Total Complexity | 160 |
Total Lines | 1251 |
Duplicated Lines | 0 % |
Complex classes like zipline.data.DataPortal often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
1 | # |
||
51 | class DataPortal(object): |
||
52 | def __init__(self, |
||
53 | env, |
||
54 | equity_daily_reader=None, |
||
55 | equity_minute_reader=None, |
||
56 | future_daily_reader=None, |
||
57 | future_minute_reader=None, |
||
58 | adjustment_reader=None): |
||
59 | self.env = env |
||
60 | |||
61 | self.views = {} |
||
62 | |||
63 | self._asset_finder = env.asset_finder |
||
64 | |||
65 | self._carrays = { |
||
66 | 'open': {}, |
||
67 | 'high': {}, |
||
68 | 'low': {}, |
||
69 | 'close': {}, |
||
70 | 'volume': {}, |
||
71 | 'sid': {}, |
||
72 | 'dt': {}, |
||
73 | } |
||
74 | |||
75 | self._adjustment_reader = adjustment_reader |
||
76 | |||
77 | # caches of sid -> adjustment list |
||
78 | self._splits_dict = {} |
||
79 | self._mergers_dict = {} |
||
80 | self._dividends_dict = {} |
||
81 | |||
82 | # Cache of sid -> the first trading day of an asset, even if that day |
||
83 | # is before 1/2/2002. |
||
84 | self._asset_start_dates = {} |
||
85 | self._asset_end_dates = {} |
||
86 | |||
87 | # Handle extra sources, like Fetcher. |
||
88 | self._augmented_sources_map = {} |
||
89 | self._extra_source_df = None |
||
90 | |||
91 | self.MINUTE_PRICE_ADJUSTMENT_FACTOR = 0.001 |
||
92 | |||
93 | self._equity_daily_reader = equity_daily_reader |
||
94 | self._equity_minute_reader = equity_minute_reader |
||
95 | self._future_daily_reader = future_daily_reader |
||
96 | self._future_minute_reader = future_minute_reader |
||
97 | |||
98 | def handle_extra_source(self, source_df, sim_params): |
||
99 | """ |
||
100 | Extra sources always have a sid column. |
||
101 | |||
102 | We expand the given data (by forward filling) to the full range of |
||
103 | the simulation dates, so that lookup is fast during simulation. |
||
104 | """ |
||
105 | if source_df is None: |
||
106 | return |
||
107 | |||
108 | self._extra_source_df = source_df |
||
109 | |||
110 | # source_df's sid column can either consist of assets we know about |
||
111 | # (such as sid(24)) or of assets we don't know about (such as |
||
112 | # palladium). |
||
113 | # |
||
114 | # In both cases, we break up the dataframe into individual dfs |
||
115 | # that only contain a single asset's information. ie, if source_df |
||
116 | # has data for PALLADIUM and GOLD, we split source_df into two |
||
117 | # dataframes, one for each. (same applies if source_df has data for |
||
118 | # AAPL and IBM). |
||
119 | # |
||
120 | # We then take each child df and reindex it to the simulation's date |
||
121 | # range by forward-filling missing values. this makes reads simpler. |
||
122 | # |
||
123 | # Finally, we store the data. For each column, we store a mapping in |
||
124 | # self.augmented_sources_map from the column to a dictionary of |
||
125 | # asset -> df. In other words, |
||
126 | # self.augmented_sources_map['days_to_cover']['AAPL'] gives us the df |
||
127 | # holding that data. |
||
128 | |||
129 | if sim_params.emission_rate == "daily": |
||
130 | source_date_index = self.env.days_in_range( |
||
131 | start=sim_params.period_start, |
||
132 | end=sim_params.period_end |
||
133 | ) |
||
134 | else: |
||
135 | source_date_index = self.env.minutes_for_days_in_range( |
||
136 | start=sim_params.period_start, |
||
137 | end=sim_params.period_end |
||
138 | ) |
||
139 | |||
140 | # Break the source_df up into one dataframe per sid. This lets |
||
141 | # us (more easily) calculate accurate start/end dates for each sid, |
||
142 | # de-dup data, and expand the data to fit the backtest start/end date. |
||
143 | grouped_by_sid = source_df.groupby(["sid"]) |
||
144 | group_names = grouped_by_sid.groups.keys() |
||
145 | group_dict = {} |
||
146 | for group_name in group_names: |
||
147 | group_dict[group_name] = grouped_by_sid.get_group(group_name) |
||
148 | |||
149 | for identifier, df in iteritems(group_dict): |
||
150 | # Before reindexing, save the earliest and latest dates |
||
151 | earliest_date = df.index[0] |
||
152 | latest_date = df.index[-1] |
||
153 | |||
154 | # Since we know this df only contains a single sid, we can safely |
||
155 | # de-dupe by the index (dt) |
||
156 | df = df.groupby(level=0).last() |
||
157 | |||
158 | # Reindex the dataframe based on the backtest start/end date. |
||
159 | # This makes reads easier during the backtest. |
||
160 | df = df.reindex(index=source_date_index, method='ffill') |
||
161 | |||
162 | if not isinstance(identifier, Asset): |
||
163 | # for fake assets we need to store a start/end date |
||
164 | self._asset_start_dates[identifier] = earliest_date |
||
165 | self._asset_end_dates[identifier] = latest_date |
||
166 | |||
167 | for col_name in df.columns.difference(['sid']): |
||
168 | if col_name not in self._augmented_sources_map: |
||
169 | self._augmented_sources_map[col_name] = {} |
||
170 | |||
171 | self._augmented_sources_map[col_name][identifier] = df |
||
172 | |||
173 | def _open_minute_file(self, field, asset): |
||
174 | sid_str = str(int(asset)) |
||
175 | |||
176 | try: |
||
177 | carray = self._carrays[field][sid_str] |
||
178 | except KeyError: |
||
179 | carray = self._carrays[field][sid_str] = \ |
||
180 | self._get_ctable(asset)[field] |
||
181 | |||
182 | return carray |
||
183 | |||
184 | def _get_ctable(self, asset): |
||
185 | sid = int(asset) |
||
186 | |||
187 | if isinstance(asset, Future): |
||
188 | if self._future_minute_reader.sid_path_func is not None: |
||
189 | path = self._future_minute_reader.sid_path_func( |
||
190 | self._future_minute_reader.rootdir, sid |
||
191 | ) |
||
192 | else: |
||
193 | path = "{0}/{1}.bcolz".format( |
