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::{collections::HashMap, str::FromStr, sync::Arc};
19
20use arrow::{
21    array::{
22        Array, BinaryArray, BinaryBuilder, StringArray, StringBuilder, UInt8Array, UInt64Array,
23    },
24    datatypes::{DataType, Field, Schema},
25    error::ArrowError,
26    record_batch::RecordBatch,
27};
28use nautilus_core::{Params, UnixNanos};
29use nautilus_model::{
30    enums::AssetClass,
31    identifiers::{InstrumentId, Symbol},
32    instruments::commodity::Commodity,
33    types::{money::Money, price::Price, quantity::Quantity},
34};
35#[allow(unused)]
36use rust_decimal::Decimal;
37#[allow(unused)]
38use serde_json::Value;
39
40use crate::arrow::{
41    ArrowSchemaProvider, EncodeToRecordBatch, EncodingError, KEY_INSTRUMENT_ID,
42    KEY_PRICE_PRECISION, extract_column, extract_column_by_name_or_index,
43    extract_optional_string_column_by_name, optional_ustr_value,
44};
45
46impl ArrowSchemaProvider for Commodity {
47    fn get_schema(metadata: Option<HashMap<String, String>>) -> Schema {
48        let fields = vec![
49            Field::new("id", DataType::Utf8, false),
50            Field::new("raw_symbol", DataType::Utf8, false),
51            Field::new("asset_class", DataType::Utf8, false),
52            Field::new("quote_currency", DataType::Utf8, false),
53            Field::new("price_precision", DataType::UInt8, false),
54            Field::new("size_precision", DataType::UInt8, false),
55            Field::new("price_increment", DataType::Utf8, false),
56            Field::new("size_increment", DataType::Utf8, false),
57            Field::new("lot_size", DataType::Utf8, true), // nullable
58            Field::new("max_quantity", DataType::Utf8, true), // nullable
59            Field::new("min_quantity", DataType::Utf8, true), // nullable
60            Field::new("max_notional", DataType::Utf8, true), // nullable
61            Field::new("min_notional", DataType::Utf8, true), // nullable
62            Field::new("max_price", DataType::Utf8, true), // nullable
63            Field::new("min_price", DataType::Utf8, true), // nullable
64            Field::new("margin_init", DataType::Utf8, false),
65            Field::new("margin_maint", DataType::Utf8, false),
66            Field::new("maker_fee", DataType::Utf8, false),
67            Field::new("taker_fee", DataType::Utf8, false),
68            Field::new("tick_scheme", DataType::Utf8, true),
69            Field::new("info", DataType::Binary, true), // nullable
70            Field::new("ts_event", DataType::UInt64, false),
71            Field::new("ts_init", DataType::UInt64, false),
72        ];
73
74        let mut final_metadata = HashMap::new();
75        final_metadata.insert("class".to_string(), "Commodity".to_string());
76
77        if let Some(meta) = metadata {
78            final_metadata.extend(meta);
79        }
80
81        Schema::new_with_metadata(fields, final_metadata)
82    }
83}
84
85impl EncodeToRecordBatch for Commodity {
86    fn encode_batch(
87        #[allow(unused)] metadata: &HashMap<String, String>,
88        data: &[Self],
89    ) -> Result<RecordBatch, ArrowError> {
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 = BinaryBuilder::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 {
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            // Encode info dict as JSON bytes (matching Python's msgspec.json.encode)
178            if let Some(ref info) = commodity.info {
179                match serde_json::to_vec(info) {
180                    Ok(json_bytes) => {
181                        info_builder.append_value(json_bytes);
182                    }
183                    Err(e) => {
184                        return Err(ArrowError::InvalidArgumentError(format!(
185                            "Failed to serialize info dict to JSON: {e}"
186                        )));
187                    }
188                }
189            } else {
190                info_builder.append_null();
191            }
192
193            ts_event_builder.append_value(commodity.ts_event.as_u64());
194            ts_init_builder.append_value(commodity.ts_init.as_u64());
195        }
196
197        let mut final_metadata = metadata.clone();
198        final_metadata.insert("class".to_string(), "Commodity".to_string());
199
200        RecordBatch::try_new(
