1 #
2 # Copyright (c) 2001 Bizar Software Pty Ltd (http://www.bizarsoftware.com.au/)
3 # This module is free software, and you may redistribute it and/or modify
4 # under the same terms as Python, so long as this copyright message and
5 # disclaimer are retained in their original form.
6 #
7 # IN NO EVENT SHALL BIZAR SOFTWARE PTY LTD BE LIABLE TO ANY PARTY FOR
8 # DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES ARISING
9 # OUT OF THE USE OF THIS CODE, EVEN IF THE AUTHOR HAS BEEN ADVISED OF THE
10 # POSSIBILITY OF SUCH DAMAGE.
11 #
12 # BIZAR SOFTWARE PTY LTD SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
13 # BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
14 # FOR A PARTICULAR PURPOSE. THE CODE PROVIDED HEREUNDER IS ON AN "AS IS"
15 # BASIS, AND THERE IS NO OBLIGATION WHATSOEVER TO PROVIDE MAINTENANCE,
16 # SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
17 #
18 # $Id: hyperdb.py,v 1.95 2003-11-16 22:56:46 jlgijsbers Exp $
20 """
21 Hyperdatabase implementation, especially field types.
22 """
24 # standard python modules
25 import sys, os, time, re
27 # roundup modules
28 import date, password
30 # configure up the DEBUG and TRACE captures
31 class Sink:
32 def write(self, content):
33 pass
34 DEBUG = os.environ.get('HYPERDBDEBUG', '')
35 if DEBUG and __debug__:
36 if DEBUG == 'stdout':
37 DEBUG = sys.stdout
38 else:
39 DEBUG = open(DEBUG, 'a')
40 else:
41 DEBUG = Sink()
42 TRACE = os.environ.get('HYPERDBTRACE', '')
43 if TRACE and __debug__:
44 if TRACE == 'stdout':
45 TRACE = sys.stdout
46 else:
47 TRACE = open(TRACE, 'w')
48 else:
49 TRACE = Sink()
50 def traceMark():
51 print >>TRACE, '**MARK', time.ctime()
52 del Sink
54 #
55 # Types
56 #
57 class String:
58 """An object designating a String property."""
59 def __init__(self, indexme='no'):
60 self.indexme = indexme == 'yes'
61 def __repr__(self):
62 ' more useful for dumps '
63 return '<%s>'%self.__class__
65 class Password:
66 """An object designating a Password property."""
67 def __repr__(self):
68 ' more useful for dumps '
69 return '<%s>'%self.__class__
71 class Date:
72 """An object designating a Date property."""
73 def __repr__(self):
74 ' more useful for dumps '
75 return '<%s>'%self.__class__
77 class Interval:
78 """An object designating an Interval property."""
79 def __repr__(self):
80 ' more useful for dumps '
81 return '<%s>'%self.__class__
83 class Link:
84 """An object designating a Link property that links to a
85 node in a specified class."""
86 def __init__(self, classname, do_journal='yes'):
87 ''' Default is to not journal link and unlink events
88 '''
89 self.classname = classname
90 self.do_journal = do_journal == 'yes'
91 def __repr__(self):
92 ' more useful for dumps '
93 return '<%s to "%s">'%(self.__class__, self.classname)
95 class Multilink:
96 """An object designating a Multilink property that links
97 to nodes in a specified class.
99 "classname" indicates the class to link to
101 "do_journal" indicates whether the linked-to nodes should have
102 'link' and 'unlink' events placed in their journal
103 """
104 def __init__(self, classname, do_journal='yes'):
105 ''' Default is to not journal link and unlink events
106 '''
107 self.classname = classname
108 self.do_journal = do_journal == 'yes'
109 def __repr__(self):
110 ' more useful for dumps '
111 return '<%s to "%s">'%(self.__class__, self.classname)
113 class Boolean:
114 """An object designating a boolean property"""
115 def __repr__(self):
116 'more useful for dumps'
117 return '<%s>' % self.__class__
119 class Number:
120 """An object designating a numeric property"""
121 def __repr__(self):
122 'more useful for dumps'
123 return '<%s>' % self.__class__
124 #
125 # Support for splitting designators
126 #
127 class DesignatorError(ValueError):
128 pass
129 def splitDesignator(designator, dre=re.compile(r'([^\d]+)(\d+)')):
130 ''' Take a foo123 and return ('foo', 123)
131 '''
132 m = dre.match(designator)
133 if m is None:
134 raise DesignatorError, '"%s" not a node designator'%designator
135 return m.group(1), m.group(2)
137 #
138 # the base Database class
139 #
140 class DatabaseError(ValueError):
141 '''Error to be raised when there is some problem in the database code
142 '''
143 pass
144 class Database:
145 '''A database for storing records containing flexible data types.
