Skip to content

Latest commit

 

History

History
82 lines (66 loc) · 3.63 KB

DBASE.md

File metadata and controls

82 lines (66 loc) · 3.63 KB
Raw

Persistency

Ostap.ZipShelve

Ostap offers very nice&efficient way to store the objects in persistent dbase. This persistency is build around shelve module and differs in two way

  1. the conntent of payload is compressed, using zlib module making the data base very compact
    • (optionally) the whole database can ve further gzip'ed using gzip module, if the extension .gz is provided. It makes data banse even more compact.
  2. in addition to the native dict interface from shelve, more extensiveinterface with more methods is supported.

Create database and write objects to it:

a = ...
import Ostap.ZipShelve as DBASE 
with DBASE.open ( 'my_dbase.db' ) as db : ## create DBASE 
  db.ls() 
  db['a'] =  a
  db['histo'] = ROOT.TH1D('h1','',10,0,1)

Reading objects from database

with DBASE.open ( 'my_dbase.db' , 'read') as db : ## read DBASE 
  db.ls() 
  b  = db['a'] 
  h2 = db['histo']

One can store in database all pickable objects, that means all python objects, all (serializeable) ROOT objects. All C++ objects with LCG/Reflex/Cint-dictionaries are also could be stored database. In practice, everything is storable, including complex combination of python&C++ objects, like dictionary of historgams and python classed, inherited from C++-base classes.

Plain ROOT.TFile

Ostap adds some decorations even for the plain ROOT.TFile, making its interface more pythonic:

rfile = ...
obj   = rfile['A/B/C/myobj']     ## READ  object form the file/directory
rfile['A/B/C/myobj2'] = object2  ## WRITE object to the file/directory 
obj  = rfile.A.B.C.myobj              ## another way to access to the object
obj  = rfile.get ( 'A/B/C/q' , None ) ## one more way to get object 
for obj in rfile : print obj          ## loop over all objects in file
for key,obj in rfile.iteritems() : print key, obj             ## another loop
for key,obj in rfile.iteritems( ROOT.TH1 ) : print key, obj   ## advanced loop, get only histograms 
for k in rfile.keys()     : print k   ## get all keys and loop over them 
for k in rfile.iterkeys() : print k   ## loop over all keys in the file
del rfile['A/B']                      ## delete the object from the file
rfile.rm ( 'A/B' )                    ## delete the object from the file
if 'A/MyHisto' in rfile          : print 'OK!' ## check presence of the key
if rfile.has_key ( 'A/MyHisto' ) : print 'OK!' ## check presence of the key
with ROOT.TFile('aa.root') as rfile : rfile.ls() ## context manager protocol

RootOnlyShelve

The module Ostap.RootShelve offers the thin wrapper over ROOT.TFile that implement shelve-interface. As a resutl one gets a ligth database build a top of underlying ROOT.TFile, where ROOT-objects could be stored:

from Ostap.RootShelve import RooOnlyShelf
db = RooOnlyShelf('mydb.root','c')
h1 = ...
db ['histogram'] = h1
db.ls()

RootShelve

The module Ostap.RootShelve offers also more sophisticated wrapper over ROOT.TFile that also implements shelve-interface and able to store ROOT and any other pickable objects

from Ostap.RootShelve import RooShelf
db = RooShelf('mydb.root','c')
h1 = ...
db ['histogram'] = h1
db ['histogramlist'] = h1,h2,h3
db.ls()

{% discussion "In details ..." %} For non-ROOT objects, database actually stores them as ROOT::TString objects carrying their pickle representation
with on-flight removal/substitutions of some magic symbol sequences, since ROOT::TString is not a real BLOB. {% enddiscussion %}