Example #1
0
    def _write(self, raw_data, filename) :
        """
        Write raw data to a compressed file.

        @arg raw_data: The raw_data to be compressed and written
        @type raw_data: byte string
        @arg filename: The intended name of the outfile
        @type filename: unicode

        @return: outfile ; The full path and name of the file written
        @rtype: unicode
        """
        result = chardet.detect(raw_data)
        if result['confidence'] > 0.5:
            encoding = result['encoding']
        else:
            encoding = 'utf-8'

        if not util.is_utf8_alias(encoding):
            raw_data = raw_data.decode(encoding).encode('utf-8')

        # Compress the data to save disk space.
        comp = bz2.BZ2Compressor()
        data = comp.compress(raw_data)
        data += comp.flush()
        out_handle = open(self._nametofile(filename), "wb")
        out_handle.write(data)
        out_handle.close()

        return out_handle.name      # return the full path to the file
Example #2
0
    def _write(self, raw_data, filename) :
        """
        Write raw data to a compressed file.

        @arg raw_data: The raw_data to be compressed and written
        @type raw_data: byte string
        @arg filename: The intended name of the outfile
        @type filename: unicode

        @return: outfile ; The full path and name of the file written
        @rtype: unicode
        """
        result = chardet.detect(raw_data)
        if result['confidence'] > 0.5:
            encoding = result['encoding']
        else:
            encoding = 'utf-8'

        if not util.is_utf8_alias(encoding):
            raw_data = raw_data.decode(encoding).encode('utf-8')

        # Compress the data to save disk space.
        comp = bz2.BZ2Compressor()
        data = comp.compress(raw_data)
        data += comp.flush()
        out_handle = open(self._nametofile(filename), "wb")
        out_handle.write(data)
        out_handle.close()

        return out_handle.name      # return the full path to the file
Example #3
0
    def _write(self, raw_data, filename):
        """
        Write raw data to a compressed file.

        :arg str raw_data: The raw_data to be compressed and written.
        :arg unicode filename: The intended name of the output filename.

        :returns: The full path and name of the file written.
        :rtype: unicode
        """
        result = chardet.detect(raw_data)
        if result['confidence'] > 0.5:
            encoding = unicode(result['encoding'])
        else:
            encoding = 'utf-8'

        if not util.is_utf8_alias(encoding):
            try:
                raw_data = raw_data.decode(encoding).encode('utf-8')
            except UnicodeDecodeError:
                self._output.addMessage(
                    __file__, 4, 'ENOPARSE',
                    'Could not decode file (using {} encoding).'.format(
                        encoding))
                return None

        # Compress the data to save disk space.
        comp = bz2.BZ2Compressor()
        data = comp.compress(raw_data)
        data += comp.flush()
        out_handle = open(self._name_to_file(filename), 'wb')
        out_handle.write(data)
        out_handle.close()

        # Return the full path to the file.
        return out_handle.name
Example #4
0
    def _write(self, raw_data, filename):
        """
        Write raw data to a compressed file.

        :arg str raw_data: The raw_data to be compressed and written.
        :arg unicode filename: The intended name of the output filename.

        :returns: The full path and name of the file written.
        :rtype: unicode
        """
        result = chardet.detect(raw_data)
        if result['confidence'] > 0.5:
            encoding = unicode(result['encoding'])
        else:
            encoding = 'utf-8'

        if not util.is_utf8_alias(encoding):
            try:
                raw_data = raw_data.decode(encoding).encode('utf-8')
            except UnicodeDecodeError:
                self._output.addMessage(
                    __file__, 4, 'ENOPARSE',
                    'Could not decode file (using {} encoding).'.format(
                        encoding))
                return None

        # Compress the data to save disk space.
        comp = bz2.BZ2Compressor()
        data = comp.compress(raw_data)
        data += comp.flush()
        out_handle = open(self._name_to_file(filename), 'wb')
        out_handle.write(data)
        out_handle.close()

        # Return the full path to the file.
        return out_handle.name