我该怎么做才能使python脚本可以从任何目录运行：脚本文件不必与.csv文件位于同一目录中？

Question

Currently, I have created a code that makes graphs from data in .csv files. However, I can only run the code if that code is present in the folder with the csv files. How can I make the the script file so that it doesn't have to be in the same directory as the .csv files.

Answer 1

Assuming you mean to include a fixed CSV file with your code, store an absolute path based on the script path:

HERE = os.path.dirname(os.path.abspath(__file__))

csv_filename = open(os.path.join(HERE, 'somefile.csv')

__file__ is the filename of the current module or script, os.path.dirname(__file__) is the directory the module resides in. For scripts, __file__ can be a relative pathname, so we use os.path.abspath() to turn that into an absolute path.

This means you can run your script from anywhere.

If you meant to make your script work with arbitrary CSV input files, use command line options:

import argparse

if __name__ == '__main__':
    parser = argparse.ArgumentParser('CSV importer')
    parser.add_argument('csvfile', type=argparse.FileType('w'),
                        default='somedefaultfilename.csv')
    options = parser.parse_args()
    import_function(options.csvfile)

where csvfile will be an open file object, so your import_function() can just do:

def import_function(csvfile):
    with csvfile:
        reader = csv.reader(csvfile)
        for row in reader:
            # etc.    

Answer 2

If you don't plan on moving around the csv files too much, the best answer is to hard code the absolute path to the csv folder into the script.

import os

csvdir = "/path/to/csv/dir"
csvfpath = os.path.join(csvdir, "myfile.csv")
csvfile = open(csvfpath)

You can also use a command line parser like argparse to let the user easily change the path to the csv files.

Using Martijn Pieters's solution will work only if you are going to be moving the folder containing both the script and csv files around. However in that case, you may as well just use relative paths to the csv files.