Total Complexity | 52 |
Total Lines | 430 |
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 | # |
||
46 | class DataPortal(object): |
||
47 | def __init__(self, |
||
48 | env, |
||
49 | sim_params=None, |
||
50 | equity_daily_reader=None, |
||
51 | equity_minute_reader=None, |
||
52 | future_daily_reader=None, |
||
53 | future_minute_reader=None, |
||
54 | adjustment_reader=None): |
||
55 | self.env = env |
||
56 | |||
57 | # Internal pointers to the current dt (can be minute) and current day. |
||
58 | # In daily mode, they point to the same thing. In minute mode, it's |
||
59 | # useful to have separate pointers to the current day and to the |
||
60 | # current minute. These pointers are updated by the |
||
61 | # AlgorithmSimulator's transform loop. |
||
62 | self.current_dt = None |
||
63 | self.current_day = None |
||
64 | |||
65 | self.views = {} |
||
66 | |||
67 | self._asset_finder = env.asset_finder |
||
68 | |||
69 | self._carrays = { |
||
70 | 'open': {}, |
||
71 | 'high': {}, |
||
72 | 'low': {}, |
||
73 | 'close': {}, |
||
74 | 'volume': {}, |
||
75 | 'sid': {}, |
||
76 | 'dt': {}, |
||
77 | } |
||
78 | |||
79 | self._adjustment_reader = adjustment_reader |
||
80 | |||
81 | # caches of sid -> adjustment list |
||
82 | self._splits_dict = {} |
||
83 | self._mergers_dict = {} |
||
84 | self._dividends_dict = {} |
||
85 | |||
86 | # Cache of sid -> the first trading day of an asset, even if that day |
||
87 | # is before 1/2/2002. |
||
88 | self._asset_start_dates = {} |
||
89 | self._asset_end_dates = {} |
||
90 | |||
91 | # Handle extra sources, like Fetcher. |
||
92 | self._augmented_sources_map = {} |
||
93 | self._extra_source_df = None |
||
94 | |||
95 | self._sim_params = sim_params |
||
96 | if self._sim_params is not None: |
||
97 | self._data_frequency = self._sim_params.data_frequency |
||
98 | else: |
||
99 | self._data_frequency = "minute" |
||
100 | |||
101 | self.MINUTE_PRICE_ADJUSTMENT_FACTOR = 0.001 |
||
102 | |||
103 | self._equity_daily_reader = equity_daily_reader |
||
104 | self._equity_minute_reader = equity_minute_reader |
||
105 | self._future_daily_reader = future_daily_reader |
||
106 | self._future_minute_reader = future_minute_reader |
||
107 | |||
108 | def _open_minute_file(self, field, asset): |
||
109 | sid_str = str(int(asset)) |
||
110 | |||
111 | try: |
||
112 | carray = self._carrays[field][sid_str] |
||
113 | except KeyError: |
||
114 | carray = self._carrays[field][sid_str] = \ |
||
115 | self._get_ctable(asset)[field] |
||
116 | |||
117 | return carray |
||
118 | |||
119 | def _get_ctable(self, asset): |
||
120 | sid = int(asset) |
||
121 | |||
122 | if isinstance(asset, Future): |
||
123 | if self._future_minute_reader.sid_path_func is not None: |
||
124 | path = self._future_minute_reader.sid_path_func( |
||
125 | self._future_minute_reader.rootdir, sid |
||
126 | ) |
||
127 | else: |
||
128 | path = "{0}/{1}.bcolz".format( |
||
129 | self._future_minute_reader.rootdir, sid) |
||
130 | elif isinstance(asset, Equity): |
||
131 | if self._equity_minute_reader.sid_path_func is not None: |
||
132 | path = self._equity_minute_reader.sid_path_func( |
||
133 | self._equity_minute_reader.rootdir, sid |
||
134 | ) |
||
135 | else: |
||
136 | path = "{0}/{1}.bcolz".format( |
||
137 | self._equity_minute_reader.rootdir, sid) |
||
138 | |||
139 | return bcolz.open(path, mode='r') |
||
140 | |||
141 | def get_spot_value(self, asset, field, dt=None): |
||
142 | """ |
||
143 | Public API method that returns a scalar value representing the value |
||
144 | of the desired asset's field at either the given dt, or this data |
||
145 | portal's current_dt. |
||
146 | |||
147 | Parameters |
||
148 | --------- |
||
149 | asset : Asset |
||
150 | The asset whose data is desired.gith |
||
