Eskil

Check-in [da2d0bc4f5]
Login

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

Overview
Comment:Split CompareBlocks in two to make it more testable.
Downloads: Tarball | ZIP archive
Timelines: family | ancestors | descendants | both | trunk
Files: files | file ages | folders
SHA1: da2d0bc4f51f52691e111624b55cb9d9f3afde51
User & Date: peter 2005-09-28 21:04:02.000
Context
2005-09-28
21:07
Detect kits and automatically mount them. check-in: ab853f4298 user: peter tags: trunk
21:04
Split CompareBlocks in two to make it more testable. check-in: da2d0bc4f5 user: peter tags: trunk
2005-07-21
19:49
Made menus into real menubar. check-in: 62b5902400 user: peter tags: trunk
Changes
Unified Diff Ignore Whitespace Patch
Changes to src/compare.tcl.
17
18
19
20
21
22
23




24
25
26
27
28
29
30
31
32
33
34
35
#  along with this program; see the file COPYING.  If not, write to
#  the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
#  Boston, MA 02111-1307, USA.
#
#----------------------------------------------------------------------
# $Revision$
#----------------------------------------------------------------------





# Compare two lines and rate how much they resemble each other.
# This has never worked well. Some day I'll sit down, think this through,
# and come up with a better algorithm.
proc compareLines2 {line1 line2} {
    set opts $::Pref(ignore)
    if {$::Pref(nocase)} {lappend opts -nocase}
    set res [eval DiffUtil::diffStrings $opts \$line1 \$line2]

    # Collect identical pieces and different pieces
    set sames {}
    set diffs1 {}







>
>
>
>




|







17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
#  along with this program; see the file COPYING.  If not, write to
#  the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
#  Boston, MA 02111-1307, USA.
#
#----------------------------------------------------------------------
# $Revision$
#----------------------------------------------------------------------

proc maxAbs {a b} {
    return [expr {abs($a) > abs($b) ? $a : $b}]
}

# Compare two lines and rate how much they resemble each other.
# This has never worked well. Some day I'll sit down, think this through,
# and come up with a better algorithm.
proc CompareLines {line1 line2} {
    set opts $::Pref(ignore)
    if {$::Pref(nocase)} {lappend opts -nocase}
    set res [eval DiffUtil::diffStrings $opts \$line1 \$line2]

    # Collect identical pieces and different pieces
    set sames {}
    set diffs1 {}
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
    foreach diff $diffs2 {
        set apa [string length $diff]
        incr sumdiff2 $apa
    }
#    puts "Same ($sames)"
#    puts "D1   ($diffs1)"
#    puts "D2   ($diffs2)"
#    puts "S $sumsame D $sumdiff1 D $sumdiff2"
    return [expr {$sumsame - [maxAbs $sumdiff1 $sumdiff2]}]
}

# Initialise a multidimensional list with some value
# This should use lrepeat once 8.5 is required
# The args are in the same order as indexes to lset/lindex
proc Linit {elem args} {
    for {set t [expr {[llength $args] - 1}]} {$t >= 0} {incr t -1} {
	set new {}
	for {set j [lindex $args $t]} {$j >= 1} {incr j -1} {
	    lappend new $elem
	}
	set elem $new
    }
    return $elem
}

# Decide how to display change blocks
# This tries to match the lines that resemble each other and put them
# next to each other.
# As compareLines2, this would need a complete rework and a
# better algorithm.

proc compareBlocks {block1 block2} {
    set size1 [llength $block1]
    set size2 [llength $block2]

    # Swap if block1 is bigger
    if {$size1 > $size2} {
        set apa $block1
        set block1 $block2
        set block2 $apa

        set size1 [llength $block1]
        set size2 [llength $block2]
        set dsym a
        set asym d
    } else {
        set dsym d
        set asym a
    }



    # Collect statistics
    set scores [Linit {} $size1 $size2]
    set emptyResult [Linit {} $size1]

    set scoresbest $emptyResult
    set origresult $emptyResult

    set j 0
    set bestsum 0
    foreach line1 $block1 {
        set bestscore -100000
        set bestline 0
        set i 0
        foreach line2 $block2 {
            set x [compareLines2 $line1 $line2]
            lset scores $j $i $x
            #puts "Score $j $i : $x"
            if {$x > $bestscore} {
                set bestscore $x
                set bestline $i
            }
            incr i







|




















|

>
|
<
<
|
|
<
|
|
|
>
|
|
<
<
<
<
<
|
>
>

|

<
>










|







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
    foreach diff $diffs2 {
        set apa [string length $diff]
        incr sumdiff2 $apa
    }
#    puts "Same ($sames)"
#    puts "D1   ($diffs1)"
#    puts "D2   ($diffs2)"
#    puts "S $sumsame  D $sumdiff1 D $sumdiff2"
    return [expr {$sumsame - [maxAbs $sumdiff1 $sumdiff2]}]
}

