re:dash Python functions

We’ve been using re:dash at Sentry lately. I’d never heard of it before (Mode being the frontrunner in this category), but it’s damned good so far.

One of the best features is the Python query runner, allowing you to manipulate data from your SQL queries. You can then output to visualizations, csv, etc. The global python functions aren’t obvious (I looked through their code for them), documenting here for others. I also make liberal use of print, which the re:dash guys have thoughtfully made print to the page.


At the end of the query, re:dash looks at the variable result. If it’s in the right shape, it’ll be show in the table below. The result is a dictionary and is the same that is returned by execute_query and get_query_result. The format should be like this:

      "rows": [
        {"user_id": 1, "anonymous_id": "s2f1"},
      "columns": [
          "name": "user_id",
          "friendly_name": "user_id",
          "type": TYPE_INTEGER,
          "name": "anonymous_id",
          "friendly_name": "anonymous_id",
          "type": TYPE_STRING,

The available value types are:



Let’s you…well, execute a query within Python.

result = execute_query("<data source name>", "your SQL query")


Grabs the results on a previous query. I prefer this, as it means while I’m debugging a python script, I’m not hitting the database over and over again.



If you want to create a new result from scratch (and just append data to it), the base result dictionary is just {"columns": [], "rows": []}.

add_result_column(result, "<new column name>", "<new friendly name for column>", "<column type>")


values should be a dictionary with the keys being column names.

add_result_row(result, "<values>")


The current result format isn’t vey friendly to data analysts. At some point, I’ll probably create a helper function that allows us to work in something closer to data frames (like R) or maybe figure out how re:dash and pandas works together.

  1. TYPE_DATETIME has some odd behavior. Still figuring out how to make it play nice with python datetime module.