Detecting line endings

[Fuzzyman]

Hello all,

I'm trying to detect line endings used in text files. I *might* be
decoding the files into unicode first (which may be encoded using
multi-byte encodings) - which is why I'm not letting Python handle the
line endings.

Is the following safe and sane :

text = open('test.txt', 'rb').read()
if encoding:
text = text.decode(encoding)
ending = '\n' # default
if '\r\n' in text:
text = text.replace('\r\n', '\n')
ending = '\r\n'
elif '\n' in text:
ending = '\n'
elif '\r' in text:
text = text.replace('\r', '\n')
ending = '\r'


My worry is that if '\n' *doesn't* signify a line break on the Mac,
then it may exist in the body of the text - and trigger ``ending =
'\n'`` prematurely ?

All the best,

Fuzzyman
http://www.voidspace.org.uk/python/index.shtml

 

[Sybren Stuvel]

Fuzzyman enlightened us with:

My worry is that if '\n' *doesn't* signify a line break on the Mac,
then it may exist in the body of the text - and trigger ``ending =
'\n'`` prematurely ?

I'd count the number of occurences of '\r\n', '\n' without a preceding
'\r' and '\r' without following '\n', and let the majority decide.

Sybren

 

[Fuzzyman]

Fuzzyman enlightened us with:

I'd count the number of occurences of '\r\n', '\n' without a preceding
'\r' and '\r' without following '\n', and let the majority decide.

Sounds reasonable, edge cases for small files be damned.

Fuzzyman
http://www.voidspace.org.uk/python/index.shtml

 

[Fuzzyman]

Fuzzyman enlightened us with:

I'd count the number of occurences of '\r\n', '\n' without a preceding
'\r' and '\r' without following '\n', and let the majority decide.

This is what I came up with. As you can see from the docstring, it
attempts to sensible(-ish) things in the event of a tie, or no line
endings at all.

Comments/corrections welcomed. I know the tests aren't very useful
(because they make no *assertions* they won't tell you if it breaks),
but you can see what's going on :

import re
import os

rn = re.compile('\r\n')
r = re.compile('\r(?!\n)')
n = re.compile('(?<!\r)\n')

# Sequence of (regex, literal, priority) for each line ending
line_ending = [(n, '\n', 3), (rn, '\r\n', 2), (r, '\r', 1)]


def find_ending(text, default=os.linesep):
"""
Given a piece of text, use a simple heuristic to determine the line
ending in use.

Returns the value assigned to default if no line endings are found.
This defaults to ``os.linesep``, the native line ending for the
machine.

If there is a tie between two endings, the priority chain is
``'\n', '\r\n', '\r'``.
"""
results = [(len(exp.findall(text)), priority, literal) for
exp, literal, priority in line_ending]
results.sort()
print results
if not sum([m[0] for m in results]):
return default
else:
return results[-1][-1]

if __name__ == '__main__':
tests = [
'hello\ngoodbye\nmy fish\n',
'hello\r\ngoodbye\r\nmy fish\r\n',
'hello\rgoodbye\rmy fish\r',
'hello\rgoodbye\n',
'',
'\r\r\r \n\n',
'\n\n \r\n\r\n',
'\n\n\r \r\r\n',
'\n\r \n\r \n\r',
]
for entry in tests:
print repr(entry)
print repr(find_ending(entry))
print

All the best,


Fuzzyman
http://www.voidspace.org.uk/python/index.shtml

 

[Alex Martelli]

Hello all,

I'm trying to detect line endings used in text files. I *might* be
decoding the files into unicode first (which may be encoded using

Open the file with 'rU' mode, and check the file object's newline
attribute.

My worry is that if '\n' *doesn't* signify a line break on the Mac,

It does, since a few years, since MacOSX is version of Unix to all
practical intents and purposes.


Alex

 

[Sybren Stuvel]

Fuzzyman enlightened us with:

This is what I came up with. [...] Comments/corrections welcomed.

You could use a little more comments in the code, but apart from that
it looks nice.

Sybren

 

[Fuzzyman]

Open the file with 'rU' mode, and check the file object's newline
attribute.

Ha, so long as it works with Python 2.2, that makes things a bit
easier.

Rats, I liked that snippet of code (I'm a great fan of list
comprehensions).

It does, since a few years, since MacOSX is version of Unix to all
practical intents and purposes.

I wondered if that might be the case. I think I've worried about this
more than enough now.

Thanks

Fuzzyman
http://www.voidspace.org.uk/python/index.shtml

 

[Fuzzyman]

Open the file with 'rU' mode, and check the file object's newline
attribute.

Do you know if this works for multi-byte encodings ? Do files have
metadata associated with them showing the line-ending in use ?

I suppose I could test this...

All the best,


Fuzzy

 

[Arthur]

Open the file with 'rU' mode, and check the file object's newline
attribute.

Do you think it would be sensible to have file.readline in universal
newline support by default?

