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::{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;
27use nautilus_model::{
28    enums::AssetClass,
29    identifiers::{InstrumentId, Symbol},
30    instruments::tokenized_asset::TokenizedAsset,
31    types::{money::Money, price::Price, quantity::Quantity},
32};
33use rust_decimal::Decimal;
34use ustr::Ustr;
35
36use super::validate_required_instrument_columns;
37use crate::arrow::{
38    ArrowSchemaProvider, EncodeToRecordBatch, EncodingError, KEY_INSTRUMENT_ID,
39    KEY_PRICE_PRECISION, KEY_SIZE_PRECISION, extract_column, extract_column_by_name,
40    extract_optional_string_column_by_name, json_string_field, metadata_with_type_name,
41    optional_ustr_value, record_batch_with_timestamps, record_batch_with_u64_timestamps,
42    timestamp_data_type,
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("tick_scheme", DataType::Utf8, true),
69            json_string_field("info", true),
70            Field::new("ts_event", timestamp_data_type(), false),
71            Field::new("ts_init", timestamp_data_type(), false),
72        ];
73
74        Schema::new_with_metadata(fields, metadata_with_type_name("TokenizedAsset", metadata))
75    }
76}
77
78impl EncodeToRecordBatch for TokenizedAsset {
79    fn encode_batch<T>(
80        #[allow(unused)] metadata: &HashMap<String, String>,
81        data: &[T],
82    ) -> Result<RecordBatch, ArrowError>
83    where
84        T: std::borrow::Borrow<Self>,
85    {
86        let mut id_builder = StringBuilder::new();
87        let mut raw_symbol_builder = StringBuilder::new();
88        let mut asset_class_builder = StringBuilder::new();
89        let mut base_currency_builder = StringBuilder::new();
90        let mut quote_currency_builder = StringBuilder::new();
91        let mut isin_builder = StringBuilder::new();
92        let mut price_precision_builder = UInt8Array::builder(data.len());
93        let mut size_precision_builder = UInt8Array::builder(data.len());
94        let mut price_increment_builder = StringBuilder::new();
95        let mut size_increment_builder = StringBuilder::new();
96        let mut multiplier_builder = StringBuilder::new();
97        let mut lot_size_builder = StringBuilder::new();
98        let mut max_quantity_builder = StringBuilder::new();
99        let mut min_quantity_builder = StringBuilder::new();
100        let mut max_notional_builder = StringBuilder::new();
101        let mut min_notional_builder = StringBuilder::new();
102        let mut max_price_builder = StringBuilder::new();
103        let mut min_price_builder = StringBuilder::new();
104        let mut margin_init_builder = StringBuilder::new();
105        let mut margin_maint_builder = StringBuilder::new();
106        let mut tick_scheme_builder = StringBuilder::new();
107        let mut info_builder = StringBuilder::new();
108        let mut ts_event_builder = UInt64Array::builder(data.len());
109        let mut ts_init_builder = UInt64Array::builder(data.len());
110
111        for ta in data.iter().map(Borrow::borrow) {
112            id_builder.append_value(ta.id.to_string());
113            raw_symbol_builder.append_value(ta.raw_symbol);
114            asset_class_builder.append_value(ta.asset_class.as_ref());
115            base_currency_builder.append_value(ta.base_currency.to_string());
116            quote_currency_builder.append_value(ta.quote_currency.to_string());
117
118            if let Some(isin) = ta.isin {
119                isin_builder.append_value(isin.as_str());
120            } else {
121                isin_builder.append_null();
122            }
123
124            price_precision_builder.append_value(ta.price_precision);
125            size_precision_builder.append_value(ta.size_precision);
126            price_increment_builder.append_value(ta.price_increment.to_string());
127            size_increment_builder.append_value(ta.size_increment.to_string());
128            multiplier_builder.append_value(ta.multiplier.to_string());
129
130            if let Some(lot_size) = ta.lot_size {
131                lot_size_builder.append_value(lot_size.to_string());
132            } else {
133                lot_size_builder.append_null();
134            }
135
136            if let Some(max_qty) = ta.max_quantity {
137                max_quantity_builder.append_value(max_qty.to_string());
138            } else {
