Skip to main content

nautilus_serialization/arrow/instrument/
commodity.rs

1// -------------------------------------------------------------------------------------------------
2//  Copyright (C) 2015-2026 Nautech Systems Pty Ltd. All rights reserved.
3//  https://nautechsystems.io
4//
5//  Licensed under the GNU Lesser General Public License Version 3.0 (the "License");
6//  You may not use this file except in compliance with the License.
7//  You may obtain a copy of the License at https://www.gnu.org/licenses/lgpl-3.0.en.html
8//
9//  Unless required by applicable law or agreed to in writing, software
10//  distributed under the License is distributed on an "AS IS" BASIS,
11//  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12//  See the License for the specific language governing permissions and
13//  limitations under the License.
14// -------------------------------------------------------------------------------------------------
15
16//! Arrow serialization for Commodity instruments.
17
18use std::{borrow::Borrow, collections::HashMap, str::FromStr, sync::Arc};
19
20use arrow::{
21    array::{Array, StringArray, StringBuilder, UInt8Array, UInt64Array},
22    datatypes::{DataType, Field, Schema},
23    error::ArrowError,
24    record_batch::RecordBatch,
25};
26use nautilus_core::{Params, UnixNanos};
27use nautilus_model::{
28    enums::AssetClass,
29    identifiers::{InstrumentId, Symbol},
30    instruments::commodity::Commodity,
31    types::{money::Money, price::Price, quantity::Quantity},
32};
33use rust_decimal::Decimal;
34
35use super::validate_required_instrument_columns;
36use crate::arrow::{
37    ArrowSchemaProvider, EncodeToRecordBatch, EncodingError, KEY_INSTRUMENT_ID,
38    KEY_PRICE_PRECISION, extract_column, extract_column_by_name,
39    extract_optional_string_column_by_name, json_string_field, metadata_with_type_name,
40    optional_ustr_value, record_batch_with_timestamps, record_batch_with_u64_timestamps,
41    timestamp_data_type,
42};
43
44impl ArrowSchemaProvider for Commodity {
45    fn get_schema(metadata: Option<HashMap<String, String>>) -> Schema {
46        let fields = vec![
47            Field::new("id", DataType::Utf8, false),
48            Field::new("raw_symbol", DataType::Utf8, false),
49            Field::new("asset_class", DataType::Utf8, false),
50            Field::new("quote_currency", DataType::Utf8, false),
51            Field::new("price_precision", DataType::UInt8, false),
52            Field::new("size_precision", DataType::UInt8, false),
53            Field::new("price_increment", DataType::Utf8, false),
54            Field::new("size_increment", DataType::Utf8, false),
55            Field::new("lot_size", DataType::Utf8, true), // nullable
56            Field::new("max_quantity", DataType::Utf8, true), // nullable
57            Field::new("min_quantity", DataType::Utf8, true), // nullable
58            Field::new("max_notional", DataType::Utf8, true), // nullable
59            Field::new("min_notional", DataType::Utf8, true), // nullable
60            Field::new("max_price", DataType::Utf8, true), // nullable
61            Field::new("min_price", DataType::Utf8, true), // nullable
62            Field::new("margin_init", DataType::Utf8, false),
63            Field::new("margin_maint", DataType::Utf8, false),
64            Field::new("tick_scheme", DataType::Utf8, true),
65            json_string_field("info", true),
66            Field::new("ts_event", timestamp_data_type(), false),
67            Field::new("ts_init", timestamp_data_type(), false),
68        ];
69
70        Schema::new_with_metadata(fields, metadata_with_type_name("Commodity", metadata))
71    }
72}
73
74impl EncodeToRecordBatch for Commodity {
75    fn encode_batch<T>(
76        #[allow(unused)] metadata: &HashMap<String, String>,
77        data: &[T],
78    ) -> Result<RecordBatch, ArrowError>
79    where
80        T: std::borrow::Borrow<Self>,
81    {
82        let mut id_builder = StringBuilder::new();
83        let mut raw_symbol_builder = StringBuilder::new();
84        let mut asset_class_builder = StringBuilder::new();
85        let mut quote_currency_builder = StringBuilder::new();
86        let mut price_precision_builder = UInt8Array::builder(data.len());
87        let mut size_precision_builder = UInt8Array::builder(data.len());
