().decode('utf-8') for t in output.split(b' ')
if len(t.strip())
]
except subprocess.CalledProcessError as cpe:
log.warning("Failed to open MDB: %s", cpe)
raise ProcessingException("Failed to extract Access database.")
def exec_command(self, command, *args):
binary = self.find_command(command)
if binary is None:
raise SystemException("Program not found: %s" % command)
cmd = [binary]
cmd.extend(args)
try:
code = subprocess.call(cmd,
timeout=self.COMMAND_TIMEOUT,
stdout=open(os.devnull, 'wb'))
except (IOError, OSError) as ose:
raise ProcessingException('Error: %s' % ose)
except subprocess.TimeoutExpired:
raise ProcessingException('Processing timed out.')
if code != 0:
raise ProcessingException('Failed: %s' % ' '.join(cmd))