![[tokkee]](http://tokkee.org/images/avatar.png){kind=avatar}
diff --git a/roundup/hyperdb.py b/roundup/hyperdb.py
index 26d324aa4f8e2dd722f244f979bfa59b35c0fe3c..c8ca0a57ff593596bf02f5b1e125f3e93e8ab15f 100644 (file)
--- a/roundup/hyperdb.py
+++ b/roundup/hyperdb.py
# BASIS, AND THERE IS NO OBLIGATION WHATSOEVER TO PROVIDE MAINTENANCE,
# SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
#
-# $Id: hyperdb.py,v 1.43 2001-12-20 06:13:24 rochecompaan Exp $
+# $Id: hyperdb.py,v 1.95 2003-11-16 22:56:46 jlgijsbers Exp $
-__doc__ = """
+"""
Hyperdatabase implementation, especially field types.
"""
# standard python modules
-import cPickle, re, string, weakref
+import sys, os, time, re
# roundup modules
import date, password
+# configure up the DEBUG and TRACE captures
+class Sink:
+ def write(self, content):
+ pass
+DEBUG = os.environ.get('HYPERDBDEBUG', '')
+if DEBUG and __debug__:
+ if DEBUG == 'stdout':
+ DEBUG = sys.stdout
+ else:
+ DEBUG = open(DEBUG, 'a')
+else:
+ DEBUG = Sink()
+TRACE = os.environ.get('HYPERDBTRACE', '')
+if TRACE and __debug__:
+ if TRACE == 'stdout':
+ TRACE = sys.stdout
+ else:
+ TRACE = open(TRACE, 'w')
+else:
+ TRACE = Sink()
+def traceMark():
+ print >>TRACE, '**MARK', time.ctime()
+del Sink
#
# Types
#
class String:
"""An object designating a String property."""
+ def __init__(self, indexme='no'):
+ self.indexme = indexme == 'yes'
def __repr__(self):
+ ' more useful for dumps '
return '<%s>'%self.__class__
class Password:
"""An object designating a Password property."""
def __repr__(self):
+ ' more useful for dumps '
return '<%s>'%self.__class__
class Date:
"""An object designating a Date property."""
def __repr__(self):
+ ' more useful for dumps '
return '<%s>'%self.__class__
class Interval:
"""An object designating an Interval property."""
def __repr__(self):
+ ' more useful for dumps '
return '<%s>'%self.__class__
class Link:
"""An object designating a Link property that links to a
node in a specified class."""
- def __init__(self, classname):
+ def __init__(self, classname, do_journal='yes'):
+ ''' Default is to not journal link and unlink events
+ '''
self.classname = classname
+ self.do_journal = do_journal == 'yes'
def __repr__(self):
+ ' more useful for dumps '
return '<%s to "%s">'%(self.__class__, self.classname)
class Multilink:
"""An object designating a Multilink property that links
to nodes in a specified class.
+
+ "classname" indicates the class to link to
+
+ "do_journal" indicates whether the linked-to nodes should have
+ 'link' and 'unlink' events placed in their journal
"""
- def __init__(self, classname):
+ def __init__(self, classname, do_journal='yes'):
+ ''' Default is to not journal link and unlink events
+ '''
self.classname = classname
+ self.do_journal = do_journal == 'yes'
def __repr__(self):
+ ' more useful for dumps '
return '<%s to "%s">'%(self.__class__, self.classname)
-class DatabaseError(ValueError):
+class Boolean:
+ """An object designating a boolean property"""
+ def __repr__(self):
+ 'more useful for dumps'
+ return '<%s>' % self.__class__
+
+class Number:
+ """An object designating a numeric property"""
+ def __repr__(self):
+ 'more useful for dumps'
+ return '<%s>' % self.__class__
+#
+# Support for splitting designators
+#
+class DesignatorError(ValueError):
pass
-
+def splitDesignator(designator, dre=re.compile(r'([^\d]+)(\d+)')):
+ ''' Take a foo123 and return ('foo', 123)
+ '''
+ m = dre.match(designator)
+ if m is None:
+ raise DesignatorError, '"%s" not a node designator'%designator
+ return m.group(1), m.group(2)
#
# the base Database class
#
+class DatabaseError(ValueError):
+ '''Error to be raised when there is some problem in the database code
+ '''
+ pass
class Database:
+ '''A database for storing records containing flexible data types.
+
+This class defines a hyperdatabase storage layer, which the Classes use to
+store their data.
+
+
+Transactions
+------------
+The Database should support transactions through the commit() and
+rollback() methods. All other Database methods should be transaction-aware,
+using data from the current transaction before looking up the database.
+
+An implementation must provide an override for the get() method so that the
+in-database value is returned in preference to the in-transaction value.
+This is necessary to determine if any values have changed during a
+transaction.
+
+
+Implementation
+--------------
+
+All methods except __repr__ must be implemented by a concrete backend Database.
+
+'''
+
# flag to set on retired entries
RETIRED_FLAG = '__hyperdb_retired'
+ def __init__(self, config, journaltag=None):
+ """Open a hyperdatabase given a specifier to some storage.
+
+ The 'storagelocator' is obtained from config.DATABASE.
+ The meaning of 'storagelocator' depends on the particular
+ implementation of the hyperdatabase. It could be a file name,
+ a directory path, a socket descriptor for a connection to a
+ database over the network, etc.
+
+ The 'journaltag' is a token that will be attached to the journal
+ entries for any edits done on the database. If 'journaltag' is
+ None, the database is opened in read-only mode: the Class.create(),
+ Class.set(), and Class.retire() methods are disabled.
+ """
+ raise NotImplementedError
+
+ def post_init(self):
+ """Called once the schema initialisation has finished.
+ If 'refresh' is true, we want to rebuild the backend
+ structures.
+ """
+ raise NotImplementedError
+
+ def refresh_database(self):
+ """Called to indicate that the backend should rebuild all tables
+ and structures. Not called in normal usage."""
+ raise NotImplementedError
+
+ def __getattr__(self, classname):
+ """A convenient way of calling self.getclass(classname)."""