88        let mut price_increment_builder = StringBuilder::new();
89        let mut size_increment_builder = StringBuilder::new();
90        let mut lot_size_builder = StringBuilder::new();
91        let mut max_quantity_builder = StringBuilder::new();
92        let mut min_quantity_builder = StringBuilder::new();
93        let mut max_notional_builder = StringBuilder::new();
94        let mut min_notional_builder = StringBuilder::new();
95        let mut max_price_builder = StringBuilder::new();
96        let mut min_price_builder = StringBuilder::new();
97        let mut margin_init_builder = StringBuilder::new();
98        let mut margin_maint_builder = StringBuilder::new();
99        let mut tick_scheme_builder = StringBuilder::new();
100        let mut info_builder = StringBuilder::new();
101        let mut ts_event_builder = UInt64Array::builder(data.len());
102        let mut ts_init_builder = UInt64Array::builder(data.len());
103
104        for commodity in data.iter().map(Borrow::borrow) {
105            id_builder.append_value(commodity.id.to_string());
106            raw_symbol_builder.append_value(commodity.raw_symbol);
107            asset_class_builder.append_value(commodity.asset_class);
108            quote_currency_builder.append_value(commodity.quote_currency.to_string());
109            price_precision_builder.append_value(commodity.price_precision);
110            size_precision_builder.append_value(commodity.size_precision);
111            price_increment_builder.append_value(commodity.price_increment.to_string());
112            size_increment_builder.append_value(commodity.size_increment.to_string());
113
114            if let Some(lot_size) = commodity.lot_size {
115                lot_size_builder.append_value(lot_size.to_string());
116            } else {
117                lot_size_builder.append_null();
118            }
119
120            if let Some(max_qty) = commodity.max_quantity {
121                max_quantity_builder.append_value(max_qty.to_string());
122            } else {
123                max_quantity_builder.append_null();
124            }
125
126            if let Some(min_qty) = commodity.min_quantity {
127                min_quantity_builder.append_value(min_qty.to_string());
128            } else {
129                min_quantity_builder.append_null();
130            }
131
132            if let Some(max_not) = commodity.max_notional {
133                max_notional_builder.append_value(max_not.to_string());
134            } else {
135                max_notional_builder.append_null();
136            }
137
138            if let Some(min_not) = commodity.min_notional {
139                min_notional_builder.append_value(min_not.to_string());
140            } else {
141                min_notional_builder.append_null();
142            }
143
144            if let Some(max_p) = commodity.max_price {
145                max_price_builder.append_value(max_p.to_string());
146            } else {
147                max_price_builder.append_null();
148            }
149
150            if let Some(min_p) = commodity.min_price {
151                min_price_builder.append_value(min_p.to_string());
152            } else {
153                min_price_builder.append_null();
154            }
155
156            margin_init_builder.append_value(commodity.margin_init.to_string());
157            margin_maint_builder.append_value(commodity.margin_maint.to_string());
158
159            if let Some(tick_scheme) = commodity.tick_scheme {
160                tick_scheme_builder.append_value(tick_scheme);
161            } else {
162                tick_scheme_builder.append_null();
163            }
164
165            if let Some(ref info) = commodity.info {
166                match serde_json::to_string(info) {
167                    Ok(json) => {
168                        info_builder.append_value(json);
169                    }
170                    Err(e) => {
171                        return Err(ArrowError::InvalidArgumentError(format!(
172                            "Failed to serialize info dict to JSON: {e}"
173                        )));
174                    }
175                }
176            } else {
177                info_builder.append_null();
178            }
179
180            ts_event_builder.append_value(commodity.ts_event.as_u64());
