wassname/catalyst
1
0
Fork 0
mirror of https://github.com/wassname/catalyst.git synced 2026-07-10 19:26:14 +08:00
Code Issues Packages Projects Releases Wiki Activity

DOC: add whatsnew entry for AssetFinderCachedEquities.

Browse Source
This commit is contained in:
Maya Tydykov
2015-11-11 17:09:33 -05:00
parent d0cb5bd513
commit df492ecf7f
3 changed files with 79 additions and 49 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/source/whatsnew/0.8.4.txt
+4 -1
View File
@@ -33,7 +33,10 @@ Bug Fixes
Performance
~~~~~~~~~~~
None
* Speeds up `AssetFinder.lookup_symbol` by adding an extension,
`AssetFinderCachedEquities`, that loads equities into dictionaries and
then directs `lookup_symbol` to these dictionaries to find matching equities
(:issue:`830`).
Maintenance and Refactorings
~~~~~~~~~~~~~~~~~~~~~~~~~~~~
tests/test_assets.py
+25
View File
@@ -457,6 +457,31 @@ class AssetFinderTestCase(TestCase):
self.assertEqual(result.symbol, 'EXISTING')
self.assertEqual(result.sid, i)
def test_lookup_symbol_from_multiple_valid(self):
df = pd.DataFrame.from_records(
[
{
'sid': 1,
'symbol': 'multiple',
'start_date': pd.Timestamp('2010-01-01'),
'end_date': pd.Timestamp('2013-01-01'),
'exchange': 'NYSE'
},
{
'sid': 2,
'symbol': 'multiple',
'start_date': pd.Timestamp('2012-01-01'),
'end_date': pd.Timestamp('2014-01-01'),
'exchange': 'NYSE'
}
]
)
self.env.write_data(equities_df=df)
finder = self.asset_finder_type(self.env.engine)
result = finder.lookup_symbol('MULTIPLE', pd.Timestamp('2012-05-05'))
self.assertEqual(result.symbol, 'MULTIPLE')
self.assertEqual(result.sid, 2)
def test_lookup_generic(self):
"""
Ensure that lookup_generic works with various permutations of inputs.
zipline/assets/assets.py
+50 -48
View File
@@ -196,7 +196,7 @@ class AssetFinder(object):
cache[sid] = asset
return asset
def get_fuzzy_candidates(self, fuzzy_symbol):
def _get_fuzzy_candidates(self, fuzzy_symbol):
candidates = sa.select(
(self.equities.c.sid,)
).where(self.equities.c.fuzzy_symbol == fuzzy_symbol).order_by(
@@ -205,7 +205,7 @@ class AssetFinder(object):
).execute().fetchall()
return candidates
def get_fuzzy_candidates_in_range(self, fuzzy_symbol, ad_value):
def _get_fuzzy_candidates_in_range(self, fuzzy_symbol, ad_value):
candidates = sa.select(
(self.equities.c.sid,)
).where(
@@ -220,10 +220,10 @@ class AssetFinder(object):
).execute().fetchall()
return candidates
def get_split_candidates_in_range(self,
company_symbol,
share_class_symbol,
ad_value):
def _get_split_candidates_in_range(self,
company_symbol,
share_class_symbol,
ad_value):
candidates = sa.select(
(self.equities.c.sid,)
).where(
@@ -239,7 +239,7 @@ class AssetFinder(object):
).execute().fetchall()
return candidates
def get_split_candidates(self, company_symbol, share_class_symbol):
def _get_split_candidates(self, company_symbol, share_class_symbol):
candidates = sa.select(
(self.equities.c.sid,)
).where(
@@ -253,10 +253,10 @@ class AssetFinder(object):
).execute().fetchall()
return candidates
def resolve_no_matching_candidates(self,
company_symbol,
share_class_symbol,
ad_value):
def _resolve_no_matching_candidates(self,
company_symbol,
share_class_symbol,
ad_value):
candidates = sa.select((self.equities.c.sid,)).where(
sa.and_(
self.equities.c.company_symbol == company_symbol,
@@ -268,10 +268,10 @@ class AssetFinder(object):
).execute().fetchall()
return candidates
def get_best_candidate(self, candidates):
def _get_best_candidate(self, candidates):
return self._retrieve_equity(candidates[0]['sid'])
def get_equities_from_candidates(self, candidates):
def _get_equities_from_candidates(self, candidates):
return list(map(
compose(self._retrieve_equity, itemgetter('sid')),
candidates,
@@ -296,35 +296,37 @@ class AssetFinder(object):
if fuzzy:
# Search for a single exact match on the fuzzy column
candidates = self.get_fuzzy_candidates_in_range(fuzzy_symbol,
ad_value)
candidates = self._get_fuzzy_candidates_in_range(fuzzy_symbol,
ad_value)
# If exactly one SID exists for fuzzy_symbol, return that sid
if len(candidates) == 1:
return self.get_best_candidate(candidates)
return self._get_best_candidate(candidates)
# Search for exact matches of the split-up company_symbol and
# share_class_symbol
candidates = self.get_split_candidates_in_range(company_symbol,
share_class_symbol,
ad_value)
candidates = self._get_split_candidates_in_range(
company_symbol,
share_class_symbol,
ad_value
)
# If exactly one SID exists for symbol, return that symbol
# If multiple SIDs exist for symbol, return latest start_date with
# end_date as a tie-breaker
if candidates:
