Python: How to ignore #comment lines when reading in a file

I recommend you don't ignore the whole line when you see a # character; just ignore the rest of the line. You can do that easily with a string method function called partition:

with open("filename") as f:
    for line in f:
        line = line.partition('#')[0]
        line = line.rstrip()
        # ... do something with line ...

partition returns a tuple: everything before the partition string, the partition string, and everything after the partition string. So, by indexing with [0] we take just the part before the partition string.

EDIT: If you are using a version of Python that doesn't have partition(), here is code you could use:

with open("filename") as f:
    for line in f:
        line = line.split('#', 1)[0]
        line = line.rstrip()
        # ... do something with line ...

This splits the string on a '#' character, then keeps everything before the split. The 1 argument makes the .split() method stop after a one split; since we are just grabbing the 0th substring (by indexing with [0]) you would get the same answer without the 1 argument, but this might be a little bit faster. (Simplified from my original code thanks to a comment from @gnr. My original code was messier for no good reason; thanks, @gnr.)

You could also just write your own version of partition(). Here is one called part():

def part(s, s_part):
    i0 = s.find(s_part)
    i1 = i0 + len(s_part)
    return (s[:i0], s[i0:i1], s[i1:])

@dalle noted that '#' can appear inside a string. It's not that easy to handle this case correctly, so I just ignored it, but I should have said something.

If your input file has simple enough rules for quoted strings, this isn't hard. It would be hard if you accepted any legal Python quoted string, because there are single-quoted, double-quoted, multiline quotes with a backslash escaping the end-of-line, triple quoted strings (using either single or double quotes), and even raw strings! The only possible way to correctly handle all that would be a complicated state machine.

But if we limit ourselves to just a simple quoted string, we can handle it with a simple state machine. We can even allow a backslash-quoted double quote inside the string.

c_backslash = '\\'
c_dquote = '"'
c_comment = '#'


def chop_comment(line):
    # a little state machine with two state varaibles:
    in_quote = False  # whether we are in a quoted string right now
    backslash_escape = False  # true if we just saw a backslash

    for i, ch in enumerate(line):
        if not in_quote and ch == c_comment:
            # not in a quote, saw a '#', it's a comment.  Chop it and return!
            return line[:i]
        elif backslash_escape:
            # we must have just seen a backslash; reset that flag and continue
            backslash_escape = False
        elif in_quote and ch == c_backslash:
            # we are in a quote and we see a backslash; escape next char
            backslash_escape = True
        elif ch == c_dquote:
            in_quote = not in_quote

    return line

I didn't really want to get this complicated in a question tagged "beginner" but this state machine is reasonably simple, and I hope it will be interesting.

This is the shortest possible form:

for line in open(filename):
  if line.startswith('#'):
    continue
  # PROCESS LINE HERE

The startswith() method on a string returns True if the string you call it on starts with the string you passed in.

While this is okay in some circumstances like shell scripts, it has two problems. First, it doesn't specify how to open the file. The default mode for opening a file is 'r', which means 'read the file in binary mode'. Since you're expecting a text file it is better to open it with 'rt'. Although this distinction is irrelevant on UNIX-like operating systems, it's important on Windows (and on pre-OS X Macs).

The second problem is the open file handle. The open() function returns a file object, and it's considered good practice to close files when you're done with them. To do that, call the close() method on the object. Now, Python will probably do this for you, eventually; in Python objects are reference-counted, and when an object's reference count goes to zero it gets freed, and at some point after an object is freed Python will call its destructor (a special method called __del__). Note that I said probably: Python has a bad habit of not actually calling the destructor on objects whose reference count drops to zero shortly before the program finishes. I guess it's in a hurry!

For short-lived programs like shell scripts, and particularly for file objects, this doesn't matter. Your operating system will automatically clean up any file handles left open when the program finishes. But if you opened the file, read the contents, then started a long computation without explicitly closing the file handle first, Python is likely to leave the file handle open during your computation. And that's bad practice.

This version will work in any 2.x version of Python, and fixes both the problems I discussed above:

f = open(file, 'rt')
for line in f:
  if line.startswith('#'):
    continue
  # PROCESS LINE HERE
f.close()

This is the best general form for older versions of Python.

As suggested by steveha, using the "with" statement is now considered best practice. If you're using 2.6 or above you should write it this way:

with open(filename, 'rt') as f:
  for line in f:
    if line.startswith('#'):
      continue
    # PROCESS LINE HERE

The "with" statement will clean up the file handle for you.

In your question you said "lines that start with #", so that's what I've shown you here. If you want to filter out lines that start with optional whitespace and then a '#', you should strip the whitespace before looking for the '#'. In that case, you should change this:

    if line.startswith('#'):

to this:

    if line.lstrip().startswith('#'):

In Python, strings are immutable, so this doesn't change the value of line. The lstrip() method returns a copy of the string with all its leading whitespace removed.

I'm coming at this late, but the problem of handling shell style (or python style) # comments is a very common one.

I've been using some code almost everytime I read a text file.
Problem is that it doesn't handle quoted or escaped comments properly. But it works for simple cases and is easy.

for line in whatever:
    line = line.split('#',1)[0].strip()
    if not line:
        continue
    # process line

A more robust solution is to use shlex:

import shlex
for line in instream:
    lex = shlex.shlex(line)
    lex.whitespace = '' # if you want to strip newlines, use '\n'
    line = ''.join(list(lex))
    if not line:
        continue
    # process decommented line

This shlex approach not only handles quotes and escapes properly, it adds a lot of cool functionality (like the ability to have files source other files if you want). I haven't tested it for speed on large files, but it is zippy enough of small stuff.

The common case when you're also splitting each input line into fields (on whitespace) is even simpler:

import shlex
for line in instream:
    fields = shlex.split(line, comments=True)
    if not fields:
        continue
    # process list of fields 

you can use startswith()

eg

for line in open("file"):
    li=line.strip()
    if not li.startswith("#"):
        print line.rstrip()