181            ts_init_builder.append_value(commodity.ts_init.as_u64());
182        }
183
184        record_batch_with_timestamps(
185            Self::get_schema(Some(metadata.clone())).into(),
186            vec![
187                Arc::new(id_builder.finish()),
188                Arc::new(raw_symbol_builder.finish()),
189                Arc::new(asset_class_builder.finish()),
190                Arc::new(quote_currency_builder.finish()),
191                Arc::new(price_precision_builder.finish()),
192                Arc::new(size_precision_builder.finish()),
193                Arc::new(price_increment_builder.finish()),
194                Arc::new(size_increment_builder.finish()),
195                Arc::new(lot_size_builder.finish()),
196                Arc::new(max_quantity_builder.finish()),
197                Arc::new(min_quantity_builder.finish()),
198                Arc::new(max_notional_builder.finish()),
199                Arc::new(min_notional_builder.finish()),
200                Arc::new(max_price_builder.finish()),
201                Arc::new(min_price_builder.finish()),
202                Arc::new(margin_init_builder.finish()),
203                Arc::new(margin_maint_builder.finish()),
204                Arc::new(tick_scheme_builder.finish()),
205                Arc::new(info_builder.finish()),
206                Arc::new(ts_event_builder.finish()),
207                Arc::new(ts_init_builder.finish()),
208            ],
209        )
210    }
211
212    fn metadata(&self) -> HashMap<String, String> {
213        let mut metadata = HashMap::new();
214        metadata.insert(KEY_INSTRUMENT_ID.to_string(), self.id.to_string());
215        metadata.insert(
216            KEY_PRICE_PRECISION.to_string(),
217            self.price_precision.to_string(),
218        );
219        metadata
220    }
221}
222
223/// Decodes [`Commodity`] instruments from a record batch.
224///
225/// Not a [`DecodeFromRecordBatch`] implementation because that trait requires `Into<Data>`.
226///
227/// # Errors
228///
229/// Returns an error if the record batch cannot be decoded.
230///
231/// [`DecodeFromRecordBatch`]: crate::arrow::DecodeFromRecordBatch
232pub fn decode_commodity_batch(
233    metadata: &HashMap<String, String>,
234    record_batch: &RecordBatch,
235) -> Result<Vec<Commodity>, EncodingError> {
236    validate_required_instrument_columns::<Commodity>(metadata, record_batch)?;
237    let record_batch = record_batch_with_u64_timestamps(record_batch)?;
238    let record_batch = &record_batch;
239    let cols = record_batch.columns();
240    let num_rows = record_batch.num_rows();
241
242    let id_values = extract_column::<StringArray>(cols, "id", 0, DataType::Utf8)?;
243    let raw_symbol_values = extract_column::<StringArray>(cols, "raw_symbol", 1, DataType::Utf8)?;
244    let asset_class_values = extract_column::<StringArray>(cols, "asset_class", 2, DataType::Utf8)?;
245    let quote_currency_values =
246        extract_column::<StringArray>(cols, "quote_currency", 3, DataType::Utf8)?;
247    let price_precision_values =
248        extract_column::<UInt8Array>(cols, "price_precision", 4, DataType::UInt8)?;
249    let size_precision_values =
250        extract_column::<UInt8Array>(cols, "size_precision", 5, DataType::UInt8)?;
251    let price_increment_values =
252        extract_column::<StringArray>(cols, "price_increment", 6, DataType::Utf8)?;
253    let size_increment_values =
254        extract_column::<StringArray>(cols, "size_increment", 7, DataType::Utf8)?;
255    let lot_size_values = cols
256        .get(8)
257        .ok_or_else(|| EncodingError::MissingColumn("lot_size", 8))?;
258    let max_quantity_values = cols
259        .get(9)
260        .ok_or_else(|| EncodingError::MissingColumn("max_quantity", 9))?;
261    let min_quantity_values = cols
262        .get(10)
263        .ok_or_else(|| EncodingError::MissingColumn("min_quantity", 10))?;
264    let max_notional_values = cols
265        .get(11)
266        .ok_or_else(|| EncodingError::MissingColumn("max_notional", 11))?;
267    let min_notional_values = cols
268        .get(12)
269        .ok_or_else(|| EncodingError::MissingColumn("min_notional", 12))?;
270    let max_price_values = cols
271        .get(13)
272        .ok_or_else(|| EncodingError::MissingColumn("max_price", 13))?;
273    let min_price_values = cols
274        .get(14)
