aboutsummaryrefslogtreecommitdiff
path: root/src/interp/guess.boot
blob: 30c4e96c2d49a8eb61df81478e691c6bd9caf07f (plain)
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
-- Copyright (c) 1991-2002, The Numerical ALgorithms Group Ltd.
-- All rights reserved.
-- Copyright (C) 2007-2010, Gabriel Dos Reis.
-- All rights reserved.
--
-- Redistribution and use in source and binary forms, with or without
-- modification, are permitted provided that the following conditions are
-- met:
--
--     - Redistributions of source code must retain the above copyright
--       notice, this list of conditions and the following disclaimer.
--
--     - Redistributions in binary form must reproduce the above copyright
--       notice, this list of conditions and the following disclaimer in
--       the documentation and/or other materials provided with the
--       distribution.
--
--     - Neither the name of The Numerical ALgorithms Group Ltd. nor the
--       names of its contributors may be used to endorse or promote products
--       derived from this software without specific prior written permission.
--
-- THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS
-- IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
-- TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
-- PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
-- OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
-- EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
-- PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
-- PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
-- LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
-- NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
-- SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.


$minThreshold := 3
$maxThreshold := 7
 
--=======================================================================
--                        Build Directories
--=======================================================================
buildOperationWordTable() ==
  $opWordTable := buildWordTable [PNAME x for x in allOperations()]
 
buildWordTable u ==
  table:= hashTable 'EQ
  for s in u repeat
    words := wordsOfString s
    key := UPCASE s.0
    HPUT(table,key,[[s,:words],:HGET(table,key)])
  for key in HKEYS table repeat
    HPUT(table,key,
      listSort(function GLESSEQP,removeDupOrderedAlist
        listSort(function GLESSEQP, HGET(table,key),function first),
          function second))
  table

measureWordTable u ==
  +/[+/[#entry for entry in HGET(u,key)] for key in HKEYS u]

removeDupOrderedAlist u ==
  -- removes duplicate entries in ordered alist
  -- (where duplicates are adjacent)
  for x in tails u repeat
    (y := rest x) and first first x = first first y => x.rest := rest y
  u
 
wordsOfString(s) == [UPCASE x for x in wordsOfStringKeepCase s]
 
wordsOfStringKeepCase s == wordsOfString1(s,0) or [COPY s]
 
wordsOfString1(s,j) ==
  k := or/[i for i in j..(MAXINDEX(s)-1) | upperCase? s.i] =>
    tailWords:=
      upperCase? s.(k+1) =>
        n:= or/[i for i in (k+2)..(MAXINDEX(s)-1)|not upperCase? s.i]
        null n => [SUBSTRING(s,k,nil)]
        n > k+1 => [SUBSTRING(s,k,n-k-1),:wordsOfString1(s,n-1)]
      m := or/[i for i in (k+2)..(MAXINDEX(s)-1) | upperCase? s.i] =>
        [SUBSTRING(s,k,m-k),:wordsOfString1(s,m)]
      [SUBSTRING(s,k,nil)]
    k > j+1 => [SUBSTRING(s,j,k-j),:tailWords]
    tailWords
  nil

wordKeys s == 
  removeDuplicates [UPCASE s.0,:fn(s,1,-1,MAXINDEX s,nil)] where fn(s,i,lastKeyIndex,n,acc) ==
    i > n => acc
    upperCase? s.i =>
--    i = lastKeyIndex + 1 => fn(s,i + 1,i,n,[s.i,:rest acc])
      fn(s,i + 1,i,n,[s.i,:acc])
    fn(s,i + 1,lastKeyIndex,n,acc)

--=======================================================================
--               Augment Function Directories
--=======================================================================
add2WordFunctionTable fn ==
--called from DEF
  $functionTable and
    null LASSOC(s := PNAME fn,HGET($functionTable,(key := UPCASE s.0))) =>
      HPUT($functionTable,key,[[s,:wordsOfString s],:HGET($functionTable,key)])
 
