Python：获取字符串索引的行号和列号？

Question

Say I have a text file I'm operating on. Something like this (hopefully this isn't too unreadable):

data_raw = open('my_data_file.dat').read()
matches = re.findall(my_regex, data_raw, re.MULTILINE)
for match in matches:
    try:
        parse(data_raw, from_=match.start(), to=match.end())
    except Exception:
        print("Error parsing data starting on line {}".format(what_do_i_put_here))
        raise

Notice in the exception handler there's a certain variable named what_do_i_put_here . My question is: how can I assign to that name so that my script will print the line number that contains the start of the 'bad region' I'm trying to work with? I don't mind re-reading the file, I just don't know what I'd do...

Answer 1

I wrote this. It's untested and inefficient but it does help my exception message be a little clearer:

def coords_of_str_index(string, index):
    """Get (line_number, col) of `index` in `string`."""
    lines = string.splitlines(True)
    curr_pos = 0
    for linenum, line in enumerate(lines):
        if curr_pos + len(line) > index:
            return linenum + 1, index-curr_pos
        curr_pos += len(line)

I haven't even tested to see if the column number is vaguely accurate. I failed to abide by YAGNI

Answer 2

Here's something a bit cleaner, and in my opinion easier to understand than your own answer:

def index_to_coordinates(s, index):
    """Returns (line_number, col) of `index` in `s`."""
    if not len(s):
        return 1, 1
    sp = s[:index+1].splitlines(keepends=True)
    return len(sp), len(sp[-1])

It works essentially the same way as your own answer, but by utilizing string slicing splitlines() actually calculates all the information you need for you without the need for any post processing.

Using the keepends=True is necessary to give correct column counts for end of line characters.

The only extra problem is the edge case of an empty string, which can easily be handled by a guard-clause.

I tested it in Python 3.8, but it probably works correctly after about version 3.4 (in some older versions len() counts code units instead of code points, and I assume it would break for any string containing characters outside of the BMP)

Answer 3

Column indexing starts with 0 so you need to extract 1 from len(sp[-1]) at the very end of your code to get the correct column value. Also, I'd perhaps return None (instead of "1.1" - which is also incorrect since it should be "1.0"...) if the lenght of string is 0 or if the string is too short to fit the index. Otherwise, it's an excellent and elegant solution Tim.

def index_to_coordinates(txt:str, index:int) -> str:
    """Returns 'line.column' of index in 'txt'."""
    if not txt or len(txt)-1 < index:
        return None
    sp = txt[:index+1].splitlines(keepends=True)
    return (f"{len(sp)}.{len(sp[-1])-1}")