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
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
|
<HTML>
<HEAD>
<TITLE>Snowball Manual</TITLE></HEAD>
<BODY BGCOLOR=WHITE>
<TABLE WIDTH=75% ALIGN=CENTER COLS=1>
<H1 ALIGN=CENTER>Snowball Manual</H1>
<TR><TD BGCOLOR="wheat">
<BR> <H2>Links to resources</H2>
<DL><DD><TABLE CELLPADDING=0>
<TR><TD><A HREF=".."> Snowball main page</A>
<TR><TD><A HREF="../runtime/use.html"> Using Snowball</A>
<TR><TD><A HREF="../algorithms/porter/stemmer.html"> Porter stemmer - a case study</A>
</TABLE></DL>
</TR>
<TR><TD>
<BR><BR>
<BR> <H2>Snowball definition</H2>
Snowball is a small string-handling language, and its name was chosen as a
tribute to SNOBOL (Farber 1964, Griswold 1968 -
see the references at the end of the
<A HREF="../texts/introduction.html">introduction</A>),
with which it shares the
concept of string patterns delivering signals that are used to control the
flow of the program.
<BR> <H2>1 Data types</H2>
The basic data types handled by Snowball are strings of characters, signed
integers, and boolean truth values, or more simply <I>strings</I>, <I>integers</I>
and <I>booleans</I>. Snowball's characters are either 8-bit wide, or
16-bit, depending on the mode of use. In particular, both 8-bit
ASCII and 16-bit Unicode are supported.
<BR> <H2>2 Names</H2>
A name in Snowball is a letter followed by zero or more letters, digits
and underlines. A name can be of type <I>string</I>, <I>integer</I>, <I>boolean</I>,
<I>routine</I>, <I>external</I> or <I>grouping</I>. All names must be declared. A
declaration has the form
<BR><PRE>
Ts ( ... )
</PRE>
where symbol <TT>T</TT> is one of <TT>string</TT>, <TT>integer</TT> etc, and the region in
brackets contains a list of names separated by whitespace. For example,
<BR><PRE>
integers ( p1 p2 )
booleans ( Y_found )
routines (
shortv
R1 R2
Step_1a Step_1b Step_1c Step_2 Step_3 Step_4 Step_5a Step_5b
)
externals ( stem )
groupings ( v v_WXY v_LSZ )
</PRE>
<TT>p1</TT> and <TT>p2</TT> are integers, <TT>Y_found</TT> is boolean, and so on. Snowball is quite
strict about the declarations, so all the names go in the same name space,
no name may be declared twice, all used names must be declared, no two
routine definitions can have the same name, etc. Names declared and
subsequently not used are merely reported in a warning message. A name may
not be one of the reserved words of Snowball.
<BR> <H2>3 Literals</H2>
A literal integer is a digit sequence, and is always interpreted as
decimal. A literal string is written between single quotes, for example,
<BR><PRE>
'aeiouy'
</PRE>
In a <TT>stringdef</TT> (see below), string may be preceded by the word <TT>hex</TT>,
or the word <TT>decimal</TT>,
in which case the contents are
interpreted as characters written out in hexadecimal, or decimal, notation.
The characters should be separated by spaces. For example,
<BR><PRE>
hex 'DA' /* is character hex DA */
hex 'D A' /* is the two characters, hex D and A (carriage
return, and line feed) */
decimal '10' /* character 10 (line feed) */
decimal '13 10' /* characters 13 and 10 (carriage return, and
line feed) */
</PRE>
The following forms are equivalent,
<BR><PRE>
hex 'd a' /* lower case also allowed */
hex '0D 000A' /* leading zeroes ignored */
hex ' D A ' /* extra spacing is harmless */
</PRE>
<TT>stringdef</TT>s define special <I>string macros</I>, to handle unusual
character combinations.
<BR><BR>
Macro <TT>m</TT> may is defined in the form <TT>stringdef m 'S'</TT>, where <TT>'S'</TT> is a
string, and <TT>m</TT> a sequence of one or more printing characters terminating
with whitespace.
<BR><BR>
Two special <I>insert characters</I> are defined by the directive
<TT>stringescapes AB</TT>, where <TT>A</TT> and <TT>B</TT> are printing characters, and <TT>A</TT> is not
single quote. (<TT>B</TT> may equal <TT>A</TT>, but then <TT>A</TT> itself can never be escaped.) For
example,
<BR><PRE>
stringescapes []
</PRE>
A subsequent occurrence of the same directive redefines the insert
characters.
<BR><BR>
Thereafter, <TT>[m]</TT> inside a string causes <TT>S</TT> to be substituted in place of <TT>m</TT>.
<BR><BR>
Immediately after the stringescapes directive, <TT>[']</TT> will substitute <TT>'</TT> and
<TT>[[]</TT> will substitute <TT>[</TT>, although macros <TT>'</TT> and <TT>[</TT> may subsequently be
redefined. A further feature is that <TT>[<I>W</I>]</TT> inside a string, where <TT><I>W</I></TT> is a
sequence of whitespace characters including one or more newlines, is
ignored. This enables long strings to be written over a number of lines.
<BR><BR>
For example,
<BR><PRE>
stringescapes []
/* special Spanish characters (in MS-DOS Latin I) */
stringdef a' hex 'A0' // a-acute
stringdef e' hex '82' // e-acute
stringdef i' hex 'A1' // i-acute
stringdef o' hex 'A2' // o-acute
stringdef u' hex 'A3' // u-acute
stringdef u" hex '81' // u-diaeresis
stringdef n~ hex 'A4' // n-tilde
/* and in the next string we define all the characters in Spanish
used to represent vowels
*/
define v 'aeiou[a'][e'][i'][o'][u'][u"]'
</PRE>
<BR> <H2>4 Routines</H2>
A routine definition has the form
<BR><PRE>
define R as C
</PRE>
where <TT>R</TT> is the routine name and <TT>C</TT> is a command, or bracketed group of
commands. So a routine is defined as a sequence of zero or more commands.
