22

I am experimenting NLTK package using Python. I tried to downloaded NLTK using nltk.download(). I got this kind of error message. How to solve this problem? Thanks.

The system I used is Ubuntu installed under VMware. The IDE is Spyder.

enter image description here

After using nltk.download('all'), it can download some packages, but it gets error message when downloading oanc_masc

enter image description here

Zulu
  • 8,765
  • 9
  • 49
  • 56
user288609
  • 12,465
  • 26
  • 85
  • 127

6 Answers6

22

To download a particular dataset/models, use the nltk.download() function, e.g. if you are looking to download the punkt sentence tokenizer, use:

$ python3
>>> import nltk
>>> nltk.download('punkt')

If you're unsure of which data/model you need, you can start out with the basic list of data + models with:

>>> import nltk
>>> nltk.download('popular')

It will download a list of "popular" resources.

Ensure that you've the latest version of NLTK because it's always improving and constantly maintain:

$ pip install --upgrade nltk

EDITED

In case anyone is avoiding errors from downloading larger datasets from nltk, from https://stackoverflow.com/a/38135306/610569

$ rm /Users/<your_username>/nltk_data/corpora/panlex_lite.zip
$ rm -r /Users/<your_username>/nltk_data/corpora/panlex_lite
$ python

>>> import nltk
>>> dler = nltk.downloader.Downloader()
>>> dler._update_index()
>>> dler._status_cache['panlex_lite'] = 'installed' # Trick the index to treat panlex_lite as it's already installed.
>>> dler.download('popular')

And if anyone wants to find nltk_data directory, see https://stackoverflow.com/a/36383314/610569

And to config nltk_data path, see https://stackoverflow.com/a/22987374/610569

alvas
  • 115,346
  • 109
  • 446
  • 738
  • 1
    thanks for the reply. I tried nltk.download('all'), it successfully proceeded with downloading some packages, but it got stuck when downloading sth related to oasc_masc, I included the related screenshot in the original post. – user288609 Dec 26 '14 at 18:48
  • 1
    what is your nltk version? what is the output of this on your terminal: `python -c "import nltk; print nltk.__version__"`? – alvas Dec 26 '14 at 18:50
  • Hi there @alvas I'm having similar issues using nltk.download('all') on Ubuntu, except I get HTTP Error 404: Not Found in both IDLE and command line. My NLTK version is 2.0b9. Do you have any idea what might be going on? – Joansy Dec 05 '15 at 23:36
  • @Joansy, Please update your NLTK. `sudo pip install nltk` or `sudo apt-get install python-nltk`. Once it's updated the problem should resolve itself. Otherwise, you would have to set the url manually. Try updating NLTK first, if it doesn't work, then come back again =) – alvas Dec 06 '15 at 00:20
8

From command line, after importing nltk, try

nltk.download('popular', halt_on_error=False)

After an error it will ask to retry broken package, just decline with n and it will continue with proper packages.

alvas
  • 115,346
  • 109
  • 446
  • 738
tolgayilmaz
  • 3,987
  • 2
  • 19
  • 19
  • I had several `UnicodeDecodeError`, and I had to launch this command several times in order to download everything, but it worked in the end. Thanks ! – CoMartel May 31 '17 at 08:30
1

a) in OSX either run

sudo /Applications/Python\ 3.6/Install\ Certificates.command

b) switch to admin user (the one you have set up with administrator privileges)

and type at command line:

/Applications/Python\ 3.6/Install\ Certificates.command

Notes:

  • "\" are necessary because they escape blank characters in file names.
  • This procedure worked if you have python 3.6 installed, otherwise change it in order to match your install python version... for this execute:

ls /Applications

and look at the python directory name you have there.

UpAndAdam
  • 4,515
  • 3
  • 28
  • 46
Alexandre
  • 11
  • 1
1

An easy(hard) way to get over this error is to do the process manually. Just go to the website https://www.nltk.org/nltk_data/ and download the required zip file and extract the contents.

In Windows, go to user/AppData/local/Programs/Python/Python(version)/lib and create a folder nltk_data. Then create the respective folder. As an example, for 'punkt' create the folder tokenizers and add the folder 'punkt' inside the extracted folder to it. This info is mostly given by the terminal itself.

Run your program. Cheers!

EDIT 1: Of course, downloading all files can be time-consuming, but it's the only option if the "urlopen error" persists.

EDIT 2 It is also mostly your router or network at fault that you are not able to download nltk files. Try changing your network and that should help.

Hyperion
  • 11
  • 2
0

First download it manually from https://www.nltk.org/nltk_data/ using an id and use a vpn for download and save it in, as shown path in error and then append /nltk_data/corpora folder and save your downloaded file in unzip format, then turn off VPN and again run same code, it works.

Glauco
  • 1,385
  • 2
  • 10
  • 20
vijay mitta
  • 1
-3

I had this error:

Resource punkt not found. Please use the NLTK Downloader to obtain the resource: import nltk nltk.download('punkt')

When I tried to solve by writing: 

import nltk

nltk.download()

my computer shut downs suddenly and anaconda also closed. When I try to open it always shows an error.

I solved the problem by writing:

import nltk

nltk.download('punkt')
double-beep
  • 5,031
  • 17
  • 33
  • 41
HaticeKübraKılınç
  • 1
  • This probably won't help. His problem was unable to nltk.download('all'), more likely only unable to nltk.download('oanc_masc') – Jzou Jan 31 '20 at 22:01