|
| 1 | +from datetime import datetime as dt |
| 2 | +import logging |
| 3 | + |
| 4 | +import pandas as pd |
| 5 | + |
| 6 | +import bson |
| 7 | +import pymongo |
| 8 | +from ..decorators import mongo_retry |
| 9 | +from ..exceptions import NoDataFoundException |
| 10 | + |
| 11 | +from .bson_store import BSONStore |
| 12 | + |
| 13 | +logger = logging.getLogger(__name__) |
| 14 | + |
| 15 | +METADATA_STORE_TYPE = 'MetadataStore' |
| 16 | + |
| 17 | +class MetadataStore(BSONStore): |
| 18 | + """ |
| 19 | + Metadata Store. This stores metadata with timestamps to allow temporal queries. |
| 20 | +
|
| 21 | + Entries are stored in the following format: |
| 22 | + 'symbol': symbol name |
| 23 | + 'metadata': metadata to be persisted |
| 24 | + 'start_time': when entry becomes effective |
| 25 | + 'end_time': (Optional) when entry expires. If not set, it is still in effect |
| 26 | +
|
| 27 | + For each symbol end_time of a entry should match start_time of the next one except for the current entry. |
| 28 | + """ |
| 29 | + |
| 30 | + @classmethod |
| 31 | + def initialize_library(cls, arctic_lib, hashed=True, **kwargs): |
| 32 | + MetadataStore(arctic_lib)._ensure_index() |
| 33 | + BSONStore.initialize_library(arctic_lib, hashed, **kwargs) |
| 34 | + |
| 35 | + @mongo_retry |
| 36 | + def _ensure_index(self): |
| 37 | + self.create_index([('symbol', pymongo.ASCENDING), ('start_time', pymongo.DESCENDING)], |
| 38 | + unique=True, background=True) |
| 39 | + |
| 40 | + def __init__(self, arctic_lib): |
| 41 | + self._arctic_lib = arctic_lib |
| 42 | + self._reset() |
| 43 | + |
| 44 | + def _reset(self): |
| 45 | + self._collection = self._arctic_lib.get_top_level_collection().metadata |
| 46 | + |
| 47 | + @mongo_retry |
| 48 | + def list_symbols(self): |
| 49 | + return self.distinct('symbol') |
| 50 | + |
| 51 | + @mongo_retry |
| 52 | + def has_symbol(self, symbol): |
| 53 | + return self.find_one({'symbol': symbol}) is not None |
| 54 | + |
| 55 | + @mongo_retry |
| 56 | + def read_history(self, symbol): |
| 57 | + """ |
| 58 | + Return all metadata saved for `symbol` |
| 59 | +
|
| 60 | + Parameters |
| 61 | + ---------- |
| 62 | + symbol : `str` |
| 63 | + symbol name for the item |
| 64 | +
|
| 65 | + Returns |
| 66 | + ------- |
| 67 | + pandas.DateFrame containing timestamps and metadata entries |
| 68 | + """ |
| 69 | + find = self.find({'symbol': symbol}, sort=[('start_time', pymongo.ASCENDING)]) |
| 70 | + times = [] |
| 71 | + entries = [] |
| 72 | + for item in find: |
| 73 | + times.append(item['start_time']) |
| 74 | + entries.append(item['metadata']) |
| 75 | + return pd.DataFrame({symbol: entries}, times) |
| 76 | + |
| 77 | + @mongo_retry |
| 78 | + def read(self, symbol): |
| 79 | + """ |
| 80 | + Return current metadata saved for `symbol` |
| 81 | +
|
| 82 | + Parameters |
| 83 | + ---------- |
| 84 | + symbol : `str` |
| 85 | + symbol name for the item |
| 86 | +
|
| 87 | + Returns |
| 88 | + ------- |
| 89 | + metadata |
| 90 | + """ |
| 91 | + res = self.find_one({'symbol': symbol}, sort=[('start_time', pymongo.DESCENDING)]) |
| 92 | + return res['metadata'] if res is not None else None |
| 93 | + |
| 94 | + |
| 95 | + def write_history(self, collection): |
| 96 | + """ |
| 97 | + Manually overwrite entire metadata history for symbols in `collection` |
| 98 | +
|
| 99 | + Parameters |
| 100 | + ---------- |
| 101 | + collection : `list of pandas.DataFrame` |
| 102 | + with symbol names as headers and timestamps as indices |
| 103 | + (the same format as output of read_history) |
| 104 | + Example: |
| 105 | + [pandas.DataFrame({'symbol': [{}]}, [datetime.datetime.utcnow()])] |
| 106 | + """ |
| 107 | + documents = [] |
| 108 | + for dataframe in collection: |
| 109 | + if len(dataframe.columns) != 1: |
| 110 | + raise ValueError('More than one symbol found in a DataFrame') |
| 111 | + symbol = dataframe.columns[0] |
| 112 | + times = dataframe.index |
| 113 | + entries = dataframe[symbol].values |
| 114 | + if self.has_symbol(symbol): |
| 115 | + self.purge(symbol) |
| 116 | + doc = {'symbol': symbol, 'metadata': entries[0], 'start_time': times[0]} |
| 117 | + for metadata, start_time in zip(entries[1:], times[1:]): |
| 118 | + if metadata == doc['metadata']: |
| 119 | + continue |
| 120 | + doc['end_time'] = start_time |
| 121 | + documents.append(doc) |
| 122 | + doc = {'symbol': symbol, 'metadata': metadata, 'start_time': start_time} |
| 123 | + documents.append(doc) |
| 124 | + |