--=======================================================================
--                       Guess Function Name
--=======================================================================
findWords(word,table) == 
  $lastWord := word
  $lastTable:= table
  $totalWords:= nil
  $countThreshold := $minThreshold
  $lastMinimum := -1
  res := findApproximateWords(word,table) 
  if null res then
    $countThreshold := $countThreshold + 2
    res := findApproximateWords(word,table) 
  $lastAlist := mySort res =>
--    $lastMinimum := first LAST $lastAlist
--    $lastWords := wordSort CDAR $lastAlist
--    $totalWords:= $lastWords
--    $lastAlist := rest  $lastAlist
--    $totalWords
      $lastMinimum := CAAR $lastAlist
      $lastWords := wordSort CDAR $lastAlist
      $totalWords:= $lastWords
      $lastAlist := rest  $lastAlist
      $totalWords
  $lastWords := nil

wordSort u == removeDuplicates listSort(function GLESSEQP,u)

more() == moreWords($lastWord,$lastTable)

moreWords(word,table) ==
  $lastAlist =>
     $lastMinimum := first LAST pp $lastAlist
     numberOfLastWords := #$lastWords
     $lastWords := "append"/(ASSOCRIGHT $lastAlist)
     if #$lastWords > numberOfLastWords then 
       trialLastAlist := 
         [p for p in $lastAlist | p.0 < $maxThreshold]
       trialLastWords := "append"/(ASSOCRIGHT trialLastAlist)
       if #trialLastWords > numberOfLastWords then
         $lastWords := trialLastWords         
     $totalWords:= wordSort [:$lastWords,:$totalWords]
     $lastAlist := nil
     $totalWords
  $countThreshold := $countThreshold + 2
  $lastAlist := findApproximateWords(word,table)
  moreWords(word,table)

findApproximateWords(word,table) ==
  count := $countThreshold
  words:= wordsOfString word
  upperWord:= UPCASE COPY word
  n := #words
  threshold:=
    n = 1 => count
    count+1
 
  --first try to break up as list of words
  alist:= nil
  for i in 1..#words repeat
    $penalty :local := (i = 1 => 0; 1)
    wordAlist:= HGET(table,UPCASE (first words).0)
    for [x,:wordList] in wordAlist repeat
      k := findApproxWordList(words,wordList,n,threshold,#wordList) 
      k =>
        k := k + $penalty
        k <= $lastMinimum => 'skip
        alist := consAlist(k,x,alist)
      
    if i = 1 and null alist then
        --no winners, so try flattening to upper case and checking again
        wordSize := SIZE word
        lastThreshold := MAX(threshold - 1,wordSize/2)
        for [x,:.] in wordAlist repeat
          k := deltaWordEntry(upperWord,UPCASE x)
          k < lastThreshold => alist := consAlist(k,x,alist)

    rotateWordList words

  alist

consAlist(x,y,alist) ==
  u := ASSOC(x,alist) => 
    u.rest := [y,:rest u]
    alist
  [[x,y],:alist]

findApproxWordList(words,wordList,n,threshold,w) ==
  val := findApproxWordList1(words,wordList,n,threshold,w)
  null val => val
--pp [val,:wordList]
  val

findApproxWordList1(words,wordList,n,threshold,w) ==
  two := threshold - 2
  n = w => 
      k := findApproxSimple(words,wordList,threshold) => k 

      n < 3 => false
      threshold := threshold - 1
      sum := 0      --next, throw out one bad word

      badWord := false
      for entry in wordList for part in words while sum < threshold repeat
        k:= deltaWordEntry(part,entry)
        k < two => sum:= sum + k
        null badWord => badWord := true
        sum := 1000
      sum < threshold => 
--      pp [2,sum,wordList]
        sum + 2

  n+1 = w =>    --assume one word is missing
      sum := 0
      badWord := false
      for entries in tails wordList for part in words
           while sum < threshold repeat
        entry := first entries
        k:= deltaWordEntry(part,entry)
        k < two => sum:= sum + k
        null badWord =>
          badWord := true
          entries := rest entries      --skip this bad word
          entry := first entries
          k := deltaWordEntry(part,entry)
          k < two => sum := sum + k
          sum := 1000
        sum := 1000
      sum < threshold => 
--      pp [3,sum,wordList]
        sum + 2
      false
  n-1 = w =>    --assume one word too many
      sum := 0  --here: KEEP it hard to satisfy
      badWord := false
      for entry in wordList for parts in tails words
           while sum < threshold repeat
        part := first parts
        k:= deltaWordEntry(part,entry)
        k < 2 => sum:= sum + k
        null badWord =>
          badWord := true
          parts := rest parts      --skip this bad word
          part := first parts
          k := deltaWordEntry(part,entry)
          k < 2 => sum := sum + k
          return (sum := 1000)
        return (sum := 1000)
      sum < threshold =>
--      pp [4,sum,wordList]
        $penalty = 1 => sum
        sum + 1
      false
  false

findApproxSimple(words,wordList,threshold) ==
  sum := 0
  --first try matching words in order
  for entry in wordList for part in words while sum < threshold repeat
    sum:= sum + deltaWordEntry(part,entry)
  sum < threshold => 
--  pp ['"--->",sum,:wordList]
    sum
  nil
      
rotateWordList u ==
  v := u
  p := first v
  while QCDR v repeat
    v.first := second v
    v := QCDR v
  v.first := p
  u

deltaWordEntry(word,entry) ==
  word = entry => 0
  word.0 ~= entry.0 => 1000
  #word > 2 and stringPrefix?(word,entry) => 1
  ABS(diff := SIZE word - SIZE entry) > 4 => 1000
  canForgeWord(word,entry)
 