147 This class defines a hyperdatabase storage layer, which the Classes use to
148 store their data.
151 Transactions
152 ------------
153 The Database should support transactions through the commit() and
154 rollback() methods. All other Database methods should be transaction-aware,
155 using data from the current transaction before looking up the database.
157 An implementation must provide an override for the get() method so that the
158 in-database value is returned in preference to the in-transaction value.
159 This is necessary to determine if any values have changed during a
160 transaction.
163 Implementation
164 --------------
166 All methods except __repr__ must be implemented by a concrete backend Database.
168 '''
170 # flag to set on retired entries
171 RETIRED_FLAG = '__hyperdb_retired'
173 def __init__(self, config, journaltag=None):
174 """Open a hyperdatabase given a specifier to some storage.
176 The 'storagelocator' is obtained from config.DATABASE.
177 The meaning of 'storagelocator' depends on the particular
178 implementation of the hyperdatabase. It could be a file name,
179 a directory path, a socket descriptor for a connection to a
180 database over the network, etc.
182 The 'journaltag' is a token that will be attached to the journal
183 entries for any edits done on the database. If 'journaltag' is
184 None, the database is opened in read-only mode: the Class.create(),
185 Class.set(), and Class.retire() methods are disabled.
186 """
187 raise NotImplementedError
189 def post_init(self):
190 """Called once the schema initialisation has finished.
191 If 'refresh' is true, we want to rebuild the backend
192 structures.
193 """
194 raise NotImplementedError
196 def refresh_database(self):
197 """Called to indicate that the backend should rebuild all tables
198 and structures. Not called in normal usage."""
199 raise NotImplementedError
201 def __getattr__(self, classname):
202 """A convenient way of calling self.getclass(classname)."""
203 raise NotImplementedError
205 def addclass(self, cl):
206 '''Add a Class to the hyperdatabase.
207 '''
208 raise NotImplementedError
210 def getclasses(self):
211 """Return a list of the names of all existing classes."""
212 raise NotImplementedError
214 def getclass(self, classname):
215 """Get the Class object representing a particular class.
217 If 'classname' is not a valid class name, a KeyError is raised.
218 """
219 raise NotImplementedError
221 def clear(self):
222 '''Delete all database contents.
223 '''
224 raise NotImplementedError
226 def getclassdb(self, classname, mode='r'):
227 '''Obtain a connection to the class db that will be used for
228 multiple actions.
229 '''
230 raise NotImplementedError
232 def addnode(self, classname, nodeid, node):
233 """Add the specified node to its class's db.
234 """
235 raise NotImplementedError
237 def serialise(self, classname, node):
238 '''Copy the node contents, converting non-marshallable data into
239 marshallable data.
240 '''
241 return node
243 def setnode(self, classname, nodeid, node):
244 '''Change the specified node.
245 '''
246 raise NotImplementedError
248 def unserialise(self, classname, node):
249 '''Decode the marshalled node data
250 '''
251 return node
253 def getnode(self, classname, nodeid, db=None, cache=1):
254 '''Get a node from the database.
256 'cache' exists for backwards compatibility, and is not used.
257 '''
258 raise NotImplementedError
260 def hasnode(self, classname, nodeid, db=None):
261 '''Determine if the database has a given node.
262 '''
263 raise NotImplementedError
265 def countnodes(self, classname, db=None):
266 '''Count the number of nodes that exist for a particular Class.
267 '''
268 raise NotImplementedError
270 def storefile(self, classname, nodeid, property, content):
271 '''Store the content of the file in the database.
