How to diff file and output stream "on-the-fly"?
Question:
I need to create a diff file using standard UNIX diff command with python subprocess module. The problem is that I must compare file and stream without creating tempopary file. I thought about using named pipes via os.mkfifo method, but didn’t reach any good result. Please, can you write a simple example on how to solve this stuff? I tried like so:
fifo = 'pipe'
os.mkfifo(fifo)
op = popen('cat ', fifo)
print >> open(fifo, 'w'), output
os.unlink(fifo)
proc = Popen(['diff', '-u', dumpfile], stdin=op, stdout=PIPE)
but it seems like diff doesn’t see the second argument.
Answers:
You can use “-” as an argument to diff to mean stdin.
You could perhaps consider using the difflib python module (I’ve linked to an example here) and create something that generates and prints the diff directly rather than relying on diff. The various function methods inside difflib can receive character buffers which can be processed into diffs of various types.
Alternatively, you can construct a shell pipeline and use process substitution like so
diff <(cat pipe) dumpfile # You compare the output of a process and a physical file without explicitly using a temporary file.
For details, check out http://tldp.org/LDP/abs/html/process-sub.html
I need to create a diff file using standard UNIX diff command with python subprocess module. The problem is that I must compare file and stream without creating tempopary file. I thought about using named pipes via os.mkfifo method, but didn’t reach any good result. Please, can you write a simple example on how to solve this stuff? I tried like so:
fifo = 'pipe'
os.mkfifo(fifo)
op = popen('cat ', fifo)
print >> open(fifo, 'w'), output
os.unlink(fifo)
proc = Popen(['diff', '-u', dumpfile], stdin=op, stdout=PIPE)
but it seems like diff doesn’t see the second argument.
You can use “-” as an argument to diff to mean stdin.
You could perhaps consider using the difflib python module (I’ve linked to an example here) and create something that generates and prints the diff directly rather than relying on diff. The various function methods inside difflib can receive character buffers which can be processed into diffs of various types.
Alternatively, you can construct a shell pipeline and use process substitution like so
diff <(cat pipe) dumpfile # You compare the output of a process and a physical file without explicitly using a temporary file.
For details, check out http://tldp.org/LDP/abs/html/process-sub.html