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::KEY_CLASS;
36use crate::arrow::{
37    ArrowSchemaProvider, EncodeToRecordBatch, EncodingError, KEY_INSTRUMENT_ID,
38    KEY_PRICE_PRECISION, extract_column, extract_column_by_name_or_index,
39    extract_optional_string_column_by_name, json_string_field, optional_ustr_value,
40    record_batch_with_timestamps, record_batch_with_u64_timestamps, timestamp_data_type,
41};
42
43impl ArrowSchemaProvider for Commodity {
44    fn get_schema(metadata: Option<HashMap<String, String>>) -> Schema {
45        let fields = vec![
46            Field::new("id", DataType::Utf8, false),
47            Field::new("raw_symbol", DataType::Utf8, false),
48            Field::new("asset_class", DataType::Utf8, false),
49            Field::new("quote_currency", DataType::Utf8, false),
50            Field::new("price_precision", DataType::UInt8, false),
51            Field::new("size_precision", DataType::UInt8, false),
52            Field::new("price_increment", DataType::Utf8, false),
53            Field::new("size_increment", DataType::Utf8, false),
54            Field::new("lot_size", DataType::Utf8, true), // nullable
55            Field::new("max_quantity", DataType::Utf8, true), // nullable
56            Field::new("min_quantity", DataType::Utf8, true), // nullable
57            Field::new("max_notional", DataType::Utf8, true), // nullable
58            Field::new("min_notional", DataType::Utf8, true), // nullable
59            Field::new("max_price", DataType::Utf8, true), // nullable
60            Field::new("min_price", DataType::Utf8, true), // nullable
61            Field::new("margin_init", DataType::Utf8, false),
62            Field::new("margin_maint", DataType::Utf8, false),
63            Field::new("maker_fee", DataType::Utf8, false),
64            Field::new("taker_fee", DataType::Utf8, false),
65            Field::new("tick_scheme", DataType::Utf8, true),
66            json_string_field("info", true),
67            Field::new("ts_event", timestamp_data_type(), false),
68            Field::new("ts_init", timestamp_data_type(), false),
69        ];
70
71        let mut final_metadata = HashMap::new();
72        final_metadata.insert(KEY_CLASS.to_string(), "Commodity".to_string());
73
74        if let Some(meta) = metadata {
75            final_metadata.extend(meta);
76        }
77
78        Schema::new_with_metadata(fields, final_metadata)
79    }
80}
81
82impl EncodeToRecordBatch for Commodity {
83    fn encode_batch<T>(
84        #[allow(unused)] metadata: &HashMap<String, String>,
85        data: &[T],
86    ) -> Result<RecordBatch, ArrowError>
87    where
88        T: std::borrow::Borrow<Self>,
89    {
90        let mut id_builder = StringBuilder::new();
91        let mut raw_symbol_builder = StringBuilder::new();
92        let mut asset_class_builder = StringBuilder::new();
93        let mut quote_currency_builder = StringBuilder::new();
94        let mut price_precision_builder = UInt8Array::builder(data.len());
95        let mut size_precision_builder = UInt8Array::builder(data.len());
96        let mut price_increment_builder = StringBuilder::new();
97        let mut size_increment_builder = StringBuilder::new();
98        let mut lot_size_builder = StringBuilder::new();
99        let mut max_quantity_builder = StringBuilder::new();
100        let mut min_quantity_builder = StringBuilder::new();
101        let mut max_notional_builder = StringBuilder::new();
102        let mut min_notional_builder = StringBuilder::new();
103        let mut max_price_builder = StringBuilder::new();
104        let mut min_price_builder = StringBuilder::new();
105        let mut margin_init_builder = StringBuilder::new();
106        let mut margin_maint_builder = StringBuilder::new();
107        let mut maker_fee_builder = StringBuilder::new();
108        let mut taker_fee_builder = StringBuilder::new();
109        let mut tick_scheme_builder = StringBuilder::new();
110        let mut info_builder = StringBuilder::new();
111        let mut ts_event_builder = UInt64Array::builder(data.len());
112        let mut ts_init_builder = UInt64Array::builder(data.len());
113
114        for commodity in data.iter().map(Borrow::borrow) {
115            id_builder.append_value(commodity.id.to_string());
116            raw_symbol_builder.append_value(commodity.raw_symbol);
117            asset_class_builder.append_value(commodity.asset_class);
118            quote_currency_builder.append_value(commodity.quote_currency.to_string());
119            price_precision_builder.append_value(commodity.price_precision);
120            size_precision_builder.append_value(commodity.size_precision);
121            price_increment_builder.append_value(commodity.price_increment.to_string());
122            size_increment_builder.append_value(commodity.size_increment.to_string());
123
124            if let Some(lot_size) = commodity.lot_size {