273 The property may be None, in which case the filename does not
274 indicate which property is being saved.
275 '''
276 raise NotImplementedError
278 def getfile(self, classname, nodeid, property):
279 '''Store the content of the file in the database.
280 '''
281 raise NotImplementedError
283 def addjournal(self, classname, nodeid, action, params):
284 ''' Journal the Action
285 'action' may be:
287 'create' or 'set' -- 'params' is a dictionary of property values
288 'link' or 'unlink' -- 'params' is (classname, nodeid, propname)
289 'retire' -- 'params' is None
290 '''
291 raise NotImplementedError
293 def getjournal(self, classname, nodeid):
294 ''' get the journal for id
295 '''
296 raise NotImplementedError
298 def pack(self, pack_before):
299 ''' pack the database
300 '''
301 raise NotImplementedError
303 def commit(self):
304 ''' Commit the current transactions.
306 Save all data changed since the database was opened or since the
307 last commit() or rollback().
308 '''
309 raise NotImplementedError
311 def rollback(self):
312 ''' Reverse all actions from the current transaction.
314 Undo all the changes made since the database was opened or the last
315 commit() or rollback() was performed.
316 '''
317 raise NotImplementedError
319 #
320 # The base Class class
321 #
322 class Class:
323 """ The handle to a particular class of nodes in a hyperdatabase.
325 All methods except __repr__ and getnode must be implemented by a
326 concrete backend Class.
327 """
329 def __init__(self, db, classname, **properties):
330 """Create a new class with a given name and property specification.
332 'classname' must not collide with the name of an existing class,
333 or a ValueError is raised. The keyword arguments in 'properties'
334 must map names to property objects, or a TypeError is raised.
335 """
336 raise NotImplementedError
338 def __repr__(self):
339 '''Slightly more useful representation
340 '''
341 return '<hyperdb.Class "%s">'%self.classname
343 # Editing nodes:
345 def create(self, **propvalues):
346 """Create a new node of this class and return its id.
348 The keyword arguments in 'propvalues' map property names to values.
350 The values of arguments must be acceptable for the types of their
351 corresponding properties or a TypeError is raised.
353 If this class has a key property, it must be present and its value
354 must not collide with other key strings or a ValueError is raised.
356 Any other properties on this class that are missing from the
357 'propvalues' dictionary are set to None.
359 If an id in a link or multilink property does not refer to a valid
360 node, an IndexError is raised.
361 """
362 raise NotImplementedError
364 _marker = []
365 def get(self, nodeid, propname, default=_marker, cache=1):
366 """Get the value of a property on an existing node of this class.
368 'nodeid' must be the id of an existing node of this class or an
369 IndexError is raised. 'propname' must be the name of a property
370 of this class or a KeyError is raised.
372 'cache' exists for backwards compatibility, and is not used.
373 """
374 raise NotImplementedError
376 # not in spec
377 def getnode(self, nodeid, cache=1):
378 ''' Return a convenience wrapper for the node.
380 'nodeid' must be the id of an existing node of this class or an
381 IndexError is raised.
383 'cache' exists for backwards compatibility, and is not used.
384 '''
385 return Node(self, nodeid)
387 def getnodeids(self, db=None):
388 '''Retrieve all the ids of the nodes for a particular Class.
389 '''
390 raise NotImplementedError
392 def set(self, nodeid, **propvalues):
393 """Modify a property on an existing node of this class.
395 'nodeid' must be the id of an existing node of this class or an
396 IndexError is raised.
398 Each key in 'propvalues' must be the name of a property of this
399 class or a KeyError is raised.
401 All values in 'propvalues' must be acceptable types for their
402 corresponding properties or a TypeError is raised.
404 If the value of the key property is set, it must not collide with
405 other key strings or a ValueError is raised.
407 If the value of a Link or Multilink property contains an invalid
408 node id, a ValueError is raised.
409 """
410 raise NotImplementedError
412 def retire(self, nodeid):
413 """Retire a node.
415 The properties on the node remain available from the get() method,
416 and the node's id is never reused.
418 Retired nodes are not returned by the find(), list(), or lookup()
419 methods, and other nodes may reuse the values of their key properties.
