|
| 1 | +//! An example program that splits a DBN file into several DBN files |
| 2 | +//! by parent symbol (from the `asset` field in the definitions schema). |
| 3 | +use std::collections::HashMap; |
| 4 | + |
| 5 | +use anyhow::Context; |
| 6 | +use async_compression::tokio::write::ZstdEncoder; |
| 7 | +use databento::{ |
| 8 | + dbn::{ |
| 9 | + decode::AsyncDbnDecoder, encode::AsyncDbnEncoder, InstrumentDefMsg, Metadata, Schema, |
| 10 | + SymbolIndex, |
| 11 | + }, |
| 12 | + historical::timeseries::GetRangeParams, |
| 13 | + HistoricalClient, |
| 14 | +}; |
| 15 | +use tokio::fs::File; |
| 16 | + |
| 17 | +#[tokio::main] |
| 18 | +async fn main() -> anyhow::Result<()> { |
| 19 | + if std::env::args().len() != 3 { |
| 20 | + anyhow::bail!( |
| 21 | + "Invalid number of arguments, expected: split_symbols FILE_PATH OUTPUT_PATTERN" |
| 22 | + ); |
| 23 | + } |
| 24 | + let file_path = std::env::args().nth(1).unwrap(); |
| 25 | + let output_pattern = std::env::args().nth(2).unwrap(); |
| 26 | + if !output_pattern.contains("{parent}") { |
| 27 | + anyhow::bail!("OUTPUT_PATTERN should contain {{parent}}"); |
| 28 | + } |
| 29 | + let mut decoder = AsyncDbnDecoder::from_zstd_file(file_path).await?; |
| 30 | + |
| 31 | + let metadata = decoder.metadata().clone(); |
| 32 | + let symbol_map = metadata.symbol_map()?; |
| 33 | + let symbols_to_parent = fetch_symbols_to_parent(&metadata).await?; |
| 34 | + let mut encoders = HashMap::<String, AsyncDbnEncoder<ZstdEncoder<File>>>::new(); |
| 35 | + while let Some(rec) = decoder.decode_record_ref().await? { |
| 36 | + let Some(symbol) = symbol_map.get_for_rec(&rec) else { |
| 37 | + eprintln!("Missing mapping for {rec:?}"); |
| 38 | + continue; |
| 39 | + }; |
| 40 | + let Some(parent) = symbols_to_parent.get(symbol) else { |
| 41 | + eprintln!("Couldn't find parent mapping for {symbol} with {rec:?}"); |
| 42 | + continue; |
| 43 | + }; |
| 44 | + if let Some(encoder) = encoders.get_mut(parent) { |
| 45 | + encoder.encode_record_ref(rec).await?; |
| 46 | + } else { |
| 47 | + let mut encoder = AsyncDbnEncoder::with_zstd( |
| 48 | + File::create_new(output_pattern.replace("{parent}", parent)) |
| 49 | + .await |
| 50 | + .with_context(|| format!("creating file for {parent}"))?, |
| 51 | + &metadata, |
| 52 | + ) |
| 53 | + .await?; |
| 54 | + encoder.encode_record_ref(rec).await?; |
| 55 | + encoders.insert(parent.clone(), encoder); |
| 56 | + }; |
| 57 | + } |
| 58 | + for (parent, encoder) in encoders { |
| 59 | + if let Err(e) = encoder.shutdown().await { |
| 60 | + eprintln!("Failed to shutdown encoder for {parent}: {e:?}"); |
| 61 | + } |
| 62 | + } |
| 63 | + |
| 64 | + Ok(()) |
| 65 | +} |
| 66 | + |
| 67 | +async fn fetch_symbols_to_parent(metadata: &Metadata) -> anyhow::Result<HashMap<String, String>> { |
| 68 | + let mut client = HistoricalClient::builder().key_from_env()?.build()?; |
| 69 | + let end = metadata.end().ok_or_else(|| { |
| 70 | + anyhow::format_err!("Missing end in metadata. This script is intended for historical data") |
| 71 | + })?; |
| 72 | + let mut res = HashMap::new(); |
| 73 | + // 2000 is the maximum number of symbols per request |
| 74 | + for chunk in metadata.symbols.chunks(2000) { |
| 75 | + let mut decoder = client |
| 76 | + .timeseries() |
| 77 | + .get_range( |
| 78 | + &GetRangeParams::builder() |
| 79 | + .dataset(metadata.dataset.clone()) |
| 80 | + .schema(Schema::Definition) |
| 81 | + .date_time_range((metadata.start(), end)) |
| 82 | + .symbols(Vec::from(chunk)) |
| 83 | + .build(), |
| 84 | + ) |
| 85 | + .await?; |
| 86 | + while let Some(def) = decoder.decode_record::<InstrumentDefMsg>().await? { |
| 87 | + res.insert(def.raw_symbol()?.to_owned(), def.asset()?.to_owned()); |
| 88 | + } |
| 89 | + } |
| 90 | + Ok(res) |
| 91 | +} |
0 commit comments