用Python 2.6+和3编写二进制文件

Question

I am writing a program that needs to write a binary file. In contains a header of strings (key-value pairs) and numeric data (which can be little-endian or big-endian) and I am looking for a way to write a program that runs unchanged in Python 2.6+ and Python 3.2+.

Can anyone suggest some best practices? Additionally, what is the right way to deal with endianness without cluttering the logic of my program with struct.pack . Should I subclass BufferedWriter ?

Thanks in advance.

Answer 1

Once you created the binary data, you just write it to a file opened in binary mode. That's all there is to it. No compatibility problems between Python 2 and 3 there at all.

Subclassing BufferedWriter is completely unecessary.

How you create the data is a different question, but again there I don't see any obvious incompatibility problems.

Answer 2

尚不清楚您是否已经具有需要匹配的预定义二进制文件格式，但是如果没有，则您只是试图以一种可以被多种编程语言（以及上的多个Python版本）读取的方式来序列化数据结构。多个平台），则可能需要查看协议缓冲区

Answer 3

You can use os.open + os.write + os.close. However, these need while loops for reliability.

I've done this using http://stromberg.dnsalias.org/~strombrg/bufsock.html (I wrote it, my former employer allowed me to opensource it) to achieve this in a deduplicating backup program that does lots of binary I/O. Doing it this way doesn't require loops, even if you're employing signals. Don't let the "bufsock" name fool you - it's also good for file I/O.

BTW, writing stuff in an endian-dependent way is usually a mistake in the long term. If you're not content with the builtin Python tools (I'm not always, EG once I needed a 3 byte integer type), it's probablty better to tear apart your numbers using divmod 256. Another alternative is to use http://stromberg.dnsalias.org/~strombrg/base255.html to get null-terminateable strings.