420 """
421 raise NotImplementedError
423 def restore(self, nodeid):
424 '''Restpre a retired node.
426 Make node available for all operations like it was before retirement.
427 '''
428 raise NotImplementedError
430 def is_retired(self, nodeid):
431 '''Return true if the node is rerired
432 '''
433 raise NotImplementedError
435 def destroy(self, nodeid):
436 """Destroy a node.
438 WARNING: this method should never be used except in extremely rare
439 situations where there could never be links to the node being
440 deleted
441 WARNING: use retire() instead
442 WARNING: the properties of this node will not be available ever again
443 WARNING: really, use retire() instead
445 Well, I think that's enough warnings. This method exists mostly to
446 support the session storage of the cgi interface.
448 The node is completely removed from the hyperdb, including all journal
449 entries. It will no longer be available, and will generally break code
450 if there are any references to the node.
451 """
453 def history(self, nodeid):
454 """Retrieve the journal of edits on a particular node.
456 'nodeid' must be the id of an existing node of this class or an
457 IndexError is raised.
459 The returned list contains tuples of the form
461 (date, tag, action, params)
463 'date' is a Timestamp object specifying the time of the change and
464 'tag' is the journaltag specified when the database was opened.
465 """
466 raise NotImplementedError
468 # Locating nodes:
469 def hasnode(self, nodeid):
470 '''Determine if the given nodeid actually exists
471 '''
472 raise NotImplementedError
474 def setkey(self, propname):
475 """Select a String property of this class to be the key property.
477 'propname' must be the name of a String property of this class or
478 None, or a TypeError is raised. The values of the key property on
479 all existing nodes must be unique or a ValueError is raised.
480 """
481 raise NotImplementedError
483 def getkey(self):
484 """Return the name of the key property for this class or None."""
485 raise NotImplementedError
487 def labelprop(self, default_to_id=0):
488 ''' Return the property name for a label for the given node.
490 This method attempts to generate a consistent label for the node.
491 It tries the following in order:
492 1. key property
493 2. "name" property
494 3. "title" property
495 4. first property from the sorted property name list
496 '''
497 raise NotImplementedError
499 def lookup(self, keyvalue):
500 """Locate a particular node by its key property and return its id.
502 If this class has no key property, a TypeError is raised. If the
503 'keyvalue' matches one of the values for the key property among
504 the nodes in this class, the matching node's id is returned;
505 otherwise a KeyError is raised.
506 """
507 raise NotImplementedError
509 def find(self, **propspec):
510 """Get the ids of nodes in this class which link to the given nodes.
512 'propspec' consists of keyword args propname={nodeid:1,}
513 'propname' must be the name of a property in this class, or a
514 KeyError is raised. That property must be a Link or Multilink
515 property, or a TypeError is raised.
517 Any node in this class whose 'propname' property links to any of the
518 nodeids will be returned. Used by the full text indexing, which knows
519 that "foo" occurs in msg1, msg3 and file7, so we have hits on these
520 issues:
522 db.issue.find(messages={'1':1,'3':1}, files={'7':1})
523 """
524 raise NotImplementedError
526 def filter(self, search_matches, filterspec, sort=(None,None),
527 group=(None,None)):
528 ''' Return a list of the ids of the active nodes in this class that
529 match the 'filter' spec, sorted by the group spec and then the
530 sort spec.
532 "filterspec" is {propname: value(s)}
533 "sort" and "group" are (dir, prop) where dir is '+', '-' or None
534 and prop is a prop name or None
535 "search_matches" is {nodeid: marker}
537 The filter must match all properties specificed - but if the
538 property value to match is a list, any one of the values in the
539 list may match for that property to match.
540 '''
541 raise NotImplementedError
543 def count(self):
544 """Get the number of nodes in this class.
546 If the returned integer is 'numnodes', the ids of all the nodes
547 in this class run from 1 to numnodes, and numnodes+1 will be the
548 id of the next node to be created in this class.
549 """
550 raise NotImplementedError
552 # Manipulating properties:
553 def getprops(self, protected=1):
554 """Return a dictionary mapping property names to property objects.