--+ Note these are optimized definitions below-- see commented out versions
--+   to understand the algorithm
canForgeWord(word,entry) ==
  forge(word,0,MAXINDEX word,entry,0,MAXINDEX entry,0)
 
forge(word,w,W,entry,e,E,n) ==
  w > W =>
    e > E => n
    QSADD1 QSPLUS(E-e,n)
  e > E => QSADD1 QSPLUS(W-w,n)
  word.w = entry.e => forge(word,w+1,W,entry,e+1,E,n)
  w=W or e=E => forge(word,w+1,W,entry,e+1,E,QSADD1 n)
  word.w=entry.(e+1) =>
    word.(w+1) = entry.e => forge(word,w+2,W,entry,e+2,E,QSADD1 n)
    forge(word,w+1,W,entry,e+2,E,QSADD1 n)
  word.(w+1)=entry.e => forge(word,w+2,W,entry,e+1,E,QSADD1 n)
 
  (deltaW := W-w) > 1 and (deltaE := E-e) > 1 =>
    --if word is long, can we delete chars to match 2 consective chars
    deltaW >= deltaE and
      (k := or/[j for j in (w+2)..(W-1) | word.j = entry.e])
        and word.(k+1) = entry.(e+1) =>
          forge(word,k+2,W,entry,e+2,E,QSPLUS(k-w,n))
    deltaW <= deltaE and
    --if word is short, can we insert chars so as to match 2 consecutive chars
      (k := or/[j for j in (e+2)..(E-1) | word.w = entry.j])
        and word.(w+1) = entry.(k+1) =>
          forge(word,w+2,W,entry,k+2,E,QSPLUS(n,k-e))
    forge(word,w+1,W,entry,e+1,E,QSADD1 n)
  --check for two consecutive matches down the line
  forge(word,w+1,W,entry,e+1,E,QSADD1 n)
 
--+ DO NOT REMOVE DEFINITIONS BELOW which explain the algorithm
--+ canForgeWord(word,entry) ==--
--+ [d,i,s,t] := forge(word,0,MAXINDEX word,entry,0,MAXINDEX entry,0,0,0,0)
--+ --d=deletions, i=insertions, s=substitutions, t=transpositions
--+ --list is formed only for tuning purposes-- remove later on
--+ d + i + s + t
 
--+forge(word,w,W,entry,e,E,d,i,s,t) ==
--+  w > W =>
--+    e > E => [d,i,s,t]
--+    [d,E-e+i+1,s,t]
--+  e > E => [W-w+d+1,i,s,t]
--+  word.w = entry.e => forge(word,w+1,W,entry,e+1,E,d,i,s,t)
--+  w=W or e=E => forge(word,w+1,W,entry,e+1,E,d,i,s+1,t)
--+  word.w=entry.(e+1) =>
--+    word.(w+1) = entry.e => forge(word,w+2,W,entry,e+2,E,d,i,s,t+1)
--+    forge(word,w+1,W,entry,e+2,E,d,i+1,s,t)
--+  word.(w+1)=entry.e => forge(word,w+2,W,entry,e+1,E,d+1,i,s,t)
--+
--+  (deltaW := W-w) > 1 and (deltaE := E-e) > 1 =>
--+    --if word is long, can we delete chars to match 2 consective chars
--+    deltaW >= deltaE and
--+      (k := or/[j for j in (w+2)..(W-1) | word.j = entry.e])
--+        and word.(k+1) = entry.(e+1) =>
--+          forge(word,k+2,W,entry,e+2,E,d+k-w,i,s,t)
--+    deltaW <= deltaE and
--+    --if word is short, can we insert chars so as to match 2 consecutive chars
--+      (k := or/[j for j in (e+2)..(E-1) | word.w = entry.j])
--+        and word.(w+1) = entry.(k+1) =>
--+          forge(word,w+2,W,entry,k+2,E,d,i+k-e,s,t)
--+    forge(word,w+1,W,entry,e+1,E,d,i,s+1,t)
--+  --check for two consecutive matches down the line
--+  forge(word,w+1,W,entry,e+1,E,d,i,s+1,t)
 
mySort u == listSort(function GLESSEQP,u)