125                lot_size_builder.append_value(lot_size.to_string());
126            } else {
127                lot_size_builder.append_null();
128            }
129
130            if let Some(max_qty) = commodity.max_quantity {
131                max_quantity_builder.append_value(max_qty.to_string());
132            } else {
133                max_quantity_builder.append_null();
134            }
135
136            if let Some(min_qty) = commodity.min_quantity {
137                min_quantity_builder.append_value(min_qty.to_string());
138            } else {
139                min_quantity_builder.append_null();
140            }
141
142            if let Some(max_not) = commodity.max_notional {
143                max_notional_builder.append_value(max_not.to_string());
144            } else {
145                max_notional_builder.append_null();
146            }
147
148            if let Some(min_not) = commodity.min_notional {
149                min_notional_builder.append_value(min_not.to_string());
150            } else {
151                min_notional_builder.append_null();
152            }
153
154            if let Some(max_p) = commodity.max_price {
155                max_price_builder.append_value(max_p.to_string());
156            } else {
157                max_price_builder.append_null();
158            }
159
160            if let Some(min_p) = commodity.min_price {
161                min_price_builder.append_value(min_p.to_string());
162            } else {
163                min_price_builder.append_null();
164            }
165
166            margin_init_builder.append_value(commodity.margin_init.to_string());
167            margin_maint_builder.append_value(commodity.margin_maint.to_string());
168            maker_fee_builder.append_value(commodity.maker_fee.to_string());
169            taker_fee_builder.append_value(commodity.taker_fee.to_string());
170
171            if let Some(tick_scheme) = commodity.tick_scheme {
172                tick_scheme_builder.append_value(tick_scheme);
173            } else {
174                tick_scheme_builder.append_null();
175            }
176
177            if let Some(ref info) = commodity.info {
178                match serde_json::to_string(info) {
179                    Ok(json) => {
180                        info_builder.append_value(json);
181                    }
182                    Err(e) => {
183                        return Err(ArrowError::InvalidArgumentError(format!(
184                            "Failed to serialize info dict to JSON: {e}"
185                        )));
186                    }
187                }
188            } else {
189                info_builder.append_null();
190            }
191
192            ts_event_builder.append_value(commodity.ts_event.as_u64());
193            ts_init_builder.append_value(commodity.ts_init.as_u64());
194        }
195
196        let mut final_metadata = metadata.clone();
197        final_metadata.insert(KEY_CLASS.to_string(), "Commodity".to_string());
198
199        record_batch_with_timestamps(
200            Self::get_schema(Some(final_metadata)).into(),
201            vec![
202                Arc::new(id_builder.finish()),
203                Arc::new(raw_symbol_builder.finish()),
204                Arc::new(asset_class_builder.finish()),
205                Arc::new(quote_currency_builder.finish()),
206                Arc::new(price_precision_builder.finish()),
207                Arc::new(size_precision_builder.finish()),
208                Arc::new(price_increment_builder.finish()),
209                Arc::new(size_increment_builder.finish()),
210                Arc::new(lot_size_builder.finish()),
211                Arc::new(max_quantity_builder.finish()),
212                Arc::new(min_quantity_builder.finish()),
213                Arc::new(max_notional_builder.finish()),
214                Arc::new(min_notional_builder.finish()),
215                Arc::new(max_price_builder.finish()),
216                Arc::new(min_price_builder.finish()),
217                Arc::new(margin_init_builder.finish()),
218                Arc::new(margin_maint_builder.finish()),
219                Arc::new(maker_fee_builder.finish()),
220                Arc::new(taker_fee_builder.finish()),
221                Arc::new(tick_scheme_builder.finish()),
222                Arc::new(info_builder.finish()),
223                Arc::new(ts_event_builder.finish()),
224                Arc::new(ts_init_builder.finish()),
225            ],
226        )
227    }
228
229    fn metadata(&self) -> HashMap<String, String> {
230        let mut metadata = HashMap::new();
231        metadata.insert(KEY_INSTRUMENT_ID.to_string(), self.id.to_string());
232        metadata.insert(
233            KEY_PRICE_PRECISION.to_string(),
234            self.price_precision.to_string(),
235        );
236        metadata
237    }
238}
239
240/// Decodes [`Commodity`] instruments from a record batch.
