39

So far pandas read through all my CSV files without any problem, however now there seems to be a problem..

When doing:

df = pd.read_csv(r'path to file', sep=';')

I get:

OSError Traceback (most recent call last) in () ----> 1 df = pd.read_csv(r'path Übersicht\Input\test\test.csv', sep=';')

c:\program files\python36\lib\site-packages\pandas\io\parsers.py in parser_f(filepath_or_buffer, sep, delimiter, header, names, index_col, usecols, squeeze, prefix, mangle_dupe_cols, dtype, engine, converters, true_values, false_values, skipinitialspace, skiprows, nrows, na_values, keep_default_na, na_filter, verbose, skip_blank_lines, parse_dates, infer_datetime_format, keep_date_col, date_parser, dayfirst, iterator, chunksize, compression, thousands, decimal, lineterminator, quotechar, quoting, escapechar, comment, encoding, dialect, tupleize_cols, error_bad_lines, warn_bad_lines, skipfooter, skip_footer, doublequote, delim_whitespace, as_recarray, compact_ints, use_unsigned, low_memory, buffer_lines, memory_map, float_precision) 703 skip_blank_lines=skip_blank_lines) 704 --> 705 return _read(filepath_or_buffer, kwds) 706 707 parser_f.name = name

c:\program files\python36\lib\site-packages\pandas\io\parsers.py in _read(filepath_or_buffer, kwds) 443 444 # Create the parser. --> 445 parser = TextFileReader(filepath_or_buffer, **kwds) 446 447 if chunksize or iterator:

c:\program files\python36\lib\site-packages\pandas\io\parsers.py in init(self, f, engine, **kwds) 812 self.options['has_index_names'] = kwds['has_index_names'] 813 --> 814 self._make_engine(self.engine) 815 816 def close(self):

c:\program files\python36\lib\site-packages\pandas\io\parsers.py in _make_engine(self, engine) 1043 def _make_engine(self, engine='c'): 1044 if engine == 'c': -> 1045 self._engine = CParserWrapper(self.f, **self.options) 1046 else: 1047 if engine == 'python':

c:\program files\python36\lib\site-packages\pandas\io\parsers.py in init(self, src, **kwds) 1682 kwds['allow_leading_cols'] = self.index_col is not False 1683 -> 1684 self._reader = parsers.TextReader(src, **kwds) 1685 1686 # XXX

pandas_libs\parsers.pyx in pandas._libs.parsers.TextReader.cinit()

pandas_libs\parsers.pyx in pandas._libs.parsers.TextReader._setup_parser_source()

OSError: Initializing from file failed

Other files in the same folder that are XLS files can be accessed without an issue.

When using the Python library like so:

import csv
file = csv.reader(open(r'pathtofile')) 

for row in file:
    print(row)
    break

df = pd.read_csv(file, sep=';')

the file is being loaded and the first line is printed. However I get:

ValueError: Invalid file path or buffer object type:

Probably because I can't use read_csv this way...

How to get the first pandas function to work? The csv does not contain any special characters except German ones. The filesize is 10MB.

Improve this question
3
  • 2
    It looks like some problem in r'path to file', is possible more specify? Commented May 27, 2018 at 13:02
  • The path works perfectly fine for xls files though... df = pd.read_csv(r'U:\folder\Input\test.csv', sep=';') Commented May 27, 2018 at 13:17
  • In my opinion is possible copy file to same folder as .py script and then try df = df.read_csv('file.csv', sep=';') first... But if pd.read_excel working then hard to find problem... Commented May 27, 2018 at 13:23

17 Answers 17

Reset to default
61 
import pandas as pd
pd.read_csv("your_file.txt", engine='python')

Try this. It totally worked for me.

Improve this answer
Sign up to request clarification or add additional context in comments.

5 Comments

This got rid of the error, but the result was not a normal Pandas dataframe. Specifically, the column headers and row indices were gone.
This helped me identify the actual error. Doing this the error changed to a Permission denied error, and then I realized I was performing a read_csv on a directory.
Helped me as well, file permission was no read, no write! odd!
But what's the reason why?
This solved it but also, in combination with the answers below, made me realize that one of the characters in the filename was greek instead of latin (this character looks the same in both)!
55

I ran into a similar problem. It turned out the CSV I had downloaded had no permissions at all. The error message from pandas did not point this out, making it hard to debug.

Check that your file have read permissions

Improve this answer

1 Comment

According to github #23784: read_csv throws wrong exception on permissions issue this has been fixed but not yet released, planned for 1.1. Please reconfirm once 1.1 is released.
22

I had the same issue, you should check your permissions.

