Skip to main content

nautilus_serialization/arrow/instrument/
tokenized_asset.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 TokenizedAsset 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;
29use nautilus_model::{
30    enums::AssetClass,
31    identifiers::{InstrumentId, Symbol},
32    instruments::tokenized_asset::TokenizedAsset,
33    types::{money::Money, price::Price, quantity::Quantity},
34};
35use rust_decimal::Decimal;
36use ustr::Ustr;
37
38use super::KEY_CLASS;
39use crate::arrow::{
40    ArrowSchemaProvider, EncodeToRecordBatch, EncodingError, KEY_INSTRUMENT_ID,
41    KEY_PRICE_PRECISION, KEY_SIZE_PRECISION, extract_column, extract_column_by_name_or_index,
42    extract_optional_string_column_by_name, optional_ustr_value,
43};
44
45impl ArrowSchemaProvider for TokenizedAsset {
46    fn get_schema(metadata: Option<HashMap<String, String>>) -> Schema {
47        let fields = vec![
48            Field::new("id", DataType::Utf8, false),
49            Field::new("raw_symbol", DataType::Utf8, false),
50            Field::new("asset_class", DataType::Utf8, false),
51            Field::new("base_currency", DataType::Utf8, false),
52            Field::new("quote_currency", DataType::Utf8, false),
53            Field::new("isin", DataType::Utf8, true), // nullable
54            Field::new("price_precision", DataType::UInt8, false),
55            Field::new("size_precision", DataType::UInt8, false),
56            Field::new("price_increment", DataType::Utf8, false),
57            Field::new("size_increment", DataType::Utf8, false),
58            Field::new("multiplier", DataType::Utf8, false),
59            Field::new("lot_size", DataType::Utf8, true), // nullable
60            Field::new("max_quantity", DataType::Utf8, true), // nullable
61            Field::new("min_quantity", DataType::Utf8, true), // nullable
62            Field::new("max_notional", DataType::Utf8, true), // nullable
63            Field::new("min_notional", DataType::Utf8, true), // nullable
64            Field::new("max_price", DataType::Utf8, true), // nullable
65            Field::new("min_price", DataType::Utf8, true), // nullable
66            Field::new("margin_init", DataType::Utf8, false),
67            Field::new("margin_maint", DataType::Utf8, false),
68            Field::new("maker_fee", DataType::Utf8, false),
69            Field::new("taker_fee", DataType::Utf8, false),
70            Field::new("tick_scheme", DataType::Utf8, true),
71            Field::new("info", DataType::Binary, true), // nullable
72            Field::new("ts_event", DataType::UInt64, false),
73            Field::new("ts_init", DataType::UInt64, false),
74        ];
75
76        let mut final_metadata = HashMap::new();
77        final_metadata.insert(KEY_CLASS.to_string(), "TokenizedAsset".to_string());
78
79        if let Some(meta) = metadata {
80            final_metadata.extend(meta);
81        }
82
83        Schema::new_with_metadata(fields, final_metadata)
84    }
85}
86
87impl EncodeToRecordBatch for TokenizedAsset {
88    fn encode_batch(
89        #[allow(unused)] metadata: &HashMap<String, String>,
90        data: &[Self],
91    ) -> Result<RecordBatch, ArrowError> {
92        let mut id_builder = StringBuilder::new();
93        let mut raw_symbol_builder = StringBuilder::new();
94        let mut asset_class_builder = StringBuilder::new();
95        let mut base_currency_builder = StringBuilder::new();
96        let mut quote_currency_builder = StringBuilder::new();
97        let mut isin_builder = StringBuilder::new();
98        let mut price_precision_builder = UInt8Array::builder(data.len());
99        let mut size_precision_builder = UInt8Array::builder(data.len());
100        let mut price_increment_builder = StringBuilder::new();
101        let mut size_increment_builder = StringBuilder::new();
102        let mut multiplier_builder = StringBuilder::new();
103        let mut lot_size_builder = StringBuilder::new();
104        let mut max_quantity_builder = StringBuilder::new();
105        let mut min_quantity_builder = StringBuilder::new();
106        let mut max_notional_builder = StringBuilder::new();
107        let mut min_notional_builder = StringBuilder::new();
108        let mut max_price_builder = StringBuilder::new();
109        let mut min_price_builder = StringBuilder::new();
110        let mut margin_init_builder = StringBuilder::new();
111        let mut margin_maint_builder = StringBuilder::new();
112        let mut maker_fee_builder = StringBuilder::new();
113        let mut taker_fee_builder = StringBuilder::new();
114        let mut tick_scheme_builder = StringBuilder::new();
115        let mut info_builder = BinaryBuilder::new();
116        let mut ts_event_builder = UInt64Array::builder(data.len());
117        let mut ts_init_builder = UInt64Array::builder(data.len());
118
119        for ta in data {
120            id_builder.append_value(ta.id.to_string());
121            raw_symbol_builder.append_value(ta.raw_symbol);
122            asset_class_builder.append_value(ta.asset_class.as_ref());
123            base_currency_builder.append_value(ta.base_currency.to_string());
124            quote_currency_builder.append_value(ta.quote_currency.to_string());
125
126            if let Some(isin) = ta.isin {
127                isin_builder.append_value(isin.as_str());
128            } else {
129                isin_builder.append_null();
130            }
131
132            price_precision_builder.append_value(ta.price_precision);
133            size_precision_builder.append_value(ta.size_precision);
