A small CLI tool for syncing data from the Canvas Data API.

NOTE: this is currently in beta, please report any bugs or issues you find!

This tool should work on Linux, OSX, and Windows. The tool uses node.js runtime, which you will need to install before being able to use it.

1. Install Node.js - Any version newer than 0.12.0 should work, best bet is to follow the instructions here

npm install -g canvas-data-cli

git clone https://github.com/instructure/canvas-data-cli.git && cd canvas-data-cli && make installLocal

The Canvas Data CLI requires a configuration file with a fields set. Canvas Data CLI uses a small javascript file as configuration file. To generate a stub of this configuration run canvasDataCli sampleConfig which will create a config.js.sample file. Rename this to a file, like config.js.

Edit the file to point to where you want to save the files as well as the file used to track the state of which data exports you have already downloaded. By default the sample config file tries to pull your API key and secret from environment variables, CD_API_KEY and CD_API_SECRET, which is more secure, however, you can also hard code the credentials in the config file.

canvas-data-cli has support for HTTP Proxies, both with and without basic authentication. To do this there are three extra options you can add to your config file. httpsProxy, proxyUsername, and proxyPassword.

If you want to simply download all the data from Canva Data, the sync command can be used to keep an up-to-date copy locally.

canvasDataCli sync -c path/to/config.js

This will start the sync process. The sync process uses the sync api endpoint to get a list of all the files. If the file does

not exist, it will download it. Otherwise, it will skip the file. After downloading all files, it will delete any unexpected files

in the directory to remove old data.

On subsequent executions, it will only download the files it doesn't have.

This process is also resumeable, if for whatever reason you have issues, it should restart and download only the files

that previously failed. One of the ways to make this more safe is that it downloads the file to a temporary name and

renames it once the process is finished. This may leave around gz.tmp files, but they should get deleted automatically once

you have a successful run.

If you run this daily, you should keep all of your data from Canvas Data up to date.

Fetches most up to date data for a single table from the API. This ignores any previously downloaded files and will redownload all the files associated with that table.

canvasDataCli fetch -c path/to/config.js -t user_dim

This will start the fetch process and download what is needed to get the most recent data for that table (in this case, the user_dim).

On subsequent executions, this will redownload all the data for that table, ignoring any previous days data.

NOTE: This only works after properly running a sync command

This command will unpack the gzipped files, concat any partitioned files, and add a header to the output file

canvasDataCli unpack -c path/to/config.js -f user_dim,account_dim

This command will unpack the user_dim and account_dim tables to a directory. Currently, you explictly have to give the files you want to unpack as this has the potential for creating very large files.

This subcommand is designed to allow users to make API calls directly. The main use case for which is debugging and development.

canvasDataCli api -c config.js -r /account/self/dump

Periodically requests data is regrouped into collections that span more than just a single day. In this case, the date that the files were generated differs from the time that the included requests were made. To make it easier to identify which files contain the requests made during a particular time range, we have the historical-requests subcommand.

canvasDataCli historical-requests -c config.js

Its output takes the form:

{
  "dumpId": "...",
  "ranges": {
    "20180315_20180330": [
      {
        "url": "...",
        "filename": "..."
      },
      {
        "url": "...",
        "filename": "..."
      }
    ],
    "20180331_20180414": [
      {
        "url": "...",
        "filename": "..."
      }
    ]
  }
}

Process:

1. Write some code
2. Write tests
3. Run make installLocal to test changes
4. Run bin/canvasDataCli .... to test your changes locally
5. Open a pull request

If you use docker, you can run tests inside a docker container

./build.sh

npm install .
npm test