After chmod 644 file.csv it worked well.

Improve this answer

1 Comment

Thanks! It was the permission, but no explicit error message about it.
6

Same problem when I was trying to load files with Japanese filenames. 

import pandas as pd
result = pd.read_csv('./result/けっこう.csv')

OSError: Initializing from file failed'

Then I added an argument engine="python". 

result = pd.read_csv('./result/けっこう.csv', engine="python")

It worked for me.

Improve this answer

1 Comment

And where did you add that argument? Please provide a complete answer.
4

I find the same problem under Win10 OS when I try to read a csv file which name is in Chinese. Then there's no problem any more after I rename my file into EN. Maybe you should ensure your full csv file path in EN.

OS:Windows 10; Python version:3.6.5; Ipython:7.0.1; Pandas:0.23.0

First, when use

import pandas as pd
answer_df = pd.read_csv('./答案.csv')

my ipython notebook raise 'OSError: Initializing from file failed'.

Then I rename my file into answers.csv

import pandas as pd
answer_df = pd.read_csv('./answers.csv')

everythings' OK.

May to help you.

Improve this answer

1 Comment

Thanks, I realized that one of the characters in my filename was not latin although it looked like it!
2

Ran into the same issue on windows. Tried to use to the solution provided by Dan Lee, but was getting abnormal results with rows and columns. I am not sure whether this was because of the Japanese characters in my csv files or not but by clearly defining the encoding format solved the problem for me.

import pandas as pd
pd.read_csv("your_file.txt", engine='python', encoding = "utf-8-sig")
Improve this answer

Comments

1

I assume that your csv file is in the same place (root). If you just want to the csv file to be read, and get the result that will show as text in your console, just do this

import csv
with open('your_file.csv', 'r') as csvFile:
    reader = csv.reader(csvFile)
    for row in reader:
        print(row)
csvFile.close()

note: the code is for Python 3, if you use Python 2 the print syntax use without brackets. Hope this will help you

Improve this answer

Comments

1

You can try using os.path.join() to build your path: 

import os

rpath = os.path.join('U:','folder','Input','test.csv')
df = pd.read_csv(rpath, sep=';')

To traverse path based on your parent directory, you can use:

os.path.pardir
Improve this answer

Comments

1

In my case, I was entering incorrectly the path.

I was working with a dataset downloaded from Kaggle as a zip. The structure of downloads was: main.zip/subfiles.zip. I unzipped the main but the solution was to unzip the subfiles.zip and then my wanted file was within the subfiles.zip.

So the path would be main/subfiles/wantedfile.csv and this was my simple fix.

Improve this answer

Comments

0

Same problem different solution here.

I had previously tried to load my file in excel, and excel had crashed but must have retained some file lock because once I force-quit excel, it loaded as expected.

Improve this answer

1 Comment

don't understand the downvote here? just offering an alternative solution in case the first one doesn't work for someone, as it didn't for me...
0

pandas read_csv OSError: Initializing from file failed

We could try chmod 600 file.csv.

Improve this answer

Comments

0

just change the permessions of csv file, It would work

chmod 750 filename.csv (in command line)

or

!chmod 750 filename.csv (in jupyter notebook)

Improve this answer

Comments

0

I had the same issue and since I have just installed another package, I realized that the problem could be due to that recent installation which may made some modifications to the pandas package. I tried to remove pandas and reinstall it and it worked perfectly as before.

First remove pandas:

conda remove pandas

Then reinstall it using: 

conda install -c anaconda pandas
Improve this answer

Comments

0

I was facing the same issue from hours and this time I tried to the zip file in reference from where I downloaded the Dataset and it works-

df=pd.read_csv(r'C:\Users\ADMIN\news.csv')-Initializing from file failed df=pd.read_csv(r'C:\Users\ADMIN\news.zip')-Worked nicely!!

Improve this answer

Comments

0

If your are facing this problem in jupyter notebook, then have you to upload the csv file in notebook and then check if your problem is solved or not. i hope this will work.

Improve this answer

Comments

0

OSError: Initializing from file failed

By this error Python indicates the file is not readable- either, file permission, file error, incorrect path. etc.

My code giving this error :

DataFrame=pd.read_csv("C:\\Users\\arindam\\Documents\\#Books & Docs\\#ML-DATA\\train.csv")

Fixed by this :

DataFrame=pd.read_csv("C:\\Users\\arindam\\Documents\\train.csv")

Make sure the path is readable by python.

Improve this answer

Comments

-4

may be the cause of the problem is that the version of pandas is too low, update the pandas to 1.0.1, it work. use the command as follows:

pip install -U pandas
Improve this answer

Comments

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.