REDCap API in Python

Description

Supports structured data extraction for REDCap projects. The API module d3b_redcap_api.redcap.REDCapStudy can be logically divided into three parts...

Part 1: The low level interface

These are generic request/response handlers implemented here because there's no "official" Python SDK from Vanderbilt for the REDCap API, and the closest thing there is (https://github.com/redcap-tools/PyCap) wasn't being actively maintained when I wanted it.

(If someone wants to replace this part using PyCap at some point, go ahead. Just don't break any of Part 3 in the process.)

Functions: _get_response, _get_json, _get_text, _act_file

Part 2: API functions

With the exception of get_records, which semi-intelligently retrieves records in batches instead of all at once to evade server errors from response timeouts, these all mirror the same/similarly named commands described in the REDCap API documentation at https://<your_redcap_server>/api/help/. For CHOP users that would be https://redcap.chop.edu/api/help.

(If someone wants to replace this part using PyCap at some point, go ahead. Just don't break any of Part 3 in the process.)

Functions:

Part 3: Whole project retrieval and structuring

Functions:

{
  <field_name>: {
    <index>: <value>,
    ...indexes
  },
  ...fields
}

{
  <event>: {                   # event data
    <instrument>: {            # event instrument data
      <record_id>: {           # subject data for this event instrument
        <instance>: {          # subject event instrument instance
          <field>: {<values>}, # set of field values
          ...fields
        },
        ...instances
      },
      ...records
    },
    ...instruments
  },
  ...events
}

There are also some extra utility functions for converting the records tree into one or more Pandas DataFrames...

Example Usage:

from d3b_redcap_api.redcap import REDCapStudy
from d3b_redcap_api.df_utils import all_dfs

r = REDCapStudy("https://redcap.chop.edu/api/", PROJECT_API_TOKEN)
study_data, errors = r.get_records_tree(raw_selectors=False)

as_dataframes = all_dfs(study_data)