I've got the following instruction in Python I'd like to understand and translate in Java
values = struct.unpack_from('>%dL' % 96, input_content, 0) What does >%dL mean? I've checked the doc from python http://docs.python.org/library/struct.html but nothing about the percentage. Shall I consider every returned value as a byte in Java and cast it to either double or long ?
Thanks for your help.
The %d is a Python string formatter similar to that found in C. What it says is put the thing that comes after the closing quote in place of the formatter, in this case 96 in place of %d. The %d specifies a signed integer decimal.
The '>%dL'%96 is an instruction to struct.unpack to say that the thing it needs to unpack is a big endian unsigned long with an unsigned integer decimal inside it. Before the '>%dL' is passed to unpack however the string formatter is resolved and '>%dL' becomes '>96L'. Have a look at the format strings section in `struct.unpack docs
http://docs.python.org/library/stdtypes.html#string-formatting
http://docs.python.org/library/struct.html?highlight=struct.unpack#format-strings
%d by %96 ... if it did, the result would be '>%96L' which is NOT a valid struct format. The second % is an OPERATOR which causes the text substitution of 96 instead of %d resulting in '>96L'. And you haven't addressed my second point.%d is replaced with 96 - reading the string formatting docs should've cleared your section objection but I spelled it out.The Python interactive prompt is your friend:
>>> '>%dL' % 96 '>96L' >>> So,
values = struct.unpack_from('>%dL' % 96, input_content, 0) is equivalent to
values = struct.unpack_from('>96L', input_content, 0) and the struct docs should tell you that '>96L' means 96 bigendian unsigned 32-bit integers.
I can't imagine why the original author wrote it in such an obfuscatory manner. It is necessary to use a technique like that to build a format if the number of items is variable, but not when it's a known constant.
If I would write the same thing in Java it could be something like:
DataInputStream inputcontent = new DataInputStream(in); long[] values = new long[96]; for (int i = 0; i < 96; i++) values[i] = inputcontent.readLong(); with the difference that the python code is using bytes that have been already loaded.
Many thanks to John Machin and Matti for their explanation.