134            price_increment_builder.append_value(ta.price_increment.to_string());
135            size_increment_builder.append_value(ta.size_increment.to_string());
136            multiplier_builder.append_value(ta.multiplier.to_string());
137
138            if let Some(lot_size) = ta.lot_size {
139                lot_size_builder.append_value(lot_size.to_string());
140            } else {
141                lot_size_builder.append_null();
142            }
143
144            if let Some(max_qty) = ta.max_quantity {
145                max_quantity_builder.append_value(max_qty.to_string());
146            } else {
147                max_quantity_builder.append_null();
148            }
149
150            if let Some(min_qty) = ta.min_quantity {
151                min_quantity_builder.append_value(min_qty.to_string());
152            } else {
153                min_quantity_builder.append_null();
154            }
155
156            if let Some(max_not) = ta.max_notional {
157                max_notional_builder.append_value(max_not.to_string());
158            } else {
159                max_notional_builder.append_null();
160            }
161
162            if let Some(min_not) = ta.min_notional {
163                min_notional_builder.append_value(min_not.to_string());
164            } else {
165                min_notional_builder.append_null();
166            }
167
168            if let Some(max_p) = ta.max_price {
169                max_price_builder.append_value(max_p.to_string());
170            } else {
171                max_price_builder.append_null();
172            }
173
174            if let Some(min_p) = ta.min_price {
175                min_price_builder.append_value(min_p.to_string());
176            } else {
177                min_price_builder.append_null();
178            }
179
180            margin_init_builder.append_value(ta.margin_init.to_string());
181            margin_maint_builder.append_value(ta.margin_maint.to_string());
182            maker_fee_builder.append_value(ta.maker_fee.to_string());
183            taker_fee_builder.append_value(ta.taker_fee.to_string());
184
185            if let Some(tick_scheme) = ta.tick_scheme {
186                tick_scheme_builder.append_value(tick_scheme);
187            } else {
188                tick_scheme_builder.append_null();
189            }
190
191            // Encode info dict as JSON bytes (matching Python's msgspec.json.encode)
192            if let Some(ref info) = ta.info {
193                match serde_json::to_vec(info) {
194                    Ok(json_bytes) => {
195                        info_builder.append_value(json_bytes);
196                    }
197                    Err(e) => {
198                        return Err(ArrowError::InvalidArgumentError(format!(
199                            "Failed to serialize info dict to JSON: {e}"
200                        )));
201                    }
202                }
203            } else {
204                info_builder.append_null();
205            }
206
207            ts_event_builder.append_value(ta.ts_event.as_u64());
208            ts_init_builder.append_value(ta.ts_init.as_u64());
209        }
210
211        let mut final_metadata = metadata.clone();
212        final_metadata.insert(KEY_CLASS.to_string(), "TokenizedAsset".to_string());
213
214        RecordBatch::try_new(
215            Self::get_schema(Some(final_metadata)).into(),
216            vec![
217                Arc::new(id_builder.finish()),
218                Arc::new(raw_symbol_builder.finish()),
219                Arc::new(asset_class_builder.finish()),
220                Arc::new(base_currency_builder.finish()),
221                Arc::new(quote_currency_builder.finish()),
222                Arc::new(isin_builder.finish()),
223                Arc::new(price_precision_builder.finish()),
224                Arc::new(size_precision_builder.finish()),
225                Arc::new(price_increment_builder.finish()),
226                Arc::new(size_increment_builder.finish()),
227                Arc::new(multiplier_builder.finish()),
228                Arc::new(lot_size_builder.finish()),
229                Arc::new(max_quantity_builder.finish()),
230                Arc::new(min_quantity_builder.finish()),
231                Arc::new(max_notional_builder.finish()),
232                Arc::new(min_notional_builder.finish()),
233                Arc::new(max_price_builder.finish()),
234                Arc::new(min_price_builder.finish()),
235                Arc::new(margin_init_builder.finish()),
236                Arc::new(margin_maint_builder.finish()),
237                Arc::new(maker_fee_builder.finish()),
238                Arc::new(taker_fee_builder.finish()),
239                Arc::new(tick_scheme_builder.finish()),
240                Arc::new(info_builder.finish()),
241                Arc::new(ts_event_builder.finish()),
242                Arc::new(ts_init_builder.finish()),
243            ],
244        )
245    }
246
247    fn metadata(&self) -> HashMap<String, String> {
248        let mut metadata = HashMap::new();
249        metadata.insert(KEY_INSTRUMENT_ID.to_string(), self.id.to_string());
250        metadata.insert(
251            KEY_PRICE_PRECISION.to_string(),
252            self.price_precision.to_string(),
253        );
254        metadata.insert(
255            KEY_SIZE_PRECISION.to_string(),
256            self.size_precision.to_string(),
257        );
258        metadata
259    }
260}
261
262/// Decodes [`TokenizedAsset`] instruments from a record batch.