+ raise NotImplementedError
+
+ def addclass(self, cl):
+ '''Add a Class to the hyperdatabase.
+ '''
+ raise NotImplementedError
+
+ def getclasses(self):
+ """Return a list of the names of all existing classes."""
+ raise NotImplementedError
+
+ def getclass(self, classname):
+ """Get the Class object representing a particular class.
+
+ If 'classname' is not a valid class name, a KeyError is raised.
+ """
+ raise NotImplementedError
+
+ def clear(self):
+ '''Delete all database contents.
+ '''
+ raise NotImplementedError
+
+ def getclassdb(self, classname, mode='r'):
+ '''Obtain a connection to the class db that will be used for
+ multiple actions.
+ '''
+ raise NotImplementedError
+
+ def addnode(self, classname, nodeid, node):
+ """Add the specified node to its class's db.
+ """
+ raise NotImplementedError
+
+ def serialise(self, classname, node):
+ '''Copy the node contents, converting non-marshallable data into
+ marshallable data.
+ '''
+ return node
+
+ def setnode(self, classname, nodeid, node):
+ '''Change the specified node.
+ '''
+ raise NotImplementedError
+
+ def unserialise(self, classname, node):
+ '''Decode the marshalled node data
+ '''
+ return node
+
+ def getnode(self, classname, nodeid, db=None, cache=1):
+ '''Get a node from the database.
+
+ 'cache' exists for backwards compatibility, and is not used.
+ '''
+ raise NotImplementedError
+
+ def hasnode(self, classname, nodeid, db=None):
+ '''Determine if the database has a given node.
+ '''
+ raise NotImplementedError
+
+ def countnodes(self, classname, db=None):
+ '''Count the number of nodes that exist for a particular Class.
+ '''
+ raise NotImplementedError
+
+ def storefile(self, classname, nodeid, property, content):
+ '''Store the content of the file in the database.
+
+ The property may be None, in which case the filename does not
+ indicate which property is being saved.
+ '''
+ raise NotImplementedError
+
+ def getfile(self, classname, nodeid, property):
+ '''Store the content of the file in the database.
+ '''
+ raise NotImplementedError
+
+ def addjournal(self, classname, nodeid, action, params):
+ ''' Journal the Action
+ 'action' may be:
+
+ 'create' or 'set' -- 'params' is a dictionary of property values
+ 'link' or 'unlink' -- 'params' is (classname, nodeid, propname)
+ 'retire' -- 'params' is None
+ '''
+ raise NotImplementedError
+
+ def getjournal(self, classname, nodeid):
+ ''' get the journal for id
+ '''
+ raise NotImplementedError
+
+ def pack(self, pack_before):
+ ''' pack the database
+ '''
+ raise NotImplementedError
+
+ def commit(self):
+ ''' Commit the current transactions.
+
+ Save all data changed since the database was opened or since the
+ last commit() or rollback().
+ '''
+ raise NotImplementedError
+
+ def rollback(self):
+ ''' Reverse all actions from the current transaction.
+
+ Undo all the changes made since the database was opened or the last
+ commit() or rollback() was performed.
+ '''
+ raise NotImplementedError
-_marker = []
#
# The base Class class
#
class Class:
- """The handle to a particular class of nodes in a hyperdatabase."""
+ """ The handle to a particular class of nodes in a hyperdatabase.
+
+ All methods except __repr__ and getnode must be implemented by a
+ concrete backend Class.
+ """
def __init__(self, db, classname, **properties):
"""Create a new class with a given name and property specification.
or a ValueError is raised. The keyword arguments in 'properties'
must map names to property objects, or a TypeError is raised.
"""
- self.classname = classname
- self.properties = properties
- self.db = weakref.proxy(db) # use a weak ref to avoid circularity
- self.key = ''
-
- # do the db-related init stuff
- db.addclass(self)
+ raise NotImplementedError
def __repr__(self):
- return '<hypderdb.Class "%s">'%self.classname
+ '''Slightly more useful representation
+ '''
+ return '<hyperdb.Class "%s">'%self.classname
# Editing nodes:
If an id in a link or multilink property does not refer to a valid
node, an IndexError is raised.
"""
- if propvalues.has_key('id'):
- raise KeyError, '"id" is reserved'
+ raise NotImplementedError
- if self.db.journaltag is None:
- raise DatabaseError, 'Database open read-only'
-
- # new node's id
- newid = str(self.count() + 1)
-
- # validate propvalues
- num_re = re.compile('^\d+$')
- for key, value in propvalues.items():
- if key == self.key:
- try:
- self.lookup(value)
- except KeyError:
- pass
- else:
- raise ValueError, 'node with key "%s" exists'%value
-
- # try to handle this property
- try:
- prop = self.properties[key]
- except KeyError:
- raise KeyError, '"%s" has no property "%s"'%(self.classname,
- key)
-
- if isinstance(prop, Link):
- if type(value) != type(''):
- raise ValueError, 'link value must be String'
- link_class = self.properties[key].classname
- # if it isn't a number, it's a key
- if not num_re.match(value):
- try:
- value = self.db.classes[link_class].lookup(value)
- except (TypeError, KeyError):
- raise IndexError, 'new property "%s": %s not a %s'%(
- key, value, link_class)
- elif not self.db.hasnode(link_class, value):
- raise IndexError, '%s has no node %s'%(link_class, value)
-
- # save off the value
- propvalues[key] = value
-
- # register the link with the newly linked node
- self.db.addjournal(link_class, value, 'link',
- (self.classname, newid, key))
-
- elif isinstance(prop, Multilink):
- if type(value) != type([]):
- raise TypeError, 'new property "%s" not a list of ids'%key
- link_class = self.properties[key].classname
- l = []
- for entry in value:
- if type(entry) != type(''):
- raise ValueError, 'link value must be String'
- # if it isn't a number, it's a key
- if not num_re.match(entry):
- try:
- entry = self.db.classes[link_class].lookup(entry)
- except (TypeError, KeyError):
- raise IndexError, 'new property "%s": %s not a %s'%(
- key, entry, self.properties[key].classname)
- l.append(entry)
- value = l
- propvalues[key] = value
-
- # handle additions
- for id in value:
- if not self.db.hasnode(link_class, id):
- raise IndexError, '%s has no node %s'%(link_class, id)
- # register the link with the newly linked node
- self.db.addjournal(link_class, id, 'link',
- (self.classname, newid, key))
-
- elif isinstance(prop, String):
- if type(value) != type(''):
- raise TypeError, 'new property "%s" not a string'%key
-
- elif isinstance(prop, Password):
- if not isinstance(value, password.Password):