||
194 | self._future_minute_reader.rootdir, sid) |
||
195 | elif isinstance(asset, Equity): |
||
196 | if self._equity_minute_reader.sid_path_func is not None: |
||
197 | path = self._equity_minute_reader.sid_path_func( |
||
198 | self._equity_minute_reader.rootdir, sid |
||
199 | ) |
||
200 | else: |
||
201 | path = "{0}/{1}.bcolz".format( |
||
202 | self._equity_minute_reader.rootdir, sid) |
||
203 | |||
204 | else: |
||
205 | # TODO: Figure out if assets should be allowed if neither, and |
||
206 | # why this code path is being hit. |
||
207 | if self._equity_minute_reader.sid_path_func is not None: |
||
208 | path = self._equity_minute_reader.sid_path_func( |
||
209 | self._equity_minute_reader.rootdir, sid |
||
210 | ) |
||
211 | else: |
||
212 | path = "{0}/{1}.bcolz".format( |
||
213 | self._equity_minute_reader.rootdir, sid) |
||
214 | |||
215 | return bcolz.open(path, mode='r') |
||
216 | |||
217 | def get_last_traded_dt(self, asset, dt, data_frequency): |
||
218 | """ |
||
219 | Given an asset and dt, returns the last traded dt from the viewpoint |
||
220 | of the given dt. |
||
221 | |||
222 | If there is a trade on the dt, the answer is dt provided. |
||
223 | """ |
||
224 | if data_frequency == 'minute': |
||
225 | return self._equity_minute_reader.get_last_traded_dt(asset, dt) |
||
226 | elif data_frequency == 'daily': |
||
227 | return self._equity_daily_reader.get_last_traded_dt(asset, dt) |
||
228 | |||
229 | def get_previous_value(self, asset, field, dt, data_frequency): |
||
230 | """ |
||
231 | Given an asset and a column and a dt, returns the previous value for |
||
232 | the same asset/column pair. If this data portal is in minute mode, |
||
233 | it's the previous minute value, otherwise it's the previous day's |
||
234 | value. |
||
235 | |||
236 | Parameters |
||
237 | --------- |
||
238 | asset : Asset |
||
239 | The asset whose data is desired. |
||
240 | |||
241 | field: string |
||
242 | The desired field of the asset. Valid values are "open", |
||
243 | "open_price", "high", "low", "close", "close_price", "volume", and |
||
244 | "price". |
||
245 | |||
246 | dt: pd.Timestamp |
||
247 | The timestamp from which to go back in time one slot. |
||
248 | |||
249 | data_frequency: string |
||
250 | The frequency of the data to query; i.e. whether the data is |
||
251 | 'daily' or 'minute' bars |
||
252 | |||
253 | Returns |
||
254 | ------- |
||
255 | The value of the desired field at the desired time. |
||
256 | """ |
||
257 | if data_frequency == 'daily': |
||
258 | prev_dt = self.env.previous_trading_day(dt) |
||
259 | elif data_frequency == 'minute': |
||
260 | prev_dt = self.env.previous_market_minute(dt) |
||
261 | |||
262 | return self.get_spot_value(asset, field, prev_dt, data_frequency) |
||
263 | |||
264 | def _check_extra_sources(self, asset, column, day): |
||
265 | # If we have an extra source with a column called "price", only look |
||
266 | # at it if it's on something like palladium and not AAPL (since our |
||
267 | # own price data always wins when dealing with assets). |
||
268 | look_in_augmented_sources = column in self._augmented_sources_map and \ |
||
269 | not (column in BASE_FIELDS and isinstance(asset, Asset)) |
||
270 | |||
271 | if look_in_augmented_sources: |
||
272 | # we're being asked for a field in an extra source |
||
273 | try: |
||
274 | return self._augmented_sources_map[column][asset].\ |
||
275 | loc[day, column] |
||
276 | except: |
||
277 | log.error( |
||
278 | "Could not find value for asset={0}, day={1}," |
||
279 | "column={2}".format( |
||
280 | str(asset), |
||
281 | str(day), |
||
282 | str(column))) |
||
283 | |||
284 | raise KeyError |
||
285 | |||
286 | def get_spot_value(self, asset, field, dt, data_frequency): |
||
287 | """ |
||
288 | Public API method that returns a scalar value representing the value |
||
289 | of the desired asset's field at either the given dt. |
||
290 | |||
291 | Parameters |
||
292 | --------- |
||
293 | asset : Asset |
||
294 | The asset whose data is desired. |
||
295 | |||
296 | field: string |
||
297 | The desired field of the asset. Valid values are "open", |
||
298 | "open_price", "high", "low", "close", "close_price", "volume", and |
||
299 | "price". |
||
300 | |||
301 | dt: pd.Timestamp |
||
302 | The timestamp for the desired value. |
||
303 | |||
304 | data_frequency: string |
||
305 | The frequency of the data to query; i.e. whether the data is |
||
306 | 'daily' or 'minute' bars |
||
307 | |||
308 | Returns |
||
309 | ------- |
||
310 | The value of the desired field at the desired time. |
||
311 | """ |
||
312 | extra_source_val = self._check_extra_sources( |
||
313 | asset, |
||
314 | field, |
||
315 | dt, |
||
316 | ) |
||
317 | |||
318 | if extra_source_val is not None: |
||
319 | return extra_source_val |
||
320 | |||
321 | if field not in BASE_FIELDS: |
||
322 | raise KeyError("Invalid column: " + str(field)) |
||
323 | |||
324 | column_to_use = BASE_FIELDS[field] |
||
325 | |||
326 | if isinstance(asset, int): |
||
327 | asset = self._asset_finder.retrieve_asset(asset) |
||
328 | |||
329 | self._check_is_currently_alive(asset, dt) |
||
330 | |||
331 | if data_frequency == "daily": |
||
332 | day_to_use = dt |
||
333 | day_to_use = normalize_date(day_to_use) |
||
334 | return self._get_daily_data(asset, column_to_use, day_to_use) |
||
335 | else: |
||
336 | if isinstance(asset, Future): |
||
337 | return self._get_minute_spot_value_future( |
||
338 | asset, column_to_use, dt) |
||
339 | else: |
||
340 | return self._get_minute_spot_value( |
||
341 | asset, column_to_use, dt) |
||
342 | |||
343 | def _get_adjusted_value(self, asset, field, dt, |
||
344 | perspective_dt, |
||
345 | data_frequency): |
||
346 | """ |
||
