Lib.rs

fred_api makes requests to FRED to download economic data, caching it to a data-store so that repeated requests to FRED can be avoided. Requires the FRED_API_KEY environment variable to be set. When requests are made to FRED or the cache, the response is written into the file debug.xml for convenient debugging.

Requirements

1. FRED_API_KEY environment variable has to be set.

2. A directory used to cache requests is required. This directory is specified in the sled::open("../fred_cache/db") function, as shown in the code example below.

A key can be obtained by creating an account at FRED , and should be set as an environment variable like

FRED_API_KEY=abcdefghijklmnopqrstuvwxyz123456

Code Example

use {
    fred_api::{build_request, capture_fields, send_request},
    sled::Db
};

# tokio_test::block_on(async {
// Set the path to the cache. The cache stores the response bytes as is, keyed by
// the `RequestSpec`.
let cache: sled::Db = sled::open("../fred_cache/db").unwrap();

// The spec string can be taken from the FRED API docs. `req_spec` uniquely
// identifies the request and is used as a key in the cache. `req` is a
// `hyper::Request`.
let (req_spec, req) = build_request("series/observations?series_id=CPGRLE01AUQ657N").unwrap();

// The boolean values determine whether to request from the cache and/or request from
// FRED. Successful FRED responses are always cached. Responses with other than HTTP
// status `Ok` status will return an error.
let bytes = send_request(&req_spec, req, true, false, &cache).await.unwrap();

// Consult the FRED API docs XML responses to determine the fields to be captured
// from the bytes. The fields must be specified in the same order they appear in the
// response.
let caps: Vec<Vec<Vec<u8>>> = capture_fields("observation", vec!("date", "value"), &bytes);
assert_eq!(caps[0][0], "   ".as_bytes()); // date
assert_eq!(caps[0][1], "   ".as_bytes()); // value
# })

Minor Details

Requests are concurrent and there is no rate limit on the requests. Hopefully making use of the cache rather than repeatedly making requests for the same data is sufficient to prevent FRED from blocking excessive requests.

capture_fields() uses regular expressions that assume the data is "well-formed" and the fields are consistently ordered. I haven't come across any problems in my own use, but I haven't tried to make these expressions robust in any way. There is always the option of just taking the bytes as is and doing one's own deserialization (which is recommended in any case because everyone's reliability/simplicity trade-offs are different).