139                max_quantity_builder.append_null();
140            }
141
142            if let Some(min_qty) = ta.min_quantity {
143                min_quantity_builder.append_value(min_qty.to_string());
144            } else {
145                min_quantity_builder.append_null();
146            }
147
148            if let Some(max_not) = ta.max_notional {
149                max_notional_builder.append_value(max_not.to_string());
150            } else {
151                max_notional_builder.append_null();
152            }
153
154            if let Some(min_not) = ta.min_notional {
155                min_notional_builder.append_value(min_not.to_string());
156            } else {
157                min_notional_builder.append_null();
158            }
159
160            if let Some(max_p) = ta.max_price {
161                max_price_builder.append_value(max_p.to_string());
162            } else {
163                max_price_builder.append_null();
164            }
165
166            if let Some(min_p) = ta.min_price {
167                min_price_builder.append_value(min_p.to_string());
168            } else {
169                min_price_builder.append_null();
170            }
171
172            margin_init_builder.append_value(ta.margin_init.to_string());
173            margin_maint_builder.append_value(ta.margin_maint.to_string());
174
175            if let Some(tick_scheme) = ta.tick_scheme {
176                tick_scheme_builder.append_value(tick_scheme);
177            } else {
178                tick_scheme_builder.append_null();
179            }
180
181            if let Some(ref info) = ta.info {
182                match serde_json::to_string(info) {
183                    Ok(json) => {
184                        info_builder.append_value(json);
185                    }
186                    Err(e) => {
187                        return Err(ArrowError::InvalidArgumentError(format!(
188                            "Failed to serialize info dict to JSON: {e}"
189                        )));
190                    }
191                }
192            } else {
193                info_builder.append_null();
194            }
195
196            ts_event_builder.append_value(ta.ts_event.as_u64());
197            ts_init_builder.append_value(ta.ts_init.as_u64());
198        }
199
200        record_batch_with_timestamps(
201            Self::get_schema(Some(metadata.clone())).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(base_currency_builder.finish()),
207                Arc::new(quote_currency_builder.finish()),
208                Arc::new(isin_builder.finish()),
209                Arc::new(price_precision_builder.finish()),
210                Arc::new(size_precision_builder.finish()),
211                Arc::new(price_increment_builder.finish()),
212                Arc::new(size_increment_builder.finish()),
213                Arc::new(multiplier_builder.finish()),
214                Arc::new(lot_size_builder.finish()),
215                Arc::new(max_quantity_builder.finish()),
216                Arc::new(min_quantity_builder.finish()),
217                Arc::new(max_notional_builder.finish()),
218                Arc::new(min_notional_builder.finish()),
219                Arc::new(max_price_builder.finish()),
220                Arc::new(min_price_builder.finish()),
221                Arc::new(margin_init_builder.finish()),
222                Arc::new(margin_maint_builder.finish()),
223                Arc::new(tick_scheme_builder.finish()),
224                Arc::new(info_builder.finish()),
225                Arc::new(ts_event_builder.finish()),
226                Arc::new(ts_init_builder.finish()),
227            ],
228        )
229    }
230
231    fn metadata(&self) -> HashMap<String, String> {
232        let mut metadata = HashMap::new();
233        metadata.insert(KEY_INSTRUMENT_ID.to_string(), self.id.to_string());
234        metadata.insert(
235            KEY_PRICE_PRECISION.to_string(),
236            self.price_precision.to_string(),
237        );
238        metadata.insert(
239            KEY_SIZE_PRECISION.to_string(),
240            self.size_precision.to_string(),
241        );
242        metadata
243    }
244}
245
246/// Decodes [`TokenizedAsset`] instruments from a record batch.
247///
248/// Not a [`DecodeFromRecordBatch`] implementation because that trait requires `Into<Data>`.