347 | Private method that returns a scalar value representing the value |
||
348 | of the desired asset's field at the given dt with adjustments applied. |
||
349 | |||
350 | Parameters |
||
351 | --------- |
||
352 | asset : Asset |
||
353 | The asset whose data is desired. |
||
354 | |||
355 | field: string |
||
356 | The desired field of the asset. Valid values are "open", |
||
357 | "open_price", "high", "low", "close", "close_price", "volume", and |
||
358 | "price". |
||
359 | |||
360 | dt: pd.Timestamp |
||
361 | The timestamp for the desired value. |
||
362 | |||
363 | perspective_dt : pd.Timestamp |
||
364 | The timestamp from which the data is being viewed back from. |
||
365 | |||
366 | data_frequency: string |
||
367 | The frequency of the data to query; i.e. whether the data is |
||
368 | 'daily' or 'minute' bars |
||
369 | |||
370 | Returns |
||
371 | ------- |
||
372 | The value of the desired field at the desired time. |
||
373 | """ |
||
374 | if isinstance(asset, int): |
||
375 | asset = self._asset_finder.retrieve_asset(asset) |
||
376 | |||
377 | spot_value = self.get_spot_value(asset, field, dt, data_frequency) |
||
378 | |||
379 | if isinstance(asset, Equity): |
||
380 | adjs = [] |
||
381 | split_adjustments = self._get_adjustment_list( |
||
382 | asset, self._splits_dict, "SPLITS" |
||
383 | ) |
||
384 | for adj_dt, adj in split_adjustments: |
||
385 | if adj_dt < dt: |
||
386 | if field != 'volume': |
||
387 | adjs.append(adj) |
||
388 | else: |
||
389 | adjs.append(1.0 / adj) |
||
390 | if adj_dt >= perspective_dt: |
||
391 | break |
||
392 | |||
393 | if field != 'volume': |
||
394 | merger_adjustments = self._get_adjustment_list( |
||
395 | asset, self._mergers_dict, "MERGERS" |
||
396 | ) |
||
397 | for adj_dt, adj in merger_adjustments: |
||
398 | if adj_dt < dt: |
||
399 | adjs.append(adj) |
||
400 | if adj_dt >= perspective_dt: |
||
401 | break |
||
402 | div_adjustments = self._get_adjustment_list( |
||
403 | asset, self._dividends_dict, "DIVIDENDS", |
||
404 | ) |
||
405 | for adj_dt, adj in div_adjustments: |
||
406 | if adj_dt < dt: |
||
407 | adjs.append(adj) |
||
408 | if adj_dt >= perspective_dt: |
||
409 | break |
||
410 | |||
411 | ratio = reduce(mul, adjs, 1.0) |
||
412 | |||
413 | spot_value *= ratio |
||
414 | |||
415 | return spot_value |
||
416 | |||
417 | def _get_minute_spot_value_future(self, asset, column, dt): |
||
418 | # Futures bcolz files have 1440 bars per day (24 hours), 7 days a week. |
||
419 | # The file attributes contain the "start_dt" and "last_dt" fields, |
||
420 | # which represent the time period for this bcolz file. |
||
421 | |||
422 | # The start_dt is midnight of the first day that this future started |
||
423 | # trading. |
||
424 | |||
425 | # figure out the # of minutes between dt and this asset's start_dt |
||
426 | start_date = self._get_asset_start_date(asset) |
||
427 | minute_offset = int((dt - start_date).total_seconds() / 60) |
||
428 | |||
429 | if minute_offset < 0: |
||
430 | # asking for a date that is before the asset's start date, no dice |
||
431 | return 0.0 |
||
432 | |||
433 | # then just index into the bcolz carray at that offset |
||
434 | carray = self._open_minute_file(column, asset) |
||
435 | result = carray[minute_offset] |
||
436 | |||
437 | # if there's missing data, go backwards until we run out of file |
||
438 | while result == 0 and minute_offset > 0: |
||
439 | minute_offset -= 1 |
||
440 | result = carray[minute_offset] |
||
441 | |||
442 | if column != 'volume': |
||
443 | return result * self.MINUTE_PRICE_ADJUSTMENT_FACTOR |
||
444 | else: |
||
445 | return result |
||
446 | |||
447 | def _get_minute_spot_value(self, asset, column, dt): |
||
448 | # if dt is before the first market minute, minute_index |
||
449 | # will be 0. if it's after the last market minute, it'll |
||
450 | # be len(minutes_for_day) |
||
451 | minute_offset_to_use = \ |
||
452 | self._equity_minute_reader._find_position_of_minute(dt) |
||
453 | |||
454 | carray = self._equity_minute_reader._open_minute_file(column, asset) |
||
455 | result = carray[minute_offset_to_use] |
||
456 | |||
457 | if result == 0: |
||
458 | # if the given minute doesn't have data, we need to seek |
||
459 | # backwards until we find data. This makes the data |
||
460 | # forward-filled. |
||
461 | |||
462 | # get this asset's start date, so that we don't look before it. |
||
463 | start_date = self._get_asset_start_date(asset) |
||
464 | start_date_idx = self._equity_minute_reader.trading_days.\ |
||
465 | searchsorted(start_date) |
||
466 | start_day_offset = start_date_idx * 390 |
||
467 | |||
468 | original_start = minute_offset_to_use |
||
469 | |||
470 | while result == 0 and minute_offset_to_use > start_day_offset: |
||
471 | minute_offset_to_use -= 1 |
||
472 | result = carray[minute_offset_to_use] |
||
473 | |||
474 | # once we've found data, we need to check whether it needs |
||
475 | # to be adjusted. |
||
476 | if result != 0: |
||
477 | minutes = self.env.market_minute_window( |
||
478 | start=dt, |
||
479 | count=(original_start - minute_offset_to_use + 1), |
||
480 | step=-1 |
||
481 | ).order() |
||
482 | |||
483 | # only need to check for adjustments if we've gone back |
||
484 | # far enough to cross the day boundary. |
||
485 | if minutes[0].date() != minutes[-1].date(): |
||
486 | # create a np array of size minutes, fill it all with |
||
487 | # the same value. and adjust the array. |
||
488 | arr = np.array([result] * len(minutes), |
||
489 | dtype=np.float64) |
||
490 | self._apply_all_adjustments( |
||
491 | data=arr, |
||
492 | asset=asset, |
||
493 | dts=minutes, |
||
494 | field=column |
||
495 | ) |
||
496 | |||
497 | # The first value of the adjusted array is the value |