263///
264/// Not a [`DecodeFromRecordBatch`] implementation because that trait requires `Into<Data>`.
265///
266/// # Errors
267///
268/// Returns an `EncodingError` if the record batch cannot be decoded.
269///
270/// [`DecodeFromRecordBatch`]: crate::arrow::DecodeFromRecordBatch
271pub fn decode_tokenized_asset_batch(
272    #[allow(unused)] metadata: &HashMap<String, String>,
273    record_batch: &RecordBatch,
274) -> Result<Vec<TokenizedAsset>, EncodingError> {
275    let cols = record_batch.columns();
276    let num_rows = record_batch.num_rows();
277
278    let id_values = extract_column::<StringArray>(cols, "id", 0, DataType::Utf8)?;
279    let raw_symbol_values = extract_column::<StringArray>(cols, "raw_symbol", 1, DataType::Utf8)?;
280    let asset_class_values = extract_column::<StringArray>(cols, "asset_class", 2, DataType::Utf8)?;
281    let base_currency_values =
282        extract_column::<StringArray>(cols, "base_currency", 3, DataType::Utf8)?;
283    let quote_currency_values =
284        extract_column::<StringArray>(cols, "quote_currency", 4, DataType::Utf8)?;
285    let isin_values = cols
286        .get(5)
287        .ok_or_else(|| EncodingError::MissingColumn("isin", 5))?;
288    let price_precision_values =
289        extract_column::<UInt8Array>(cols, "price_precision", 6, DataType::UInt8)?;
290    let size_precision_values =
291        extract_column::<UInt8Array>(cols, "size_precision", 7, DataType::UInt8)?;
292    let price_increment_values =
293        extract_column::<StringArray>(cols, "price_increment", 8, DataType::Utf8)?;
294    let size_increment_values =
295        extract_column::<StringArray>(cols, "size_increment", 9, DataType::Utf8)?;
296    let multiplier_values = extract_column::<StringArray>(cols, "multiplier", 10, DataType::Utf8)?;
297    let lot_size_values = cols
298        .get(11)
299        .ok_or_else(|| EncodingError::MissingColumn("lot_size", 11))?;
300    let max_quantity_values = cols
301        .get(12)
302        .ok_or_else(|| EncodingError::MissingColumn("max_quantity", 12))?;
303    let min_quantity_values = cols
304        .get(13)
305        .ok_or_else(|| EncodingError::MissingColumn("min_quantity", 13))?;
306    let max_notional_values = cols
307        .get(14)
308        .ok_or_else(|| EncodingError::MissingColumn("max_notional", 14))?;
309    let min_notional_values = cols
310        .get(15)
311        .ok_or_else(|| EncodingError::MissingColumn("min_notional", 15))?;
312    let max_price_values = cols
313        .get(16)
314        .ok_or_else(|| EncodingError::MissingColumn("max_price", 16))?;
315    let min_price_values = cols
316        .get(17)
317        .ok_or_else(|| EncodingError::MissingColumn("min_price", 17))?;
318    let margin_init_values =
319        extract_column::<StringArray>(cols, "margin_init", 18, DataType::Utf8)?;
320    let margin_maint_values =
321        extract_column::<StringArray>(cols, "margin_maint", 19, DataType::Utf8)?;
322    let maker_fee_values = extract_column::<StringArray>(cols, "maker_fee", 20, DataType::Utf8)?;
323    let taker_fee_values = extract_column::<StringArray>(cols, "taker_fee", 21, DataType::Utf8)?;
324    let tick_scheme_values = extract_optional_string_column_by_name(record_batch, "tick_scheme")?;
325    let info_values =
326        extract_column_by_name_or_index::<BinaryArray>(record_batch, "info", 22, DataType::Binary)?;
327    let ts_event_values = extract_column_by_name_or_index::<UInt64Array>(
328        record_batch,
329        "ts_event",
330        23,
331        DataType::UInt64,
332    )?;
333    let ts_init_values = extract_column_by_name_or_index::<UInt64Array>(
334        record_batch,
335        "ts_init",
336        24,
337        DataType::UInt64,
338    )?;
339
340    let mut result = Vec::with_capacity(num_rows);
341
342    for i in 0..num_rows {
343        let id = InstrumentId::from_str(id_values.value(i))
344            .map_err(|e| EncodingError::ParseError("id", format!("row {i}: {e}")))?;
345        let raw_symbol = Symbol::from(raw_symbol_values.value(i));
346        let asset_class = AssetClass::from_str(asset_class_values.value(i))
