sys.stdin on windows

[zugnush]

I often grep particular patterns out of large logfiles and then
pipeline the output to sort and uniq -c
I thought today to knock up a script to do the counting in a python
dict.

This seems work in linux

$ cat count.py
#!/usr/bin/env python
import sys
from collections import defaultdict
accumulator=defaultdict(int)
for line in sys.stdin.readlines():
accumulator[line.strip()]+=1
print "contents,count"
for key in accumulator.keys():
print key,",",accumulator[key]

$ cat test | ./count.py
contents,count
, 1
23 , 1
1 , 1
3 , 2
2 , 2
5 , 3

When I try to run the same thing on windows I get
IOError: [Error 9] Bad file descriptor

How can I make this more windows friendly?

Thanks
Neil

[Gabriel Genellina]

En Wed, 03 Sep 2008 06:16:03 -0300, zu*****@gmail.com <zu*****@gmail.com>
escribi�:

I often grep particular patterns out of large logfiles and then
pipeline the output to sort and uniq -c
I thought today to knock up a script to do the counting in a python
dict.

This seems work in linux

$ cat count.py
#!/usr/bin/env python
import sys
from collections import defaultdict
accumulator=defaultdict(int)
for line in sys.stdin.readlines():
accumulator[line.strip()]+=1
print "contents,count"
for key in accumulator.keys():
print key,",",accumulator[key]

$ cat test | ./count.py
contents,count
, 1
23 , 1
1 , 1
3 , 2
2 , 2
5 , 3

When I try to run the same thing on windows I get
IOError: [Error 9] Bad file descriptor

How can I make this more windows friendly?

Explicitely invoking the interpreter worked for me. That is, these two
commands worked fine:

type test.txt | python count.py
python count.py < test.txt

But I cannot explain *why* it doesn't work the other way.

--
Gabriel Genellina

[Tim Golden]

Gabriel Genellina wrote:

En Wed, 03 Sep 2008 06:16:03 -0300, zu*****@gmail.com
<zu*****@gmail.comescribi�:

>I often grep particular patterns out of large logfiles and then
pipeline the output to sort and uniq -c
I thought today to knock up a script to do the counting in a python
dict.

This seems work in linux

$ cat count.py
#!/usr/bin/env python
import sys
from collections import defaultdict
accumulator=defaultdict(int)
for line in sys.stdin.readlines():
accumulator[line.strip()]+=1
print "contents,count"
for key in accumulator.keys():
print key,",",accumulator[key]

$ cat test | ./count.py
contents,count
, 1
23 , 1
1 , 1
3 , 2
2 , 2
5 , 3

When I try to run the same thing on windows I get
IOError: [Error 9] Bad file descriptor

How can I make this more windows friendly?

Explicitely invoking the interpreter worked for me. That is, these two
commands worked fine:

type test.txt | python count.py
python count.py < test.txt

But I cannot explain *why* it doesn't work the other way.

Known bug in NT-based file association. I'll try
to find an online reference, but that's basically
what it comes to. I think you can faff-about with
batch files to achieve the effect, but I can't
quite remember.

http://support.microsoft.com/kb/321788

TJG

[zugnush]

On Sep 3, 11:16*pm, Tim Golden <[e-mail]:

Gabriel Genellina wrote:

En Wed, 03 Sep 2008 06:16:03 -0300, [e-mail]
<[e-mail] :

I often grep particular patterns out of large logfiles and then
pipeline the output to sort and uniq -c
I thought today to knock up a script to do the counting in a python
dict.

This seems work in linux

$ cat count.py
#!/usr/bin/env python
import sys
from collections import defaultdict
accumulator=defaultdict(int)
for line in sys.stdin.readlines():
* * accumulator[line.strip()]+=1
print "contents,count"
for key in accumulator.keys():
* * print key,",",accumulator[key]

$ cat test | ./count.py
contents,count
*, 1
23 , 1
1 , 1
3 , 2
2 , 2
5 , 3

When I try to run the same thing on windows I get
IOError: [Error 9] Bad file descriptor

How can I make this more windows friendly?

Explicitely invoking the interpreter worked for me. That is, these two
commands worked fine:

type test.txt | python count.py
python count.py < test.txt

But I cannot explain *why* it doesn't work the other way.

Known bug in NT-based file association. I'll try
to find an online reference, but that's basically
what it comes to. I think you can faff-about with
batch files to achieve the effect, but I can't
quite remember.

http://support.microsoft.com/kb/321788

TJG

Thanks.

I'll ues the explicit python call.

[Gabriel Genellina]

En Wed, 03 Sep 2008 07:16:12 -0300, Tim Golden <ma**@timgolden.me.uk>
escribi�:

Gabriel Genellina wrote:

>En Wed, 03 Sep 2008 06:16:03 -0300, zu*****@gmail.com
<zu*****@gmail.comescribi�:

>>When I try to run the same thing on windows I get
IOError: [Error 9] Bad file descriptor

How can I make this more windows friendly?

Explicitely invoking the interpreter worked for me. That is, these two
commands worked fine:
type test.txt | python count.py
python count.py < test.txt
But I cannot explain *why* it doesn't work the other way.

Known bug in NT-based file association. I'll try to find an online
reference, but that's basically
what it comes to. I think you can faff-about with
batch files to achieve the effect, but I can't
quite remember.

http://support.microsoft.com/kb/321788

Uhmm... That KB article says the bug was corrected in Windows XP SP1, but
I have SP3 installed and the test failed. Updating the registry by hand
solved the problem. A regression maybe?

--
Gabriel Genellina