151 | |||
152 | field: string |
||
153 | The desired field of the asset. Valid values are "open", |
||
154 | "open_price", "high", "low", "close", "close_price", "volume", and |
||
155 | "price". |
||
156 | |||
157 | dt: pd.Timestamp |
||
158 | (Optional) The timestamp for the desired value. |
||
159 | |||
160 | Returns |
||
161 | ------- |
||
162 | The value of the desired field at the desired time. |
||
163 | """ |
||
164 | if field not in BASE_FIELDS: |
||
165 | raise KeyError("Invalid column: " + str(field)) |
||
166 | |||
167 | column_to_use = BASE_FIELDS[field] |
||
168 | |||
169 | if isinstance(asset, int): |
||
170 | asset = self._asset_finder.retrieve_asset(asset) |
||
171 | |||
172 | self._check_is_currently_alive(asset, dt) |
||
173 | |||
174 | if self._data_frequency == "daily": |
||
175 | day_to_use = dt or self.current_day |
||
176 | day_to_use = normalize_date(day_to_use) |
||
177 | return self._get_daily_data(asset, column_to_use, day_to_use) |
||
178 | else: |
||
179 | dt_to_use = dt or self.current_dt |
||
180 | |||
181 | if isinstance(asset, Future): |
||
182 | return self._get_minute_spot_value_future( |
||
183 | asset, column_to_use, dt_to_use) |
||
184 | else: |
||
185 | return self._get_minute_spot_value( |
||
186 | asset, column_to_use, dt_to_use) |
||
187 | |||
188 | def _get_minute_spot_value_future(self, asset, column, dt): |
||
189 | # Futures bcolz files have 1440 bars per day (24 hours), 7 days a week. |
||
190 | # The file attributes contain the "start_dt" and "last_dt" fields, |
||
191 | # which represent the time period for this bcolz file. |
||
192 | |||
193 | # The start_dt is midnight of the first day that this future started |
||
194 | # trading. |
||
195 | |||
196 | # figure out the # of minutes between dt and this asset's start_dt |
||
197 | start_date = self._get_asset_start_date(asset) |
||
198 | minute_offset = int((dt - start_date).total_seconds() / 60) |
||
199 | |||
200 | if minute_offset < 0: |
||
201 | # asking for a date that is before the asset's start date, no dice |
||
202 | return 0.0 |
||
203 | |||
204 | # then just index into the bcolz carray at that offset |
||
205 | carray = self._open_minute_file(column, asset) |
||
206 | result = carray[minute_offset] |
||
207 | |||
208 | # if there's missing data, go backwards until we run out of file |
||
209 | while result == 0 and minute_offset > 0: |
||
210 | minute_offset -= 1 |
||
211 | result = carray[minute_offset] |
||
212 | |||
213 | if column != 'volume': |
||
214 | return result * self.MINUTE_PRICE_ADJUSTMENT_FACTOR |
||
215 | else: |
||
216 | return result |
||
217 | |||
218 | def _get_minute_spot_value(self, asset, column, dt): |
||
219 | # if dt is before the first market minute, minute_index |
||
220 | # will be 0. if it's after the last market minute, it'll |
||
221 | # be len(minutes_for_day) |
||
222 | given_day = pd.Timestamp(dt.date(), tz='utc') |
||
223 | day_index = self._equity_minute_reader.trading_days.searchsorted( |
||
224 | given_day) |
||
225 | |||
226 | # if dt is before the first market minute, minute_index |
||
227 | # will be 0. if it's after the last market minute, it'll |
||
228 | # be len(minutes_for_day) |
||
229 | minute_index = self.env.market_minutes_for_day(given_day).\ |
||
230 | searchsorted(dt) |
||
231 | |||
232 | minute_offset_to_use = (day_index * 390) + minute_index |
||
233 | |||
234 | carray = self._equity_minute_reader._open_minute_file(column, asset) |
||
235 | result = carray[minute_offset_to_use] |
||
236 | |||
237 | if result == 0: |
||
238 | # if the given minute doesn't have data, we need to seek |
||
239 | # backwards until we find data. This makes the data |
||
240 | # forward-filled. |
||
241 | |||
242 | # get this asset's start date, so that we don't look before it. |
||