347            .map_err(|e| EncodingError::ParseError("asset_class", format!("row {i}: {e}")))?;
348        let base_currency = super::decode_currency(
349            base_currency_values.value(i),
350            "base_currency",
351            "tokenized_asset.base_currency",
352            i,
353        )?;
354        let quote_currency = super::decode_currency(
355            quote_currency_values.value(i),
356            "quote_currency",
357            "tokenized_asset.quote_currency",
358            i,
359        )?;
360
361        let isin = if isin_values.is_null(i) {
362            None
363        } else {
364            let isin_str = isin_values
365                .as_any()
366                .downcast_ref::<StringArray>()
367                .ok_or_else(|| EncodingError::ParseError("isin", format!("row {i}: invalid type")))?
368                .value(i);
369            Some(Ustr::from(isin_str))
370        };
371
372        let price_prec = price_precision_values.value(i);
373        let size_prec = size_precision_values.value(i);
374
375        let price_increment = Price::from_str(price_increment_values.value(i))
376            .map_err(|e| EncodingError::ParseError("price_increment", format!("row {i}: {e}")))?;
377        let size_increment = Quantity::from_str(size_increment_values.value(i))
378            .map_err(|e| EncodingError::ParseError("size_increment", format!("row {i}: {e}")))?;
379        let multiplier = Quantity::from_str(multiplier_values.value(i))
380            .map_err(|e| EncodingError::ParseError("multiplier", format!("row {i}: {e}")))?;
381
382        let lot_size = if lot_size_values.is_null(i) {
383            None
384        } else {
385            let lot_size_str = lot_size_values
386                .as_any()
387                .downcast_ref::<StringArray>()
388                .ok_or_else(|| {
389                    EncodingError::ParseError("lot_size", format!("row {i}: invalid type"))
390                })?
391                .value(i);
392            Some(
393                Quantity::from_str(lot_size_str)
394                    .map_err(|e| EncodingError::ParseError("lot_size", format!("row {i}: {e}")))?,
395            )
396        };
397
398        let max_quantity =
399            if max_quantity_values.is_null(i) {
400                None
401            } else {
402                let max_qty_str = max_quantity_values
403                    .as_any()
404                    .downcast_ref::<StringArray>()
405                    .ok_or_else(|| {
406                        EncodingError::ParseError("max_quantity", format!("row {i}: invalid type"))
407                    })?
408                    .value(i);
409                Some(Quantity::from_str(max_qty_str).map_err(|e| {
410                    EncodingError::ParseError("max_quantity", format!("row {i}: {e}"))
411                })?)
412            };
413
414        let min_quantity =
415            if min_quantity_values.is_null(i) {
416                None
417            } else {
418                let min_qty_str = min_quantity_values
419                    .as_any()
420                    .downcast_ref::<StringArray>()
421                    .ok_or_else(|| {
422                        EncodingError::ParseError("min_quantity", format!("row {i}: invalid type"))
423                    })?
424                    .value(i);
425                Some(Quantity::from_str(min_qty_str).map_err(|e| {
426                    EncodingError::ParseError("min_quantity", format!("row {i}: {e}"))
427                })?)
428            };
429
430        let max_notional =
431            if max_notional_values.is_null(i) {
432                None
433            } else {
434                let max_not_str = max_notional_values
435                    .as_any()
436                    .downcast_ref::<StringArray>()
437                    .ok_or_else(|| {
438                        EncodingError::ParseError("max_notional", format!("row {i}: invalid type"))
439                    })?
440                    .value(i);
441                Some(Money::from_str(max_not_str).map_err(|e| {
442                    EncodingError::ParseError("max_notional", format!("row {i}: {e}"))
443                })?)