# Initialise a multidimensional list with some value
# This should use lrepeat once 8.5 is required
# The args are in the same order as indexes to lset/lindex
proc Linit {elem args} {
    for {set t [expr {[llength $args] - 1}]} {$t >= 0} {incr t -1} {
	set new {}
	for {set j [lindex $args $t]} {$j >= 1} {incr j -1} {
	    lappend new $elem
	}
	set elem $new
    }
    return $elem
}

# Decide how to display change blocks
# This tries to match the lines that resemble each other and put them
# next to each other.
# As CompareLines, this would need a complete rework and a
# better algorithm.
#
# Constraint: block1 may not be longer than block2


#
# Result is a list with one element per row in block1.

# The element is the index of the matching row in block2, and could be
# out of range.
proc CompareBlocks2 {block1 block2 scoresName} {
    upvar 1 $scoresName scores
    set size1 [llength $block1]
    set size2 [llength $block2]






    # A "constant", so I don't need to create it more than once
    set emptyResult [Linit {} $size1]

    # Collect statistics about each pair of lines.
    set scores [Linit {} $size1 $size2]

    # Store the best match for each item
    set scoresbest $emptyResult
    set origresult $emptyResult

    set j 0
    set bestsum 0
    foreach line1 $block1 {
        set bestscore -100000
        set bestline 0
        set i 0
        foreach line2 $block2 {
            set x [CompareLines $line1 $line2]
            lset scores $j $i $x
            #puts "Score $j $i : $x"
            if {$x > $bestscore} {
                set bestscore $x
                set bestline $i
            }
            incr i
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
    # result since it has to be in order.

    #puts "Origresult: $origresult"

    # If the size is 1, it is automatically in order so we
    # don't need further processing.

    if {$size1 > 1} {



        # Start with a check if the theoretical best works, since often that
        # is the case.
        set order 1
        set result $origresult
        for {set i 0} {$i < ($size1 - 1)} {incr i} {
            if {[lindex $result $i] >= [lindex $result [expr {$i + 1}]]} {
                set order 0
                break
            }
        }
        #if {$order} {puts "ORDER"}


    }

    set bestresult $origresult
    set bestscoresum -100000

    if {$size1 > 1 && $order == 0} {
        # Look through the obvious "subblock" alternatives

        for {set startj 0} {$startj < ($size2 - $size1 + 1)} {incr startj} {
	    set sum 0
	    set result $emptyResult
	    for {set i 0 ; set j $startj} {$i < $size1} {incr i ; incr j} {
		lset result $i $j
		incr sum [lindex $scores $i $j]
	    }
	    #puts "Subblock $startj sum: $sum"
            if {$sum > $bestscoresum} {
                #puts "New best: $sum ($bestscoresum)"
                set bestresult $result
                set bestscoresum $sum
            }
	}

        # If we reach 75% if the theoretical best, we take it
        while {$bestscoresum < (3 * $bestsum / 4)} {
            #puts "Outer: $scoresbest"
	    # The outer loop restarts from the "best mapping"
	    set result $origresult
            set mark [Linit 0 $size1]
            set high $mark

            # If result is in order, no problem.
            # Otherwise, try to adjust result to make it ordered

            while {1} {
                #puts "Inner: $scoresbest"
		# The inner loop tries to get the result in order
                set besti 0
                set bestscore -100000
                set order 1
                for {set i 0} {$i < $size1} {incr i} {
                    if {[lindex $mark $i] == 0} {
                        for {set j [expr {$i + 1}]} {$j < $size1} {incr j} {
                            if {[lindex $mark $j] == 0} break
                        }
                        if {$j < $size1 && \
                                [lindex $result $i] >= [lindex $result $j]} {
                            set order 0
                        }
                        set x [lindex $scoresbest $i]
                        if {$x > $bestscore} {
                            set bestscore $x
                            set besti $i
                        }
                    }
                }
                #puts "Best $besti order $order sc $bestscore"
                if {$order} break
                lset mark $besti 1
                set bestr [lindex $result $besti]
                for {set i 0} {$i < $besti} {incr i} {
                    if {[lindex $mark $i] == 0 && \
                            [lindex $result $i] >= $bestr} {
                        lset mark $i 2
                    }
                }
                for {set i [expr {$besti + 1}]} {$i < $size1} {incr i} {
                    if {[lindex $mark $i] == 0 && \
                            [lindex $result $i] <= $bestr} {
                        lset mark $i 2
                    }
                }
            }