Snowball routines do not (at present) take parameters. For example,
<BR><PRE>
define Step_5b as ( // this defines Step_5b
['l'] // three commands here: [, 'l' and ]
R2 'l' // two commands, R2 and 'l'
delete // delete is one command
)
define R1 as $p1 <= cursor
/* R1 is defined as the single command "$p1 <= cursor" */
</PRE>
A routine is called simply by using its name, <TT>R</TT>, as a command.
<BR> <H2>5 Commands and signals</H2>
The flow of control in Snowball is arranged by the implicit use of
<I>signals</I>, rather than the explicit use of constructs like the <TT>if</TT>,
<TT>then</TT>, <TT>break</TT> of C. The scheme is designed for handling strings, but is
perhaps easier to introduce using integers. Suppose <TT>x</TT>, <TT>y</TT>, <TT>z</TT> ... are
integers. The command
<BR><PRE>
$x = 1
</PRE>
sets <TT>x</TT> to 1. The command
<BR><PRE>
$x > 0
</PRE>
tests if <TT>x</TT> is greater than zero. Both commands give a signal <B><I>t</I></B> or <B><I>f</I></B>,
(<I>true</I> or <I>false</I>), but while the second command gives <B><I>t</I></B> if <TT>x</TT> is greater
than zero and <B><I>f</I></B> otherwise, the first command always gives <B><I>t</I></B>. In Snowball,
every command gives a <B><I>t</I></B> or <B><I>f</I></B> signal. A sequence of commands can be turned
into as a single command by putting them in a list surrounded by round
brackets:
<BR><PRE>
( C<SUB>1</SUB> C<SUB>2</SUB> C<SUB>3</SUB> ... C<SUB>i</SUB> C<SUB>i+1</SUB> ... )
</PRE>
When this is obeyed, <TT>C<SUB>i+1</SUB></TT> will be obeyed if each of the preceding <TT>C<SUB>1</SUB></TT> ...
<TT>C<SUB>i</SUB></TT> give <B><I>t</I></B>, but as soon as a <TT>C<SUB>i</SUB></TT> gives <B><I>f</I></B>, the subsequent <TT>C<SUB>i+1</SUB> C<SUB>i+2</SUB></TT> ...
are ignored, and the whole sequence gives signal <B><I>f</I></B>. If all the <TT>C<SUB>i</SUB></TT> give <B><I>t</I></B>,
however, the bracketed command sequence also gives <B><I>t</I></B>. So,
<BR><PRE>
$x > 0 $y = 1
</PRE>
sets <TT>y</TT> to 1 if <TT>x</TT> is greater than zero. If <TT>x</TT> is less than or equal to zero
the two commands give <B><I>f</I></B>.
<BR><BR>
If <TT>C<SUB>1</SUB></TT> and <TT>C<SUB>2</SUB></TT> are commands, we can build up the larger commands,
<DL><DD><DL>
<DT><TT>C<SUB>1</SUB> or C<SUB>2</SUB></TT>
<DD>- Do <TT>C<SUB>1</SUB></TT>. If it gives <B><I>t</I></B> ignore <TT>C<SUB>2</SUB></TT>, otherwise do <TT>C<SUB>2</SUB></TT>. The resulting
signal is <B><I>t</I></B> if and only <TT>C<SUB>1</SUB></TT> or <TT>C<SUB>2</SUB></TT> gave <B><I>t</I></B>.
<DT><TT>C<SUB>1</SUB> and C<SUB>2</SUB></TT>
<DD>- Do <TT>C<SUB>1</SUB></TT>. If it gives <B><I>f</I></B> ignore <TT>C<SUB>2</SUB></TT>, otherwise do <TT>C<SUB>2</SUB></TT>. The resulting
signal is <B><I>t</I></B> if and only <TT>C<SUB>1</SUB></TT> and <TT>C<SUB>2</SUB></TT> gave <B><I>t</I></B>.
<DT><TT>not C</TT>
<DD>- Do <TT>C</TT>. The resulting signal is <B><I>t</I></B> if <TT>C</TT> gave <B><I>f</I></B>, otherwise <B><I>f</I></B>.
<DT><TT>try C</TT>
<DD>- Do <TT>C</TT>. The resulting signal is <B><I>t</I></B> whatever the signal of <TT>C</TT>.
<DT><TT>fail C</TT>
<DD>- Do <TT>C</TT>. The resulting signal is <B><I>f</I></B> whatever the signal of <TT>C</TT>.
</DL></DL>
So for example,
<DL><DD><DL>
<DT><TT>($x > 0 $y = 1) or ($y = 0)</TT>
<DD>- sets <TT>y</TT> to 1 if <TT>x</TT> is greater than zero, otherwise to zero.
<DT><TT>try( ($x > 0) and ($z > 0) $y = 1)</TT>
<DD>- sets <TT>y</TT> to 1 if both <TT>x</TT> and <TT>z</TT> are greater than 0, and gives <B><I>t</I></B>.
</DL></DL>
This last example is the same as
<BR><PRE>
try($x > 0 $z > 0 $y = 1)
</PRE>
so that <TT>and</TT> seems unnecessary here. But we will see that <TT>and</TT> has a
particular significance in string commands.
