This file is indexed.

/usr/bin/delta is in swap-cwm 1.2.1-5.

This file is owned by root:root, with mode 0o755.

The actual contents of the file can be viewed below.

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
#!/usr/bin/python
"""
Find differences between two RDF graphs, using
functional and inverse functional properties to identify bnodes
in the patch file.

--from=uri      -f uri     from-file
--to=uri        -t uri     file against which to check for differences
--meta=uri      -m uri     file with metadata to be assumed (in addition to schemas)
--help          -h         print this help message
--verbose=      -v 1       verbose mode (can be 0 to 10, 0 is default)
--granularity=  -g 0       g=0 - lots of little diffs.
                           g=1, fewer diffs (default)

If from-file but not to-file is given, from-file is smushed and output
Uris are relative to present working directory.

For motivation and explanation, see  <http://www.w3.org/DesignIssues/Diff>

$Id: delta.py,v 1.11 2007/06/26 02:36:15 syosi Exp $
http://www.w3.org/2000/10/swap/diff.py
"""



import string, getopt
try:
    Set = set               # Python2.4 and on
except NameError:
    from sets import Set    # Python2.3 and on
import string
import sys




# http://www.w3.org/2000/10/swap/
try:
    from swap import llyn, diag
    from swap.myStore import loadMany
    from swap.diag import verbosity, setVerbosity, progress
    from swap import notation3          # N3 parsers and generators


    from swap.RDFSink import FORMULA, LITERAL, ANONYMOUS, Logic_NS
    from swap import uripath
    from swap.uripath import base
    from swap.myStore import  Namespace
    from swap import myStore
    from swap.notation3 import RDF_NS_URI
    from swap.llyn import Formula, CONTEXT, PRED, SUBJ, OBJ
except ImportError:
    import llyn, diag
    from myStore import loadMany
    from diag import verbosity, setVerbosity, progress
    import notation3            # N3 parsers and generators


    from RDFSink import FORMULA, LITERAL, ANONYMOUS, Logic_NS
    import uripath
    from uripath import base
    from myStore import  Namespace
    import myStore
    from notation3 import RDF_NS_URI
    from llyn import Formula, CONTEXT, PRED, SUBJ, OBJ

#daml = Namespace("http://www.daml.org/2001/03/daml+oil#")
OWL = Namespace("http://www.w3.org/2002/07/owl#")
RDF = Namespace("http://www.w3.org/1999/02/22-rdf-syntax-ns#")
LOG = Namespace("http://www.w3.org/2000/10/swap/log#")
DELTA = Namespace("http://www.w3.org/2004/delta#")

def debugBnode(n, f):
    progress("For node %s" % `n`)
    for s in f.statementsMatching(subj=n):
        progress("     %s  %s; # could be ifp?" %(`s.predicate()`, `s.object()`)) 
    for s in f.statementsMatching(obj=n):
        progress("     is %s of  %s; # could be fp?" %(`s.predicate()`, `s.subject()`)) 

def lookUp(predicates, assumptions=Set()):
    """Look up all the schemas for the predicates given"""
    global verbose
    schemas = assumptions
    for pred in predicates:
        if verbose > 3: progress("Predicate: %s" % `pred`)
        u = pred.uriref()
        hash = u.find("#")
        if hash <0:
            if verbose > 1: progress("Warning: Predicate <%s> looks like web resource not Property" % u)
        else:
            schemas.add(u[:hash])
    if verbose > 2:
        for r in schemas:
            progress("Metadata to be loaded: ", r) 
    if schemas:
        return loadMany([(x) for x in schemas])
    return myStore.store.newFormula() # Empty formula
    
def nailFormula(f, assumptions=Set()):
    """Smush the formula.
    Build a dictionary of nodes which are indirectly identified
    by [inverse] functonal properties."""
    global verbose
    cc, predicates, ss, oo = getParts(f)
    nodes = ss | oo
    sofar = {}
    bnodes = Set()
    for node in nodes:
        if node.generated() or node in f.existentials():
            bnodes.add(node)
            if verbose >=5: progress("Blank node: %s" % `node`)
        else:
            if verbose >=5: progress("Fixed node: %s" % `node`)
        sofar[node] = []