- raise TypeError, 'new property "%s" not a Password'%key
-
- elif isinstance(prop, Date):
- if not isinstance(value, date.Date):
- raise TypeError, 'new property "%s" not a Date'%key
-
- elif isinstance(prop, Interval):
- if not isinstance(value, date.Interval):
- raise TypeError, 'new property "%s" not an Interval'%key
-
- # make sure there's data where there needs to be
- for key, prop in self.properties.items():
- if propvalues.has_key(key):
- continue
- if key == self.key:
- raise ValueError, 'key property "%s" is required'%key
- if isinstance(prop, Multilink):
- propvalues[key] = []
- else:
- propvalues[key] = None
-
- # convert all data to strings
- for key, prop in self.properties.items():
- if isinstance(prop, Date):
- propvalues[key] = propvalues[key].get_tuple()
- elif isinstance(prop, Interval):
- propvalues[key] = propvalues[key].get_tuple()
- elif isinstance(prop, Password):
- propvalues[key] = str(propvalues[key])
-
- # done
- self.db.addnode(self.classname, newid, propvalues)
- self.db.addjournal(self.classname, newid, 'create', propvalues)
- return newid
-
- def get(self, nodeid, propname, default=_marker):
+ _marker = []
+ def get(self, nodeid, propname, default=_marker, cache=1):
"""Get the value of a property on an existing node of this class.
'nodeid' must be the id of an existing node of this class or an
IndexError is raised. 'propname' must be the name of a property
of this class or a KeyError is raised.
- """
- if propname == 'id':
- return nodeid
-
- # get the node's dict
- d = self.db.getnode(self.classname, nodeid)
- if not d.has_key(propname) and default is not _marker:
- return default
- # get the value
- prop = self.properties[propname]
+ 'cache' exists for backwards compatibility, and is not used.
+ """
+ raise NotImplementedError
- # possibly convert the marshalled data to instances
- if isinstance(prop, Date):
- return date.Date(d[propname])
- elif isinstance(prop, Interval):
- return date.Interval(d[propname])
- elif isinstance(prop, Password):
- p = password.Password()
- p.unpack(d[propname])
- return p
+ # not in spec
+ def getnode(self, nodeid, cache=1):
+ ''' Return a convenience wrapper for the node.
- return d[propname]
+ 'nodeid' must be the id of an existing node of this class or an
+ IndexError is raised.
- # XXX not in spec
- def getnode(self, nodeid):
- ''' Return a convenience wrapper for the node
+ 'cache' exists for backwards compatibility, and is not used.
'''
return Node(self, nodeid)
+ def getnodeids(self, db=None):
+ '''Retrieve all the ids of the nodes for a particular Class.
+ '''
+ raise NotImplementedError
+
def set(self, nodeid, **propvalues):
"""Modify a property on an existing node of this class.
If the value of a Link or Multilink property contains an invalid
node id, a ValueError is raised.
"""
- if not propvalues:
- return
-
- if propvalues.has_key('id'):
- raise KeyError, '"id" is reserved'
-
- if self.db.journaltag is None:
- raise DatabaseError, 'Database open read-only'
-
- node = self.db.getnode(self.classname, nodeid)
- if node.has_key(self.db.RETIRED_FLAG):
- raise IndexError
- num_re = re.compile('^\d+$')
- for key, value in propvalues.items():
- # check to make sure we're not duplicating an existing key
- if key == self.key and node[key] != value:
- try:
- self.lookup(value)
- except KeyError:
- pass
- else:
- raise ValueError, 'node with key "%s" exists'%value
-
- # this will raise the KeyError if the property isn't valid
- # ... we don't use getprops() here because we only care about
- # the writeable properties.
- prop = self.properties[key]
-
- if isinstance(prop, Link):
- link_class = self.properties[key].classname
- # if it isn't a number, it's a key
- if type(value) != type(''):
- raise ValueError, 'link value must be String'
- if not num_re.match(value):
- try:
- value = self.db.classes[link_class].lookup(value)
- except (TypeError, KeyError):
- raise IndexError, 'new property "%s": %s not a %s'%(
- key, value, self.properties[key].classname)
-
- if not self.db.hasnode(link_class, value):
- raise IndexError, '%s has no node %s'%(link_class, value)
-
- # register the unlink with the old linked node
- if node[key] is not None:
- self.db.addjournal(link_class, node[key], 'unlink',
- (self.classname, nodeid, key))
-
- # register the link with the newly linked node
- if value is not None:
- self.db.addjournal(link_class, value, 'link',
- (self.classname, nodeid, key))
-
- elif isinstance(prop, Multilink):
- if type(value) != type([]):
- raise TypeError, 'new property "%s" not a list of ids'%key
- link_class = self.properties[key].classname
- l = []
- for entry in value:
- # if it isn't a number, it's a key
- if type(entry) != type(''):
- raise ValueError, 'link value must be String'
- if not num_re.match(entry):
- try:
- entry = self.db.classes[link_class].lookup(entry)
- except (TypeError, KeyError):
- raise IndexError, 'new property "%s": %s not a %s'%(
- key, entry, self.properties[key].classname)
- l.append(entry)
- value = l
- propvalues[key] = value
-
- #handle removals
- l = node[key]
- for id in l[:]:
- if id in value:
- continue
- # register the unlink with the old linked node
- self.db.addjournal(link_class, id, 'unlink',
- (self.classname, nodeid, key))
- l.remove(id)
-
- # handle additions
- for id in value:
- if not self.db.hasnode(link_class, id):
- raise IndexError, '%s has no node %s'%(link_class, id)
- if id in l:
- continue
- # register the link with the newly linked node
- self.db.addjournal(link_class, id, 'link',
- (self.classname, nodeid, key))
- l.append(id)
-
- elif isinstance(prop, String):
- if value is not None and type(value) != type(''):
- raise TypeError, 'new property "%s" not a string'%key
-
- elif isinstance(prop, Password):
- if not isinstance(value, password.Password):
- raise TypeError, 'new property "%s" not a Password'% key
- propvalues[key] = value = str(value)
-
- elif isinstance(prop, Date):
- if not isinstance(value, date.Date):
- raise TypeError, 'new property "%s" not a Date'% key
- propvalues[key] = value = value.get_tuple()
-
- elif isinstance(prop, Interval):
- if not isinstance(value, date.Interval):
- raise TypeError, 'new property "%s" not an Interval'% key
- propvalues[key] = value = value.get_tuple()
-
- node[key] = value
-
- self.db.setnode(self.classname, nodeid, node)
- self.db.addjournal(self.classname, nodeid, 'set', propvalues)
+ raise NotImplementedError
def retire(self, nodeid):
"""Retire a node.