201            Self::get_schema(Some(final_metadata)).into(),
202            vec![
203                Arc::new(id_builder.finish()),
204                Arc::new(raw_symbol_builder.finish()),
205                Arc::new(asset_class_builder.finish()),
206                Arc::new(quote_currency_builder.finish()),
207                Arc::new(price_precision_builder.finish()),
208                Arc::new(size_precision_builder.finish()),
209                Arc::new(price_increment_builder.finish()),
210                Arc::new(size_increment_builder.finish()),
211                Arc::new(lot_size_builder.finish()),
212                Arc::new(max_quantity_builder.finish()),
213                Arc::new(min_quantity_builder.finish()),
214                Arc::new(max_notional_builder.finish()),
215                Arc::new(min_notional_builder.finish()),
216                Arc::new(max_price_builder.finish()),
217                Arc::new(min_price_builder.finish()),
218                Arc::new(margin_init_builder.finish()),
219                Arc::new(margin_maint_builder.finish()),
220                Arc::new(maker_fee_builder.finish()),
221                Arc::new(taker_fee_builder.finish()),
222                Arc::new(tick_scheme_builder.finish()),
223                Arc::new(info_builder.finish()),
224                Arc::new(ts_event_builder.finish()),
225                Arc::new(ts_init_builder.finish()),
226            ],
227        )
228    }
229
230    fn metadata(&self) -> HashMap<String, String> {
231        let mut metadata = HashMap::new();
232        metadata.insert(KEY_INSTRUMENT_ID.to_string(), self.id.to_string());
233        metadata.insert(
234            KEY_PRICE_PRECISION.to_string(),
235            self.price_precision.to_string(),
236        );
237        metadata
238    }
239}
240
241/// Helper function to decode Commodity from RecordBatch
242/// (Cannot implement DecodeFromRecordBatch trait due to `Into<Data>` bound)
243///
244/// # Errors
245///
246/// Returns an `EncodingError` if the RecordBatch cannot be decoded.
247pub fn decode_commodity_batch(
248    #[allow(unused)] metadata: &HashMap<String, String>,
249    record_batch: &RecordBatch,
250) -> Result<Vec<Commodity>, EncodingError> {
251    let cols = record_batch.columns();
252    let num_rows = record_batch.num_rows();
253
254    let id_values = extract_column::<StringArray>(cols, "id", 0, DataType::Utf8)?;
255    let raw_symbol_values = extract_column::<StringArray>(cols, "raw_symbol", 1, DataType::Utf8)?;
256    let asset_class_values = extract_column::<StringArray>(cols, "asset_class", 2, DataType::Utf8)?;
257    let quote_currency_values =
258        extract_column::<StringArray>(cols, "quote_currency", 3, DataType::Utf8)?;
259    let price_precision_values =
260        extract_column::<UInt8Array>(cols, "price_precision", 4, DataType::UInt8)?;
261    let size_precision_values =
262        extract_column::<UInt8Array>(cols, "size_precision", 5, DataType::UInt8)?;
263    let price_increment_values =
264        extract_column::<StringArray>(cols, "price_increment", 6, DataType::Utf8)?;
265    let size_increment_values =
266        extract_column::<StringArray>(cols, "size_increment", 7, DataType::Utf8)?;
267    let lot_size_values = cols
268        .get(8)
269        .ok_or_else(|| EncodingError::MissingColumn("lot_size", 8))?;
270    let max_quantity_values = cols
271        .get(9)
272        .ok_or_else(|| EncodingError::MissingColumn("max_quantity", 9))?;
273    let min_quantity_values = cols
274        .get(10)
275        .ok_or_else(|| EncodingError::MissingColumn("min_quantity", 10))?;
276    let max_notional_values = cols
277        .get(11)
278        .ok_or_else(|| EncodingError::MissingColumn("max_notional", 11))?;
279    let min_notional_values = cols
280        .get(12)
281        .ok_or_else(|| EncodingError::MissingColumn("min_notional", 12))?;
282    let max_price_values = cols
283        .get(13)
284        .ok_or_else(|| EncodingError::MissingColumn("max_price", 13))?;
285    let min_price_values = cols
286        .get(14)
287        .ok_or_else(|| EncodingError::MissingColumn("min_price", 14))?;
288    let margin_init_values =
289        extract_column::<StringArray>(cols, "margin_init", 15, DataType::Utf8)?;
290    let margin_maint_values =
291        extract_column::<StringArray>(cols, "margin_maint", 16, DataType::Utf8)?;