555 If the "protected" flag is true, we include protected properties -
556 those which may not be modified.
557 """
558 raise NotImplementedError
560 def addprop(self, **properties):
561 """Add properties to this class.
563 The keyword arguments in 'properties' must map names to property
564 objects, or a TypeError is raised. None of the keys in 'properties'
565 may collide with the names of existing properties, or a ValueError
566 is raised before any properties have been added.
567 """
568 raise NotImplementedError
570 def index(self, nodeid):
571 '''Add (or refresh) the node to search indexes
572 '''
573 raise NotImplementedError
575 def safeget(self, nodeid, propname, default=None):
576 """Safely get the value of a property on an existing node of this class.
578 Return 'default' if the node doesn't exist.
579 """
580 try:
581 return self.get(nodeid, propname)
582 except IndexError:
583 return default
585 class HyperdbValueError(ValueError):
586 ''' Error converting a raw value into a Hyperdb value '''
587 pass
589 def convertLinkValue(db, propname, prop, value, idre=re.compile('\d+')):
590 ''' Convert the link value (may be id or key value) to an id value. '''
591 linkcl = db.classes[prop.classname]
592 if not idre.match(value):
593 if linkcl.getkey():
594 try:
595 value = linkcl.lookup(value)
596 except KeyError, message:
597 raise HyperdbValueError, 'property %s: %r is not a %s.'%(
598 propname, value, prop.classname)
599 else:
600 raise HyperdbValueError, 'you may only enter ID values '\
601 'for property %s'%propname
602 return value
604 def fixNewlines(text):
605 """ Homogenise line endings.
607 Different web clients send different line ending values, but
608 other systems (eg. email) don't necessarily handle those line
609 endings. Our solution is to convert all line endings to LF.
610 """
611 text = text.replace('\r\n', '\n')
612 return text.replace('\r', '\n')
614 def rawToHyperdb(db, klass, itemid, propname, value,
615 pwre=re.compile(r'{(\w+)}(.+)')):
616 ''' Convert the raw (user-input) value to a hyperdb-storable value. The
617 value is for the "propname" property on itemid (may be None for a
618 new item) of "klass" in "db".
620 The value is usually a string, but in the case of multilink inputs
621 it may be either a list of strings or a string with comma-separated
622 values.
623 '''
624 properties = klass.getprops()
626 # ensure it's a valid property name
627 propname = propname.strip()
628 try:
629 proptype = properties[propname]
630 except KeyError:
631 raise HyperdbValueError, '%r is not a property of %s'%(propname,
632 klass.classname)
634 # if we got a string, strip it now
635 if isinstance(value, type('')):
636 value = value.strip()
638 # convert the input value to a real property value
639 if isinstance(proptype, String):
640 # fix the CRLF/CR -> LF stuff
641 value = fixNewlines(value)
642 if isinstance(proptype, Password):
643 m = pwre.match(value)
644 if m:
645 # password is being given to us encrypted
646 p = password.Password()
647 p.scheme = m.group(1)
648 if p.scheme not in 'SHA crypt plaintext'.split():
649 raise HyperdbValueError, 'property %s: unknown encryption '\
650 'scheme %r'%(propname, p.scheme)
651 p.password = m.group(2)
652 value = p
653 else:
654 try:
655 value = password.Password(value)
656 except password.PasswordValueError, message:
657 raise HyperdbValueError, 'property %s: %s'%(propname, message)
658 elif isinstance(proptype, Date):
659 try:
660 tz = db.getUserTimezone()
661 value = date.Date(value).local(tz)
662 except ValueError, message:
663 raise HyperdbValueError, 'property %s: %r is an invalid '\
664 'date (%s)'%(propname, value, message)
665 elif isinstance(proptype, Interval):
666 try:
667 value = date.Interval(value)
668 except ValueError, message:
669 raise HyperdbValueError, 'property %s: %r is an invalid '\
670 'date interval (%s)'%(propname, value, message)
671 elif isinstance(proptype, Link):
672 if value == '-1' or not value:
673 value = None
674 else:
675 value = convertLinkValue(db, propname, proptype, value)
677 elif isinstance(proptype, Multilink):
678 # get the current item value if it's not a new item
679 if itemid and not itemid.startswith('-'):
680 curvalue = klass.get(itemid, propname)
681 else:
682 curvalue = []
684 # if the value is a comma-separated string then split it now
685 if isinstance(value, type('')):
686 value = value.split(',')
688 # handle each add/remove in turn
689 # keep an extra list for all items that are
690 # definitely in the new list (in case of e.g.