Retired nodes are not returned by the find(), list(), or lookup()
methods, and other nodes may reuse the values of their key properties.
"""
- if self.db.journaltag is None:
- raise DatabaseError, 'Database open read-only'
- node = self.db.getnode(self.classname, nodeid)
- node[self.db.RETIRED_FLAG] = 1
- self.db.setnode(self.classname, nodeid, node)
- self.db.addjournal(self.classname, nodeid, 'retired', None)
+ raise NotImplementedError
+
+ def restore(self, nodeid):
+ '''Restpre a retired node.
+
+ Make node available for all operations like it was before retirement.
+ '''
+ raise NotImplementedError
+
+ def is_retired(self, nodeid):
+ '''Return true if the node is rerired
+ '''
+ raise NotImplementedError
+
+ def destroy(self, nodeid):
+ """Destroy a node.
+
+ WARNING: this method should never be used except in extremely rare
+ situations where there could never be links to the node being
+ deleted
+ WARNING: use retire() instead
+ WARNING: the properties of this node will not be available ever again
+ WARNING: really, use retire() instead
+
+ Well, I think that's enough warnings. This method exists mostly to
+ support the session storage of the cgi interface.
+
+ The node is completely removed from the hyperdb, including all journal
+ entries. It will no longer be available, and will generally break code
+ if there are any references to the node.
+ """
def history(self, nodeid):
"""Retrieve the journal of edits on a particular node.
'date' is a Timestamp object specifying the time of the change and
'tag' is the journaltag specified when the database was opened.
"""
- return self.db.getjournal(self.classname, nodeid)
+ raise NotImplementedError
# Locating nodes:
+ def hasnode(self, nodeid):
+ '''Determine if the given nodeid actually exists
+ '''
+ raise NotImplementedError
def setkey(self, propname):
"""Select a String property of this class to be the key property.
None, or a TypeError is raised. The values of the key property on
all existing nodes must be unique or a ValueError is raised.
"""
- # TODO: validate that the property is a String!
- self.key = propname
+ raise NotImplementedError
def getkey(self):
"""Return the name of the key property for this class or None."""
- return self.key
+ raise NotImplementedError
def labelprop(self, default_to_id=0):
''' Return the property name for a label for the given node.
3. "title" property
4. first property from the sorted property name list
'''
- k = self.getkey()
- if k:
- return k
- props = self.getprops()
- if props.has_key('name'):
- return 'name'
- elif props.has_key('title'):
- return 'title'
- if default_to_id:
- return 'id'
- props = props.keys()
- props.sort()
- return props[0]
-
- # TODO: set up a separate index db file for this? profile?
+ raise NotImplementedError
+
def lookup(self, keyvalue):
"""Locate a particular node by its key property and return its id.
the nodes in this class, the matching node's id is returned;
otherwise a KeyError is raised.
"""
- cldb = self.db.getclassdb(self.classname)
- for nodeid in self.db.getnodeids(self.classname, cldb):
- node = self.db.getnode(self.classname, nodeid, cldb)
- if node.has_key(self.db.RETIRED_FLAG):
- continue
- if node[self.key] == keyvalue:
- return nodeid
- raise KeyError, keyvalue
-
- # XXX: change from spec - allows multiple props to match
- def find(self, **propspec):
- """Get the ids of nodes in this class which link to a given node.
+ raise NotImplementedError
- 'propspec' consists of keyword args propname=nodeid
- 'propname' must be the name of a property in this class, or a
- KeyError is raised. That property must be a Link or Multilink
- property, or a TypeError is raised.
+ def find(self, **propspec):
+ """Get the ids of nodes in this class which link to the given nodes.
- 'nodeid' must be the id of an existing node in the class linked
- to by the given property, or an IndexError is raised.
- """
- propspec = propspec.items()
- for propname, nodeid in propspec:
- # check the prop is OK
- prop = self.properties[propname]
- if not isinstance(prop, Link) and not isinstance(prop, Multilink):
- raise TypeError, "'%s' not a Link/Multilink property"%propname
- if not self.db.hasnode(prop.classname, nodeid):
- raise ValueError, '%s has no node %s'%(prop.classname, nodeid)
-
- # ok, now do the find
- cldb = self.db.getclassdb(self.classname)
- l = []
- for id in self.db.getnodeids(self.classname, cldb):
- node = self.db.getnode(self.classname, id, cldb)
- if node.has_key(self.db.RETIRED_FLAG):
- continue
- for propname, nodeid in propspec:
- property = node[propname]
- if isinstance(prop, Link) and nodeid == property:
- l.append(id)
- elif isinstance(prop, Multilink) and nodeid in property:
- l.append(id)
- return l
+ 'propspec' consists of keyword args propname={nodeid:1,}
+ 'propname' must be the name of a property in this class, or a
+ KeyError is raised. That property must be a Link or Multilink
+ property, or a TypeError is raised.
- def stringFind(self, **requirements):
- """Locate a particular node by matching a set of its String
- properties in a caseless search.