<BR><BR>
When a ‘monadic’ construct like <TT>not</TT>, <TT>try</TT> or <TT>fail</TT> is not followed by a
round bracket, the construct applies to the shortest following valid command.
So for example
<BR><PRE>
try not $x < 1 $z > 0
</PRE>
would mean
<BR><PRE>
try ( not ( $x < 1 ) ) $z > 0
</PRE>
because <TT>$x < 1</TT> is the shortest valid command following <TT>not</TT>, and then
<TT>not $x < 1</TT> is the shortest valid command following <TT>try</TT>.
<BR><BR>
The ‘diadic’ constructs like <TT>and</TT> and <TT>or</TT> must sit in a bracketed list
of commands anyway, for example,
<BR><PRE>
( C<SUB>1</SUB> C<SUB>2</SUB> and C<SUB>3</SUB> C<SUB>4</SUB> or C<SUB>5</SUB> )
</PRE>
And then in this case <TT>C<SUB>2</SUB></TT> and <TT>C<SUB>3</SUB></TT> are connected by the <TT>and</TT>; <TT>C<SUB>4</SUB></TT> and <TT>C<SUB>5</SUB></TT> are
connected by the <TT>or</TT>. So
<BR><PRE>
$x > 0 not $y > 0 or not $z > 0 $t > 0
</PRE>
means
<BR><PRE>
$x > 0 ((not ($y > 0)) or (not ($z > 0))) $t > 0
</PRE>
<TT>and</TT> and <TT>or</TT> are equally binding, and bind from left to right,
so <TT>C<SUB>1</SUB> or C<SUB>2</SUB> and C<SUB>3</SUB></TT> means <TT>(C<SUB>1</SUB> or C<SUB>2</SUB>) and C<SUB>3</SUB></TT> etc.
<BR> <H2>6 AEs and integer commands</H2>
An AE (arithmetic expression) consists of integer names and literal
numbers connected by diadic <TT>+</TT>, <TT>-</TT>, <TT>*</TT> and <TT>/</TT>, and monadic <TT>-</TT>, with the same
binding powers and semantics as C. An integer command has the form
<BR><PRE>
$X <I>op</I> AE
</PRE>
where <TT>X</TT> is an integer name and <I>op</I> is one of the six tests <TT>==</TT>, <TT>!=</TT>, <TT>>=</TT>, <TT>></TT>,
<TT><=</TT>, <TT><</TT>, or five assignments <TT>=</TT>, <TT>+=</TT>, <TT>-=</TT>, <TT>*=</TT>, <TT>/=</TT>. Again, the meanings are the
same as in C.
<BR><BR>
As well as integer names and literal numbers, the following may be used in
AEs:
<DL><DD><TABLE CELLPADDING=0>
<TR><TD><TT>minint</TT> <TD></TD><TD> - the smallest negative number
<TR><TD><TT>maxint</TT> <TD></TD><TD> - the largest positive number
<TR><TD><TT>sizeof s</TT> <TD></TD><TD> - the number of characters in <TT>s</TT>, where <TT>s</TT> is the name of a string
<TR><TD><TT>cursor</TT> <TD></TD><TD> - the current value of the string <I>cursor</I>
<TR><TD><TT>limit</TT> <TD></TD><TD> - the current value of the string <I>limit</I>
<TR><TD><TT>size</TT> <TD></TD><TD> - the size of the string, in characters
</TABLE></DL>
The <I>cursor</I> and <I>limit</I> concepts are explained below.
<BR><BR>
Examples of integer commands are,
<BR><PRE>
$p1 <= cursor // signal is f if the cursor is before position p1
$p1 = limit // set p1 to the string limit
</PRE>
<BR> <H2>7 String commands</H2>
If <TT>s</TT> is a string name, a string command has the form
<BR><PRE>
$s C
</PRE>
where <TT>C</TT> is a command that operate on the string. Strings can be processed
left-to-right or right-to-left, but we will describe only the
left-to-right case for now. The string has a <I>cursor</I>, which we will
denote by <B><I>c</I></B>, and a limit point, or <I>limit</I>, which we will denote by <B><I>l</I></B>. <B><I>c</I></B>
advances towards <B><I>l</I></B> in the course of a string command, but the various
constructs <TT>and</TT>, <TT>or</TT>, <TT>not</TT> etc have side-effects which keep moving it
backwards. Initially <B><I>c</I></B> is at the start and <B><I>l</I></B> the end of the string. For
example,
<BR><PRE>
'a|n|i|m|a|d|v|e|r|s|i|o|n'
| |
c l
</PRE>
<B><I>c</I></B>, and <B><I>l</I></B>, mark the boundaries between characters, and not
characters themselves. The characters between <B><I>c</I></B> and <B><I>l</I></B> will be denoted by
<B><I>c:l</I></B>.
<BR><BR>
If <TT>C</TT> gives <B><I>t</I></B>, the cursor <B><I>c</I></B> will have a new, well-defined value. But if <TT>C</TT>
gives <B><I>f</I></B>, <B><I>c</I></B> is undefined. Its later value will in fact be determined by the
outer context of commands in which <TT>C</TT> came to be obeyed, not by <TT>C</TT> itself.
<BR><BR>
Here is a list of the commands that can be used to operate on strings.
