{
  "id": 24081,
  "title": "how can I download the dataset on a remote instance?",
  "url": "/competitions/bosch-production-line-performance/discussion/24081",
  "author_name": "",
  "post_date": "2016-10-04T16:21:20Z",
  "votes": 1,
  "comment_count": 18,
  "views": 1202,
  "content": "<p>Does anyone have a method to download datasets from a remote instance that works for them?</p>\n\n<p>I've been trying the methods posted on other Kaggle forums with no luck. I've resorted to rural internet service to sftp this dataset and it's uploading at ~50KB/s -- so maybe it'll be done in a week after restarts?</p>\n\n<p>Surely other people have done this before for AWS instances. I've tried wget with a cookies file, Lynx, Lynx+wget and none of them work for me.</p>\n\n<p>Does Kaggle provide a direct means to download the files? Why don't they just set up security keys and be done with it? Where's the harm if my Kaggle profile held my public key?</p>",
  "messages": [
    {
      "id": "137748",
      "postDate": "10/04/2016 16:21:20",
      "content": "<p>Does anyone have a method to download datasets from a remote instance that works for them?</p>\n\n<p>I've been trying the methods posted on other Kaggle forums with no luck. I've resorted to rural internet service to sftp this dataset and it's uploading at ~50KB/s -- so maybe it'll be done in a week after restarts?</p>\n\n<p>Surely other people have done this before for AWS instances. I've tried wget with a cookies file, Lynx, Lynx+wget and none of them work for me.</p>\n\n<p>Does Kaggle provide a direct means to download the files? Why don't they just set up security keys and be done with it? Where's the harm if my Kaggle profile held my public key?</p>",
      "rawMarkdown": "Does anyone have a method to download datasets from a remote instance that works for them?\r\n\r\nI've been trying the methods posted on other Kaggle forums with no luck. I've resorted to rural internet service to sftp this dataset and it's uploading at ~50KB/s -- so maybe it'll be done in a week after restarts?\r\n\r\nSurely other people have done this before for AWS instances. I've tried wget with a cookies file, Lynx, Lynx+wget and none of them work for me.\r\n\r\nDoes Kaggle provide a direct means to download the files? Why don't they just set up security keys and be done with it? Where's the harm if my Kaggle profile held my public key?",
      "votes": null
    },
    {
      "id": "137750",
      "postDate": "10/04/2016 16:51:14",
      "content": "<p>There are a few methods. You can download and put the data to Dropbox then use wget to download the Dropbox shared URL. For example, given Dropbox URL </p>\n\n<p><a href=\"https://www.dropbox.com/s/7ob8wvcdd5k1bnf/helloworld.PNG?dl=0\">https://www.dropbox.com/s/7ob8wvcdd5k1bnf/helloworld.PNG?dl=0</a>\nyou can run </p>\n\n<p><code>wget https://www.dropbox.com/s/7ob8wvcdd5k1bnf/helloworld.PNG</code>\nNote that ?dl=0 is removed.</p>\n\n<p>Also you can convert csv data to binary data and put them into dropbox for faster upload/download.</p>",
      "rawMarkdown": "There are a few methods. You can download and put the data to Dropbox then use wget to download the Dropbox shared URL. For example, given Dropbox URL \r\n\r\nhttps://www.dropbox.com/s/7ob8wvcdd5k1bnf/helloworld.PNG?dl=0\r\nyou can run \r\n\r\n`wget https://www.dropbox.com/s/7ob8wvcdd5k1bnf/helloworld.PNG`\r\nNote that ?dl=0 is removed.\r\n\r\nAlso you can convert csv data to binary data and put them into dropbox for faster upload/download.",
      "votes": null
    },
    {
      "id": "137751",
      "postDate": "10/04/2016 16:54:38",
      "content": "<p>Check <a href=\"https://www.kaggle.com/c/belkin-energy-disaggregation-competition/forums/t/5118/downloading-data-via-wget/112602\">this</a> thread.\nI've used the solution described in the first four posts in the past. It should work.</p>",
      "rawMarkdown": "Check [this][1] thread.\r\nI've used the solution described in the first four posts in the past. It should work.\r\n\r\n  [1]: https://www.kaggle.com/c/belkin-energy-disaggregation-competition/forums/t/5118/downloading-data-via-wget/112602",
      "votes": null
    },
    {
      "id": "137758",
      "postDate": "10/04/2016 21:00:49",
