working with recursive

This commit is contained in:
Palash Tyagi
2024-11-07 20:26:50 +00:00
parent b14a59b259
commit 63c1cf64b2
5 changed files with 196 additions and 317 deletions

View File

@@ -1,8 +1,11 @@
use log;
use macrosynergy_dataquery::oauth_client::OAuthClient;
use macrosynergy_dataquery::requester;
use macrosynergy_dataquery::requester::*;
use macrosynergy_dataquery::timeseries::*;
use rand::seq::SliceRandom;
use std::collections::HashSet;
use std::time::Instant;
fn main() {
let mut requester = requester::DQRequester::new(OAuthClient::default());
@@ -10,7 +13,6 @@ fn main() {
let responses = requester.get_catalogue("JPMAQS", 1000).unwrap();
let mut tickers_vec = vec![];
for rsp in responses {
let response: DQCatalogueResponse = serde_json::from_str(&rsp.to_string()).unwrap();
@@ -41,12 +43,31 @@ fn main() {
}
}
let rsp = requester.get_timeseries(DQTimeseriesRequestArgs {
expressions: expressions_vec[0..40].to_vec(),
..Default::default()
});
// let rsp = requester.get_timeseries(DQTimeseriesRequestArgs {
// expressions: expressions_vec[0..40].to_vec(),
// ..Default::default()
// });
for response in rsp {
// for response in rsp {
// let dq_response: DQTimeSeriesResponse =
// serde_json::from_str(&response.text().unwrap()).unwrap();
// for ts_group in dq_response.get_timeseries_by_ticker() {
// let jpmi = JPMaQSIndicator::new(ts_group).unwrap();
// println!("{:?}", jpmi.as_qdf().unwrap());
// }
// }
// fetch first 50 expressions
let start = Instant::now();
let responses = requester
.get_timeseries(DQTimeseriesRequestArgs {
expressions: expressions_vec[0..50].to_vec(),
..Default::default()
})
.unwrap();
for response in responses {
let dq_response: DQTimeSeriesResponse =
serde_json::from_str(&response.text().unwrap()).unwrap();
for ts_group in dq_response.get_timeseries_by_ticker() {

View File

@@ -167,10 +167,14 @@ impl DQRequester {
Ok(responses)
}
pub fn get_single_timeseries_batch(
pub fn _fetch_single_timeseries_batch(
&mut self,
args: DQTimeseriesRequestArgs,
) -> Result<reqwest::blocking::Response, Box<dyn Error>> {
log::info!(
"Fetching timeseries batch with {} expressions",
args.expressions.len()
);
if args.expressions.len() < 1 || args.expressions.len() > 20 {
return Err("Number of expressions must be between 1 and 20".into());
}
@@ -179,49 +183,94 @@ impl DQRequester {
let endpoint = format!("{}?{}", TIMESERIES_ENDPOINT, query_string);
let response = self._request(reqwest::Method::GET, &endpoint)?;
log::info!("Got response: {:?}", response.status());
Ok(response)
}
pub fn get_single_timeseries_with_defaults(
&mut self,
expressions: Vec<&str>,
) -> Result<reqwest::blocking::Response, Box<dyn Error>> {
let args = DQTimeseriesRequestArgs {
expressions: expressions.into_iter().map(|s| s.to_string()).collect(),
..Default::default()
};
// pub fn get_timeseries(
// &mut self,
// args: DQTimeseriesRequestArgs,
// ) -> Vec<reqwest::blocking::Response> {
// let expressions_chunks: Vec<Vec<String>> = args
// .expressions
// .chunks(20)
// .map(|chunk| chunk.to_vec())
// .collect();
self.get_single_timeseries_batch(args)
}
// let mut responses: Vec<Result<reqwest::blocking::Response, Box<dyn Error>>> = Vec::new();
pub fn get_timeseries(
// for expressions in expressions_chunks {
// let mut args = args.clone();
// args.update_expressions(expressions);
// let response = self.get_single_timeseries_batch(args);
// responses.push(response);
// // just sleep for a bit not threadsleep
// // println!("Sleeping for 500ms");
// std::thread::sleep(std::time::Duration::from_millis(500));
// }
// responses.into_iter().map(|r| r.unwrap()).collect()
// }
fn _fetch_timeseries_recursive(
&mut self,
args: DQTimeseriesRequestArgs,
) -> Vec<reqwest::blocking::Response> {
let expressions_chunks: Vec<Vec<String>> = args
max_retries: u32,
) -> Result<Vec<reqwest::blocking::Response>, Box<dyn Error>> {
let expression_batches: Vec<Vec<String>> = args
.expressions
.chunks(20)
.map(|chunk| chunk.to_vec())
.collect();
let mut responses: Vec<Result<reqwest::blocking::Response, Box<dyn Error>>> = Vec::new();
let mut okay_responses: Vec<reqwest::blocking::Response> = Vec::new();
let mut failed_batches: Vec<Vec<String>> = Vec::new();
for expressions in expressions_chunks {
for expr_batch in expression_batches {
let mut args = args.clone();
args.update_expressions(expressions);
println!(
"Requesting timeseries for expressions: {:?}",
args.expressions
);
let response = self.get_single_timeseries_batch(args);
responses.push(response);
// just sleep for a bit not threadsleep
println!("Sleeping for 500ms");
std::thread::sleep(std::time::Duration::from_millis(500));
args.update_expressions(expr_batch.clone());
let response = self._fetch_single_timeseries_batch(args);
std::thread::sleep(std::time::Duration::from_millis(200));
match response {
Ok(r) => {
okay_responses.push(r);
log::info!("Got batch: {:?}", expr_batch);
}
Err(e) => {
failed_batches.push(expr_batch.clone());
log::error!("Failed to get batch: {:?} : {:?}", expr_batch, e);
}
}
}
responses.into_iter().map(|r| r.unwrap()).collect()
if !failed_batches.is_empty() && max_retries == 0 {
return Err("Max retries reached".into());
}
if !failed_batches.is_empty() && max_retries > 0 {
log::info!("Retrying failed batches");
let mut new_args = args.clone();
new_args.update_expressions(failed_batches.concat());
log::info!("Retrying with {} failed expressions", failed_batches.len());
let mut retry_responses =
self._fetch_timeseries_recursive(new_args, max_retries - 1)?;
okay_responses.append(&mut retry_responses);
}
log::info!("Returning {} responses", okay_responses.len());
Ok(okay_responses)
}
pub fn get_timeseries(
&mut self,
args: DQTimeseriesRequestArgs,
) -> Result<Vec<reqwest::blocking::Response>, Box<dyn Error>> {
let max_retries = 5;
println!(
"Invoking recursive function for {:?} expressions",
args.expressions.len()
);
self._fetch_timeseries_recursive(args, max_retries)
}
}

View File

@@ -108,6 +108,18 @@ impl DQTimeSeries {
}
impl DQTimeSeriesResponse {
pub fn list_expressions(&self) -> Vec<String> {
self.instruments
.iter()
.flat_map(|instrument| {
instrument
.attributes
.iter()
.map(|attribute| attribute.expression.clone())
})
.collect()
}
/// Return a list of all DQTimeSeries in the response
pub fn get_all_timeseries(&self) -> Vec<DQTimeSeries> {
self.instruments