292    let maker_fee_values = extract_column::<StringArray>(cols, "maker_fee", 17, DataType::Utf8)?;
293    let taker_fee_values = extract_column::<StringArray>(cols, "taker_fee", 18, DataType::Utf8)?;
294    let tick_scheme_values = extract_optional_string_column_by_name(record_batch, "tick_scheme")?;
295    let info_values =
296        extract_column_by_name_or_index::<BinaryArray>(record_batch, "info", 19, DataType::Binary)?;
297    let ts_event_values = extract_column_by_name_or_index::<UInt64Array>(
298        record_batch,
299        "ts_event",
300        20,
301        DataType::UInt64,
302    )?;
303    let ts_init_values = extract_column_by_name_or_index::<UInt64Array>(
304        record_batch,
305        "ts_init",
306        21,
307        DataType::UInt64,
308    )?;
309
310    let mut result = Vec::with_capacity(num_rows);
311
312    for i in 0..num_rows {
313        let id = InstrumentId::from_str(id_values.value(i))
314            .map_err(|e| EncodingError::ParseError("id", format!("row {i}: {e}")))?;
315        let raw_symbol = Symbol::from(raw_symbol_values.value(i));
316        let asset_class = AssetClass::from_str(asset_class_values.value(i))
317            .map_err(|e| EncodingError::ParseError("asset_class", format!("row {i}: {e}")))?;
318        let quote_currency = super::decode_currency(
319            quote_currency_values.value(i),
320            "quote_currency",
321            "commodity.quote_currency",
322            i,
323        )?;
324        let price_prec = price_precision_values.value(i);
325        let size_prec = size_precision_values.value(i);
326
327        let price_increment = Price::from_str(price_increment_values.value(i))
328            .map_err(|e| EncodingError::ParseError("price_increment", format!("row {i}: {e}")))?;
329        let size_increment = Quantity::from_str(size_increment_values.value(i))
330            .map_err(|e| EncodingError::ParseError("size_increment", format!("row {i}: {e}")))?;
331
332        let lot_size = if lot_size_values.is_null(i) {
333            None
334        } else {
335            let lot_size_str = lot_size_values
336                .as_any()
337                .downcast_ref::<StringArray>()
338                .ok_or_else(|| {
339                    EncodingError::ParseError("lot_size", format!("row {i}: invalid type"))
340                })?
341                .value(i);
342            Some(
343                Quantity::from_str(lot_size_str)
344                    .map_err(|e| EncodingError::ParseError("lot_size", format!("row {i}: {e}")))?,
345            )
346        };
347
348        let max_quantity =
349            if max_quantity_values.is_null(i) {
350                None
351            } else {
352                let max_qty_str = max_quantity_values
353                    .as_any()
354                    .downcast_ref::<StringArray>()
355                    .ok_or_else(|| {
356                        EncodingError::ParseError("max_quantity", format!("row {i}: invalid type"))
357                    })?
358                    .value(i);
359                Some(Quantity::from_str(max_qty_str).map_err(|e| {
360                    EncodingError::ParseError("max_quantity", format!("row {i}: {e}"))
361                })?)
362            };
363
364        let min_quantity =
365            if min_quantity_values.is_null(i) {
366                None
367            } else {
368                let min_qty_str = min_quantity_values
369                    .as_any()
370                    .downcast_ref::<StringArray>()
371                    .ok_or_else(|| {
372                        EncodingError::ParseError("min_quantity", format!("row {i}: invalid type"))
373                    })?
374                    .value(i);
375                Some(Quantity::from_str(min_qty_str).map_err(|e| {
376                    EncodingError::ParseError("min_quantity", format!("row {i}: {e}"))
377                })?)
378            };
379
380        let max_notional =
381            if max_notional_values.is_null(i) {
382                None
383            } else {
384                let max_not_str = max_notional_values
385                    .as_any()
386                    .downcast_ref::<StringArray>()
387                    .ok_or_else(|| {
388                        EncodingError::ParseError("max_notional", format!("row {i}: invalid type"))
389                    })?