||
498 | # we want. |
||
499 | result = arr[0] |
||
500 | |||
501 | if column != 'volume': |
||
502 | return result * self.MINUTE_PRICE_ADJUSTMENT_FACTOR |
||
503 | else: |
||
504 | return result |
||
505 | |||
506 | def _get_daily_data(self, asset, column, dt): |
||
507 | while True: |
||
508 | try: |
||
509 | value = self._equity_daily_reader.spot_price(asset, dt, column) |
||
510 | if value != -1: |
||
511 | return value |
||
512 | else: |
||
513 | dt -= tradingcalendar.trading_day |
||
514 | except NoDataOnDate: |
||
515 | return 0 |
||
516 | |||
517 | def _get_history_daily_window(self, assets, end_dt, bar_count, |
||
518 | field_to_use): |
||
519 | """ |
||
520 | Internal method that returns a dataframe containing history bars |
||
521 | of daily frequency for the given sids. |
||
522 | """ |
||
523 | day_idx = tradingcalendar.trading_days.searchsorted(end_dt.date()) |
||
524 | days_for_window = tradingcalendar.trading_days[ |
||
525 | (day_idx - bar_count + 1):(day_idx + 1)] |
||
526 | |||
527 | if len(assets) == 0: |
||
528 | return pd.DataFrame(None, |
||
529 | index=days_for_window, |
||
530 | columns=None) |
||
531 | |||
532 | data = [] |
||
533 | |||
534 | for asset in assets: |
||
535 | if isinstance(asset, Future): |
||
536 | data.append(self._get_history_daily_window_future( |
||
537 | asset, days_for_window, end_dt, field_to_use |
||
538 | )) |
||
539 | else: |
||
540 | data.append(self._get_history_daily_window_equity( |
||
541 | asset, days_for_window, end_dt, field_to_use |
||
542 | )) |
||
543 | |||
544 | return pd.DataFrame( |
||
545 | np.array(data).T, |
||
546 | index=days_for_window, |
||
547 | columns=assets |
||
548 | ) |
||
549 | |||
550 | def _get_history_daily_window_future(self, asset, days_for_window, |
||
551 | end_dt, column): |
||
552 | # Since we don't have daily bcolz files for futures (yet), use minute |
||
553 | # bars to calculate the daily values. |
||
554 | data = [] |
||
555 | data_groups = [] |
||
556 | |||
557 | # get all the minutes for the days NOT including today |
||
558 | for day in days_for_window[:-1]: |
||
559 | minutes = self.env.market_minutes_for_day(day) |
||
560 | |||
561 | values_for_day = np.zeros(len(minutes), dtype=np.float64) |
||
562 | |||
563 | for idx, minute in enumerate(minutes): |
||
564 | minute_val = self._get_minute_spot_value_future( |
||
565 | asset, column, minute |
||
566 | ) |
||
567 | |||
568 | values_for_day[idx] = minute_val |
||
569 | |||
570 | data_groups.append(values_for_day) |
||
571 | |||
572 | # get the minutes for today |
||
573 | last_day_minutes = pd.date_range( |
||
574 | start=self.env.get_open_and_close(end_dt)[0], |
||
575 | end=end_dt, |
||
576 | freq="T" |
||
577 | ) |
||
578 | |||
579 | values_for_last_day = np.zeros(len(last_day_minutes), dtype=np.float64) |
||
580 | |||
581 | for idx, minute in enumerate(last_day_minutes): |
||
582 | minute_val = self._get_minute_spot_value_future( |
||
583 | asset, column, minute |
||
584 | ) |
||
585 | |||
586 | values_for_last_day[idx] = minute_val |
||
587 | |||
588 | data_groups.append(values_for_last_day) |
||
589 | |||
590 | for group in data_groups: |
||
591 | if len(group) == 0: |
||
592 | continue |
||
593 | |||
594 | if column == 'volume': |
||
595 | data.append(np.sum(group)) |
||
596 | elif column == 'open': |
||
597 | data.append(group[0]) |
||
598 | elif column == 'close': |
||
599 | data.append(group[-1]) |
||
600 | elif column == 'high': |
||
601 | data.append(np.amax(group)) |
||
602 | elif column == 'low': |
||
603 | data.append(np.amin(group)) |
||
604 | |||
605 | return data |
||
606 | |||
607 | def _get_history_daily_window_equity(self, asset, days_for_window, |
||
608 | end_dt, field_to_use): |
||
609 | sid = int(asset) |
||
610 | ends_at_midnight = end_dt.hour == 0 and end_dt.minute == 0 |
||
611 | |||
612 | # get the start and end dates for this sid |
||
613 | end_date = self._get_asset_end_date(asset) |
||
614 | |||
615 | if ends_at_midnight or (days_for_window[-1] > end_date): |
||
616 | # two cases where we use daily data for the whole range: |
||
617 | # 1) the history window ends at midnight utc. |
||
618 | # 2) the last desired day of the window is after the |
||
619 | # last trading day, use daily data for the whole range. |
||
620 | return self._get_daily_window_for_sid( |
||
621 | asset, |
||
622 | field_to_use, |
||
623 | days_for_window, |
||
624 | extra_slot=False |
||
625 | ) |
||
626 | else: |
||
627 | # for the last day of the desired window, use minute |
||
628 | # data and aggregate it. |
||
629 | all_minutes_for_day = self.env.market_minutes_for_day( |
||
630 | pd.Timestamp(end_dt.date())) |
||
631 | |||
632 | last_minute_idx = all_minutes_for_day.searchsorted(end_dt) |
||
633 | |||
634 | # these are the minutes for the partial day |
||
635 | minutes_for_partial_day =\ |
||
636 | all_minutes_for_day[0:(last_minute_idx + 1)] |
||
637 | |||
638 | daily_data = self._get_daily_window_for_sid( |
||
639 | sid, |
||
640 | field_to_use, |
||
641 | days_for_window[0:-1] |
||
642 | ) |
||
643 | |||
644 | minute_data = self._get_minute_window_for_equity( |
||
645 | sid, |
||
646 | field_to_use, |
||
647 | minutes_for_partial_day |
||
648 | ) |
||
649 | |||
650 | if field_to_use == 'volume': |
||
651 | minute_value = np.sum(minute_data) |
||
652 | elif field_to_use == 'open': |
||
653 | minute_value = minute_data[0] |
||
654 | elif field_to_use == 'close': |
||
655 | minute_value = minute_data[-1] |
||
656 | elif field_to_use == 'high': |
||
657 | minute_value = np.amax(minute_data) |
||
658 | elif field_to_use == 'low': |
||
659 | minute_value = np.amin(minute_data) |
||
660 | |||
661 | # append the partial day. |
||
662 | daily_data[-1] = minute_value |
||
663 | |||
664 | return daily_data |