    meta = lookUp(predicates, assumptions)
    ifps = predicates & Set(meta.each(pred=RDF.type, obj=OWL.InverseFunctionalProperty))
    fps = predicates & Set(meta.each(pred=RDF.type, obj=OWL.FunctionalProperty))
    if verbose > 1:
        for p in fps:  progress("Functional Property:", p)
        for p in ifps: progress("Inverse Functional: ", p)
    
    definitions = []
    if len(bnodes) == 0:
        if verbose > 0: progress("No bnodes in graph")
        return  bnodes, definitions

    a = float(len(bnodes))/len(nodes)
    if verbose > 1: progress("Proportion of bodes which are blank: %f" % a)
#    if a == 0: return bnodes, definitions

    loose = bnodes.copy()
    equivs = Set()
    #  Note possible optmization: First pass only like this, 
    # future passes work from newNodes.
    while loose:
        newNailed = Set()
        for preds, inverse, char in ((fps, 0, "!"), (ifps, 1, "^")):
            for pred in preds:
                if verbose > 3: progress("Predicate", pred)
                ss = f.statementsMatching(pred=pred)
                for s in ss:
                    if inverse: y, x = s.object(), s.subject()
                    else: x, y = s.object(), s.subject()
                    if not x.generated(): continue  # Only anchor bnodes
                    if y not in loose:  # y is the possible anchor
                        defi = (x, inverse, pred, y)
                        if x in loose:   # this node
                            if verbose > 4: progress("   Nailed %s as %s%s%s" % (`x`, `y`, `char`, `pred`))
                            loose.discard(x)
                            newNailed.add(x)
                        else:
                            if verbose >=6 : progress(
                                "   (ignored %s as %s%s%s)" % (`x`, `y`, `char`, `pred`))
                        definitions.append(defi)
#                       if verbose > 9: progress("   Definition[x] is now", definition[x])
                        if inverse: equivalentSet = Set(f.each(obj=y, pred=pred))
                        else: equivalentSet = Set(f.each(subj=y, pred=pred))
                        if len(equivalentSet) > 1: equivs.add(equivalentSet)

        if not newNailed:
            if verbose > 1:
                progress("Failed to nail nodes:", loose)
                if verbose > 3:
                    for n in loose:
                        debugBnode(n, f)
            break

# At this point if we still have loose nodes, we have failed with ifps and fps.
# Diff may not be strong. (It might still be: the diffs might not involve weak definitions)

    weak = loose.copy()  # Remember
    if verbose > 0: progress("\nFailed to directly nail everything, looking for weak nailings.")
#    sys.exit(-1)  #@@@
    while loose:
        newNailed = Set()
        if verbose>2:
            progress()
            progress("Pass: loose = %s" % loose)
        for x in loose.copy():
            if verbose>3: progress("Checking weakly node %s" % x)
            for s in f.statementsMatching(obj=x):
                pred, y = s.predicate(), s.subject()
                if y in loose:
                    if verbose > 4: progress("Can't nail to loose %s" % y)
                    continue    # Can't nail to something loose
                others = f.each(subj=y, pred=pred)
                # @@ Should ignore redirected equivalent nodes in others
                if len(others) != 1:
                    if verbose>4: progress("Can't nail: %s all are %s of %s." % (others, pred, y))
                    continue  # Defn would be ambiguous in this graph
                defi = (x, 0, pred, y)
                if verbose >4: progress("   Weakly-nailed %s as %s%s%s" % (x, y, "!", pred))
                loose.discard(x)
                newNailed.add(x)
                definitions.append(defi)
                break # progress
            else:
                for s in f.statementsMatching(subj=x):
                    pred, obj = s.predicate(), s.object()
                    if obj in loose:
                        if verbose >4: progress("Can't nail to loose %s" % obj)
                        continue    # Can't nail to something loose
                    others = f.each(obj=obj, pred=pred)
                    # @@ Should ignore redirected equivalent nodes in others
                    if len(others) != 1:
                        if verbose >4: progress(
                        "Can't nail: %s all have %s of %s." % (others, pred, obj))
                        continue  # Defn would be ambiguous in this graph
                    defi = (x, 1, pred, obj)
                    if verbose>2: progress("   Weakly-nailed %s as %s%s%s" % (`x`, `obj`, "^", `pred`))
                    loose.discard(x)
                    newNailed.add(x)
                    definitions.append(defi)
                    break # progress