<H4>a) Setting a value</H4>
<DL>
<DT><TT>= S</TT>
<DD>where <TT>S</TT> is the name of a string or a literal string. <B><I>c:l</I></B> is set equal
to <TT>S</TT>, and <B><I>l</I></B> is adjusted to point to the end of the copied string. The
signal is <B><I>t</I></B>. For example,
<BR><PRE>
$x = 'animadversion' /* literal string */
$y = x /* string name */
</PRE>
</DL>
<H4>b) Basic tests</H4>
<DL>
<DT><TT>S</TT>
<DD>here and below, <TT>S</TT> is the name of a string or a literal string. If <B><I>c:l</I></B>
begins with the substring <TT>S</TT>, <B><I>c</I></B> is repositioned to the end of this
substring, and the signal is <B><I>t</I></B>. Otherwise the signal is <B><I>f</I></B>. For example,
<BR><PRE>
$x 'anim' /* gives t, assuming the string is 'animadversion' */
$x ('anim' 'ad' 'vers')
/* ditto */
$t = 'anim'
$x t /* ditto */
</PRE>
<DT><TT>C<SUB>1</SUB> or C<SUB>2</SUB></TT>
<DD>This is like the case for integers described above, but the extra
touch is that if <TT>C<SUB>1</SUB></TT> gives <B><I>f</I></B>, <B><I>c</I></B> is set back to its old position after
<TT>C<SUB>1</SUB></TT> has given <B><I>f</I></B> and before <TT>C<SUB>2</SUB></TT> is tried, so that the test takes place on
the same point in the string. So we have
<BR><PRE>
$x ('anim' /* signal t */
'ation' /* signal f */
) or
( 'an' /* signal t - from the beginning */
)
</PRE>
<DT><TT>true</TT>, <TT>false</TT>
<DD><TT>true</TT> is a dummy command that generates signal <B><I>t</I></B>. <TT>false</TT> generates
signal <B><I>f</I></B>. They are sometimes useful for emphasis,
<BR><PRE>
define start_off as true // nothing to do
define exception_list as false // put in among(...) list later
</PRE>
<TT>true</TT> is equivalent to <TT>()</TT>
<DT><TT>C<SUB>1</SUB> and C<SUB>2</SUB></TT>
<DD>And similarly <B><I>c</I></B> is set back to its old position after <TT>C<SUB>1</SUB></TT> has given <B><I>t</I></B>
and before <TT>C<SUB>2</SUB></TT> is tried. So,
<BR><PRE>
$x 'anim' and 'an' /* signal t */
$x ('anim' 'an') /* signal f, since 'an' and 'ad' mis-match */
</PRE>
<DT><TT>not C</TT>
<DT><TT>try C</TT>
<DD>These are like the integer tests, with the added feature that <B><I>c</I></B> is set
back to its old position after an <B><I>f</I></B> signal is turned into <B><I>t</I></B>. So,
<BR><PRE>
$x (not 'animation' not 'immersion')
/* both tests are done at the start of the string */
$x (try 'animus' try 'an'
'imad')
/* - gives t */
</PRE>
<DL><DD><TABLE CELLPADDING=0>
<TR><TD> <TT>try C</TT> <TD></TD><TD> is equivalent to <TD></TD><TD> <TT>C or true</TT>
</TABLE></DL>
<DT><TT>test C</TT>
<DD>This does command <TT>C</TT> but without advancing <B><I>c</I></B>. Its signal is the same as
the signal of <TT>C</TT>, but following signal <B><I>t</I></B>, <B><I>c</I></B> is set back to its old
value.
<DL><DD><TABLE CELLPADDING=0>
<TR><TD> <TT>test C</TT> <TD></TD><TD> is equivalent to <TD></TD><TD> <TT>not not C</TT>
<TR><TD> <TT>test C<SUB>1</SUB> C<SUB>2</SUB></TT> <TD></TD><TD> is equivalent to <TD></TD><TD> <TT>C<SUB>1</SUB> and C<SUB>2</SUB></TT>
</TABLE></DL>
<DT><TT>fail C</TT>
<DD>This does <TT>C</TT> and gives signal <B><I>f</I></B>. It is equivalent to <TT>C false</TT>. Like
<TT>false</TT> it is useful, but only rarely.
<DT><TT>do C</TT>
<DD>This does <TT>C</TT>, puts <B><I>c</I></B> back to its old value and gives signal <B><I>t</I></B>. It is
very useful as a way of suppressing the side effect of <B><I>f</I></B> signals and
cursor movement.
<DL><DD><TABLE CELLPADDING=0>
<TR><TD> <TT>do C</TT> <TD></TD><TD> is equivalent to <TD></TD><TD> <TT>try test C</TT>
<TR><TD> <TD></TD><TD> or <TD></TD><TD> <TT>test try C</TT>
</TABLE></DL>
<DT><TT>goto C</TT>
<DD><B><I>c</I></B> is moved right until obeying <TT>C</TT> gives <B><I>t</I></B>. But if <B><I>c</I></B> cannot be moved
right because it is at <B><I>l</I></B> the signal is <B><I>f</I></B>. <B><I>c</I></B> is set back to the position
it had before the last obeying of <TT>C</TT>, so the effect is to leave <B><I>c</I></B> before
the pattern which matched against <TT>C</TT>.
<BR><PRE>
$x goto 'ad' /* positions c after 'anim' */
$x goto 'ax' /* signal f */
</PRE>
<DT><TT>gopast C</TT>
<DD>Like goto, but <B><I>c</I></B> is not set back, so the effect is to leave <B><I>c</I></B> after
the pattern which matched against <TT>C</TT>.