+ Any node in this class whose 'propname' property links to any of the
+ nodeids will be returned. Used by the full text indexing, which knows
+ that "foo" occurs in msg1, msg3 and file7, so we have hits on these
+ issues:
- If the property is not a String property, a TypeError is raised.
-
- The return is a list of the id of all nodes that match.
+ db.issue.find(messages={'1':1,'3':1}, files={'7':1})
"""
- for propname in requirements.keys():
- prop = self.properties[propname]
- if isinstance(not prop, String):
- raise TypeError, "'%s' not a String property"%propname
- requirements[propname] = requirements[propname].lower()
- l = []
- cldb = self.db.getclassdb(self.classname)
- for nodeid in self.db.getnodeids(self.classname, cldb):
- node = self.db.getnode(self.classname, nodeid, cldb)
- if node.has_key(self.db.RETIRED_FLAG):
- continue
- for key, value in requirements.items():
- if node[key] and node[key].lower() != value:
- break
- else:
- l.append(nodeid)
- return l
-
- def list(self):
- """Return a list of the ids of the active nodes in this class."""
- l = []
- cn = self.classname
- cldb = self.db.getclassdb(cn)
- for nodeid in self.db.getnodeids(cn, cldb):
- node = self.db.getnode(cn, nodeid, cldb)
- if node.has_key(self.db.RETIRED_FLAG):
- continue
- l.append(nodeid)
- l.sort()
- return l
+ raise NotImplementedError
- # XXX not in spec
- def filter(self, filterspec, sort, group, num_re = re.compile('^\d+$')):
+ def filter(self, search_matches, filterspec, sort=(None,None),
+ group=(None,None)):
''' Return a list of the ids of the active nodes in this class that
match the 'filter' spec, sorted by the group spec and then the
- sort spec
- '''
- cn = self.classname
+ sort spec.
- # optimise filterspec
- l = []
- props = self.getprops()
- for k, v in filterspec.items():
- propclass = props[k]
- if isinstance(propclass, Link):
- if type(v) is not type([]):
- v = [v]
- # replace key values with node ids
- u = []
- link_class = self.db.classes[propclass.classname]
- for entry in v:
- if entry == '-1': entry = None
- elif not num_re.match(entry):
- try:
- entry = link_class.lookup(entry)
- except (TypeError,KeyError):
- raise ValueError, 'property "%s": %s not a %s'%(
- k, entry, self.properties[k].classname)
- u.append(entry)
-
- l.append((0, k, u))
- elif isinstance(propclass, Multilink):
- if type(v) is not type([]):
- v = [v]
- # replace key values with node ids
- u = []
- link_class = self.db.classes[propclass.classname]
- for entry in v:
- if not num_re.match(entry):
- try:
- entry = link_class.lookup(entry)
- except (TypeError,KeyError):
- raise ValueError, 'new property "%s": %s not a %s'%(
- k, entry, self.properties[k].classname)
- u.append(entry)
- l.append((1, k, u))
- elif isinstance(propclass, String):
- # simple glob searching
- v = re.sub(r'([\|\{\}\\\.\+\[\]\(\)])', r'\\\1', v)
- v = v.replace('?', '.')
- v = v.replace('*', '.*?')
- l.append((2, k, re.compile(v, re.I)))
- else:
- l.append((6, k, v))
- filterspec = l
+ "filterspec" is {propname: value(s)}
+ "sort" and "group" are (dir, prop) where dir is '+', '-' or None
+ and prop is a prop name or None
+ "search_matches" is {nodeid: marker}
- # now, find all the nodes that are active and pass filtering
- l = []
- cldb = self.db.getclassdb(cn)
- for nodeid in self.db.getnodeids(cn, cldb):
- node = self.db.getnode(cn, nodeid, cldb)
- if node.has_key(self.db.RETIRED_FLAG):
- continue
- # apply filter
- for t, k, v in filterspec:
- # this node doesn't have this property, so reject it
- if not node.has_key(k): break
-
- if t == 0 and node[k] not in v:
- # link - if this node'd property doesn't appear in the
- # filterspec's nodeid list, skip it
- break
- elif t == 1:
- # multilink - if any of the nodeids required by the
- # filterspec aren't in this node's property, then skip
- # it
- for value in v:
- if value not in node[k]:
- break
- else:
- continue
- break
- elif t == 2 and not v.search(node[k]):
- # RE search
- break
- elif t == 6 and node[k] != v:
- # straight value comparison for the other types
- break
- else:
- l.append((nodeid, node))
- l.sort()
-
- # optimise sort
- m = []
- for entry in sort:
- if entry[0] != '-':
- m.append(('+', entry))
- else:
- m.append((entry[0], entry[1:]))
- sort = m
-
- # optimise group
- m = []
- for entry in group:
- if entry[0] != '-':
- m.append(('+', entry))
- else:
- m.append((entry[0], entry[1:]))
- group = m
- # now, sort the result
- def sortfun(a, b, sort=sort, group=group, properties=self.getprops(),
- db = self.db, cl=self):
- a_id, an = a
- b_id, bn = b
- # sort by group and then sort
- for list in group, sort:
- for dir, prop in list:
- # sorting is class-specific
- propclass = properties[prop]
-
- # handle the properties that might be "faked"
- # also, handle possible missing properties
- try:
- if not an.has_key(prop):
- an[prop] = cl.get(a_id, prop)
- av = an[prop]
- except KeyError:
- # the node doesn't have a value for this property
- if isinstance(propclass, Multilink): av = []
- else: av = ''
- try:
- if not bn.has_key(prop):
- bn[prop] = cl.get(b_id, prop)
- bv = bn[prop]
- except KeyError:
- # the node doesn't have a value for this property
- if isinstance(propclass, Multilink): bv = []
- else: bv = ''
-
- # String and Date values are sorted in the natural way
- if isinstance(propclass, String):
- # clean up the strings
- if av and av[0] in string.uppercase:
- av = an[prop] = av.lower()
- if bv and bv[0] in string.uppercase:
- bv = bn[prop] = bv.lower()
- if (isinstance(propclass, String) or
- isinstance(propclass, Date)):
- # it might be a string that's really an integer
- try:
- av = int(av)
- bv = int(bv)
- except:
- pass
- if dir == '+':
- r = cmp(av, bv)
- if r != 0: return r
- elif dir == '-':
- r = cmp(bv, av)
- if r != 0: return r
-
- # Link properties are sorted according to the value of
- # the "order" property on the linked nodes if it is
- # present; or otherwise on the key string of the linked
- # nodes; or finally on the node ids.