        if not newNailed:
            if verbose>0:
                progress("Failed to even weakly nail nodes:", loose)
                for n in loose:
                    progress("For node %s" % n)
                    for s in f.statementsMatching(subj=n):
                        progress("     %s  %s; # could be ifp?" %(`s.predicate()`, `s.object()`)) 
                    for s in f.statementsMatching(obj=n):
                        progress("     is %s of  %s; # could be fp?" %(s.predicate(), s.subject())) 
                

            raise ValueError("Graph insufficiently labelled for nodes: %s" % loose)

    if verbose>0 and not weak: progress("Graph is solid.")
    if verbose>0 and weak: progress("Graph is NOT solid.")
    
    f.reopen()
    for es in equivs:
        if verbose>3: progress("Equivalent: ", es)
        prev = None
        for x in es:
            if prev:
                f.add(x, OWL.sameAs, prev)
            prev = x
    return bnodes, definitions

def removeCommon(f, g, match):
    """Find common statements from f and g
    match gives the dictionary mapping bnodes in f to bnodes in g"""
    only_f, common_g = Set(), Set()
    for st in f.statements[:]:
        s, p, o = st.spo()
        assert s not in f._redirections 
        assert o not in f._redirections
        if s.generated(): sg = match.get(s, None)
        else: sg = s
        if o.generated(): og = match.get(o, None)
        else: og = o
        if og != None and sg != None:
            gsts = g.statementsMatching(subj=sg, pred=p, obj=og)
            if len(gsts) == 1:
                if verbose>4: progress("Statement in both", st)
                common_g.add(gsts[0])
                continue
        only_f.add(st)
    return only_f, Set(g.statements)-common_g

def patches(delta, f, only_f, originalBnodes, definitions, deleting=0):
    """Generate patches in patch formula, for the remaining statements in f
    given the bnodes and definitions for f."""
    todo = only_f.copy()
    if deleting:
        patchVerb = DELTA.deletion
    else:
        patchVerb = DELTA.insertion
    if verbose>2: progress("*********** PATCHES: %s, with  %i to do" %(patchVerb, len(todo)))
    while todo:

        # find a contiguous subgraph defined in the given graph
        bnodesToDo = Set()
        bnodes = Set()
        rhs = delta.newFormula()
        lhs = delta.newFormula()  # left hand side of patch
        newStatements = Set()
        for seed in todo: break # pick one #@2 fn?
        statementsToDo = Set([seed])
        if verbose>3: progress("Seed:", seed)
        subgraph = statementsToDo
        while statementsToDo or bnodesToDo:
            for st in statementsToDo:
                s, p, o = st.spo()
                for x in s, p, o:
                    if x.generated() and x not in bnodes: # and x not in commonBnodes:
                        if verbose>4: progress("   Bnode ", x)
                        bnodesToDo.add(x)
                        bnodes.add(x)
                rhs.add(s, p, o)
            statementsToDo = Set()
            for x in bnodesToDo:
                bnodes.add(x)
                ss = (f.statementsMatching(subj=x)
                    + f.statementsMatching(pred=x)
                    + f.statementsMatching(obj=x))
                for z in ss:
                    if z in only_f:
                        newStatements.add(z)
                if verbose>3: progress("    New statements from %s: %s" % (x, newStatements))
                statementsToDo = statementsToDo | newStatements
                subgraph = subgraph |newStatements
            bnodesToDo = Set()