<BR><PRE>
$x gopast 'ad' /* positions c after 'animad' */
</PRE>
<DT><TT>repeat C</TT>
<DD><TT>C</TT> is repeated until it gives <B><I>f</I></B>. When this happens <B><I>c</I></B> is set back to the
position it had before the last repetition of <TT>C</TT>, and <TT>repeat C</TT> gives
signal <B><I>t</I></B>. For example,
<BR><PRE>
$x repeat gopast 'a' /* position c after the last 'a' */
</PRE>
<DT><TT>loop AE C</TT>
<DD>This is like <TT>C C ... C</TT> written out AE times, where AE is an arithmetic
expression. For example,
<BR><PRE>
$x loop 2 gopast ('a' or 'e' or 'i' or 'o' or 'u')
/* position c after the second vowel */
</PRE>
The equivalent expression in C has the shape,
<BR><PRE>
{ int i;
int limit = AE;
for (i = 0; i < limit; i++) C;
}
</PRE>
<DT><TT>atleast AE C</TT>
<DD>This is equivalent to <TT>loop AE C repeat C</TT>.
<DT><TT>hop AE</TT>
<DD>moves <B><I>c</I></B> AE character positions towards <B><I>l</I></B>, but if AE is negative, or if
there are less than AE characters between <B><I>c</I></B> and <B><I>l</I></B> the signal is <B><I>f</I></B>.
For example,
<BR><PRE>
test hop 3
</PRE>
tests that <B><I>c:l</I></B> contains more than 2 characters.
<DT><TT>next</TT>
<DD>is equivalent to <TT>hop 1</TT>.
</DL>
<H4>c) Moving text about</H4>
We have seen in (a) that <TT>$x = y</TT>, when <TT>x</TT> and <TT>y</TT> are strings, sets <B><I>c:l</I></B> of <TT>x</TT>
to the value of <TT>y</TT>. Conversely
<BR><PRE>
$x => y
</PRE>
sets the value of <TT>y</TT> to the <B><I>c:l</I></B> region of <TT>x</TT>.
<BR><BR>
A more delicate mechanism for pushing text around is to define a substring,
or <I>slice</I> of the string being tested. Then
<DL>
<DT><TT>[</TT>
<DD>sets the left-end of the slice to <B><I>c</I></B>,
<DT><TT>]</TT>
<DD>sets the right-end of the slice to <B><I>c</I></B>,
<DT><TT>-> s</TT>
<DD>moves the slice to variable <TT>s</TT>,
<DT><TT><- S</TT>
<DD>replaces the slice with variable (or literal) <TT>S</TT>.
</DL>
For example
<BR><PRE>
/* assume x holds 'animadversion' */
$x ( [ // '[animadversion' - [ set as indicated
loop 2 gopast 'a'
// '[anima|dversion' - c is marked by '|'
] // '[anima]dversion' - ] set as indicated
-> y // y is 'anima'
)
</PRE>
For any string, the slice ends should be assumed to be unset until they are
set with the two commands <TT>[</TT>, <TT>]</TT>. Thereafter the slice ends will retain
the same values until altered.
<DL>
<DT><TT>delete</TT>
<DD>is equivalent to <TT><- ''</TT>
</DL>
This next example deletes all vowels in x,
<BR><PRE>
define vowel ('a' or 'e' or 'i' or 'o' or 'u')
....
$ x repeat ( gopast([vowel]) delete )
</PRE>
As this example shows, the slice markers <TT>[</TT> and <TT>]</TT> often appear as
pairs in a bracketed style, which makes for easy reading of the Snowball
scripts. But it must be remembered that, unusually in a computer
programming language, they are not true brackets.
<BR><BR>
More simply, text can be inserted at <B><I>c</I></B>.
<DL>
<DT><TT>insert S</TT>
<DD>insert variable or literal <TT>S</TT> before <B><I>c</I></B>, moving <B><I>c</I></B> to the right of the
insert. <TT><+</TT> is a synonym for <TT>insert</TT>.
<DT><TT>attach S</TT>
<DD>the same, but leave <B><I>c</I></B> at the left of the insert.
</DL>
<H4>d) Marks</H4>
The cursor, <B><I>c</I></B>, (and the limit, <B><I>l</I></B>) can be thought of as having a numeric
value, from zero upwards:
<BR><PRE>
| a | n | i | m | a | d | v | e | r | s | i | o | n |
0 1 2 3 4 5 6 7 8 9 10 11 12 13
</PRE>
It is these numeric values of <B><I>c</I></B> and <B><I>l</I></B> which are accessible through
<TT>cursor</TT> and <TT>limit</TT> in arithmetic expressions.
<DL>
<DT><TT>setmark X</TT>
<DD>sets <TT>X</TT> to the current value of <B><I>c</I></B>, where <TT>X</TT> is an integer variable.
<DT><TT>tomark AE</TT>
<DD>moves <B><I>c</I></B> forward to the position given by AE,
<DT><TT>atmark AE</TT>
<DD>tests if <B><I>c</I></B> is at position AE (<B><I>t</I></B> or <B><I>f</I></B> signal).
</DL>
In the case of <TT>tomark AE</TT>, a similar fail condition occurs as with <TT>hop AE</TT>.
If <B><I>c</I></B> is already beyond AE, or if position <B><I>l</I></B> is before position AE, the
signal is <B><I>f</I></B>.
<BR><BR>
In the stemming algorithms, certain regions of the word are defined by
setting marks, and later the failure condition of <TT>tomark</TT> is used to see if
<B><I>c</I></B> is inside a particular region.
<BR><BR>
Two other commands put <B><I>c</I></B> at <B><I>l</I></B>, and test if <B><I>c</I></B> is at <B><I>l</I></B>,
<DL>
<DT><TT>tolimit</TT>
<DD>moves <B><I>c</I></B> forward to <B><I>l</I></B> (signal <B><I>t</I></B> always),
<DT><TT>atlimit</TT>
<DD>tests if <B><I>c</I></B> is at <B><I>l</I></B> (<B><I>t</I></B> or <B><I>f</I></B> signal).