275        .ok_or_else(|| EncodingError::MissingColumn("min_price", 14))?;
276    let margin_init_values =
277        extract_column::<StringArray>(cols, "margin_init", 15, DataType::Utf8)?;
278    let margin_maint_values =
279        extract_column::<StringArray>(cols, "margin_maint", 16, DataType::Utf8)?;
280    let tick_scheme_values = extract_optional_string_column_by_name(record_batch, "tick_scheme")?;
281    let info_values = extract_column_by_name::<StringArray>(record_batch, "info", DataType::Utf8)?;
282    let ts_event_values =
283        extract_column_by_name::<UInt64Array>(record_batch, "ts_event", DataType::UInt64)?;
284    let ts_init_values =
285        extract_column_by_name::<UInt64Array>(record_batch, "ts_init", DataType::UInt64)?;
286
287    let mut result = Vec::with_capacity(num_rows);
288
289    for i in 0..num_rows {
290        let id = InstrumentId::from_str(id_values.value(i))
291            .map_err(|e| EncodingError::ParseError("id", format!("row {i}: {e}")))?;
292        let raw_symbol = Symbol::from(raw_symbol_values.value(i));
293        let asset_class = AssetClass::from_str(asset_class_values.value(i))
294            .map_err(|e| EncodingError::ParseError("asset_class", format!("row {i}: {e}")))?;
295        let quote_currency = super::decode_currency(
296            quote_currency_values.value(i),
297            "quote_currency",
298            "commodity.quote_currency",
299            i,
300        )?;
301        let price_prec = price_precision_values.value(i);
302        let size_prec = size_precision_values.value(i);
303
304        let price_increment = Price::from_str(price_increment_values.value(i))
305            .map_err(|e| EncodingError::ParseError("price_increment", format!("row {i}: {e}")))?;
306        let size_increment = Quantity::from_str(size_increment_values.value(i))
307            .map_err(|e| EncodingError::ParseError("size_increment", format!("row {i}: {e}")))?;
308
309        let lot_size = if lot_size_values.is_null(i) {
310            None
311        } else {
312            let lot_size_str = lot_size_values
313                .as_any()
314                .downcast_ref::<StringArray>()
315                .ok_or_else(|| {
316                    EncodingError::ParseError("lot_size", format!("row {i}: invalid type"))
317                })?
318                .value(i);
319            Some(
320                Quantity::from_str(lot_size_str)
321                    .map_err(|e| EncodingError::ParseError("lot_size", format!("row {i}: {e}")))?,
322            )
323        };
324
325        let max_quantity =
326            if max_quantity_values.is_null(i) {
327                None
328            } else {
329                let max_qty_str = max_quantity_values
330                    .as_any()
331                    .downcast_ref::<StringArray>()
332                    .ok_or_else(|| {
333                        EncodingError::ParseError("max_quantity", format!("row {i}: invalid type"))
334                    })?
335                    .value(i);
336                Some(Quantity::from_str(max_qty_str).map_err(|e| {
337                    EncodingError::ParseError("max_quantity", format!("row {i}: {e}"))
338                })?)
339            };
340
341        let min_quantity =
342            if min_quantity_values.is_null(i) {
343                None
344            } else {
345                let min_qty_str = min_quantity_values
346                    .as_any()
347                    .downcast_ref::<StringArray>()
348                    .ok_or_else(|| {
349                        EncodingError::ParseError("min_quantity", format!("row {i}: invalid type"))
350                    })?
351                    .value(i);
352                Some(Quantity::from_str(min_qty_str).map_err(|e| {
353                    EncodingError::ParseError("min_quantity", format!("row {i}: {e}"))
354                })?)
355            };
356
357        let max_notional =
358            if max_notional_values.is_null(i) {
359                None
360            } else {
361                let max_not_str = max_notional_values
362                    .as_any()
363                    .downcast_ref::<StringArray>()
364                    .ok_or_else(|| {
365                        EncodingError::ParseError("max_notional", format!("row {i}: invalid type"))
366                    })?