||
665 | |||
666 | def _get_history_minute_window(self, assets, end_dt, bar_count, |
||
667 | field_to_use): |
||
668 | """ |
||
669 | Internal method that returns a dataframe containing history bars |
||
670 | of minute frequency for the given sids. |
||
671 | """ |
||
672 | # get all the minutes for this window |
||
673 | minutes_for_window = self.env.market_minute_window( |
||
674 | end_dt, bar_count, step=-1)[::-1] |
||
675 | |||
676 | first_trading_day = self._equity_minute_reader.first_trading_day |
||
677 | |||
678 | # but then cut it down to only the minutes after |
||
679 | # the first trading day. |
||
680 | modified_minutes_for_window = minutes_for_window[ |
||
681 | minutes_for_window.slice_indexer(first_trading_day)] |
||
682 | |||
683 | modified_minutes_length = len(modified_minutes_for_window) |
||
684 | |||
685 | if modified_minutes_length == 0: |
||
686 | raise ValueError("Cannot calculate history window that ends" |
||
687 | "before 2002-01-02 14:31 UTC!") |
||
688 | |||
689 | data = [] |
||
690 | bars_to_prepend = 0 |
||
691 | nans_to_prepend = None |
||
692 | |||
693 | if modified_minutes_length < bar_count: |
||
694 | first_trading_date = first_trading_day.date() |
||
695 | if modified_minutes_for_window[0].date() == first_trading_date: |
||
696 | # the beginning of the window goes before our global trading |
||
697 | # start date |
||
698 | bars_to_prepend = bar_count - modified_minutes_length |
||
699 | nans_to_prepend = np.repeat(np.nan, bars_to_prepend) |
||
700 | |||
701 | if len(assets) == 0: |
||
702 | return pd.DataFrame( |
||
703 | None, |
||
704 | index=modified_minutes_for_window, |
||
705 | columns=None |
||
706 | ) |
||
707 | |||
708 | for asset in assets: |
||
709 | asset_minute_data = self._get_minute_window_for_asset( |
||
710 | asset, |
||
711 | field_to_use, |
||
712 | modified_minutes_for_window |
||
713 | ) |
||
714 | |||
715 | if bars_to_prepend != 0: |
||
716 | asset_minute_data = np.insert(asset_minute_data, 0, |
||
717 | nans_to_prepend) |
||
718 | |||
719 | data.append(asset_minute_data) |
||
720 | |||
721 | return pd.DataFrame( |
||
722 | np.array(data).T, |
||
723 | index=minutes_for_window, |
||
724 | columns=assets |
||
725 | ) |
||
726 | |||
727 | def get_history_window(self, assets, end_dt, bar_count, frequency, field, |
||
728 | ffill=True): |
||
729 | """ |
||
730 | Public API method that returns a dataframe containing the requested |
||
731 | history window. Data is fully adjusted. |
||
732 | |||
733 | Parameters |
||
734 | --------- |
||
735 | assets : list of zipline.data.Asset objects |
||
736 | The assets whose data is desired. |
||
737 | |||
738 | bar_count: int |
||
739 | The number of bars desired. |
||
740 | |||
741 | frequency: string |
||
742 | "1d" or "1m" |
||
743 | |||
744 | field: string |
||
745 | The desired field of the asset. |
||
746 | |||
747 | ffill: boolean |
||
748 | Forward-fill missing values. Only has effect if field |
||
749 | is 'price'. |
||
750 | |||
751 | Returns |
||
752 | ------- |
||
753 | A dataframe containing the requested data. |
||
754 | """ |
||
755 | try: |
||
756 | field_to_use = BASE_FIELDS[field] |
||
757 | except KeyError: |
||
758 | raise ValueError("Invalid history field: " + str(field)) |
||
759 | |||
760 | # sanity check in case sids were passed in |
||
761 | assets = np.array([ |
||
762 | (self.env.asset_finder.retrieve_asset(asset) if |
||
763 | isinstance(asset, int) else asset) for asset in assets]) |
||
764 | |||
765 | if frequency == "1d": |
||
766 | df = self._get_history_daily_window(assets, end_dt, bar_count, |
||
767 | field_to_use) |
||
768 | elif frequency == "1m": |
||
769 | df = self._get_history_minute_window(assets, end_dt, bar_count, |
||
770 | field_to_use) |
||
771 | else: |
||
772 | raise ValueError("Invalid frequency: {0}".format(frequency)) |
||
773 | |||
774 | # forward-fill if needed |
||
775 | if field == "price" and ffill: |
||
776 | if frequency == "1m": |
||
777 | data_frequency = 'minute' |
||
778 | elif frequency == "1d": |
||
779 | data_frequency = 'daily' |
||
780 | else: |
||
781 | raise Exception( |
||
782 | "Only 1d and 1m are supported for forward-filling.") |
||
783 | |||
784 | dt_to_fill = df.index[0] |
||
785 | |||
786 | perspective_dt = df.index[-1] |
||
787 | assets_with_leading_nan = np.where(pd.isnull(df.iloc[0]))[0] |
||
788 | for missing_loc in assets_with_leading_nan: |
||
789 | asset = assets[missing_loc] |
||
790 | previous_dt = self.get_last_traded_dt( |
||
791 | asset, dt_to_fill, data_frequency) |
||
792 | if pd.isnull(previous_dt): |
||
793 | continue |
||
794 | previous_value = self._get_adjusted_value( |
||
795 | asset, |
||
796 | field, |
||
797 | previous_dt, |
||
798 | perspective_dt, |
||
799 | data_frequency, |
||
800 | ) |
||
801 | df.iloc[0, missing_loc] = previous_value |
||
802 | |||
803 | df.fillna(method='ffill', inplace=True) |
||
804 | |||
805 | return df |
||
806 | |||
807 | def _get_minute_window_for_asset(self, asset, field, minutes_for_window): |
||
808 | """ |
||
809 | Internal method that gets a window of adjusted minute data for an asset |
||
810 | and specified date range. Used to support the history API method for |
||
811 | minute bars. |
||
812 | |||
813 | Missing bars are filled with NaN. |
||
814 | |||
815 | Parameters |
||
816 | ---------- |
||
817 | asset : Asset |
||
818 | The asset whose data is desired. |
||
819 | |||
820 | field: string |
||
821 | The specific field to return. "open", "high", "close_price", etc. |
||
822 | |||
823 | minutes_for_window: pd.DateTimeIndex |
||
824 | The list of minutes representing the desired window. Each minute |
||
825 | is a pd.Timestamp. |
||
826 | |||
827 | Returns |
||
828 | ------- |
||
829 | A numpy array with requested values. |