243 | start_date = self._get_asset_start_date(asset) |
||
244 | start_date_idx = self._equity_minute_reader.trading_days.\ |
||
245 | searchsorted(start_date) |
||
246 | start_day_offset = start_date_idx * 390 |
||
247 | |||
248 | original_start = minute_offset_to_use |
||
249 | |||
250 | while result == 0 and minute_offset_to_use > start_day_offset: |
||
251 | minute_offset_to_use -= 1 |
||
252 | result = carray[minute_offset_to_use] |
||
253 | |||
254 | # once we've found data, we need to check whether it needs |
||
255 | # to be adjusted. |
||
256 | if result != 0: |
||
257 | minutes = self.env.market_minute_window( |
||
258 | start=(dt or self.current_dt), |
||
259 | count=(original_start - minute_offset_to_use + 1), |
||
260 | step=-1 |
||
261 | ).order() |
||
262 | |||
263 | # only need to check for adjustments if we've gone back |
||
264 | # far enough to cross the day boundary. |
||
265 | if minutes[0].date() != minutes[-1].date(): |
||
266 | # create a np array of size minutes, fill it all with |
||
267 | # the same value. and adjust the array. |
||
268 | arr = np.array([result] * len(minutes), |
||
269 | dtype=np.float64) |
||
270 | self._apply_all_adjustments( |
||
271 | data=arr, |
||
272 | asset=asset, |
||
273 | dts=minutes, |
||
274 | field=column |
||
275 | ) |
||
276 | |||
277 | # The first value of the adjusted array is the value |
||
278 | # we want. |
||
279 | result = arr[0] |
||
280 | |||
281 | if column != 'volume': |
||
282 | return result * self.MINUTE_PRICE_ADJUSTMENT_FACTOR |
||
283 | else: |
||
284 | return result |
||
285 | |||
286 | def _get_daily_data(self, asset, column, dt): |
||
287 | while True: |
||
288 | try: |
||
289 | value = self._equity_daily_reader.spot_price( |
||
290 | asset, dt, column) |
||
291 | if value != -1: |
||
292 | return value |
||
293 | else: |
||
294 | dt -= tradingcalendar.trading_day |
||
295 | except NoDataOnDate: |
||
296 | return 0 |
||
297 | |||
298 | def _apply_all_adjustments(self, data, asset, dts, field, |
||
299 | price_adj_factor=1.0): |
||
300 | """ |
||
301 | Internal method that applies all the necessary adjustments on the |
||
302 | given data array. |
||
303 | |||
304 | The adjustments are: |
||
305 | - splits |
||
306 | - if field != "volume": |
||
307 | - mergers |
||
308 | - dividends |
||
309 | - * 0.001 |
||
310 | - any zero fields replaced with NaN |
||
311 | - all values rounded to 3 digits after the decimal point. |
||
312 | |||
313 | Parameters |
||
314 | ---------- |
||
315 | data : np.array |
||
316 | The data to be adjusted. |
||
317 | |||
318 | asset: Asset |
||
319 | The asset whose data is being adjusted. |
||
320 | |||
321 | dts: pd.DateTimeIndex |
||
322 | The list of minutes or days representing the desired window. |
||
323 | |||
324 | field: string |
||
325 | The field whose values are in the data array. |
||
326 | |||
327 | price_adj_factor: float |
||
328 | Factor with which to adjust OHLC values. |
||
329 | Returns |
||
330 | ------- |
||
331 | None. The data array is modified in place. |
||
332 | """ |
||
333 | self._apply_adjustments_to_window( |
||
334 | self._get_adjustment_list( |
||
335 | asset, self._splits_dict, "SPLITS" |
||
336 | ), |
||
337 | data, |
||
338 | dts, |
||
339 | field != 'volume' |
||
340 | ) |
||
341 | |||
342 | if field != 'volume': |
||
343 | self._apply_adjustments_to_window( |
||
344 | self._get_adjustment_list( |
||
345 | asset, self._mergers_dict, "MERGERS" |
||
346 | ), |
||
347 | data, |
||
348 | dts, |
||
349 | True |
||
350 | ) |
||
351 | |||
352 | self._apply_adjustments_to_window( |
||
353 | self._get_adjustment_list( |
||
354 | asset, self._dividends_dict, "DIVIDENDS" |
||
355 | ), |
||
356 | data, |
||
357 | dts, |
||
358 | True |
||
359 | ) |
||
360 | |||
361 | data *= price_adj_factor |
||
362 | |||