- elif isinstance(propclass, Link):
- link = db.classes[propclass.classname]
- if av is None and bv is not None: return -1
- if av is not None and bv is None: return 1
- if av is None and bv is None: continue
- if link.getprops().has_key('order'):
- if dir == '+':
- r = cmp(link.get(av, 'order'),
- link.get(bv, 'order'))
- if r != 0: return r
- elif dir == '-':
- r = cmp(link.get(bv, 'order'),
- link.get(av, 'order'))
- if r != 0: return r
- elif link.getkey():
- key = link.getkey()
- if dir == '+':
- r = cmp(link.get(av, key), link.get(bv, key))
- if r != 0: return r
- elif dir == '-':
- r = cmp(link.get(bv, key), link.get(av, key))
- if r != 0: return r
- else:
- if dir == '+':
- r = cmp(av, bv)
- if r != 0: return r
- elif dir == '-':
- r = cmp(bv, av)
- if r != 0: return r
-
- # Multilink properties are sorted according to how many
- # links are present.
- elif isinstance(propclass, Multilink):
- if dir == '+':
- r = cmp(len(av), len(bv))
- if r != 0: return r
- elif dir == '-':
- r = cmp(len(bv), len(av))
- if r != 0: return r
- # end for dir, prop in list:
- # end for list in sort, group:
- # if all else fails, compare the ids
- return cmp(a[0], b[0])
-
- l.sort(sortfun)
- return [i[0] for i in l]
+ The filter must match all properties specificed - but if the
+ property value to match is a list, any one of the values in the
+ list may match for that property to match.
+ '''
+ raise NotImplementedError
def count(self):
"""Get the number of nodes in this class.
in this class run from 1 to numnodes, and numnodes+1 will be the
id of the next node to be created in this class.
"""
- return self.db.countnodes(self.classname)
+ raise NotImplementedError
# Manipulating properties:
-
def getprops(self, protected=1):
"""Return a dictionary mapping property names to property objects.
If the "protected" flag is true, we include protected properties -
- those which may not be modified."""
- d = self.properties.copy()
- if protected:
- d['id'] = String()
- return d
+ those which may not be modified.
+ """
+ raise NotImplementedError
def addprop(self, **properties):
"""Add properties to this class.
may collide with the names of existing properties, or a ValueError
is raised before any properties have been added.
"""
- for key in properties.keys():
- if self.properties.has_key(key):
- raise ValueError, key
- self.properties.update(properties)
+ raise NotImplementedError
+
+ def index(self, nodeid):
+ '''Add (or refresh) the node to search indexes
+ '''
+ raise NotImplementedError
+
+ def safeget(self, nodeid, propname, default=None):
+ """Safely get the value of a property on an existing node of this class.
+
+ Return 'default' if the node doesn't exist.
+ """
+ try:
+ return self.get(nodeid, propname)
+ except IndexError:
+ return default
+
+class HyperdbValueError(ValueError):
+ ''' Error converting a raw value into a Hyperdb value '''
+ pass
+def convertLinkValue(db, propname, prop, value, idre=re.compile('\d+')):
+ ''' Convert the link value (may be id or key value) to an id value. '''
+ linkcl = db.classes[prop.classname]
+ if not idre.match(value):
+ if linkcl.getkey():
+ try:
+ value = linkcl.lookup(value)
+ except KeyError, message:
+ raise HyperdbValueError, 'property %s: %r is not a %s.'%(
+ propname, value, prop.classname)
+ else:
+ raise HyperdbValueError, 'you may only enter ID values '\
+ 'for property %s'%propname
+ return value
+
+def fixNewlines(text):
+ """ Homogenise line endings.
+
+ Different web clients send different line ending values, but
+ other systems (eg. email) don't necessarily handle those line
+ endings. Our solution is to convert all line endings to LF.
+ """
+ text = text.replace('\r\n', '\n')
+ return text.replace('\r', '\n')
+
+def rawToHyperdb(db, klass, itemid, propname, value,
+ pwre=re.compile(r'{(\w+)}(.+)')):
+ ''' Convert the raw (user-input) value to a hyperdb-storable value. The
+ value is for the "propname" property on itemid (may be None for a
+ new item) of "klass" in "db".
+
+ The value is usually a string, but in the case of multilink inputs
+ it may be either a list of strings or a string with comma-separated
+ values.
+ '''
+ properties = klass.getprops()
+
+ # ensure it's a valid property name
+ propname = propname.strip()
+ try:
+ proptype = properties[propname]
+ except KeyError:
+ raise HyperdbValueError, '%r is not a property of %s'%(propname,
+ klass.classname)
+
+ # if we got a string, strip it now
+ if isinstance(value, type('')):
+ value = value.strip()
+
+ # convert the input value to a real property value
+ if isinstance(proptype, String):
+ # fix the CRLF/CR -> LF stuff
+ value = fixNewlines(value)
+ if isinstance(proptype, Password):
+ m = pwre.match(value)
+ if m:
+ # password is being given to us encrypted
+ p = password.Password()
+ p.scheme = m.group(1)
+ if p.scheme not in 'SHA crypt plaintext'.split():
+ raise HyperdbValueError, 'property %s: unknown encryption '\
+ 'scheme %r'%(propname, p.scheme)
+ p.password = m.group(2)
+ value = p
+ else:
+ try:
+ value = password.Password(value)
+ except password.PasswordValueError, message:
+ raise HyperdbValueError, 'property %s: %s'%(propname, message)
+ elif isinstance(proptype, Date):
+ try:
+ tz = db.getUserTimezone()
+ value = date.Date(value).local(tz)
+ except ValueError, message:
+ raise HyperdbValueError, 'property %s: %r is an invalid '\
+ 'date (%s)'%(propname, value, message)
+ elif isinstance(proptype, Interval):
+ try:
+ value = date.Interval(value)
+ except ValueError, message:
+ raise HyperdbValueError, 'property %s: %r is an invalid '\
+ 'date interval (%s)'%(propname, value, message)
+ elif isinstance(proptype, Link):
+ if value == '-1' or not value:
+ value = None
+ else:
+ value = convertLinkValue(db, propname, proptype, value)
+
+ elif isinstance(proptype, Multilink):
+ # get the current item value if it's not a new item
+ if itemid and not itemid.startswith('-'):
+ curvalue = klass.get(itemid, propname)
+ else:
+ curvalue = []
+
+ # if the value is a comma-separated string then split it now
+ if isinstance(value, type('')):
+ value = value.split(',')
+
+ # handle each add/remove in turn
+ # keep an extra list for all items that are
+ # definitely in the new list (in case of e.g.