||
830 | """ |
||
831 | if isinstance(asset, int): |
||
832 | asset = self.env.asset_finder.retrieve_asset(asset) |
||
833 | |||
834 | if isinstance(asset, Future): |
||
835 | return self._get_minute_window_for_future(asset, field, |
||
836 | minutes_for_window) |
||
837 | else: |
||
838 | return self._get_minute_window_for_equity(asset, field, |
||
839 | minutes_for_window) |
||
840 | |||
841 | def _get_minute_window_for_future(self, asset, field, minutes_for_window): |
||
842 | # THIS IS TEMPORARY. For now, we are only exposing futures within |
||
843 | # equity trading hours (9:30 am to 4pm, Eastern). The easiest way to |
||
844 | # do this is to simply do a spot lookup for each desired minute. |
||
845 | return_data = np.zeros(len(minutes_for_window), dtype=np.float64) |
||
846 | for idx, minute in enumerate(minutes_for_window): |
||
847 | return_data[idx] = \ |
||
848 | self._get_minute_spot_value_future(asset, field, minute) |
||
849 | |||
850 | # Note: an improvement could be to find the consecutive runs within |
||
851 | # minutes_for_window, and use them to read the underlying ctable |
||
852 | # more efficiently. |
||
853 | |||
854 | # Once futures are on 24-hour clock, then we can just grab all the |
||
855 | # requested minutes in one shot from the ctable. |
||
856 | |||
857 | # no adjustments for futures, yay. |
||
858 | return return_data |
||
859 | |||
860 | def _get_minute_window_for_equity(self, asset, field, minutes_for_window): |
||
861 | # each sid's minutes are stored in a bcolz file |
||
862 | # the bcolz file has 390 bars per day, starting at 1/2/2002, regardless |
||
863 | # of when the asset started trading and regardless of half days. |
||
864 | # for a half day, the second half is filled with zeroes. |
||
865 | |||
866 | # find the position of start_dt in the entire timeline, go back |
||
867 | # bar_count bars, and that's the unadjusted data |
||
868 | raw_data = self._equity_minute_reader._open_minute_file(field, asset) |
||
869 | |||
870 | try: |
||
871 | start_idx = self._equity_minute_reader._find_position_of_minute( |
||
872 | minutes_for_window[0]) |
||
873 | except KeyError: |
||
874 | start_idx = 0 |
||
875 | |||
876 | try: |
||
877 | end_idx = self._equity_minute_reader._find_position_of_minute( |
||
878 | minutes_for_window[-1]) + 1 |
||
879 | except KeyError: |
||
880 | end_idx = 0 |
||
881 | |||
882 | if end_idx == 0: |
||
883 | # No data to return for minute window. |
||
884 | return np.full(len(minutes_for_window), np.nan) |
||
885 | |||
886 | return_data = np.zeros(len(minutes_for_window), dtype=np.float64) |
||
887 | |||
888 | data_to_copy = raw_data[start_idx:end_idx] |
||
889 | |||
890 | num_minutes = len(minutes_for_window) |
||
891 | |||
892 | # data_to_copy contains all the zeros (from 1pm to 4pm of an early |
||
893 | # close). num_minutes is the number of actual trading minutes. if |
||
894 | # these two have different lengths, that means that we need to trim |
||
895 | # away data due to early closes. |
||
896 | if len(data_to_copy) != num_minutes: |
||
897 | # get a copy of the minutes in Eastern time, since we depend on |
||
898 | # an early close being at 1pm Eastern. |
||
899 | eastern_minutes = minutes_for_window.tz_convert("US/Eastern") |
||
900 | |||
901 | # accumulate a list of indices of the last minute of an early |
||
902 | # close day. For example, if data_to_copy starts at 12:55 pm, and |
||
903 | # there are five minutes of real data before 180 zeroes, we would |
||
904 | # put 5 into last_minute_idx_of_early_close_day, because the fifth |
||
905 | # minute is the last "real" minute of the day. |
||
906 | last_minute_idx_of_early_close_day = [] |
||
907 | for minute_idx, minute_dt in enumerate(eastern_minutes): |
||
908 | if minute_idx == (num_minutes - 1): |
||
909 | break |
||
910 | |||
911 | if minute_dt.hour == 13 and minute_dt.minute == 0: |
||
912 | next_minute = eastern_minutes[minute_idx + 1] |
||
913 | if next_minute.hour != 13: |
||
914 | # minute_dt is the last minute of an early close day |
||
915 | last_minute_idx_of_early_close_day.append(minute_idx) |
||
916 | |||
917 | # spin through the list of early close markers, and use them to |
||
918 | # chop off 180 minutes at a time from data_to_copy. |
||
919 | for idx, early_close_minute_idx in \ |
||
920 | enumerate(last_minute_idx_of_early_close_day): |
||
921 | early_close_minute_idx -= (180 * idx) |
||
922 | data_to_copy = np.delete( |
||
923 | data_to_copy, |
||
924 | range( |
||
925 | early_close_minute_idx + 1, |
||
926 | early_close_minute_idx + 181 |
||
927 | ) |
||
928 | ) |
||
929 | |||
930 | return_data[0:len(data_to_copy)] = data_to_copy |
||
931 | |||
932 | self._apply_all_adjustments( |
||
933 | return_data, |
||
934 | asset, |
||
935 | minutes_for_window, |
||
936 | field, |
||
937 | self.MINUTE_PRICE_ADJUSTMENT_FACTOR |
||
938 | ) |
||
939 | |||
940 | return return_data |
||
941 | |||
942 | def _apply_all_adjustments(self, data, asset, dts, field, |
||
943 | price_adj_factor=1.0): |
||
944 | """ |
||
945 | Internal method that applies all the necessary adjustments on the |
||
946 | given data array. |
||
947 | |||
948 | The adjustments are: |
||
949 | - splits |
||
950 | - if field != "volume": |
||
951 | - mergers |
||
952 | - dividends |
||
953 | - * 0.001 |
||
954 | - any zero fields replaced with NaN |
||
955 | - all values rounded to 3 digits after the decimal point. |
||
956 | |||
957 | Parameters |
||
958 | ---------- |
||
959 | data : np.array |
||
960 | The data to be adjusted. |
||
961 | |||
962 | asset: Asset |
||
963 | The asset whose data is being adjusted. |
||
964 | |||
965 | dts: pd.DateTimeIndex |
||
966 | The list of minutes or days representing the desired window. |
||
967 | |||