            set prev $size2
            for {set i [expr {$size1 - 1}]} {$i >= 0} {incr i -1} {
                if {[lindex $mark $i] != 2} {
                    set prev [lindex $result $i]
                } else {
                    lset high $i [expr {$prev - 1}]
                }
            }
            set prev -1
            for {set i 0} {$i < $size1} {incr i} {
                if {[lindex $mark $i] != 2} {
                    set prev [lindex $result $i]
                } else {
                    if {[lindex $high $i] > $prev} {
                        incr prev
                        lset result $i $prev
                    } else {
                        lset result $i -1
                    }
                }
            }
            set scoresum 0
            for {set i 0} {$i < $size1} {incr i} {
                set j [lindex $result $i]
                set sc [lindex $scores $i $j] ;# FIXA: can this fail?
                if {[string is integer -strict $sc]} {
                    #puts "Score: $i $j [lindex $scores $i $j]"
                    incr scoresum $sc
                }
            }
            #puts "Scoresum: $scoresum ($bestscoresum)"

	    # If it was not an improvement over previous iteration, quit
            if {$scoresum <= $bestscoresum} {
                break
	    }

	    set bestresult $result
	    set bestscoresum $scoresum

	    # We are redoing from start, but try to improve by
	    # ignoring the most awkwardly placed line.
	    set mostp -1
	    set mosti 0
	    for {set i 0} {$i < $size1} {incr i} {
		if {[lindex $mark $i] == 1} {
		    if {abs([lindex $result $i] - $i) > $mostp} {
			set mostp [expr {abs([lindex $result $i] - $i)}]
                        set mosti $i
		    }
		}
	    }
	    #puts "Most $mosti $mostp"
	    lset scoresbest $mosti 0
        }
    }


    set result $bestresult


























    # Collect the result into diff-like codes to use as display info.

    set apa {}
    set t1 0
    set t2 0
    while {$t1 < $size1 || $t2 < $size2} {
        if {$t1 < $size1} {
            set r [lindex $result $t1]
            if {$r < $t2 || $t2 >= $size2} {

                lappend apa $dsym
                incr t1
            } elseif {$r == $t2} {
                #if {[string match Wm* [lindex $block2 $t2]]} {
                #    puts "Left : [lindex $block1 $t1]"
                #    puts "Right: [lindex $block2 $t2]"
                #    puts "Score: $scores($t1,$t2)"
                #}

                # If the score is too bad, don't do line parsing.
                if {[lindex $scores $t1 $t2] < 0} {
                    lappend apa "C"
                } else {
                    lappend apa "c"
                }
                incr t1
                incr t2
            } else {

                lappend apa $asym
                incr t2
            }
        } else {

            lappend apa $asym
            incr t2
        }
    }
    return $apa
}








|
>
>
>
|
|
|
|
|
|
|
|
|
|
|
>
>





<
|

|
|
|
|
|
|
|
|
|
|
|
|
|
|

|
|
|
|
|
|
|

|
|

|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|

|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|

|
|

|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
>
|
|
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>










>


















>




>







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
    # result since it has to be in order.

    #puts "Origresult: $origresult"

    # If the size is 1, it is automatically in order so we
    # don't need further processing.

    if {$size1 == 1} {
        return $origresult
    }

    # Start with a check if the theoretical best works, since often that
    # is the case.
    set order 1
    set result $origresult
    for {set i 0} {$i < ($size1 - 1)} {incr i} {
        if {[lindex $result $i] >= [lindex $result [expr {$i + 1}]]} {
            set order 0
            break
        }
    }
    if {$order} {
        #puts "ORDER"
        return $origresult
    }

    set bestresult $origresult
    set bestscoresum -100000


    # Look through the obvious "subblock" alternatives

    for {set startj 0} {$startj < ($size2 - $size1 + 1)} {incr startj} {
        set sum 0
        set result $emptyResult
        for {set i 0 ; set j $startj} {$i < $size1} {incr i ; incr j} {
            lset result $i $j
            incr sum [lindex $scores $i $j]
        }
        #puts "Subblock $startj sum: $sum"
        if {$sum > $bestscoresum} {
            #puts "New best: $sum ($bestscoresum)"
            set bestresult $result
            set bestscoresum $sum
        }
    }

    # If we reach 75% if the theoretical best, we take it
    while {$bestscoresum < (3 * $bestsum / 4)} {
        #puts "Outer: $scoresbest"
        # The outer loop restarts from the "best mapping"
        set result $origresult
        set mark [Linit 0 $size1]
        set high $mark