      "content": "<p>Thanks @Banana Kitty. I have downloaded the Bosch dataset to to my computer, it's the upload rate that's my problem. So whether I upload to Dropbox or my remote instance will take the same amount of time. I would need a direct way of getting the datasets to Dropbox for that to work.</p>\n\n<p>@Rohan, thanks for the link. That's one of them that I tried and I couldn't get any of the methods to work.  On my browser I logged into Kaggle, started the download and accepted the terms. Then took the cookie info, uploaded it to my remote and tried the wget method. No luck...</p>",
      "rawMarkdown": "Thanks @Banana Kitty. I have downloaded the Bosch dataset to to my computer, it's the upload rate that's my problem. So whether I upload to Dropbox or my remote instance will take the same amount of time. I would need a direct way of getting the datasets to Dropbox for that to work.\r\n\r\n@Rohan, thanks for the link. That's one of them that I tried and I couldn't get any of the methods to work.  On my browser I logged into Kaggle, started the download and accepted the terms. Then took the cookie info, uploaded it to my remote and tried the wget method. No luck...",
      "votes": null
    },
    {
      "id": "137765",
      "postDate": "10/04/2016 22:06:06",
      "content": "<p>This failed to work for me too, initially. I fixed it by changing <strong>%2F</strong> in the third column of the cookies file to <strong>/</strong></p>\n\n<p>e.g.</p>\n\n<pre><code>.kaggle.com TRUE    %2F FALSE   1467890123 ...\n</code></pre>\n\n<p>to</p>\n\n<pre><code>.kaggle.com TRUE    /   FALSE   1467890123 ...\n</code></pre>",
      "rawMarkdown": "This failed to work for me too, initially. I fixed it by changing **%2F** in the third column of the cookies file to **/**\r\n\r\ne.g.\r\n\r\n    .kaggle.com\tTRUE\t%2F\tFALSE\t1467890123 ...\r\n\r\nto\r\n\r\n    .kaggle.com\tTRUE\t/\tFALSE\t1467890123 ...",
      "votes": null
    },
    {
      "id": "137774",
      "postDate": "10/05/2016 00:06:14",
      "content": "<p>Thanks @James. That didn't turn out to be my problem.</p>\n\n<p>Actually I don't know what my problem was. I tried it again and it worked. Using the Chrome <em>cookie.txt export</em> extension I did these steps,</p>\n\n<ol>\n<li>on the download page, right click on one file to download and copy the link</li>\n<li>open a new tab and paste the link into it </li>\n<li>after the download starts, pause it from the lower download bar</li>\n<li>click on the cookie.txt export extension icon on the upper right bar</li>\n<li>copy the text from the pop up and follow the instructions</li>\n<li>of course, make sure the cookies.txt file is in the current directory when you run wget</li>\n<li>repeat for the other files, changing the file name</li>\n</ol>\n\n<p>The files will be in a sequence of folders starting at www.kaggle.com</p>",
      "rawMarkdown": "Thanks @James. That didn't turn out to be my problem.\r\n\r\nActually I don't know what my problem was. I tried it again and it worked. Using the Chrome *cookie.txt export* extension I did these steps,\r\n\r\n 1. on the download page, right click on one file to download and copy the link\r\n 2. open a new tab and paste the link into it \r\n 3. after the download starts, pause it from the lower download bar\r\n 4. click on the cookie.txt export extension icon on the upper right bar\r\n 5. copy the text from the pop up and follow the instructions\r\n 6. of course, make sure the cookies.txt file is in the current directory when you run wget\r\n 7. repeat for the other files, changing the file name\r\n\r\nThe files will be in a sequence of folders starting at www.kaggle.com",
      "votes": null
    },
    {
      "id": "137778",
      "postDate": "10/05/2016 01:46:16",
      "content": "<p>You can upload files using Cyberduck. <a href=\"http://cyberduck.en.softonic.com/mac\">http://cyberduck.en.softonic.com/mac</a></p>",
      "rawMarkdown": "You can upload files using Cyberduck. http://cyberduck.en.softonic.com/mac",
      "votes": null
    },
    {
      "id": "137787",
      "postDate": "10/05/2016 03:51:50",
      "content": "<p>Hi r3m,</p>\n\n<p>Did you include https? Not sure if it will make any difference for you.</p>\n\n<p>Good luck.</p>",