968 | field: string |
||
969 | The field whose values are in the data array. |
||
970 | |||
971 | price_adj_factor: float |
||
972 | Factor with which to adjust OHLC values. |
||
973 | Returns |
||
974 | ------- |
||
975 | None. The data array is modified in place. |
||
976 | """ |
||
977 | self._apply_adjustments_to_window( |
||
978 | self._get_adjustment_list( |
||
979 | asset, self._splits_dict, "SPLITS" |
||
980 | ), |
||
981 | data, |
||
982 | dts, |
||
983 | field != 'volume' |
||
984 | ) |
||
985 | |||
986 | if field != 'volume': |
||
987 | self._apply_adjustments_to_window( |
||
988 | self._get_adjustment_list( |
||
989 | asset, self._mergers_dict, "MERGERS" |
||
990 | ), |
||
991 | data, |
||
992 | dts, |
||
993 | True |
||
994 | ) |
||
995 | |||
996 | self._apply_adjustments_to_window( |
||
997 | self._get_adjustment_list( |
||
998 | asset, self._dividends_dict, "DIVIDENDS" |
||
999 | ), |
||
1000 | data, |
||
1001 | dts, |
||
1002 | True |
||
1003 | ) |
||
1004 | |||
1005 | data *= price_adj_factor |
||
1006 | |||
1007 | # if anything is zero, it's a missing bar, so replace it with NaN. |
||
1008 | # we only want to do this for non-volume fields, because a missing |
||
1009 | # volume should be 0. |
||
1010 | data[data == 0] = np.NaN |
||
1011 | |||
1012 | np.around(data, 3, out=data) |
||
1013 | |||
1014 | def _get_daily_window_for_sid(self, asset, field, days_in_window, |
||
1015 | extra_slot=True): |
||
1016 | """ |
||
1017 | Internal method that gets a window of adjusted daily data for a sid |
||
1018 | and specified date range. Used to support the history API method for |
||
1019 | daily bars. |
||
1020 | |||
1021 | Parameters |
||
1022 | ---------- |
||
1023 | asset : Asset |
||
1024 | The asset whose data is desired. |
||
1025 | |||
1026 | start_dt: pandas.Timestamp |
||
1027 | The start of the desired window of data. |
||
1028 | |||
1029 | bar_count: int |
||
1030 | The number of days of data to return. |
||
1031 | |||
1032 | field: string |
||
1033 | The specific field to return. "open", "high", "close_price", etc. |
||
1034 | |||
1035 | extra_slot: boolean |
||
1036 | Whether to allocate an extra slot in the returned numpy array. |
||
1037 | This extra slot will hold the data for the last partial day. It's |
||
1038 | much better to create it here than to create a copy of the array |
||
1039 | later just to add a slot. |
||
1040 | |||
1041 | Returns |
||
1042 | ------- |
||
1043 | A numpy array with requested values. Any missing slots filled with |
||
1044 | nan. |
||
1045 | |||
1046 | """ |
||
1047 | bar_count = len(days_in_window) |
||
1048 | # create an np.array of size bar_count |
||
1049 | if extra_slot: |
||
1050 | return_array = np.zeros((bar_count + 1,)) |
||
1051 | else: |
||
1052 | return_array = np.zeros((bar_count,)) |
||
1053 | |||
1054 | return_array[:] = np.NAN |
||
1055 | |||
1056 | start_date = self._get_asset_start_date(asset) |
||
1057 | end_date = self._get_asset_end_date(asset) |
||
1058 | day_slice = days_in_window.slice_indexer(start_date, end_date) |
||
1059 | active_days = days_in_window[day_slice] |
||
1060 | |||
1061 | if active_days.shape[0]: |
||
1062 | data = self._equity_daily_reader.history_window(field, |
||
1063 | active_days[0], |
||
1064 | active_days[-1], |
||
1065 | asset) |
||
1066 | return_array[day_slice] = data |
||
1067 | self._apply_all_adjustments( |
||
1068 | return_array, |
||
1069 | asset, |
||
1070 | active_days, |
||
1071 | field, |
||
1072 | ) |
||
1073 | |||
1074 | return return_array |
||
1075 | |||
1076 | @staticmethod |
||
1077 | def _apply_adjustments_to_window(adjustments_list, window_data, |
||
1078 | dts_in_window, multiply): |
||
1079 | if len(adjustments_list) == 0: |
||
1080 | return |
||
1081 | |||
1082 | # advance idx to the correct spot in the adjustments list, based on |
||
1083 | # when the window starts |
||
1084 | idx = 0 |
||
1085 | |||
1086 | while idx < len(adjustments_list) and dts_in_window[0] >\ |
||
1087 | adjustments_list[idx][0]: |
||
1088 | idx += 1 |
||
1089 | |||
1090 | # if we've advanced through all the adjustments, then there's nothing |
||
1091 | # to do. |
||
1092 | if idx == len(adjustments_list): |
||
1093 | return |
||
1094 | |||
1095 | while idx < len(adjustments_list): |
||
1096 | adjustment_to_apply = adjustments_list[idx] |
||
1097 | |||
1098 | if adjustment_to_apply[0] > dts_in_window[-1]: |
||
1099 | break |
||
1100 | |||
1101 | range_end = dts_in_window.searchsorted(adjustment_to_apply[0]) |
||
1102 | if multiply: |
||
1103 | window_data[0:range_end] *= adjustment_to_apply[1] |
||
1104 | else: |
||
1105 | window_data[0:range_end] /= adjustment_to_apply[1] |
||
1106 | |||
1107 | idx += 1 |
||
1108 | |||
1109 | def _get_adjustment_list(self, asset, adjustments_dict, table_name): |
||
1110 | """ |
||
1111 | Internal method that returns a list of adjustments for the given sid. |
||
1112 | |||
1113 | Parameters |
||
1114 | ---------- |
||
1115 | asset : Asset |
||
1116 | The asset for which to return adjustments. |
||
1117 | |||
1118 | adjustments_dict: dict |
||
1119 | A dictionary of sid -> list that is used as a cache. |
||
1120 | |||
1121 | table_name: string |
||
1122 | The table that contains this data in the adjustments db. |
||
1123 | |||
1124 | Returns |
||
1125 | ------- |
||
1126 | adjustments: list |
||
1127 | A list of [multiplier, pd.Timestamp], earliest first |
||
1128 | |||
1129 | """ |
||
1130 | if self._adjustment_reader is None: |
||
1131 | return [] |
||
1132 | |||
1133 | sid = int(asset) |
||
1134 | |||
1135 | try: |
||
1136 | adjustments = adjustments_dict[sid] |
||
1137 | except KeyError: |
||
1138 | adjustments = adjustments_dict[sid] = self._adjustment_reader.\ |
||
1139 | get_adjustments_for_sid(table_name, sid) |