+ # <propname>=A,+B, which should replace the old
+ # list with A,B)
+ set = 1
+ newvalue = []
+ for item in value:
+ item = item.strip()
+
+ # skip blanks
+ if not item: continue
+
+ # handle +/-
+ remove = 0
+ if item.startswith('-'):
+ remove = 1
+ item = item[1:]
+ set = 0
+ elif item.startswith('+'):
+ item = item[1:]
+ set = 0
+
+ # look up the value
+ itemid = convertLinkValue(db, propname, proptype, item)
+
+ # perform the add/remove
+ if remove:
+ try:
+ curvalue.remove(itemid)
+ except ValueError:
+ raise HyperdbValueError, 'property %s: %r is not ' \
+ 'currently an element'%(propname, item)
+ else:
+ newvalue.append(itemid)
+ if itemid not in curvalue:
+ curvalue.append(itemid)
+
+ # that's it, set the new Multilink property value,
+ # or overwrite it completely
+ if set:
+ value = newvalue
+ else:
+ value = curvalue
+
+ # TODO: one day, we'll switch to numeric ids and this will be
+ # unnecessary :(
+ value = [int(x) for x in value]
+ value.sort()
+ value = [str(x) for x in value]
+ elif isinstance(proptype, Boolean):
+ value = value.strip()
+ value = value.lower() in ('yes', 'true', 'on', '1')
+ elif isinstance(proptype, Number):
+ value = value.strip()
+ try:
+ value = float(value)
+ except ValueError:
+ raise HyperdbValueError, 'property %s: %r is not a number'%(
+ propname, value)
+ return value
+
+class FileClass:
+ ''' A class that requires the "content" property and stores it on
+ disk.
+ '''
+ pass
-# XXX not in spec
class Node:
''' A convenience wrapper for the given node
'''
- def __init__(self, cl, nodeid):
+ def __init__(self, cl, nodeid, cache=1):
self.__dict__['cl'] = cl
self.__dict__['nodeid'] = nodeid
def keys(self, protected=1):
return l
def has_key(self, name):
return self.cl.getprops().has_key(name)
+ def get(self, name, default=None):
+ if self.has_key(name):
+ return self[name]
+ else:
+ return default
def __getattr__(self, name):
if self.__dict__.has_key(name):
return self.__dict__[name]
return self.cl.retire(self.nodeid)
-def Choice(name, *options):
- cl = Class(db, name, name=hyperdb.String(), order=hyperdb.String())
+def Choice(name, db, *options):
+ '''Quick helper to create a simple class with choices
+ '''
+ cl = Class(db, name, name=String(), order=String())
for i in range(len(options)):
- cl.create(name=option[i], order=i)
+ cl.create(name=options[i], order=i)
return hyperdb.Link(name)
-#
-# $Log: not supported by cvs2svn $
-# Revision 1.42 2001/12/16 10:53:37 richard
-# take a copy of the node dict so that the subsequent set
-# operation doesn't modify the oldvalues structure
-#
-# Revision 1.41 2001/12/15 23:47:47 richard
-# Cleaned up some bare except statements
-#
-# Revision 1.40 2001/12/14 23:42:57 richard
-# yuck, a gdbm instance tests false :(
-# I've left the debugging code in - it should be removed one day if we're ever
-# _really_ anal about performace :)
-#
-# Revision 1.39 2001/12/02 05:06:16 richard
-# . We now use weakrefs in the Classes to keep the database reference, so
-# the close() method on the database is no longer needed.
-# I bumped the minimum python requirement up to 2.1 accordingly.
-# . #487480 ] roundup-server
-# . #487476 ] INSTALL.txt
-#
-# I also cleaned up the change message / post-edit stuff in the cgi client.
-# There's now a clearly marked "TODO: append the change note" where I believe
-# the change note should be added there. The "changes" list will obviously
-# have to be modified to be a dict of the changes, or somesuch.
-#
-# More testing needed.
-#
-# Revision 1.38 2001/12/01 07:17:50 richard
-# . We now have basic transaction support! Information is only written to
-# the database when the commit() method is called. Only the anydbm
-# backend is modified in this way - neither of the bsddb backends have been.
-# The mail, admin and cgi interfaces all use commit (except the admin tool
-# doesn't have a commit command, so interactive users can't commit...)
-# . Fixed login/registration forwarding the user to the right page (or not,
-# on a failure)
-#
-# Revision 1.37 2001/11/28 21:55:35 richard
-# . login_action and newuser_action return values were being ignored
-# . Woohoo! Found that bloody re-login bug that was killing the mail
-# gateway.
-# (also a minor cleanup in hyperdb)
-#
-# Revision 1.36 2001/11/27 03:16:09 richard
-# Another place that wasn't handling missing properties.
-#
-# Revision 1.35 2001/11/22 15:46:42 jhermann
-# Added module docstrings to all modules.
-#
-# Revision 1.34 2001/11/21 04:04:43 richard
-# *sigh* more missing value handling
-#
-# Revision 1.33 2001/11/21 03:40:54 richard
-# more new property handling
-#
-# Revision 1.32 2001/11/21 03:11:28 richard
-# Better handling of new properties.