        if verbose>3: progress("Subgraph of %i statements (%i left):\n\t%s\n" 
                %(len(subgraph), len(todo), subgraph))
        todo = todo - subgraph
        
        
        undefined = bnodes.copy()
        for x, inverse, pred, y in definitions:
            if x in undefined:
                if inverse: s, p, o = x, pred, y
                else: s, p, o = y, pred, x
                if verbose > 4: progress("Declaring variable %s" % x.uriref())
                if deleting:
                    delta.declareUniversal(x)
                    lhs.add(subj=s, pred=p, obj=o)
                else: # inserting
                    if x in originalBnodes:
                        delta.declareUniversal(x)
                        lhs.add(subj=s, pred=p, obj=o)
                    else:
                        rhs.declareExistential(x)
                if y.generated():
                    undefined.add(y)
                undefined.discard(x)
        if undefined:
            progress("Still haven't defined bnodes %s" % undefined)
            for n in undefined:
                debugBnode(n, f)
            raise RuntimeError("BNodes still undefined", undefined)
        lhs = lhs.close()
        rhs = rhs.close()
        delta.add(subj=lhs, pred=patchVerb, obj=rhs)
        if verbose >1: progress("PATCH: %s %s %s\n" %(lhs.n3String(), `patchVerb`, rhs.n3String()))
    return

def consolidate(delta, patchVerb):
    """Consolidate patches
    
    Where the same left hand side applies to more than 1 RHS formula,
    roll those RHS formulae into one, to make the dif file more readable
    and faster to execute in some implementations
    """
    agenda = {}
    if verbose >3: progress("Consolidating %s" % patchVerb)
    for s in delta.statementsMatching(pred=patchVerb):
        list = agenda.get(s.subject(), None)
        if list == None:
            list = []
            agenda[s.subject()] = list
        list.append(s)
    for lhs, list in agenda.items():
        if verbose >3: progress("Patches lhs= %s: %s" %(lhs, list))
        if len(list) > 1:
            rhs = delta.newFormula()
            for s in list:
                delta.store.copyFormula(s.object(), rhs)
                delta.removeStatement(s)
            delta.add(subj=lhs, pred=patchVerb, obj=rhs.close())

    
def differences(f, g, assumptions):
    """Smush the formulae.  Compare them, generating patch instructions."""
    global lumped
    
# Cross-map nodes:

    g_bnodes, g_definitions = nailFormula(g, assumptions)
    bnodes, definitions = nailFormula(f, assumptions)
    if verbose > 1: progress("\n Done nailing")
    definitions.reverse()  # go back down list @@@ reverse the g list too? @@@
    g_definitions.reverse()     # @@ needed for the patch generation
    
    unmatched = bnodes.copy()
    match = {}  # Mapping of nodes in f to nodes in g
    for x, inverse, pred, y in definitions:
        if x in match: continue # done already

        if x in f._redirections:
            if verbose > 3: progress("Redirected %s to %s. Ignoring" % (`x`, `f._redirections[x]`))
            unmatched.discard(x)
            continue

        if verbose > 3: progress("Definition of %s = %s%s%s"% (`x`, `y` , ".!^"[inverse], `pred`))

        if y.generated():
            while y in f._redirections:
                y = f._redirections[y]
                if verbose>4: progress(" redirected to  %s = %s%s%s"% (`x`,  `y`, "!^"[inverse], `pred`))
            yg = match.get(y, None)
            if yg == None:
                if verbose>4: progress("  Had definition for %s in terms of %s which is not matched"%(`x`,`y`))
                continue
        else:
            yg = y

        if inverse:  # Inverse functional property like ssn
            matches = Set(g.each(obj=yg, pred=pred))
        else: matches = Set(g.each(subj=yg, pred=pred))
        if len(matches) == 0:
            continue   # This is normal - the node does not exist in the other graph
#           raise RuntimeError("Can't match %s" % x)

