Merge pull request #34 from fact-project/ontime_from_db

Ontime from db

fact/factdb/__init__.py

@@ -1,3 +1,3 @@
-from .database import factdata_db
-from .utils import get_ontime_by_source, get_ontime_by_source_and_runtype
+from .database import factdata_db, connect_database
+from .utils import get_ontime_by_source, get_ontime_by_source_and_runtype, get_correct_ontime
 from .models import *

fact/factdb/utils.py

@@ -2,10 +2,67 @@
 import peewee
 from .models import RunInfo, Source, RunType
 from ..credentials import create_factdb_engine
+from ..time import night_integer
+
+
+SECOND = peewee.SQL('SECOND')
+
+
+def read_into_dataframe(query, engine=None):
+    ''' read the result of a peewee query object into a pandas DataFrame '''
+
+    sql, params = query.sql()
+    df = pd.read_sql_query(sql, engine or create_factdb_engine(), params=params)
+
+    return df
+
+
+def get_correct_ontime(start=None, end=None, engine=None):
+    '''
+    The database field fOnTime underestimates the real ontime by about 5 seconds
+    because of how the number is calculated from the FTM auxfiles.
+    A better estimate can be obtained by taking (fRunStop - fRunStart) * fEffectiveOn.
+
+    Parameters
+    ----------
+    start : int or datetime.date
+        First night to select, either in fact int format or as date
+    end : int or datetime.date
+        Last night to select, either in fact int format or as date
+    engine: sqlalchemy.Engine
+        The engine connected to the database.
+        If None, fact.credentials.create_factdb_engine will be used to create one.
+
+    Source: D. Neise, A. Biland. Also see github.com/dneise/about_fact_ontime
+    '''
+
+    duration = (peewee.fn.TIMESTAMPDIFF(SECOND, RunInfo.frunstart, RunInfo.frunstop))
+
+    query = RunInfo.select(
+        RunInfo.fnight.alias('night'),
+        RunInfo.frunid.alias('run_id'),
+        RunInfo.frunstart.alias('start'),
+        RunInfo.frunstart.alias('stop'),
+        duration.alias('duration'),
+        RunInfo.feffectiveon.alias('relative_ontime'),
+    )
+
+    if start is not None:
+        start = night_integer(start) if not isinstance(start, int) else start
+        query = query.where(RunInfo.fnight >= start)
+
+    if end is not None:
+        end = night_integer(end) if not isinstance(end, int) else end
+        query = query.where(RunInfo.fnight <= end)
+
+    df = read_into_dataframe(query, engine=engine)
+    df['ontime'] = df['duration'] * df['relative_ontime']
+
+    return df
 
 
 def get_ontime_by_source_and_runtype(engine=None):
-    query, params = (
+    query = (
         RunInfo
         .select(
             peewee.fn.SUM(RunInfo.fontime).alias('ontime'),
@@ -16,9 +73,8 @@ def get_ontime_by_source_and_runtype(engine=None):
         .switch(RunInfo)
         .join(RunType, on=RunType.fruntypekey == RunInfo.fruntypekey)
         .group_by(Source.fsourcename, RunType.fruntypename)
-        .sql()
     )
-    df = pd.read_sql(query, engine or create_factdb_engine(), params=params)
+    df = read_into_dataframe(query, engine or create_factdb_engine())
     df.set_index(['source', 'runtype'], inplace=True)
 
     return df
@@ -38,8 +94,9 @@ def get_ontime_by_source(runtype=None, engine=None):
     if runtype is not None:
         query = query.where(RunType.fruntypename == runtype)
 
-    query, params = query.group_by(Source.fsourcename).sql()
-    df = pd.read_sql(query, engine or create_factdb_engine(), params=params)
+    query = query.group_by(Source.fsourcename)
+
+    df = read_into_dataframe(query, engine or create_factdb_engine())
     df.set_index('source', inplace=True)
 
     return df

setup.py

@@ -3,7 +3,7 @@
 
 setup(
     name='pyfact',
-    version='0.8.5',
+    version='0.8.6',
     description='A module containing useful methods for working with fact',
     url='http://github.com/fact-project/pyfact',
     author='Maximilian Noethe, Dominik Neise',