367                    .value(i);
368
369                Some(Money::from_str_exact(max_not_str).map_err(|e| {
370                    EncodingError::ParseError("max_notional", format!("row {i}: {e}"))
371                })?)
372            };
373
374        let min_notional =
375            if min_notional_values.is_null(i) {
376                None
377            } else {
378                let min_not_str = min_notional_values
379                    .as_any()
380                    .downcast_ref::<StringArray>()
381                    .ok_or_else(|| {
382                        EncodingError::ParseError("min_notional", format!("row {i}: invalid type"))
383                    })?
384                    .value(i);
385
386                Some(Money::from_str_exact(min_not_str).map_err(|e| {
387                    EncodingError::ParseError("min_notional", format!("row {i}: {e}"))
388                })?)
389            };
390
391        let max_price = if max_price_values.is_null(i) {
392            None
393        } else {
394            let max_p_str = max_price_values
395                .as_any()
396                .downcast_ref::<StringArray>()
397                .ok_or_else(|| {
398                    EncodingError::ParseError("max_price", format!("row {i}: invalid type"))
399                })?
400                .value(i);
401            Some(
402                Price::from_str(max_p_str)
403                    .map_err(|e| EncodingError::ParseError("max_price", format!("row {i}: {e}")))?,
404            )
405        };
406
407        let min_price = if min_price_values.is_null(i) {
408            None
409        } else {
410            let min_p_str = min_price_values
411                .as_any()
412                .downcast_ref::<StringArray>()
413                .ok_or_else(|| {
414                    EncodingError::ParseError("min_price", format!("row {i}: invalid type"))
415                })?
416                .value(i);
417            Some(
418                Price::from_str(min_p_str)
419                    .map_err(|e| EncodingError::ParseError("min_price", format!("row {i}: {e}")))?,
420            )
421        };
422
423        let margin_init = Decimal::from_str(margin_init_values.value(i))
424            .map_err(|e| EncodingError::ParseError("margin_init", format!("row {i}: {e}")))?;
425        let margin_maint = Decimal::from_str(margin_maint_values.value(i))
426            .map_err(|e| EncodingError::ParseError("margin_maint", format!("row {i}: {e}")))?;
427
428        let info = if info_values.is_null(i) {
429            None
430        } else {
431            let info_json = info_values
432                .as_any()
433                .downcast_ref::<StringArray>()
434                .ok_or_else(|| EncodingError::ParseError("info", format!("row {i}: invalid type")))?
435                .value(i);
436
437            match serde_json::from_str::<Params>(info_json) {
438                Ok(info_dict) => Some(info_dict),
439                Err(e) => {
440                    return Err(EncodingError::ParseError(
441                        "info",
442                        format!("row {i}: failed to deserialize JSON: {e}"),
443                    ));
444                }
445            }
446        };
447
448        let ts_event = UnixNanos::from(ts_event_values.value(i));
449        let ts_init = UnixNanos::from(ts_init_values.value(i));
450
451        let tick_scheme = optional_ustr_value(tick_scheme_values, i);
452
453        let commodity = Commodity::builder()
454            .instrument_id(id)
455            .raw_symbol(raw_symbol)
456            .asset_class(asset_class)
457            .quote_currency(quote_currency)
458            .price_precision(price_prec)
459            .size_precision(size_prec)
460            .price_increment(price_increment)
461            .size_increment(size_increment)
462            .maybe_lot_size(lot_size)
463            .maybe_max_quantity(max_quantity)
464            .maybe_min_quantity(min_quantity)
465            .maybe_max_notional(max_notional)
466            .maybe_min_notional(min_notional)
467            .maybe_max_price(max_price)
468            .maybe_min_price(min_price)
469            .margin_init(margin_init)
470            .margin_maint(margin_maint)
471            .maybe_tick_scheme(tick_scheme)
472            .maybe_info(info)
473            .ts_event(ts_event)
474            .ts_init(ts_init)
475            .build()
476            .map_err(|e| super::instrument_validation_error::<Commodity>(i, e))?;
477
478        result.push(commodity);
479    }
480
481    Ok(result)
482}