249///
250/// # Errors
251///
252/// Returns an error if the record batch cannot be decoded.
253///
254/// [`DecodeFromRecordBatch`]: crate::arrow::DecodeFromRecordBatch
255pub fn decode_tokenized_asset_batch(
256    metadata: &HashMap<String, String>,
257    record_batch: &RecordBatch,
258) -> Result<Vec<TokenizedAsset>, EncodingError> {
259    validate_required_instrument_columns::<TokenizedAsset>(metadata, record_batch)?;
260    let record_batch = record_batch_with_u64_timestamps(record_batch)?;
261    let record_batch = &record_batch;
262    let cols = record_batch.columns();
263    let num_rows = record_batch.num_rows();
264
265    let id_values = extract_column::<StringArray>(cols, "id", 0, DataType::Utf8)?;
266    let raw_symbol_values = extract_column::<StringArray>(cols, "raw_symbol", 1, DataType::Utf8)?;
267    let asset_class_values = extract_column::<StringArray>(cols, "asset_class", 2, DataType::Utf8)?;
268    let base_currency_values =
269        extract_column::<StringArray>(cols, "base_currency", 3, DataType::Utf8)?;
270    let quote_currency_values =
271        extract_column::<StringArray>(cols, "quote_currency", 4, DataType::Utf8)?;
272    let isin_values = cols
273        .get(5)
274        .ok_or_else(|| EncodingError::MissingColumn("isin", 5))?;
275    let price_precision_values =
276        extract_column::<UInt8Array>(cols, "price_precision", 6, DataType::UInt8)?;
277    let size_precision_values =
278        extract_column::<UInt8Array>(cols, "size_precision", 7, DataType::UInt8)?;
279    let price_increment_values =
280        extract_column::<StringArray>(cols, "price_increment", 8, DataType::Utf8)?;
281    let size_increment_values =
282        extract_column::<StringArray>(cols, "size_increment", 9, DataType::Utf8)?;
283    let multiplier_values = extract_column::<StringArray>(cols, "multiplier", 10, DataType::Utf8)?;
284    let lot_size_values = cols
285        .get(11)
286        .ok_or_else(|| EncodingError::MissingColumn("lot_size", 11))?;
287    let max_quantity_values = cols
288        .get(12)
289        .ok_or_else(|| EncodingError::MissingColumn("max_quantity", 12))?;
290    let min_quantity_values = cols
291        .get(13)
292        .ok_or_else(|| EncodingError::MissingColumn("min_quantity", 13))?;
293    let max_notional_values = cols
294        .get(14)
295        .ok_or_else(|| EncodingError::MissingColumn("max_notional", 14))?;
296    let min_notional_values = cols
297        .get(15)
298        .ok_or_else(|| EncodingError::MissingColumn("min_notional", 15))?;
299    let max_price_values = cols
300        .get(16)
301        .ok_or_else(|| EncodingError::MissingColumn("max_price", 16))?;
302    let min_price_values = cols
303        .get(17)
304        .ok_or_else(|| EncodingError::MissingColumn("min_price", 17))?;
305    let margin_init_values =
306        extract_column::<StringArray>(cols, "margin_init", 18, DataType::Utf8)?;
307    let margin_maint_values =
308        extract_column::<StringArray>(cols, "margin_maint", 19, DataType::Utf8)?;
309    let tick_scheme_values = extract_optional_string_column_by_name(record_batch, "tick_scheme")?;
310    let info_values = extract_column_by_name::<StringArray>(record_batch, "info", DataType::Utf8)?;
311    let ts_event_values =
312        extract_column_by_name::<UInt64Array>(record_batch, "ts_event", DataType::UInt64)?;
313    let ts_init_values =
314        extract_column_by_name::<UInt64Array>(record_batch, "ts_init", DataType::UInt64)?;
315
316    let mut result = Vec::with_capacity(num_rows);
317
318    for i in 0..num_rows {
319        let id = InstrumentId::from_str(id_values.value(i))
320            .map_err(|e| EncodingError::ParseError("id", format!("row {i}: {e}")))?;
321        let raw_symbol = Symbol::from(raw_symbol_values.value(i));
322        let asset_class = AssetClass::from_str(asset_class_values.value(i))
323            .map_err(|e| EncodingError::ParseError("asset_class", format!("row {i}: {e}")))?;
324        let base_currency = super::decode_currency(
325            base_currency_values.value(i),
326            "base_currency",
327            "tokenized_asset.base_currency",
328            i,
329        )?;
330        let quote_currency = super::decode_currency(
331            quote_currency_values.value(i),
332            "quote_currency",
333            "tokenized_asset.quote_currency",
334            i,
335        )?;
336
337        let isin = if isin_values.is_null(i) {
338            None
339        } else {
340            let isin_str = isin_values
341                .as_any()
342                .downcast_ref::<StringArray>()
343                .ok_or_else(|| EncodingError::ParseError("isin", format!("row {i}: invalid type")))?