691 # <propname>=A,+B, which should replace the old
692 # list with A,B)
693 set = 1
694 newvalue = []
695 for item in value:
696 item = item.strip()
698 # skip blanks
699 if not item: continue
701 # handle +/-
702 remove = 0
703 if item.startswith('-'):
704 remove = 1
705 item = item[1:]
706 set = 0
707 elif item.startswith('+'):
708 item = item[1:]
709 set = 0
711 # look up the value
712 itemid = convertLinkValue(db, propname, proptype, item)
714 # perform the add/remove
715 if remove:
716 try:
717 curvalue.remove(itemid)
718 except ValueError:
719 raise HyperdbValueError, 'property %s: %r is not ' \
720 'currently an element'%(propname, item)
721 else:
722 newvalue.append(itemid)
723 if itemid not in curvalue:
724 curvalue.append(itemid)
726 # that's it, set the new Multilink property value,
727 # or overwrite it completely
728 if set:
729 value = newvalue
730 else:
731 value = curvalue
733 # TODO: one day, we'll switch to numeric ids and this will be
734 # unnecessary :(
735 value = [int(x) for x in value]
736 value.sort()
737 value = [str(x) for x in value]
738 elif isinstance(proptype, Boolean):
739 value = value.strip()
740 value = value.lower() in ('yes', 'true', 'on', '1')
741 elif isinstance(proptype, Number):
742 value = value.strip()
743 try:
744 value = float(value)
745 except ValueError:
746 raise HyperdbValueError, 'property %s: %r is not a number'%(
747 propname, value)
748 return value
750 class FileClass:
751 ''' A class that requires the "content" property and stores it on
752 disk.
753 '''
754 pass
756 class Node:
757 ''' A convenience wrapper for the given node
758 '''
759 def __init__(self, cl, nodeid, cache=1):
760 self.__dict__['cl'] = cl
761 self.__dict__['nodeid'] = nodeid
762 def keys(self, protected=1):
763 return self.cl.getprops(protected=protected).keys()
764 def values(self, protected=1):
765 l = []
766 for name in self.cl.getprops(protected=protected).keys():
767 l.append(self.cl.get(self.nodeid, name))
768 return l
769 def items(self, protected=1):
770 l = []
771 for name in self.cl.getprops(protected=protected).keys():
772 l.append((name, self.cl.get(self.nodeid, name)))
773 return l
774 def has_key(self, name):
775 return self.cl.getprops().has_key(name)
776 def get(self, name, default=None):
777 if self.has_key(name):
778 return self[name]
779 else:
780 return default
781 def __getattr__(self, name):
782 if self.__dict__.has_key(name):
783 return self.__dict__[name]
784 try:
785 return self.cl.get(self.nodeid, name)
786 except KeyError, value:
787 # we trap this but re-raise it as AttributeError - all other
788 # exceptions should pass through untrapped
789 pass
790 # nope, no such attribute
791 raise AttributeError, str(value)
792 def __getitem__(self, name):
793 return self.cl.get(self.nodeid, name)
794 def __setattr__(self, name, value):
795 try:
796 return self.cl.set(self.nodeid, **{name: value})
797 except KeyError, value:
798 raise AttributeError, str(value)
799 def __setitem__(self, name, value):
800 self.cl.set(self.nodeid, **{name: value})
801 def history(self):
802 return self.cl.history(self.nodeid)
803 def retire(self):
804 return self.cl.retire(self.nodeid)
807 def Choice(name, db, *options):
808 '''Quick helper to create a simple class with choices
809 '''
810 cl = Class(db, name, name=String(), order=String())
811 for i in range(len(options)):
812 cl.create(name=options[i], order=i)
813 return hyperdb.Link(name)
815 # vim: set filetype=python ts=4 sw=4 et si