Question

I need to add a single line to the first line of a text file and it looks like the only options available to me are more lines of code than I would expect from python. Something like this:

f = open('filename','r')
temp = f.read()
f.close()

f = open('filename', 'w')
f.write("#testfirstline")

f.write(temp)
f.close()

Is there no easier way? Additionally, I see this two-handle example more often than opening a single handle for reading and writing ('r+') - why is that?

Was it helpful?

Solution

Python makes a lot of things easy and contains libraries and wrappers for a lot of common operations, but the goal is not to hide fundamental truths.

The fundamental truth you are encountering here is that you generally can't prepend data to an existing flat structure without rewriting the entire structure. This is true regardless of language.

There are ways to save a filehandle or make your code less readable, many of which are provided in other answers, but none change the fundamental operation: You must read in the existing file, then write out the data you want to prepend, followed by the existing data you read in.

By all means save yourself the filehandle, but don't go looking to pack this operation into as few lines of code as possible. In fact, never go looking for the fewest lines of code -- that's obfuscation, not programming.

OTHER TIPS

I would stick with separate reads and writes, but we certainly can express each more concisely:

Python2:

with file('filename', 'r') as original: data = original.read()
with file('filename', 'w') as modified: modified.write("new first line\n" + data)

Python3:

with open('filename', 'r') as original: data = original.read()
with open('filename', 'w') as modified: modified.write("new first line\n" + data)

Note: file() function is not available in python3.

Other approach:

with open("infile") as f1:
    with open("outfile", "w") as f2:
        f2.write("#test firstline")
        for line in f1:
            f2.write(line)

or a one liner:

open("outfile", "w").write("#test firstline\n" + open("infile").read())

Thanks for the opportunity to think about this problem :)

Cheers

with open("file", "r+") as f: s = f.read(); f.seek(0); f.write("prepend\n" + s)

You can save one write call with this:

f.write('#testfirstline\n' + temp)

When using 'r+', you would have to rewind the file after reading and before writing.

Here's a 3 liner that I think is clear and flexible. It uses the list.insert function, so if you truly want to prepend to the file use l.insert(0, 'insert_str'). When I actually did this for a Python Module I am developing, I used l.insert(1, 'insert_str') because I wanted to skip the '# -- coding: utf-8 --' string at line 0. Here is the code.

f = open(file_path, 'r'); s = f.read(); f.close()
l = s.splitlines(); l.insert(0, 'insert_str'); s = '\n'.join(l)
f = open(file_path, 'w'); f.write(s); f.close()

This does the job without reading the whole file into memory, though it may not work on Windows

def prepend_line(path, line):
    with open(path, 'r') as old:
        os.unlink(path)
        with open(path, 'w') as new:
            new.write(str(line) + "\n")
            shutil.copyfileobj(old, new)

One possibility is the following:

import os
open('tempfile', 'w').write('#testfirstline\n' + open('filename', 'r').read())
os.rename('tempfile', 'filename')

If you wish to prepend in the file after a specific text then you can use the function below.

def prepend_text(file, text, after=None):
    ''' Prepend file with given raw text '''
    f_read = open(file, 'r')
    buff = f_read.read()
    f_read.close()
    f_write = open(file, 'w')
    inject_pos = 0
    if after:
        pattern = after
        inject_pos = buff.find(pattern)+len(pattern)
    f_write.write(buff[:inject_pos] + text + buff[inject_pos:])
    f_write.close()

So first you open the file, read it and save it all into one string. Then we try to find the character number in the string where the injection will happen. Then with a single write and some smart indexing of the string we can rewrite the whole file including the injected text now.

Am I not seeing something or couldn't we just use a buffer large-enough to read-in the input file in parts (instead of the whole content) and with this buffer traverse the file while it is open and keep exchanging file<->buffer contents?

This seems much more efficient (for big files especially) than reading the whole content in memory, modifying it in memory and writing it back to the same file or (even worse) a different one. Sorry that now I don't have time to implement a sample snippet, I'll get back to this later, but maybe you get the idea.

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top