344                .value(i);
345            Some(Ustr::from(isin_str))
346        };
347
348        let price_prec = price_precision_values.value(i);
349        let size_prec = size_precision_values.value(i);
350
351        let price_increment = Price::from_str(price_increment_values.value(i))
352            .map_err(|e| EncodingError::ParseError("price_increment", format!("row {i}: {e}")))?;
353        let size_increment = Quantity::from_str(size_increment_values.value(i))
354            .map_err(|e| EncodingError::ParseError("size_increment", format!("row {i}: {e}")))?;
355        let multiplier = Quantity::from_str(multiplier_values.value(i))
356            .map_err(|e| EncodingError::ParseError("multiplier", format!("row {i}: {e}")))?;
357
358        let lot_size = if lot_size_values.is_null(i) {
359            None
360        } else {
361            let lot_size_str = lot_size_values
362                .as_any()
363                .downcast_ref::<StringArray>()
364                .ok_or_else(|| {
365                    EncodingError::ParseError("lot_size", format!("row {i}: invalid type"))
366                })?
367                .value(i);
368            Some(
369                Quantity::from_str(lot_size_str)
370                    .map_err(|e| EncodingError::ParseError("lot_size", format!("row {i}: {e}")))?,
371            )
372        };
373
374        let max_quantity =
375            if max_quantity_values.is_null(i) {
376                None
377            } else {
378                let max_qty_str = max_quantity_values
379                    .as_any()
380                    .downcast_ref::<StringArray>()
381                    .ok_or_else(|| {
382                        EncodingError::ParseError("max_quantity", format!("row {i}: invalid type"))
383                    })?
384                    .value(i);
385                Some(Quantity::from_str(max_qty_str).map_err(|e| {
386                    EncodingError::ParseError("max_quantity", format!("row {i}: {e}"))
387                })?)
388            };
389
390        let min_quantity =
391            if min_quantity_values.is_null(i) {
392                None
393            } else {
394                let min_qty_str = min_quantity_values
395                    .as_any()
396                    .downcast_ref::<StringArray>()
397                    .ok_or_else(|| {
398                        EncodingError::ParseError("min_quantity", format!("row {i}: invalid type"))
399                    })?
400                    .value(i);
401                Some(Quantity::from_str(min_qty_str).map_err(|e| {
402                    EncodingError::ParseError("min_quantity", format!("row {i}: {e}"))
403                })?)
404            };
405
406        let max_notional =
407            if max_notional_values.is_null(i) {
408                None
409            } else {
410                let max_not_str = max_notional_values
411                    .as_any()
412                    .downcast_ref::<StringArray>()
413                    .ok_or_else(|| {
414                        EncodingError::ParseError("max_notional", format!("row {i}: invalid type"))
415                    })?
416                    .value(i);
417
418                Some(Money::from_str_exact(max_not_str).map_err(|e| {
419                    EncodingError::ParseError("max_notional", format!("row {i}: {e}"))
420                })?)