241///
242/// Not a [`DecodeFromRecordBatch`] implementation because that trait requires `Into<Data>`.
243///
244/// # Errors
245///
246/// Returns an `EncodingError` if the record batch cannot be decoded.
247///
248/// [`DecodeFromRecordBatch`]: crate::arrow::DecodeFromRecordBatch
249pub fn decode_commodity_batch(
250    #[allow(unused)] metadata: &HashMap<String, String>,
251    record_batch: &RecordBatch,
252) -> Result<Vec<Commodity>, EncodingError> {
253    let record_batch = record_batch_with_u64_timestamps(record_batch)?;
254    let record_batch = &record_batch;
255    let cols = record_batch.columns();
256    let num_rows = record_batch.num_rows();
257
258    let id_values = extract_column::<StringArray>(cols, "id", 0, DataType::Utf8)?;
259    let raw_symbol_values = extract_column::<StringArray>(cols, "raw_symbol", 1, DataType::Utf8)?;
260    let asset_class_values = extract_column::<StringArray>(cols, "asset_class", 2, DataType::Utf8)?;
261    let quote_currency_values =
262        extract_column::<StringArray>(cols, "quote_currency", 3, DataType::Utf8)?;
263    let price_precision_values =
264        extract_column::<UInt8Array>(cols, "price_precision", 4, DataType::UInt8)?;
265    let size_precision_values =
266        extract_column::<UInt8Array>(cols, "size_precision", 5, DataType::UInt8)?;
267    let price_increment_values =
268        extract_column::<StringArray>(cols, "price_increment", 6, DataType::Utf8)?;
269    let size_increment_values =
270        extract_column::<StringArray>(cols, "size_increment", 7, DataType::Utf8)?;
271    let lot_size_values = cols
272        .get(8)
273        .ok_or_else(|| EncodingError::MissingColumn("lot_size", 8))?;
274    let max_quantity_values = cols
275        .get(9)
276        .ok_or_else(|| EncodingError::MissingColumn("max_quantity", 9))?;
277    let min_quantity_values = cols
278        .get(10)
279        .ok_or_else(|| EncodingError::MissingColumn("min_quantity", 10))?;
280    let max_notional_values = cols
281        .get(11)
282        .ok_or_else(|| EncodingError::MissingColumn("max_notional", 11))?;
283    let min_notional_values = cols
284        .get(12)
285        .ok_or_else(|| EncodingError::MissingColumn("min_notional", 12))?;
286    let max_price_values = cols
287        .get(13)
288        .ok_or_else(|| EncodingError::MissingColumn("max_price", 13))?;
289    let min_price_values = cols
290        .get(14)
291        .ok_or_else(|| EncodingError::MissingColumn("min_price", 14))?;
292    let margin_init_values =
293        extract_column::<StringArray>(cols, "margin_init", 15, DataType::Utf8)?;
294    let margin_maint_values =
295        extract_column::<StringArray>(cols, "margin_maint", 16, DataType::Utf8)?;
296    let maker_fee_values = extract_column::<StringArray>(cols, "maker_fee", 17, DataType::Utf8)?;
297    let taker_fee_values = extract_column::<StringArray>(cols, "taker_fee", 18, DataType::Utf8)?;
298    let tick_scheme_values = extract_optional_string_column_by_name(record_batch, "tick_scheme")?;
299    let info_values =
300        extract_column_by_name_or_index::<StringArray>(record_batch, "info", 19, DataType::Utf8)?;
301    let ts_event_values = extract_column_by_name_or_index::<UInt64Array>(
302        record_batch,
303        "ts_event",
304        20,
305        DataType::UInt64,
306    )?;
307    let ts_init_values = extract_column_by_name_or_index::<UInt64Array>(
308        record_batch,
309        "ts_init",
310        21,
311        DataType::UInt64,
312    )?;
313
314    let mut result = Vec::with_capacity(num_rows);
315
316    for i in 0..num_rows {
317        let id = InstrumentId::from_str(id_values.value(i))
318            .map_err(|e| EncodingError::ParseError("id", format!("row {i}: {e}")))?;
319        let raw_symbol = Symbol::from(raw_symbol_values.value(i));
320        let asset_class = AssetClass::from_str(asset_class_values.value(i))