        if len(matches) > 1:
            raise RuntimeError("""Rats. Wheras in the first graph %s%s%s uniquely selects %s,
                    in the other graph there are more than 1 matches: %s""" % (`y`, "!^"[inverse], `pred`, `x`,  `matches`))
        for q in matches:  # pick only one  @@ python function?
            z = q
            break
        if verbose > 2:
            progress("Found match for %s in %s " % (`x`,`z`))
        match[x] = z
        unmatched.discard(x)

    if len(unmatched) > 0:
        if verbose >1:
            progress("Failed to match all nodes:", unmatched)
            for n in unmatched:
                debugBnode(n, f)

    # Find common parts
    only_f, only_g = removeCommon(f,g, match)

    delta = f.newFormula()
    if len(only_f) == 0 and len(only_g) == 0:
        return delta

    f = f.close()    #  We are not going to mess with them any more
    g = g.close()
    
    common = Set([match[x] for x in match])

    if verbose>2: progress("Common bnodes (as named in g)", common)
    patches(delta, f, only_f, Set(), definitions, deleting=1)
    patches(delta, g, only_g, common, g_definitions, deleting=0)
    if lumped:
        consolidate(delta, delta.store.insertion)
        consolidate(delta, delta.store.deletion)
    return delta

        
     
def getParts(f, meta=None):
    """Make lists of all node IDs and arc types
    """
    values = [Set([]),Set([]),Set([]),Set([])]
    for s in f.statements:
        for p in SUBJ, PRED, OBJ:
            x = s[p]
            values[p].add(x)
    return values


def loadFiles(files):
    graph = myStore.formula()
    graph.setClosureMode("e")    # Implement sameAs by smushing
    if verbose>0: progress("Loading %s..." % files)
    graph = myStore.loadMany(files, openFormula=graph)
    if verbose>0: progress("Loaded", graph)
    return graph

def usage():
    sys.stderr.write(__doc__)
    
def main():
    testFiles = []
    diffFiles = []
    assumptions = Set()
    global ploughOn # even if error
    ploughOn = 0
    global verbose
    global lumped
    verbose = 0
    lumped = 1
    try:
        opts, args = getopt.getopt(sys.argv[1:], "hf:t:m:v:g:",
            ["help", "from=", "to=", "meta=", "verbose=", "granularity="])
    except getopt.GetoptError:
        # print help information and exit:
        usage()
        sys.exit(2)
    output = None
    for o, a in opts:
        if o in ("-h", "--help"):
            usage()
            sys.exit(2)
        if o in ("-v", "--verbose"):
            try: verbose = int(a)
            except ValueError: verbose = 10
        if o in ("-g", "--granularity"):
            try: lumped = int(a)
            except ValueError: lumped = 0
        if o in ("-f", "--from"):
            testFiles.append(a)
        if o in ("-t", "--to"):
            diffFiles.append(a)
        if o in ("-m", "--meta"):
            assumptions.add(a)

    

#    if testFiles == []: testFiles = [ "/dev/stdin" ]
    if testFiles == []:
        usage()
        sys.exit(2)
    graph = loadFiles(testFiles)
    version = "$Id: delta.py,v 1.11 2007/06/26 02:36:15 syosi Exp $"[1:-1]
    if diffFiles == []:
        nailFormula(graph, assumptions)
        if verbose > 1: print "# Smush generated by " + version
        print graph.close().n3String(base=base(), flags="a")
        sys.exit(0)
        
    graph2 = loadFiles(diffFiles)
    delta = differences(graph, graph2, assumptions)
    if verbose >1: print "# Differences by " + version
    print delta.close().n3String(base=base())
#    sys.exit(len(delta))
#    sys.exit(0)   # didn't crash
    if delta.contains():        # Any statements in delta at all?
        sys.exit(1)
    else:
        sys.exit(0)
    
        
                
if __name__ == "__main__":
    main()