Cyclone-Team/gramps
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

reference_map implementation.

Browse Source 
svn: r5555
This commit is contained in:
Richard Taylor 2005-12-15 17:32:10 +00:00
parent db364e3d26
commit 6a81fa36bc
3 changed files with 209 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
gramps2/ChangeLog
View File

@ -1,3 +1,10 @@
2005-12-15 Richard Taylor <rjt-gramps@thegrindstone.me.uk>
* src/GrampsBSDDB.py: implemented get_backlink_handles that uses
the reference_map index, _update_reference_map that populates
the reference_map and supporting code.
* src/GrampsDbBase.py: added hooks to update reference_map when
primary objects are commited.
2005-12-15 Richard Taylor <rjt-gramps@thegrindstone.me.uk>
* src/GrampsDbBase.py: in preparation for new reference_map,
added a sequencial scan implementation of get_backlink_handles

171
gramps2/src/GrampsBSDDB.py
View File

@ -29,6 +29,7 @@ Provides the Berkeley DB (BSDDB) database backend for GRAMPS
# Standard python modules
#
#-------------------------------------------------------------------------
import cPickle
import os
import time
import locale
@ -62,6 +63,17 @@ def find_eventname(key,data):
def find_repository_type(key,data):
return str(data[2])
# Secondary database key lookups for reference_map table
# reference_map data values are of the form:
# ((primary_object_class_name, primary_object_handle),
# (referenced_object_class_name, referenced_object_handle))
def find_primary_handle(key,data):
return (data)[0][1]
def find_referenced_handle(key,data):
return (data)[1][1]
class GrampsBSDDBCursor(GrampsCursor):
def __init__(self,source):
@ -76,6 +88,16 @@ class GrampsBSDDBCursor(GrampsCursor):
def close(self):
self.cursor.close()
class GrampsBSDDBDupCursor(GrampsBSDDBCursor):
"""Cursor that includes handling for duplicate keys"""
def set(self,key):
return self.cursor.set(key)
def next_dup(self):
return self.cursor.next_dup()
#-------------------------------------------------------------------------
#
# GrampsBSDDB
@ -119,6 +141,23 @@ class GrampsBSDDB(GrampsDbBase):
def get_repository_cursor(self):
return GrampsBSDDBCursor(self.repository_map)
# cursors for lookups in the reference_map for back reference
# lookups. The reference_map has three indexes:
# the main index: a tuple of (primary_handle,referenced_handle)
# the primary_handle index: the primary_handle
# the referenced_handle index: the referenced_handle
# the main index is unique, the others allow duplicate entries.
def get_reference_map_cursor(self):
return GrampsBSDDBCursor(self.reference_map)
def get_reference_map_primary_cursor(self):
return GrampsBSDDBDupCursor(self.reference_map_primary_map)
def get_reference_map_referenced_cursor(self):
return GrampsBSDDBDupCursor(self.reference_map_referenced_map)
def version_supported(self):
return (self.metadata.get('version',0) <= _DBVERSION and
self.metadata.get('version',0) >= _MINVERSION)
@ -158,6 +197,10 @@ class GrampsBSDDB(GrampsDbBase):
self.person_map = self.dbopen(name, "person")
self.repository_map = self.dbopen(name, "repository")
# index tables used just for speeding up searches
self.reference_map = self.dbopen(name, "reference_map")
if not self.readonly:
self.undodb = db.DB()
self.undodb.open(self.undolog, db.DB_RECNO, db.DB_CREATE)
@ -249,6 +292,20 @@ class GrampsBSDDB(GrampsDbBase):
self.rid_trans.open(self.save_name, "ridtrans",
db.DB_HASH, flags=openflags)
self.reference_map_primary_map = db.DB(self.env)
self.reference_map_primary_map.set_flags(db.DB_DUP)
self.reference_map_primary_map.open(self.save_name,
"reference_map_primary_map",
db.DB_BTREE, flags=openflags)
self.reference_map_referenced_map = db.DB(self.env)
self.reference_map_referenced_map.set_flags(db.DB_DUP)
self.reference_map_referenced_map.open(self.save_name,
"reference_map_referenced_map",
db.DB_BTREE, flags=openflags)
if not self.readonly:
self.person_map.associate(self.surnames, find_surname, openflags)
self.person_map.associate(self.id_trans, find_idmap, openflags)
@ -261,6 +318,12 @@ class GrampsBSDDB(GrampsDbBase):
self.place_map.associate(self.pid_trans, find_idmap, openflags)