321            .map_err(|e| EncodingError::ParseError("asset_class", format!("row {i}: {e}")))?;
322        let quote_currency = super::decode_currency(
323            quote_currency_values.value(i),
324            "quote_currency",
325            "commodity.quote_currency",
326            i,
327        )?;
328        let price_prec = price_precision_values.value(i);
329        let size_prec = size_precision_values.value(i);
330
331        let price_increment = Price::from_str(price_increment_values.value(i))
332            .map_err(|e| EncodingError::ParseError("price_increment", format!("row {i}: {e}")))?;
333        let size_increment = Quantity::from_str(size_increment_values.value(i))
334            .map_err(|e| EncodingError::ParseError("size_increment", format!("row {i}: {e}")))?;
335
336        let lot_size = if lot_size_values.is_null(i) {
337            None
338        } else {
339            let lot_size_str = lot_size_values
340                .as_any()
341                .downcast_ref::<StringArray>()
342                .ok_or_else(|| {
343                    EncodingError::ParseError("lot_size", format!("row {i}: invalid type"))
344                })?
345                .value(i);
346            Some(
347                Quantity::from_str(lot_size_str)
348                    .map_err(|e| EncodingError::ParseError("lot_size", format!("row {i}: {e}")))?,
349            )
350        };
351
352        let max_quantity =
353            if max_quantity_values.is_null(i) {
354                None
355            } else {
356                let max_qty_str = max_quantity_values
357                    .as_any()
358                    .downcast_ref::<StringArray>()
359                    .ok_or_else(|| {
360                        EncodingError::ParseError("max_quantity", format!("row {i}: invalid type"))
361                    })?
362                    .value(i);
363                Some(Quantity::from_str(max_qty_str).map_err(|e| {
364                    EncodingError::ParseError("max_quantity", format!("row {i}: {e}"))
365                })?)
366            };
367
368        let min_quantity =
369            if min_quantity_values.is_null(i) {
370                None
371            } else {
372                let min_qty_str = min_quantity_values
373                    .as_any()
374                    .downcast_ref::<StringArray>()
375                    .ok_or_else(|| {
376                        EncodingError::ParseError("min_quantity", format!("row {i}: invalid type"))
377                    })?
378                    .value(i);
379                Some(Quantity::from_str(min_qty_str).map_err(|e| {
380                    EncodingError::ParseError("min_quantity", format!("row {i}: {e}"))
381                })?)
382            };
383
384        let max_notional =
385            if max_notional_values.is_null(i) {
386                None
387            } else {
388                let max_not_str = max_notional_values
389                    .as_any()
390                    .downcast_ref::<StringArray>()
391                    .ok_or_else(|| {
392                        EncodingError::ParseError("max_notional", format!("row {i}: invalid type"))
393                    })?
394                    .value(i);
395                Some(Money::from_str(max_not_str).map_err(|e| {
396                    EncodingError::ParseError("max_notional", format!("row {i}: {e}"))
397                })?)
398            };
399
400        let min_notional =
401            if min_notional_values.is_null(i) {
402                None
403            } else {
404                let min_not_str = min_notional_values
405                    .as_any()
406                    .downcast_ref::<StringArray>()
407                    .ok_or_else(|| {
408                        EncodingError::ParseError("min_notional", format!("row {i}: invalid type"))
409                    })?