</DL>
<H4>e) Changing <B><I>l</I></B></H4>
In this account of string commands we see <B><I>c</I></B> moving right towards <B><I>l</I></B>, while
<B><I>l</I></B> stays fixed at the end. In fact <B><I>l</I></B> can be reset to a new position between
<B><I>c</I></B> and its old position, to act as a shorter barrier for the movement of <B><I>c</I></B>.
<DL>
<DT><TT>setlimit C<SUB>1</SUB> for C<SUB>2</SUB></TT>
<DD><TT>C<SUB>1</SUB></TT> is obeyed, and if it gives <B><I>f</I></B> the final value of <B><I>c</I></B> becomes the new
position of <B><I>l</I></B>. <B><I>c</I></B> is then set back to its old value before <TT>C<SUB>1</SUB></TT> was
obeyed, and <TT>C<SUB>2</SUB></TT> is obeyed. Finally <B><I>l</I></B> is set back to its old position.
The signal is <B><I>f</I></B> if either <TT>C<SUB>1</SUB></TT> or <TT>C<SUB>2</SUB></TT> gives <B><I>f</I></B>, otherwise <B><I>t</I></B>.
For example,
<BR><PRE>
$x ( setlimit goto 's' // 'animadver}sion' new l as marked '}'
for // below, '|' marks c after each goto
( goto 'a' and // '|animadver}sion'
goto 'e' and // 'animadv|er}sion'
goto 'i' and // 'an|imadver}sion'
)
)
</PRE>
This checks that x has characters ‘a’, ‘e’ and ‘i’ before the first
‘s’.
</DL>
<H4>f) Backward processing</H4>
String commands have been described with <B><I>c</I></B> to the left of <B><I>l</I></B> and moving
right. But the process can be reversed.
<DL>
<DT><TT>backwards C</TT>
<DD><B><I>c</I></B> and <B><I>l</I></B> are swapped over, and <B><I>c</I></B> moves left towards <B><I>l</I></B>. <TT>C</TT> is obeyed, the
signal given by <TT>C</TT> becomes the signal of <TT>backwards C</TT>, and <B><I>c</I></B> and <B><I>l</I></B> are
swapped back to their old values (except that <B><I>l</I></B> may have been adjusted
because of deletions and insertions). <TT>C</TT> cannot contain another
<TT>backwards</TT> command.
<DT><TT>reverse C</TT>
<DD>A similar idea, but here <B><I>c</I></B> simply moves left instead of moving right,
with the beginning of the string as the limit, <B><I>l</I></B>. <TT>C</TT> can contain other
<TT>reverse</TT> commands, but it cannot contain commands to do deletions or
insertions - it must be used for testing only. (Without this
restriction Snowball's semantics would become very untidy.)
</DL>
Forward and backward processing are entirely symmetric, except that forward
processing is the default direction, and literals strings are always
written out forwards, even when they are being tested backwards. So the
following are equivalent,
<BR><PRE>
$x (
'ani' 'mad' 'version' atlimit
)
$x backwards (
'version' 'mad' 'ani' atlimit
)
</PRE>
If a routine is defined for backwards mode processing, it must be included
inside a <TT>backwardmode(...)</TT> declaration.
<H4>g) <TT>substring</TT> and <TT>among</TT></H4>
The use of <TT>substring</TT> and <TT>among</TT> is central to the implementation of the
stemming algorithms. It is like a case switch on strings. In its simpler
form,
<BR><PRE>
substring among('S<SUB>1</SUB>' 'S<SUB>2</SUB>' 'S<SUB>3</SUB>' ...)
</PRE>
searches for the longest matching substring <TT>'S<SUB>1</SUB>'</TT> or <TT>'S<SUB>2</SUB>'</TT> or <TT>'S<SUB>3</SUB>'</TT> ... from
position <B><I>c</I></B>. (The <TT>'S<SUB>i</SUB>'</TT> must all be different.) So this has the same
semantics as
<BR><PRE>
('S<SUB>1</SUB>' or 'S<SUB>2</SUB>' or 'S<SUB>3</SUB>' ...)
</PRE>
- so long as the <TT>'S<SUB>i</SUB>'</TT> are written out in decreasing order of length.
<BR><BR>
<TT>substring</TT> may be omitted, in which case it is attached to its following
<TT>among</TT>, so
<BR><PRE>
among(...)
</PRE>
without a preceding <TT>substring</TT> is equivalent to
<BR><PRE>
(substring among(...))
</PRE>
<TT>substring</TT> may also be detached from its <TT>among</TT>, although it must
precede it textually in the same routine in which the <TT>among</TT> appears.
The more general form of <TT>substring ... among</TT> is,
<BR><PRE>
substring
...
among( 'S<SUB>11</SUB>' 'S<SUB>12</SUB>' ... (C<SUB>1</SUB>)
'S<SUB>21</SUB>' 'S<SUB>22</SUB>' ... (C<SUB>2</SUB>)
...
'S<SUB>n1</SUB>' 'S<SUB>n2</SUB>' ... (C<SUB>n</SUB>)
)
</PRE>
Obeying <TT>substring</TT> searches for a longest match among the <TT>'S<SUB>ij</SUB>'</TT>. The
signal from <TT>substring</TT> is <B><I>t</I></B> if a match is found, otherwise <B><I>f</I></B>. When the
<TT>among</TT> comes to be obeyed, the <TT>C<SUB>i</SUB></TT> corresponding to the matched <TT>'S<SUB>ij</SUB>'</TT> is
obeyed, and its signal becomes the signal of the <TT>among</TT> command.