390                    .value(i);
391                Some(Money::from_str(max_not_str).map_err(|e| {
392                    EncodingError::ParseError("max_notional", format!("row {i}: {e}"))
393                })?)
394            };
395
396        let min_notional =
397            if min_notional_values.is_null(i) {
398                None
399            } else {
400                let min_not_str = min_notional_values
401                    .as_any()
402                    .downcast_ref::<StringArray>()
403                    .ok_or_else(|| {
404                        EncodingError::ParseError("min_notional", format!("row {i}: invalid type"))
405                    })?
406                    .value(i);
407                Some(Money::from_str(min_not_str).map_err(|e| {
408                    EncodingError::ParseError("min_notional", format!("row {i}: {e}"))
409                })?)
410            };
411
412        let max_price = if max_price_values.is_null(i) {
413            None
414        } else {
415            let max_p_str = max_price_values
416                .as_any()
417                .downcast_ref::<StringArray>()
418                .ok_or_else(|| {
419                    EncodingError::ParseError("max_price", format!("row {i}: invalid type"))
420                })?
421                .value(i);
422            Some(
423                Price::from_str(max_p_str)
424                    .map_err(|e| EncodingError::ParseError("max_price", format!("row {i}: {e}")))?,
425            )
426        };
427
428        let min_price = if min_price_values.is_null(i) {
429            None
430        } else {
431            let min_p_str = min_price_values
432                .as_any()
433                .downcast_ref::<StringArray>()
434                .ok_or_else(|| {
435                    EncodingError::ParseError("min_price", format!("row {i}: invalid type"))
436                })?
437                .value(i);
438            Some(
439                Price::from_str(min_p_str)
440                    .map_err(|e| EncodingError::ParseError("min_price", format!("row {i}: {e}")))?,
441            )
442        };
443
444        let margin_init = Decimal::from_str(margin_init_values.value(i))
445            .map_err(|e| EncodingError::ParseError("margin_init", format!("row {i}: {e}")))?;
446        let margin_maint = Decimal::from_str(margin_maint_values.value(i))
447            .map_err(|e| EncodingError::ParseError("margin_maint", format!("row {i}: {e}")))?;
448        let maker_fee = Decimal::from_str(maker_fee_values.value(i))
449            .map_err(|e| EncodingError::ParseError("maker_fee", format!("row {i}: {e}")))?;
450        let taker_fee = Decimal::from_str(taker_fee_values.value(i))
451            .map_err(|e| EncodingError::ParseError("taker_fee", format!("row {i}: {e}")))?;
452
453        // Decode info dict from JSON bytes (matching Python's msgspec.json.decode)
454        let info = if info_values.is_null(i) {
455            None
456        } else {
457            let info_bytes = info_values
458                .as_any()
459                .downcast_ref::<BinaryArray>()
460                .ok_or_else(|| EncodingError::ParseError("info", format!("row {i}: invalid type")))?
461                .value(i);
462
463            match serde_json::from_slice::<Params>(info_bytes) {
464                Ok(info_dict) => Some(info_dict),
465                Err(e) => {
466                    return Err(EncodingError::ParseError(
467                        "info",
468                        format!("row {i}: failed to deserialize JSON: {e}"),
469                    ));
470                }
471            }
472        };
473
474        let ts_event = UnixNanos::from(ts_event_values.value(i));
475        let ts_init = UnixNanos::from(ts_init_values.value(i));
476
477        let tick_scheme = optional_ustr_value(tick_scheme_values, i);
478
479        let commodity = Commodity::new_checked(
480            id,
481            raw_symbol,
482            asset_class,
483            quote_currency,
484            price_prec,
485            size_prec,
486            price_increment,
487            size_increment,
488            lot_size,
489            max_quantity,
490            min_quantity,
491            max_notional,
492            min_notional,
493            max_price,
494            min_price,
495            Some(margin_init),
496            Some(margin_maint),
497            Some(maker_fee),
498            Some(taker_fee),
499            tick_scheme,
500            info,
501            ts_event,
502            ts_init,
503        )
504        .map_err(|e| super::instrument_validation_error::<Commodity>(i, e))?;
505
506        result.push(commodity);
507    }
508
509    Ok(result)
510}