444            };
445
446        let min_notional =
447            if min_notional_values.is_null(i) {
448                None
449            } else {
450                let min_not_str = min_notional_values
451                    .as_any()
452                    .downcast_ref::<StringArray>()
453                    .ok_or_else(|| {
454                        EncodingError::ParseError("min_notional", format!("row {i}: invalid type"))
455                    })?
456                    .value(i);
457                Some(Money::from_str(min_not_str).map_err(|e| {
458                    EncodingError::ParseError("min_notional", format!("row {i}: {e}"))
459                })?)
460            };
461
462        let max_price = if max_price_values.is_null(i) {
463            None
464        } else {
465            let max_p_str = max_price_values
466                .as_any()
467                .downcast_ref::<StringArray>()
468                .ok_or_else(|| {
469                    EncodingError::ParseError("max_price", format!("row {i}: invalid type"))
470                })?
471                .value(i);
472            Some(
473                Price::from_str(max_p_str)
474                    .map_err(|e| EncodingError::ParseError("max_price", format!("row {i}: {e}")))?,
475            )
476        };
477
478        let min_price = if min_price_values.is_null(i) {
479            None
480        } else {
481            let min_p_str = min_price_values
482                .as_any()
483                .downcast_ref::<StringArray>()
484                .ok_or_else(|| {
485                    EncodingError::ParseError("min_price", format!("row {i}: invalid type"))
486                })?
487                .value(i);
488            Some(
489                Price::from_str(min_p_str)
490                    .map_err(|e| EncodingError::ParseError("min_price", format!("row {i}: {e}")))?,
491            )
492        };
493
494        let margin_init = Decimal::from_str(margin_init_values.value(i))
495            .map_err(|e| EncodingError::ParseError("margin_init", format!("row {i}: {e}")))?;
496        let margin_maint = Decimal::from_str(margin_maint_values.value(i))
497            .map_err(|e| EncodingError::ParseError("margin_maint", format!("row {i}: {e}")))?;
498        let maker_fee = Decimal::from_str(maker_fee_values.value(i))
499            .map_err(|e| EncodingError::ParseError("maker_fee", format!("row {i}: {e}")))?;
500        let taker_fee = Decimal::from_str(taker_fee_values.value(i))
501            .map_err(|e| EncodingError::ParseError("taker_fee", format!("row {i}: {e}")))?;
502
503        // Decode info dict from JSON bytes (matching Python's msgspec.json.decode)
504        let info = if info_values.is_null(i) {
505            None
506        } else {
507            let info_bytes = info_values
508                .as_any()
509                .downcast_ref::<BinaryArray>()
510                .ok_or_else(|| EncodingError::ParseError("info", format!("row {i}: invalid type")))?
511                .value(i);
512
513            match serde_json::from_slice::<Params>(info_bytes) {
514                Ok(info_dict) => Some(info_dict),
515                Err(e) => {
516                    return Err(EncodingError::ParseError(
517                        "info",
518                        format!("row {i}: failed to deserialize JSON: {e}"),
519                    ));
520                }
521            }
522        };
523
524        let ts_event = nautilus_core::UnixNanos::from(ts_event_values.value(i));
525        let ts_init = nautilus_core::UnixNanos::from(ts_init_values.value(i));
526
527        let tick_scheme = optional_ustr_value(tick_scheme_values, i);
528
529        let tokenized_asset = TokenizedAsset::builder()
530            .instrument_id(id)
531            .raw_symbol(raw_symbol)
532            .asset_class(asset_class)
533            .base_currency(base_currency)
534            .quote_currency(quote_currency)
535            .maybe_isin(isin)
536            .price_precision(price_prec)
537            .size_precision(size_prec)
538            .price_increment(price_increment)
539            .size_increment(size_increment)
540            .multiplier(multiplier)
541            .maybe_lot_size(lot_size)
542            .maybe_max_quantity(max_quantity)
543            .maybe_min_quantity(min_quantity)
544            .maybe_max_notional(max_notional)
545            .maybe_min_notional(min_notional)
546            .maybe_max_price(max_price)
547            .maybe_min_price(min_price)
548            .margin_init(margin_init)
549            .margin_maint(margin_maint)
550            .maker_fee(maker_fee)
551            .taker_fee(taker_fee)
552            .maybe_tick_scheme(tick_scheme)
553            .maybe_info(info)
554            .ts_event(ts_event)
555            .ts_init(ts_init)
556            .build()
557            .map_err(|e| super::instrument_validation_error::<TokenizedAsset>(i, e))?;
558
559        result.push(tokenized_asset);
560    }
561
562    Ok(result)
563}