410                    .value(i);
411                Some(Money::from_str(min_not_str).map_err(|e| {
412                    EncodingError::ParseError("min_notional", format!("row {i}: {e}"))
413                })?)
414            };
415
416        let max_price = if max_price_values.is_null(i) {
417            None
418        } else {
419            let max_p_str = max_price_values
420                .as_any()
421                .downcast_ref::<StringArray>()
422                .ok_or_else(|| {
423                    EncodingError::ParseError("max_price", format!("row {i}: invalid type"))
424                })?
425                .value(i);
426            Some(
427                Price::from_str(max_p_str)
428                    .map_err(|e| EncodingError::ParseError("max_price", format!("row {i}: {e}")))?,
429            )
430        };
431
432        let min_price = if min_price_values.is_null(i) {
433            None
434        } else {
435            let min_p_str = min_price_values
436                .as_any()
437                .downcast_ref::<StringArray>()
438                .ok_or_else(|| {
439                    EncodingError::ParseError("min_price", format!("row {i}: invalid type"))
440                })?
441                .value(i);
442            Some(
443                Price::from_str(min_p_str)
444                    .map_err(|e| EncodingError::ParseError("min_price", format!("row {i}: {e}")))?,
445            )
446        };
447
448        let margin_init = Decimal::from_str(margin_init_values.value(i))
449            .map_err(|e| EncodingError::ParseError("margin_init", format!("row {i}: {e}")))?;
450        let margin_maint = Decimal::from_str(margin_maint_values.value(i))
451            .map_err(|e| EncodingError::ParseError("margin_maint", format!("row {i}: {e}")))?;
452        let maker_fee = Decimal::from_str(maker_fee_values.value(i))
453            .map_err(|e| EncodingError::ParseError("maker_fee", format!("row {i}: {e}")))?;
454        let taker_fee = Decimal::from_str(taker_fee_values.value(i))
455            .map_err(|e| EncodingError::ParseError("taker_fee", format!("row {i}: {e}")))?;
456
457        let info = if info_values.is_null(i) {
458            None
459        } else {
460            let info_json = info_values
461                .as_any()
462                .downcast_ref::<StringArray>()
463                .ok_or_else(|| EncodingError::ParseError("info", format!("row {i}: invalid type")))?
464                .value(i);
465
466            match serde_json::from_str::<Params>(info_json) {
467                Ok(info_dict) => Some(info_dict),
468                Err(e) => {
469                    return Err(EncodingError::ParseError(
470                        "info",
471                        format!("row {i}: failed to deserialize JSON: {e}"),
472                    ));
473                }
474            }
475        };
476
477        let ts_event = UnixNanos::from(ts_event_values.value(i));
478        let ts_init = UnixNanos::from(ts_init_values.value(i));
479
480        let tick_scheme = optional_ustr_value(tick_scheme_values, i);
481
482        let commodity = Commodity::builder()
483            .instrument_id(id)
484            .raw_symbol(raw_symbol)
485            .asset_class(asset_class)
486            .quote_currency(quote_currency)
487            .price_precision(price_prec)
488            .size_precision(size_prec)
489            .price_increment(price_increment)
490            .size_increment(size_increment)
491            .maybe_lot_size(lot_size)
492            .maybe_max_quantity(max_quantity)
493            .maybe_min_quantity(min_quantity)
494            .maybe_max_notional(max_notional)
495            .maybe_min_notional(min_notional)
496            .maybe_max_price(max_price)
497            .maybe_min_price(min_price)
498            .margin_init(margin_init)
499            .margin_maint(margin_maint)
500            .maker_fee(maker_fee)
501            .taker_fee(taker_fee)
502            .maybe_tick_scheme(tick_scheme)
503            .maybe_info(info)
504            .ts_event(ts_event)
505            .ts_init(ts_init)
506            .build()
507            .map_err(|e| super::instrument_validation_error::<Commodity>(i, e))?;
508
509        result.push(commodity);
510    }
511
512    Ok(result)
513}