||
1140 | |||
1141 | return adjustments |
||
1142 | |||
1143 | def _check_is_currently_alive(self, asset, dt): |
||
1144 | sid = int(asset) |
||
1145 | |||
1146 | if sid not in self._asset_start_dates: |
||
1147 | self._get_asset_start_date(asset) |
||
1148 | |||
1149 | start_date = self._asset_start_dates[sid] |
||
1150 | if self._asset_start_dates[sid] > dt: |
||
1151 | raise NoTradeDataAvailableTooEarly( |
||
1152 | sid=sid, |
||
1153 | dt=normalize_date(dt), |
||
1154 | start_dt=start_date |
||
1155 | ) |
||
1156 | |||
1157 | end_date = self._asset_end_dates[sid] |
||
1158 | if self._asset_end_dates[sid] < dt: |
||
1159 | raise NoTradeDataAvailableTooLate( |
||
1160 | sid=sid, |
||
1161 | dt=normalize_date(dt), |
||
1162 | end_dt=end_date |
||
1163 | ) |
||
1164 | |||
1165 | def _get_asset_start_date(self, asset): |
||
1166 | self._ensure_asset_dates(asset) |
||
1167 | return self._asset_start_dates[asset] |
||
1168 | |||
1169 | def _get_asset_end_date(self, asset): |
||
1170 | self._ensure_asset_dates(asset) |
||
1171 | return self._asset_end_dates[asset] |
||
1172 | |||
1173 | def _ensure_asset_dates(self, asset): |
||
1174 | sid = int(asset) |
||
1175 | |||
1176 | if sid not in self._asset_start_dates: |
||
1177 | self._asset_start_dates[sid] = asset.start_date |
||
1178 | self._asset_end_dates[sid] = asset.end_date |
||
1179 | |||
1180 | def get_splits(self, sids, dt): |
||
1181 | """ |
||
1182 | Returns any splits for the given sids and the given dt. |
||
1183 | |||
1184 | Parameters |
||
1185 | ---------- |
||
1186 | sids : list |
||
1187 | Sids for which we want splits. |
||
1188 | |||
1189 | dt: pd.Timestamp |
||
1190 | The date for which we are checking for splits. Note: this is |
||
1191 | expected to be midnight UTC. |
||
1192 | |||
1193 | Returns |
||
1194 | ------- |
||
1195 | list: List of splits, where each split is a (sid, ratio) tuple. |
||
1196 | """ |
||
1197 | if self._adjustment_reader is None or len(sids) == 0: |
||
1198 | return {} |
||
1199 | |||
1200 | # convert dt to # of seconds since epoch, because that's what we use |
||
1201 | # in the adjustments db |
||
1202 | seconds = int(dt.value / 1e9) |
||
1203 | |||
1204 | splits = self._adjustment_reader.conn.execute( |
||
1205 | "SELECT sid, ratio FROM SPLITS WHERE effective_date = ?", |
||
1206 | (seconds,)).fetchall() |
||
1207 | |||
1208 | sids_set = set(sids) |
||
1209 | splits = [split for split in splits if split[0] in sids_set] |
||
1210 | |||
1211 | return splits |
||
1212 | |||
1213 | def get_stock_dividends(self, sid, trading_days): |
||
1214 | """ |
||
1215 | Returns all the stock dividends for a specific sid that occur |
||
1216 | in the given trading range. |
||
1217 | |||
1218 | Parameters |
||
1219 | ---------- |
||
1220 | sid: int |
||
1221 | The asset whose stock dividends should be returned. |
||
1222 | |||
1223 | trading_days: pd.DatetimeIndex |
||
1224 | The trading range. |
||
1225 | |||
1226 | Returns |
||
1227 | ------- |
||
1228 | list: A list of objects with all relevant attributes populated. |
||
1229 | All timestamp fields are converted to pd.Timestamps. |
||
1230 | """ |
||
1231 | |||
1232 | if self._adjustment_reader is None: |
||
1233 | return [] |
||
1234 | |||
1235 | if len(trading_days) == 0: |
||
1236 | return [] |
||
1237 | |||
1238 | start_dt = trading_days[0].value / 1e9 |
||
1239 | end_dt = trading_days[-1].value / 1e9 |
||
1240 | |||
1241 | dividends = self._adjustment_reader.conn.execute( |
||
1242 | "SELECT * FROM stock_dividend_payouts WHERE sid = ? AND " |
||
1243 | "ex_date > ? AND pay_date < ?", (int(sid), start_dt, end_dt,)).\ |
||
1244 | fetchall() |
||
1245 | |||
1246 | dividend_info = [] |
||
1247 | for dividend_tuple in dividends: |
||
1248 | dividend_info.append({ |
||
1249 | "declared_date": dividend_tuple[1], |
||
1250 | "ex_date": pd.Timestamp(dividend_tuple[2], unit="s"), |
||
1251 | "pay_date": pd.Timestamp(dividend_tuple[3], unit="s"), |
||
1252 | "payment_sid": dividend_tuple[4], |
||
1253 | "ratio": dividend_tuple[5], |
||
1254 | "record_date": pd.Timestamp(dividend_tuple[6], unit="s"), |
||
1255 | "sid": dividend_tuple[7] |
||
1256 | }) |
||
1257 | |||
1258 | return dividend_info |
||
1259 | |||
1260 | def contains(self, asset, field): |
||
1261 | return field in BASE_FIELDS or \ |
||
1262 | (field in self._augmented_sources_map and |
||
1263 | asset in self._augmented_sources_map[field]) |
||
1264 | |||
1265 | def get_fetcher_assets(self, day): |
||
1266 | """ |
||
1267 | Returns a list of assets for the current date, as defined by the |
||
1268 | fetcher data. |
||
1269 | |||
1270 | Notes |
||
1271 | ----- |
||
1272 | Data is forward-filled. If there is no fetcher data defined for day |
||
1273 | N, we use day N-1's data (if available, otherwise we keep going back). |
||
1274 | |||
1275 | Returns |
||
1276 | ------- |
||
1277 | list: a list of Asset objects. |
||
1278 | """ |
||
1279 | # return a list of assets for the current date, as defined by the |
||
1280 | # fetcher source |
||
1281 | if self._extra_source_df is None: |
||
1282 | return [] |
||
1283 | |||
1284 | if day in self._extra_source_df.index: |
||
1285 | date_to_use = day |
||
1286 | else: |
||
1287 | # current day isn't in the fetcher df, go back the last |
||
1288 | # available day |
||
1289 | idx = self._extra_source_df.index.searchsorted(day) |
||
1290 | if idx == 0: |
||
1291 | return [] |
||
1292 | |||
1293 | date_to_use = self._extra_source_df.index[idx - 1] |
||
1294 | |||
1295 | asset_list = self._extra_source_df.loc[date_to_use]["sid"] |
||
1296 | |||
1297 | # make sure they're actually assets |
||
1298 | asset_list = [asset for asset in asset_list |
||
1299 | if isinstance(asset, Asset)] |
||
1300 | |||
1301 | return asset_list |
||
1302 |