<BR><BR>
<TT>substring/among</TT> pairs must match up textually inside each routine
definition. But there is no problem with an <TT>among</TT> containing other
<TT>substring/among</TT> pairs, and <TT>substring</TT> is optional before <TT>among</TT> anyway.
The essential constraint is that two <TT>substring</TT>s must be separated by an
<TT>among</TT>, and each <TT>substring</TT> must be followed by an <TT>among</TT>.
<BR><BR>
The effect of obeying <TT>among</TT> when the preceding <TT>substring</TT> is not obeyed
is undefined. This would happen for example here,
<BR><PRE>
try($x != 617 substring)
among(...) // 'substring' is bypassed in the exceptional case where x == 617
</PRE>
The significance of separating the <TT>substring</TT> from the <TT>among</TT> is to allow
them to work in different contexts. For example,
<BR><PRE>
setlimit tomark L for substring
among( 'S<SUB>11</SUB>' 'S<SUB>12</SUB>' ... (C<SUB>1</SUB>)
...
'S<SUB>n1</SUB>' 'S<SUB>n2</SUB>' ... (C<SUB>n</SUB>)
)
</PRE>
Here the test for the longest <TT>'S<SUB>ij</SUB>'</TT> is constrained to the region between <B><I>c</I></B>
and the mark point given by integer <TT>L</TT>. But the commands <TT>C<SUB>i</SUB></TT> operate outside
this limit. Another example is
<BR><PRE>
reverse substring
among( 'S<SUB>11</SUB>' 'S<SUB>12</SUB>' ... (C<SUB>1</SUB>)
...
'S<SUB>n1</SUB>' 'S<SUB>n2</SUB>' ... (C<SUB>n</SUB>)
)
</PRE>
The substring test is in the opposite direction in the string to the
direction of the commands <TT>C<SUB>i</SUB></TT>.
<BR><BR>
The last <TT>(C<SUB>n</SUB>)</TT> may be omitted, in which case <TT>(true)</TT> is assumed.
<BR><BR>
Another possible abbreviation is that when <TT>substring</TT> is omitted, a
construct such as
<BR><PRE>
among( 'S<SUB>11</SUB>' 'S<SUB>12</SUB>' ... (C C<SUB>1</SUB>)
'S<SUB>21</SUB>' 'S<SUB>22</SUB>' ... (C C<SUB>2</SUB>)
...
'S<SUB>n1</SUB>' 'S<SUB>n2</SUB>' ... (C C<SUB>n</SUB>)
)
</PRE>
can be written
<BR><PRE>
among( (C)
'S<SUB>11</SUB>' 'S<SUB>12</SUB>' ... (C<SUB>1</SUB>)
'S<SUB>21</SUB>' 'S<SUB>22</SUB>' ... (C<SUB>2</SUB>)
...
'S<SUB>n1</SUB>' 'S<SUB>n2</SUB>' ... (C<SUB>n</SUB>)
)
</PRE>
and this is just equivalent to
<BR><PRE>
substring C
among( 'S<SUB>11</SUB>' 'S<SUB>12</SUB>' ... (C<SUB>1</SUB>)
'S<SUB>21</SUB>' 'S<SUB>22</SUB>' ... (C<SUB>2</SUB>)
...
'S<SUB>n1</SUB>' 'S<SUB>n2</SUB>' ... (C<SUB>n</SUB>)
)
</PRE>
In its most general form, each string <TT>'S<SUB>ij</SUB>'</TT> may be optionally followed by a
routine name,
<BR><PRE>
among( (C)
'S<SUB>11</SUB>' R<SUB>11</SUB> 'S<SUB>12</SUB>' R<SUB>12</SUB> ... (C<SUB>1</SUB>)
'S<SUB>21</SUB>' R<SUB>21</SUB> 'S<SUB>22</SUB>' R<SUB>22</SUB> ... (C<SUB>2</SUB>)
...
'S<SUB>n1</SUB>' R<SUB>n1</SUB> 'S<SUB>n2</SUB>' R<SUB>n1</SUB> ... (C<SUB>n</SUB>)
)
</PRE>
So here each <TT>R<SUB>ij</SUB></TT> is either a routine name or is null. If null, it is equivalent
to a routine which simply returns signal <B><I>t</I></B>,
<BR><PRE>
define null as true
</PRE>
- so we can imagine each <TT>'S<SUB>ij</SUB>'</TT> having its associated routine
<TT>R<SUB>ij</SUB></TT>. Then obeying the <TT>among</TT> causes a search for the longest
<TT>'S<SUB>ij</SUB>'</TT> whose corresponding routine
<TT>R<SUB>ij</SUB></TT> gives <B><I>t</I></B>. The routines
<TT>R<SUB>ij</SUB></TT> should be written without any side-effects, other than the inevitable cursor
movement. (<B><I>c</I></B> is in any case set back to its old value following a call of
<TT>R<SUB>ij</SUB></TT>.)
<BR> <H2>8 Booleans</H2>
<TT>set B</TT> and <TT>unset B</TT> set <TT>B</TT> to true and false respectively, where <TT>B</TT> is a
boolean name. <TT>B</TT> as a command gives a signal <B><I>t</I></B> if it is set true, <B><I>f</I></B>
otherwise. For example,
<BR><PRE>
booleans ( Y_found ) // declare the boolean
....
unset Y_found // unset it
do ( ['y'] <-'Y' set Y_found )
/* if c:l begins 'y' replace it by 'Y' and set Y_found */
do repeat(goto (v ['y']) <-'Y' set Y_found)
/* repeatedy move down the string looking for v 'y' and
replacing 'y' with 'Y'. Whenever the replacement takes
place set Y_found. v is a test for a vowel, defined as
a grouping (see below). */
/* Y_found means there are some letters Y in the string.
