I'm using merge to push objects into a db. (Long story short: I have
a set of objects that have to behave identically if they are
persisted or not, and may be retrieved or sent to a set of databases.
Hence the use of merge.) However, merged objects appear object
joined by a relation and backrefs twice. Is this a bug or an I making
some subtle mistake?
A condensed demonstration - the names of the mapped objects and
tables are unimportant, but a Biosequence has a one-to-many with a
set of Annotations, established by a backref::
from sqlalchemy import *
from sqlalchemy.orm import *
metadata = MetaData()
class Biosequence (object):
pass
class Annotation (object):
pass
table_biosequences = Table ('biosequences', metadata,
Column ('identifier', Integer(), primary_key=True),
Column ('dummy', String(16)),
)
table_seqannotations = Table ('seqannotations', metadata,
Column ('identifier', Integer(), primary_key=True),
Column ('biosequence_id', Integer(), ForeignKey
('biosequences.identifier')),
)
mapper (Annotation, table_seqannotations)
mapper (Biosequence, table_biosequences,
properties={
'annotations':relation (
Annotation,
backref='biosequence_ref',
lazy=False,
cascade="all, delete, delete-orphan",
),
},
)
engine = create_engine ('sqlite:///', convert_unicode=True,
# echo=True,
)
sessionfactory = sessionmaker (bind=engine, autoflush=False,
transactional=False)
session = sessionfactory()
metadata.create_all (bind=engine, checkfirst=True)
# make an object with 3 'children'
bseq = Biosequence()
anns = [Annotation() for x in range (3)]
bseq.annotations = anns
print "Original num of anns:", len (bseq.annotations)
# the merged copy has 6 children
merge_bseq = session.merge (bseq)
session.flush()
print "Now num of anns still:", len (bseq.annotations)
print "Ids:", [x.identifier for x in bseq.annotations]
print "Merged copy num of anns:", len (merge_bseq.annotations)
print "Ids:", [x.identifier for x in merge_bseq.annotations]
# as does the return!
results = session.query (Biosequence)
print "Number of results", results.count()
returned_bseq = results.one()
print "Returned num of anns:", len (returned_bseq.annotations)
print "Ids:", [x.identifier for x in returned_bseq.annotations]
# make an new object with 3 children
bseq2 = Biosequence()
bseq2.annotations = [Annotation() for x in range (3)]
print "New obj num of anns:", len (bseq.annotations)
session.save_or_update (bseq2)
session.flush()
# and it works as expected
results = session.query (Biosequence)
print "Number of retreived objects", results.count()
returned_bseqs = results.all()
for item in returned_bseqs:
print "Returned num of anns:", len (item.annotations)
print "Ids:", [x.identifier for x in item.annotations]
Which returns::
Original num of anns: 3
Now num of anns still: 3
Ids: [None, None, None]
Merged copy num of anns: 6
Ids: [1, 1, 2, 2, 3, 3]
Number of results 1
Returned num of anns: 6
Ids: [1, 1, 2, 2, 3, 3]
New obj num of anns: 3
Number of retreived objects 2
Returned num of anns: 6
Ids: [1, 1, 2, 2, 3, 3]
Returned num of anns: 3
Ids: [4, 5, 6]
In summary: if an object is stored using 'save_or_update', when later
retrieved, it has the correct number of related children. If merged,
the returned merge copy has each child twice. If this object is
retrieved later, it still has every child twice. Looking at the db,
the correct number of child rows is being stored and the relations
are correct. It therefore looks like there's a problem in the session
for merged objects.
Any pointers?
--
Dr Paul-Michael Agapow: VieDigitale / Inst. for Animal Health
[EMAIL PROTECTED] / [EMAIL PROTECTED]
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"sqlalchemy" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/sqlalchemy?hl=en
-~----------~----~----~----~------~----~------~--~---