self.media_map.associate(self.oid_trans, find_idmap, openflags)
self.source_map.associate(self.sid_trans, find_idmap, openflags)
self.reference_map.associate(self.reference_map_primary_map,
find_primary_handle,
openflags)
self.reference_map.associate(self.reference_map_referenced_map,
find_referenced_handle,
openflags)
def rebuild_secondary(self,callback=None):
@ -370,6 +433,114 @@ class GrampsBSDDB(GrampsDbBase):
self.repository_map[key] = self.repository_map[key]
self.repository_map.sync()
def find_backlink_handles(self, handle, include_classes=None):
"""
Find all objects that hold a reference to the object handle.
Returns an interator over alist of (class_name,handle) tuples.
@param handle: handle of the object to search for.
@type handle: database handle
@param include_classes: list of class names to include in the results.
Default: None means include all classes.
@type include_classes: list of class names
This default implementation does a sequencial scan through all
the primary object databases and is very slow. Backends can
override this method to provide much faster implementations that
make use of additional capabilities of the backend.
Note that this is a generator function, it returns a iterator for
use in loops. If you want a list of the results use:
result_list = [i for i in find_backlink_handles(handle)]
"""
# Use the secondary index to locate all the reference_map entries
# that include a reference to the object we are looking for.
referenced_cur = self.get_reference_map_referenced_cursor()
ret = referenced_cur.set(handle)
while (ret is not None):
(key,data) = ret
# data values are of the form:
# ((primary_object_class_name, primary_object_handle),
# (referenced_object_class_name, referenced_object_handle))
# so we need the first tuple to give us the type to compare
data = cPickle.loads(data)
if include_classes == None or data[0][0] in include_classes:
yield data[0]
ret = referenced_cur.next_dup()
referenced_cur.close()
return
def _update_reference_map(self, obj, class_name):
# Add references to the reference_map for all primary object referenced
# from the primary object 'obj' or any of its secondary objects.
# FIXME: this needs to be properly integrated into the transaction
# framework so that the reference_map changes are part of the
# transaction
handle = obj.get_handle()
# First thing to do is get hold of all rows in the reference_map
# table that hold a reference from this primary obj. This means finding
# all the rows that have this handle somewhere in the list of (class_name,handle)
# pairs.
# The primary_map secondary index allows us to look this up quickly.
existing_references = set()
primary_cur = self.get_reference_map_primary_cursor()
ret = primary_cur.set(handle)
while (ret is not None):
(key,data) = ret
# data values are of the form:
# ((primary_object_class_name, primary_object_handle),
# (referenced_object_class_name, referenced_object_handle))
# so we need the second tuple give us a reference that we can
# compare with what is returned from get_referenced_handles_recursively
# Looks like there is a bug in the set() and next_dup() methods
# because they do not run the data through cPickle.loads before
# returning it, so we have to here.
existing_references.add(cPickle.loads(data)[1])
ret = primary_cur.next_dup()
primary_cur.close()
# Once we have the list of rows that already have a reference we need to compare
# it with the list of objects that are still references from the primary object.
current_references = set(obj.get_referenced_handles_recursively())
no_longer_required_references = existing_references.difference(current_references)
new_references = current_references.difference(existing_references)
# handle addition of new references
if len(new_references) > 0:
for (ref_class_name,ref_handle) in new_references:
self.reference_map[str((handle,ref_handle),)] = ((class_name,handle),
(ref_class_name,ref_handle),)
# handle deletion of old references
if len(no_longer_required_references) > 0:
for (ref_class_name,ref_handle) in no_longer_required_references:
self.reference_map.delete(str((handle,ref_handle),))
def abort_changes(self):
while self.undo():
pass