      "rawMarkdown": "Hi r3m,\r\n\r\nDid you include https? Not sure if it will make any difference for you.\r\n\r\nGood luck.",
      "votes": null
    },
    {
      "id": "137793",
      "postDate": "10/05/2016 04:52:10",
      "content": "<p>To download to an Amazon EC2 node, I just use <code>links</code> on the remote node (I use ubuntu-based AMIs).  You don't <em>have to</em> supply a file with the cookies, you can just enter your login username and password from within <code>links</code>.  (Indeed, I don't recall ever actually getting the cookie-file approach to work myself.)  </p>\n\n<p>It can be a bit confusing because the site is not exactly designed to be read in a text-only browser, so you may fumble a bit at first with the login and password fields.  Also, you will probably want to get past the &quot;accepting the competition rules&quot; dialog on your local machine before you attempt the download through <code>links</code>.  But beyond that it has always worked for me.</p>",
      "rawMarkdown": "To download to an Amazon EC2 node, I just use `links` on the remote node (I use ubuntu-based AMIs).  You don't *have to* supply a file with the cookies, you can just enter your login username and password from within `links`.  (Indeed, I don't recall ever actually getting the cookie-file approach to work myself.)  \r\n\r\nIt can be a bit confusing because the site is not exactly designed to be read in a text-only browser, so you may fumble a bit at first with the login and password fields.  Also, you will probably want to get past the \"accepting the competition rules\" dialog on your local machine before you attempt the download through `links`.  But beyond that it has always worked for me.",
      "votes": null
    },
    {
      "id": "137794",
      "postDate": "10/05/2016 04:52:33",
      "content": "<p>Hi @Banana Kitty. Yeah I did. Actually I got it working (see my previous post). Thanks!</p>",
      "rawMarkdown": "Hi @Banana Kitty. Yeah I did. Actually I got it working (see my previous post). Thanks!",
      "votes": null
    },
    {
      "id": "137795",
      "postDate": "10/05/2016 04:58:45",
      "content": "<p>@Tod. Yeah I couldn't get past the login with lynx. I tried a few times. Mostly my fault there since I don't know lynx well. </p>\n\n<p>Cookies/wget worked after I opened a download session on Chrome (using the direct link to the file) in my local computer, paused the download, and used wget on the remote. Maybe having the open session did something??</p>",
      "rawMarkdown": "Tod. Yeah I couldn't get past the login with lynx. I tried a few times. Mostly my fault there since I don't know lynx well. \r\n\r\nCookies/wget worked after I opened a download session on Chrome (using the direct link to the file) in my local computer, paused the download, and used wget on the remote. Maybe having the open session did something??",
      "votes": null
    },
    {
      "id": "137796",
      "postDate": "10/05/2016 05:02:22",
      "content": "<p>@Scott. I needed a solution running on my remote machine, so Cyberduck wouldn't help me there. But it does look interesting. Thanks!</p>",
      "rawMarkdown": "Scott. I needed a solution running on my remote machine, so Cyberduck wouldn't help me there. But it does look interesting. Thanks!",
      "votes": null
    },
    {
      "id": "137799",
      "postDate": "10/05/2016 05:13:32",
      "content": "<p>@r3m I mean literally <code>links</code>, which is a text-based browser distinct from <code>lynx</code>.  I haven't attempted it with <code>lynx</code>, so it is possible that <code>lynx</code> is less agreeable for this task, I wouldn't know.  Sounds like you got it sorted anyway...</p>",
      "rawMarkdown": "r3m I mean literally `links`, which is a text-based browser distinct from `lynx`.  I haven't attempted it with `lynx`, so it is possible that `lynx` is less agreeable for this task, I wouldn't know.  Sounds like you got it sorted anyway...",
      "votes": null
    },
    {
      "id": "137800",
      "postDate": "10/05/2016 05:22:45",
      "content": "<p>Ohh... sorry @Tod, I just assumed that's the way you spelled it. Never heard of it before.</p>",
      "rawMarkdown": "Ohh... sorry @Tod, I just assumed that's the way you spelled it. Never heard of it before.",