I just got flummoxed by this issue, working with a (pre-alpha) package
by very experienced Python programmers who sent file.readline to
tokenizer.py without universal newline support. Went on a long (and
educational) journey trying to figure out why my file was not being
processed as expected.

Are there circumstances that it would be sensible to have tokenizer
process files without universal newline support?

The result here was having tokenizer detect indentation inconstancies
that did not exist - in the sense that the files were compiled and ran
fine by Python.exe.

Art

 

[Arthur]

Alex Martelli wrote:

I just got flummoxed by this issue, working with a (pre-alpha) package
by very experienced Python programmers who sent file.readline to
tokenizer.py without universal newline support. Went on a long (and
educational) journey trying to figure out why my file was not being
processed as expected.

For example, the widely used MoinMoin source code colorizer sends files
to tokenizer without universal newline support:

http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/52298

Is my premise that tokenizer needs universal newline support to be
reliable correct?

What else could put it out of sync with the complier?

Art

 

[Bengt Richter]

Hello all,

I'm trying to detect line endings used in text files. I *might* be
decoding the files into unicode first (which may be encoded using
multi-byte encodings) - which is why I'm not letting Python handle the
line endings.

Is the following safe and sane :

text = open('test.txt', 'rb').read()
if encoding:
text = text.decode(encoding)
ending = '\n' # default
if '\r\n' in text:
text = text.replace('\r\n', '\n')
ending = '\r\n'
elif '\n' in text:
ending = '\n'
elif '\r' in text:
text = text.replace('\r', '\n')
ending = '\r'


My worry is that if '\n' *doesn't* signify a line break on the Mac,
then it may exist in the body of the text - and trigger ``ending =
'\n'`` prematurely ?

Are you guaranteed that text bodies don't contain escape or quoting
mechanisms for binary data where it would be a mistake to convert
or delete an '\r' ? (E.g., I think XML CDATA might be an example).

Regards,
Bengt Richter

 

[Alex Martelli]

Do you know if this works for multi-byte encodings ? Do files have

You mean when you open them with the codecs module?

metadata associated with them showing the line-ending in use ?

Not in the filesystems I'm familiar with (they did use to, in
filesystems used on VMS and other ancient OSs, but that was a very long
time ago).


Alex

 

[Fuzzyman]

Are you guaranteed that text bodies don't contain escape or quoting
mechanisms for binary data where it would be a mistake to convert
or delete an '\r' ? (E.g., I think XML CDATA might be an example).

My personal use case is for reading config files in arbitrary encodings
(so it's not an issue).

How would Python handle opening such files when not in binary mode ?
That may be an issue even on Linux - if you open a windows file and
use splitlines does Python convert '\r\n' to '\n' ? (or does it leave
the extra '\r's in place, which is *different to the behaviour under
windows).

All the best,

Fuzzyman
http://www.voidspace.org.uk/python/index.shtml

 

[Fuzzyman]

You mean when you open them with the codecs module?

No, if I open a UTF16 encoded file in universal mode - will it still
have the correct lineending attribute ?

I can't open with a codec unless an encoding is explicitly supplied. I
still want to detect UTF16 even if the encoding isn't specified.

As I said, I ought to test this... Without metadata I wonder how Python
determines it ?

All the best,

Fuzzyman
http://www.voidspace.org.uk/python/index.shtml

 

[ajsiegel]

Arthur wrote:

Is my premise that tokenizer needs universal newline support to be
reliable correct?

What else could put it out of sync with the complier?

Anybody out there?

Is my question, and the real world issue that provked it, unclear.

Is the answer too obvious?

Have I made *everybody's* kill list?

Isn't it a prima facie issue if the tokenizer fails in ways
incompatible with what the compiler is seeing?

Is this just easy, and I am making it hard? As I apparently do with
Python more generally.

Art

 

[Alex Martelli]

I can't open with a codec unless an encoding is explicitly supplied. I
still want to detect UTF16 even if the encoding isn't specified.

As I said, I ought to test this... Without metadata I wonder how Python
determines it ?

It doesn't. Python doesn't even try to guess: nor would any other
sensible programming language.


Alex

 

[Fuzzyman]

It doesn't. Python doesn't even try to guess: nor would any other
sensible programming language.

Right, so opening in "rU" mode and testing the 'newline' attribute
*won't* work for UTF16 encoded files. (Which was what I was asking.)

I'll have to read, determine encoding, decode, then *either* use my
code to determine line endings *or* use ``splitlines(True)``.

All the best,

Fuzzyman
http://www.voidspace.org.uk/python/index.shtml

 

[Fuzzyman]

Just to confirm, for a UTF16 encoded file, the newlines attribute is
``None``.

All the best,

Fuzzyman
http://www.voidspace.org.uk/python/index.shtml

 

[Fuzzyman]

Just to confirm, for a UTF16 encoded file, the newlines attribute is
``None``.

Hmmm... having read the documentation, the newlines attribute remains
None until some newlines are encountered.

I don't think it's technique is any better than mine though. ;-)

Fuzzy
http://www.voidspace.org.uk/python/index.shtml