421            };
422
423        let min_notional =
424            if min_notional_values.is_null(i) {
425                None
426            } else {
427                let min_not_str = min_notional_values
428                    .as_any()
429                    .downcast_ref::<StringArray>()
430                    .ok_or_else(|| {
431                        EncodingError::ParseError("min_notional", format!("row {i}: invalid type"))
432                    })?
433                    .value(i);
434
435                Some(Money::from_str_exact(min_not_str).map_err(|e| {
436                    EncodingError::ParseError("min_notional", format!("row {i}: {e}"))
437                })?)
438            };
439
440        let max_price = if max_price_values.is_null(i) {
441            None
442        } else {
443            let max_p_str = max_price_values
444                .as_any()
445                .downcast_ref::<StringArray>()
446                .ok_or_else(|| {
447                    EncodingError::ParseError("max_price", format!("row {i}: invalid type"))
448                })?
449                .value(i);
450            Some(
451                Price::from_str(max_p_str)
452                    .map_err(|e| EncodingError::ParseError("max_price", format!("row {i}: {e}")))?,
453            )
454        };
455
456        let min_price = if min_price_values.is_null(i) {
457            None
458        } else {
459            let min_p_str = min_price_values
460                .as_any()
461                .downcast_ref::<StringArray>()
462                .ok_or_else(|| {
463                    EncodingError::ParseError("min_price", format!("row {i}: invalid type"))
464                })?
465                .value(i);
466            Some(
467                Price::from_str(min_p_str)
468                    .map_err(|e| EncodingError::ParseError("min_price", format!("row {i}: {e}")))?,
469            )
470        };
471
472        let margin_init = Decimal::from_str(margin_init_values.value(i))
473            .map_err(|e| EncodingError::ParseError("margin_init", format!("row {i}: {e}")))?;
474        let margin_maint = Decimal::from_str(margin_maint_values.value(i))
475            .map_err(|e| EncodingError::ParseError("margin_maint", format!("row {i}: {e}")))?;
476
477        let info = if info_values.is_null(i) {
478            None
479        } else {
480            let info_json = info_values
481                .as_any()
482                .downcast_ref::<StringArray>()
483                .ok_or_else(|| EncodingError::ParseError("info", format!("row {i}: invalid type")))?
484                .value(i);
485
486            match serde_json::from_str::<Params>(info_json) {
487                Ok(info_dict) => Some(info_dict),
488                Err(e) => {
489                    return Err(EncodingError::ParseError(
490                        "info",
491                        format!("row {i}: failed to deserialize JSON: {e}"),
492                    ));
493                }
494            }
495        };
496
497        let ts_event = nautilus_core::UnixNanos::from(ts_event_values.value(i));
498        let ts_init = nautilus_core::UnixNanos::from(ts_init_values.value(i));
499
500        let tick_scheme = optional_ustr_value(tick_scheme_values, i);
501
502        let tokenized_asset = TokenizedAsset::builder()
503            .instrument_id(id)
504            .raw_symbol(raw_symbol)
505            .asset_class(asset_class)
506            .base_currency(base_currency)
507            .quote_currency(quote_currency)
508            .maybe_isin(isin)
509            .price_precision(price_prec)
510            .size_precision(size_prec)
511            .price_increment(price_increment)
512            .size_increment(size_increment)
513            .multiplier(multiplier)
514            .maybe_lot_size(lot_size)
515            .maybe_max_quantity(max_quantity)
516            .maybe_min_quantity(min_quantity)
517            .maybe_max_notional(max_notional)
518            .maybe_min_notional(min_notional)
519            .maybe_max_price(max_price)
520            .maybe_min_price(min_price)
521            .margin_init(margin_init)
522            .margin_maint(margin_maint)
523            .maybe_tick_scheme(tick_scheme)
524            .maybe_info(info)
525            .ts_event(ts_event)
526            .ts_init(ts_init)
527            .build()
528            .map_err(|e| super::instrument_validation_error::<TokenizedAsset>(i, e))?;
529
530        result.push(tokenized_asset);
531    }
532
533    Ok(result)
534}