      "votes": null
    },
    {
      "id": "137836",
      "postDate": "10/05/2016 11:57:42",
      "content": "<p>If you have the datasets on one machine, you could send them to a remote machine, check the <code>scp</code> or <code>rsync</code> command.</p>\n\n<p>or you could use the remote machine to download the datasets directly. I recommend to use the <a href=\"https://github.com/floydwch/kaggle-cli\">unofficial kaggle command line tool</a>. It has some problems at the moment but you are able to download the datasets. </p>",
      "rawMarkdown": "If you have the datasets on one machine, you could send them to a remote machine, check the `scp` or `rsync` command.\r\n\r\nor you could use the remote machine to download the datasets directly. I recommend to use the [unofficial kaggle command line tool][1]. It has some problems at the moment but you are able to download the datasets. \r\n\r\n  [1]: https://github.com/floydwch/kaggle-cli",
      "votes": null
    },
    {
      "id": "137838",
      "postDate": "10/05/2016 12:01:31",
      "content": "<p>If you have the datasets on one machine, you could send them to a remote machine, check the <code>scp</code> or <code>rsync</code> command.</p>\n\n<p>or you could use the remote machine to download the datasets directly. I recommend to use the <a href=\"https://github.com/floydwch/kaggle-cli\">unofficial kaggle command line tool</a>. It has some problems at the moment but you are able to download the datasets. </p>",
      "rawMarkdown": "If you have the datasets on one machine, you could send them to a remote machine, check the `scp` or `rsync` command.\r\n\r\nor you could use the remote machine to download the datasets directly. I recommend to use the [unofficial kaggle command line tool][1]. It has some problems at the moment but you are able to download the datasets. \r\n\r\n  [1]: https://github.com/floydwch/kaggle-cli",
      "votes": null
    },
    {
      "id": "137839",
      "postDate": "10/05/2016 12:01:36",
      "content": "<p>If you have the datasets on one machine, you could send them to a remote machine, check the <code>scp</code> or <code>rsync</code> command.</p>\n\n<p>or you could use the remote machine to download the datasets directly. I recommend to use the <a href=\"https://github.com/floydwch/kaggle-cli\">unofficial kaggle command line tool</a>. It has some problems at the moment but you are able to download the datasets. </p>",
      "rawMarkdown": "If you have the datasets on one machine, you could send them to a remote machine, check the `scp` or `rsync` command.\r\n\r\nor you could use the remote machine to download the datasets directly. I recommend to use the [unofficial kaggle command line tool][1]. It has some problems at the moment but you are able to download the datasets. \r\n\r\n  [1]: https://github.com/floydwch/kaggle-cli",
      "votes": null
    },
    {
      "id": "138049",
      "postDate": "10/06/2016 20:10:22",
      "content": "<p>[quote=YiTang;137839]</p>\n\n<p>...or you could use the remote machine to download the datasets directly. I recommend to use the <a href=\"https://github.com/floydwch/kaggle-cli\">unofficial kaggle command line tool</a>. It has some problems at the moment but you are able to download the datasets. </p>\n\n<p>[/quote]</p>\n\n<p>Thanks @YiTang. Kaggle-CLI may be exactly what I'm looking for. </p>",
      "rawMarkdown": "[quote=YiTang;137839]\r\n\r\n...or you could use the remote machine to download the datasets directly. I recommend to use the [unofficial kaggle command line tool][1]. It has some problems at the moment but you are able to download the datasets. \r\n\r\n  [1]: https://github.com/floydwch/kaggle-cli\r\n\r\n[/quote]\r\n\r\nThanks @YiTang. Kaggle-CLI may be exactly what I'm looking for.",
      "votes": null
    },
    {
      "id": "138832",
      "postDate": "10/11/2016 03:44:05",
      "content": "<ol>\n<li>Get a Chrome plugin that lets you copy your cookies to text file. Save this as <code>cookies.txt</code> in your data folder.</li>\n<li>Right-click and copy the kaggle data url from the downloads page.</li>\n<li>Use <code>wget -x --load-cookies cookies.txt url</code></li>\n<li>It will save the download in a nested set of folders corresponding to the link backslashes. </li>\n</ol>",