363 | # if anything is zero, it's a missing bar, so replace it with NaN. |
||
364 | # we only want to do this for non-volume fields, because a missing |
||
365 | # volume should be 0. |
||
366 | data[data == 0] = np.NaN |
||
367 | |||
368 | np.around(data, 3, out=data) |
||
369 | |||
370 | @staticmethod |
||
371 | def _apply_adjustments_to_window(adjustments_list, window_data, |
||
372 | dts_in_window, multiply): |
||
373 | if len(adjustments_list) == 0: |
||
374 | return |
||
375 | |||
376 | # advance idx to the correct spot in the adjustments list, based on |
||
377 | # when the window starts |
||
378 | idx = 0 |
||
379 | |||
380 | while idx < len(adjustments_list) and dts_in_window[0] >\ |
||
381 | adjustments_list[idx][0]: |
||
382 | idx += 1 |
||
383 | |||
384 | # if we've advanced through all the adjustments, then there's nothing |
||
385 | # to do. |
||
386 | if idx == len(adjustments_list): |
||
387 | return |
||
388 | |||
389 | while idx < len(adjustments_list): |
||
390 | adjustment_to_apply = adjustments_list[idx] |
||
391 | |||
392 | if adjustment_to_apply[0] > dts_in_window[-1]: |
||
393 | break |
||
394 | |||
395 | range_end = dts_in_window.searchsorted(adjustment_to_apply[0]) |
||
396 | if multiply: |
||
397 | window_data[0:range_end] *= adjustment_to_apply[1] |
||
398 | else: |
||
399 | window_data[0:range_end] /= adjustment_to_apply[1] |
||
400 | |||
401 | idx += 1 |
||
402 | |||
403 | def _get_adjustment_list(self, asset, adjustments_dict, table_name): |
||
404 | """ |
||
405 | Internal method that returns a list of adjustments for the given sid. |
||
406 | |||
407 | Parameters |
||
408 | ---------- |
||
409 | asset : Asset |
||
410 | The asset for which to return adjustments. |
||
411 | |||
412 | adjustments_dict: dict |
||
413 | A dictionary of sid -> list that is used as a cache. |
||
414 | |||
415 | table_name: string |
||
416 | The table that contains this data in the adjustments db. |
||
417 | |||
418 | Returns |
||
419 | ------- |
||
420 | adjustments: list |
||
421 | A list of [multiplier, pd.Timestamp], earliest first |
||
422 | |||
423 | """ |
||
424 | if self._adjustment_reader is None: |
||
425 | return [] |
||
426 | |||
427 | sid = int(asset) |
||
428 | |||
429 | try: |
||
430 | adjustments = adjustments_dict[sid] |
||
431 | except KeyError: |
||
432 | adjustments = adjustments_dict[sid] = self._adjustment_reader.\ |
||
433 | get_adjustments_for_sid(table_name, sid) |
||
434 | |||
435 | return adjustments |
||
436 | |||
437 | def _check_is_currently_alive(self, asset, dt): |
||
438 | if dt is None: |
||
439 | dt = self.current_day |
||
440 | |||
441 | sid = int(asset) |
||
442 | |||
443 | if sid not in self._asset_start_dates: |
||
444 | self._get_asset_start_date(asset) |
||
445 | |||
446 | start_date = self._asset_start_dates[sid] |
||
447 | if self._asset_start_dates[sid] > dt: |
||
448 | raise NoTradeDataAvailableTooEarly( |
||
449 | sid=sid, |
||
450 | dt=dt, |
||
451 | start_dt=start_date |
||
452 | ) |
||
453 | |||
454 | end_date = self._asset_end_dates[sid] |
||
455 | if self._asset_end_dates[sid] < dt: |
||
456 | raise NoTradeDataAvailableTooLate( |
||
457 | sid=sid, |
||
458 | dt=dt, |
||
459 | end_dt=end_date |
||
460 | ) |
||
461 | |||
462 | def _get_asset_start_date(self, asset): |
||
463 | self._ensure_asset_dates(asset) |
||
464 | return self._asset_start_dates[asset] |
||
465 | |||
466 | def _get_asset_end_date(self, asset): |
||
467 | self._ensure_asset_dates(asset) |
||
468 | return self._asset_end_dates[asset] |
||
469 | |||
470 | def _ensure_asset_dates(self, asset): |
||
471 | sid = int(asset) |
||
472 | |||
473 | if sid not in self._asset_start_dates: |
||
474 | self._asset_start_dates[sid] = asset.start_date |
||
475 | self._asset_end_dates[sid] = asset.end_date |
||
476 |