return self.get_best_candidate(candidates)
return self._get_best_candidate(candidates)
# If no SID exists for symbol, return SID with the
# highest-but-not-over end_date
elif not candidates:
candidates = self.resolve_no_matching_candidates(
candidates = self._resolve_no_matching_candidates(
company_symbol,
share_class_symbol,
ad_value
)
if candidates:
return self.get_best_candidate(candidates)
return self._get_best_candidate(candidates)
raise SymbolNotFound(symbol=symbol)
@@ -332,20 +334,20 @@ class AssetFinder(object):
# If this is a fuzzy look-up, check if there is exactly one match
# for the fuzzy symbol
if fuzzy:
candidates = self.get_fuzzy_candidates(fuzzy_symbol)
candidates = self._get_fuzzy_candidates(fuzzy_symbol)
if len(candidates) == 1:
return self.get_best_candidate(candidates)
return self._get_best_candidate(candidates)
candidates = self.get_split_candidates(company_symbol,
share_class_symbol)
candidates = self._get_split_candidates(company_symbol,
share_class_symbol)
if len(candidates) == 1:
return self.get_best_candidate(candidates)
return self._get_best_candidate(candidates)
elif not candidates:
raise SymbolNotFound(symbol=symbol)
else:
raise MultipleSymbolsFound(
symbol=symbol,
options=self.get_equities_from_candidates(candidates)
options=self._get_equities_from_candidates(candidates)
)
def lookup_future_symbol(self, symbol):
@@ -741,9 +743,9 @@ class AssetFinderCachedEquities(AssetFinder):
self.fuzzy_symbol_hashed_equities = {}
self.company_share_class_hashed_equities = {}
self.hashed_equities = sa.select(self.equities.c).execute().fetchall()
self.load_hashed_equities()
self._load_hashed_equities()
def load_hashed_equities(self):
def _load_hashed_equities(self):
"""
Populates two maps - fuzzy symbol to list of equities having that
fuzzy symbol and company symbol/share class symbol to list of
@@ -753,7 +755,7 @@ class AssetFinderCachedEquities(AssetFinder):
company_symbol = equity['company_symbol']
share_class_symbol = equity['share_class_symbol']
fuzzy_symbol = equity['fuzzy_symbol']
asset = self.convert_row_to_equity(equity)
asset = self._convert_row_to_equity(equity)
self.company_share_class_hashed_equities.setdefault(
(company_symbol, share_class_symbol),
[]
@@ -762,7 +764,7 @@ class AssetFinderCachedEquities(AssetFinder):
fuzzy_symbol, []
).append(asset)
def convert_row_to_equity(self, equity):
def _convert_row_to_equity(self, equity):
"""
Converts a SQLAlchemy equity row to an Equity object.
"""
@@ -771,13 +773,13 @@ class AssetFinderCachedEquities(AssetFinder):
asset = Equity(**data)
return asset
def get_fuzzy_candidates(self, fuzzy_symbol):
def _get_fuzzy_candidates(self, fuzzy_symbol):
if fuzzy_symbol in self.fuzzy_symbol_hashed_equities:
return self.fuzzy_symbol_hashed_equities[fuzzy_symbol]
return []
def get_fuzzy_candidates_in_range(self, fuzzy_symbol, ad_value):
equities = self.get_fuzzy_candidates(fuzzy_symbol)
def _get_fuzzy_candidates_in_range(self, fuzzy_symbol, ad_value):
equities = self._get_fuzzy_candidates(fuzzy_symbol)
fuzzy_candidates = []
for equity in equities:
if (equity.start_date.value <=
@@ -786,18 +788,18 @@ class AssetFinderCachedEquities(AssetFinder):
fuzzy_candidates.append(equity)
return fuzzy_candidates
def get_split_candidates(self, company_symbol, share_class_symbol):
def _get_split_candidates(self, company_symbol, share_class_symbol):
if (company_symbol, share_class_symbol) in \
self.company_share_class_hashed_equities:
return self.company_share_class_hashed_equities[(
company_symbol, share_class_symbol)]
return []
def get_split_candidates_in_range(self,
company_symbol,
share_class_symbol,
ad_value):
equities = self.get_split_candidates(
def _get_split_candidates_in_range(self,
company_symbol,
share_class_symbol,
ad_value):
equities = self._get_split_candidates(
company_symbol, share_class_symbol
)
best_candidates = []
@@ -814,11 +816,11 @@ class AssetFinderCachedEquities(AssetFinder):
)
return best_candidates
def resolve_no_matching_candidates(self,
company_symbol,
share_class_symbol,
ad_value):
equities = self.get_split_candidates(
def _resolve_no_matching_candidates(self,
company_symbol,
share_class_symbol,
ad_value):
equities = self._get_split_candidates(
company_symbol, share_class_symbol
)
partial_candidates = []
@@ -833,8 +835,8 @@ class AssetFinderCachedEquities(AssetFinder):
)
return partial_candidates
def get_best_candidate(self, candidates):
def _get_best_candidate(self, candidates):
return candidates[0]
def get_equities_from_candidates(self, candidates):
def _get_equities_from_candidates(self, candidates):
return candidates