      "rawMarkdown": "1. Get a Chrome plugin that lets you copy your cookies to text file. Save this as `cookies.txt` in your data folder.\r\n2. Right-click and copy the kaggle data url from the downloads page.\r\n3. Use `wget -x --load-cookies cookies.txt url`\r\n4. It will save the download in a nested set of folders corresponding to the link backslashes.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 137750,
      "author_name": "",
      "author_url": "",
      "post_date": "10/04/2016 16:51:14",
      "content": "<p>There are a few methods. You can download and put the data to Dropbox then use wget to download the Dropbox shared URL. For example, given Dropbox URL </p>\n\n<p><a href=\"https://www.dropbox.com/s/7ob8wvcdd5k1bnf/helloworld.PNG?dl=0\">https://www.dropbox.com/s/7ob8wvcdd5k1bnf/helloworld.PNG?dl=0</a>\nyou can run </p>\n\n<p><code>wget https://www.dropbox.com/s/7ob8wvcdd5k1bnf/helloworld.PNG</code>\nNote that ?dl=0 is removed.</p>\n\n<p>Also you can convert csv data to binary data and put them into dropbox for faster upload/download.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137751,
      "author_name": "rohanrao",
      "author_url": "",
      "post_date": "10/04/2016 16:54:38",
      "content": "<p>Check <a href=\"https://www.kaggle.com/c/belkin-energy-disaggregation-competition/forums/t/5118/downloading-data-via-wget/112602\">this</a> thread.\nI've used the solution described in the first four posts in the past. It should work.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137758,
      "author_name": "rshpeley",
      "author_url": "",
      "post_date": "10/04/2016 21:00:49",
      "content": "<p>Thanks @Banana Kitty. I have downloaded the Bosch dataset to to my computer, it's the upload rate that's my problem. So whether I upload to Dropbox or my remote instance will take the same amount of time. I would need a direct way of getting the datasets to Dropbox for that to work.</p>\n\n<p>@Rohan, thanks for the link. That's one of them that I tried and I couldn't get any of the methods to work.  On my browser I logged into Kaggle, started the download and accepted the terms. Then took the cookie info, uploaded it to my remote and tried the wget method. No luck...</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137765,
      "author_name": "jtrotman",
      "author_url": "",
      "post_date": "10/04/2016 22:06:06",
      "content": "<p>This failed to work for me too, initially. I fixed it by changing <strong>%2F</strong> in the third column of the cookies file to <strong>/</strong></p>\n\n<p>e.g.</p>\n\n<pre><code>.kaggle.com TRUE    %2F FALSE   1467890123 ...\n</code></pre>\n\n<p>to</p>\n\n<pre><code>.kaggle.com TRUE    /   FALSE   1467890123 ...\n</code></pre>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137774,
      "author_name": "rshpeley",
      "author_url": "",
      "post_date": "10/05/2016 00:06:14",
      "content": "<p>Thanks @James. That didn't turn out to be my problem.</p>\n\n<p>Actually I don't know what my problem was. I tried it again and it worked. Using the Chrome <em>cookie.txt export</em> extension I did these steps,</p>\n\n<ol>\n<li>on the download page, right click on one file to download and copy the link</li>\n<li>open a new tab and paste the link into it </li>\n<li>after the download starts, pause it from the lower download bar</li>\n<li>click on the cookie.txt export extension icon on the upper right bar</li>\n<li>copy the text from the pop up and follow the instructions</li>\n<li>of course, make sure the cookies.txt file is in the current directory when you run wget</li>\n<li>repeat for the other files, changing the file name</li>\n</ol>\n\n<p>The files will be in a sequence of folders starting at www.kaggle.com</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137778,
      "author_name": "liouscott",
      "author_url": "",
      "post_date": "10/05/2016 01:46:16",
      "content": "<p>You can upload files using Cyberduck. <a href=\"http://cyberduck.en.softonic.com/mac\">http://cyberduck.en.softonic.com/mac</a></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137787,
      "author_name": "",
      "author_url": "",
      "post_date": "10/05/2016 03:51:50",