32
gramps2/src/GrampsDbBase.py
View File

@ -307,6 +307,8 @@ class GrampsDbBase(GrampsDBCallback.GrampsDBCallback):
self.emit('event-rebuild')
self.emit('repository-rebuild')
def _commit_base(self, obj, data_map, key, update_list, add_list,
transaction, change_time):
"""
@ -339,6 +341,9 @@ class GrampsDbBase(GrampsDBCallback.GrampsDBCallback):
Commits the specified Person to the database, storing the changes
as part of the transaction.
"""
self._update_reference_map(person,'Person')
old_data = self._commit_base(
person, self.person_map, PERSON_KEY, transaction.person_update,
transaction.person_add, transaction, change_time)
@ -361,6 +366,9 @@ class GrampsDbBase(GrampsDBCallback.GrampsDBCallback):
Commits the specified MediaObject to the database, storing the changes
as part of the transaction.
"""
self._update_reference_map(obj,'MediaObject')
self._commit_base(obj, self.media_map, MEDIA_KEY,
transaction.media_update, transaction.media_add,
transaction, change_time)
@ -370,6 +378,9 @@ class GrampsDbBase(GrampsDBCallback.GrampsDBCallback):
Commits the specified Source to the database, storing the changes
as part of the transaction.
"""
self._update_reference_map(source,'Source')
self._commit_base(source, self.source_map, SOURCE_KEY,
transaction.source_update, transaction.source_add,
transaction, change_time)
@ -379,6 +390,9 @@ class GrampsDbBase(GrampsDBCallback.GrampsDBCallback):
Commits the specified Place to the database, storing the changes
as part of the transaction.
"""
self._update_reference_map(place,'Place')
self._commit_base(place, self.place_map, PLACE_KEY,
transaction.place_update, transaction.place_add,
transaction, change_time)
@ -398,6 +412,9 @@ class GrampsDbBase(GrampsDBCallback.GrampsDBCallback):
Commits the specified Event to the database, storing the changes
as part of the transaction.
"""
self._update_reference_map(event,'Event')
self._commit_base(event, self.event_map, EVENT_KEY,
transaction.event_update, transaction.event_add,
transaction, change_time)
@ -407,6 +424,9 @@ class GrampsDbBase(GrampsDBCallback.GrampsDBCallback):
Commits the specified Family to the database, storing the changes
as part of the transaction.
"""
self._update_reference_map(family,'Family')
self._commit_base(family, self.family_map, FAMILY_KEY,
transaction.family_update, transaction.family_add,
transaction, change_time)
@ -419,6 +439,9 @@ class GrampsDbBase(GrampsDBCallback.GrampsDBCallback):
Commits the specified Repository to the database, storing the changes
as part of the transaction.
"""
self._update_reference_map(repository,'Repository')
self._commit_base(repository, self.repository_map, REPOSITORY_KEY,
transaction.repository_update, transaction.repository_add,
transaction, change_time)
@ -1446,11 +1469,16 @@ class GrampsDbBase(GrampsDBCallback.GrampsDBCallback):
default = [(1,1),(0,5),(0,6),(1,2),(1,3),(0,4),(0,7),(0,8),(0,9),(0,10)]
return self._get_column_order(REPOSITORY_COL_KEY,default)
def _update_reference_map(self,obj, class_name):
"""Called each time an object is writen to the database. This can
be used by subclasses to update any additional index tables that might
need to be changed."""
pass
def find_backlink_handles(self, handle, include_classes=None):
"""
Find all objects that hold a reference to the object handle.
Returns an interator over alist of (handle,class_name) tuples.
Returns an interator over alist of (class_name,handle) tuples.
@param handle: handle of the object to search for.
@type handle: database handle
@ -1507,7 +1535,7 @@ class GrampsDbBase(GrampsDBCallback.GrampsDBCallback):
obj.unserialize(val)
if obj.has_source_reference(handle):
yield (found_handle,primary_table_name)
yield (primary_table_name,found_handle)
data = cursor.next()