-#
-# Revision 1.31 2001/11/12 22:01:06 richard
-# Fixed issues with nosy reaction and author copies.
-#
-# Revision 1.30 2001/11/09 10:11:08 richard
-# . roundup-admin now handles all hyperdb exceptions
-#
-# Revision 1.29 2001/10/27 00:17:41 richard
-# Made Class.stringFind() do caseless matching.
-#
-# Revision 1.28 2001/10/21 04:44:50 richard
-# bug #473124: UI inconsistency with Link fields.
-# This also prompted me to fix a fairly long-standing usability issue -
-# that of being able to turn off certain filters.
-#
-# Revision 1.27 2001/10/20 23:44:27 richard
-# Hyperdatabase sorts strings-that-look-like-numbers as numbers now.
-#
-# Revision 1.26 2001/10/16 03:48:01 richard
-# admin tool now complains if a "find" is attempted with a non-link property.
-#
-# Revision 1.25 2001/10/11 00:17:51 richard
-# Reverted a change in hyperdb so the default value for missing property
-# values in a create() is None and not '' (the empty string.) This obviously
-# breaks CSV import/export - the string 'None' will be created in an
-# export/import operation.
-#
-# Revision 1.24 2001/10/10 03:54:57 richard
-# Added database importing and exporting through CSV files.
-# Uses the csv module from object-craft for exporting if it's available.
-# Requires the csv module for importing.
-#
-# Revision 1.23 2001/10/09 23:58:10 richard
-# Moved the data stringification up into the hyperdb.Class class' get, set
-# and create methods. This means that the data is also stringified for the
-# journal call, and removes duplication of code from the backends. The
-# backend code now only sees strings.
-#
-# Revision 1.22 2001/10/09 07:25:59 richard
-# Added the Password property type. See "pydoc roundup.password" for
-# implementation details. Have updated some of the documentation too.
-#
-# Revision 1.21 2001/10/05 02:23:24 richard
-# . roundup-admin create now prompts for property info if none is supplied
-# on the command-line.
-# . hyperdb Class getprops() method may now return only the mutable
-# properties.
-# . Login now uses cookies, which makes it a whole lot more flexible. We can
-# now support anonymous user access (read-only, unless there's an
-# "anonymous" user, in which case write access is permitted). Login
-# handling has been moved into cgi_client.Client.main()
-# . The "extended" schema is now the default in roundup init.
-# . The schemas have had their page headings modified to cope with the new
-# login handling. Existing installations should copy the interfaces.py
-# file from the roundup lib directory to their instance home.
-# . Incorrectly had a Bizar Software copyright on the cgitb.py module from
-# Ping - has been removed.
-# . Fixed a whole bunch of places in the CGI interface where we should have
-# been returning Not Found instead of throwing an exception.
-# . Fixed a deviation from the spec: trying to modify the 'id' property of
-# an item now throws an exception.
-#
-# Revision 1.20 2001/10/04 02:12:42 richard
-# Added nicer command-line item adding: passing no arguments will enter an
-# interactive more which asks for each property in turn. While I was at it, I
-# fixed an implementation problem WRT the spec - I wasn't raising a
-# ValueError if the key property was missing from a create(). Also added a
-# protected=boolean argument to getprops() so we can list only the mutable
-# properties (defaults to yes, which lists the immutables).
-#
-# Revision 1.19 2001/08/29 04:47:18 richard
-# Fixed CGI client change messages so they actually include the properties
-# changed (again).
-#
-# Revision 1.18 2001/08/16 07:34:59 richard
-# better CGI text searching - but hidden filter fields are disappearing...
-#
-# Revision 1.17 2001/08/16 06:59:58 richard
-# all searches use re now - and they're all case insensitive
-#
-# Revision 1.16 2001/08/15 23:43:18 richard
-# Fixed some isFooTypes that I missed.
-# Refactored some code in the CGI code.
-#
-# Revision 1.15 2001/08/12 06:32:36 richard
-# using isinstance(blah, Foo) now instead of isFooType
-#
-# Revision 1.14 2001/08/07 00:24:42 richard
-# stupid typo
-#
-# Revision 1.13 2001/08/07 00:15:51 richard
-# Added the copyright/license notice to (nearly) all files at request of
-# Bizar Software.
-#
-# Revision 1.12 2001/08/02 06:38:17 richard
-# Roundupdb now appends "mailing list" information to its messages which
-# include the e-mail address and web interface address. Templates may
-# override this in their db classes to include specific information (support
-# instructions, etc).
-#
-# Revision 1.11 2001/08/01 04:24:21 richard
-# mailgw was assuming certain properties existed on the issues being created.
-#
-# Revision 1.10 2001/07/30 02:38:31 richard
-# get() now has a default arg - for migration only.
-#
-# Revision 1.9 2001/07/29 09:28:23 richard
-# Fixed sorting by clicking on column headings.
-#
-# Revision 1.8 2001/07/29 08:27:40 richard
-# Fixed handling of passed-in values in form elements (ie. during a
-# drill-down)
-#
-# Revision 1.7 2001/07/29 07:01:39 richard
-# Added vim command to all source so that we don't get no steenkin' tabs :)
-#
-# Revision 1.6 2001/07/29 05:36:14 richard
-# Cleanup of the link label generation.
-#
-# Revision 1.5 2001/07/29 04:05:37 richard
-# Added the fabricated property "id".
-#
-# Revision 1.4 2001/07/27 06:25:35 richard
-# Fixed some of the exceptions so they're the right type.
-# Removed the str()-ification of node ids so we don't mask oopsy errors any
-# more.
-#
-# Revision 1.3 2001/07/27 05:17:14 richard
-# just some comments
-#
-# Revision 1.2 2001/07/22 12:09:32 richard
-# Final commit of Grande Splite
-#
-# Revision 1.1 2001/07/22 11:58:35 richard
-# More Grande Splite
-#
-#
# vim: set filetype=python ts=4 sw=4 et si