      "content": "<p>Hi r3m,</p>\n\n<p>Did you include https? Not sure if it will make any difference for you.</p>\n\n<p>Good luck.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137793,
      "author_name": "tmwright",
      "author_url": "",
      "post_date": "10/05/2016 04:52:10",
      "content": "<p>To download to an Amazon EC2 node, I just use <code>links</code> on the remote node (I use ubuntu-based AMIs).  You don't <em>have to</em> supply a file with the cookies, you can just enter your login username and password from within <code>links</code>.  (Indeed, I don't recall ever actually getting the cookie-file approach to work myself.)  </p>\n\n<p>It can be a bit confusing because the site is not exactly designed to be read in a text-only browser, so you may fumble a bit at first with the login and password fields.  Also, you will probably want to get past the &quot;accepting the competition rules&quot; dialog on your local machine before you attempt the download through <code>links</code>.  But beyond that it has always worked for me.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137794,
      "author_name": "rshpeley",
      "author_url": "",
      "post_date": "10/05/2016 04:52:33",
      "content": "<p>Hi @Banana Kitty. Yeah I did. Actually I got it working (see my previous post). Thanks!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137795,
      "author_name": "rshpeley",
      "author_url": "",
      "post_date": "10/05/2016 04:58:45",
      "content": "<p>@Tod. Yeah I couldn't get past the login with lynx. I tried a few times. Mostly my fault there since I don't know lynx well. </p>\n\n<p>Cookies/wget worked after I opened a download session on Chrome (using the direct link to the file) in my local computer, paused the download, and used wget on the remote. Maybe having the open session did something??</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137796,
      "author_name": "rshpeley",
      "author_url": "",
      "post_date": "10/05/2016 05:02:22",
      "content": "<p>@Scott. I needed a solution running on my remote machine, so Cyberduck wouldn't help me there. But it does look interesting. Thanks!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137799,
      "author_name": "tmwright",
      "author_url": "",
      "post_date": "10/05/2016 05:13:32",
      "content": "<p>@r3m I mean literally <code>links</code>, which is a text-based browser distinct from <code>lynx</code>.  I haven't attempted it with <code>lynx</code>, so it is possible that <code>lynx</code> is less agreeable for this task, I wouldn't know.  Sounds like you got it sorted anyway...</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137800,
      "author_name": "rshpeley",
      "author_url": "",
      "post_date": "10/05/2016 05:22:45",
      "content": "<p>Ohh... sorry @Tod, I just assumed that's the way you spelled it. Never heard of it before.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137836,
      "author_name": "yimacs",
      "author_url": "",
      "post_date": "10/05/2016 11:57:42",
      "content": "<p>If you have the datasets on one machine, you could send them to a remote machine, check the <code>scp</code> or <code>rsync</code> command.</p>\n\n<p>or you could use the remote machine to download the datasets directly. I recommend to use the <a href=\"https://github.com/floydwch/kaggle-cli\">unofficial kaggle command line tool</a>. It has some problems at the moment but you are able to download the datasets. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137838,
      "author_name": "yimacs",
      "author_url": "",
      "post_date": "10/05/2016 12:01:31",
      "content": "<p>If you have the datasets on one machine, you could send them to a remote machine, check the <code>scp</code> or <code>rsync</code> command.</p>\n\n<p>or you could use the remote machine to download the datasets directly. I recommend to use the <a href=\"https://github.com/floydwch/kaggle-cli\">unofficial kaggle command line tool</a>. It has some problems at the moment but you are able to download the datasets. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 137839,
      "author_name": "yimacs",
      "author_url": "",
      "post_date": "10/05/2016 12:01:36",