        # If result is in order, no problem.
        # Otherwise, try to adjust result to make it ordered

        while {1} {
            #puts "Inner: $scoresbest"
            # The inner loop tries to get the result in order
            set besti 0
            set bestscore -100000
            set order 1
            for {set i 0} {$i < $size1} {incr i} {
                if {[lindex $mark $i] == 0} {
                    for {set j [expr {$i + 1}]} {$j < $size1} {incr j} {
                        if {[lindex $mark $j] == 0} break
                    }
                    if {$j < $size1 && \
                            [lindex $result $i] >= [lindex $result $j]} {
                        set order 0
                    }
                    set x [lindex $scoresbest $i]
                    if {$x > $bestscore} {
                        set bestscore $x
                        set besti $i
                    }
                }
            }
            #puts "Best $besti order $order sc $bestscore"
            if {$order} break
            lset mark $besti 1
            set bestr [lindex $result $besti]
            for {set i 0} {$i < $besti} {incr i} {
                if {[lindex $mark $i] == 0 && \
                        [lindex $result $i] >= $bestr} {
                    lset mark $i 2
                }
            }
            for {set i [expr {$besti + 1}]} {$i < $size1} {incr i} {
                if {[lindex $mark $i] == 0 && \
                        [lindex $result $i] <= $bestr} {
                    lset mark $i 2
                }
            }
        }

        set prev $size2
        for {set i [expr {$size1 - 1}]} {$i >= 0} {incr i -1} {
            if {[lindex $mark $i] != 2} {
                set prev [lindex $result $i]
            } else {
                lset high $i [expr {$prev - 1}]
            }
        }
        set prev -1
        for {set i 0} {$i < $size1} {incr i} {
            if {[lindex $mark $i] != 2} {
                set prev [lindex $result $i]
            } else {
                if {[lindex $high $i] > $prev} {
                    incr prev
                    lset result $i $prev
                } else {
                    lset result $i -1
                }
            }
        }
        set scoresum 0
        for {set i 0} {$i < $size1} {incr i} {
            set j [lindex $result $i]
            set sc [lindex $scores $i $j] ;# FIXA: can this fail?
            if {[string is integer -strict $sc]} {
                #puts "Score: $i $j [lindex $scores $i $j]"
                incr scoresum $sc
            }
        }
        #puts "Scoresum: $scoresum ($bestscoresum)"
        
        # If it was not an improvement over previous iteration, quit
        if {$scoresum <= $bestscoresum} {
            break
        }

        set bestresult $result
        set bestscoresum $scoresum

        # We are redoing from start, but try to improve by
        # ignoring the most awkwardly placed line.
        set mostp -1
        set mosti 0
        for {set i 0} {$i < $size1} {incr i} {
            if {[lindex $mark $i] == 1} {
                if {abs([lindex $result $i] - $i) > $mostp} {
                    set mostp [expr {abs([lindex $result $i] - $i)}]
                    set mosti $i
                }
            }
        }
        #puts "Most $mosti $mostp"
        lset scoresbest $mosti 0
    }

    return $bestresult
}

# Decide how to display change blocks
# This tries to match the lines that resemble each other and put them
# next to each other.
# Returns diff-like codes to use as display info.
proc compareBlocks {block1 block2} {
    set size1 [llength $block1]
    set size2 [llength $block2]

    # Things below assume that block1 is not bigger than block2.
    # Swap if block1 is bigger
    if {$size1 > $size2} {
        set apa $block1
        set block1 $block2
        set block2 $apa
        set size1 [llength $block1]
        set size2 [llength $block2]
        # Swap output symbols
        set dsym a
        set asym d
    } else {
        set dsym d
        set asym a
    }

    set result [CompareBlocks2 $block1 $block2 scores]

    # Collect the result into diff-like codes to use as display info.

    set apa {}
    set t1 0
    set t2 0
    while {$t1 < $size1 || $t2 < $size2} {
        if {$t1 < $size1} {
            set r [lindex $result $t1]
            if {$r < $t2 || $t2 >= $size2} {
                # Deleted row
                lappend apa $dsym
                incr t1
            } elseif {$r == $t2} {
                #if {[string match Wm* [lindex $block2 $t2]]} {
                #    puts "Left : [lindex $block1 $t1]"
                #    puts "Right: [lindex $block2 $t2]"
                #    puts "Score: $scores($t1,$t2)"
                #}

                # If the score is too bad, don't do line parsing.
                if {[lindex $scores $t1 $t2] < 0} {
                    lappend apa "C"
                } else {
                    lappend apa "c"
                }
                incr t1
                incr t2
            } else {
                # Added row
                lappend apa $asym
                incr t2
            }
        } else {
            # Added row
            lappend apa $asym
            incr t2
        }
    }
    return $apa
}