| 125 | + self.insert_many(documents) |
| 126 | + |
| 127 | + def append(self, symbol, metadata, start_time=None): |
| 128 | + """ |
| 129 | + Update metadata entry for `symbol` |
| 130 | +
|
| 131 | + Parameters |
| 132 | + ---------- |
| 133 | + symbol : `str` |
| 134 | + symbol name for the item |
| 135 | + metadata : `dict` |
| 136 | + to be persisted |
| 137 | + start_time : `datetime.datetime` |
| 138 | + when metadata becomes effective |
| 139 | + Default: datetime.datetime.utcnow() |
| 140 | + """ |
| 141 | + if start_time is None: |
| 142 | + start_time = dt.utcnow() |
| 143 | + old_metadata = self.find_one({'symbol': symbol}, sort=[('start_time', pymongo.DESCENDING)]) |
| 144 | + if old_metadata is not None: |
| 145 | + if old_metadata['start_time'] >= start_time: |
| 146 | + raise ValueError('start_time={} is earlier than the last metadata @{}'.format(start_time, |
| 147 | + old_metadata['start_time'])) |
| 148 | + if old_metadata['metadata'] == metadata: |
| 149 | + logger.warning('No change to metadata') |
| 150 | + return metadata |
| 151 | + elif metadata is None: |
| 152 | + return |
| 153 | + |
| 154 | + self.find_one_and_update({'symbol': symbol}, {'$set': {'end_time': start_time}}, |
| 155 | + sort=[('start_time', pymongo.DESCENDING)]) |
| 156 | + document = {'_id': bson.ObjectId(), 'symbol': symbol, 'metadata': metadata, 'start_time': start_time} |
| 157 | + mongo_retry(self.insert_one)(document) |
| 158 | + |
| 159 | + logger.debug('Finished writing metadata for %s', symbol) |
| 160 | + return document |
| 161 | + |
| 162 | + def prepend(self, symbol, metadata, start_time=None): |
| 163 | + """ |
| 164 | + Prepend a metadata entry for `symbol` |
| 165 | +
|
| 166 | + Parameters |
| 167 | + ---------- |
| 168 | + symbol : `str` |
| 169 | + symbol name for the item |
| 170 | + metadata : `dict` |
| 171 | + to be persisted |
| 172 | + start_time : `datetime.datetime` |
| 173 | + when metadata becomes effective |
| 174 | + Default: datetime.datetime.min |
| 175 | + """ |
| 176 | + if metadata is None: |
| 177 | + return |
| 178 | + if start_time is None: |
| 179 | + start_time = dt.min |
| 180 | + old_metadata = self.find_one({'symbol': symbol}, sort=[('start_time', pymongo.ASCENDING)]) |
| 181 | + if old_metadata is not None: |
| 182 | + if old_metadata['start_time'] <= start_time: |
| 183 | + raise ValueError('start_time={} is later than the first metadata @{}'.format(start_time, |
| 184 | + old_metadata['start_time'])) |
| 185 | + if old_metadata['metadata'] == metadata: |
| 186 | + self.find_one_and_update({'symbol': symbol}, {'$set': {'start_time': start_time}}, |
| 187 | + sort=[('start_time', pymongo.ASCENDING)]) |
| 188 | + return metadata |
| 189 | + end_time = old_metadata.get('start_time') |
| 190 | + else: |
| 191 | + end_time = None |
| 192 | + |
| 193 | + document = {'_id': bson.ObjectId(), 'symbol': symbol, 'metadata': metadata, 'start_time': start_time} |
| 194 | + if end_time is not None: |
| 195 | + document['end_time'] = end_time |
| 196 | + mongo_retry(self.insert_one)(document) |
| 197 | + |
| 198 | + logger.debug('Finished writing metadata for %s', symbol) |
| 199 | + return document |
| 200 | + |
| 201 | + def pop(self, symbol): |
| 202 | + """ |
| 203 | + Delete current metadata of `symbol` |
| 204 | +
|
| 205 | + Parameters |
| 206 | + ---------- |
| 207 | + symbol : `str` |
| 208 | + symbol name to delete |
| 209 | +
|
| 210 | + Returns |
| 211 | + ------- |
| 212 | + Deleted metadata |
| 213 | + """ |
| 214 | + last_metadata = self.find_one({'symbol': symbol}, sort=[('start_time', pymongo.DESCENDING)]) |
| 215 | + if last_metadata is None: |
| 216 | + raise NoDataFoundException('No metadata found for symbol {}'.format(symbol)) |
| 217 | + |
| 218 | + self.find_one_and_delete({'symbol': symbol}, sort=[('start_time', pymongo.DESCENDING)]) |
| 219 | + mongo_retry(self.find_one_and_update)({'symbol': symbol}, {'$unset': {'end_time': ''}}, |
| 220 | + sort=[('start_time', pymongo.DESCENDING)]) |
| 221 | + |
| 222 | + return last_metadata |
| 223 | + |
| 224 | + @mongo_retry |
| 225 | + def purge(self, symbol): |
| 226 | + """ |
| 227 | + Delete all metadata of `symbol` |
| 228 | +
|
| 229 | + Parameters |
| 230 | + ---------- |
| 231 | + symbol : `str` |
| 232 | + symbol name to delete |
| 233 | + """ |
| 234 | + logger.warning("Deleting entire metadata history for %r from %r" % (symbol, self._arctic_lib.get_name())) |
| 235 | + self.delete_many({'symbol': symbol}) |
0 commit comments