      "content": "<p>If you have the datasets on one machine, you could send them to a remote machine, check the <code>scp</code> or <code>rsync</code> command.</p>\n\n<p>or you could use the remote machine to download the datasets directly. I recommend to use the <a href=\"https://github.com/floydwch/kaggle-cli\">unofficial kaggle command line tool</a>. It has some problems at the moment but you are able to download the datasets. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 138049,
      "author_name": "rshpeley",
      "author_url": "",
      "post_date": "10/06/2016 20:10:22",
      "content": "<p>[quote=YiTang;137839]</p>\n\n<p>...or you could use the remote machine to download the datasets directly. I recommend to use the <a href=\"https://github.com/floydwch/kaggle-cli\">unofficial kaggle command line tool</a>. It has some problems at the moment but you are able to download the datasets. </p>\n\n<p>[/quote]</p>\n\n<p>Thanks @YiTang. Kaggle-CLI may be exactly what I'm looking for. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 138832,
      "author_name": "th3o6a1d",
      "author_url": "",
      "post_date": "10/11/2016 03:44:05",
      "content": "<ol>\n<li>Get a Chrome plugin that lets you copy your cookies to text file. Save this as <code>cookies.txt</code> in your data folder.</li>\n<li>Right-click and copy the kaggle data url from the downloads page.</li>\n<li>Use <code>wget -x --load-cookies cookies.txt url</code></li>\n<li>It will save the download in a nested set of folders corresponding to the link backslashes. </li>\n</ol>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "137748": "Does anyone have a method to download datasets from a remote instance that works for them?\r\n\r\nI've been trying the methods posted on other Kaggle forums with no luck. I've resorted to rural internet service to sftp this dataset and it's uploading at ~50KB/s -- so maybe it'll be done in a week after restarts?\r\n\r\nSurely other people have done this before for AWS instances. I've tried wget with a cookies file, Lynx, Lynx+wget and none of them work for me.\r\n\r\nDoes Kaggle provide a direct means to download the files? Why don't they just set up security keys and be done with it? Where's the harm if my Kaggle profile held my public key?",
    "137750": "There are a few methods. You can download and put the data to Dropbox then use wget to download the Dropbox shared URL. For example, given Dropbox URL \r\n\r\nhttps://www.dropbox.com/s/7ob8wvcdd5k1bnf/helloworld.PNG?dl=0\r\nyou can run \r\n\r\n`wget https://www.dropbox.com/s/7ob8wvcdd5k1bnf/helloworld.PNG`\r\nNote that ?dl=0 is removed.\r\n\r\nAlso you can convert csv data to binary data and put them into dropbox for faster upload/download.",
    "137751": "Check [this][1] thread.\r\nI've used the solution described in the first four posts in the past. It should work.\r\n\r\n  [1]: https://www.kaggle.com/c/belkin-energy-disaggregation-competition/forums/t/5118/downloading-data-via-wget/112602",
    "137758": "Thanks @Banana Kitty. I have downloaded the Bosch dataset to to my computer, it's the upload rate that's my problem. So whether I upload to Dropbox or my remote instance will take the same amount of time. I would need a direct way of getting the datasets to Dropbox for that to work.\r\n\r\n@Rohan, thanks for the link. That's one of them that I tried and I couldn't get any of the methods to work.  On my browser I logged into Kaggle, started the download and accepted the terms. Then took the cookie info, uploaded it to my remote and tried the wget method. No luck...",
    "137765": "This failed to work for me too, initially. I fixed it by changing **%2F** in the third column of the cookies file to **/**\r\n\r\ne.g.\r\n\r\n    .kaggle.com\tTRUE\t%2F\tFALSE\t1467890123 ...\r\n\r\nto\r\n\r\n    .kaggle.com\tTRUE\t/\tFALSE\t1467890123 ...",
    "137774": "Thanks @James. That didn't turn out to be my problem.\r\n\r\nActually I don't know what my problem was. I tried it again and it worked. Using the Chrome *cookie.txt export* extension I did these steps,\r\n\r\n 1. on the download page, right click on one file to download and copy the link\r\n 2. open a new tab and paste the link into it \r\n 3. after the download starts, pause it from the lower download bar\r\n 4. click on the cookie.txt export extension icon on the upper right bar\r\n 5. copy the text from the pop up and follow the instructions\r\n 6. of course, make sure the cookies.txt file is in the current directory when you run wget\r\n 7. repeat for the other files, changing the file name\r\n\r\nThe files will be in a sequence of folders starting at www.kaggle.com",
    "137778": "You can upload files using Cyberduck. http://cyberduck.en.softonic.com/mac",
    "137787": "Hi r3m,\r\n\r\nDid you include https? Not sure if it will make any difference for you.\r\n\r\nGood luck.",
    "137793": "To download to an Amazon EC2 node, I just use `links` on the remote node (I use ubuntu-based AMIs).  You don't *have to* supply a file with the cookies, you can just enter your login username and password from within `links`.  (Indeed, I don't recall ever actually getting the cookie-file approach to work myself.)  \r\n\r\nIt can be a bit confusing because the site is not exactly designed to be read in a text-only browser, so you may fumble a bit at first with the login and password fields.  Also, you will probably want to get past the \"accepting the competition rules\" dialog on your local machine before you attempt the download through `links`.  But beyond that it has always worked for me.",
    "137794": "Hi @Banana Kitty. Yeah I did. Actually I got it working (see my previous post). Thanks!",
    "137795": "Tod. Yeah I couldn't get past the login with lynx. I tried a few times. Mostly my fault there since I don't know lynx well. \r\n\r\nCookies/wget worked after I opened a download session on Chrome (using the direct link to the file) in my local computer, paused the download, and used wget on the remote. Maybe having the open session did something??",
    "137796": "Scott. I needed a solution running on my remote machine, so Cyberduck wouldn't help me there. But it does look interesting. Thanks!",
    "137799": "r3m I mean literally `links`, which is a text-based browser distinct from `lynx`.  I haven't attempted it with `lynx`, so it is possible that `lynx` is less agreeable for this task, I wouldn't know.  Sounds like you got it sorted anyway...",
    "137800": "Ohh... sorry @Tod, I just assumed that's the way you spelled it. Never heard of it before.",
    "137836": "If you have the datasets on one machine, you could send them to a remote machine, check the `scp` or `rsync` command.\r\n\r\nor you could use the remote machine to download the datasets directly. I recommend to use the [unofficial kaggle command line tool][1]. It has some problems at the moment but you are able to download the datasets. \r\n\r\n  [1]: https://github.com/floydwch/kaggle-cli",
    "137838": "If you have the datasets on one machine, you could send them to a remote machine, check the `scp` or `rsync` command.\r\n\r\nor you could use the remote machine to download the datasets directly. I recommend to use the [unofficial kaggle command line tool][1]. It has some problems at the moment but you are able to download the datasets. \r\n\r\n  [1]: https://github.com/floydwch/kaggle-cli",
    "137839": "If you have the datasets on one machine, you could send them to a remote machine, check the `scp` or `rsync` command.\r\n\r\nor you could use the remote machine to download the datasets directly. I recommend to use the [unofficial kaggle command line tool][1]. It has some problems at the moment but you are able to download the datasets. \r\n\r\n  [1]: https://github.com/floydwch/kaggle-cli",
    "138049": "[quote=YiTang;137839]\r\n\r\n...or you could use the remote machine to download the datasets directly. I recommend to use the [unofficial kaggle command line tool][1]. It has some problems at the moment but you are able to download the datasets. \r\n\r\n  [1]: https://github.com/floydwch/kaggle-cli\r\n\r\n[/quote]\r\n\r\nThanks @YiTang. Kaggle-CLI may be exactly what I'm looking for.",
    "138832": "1. Get a Chrome plugin that lets you copy your cookies to text file. Save this as `cookies.txt` in your data folder.\r\n2. Right-click and copy the kaggle data url from the downloads page.\r\n3. Use `wget -x --load-cookies cookies.txt url`\r\n4. It will save the download in a nested set of folders corresponding to the link backslashes."
  },
  "source": "meta"
}