Later we can use this to trigger a conversion back to
lower case y. */
....
do (Y_found repeat(goto (['Y']) <- 'y')
</PRE>
<BR> <H2>9 Groupings</H2>
A grouping brings characters together and enables them to be looked for
with a single test.
<BR><BR>
If <TT>G</TT> is declared as a grouping, it can be defined by
<BR><PRE>
define G G<SUB>1</SUB> <I>op</I> G<SUB>2</SUB> <I>op</I> G<SUB>3</SUB> ...
</PRE>
where <I>op</I> is <TT>+</TT> or <TT>-</TT>, and <TT>G<SUB>1</SUB></TT>, <TT>G<SUB>2</SUB></TT>, <TT>G<SUB>3</SUB></TT> are literal strings, or groupings that
have already been defined. (There can be zero or more of these additional
<I>op</I> components). For example,
<BR><PRE>
define capital_letter 'ABDEFGHIJKLMNOPQRSTUVWXYZ'
define small_letter 'abdefghijklmnopqrstuvwxyz'
define letter capital_letter + small_letter
define vowel 'aeiou' + 'AEIOU'
define consonant letter - vowel
define digit '0123456789'
define alphanumeric letter + digit
</PRE>
Once <TT>G</TT> is defined, it can be used as a command, and is equivalent to a test
<BR><PRE>
'ch1' or 'ch2' or ...
</PRE>
where <TT>ch1</TT>, <TT>ch2</TT> ... list all the characters in the grouping.
<BR><BR>
<TT>non G</TT> is the converse test, and matches any character except the
characters of <TT>G</TT>. Note that <TT>non G</TT> is not the same as <TT>not G</TT>, in fact
<BR><PRE>
non G is equivalent to (not G next)
</PRE>
<TT>non</TT> may be optionally followed by hyphen, so one may write
<BR><PRE>
non-vowel
non-digit
</PRE>
etc.
<BR> <H2>10 A Snowball program</H2>
A complete program consists of a sequence of declarations followed by a
sequence of definitions of groupings and routines. Routines which are
implicitly defined as operating on <B><I>c:l</I></B> from right to left must be included
in a <TT>backwardmode(...)</TT> declaration.
<BR><BR>
A Snowball program is called up via a simple
<A HREF="../runtime/use.html">API</A>
through its defined
externals. For example,
<BR><PRE>
externals ( stem1 stem2 )
....
define stem1 as ( ... /* stem1 commands */ )
define stem2 as ( ... /* stem2 commands */ )
</PRE>
The API also allows a current string to be defined, and this becomes the
<B><I>c:l</I></B> string for the external routine to work on. Its final value is the
result handed back through the API.
<BR><BR>
The strings, integers and booleans are accessible from any point in the
program, and exist throughout the running of the Snowball program. They are
therefore like static declarations in C.
<BR> <H2>11 Comments, and other whitespace fillers</H2>
At a deeper level, a program is a sequence of <I>tokens</I>, interspersed with
whitespace. Names, reserved words, literal numbers and strings are all
tokens. Various symbols, made up of non-alphanumerics, are also tokens.
<BR><BR>
A name, reserved word or number is terminated by the first character that
cannot form part of it. A symbol is recognised as the longest sequence of
characters that forms a valid symbol. So <TT>+=-</TT> is two symbols, <TT>+=</TT> and
<TT>-</TT>, because <TT>+=</TT> is a valid symbol in the language while <TT>+=-</TT> is not.
Whitespace separates tokens but is otherwise ignored. This of course is
like C.
<BR><BR>
Anywhere that whitespace can occur, there may also occur:
<BR><BR>
(a) Comments, in the usual multi-line <TT>/* .... */</TT> or single line
<TT>// ...</TT> format.
<BR><BR>
(b) Get directives. These are like <TT>#include</TT> commands in C, and have the form
<TT>get 'S'</TT>, where <TT>'S'</TT> is a literal string. For example,
<BR><PRE>
get '/home/martin/snowball/main-hdr' // include the file contents
</PRE>
(c) <TT>stringescapes XY</TT> where <TT>X</TT> and <TT>Y</TT> are any two printing characters.
<BR><BR>
(d) <TT>stringdef m 'S'</TT> where <TT>m</TT> is sequence of characters not including
whitespace and terminated with whitespace, and <TT>'S'</TT> is a literal string.
<BR><BR>
</TR>
<TR><TD BGCOLOR="lightblue">
<BR> <H2>Snowball syntax</H2>
<DL><DD>
<TT>||</TT> is used for alternatives, <TT>[<I>X</I>]</TT> means that <I>X</I> is
optional, and <TT>[<I>X</I>]*</TT> means that <I>X</I> is repreated zero or more
times. meta-symbols are defined on the left. <TT><char></TT> means any
character.
<BR><BR>
The definition of <TT>literal string</TT> does not allow for the escaping
conventions established by the <TT>stringescapes</TT> directive. The command
<TT>?</TT> is a debugging aid.
<BR><BR>
<BR><PRE>
<letter> ::= a || b || ... || z || A || B || ... || Z
<digit> ::= 0 || 1 || ... || 9
<name> ::= <letter> [ <letter> || <digit> || _ ]*
<s_name> ::= <name>
<i_name> ::= <name>
<b_name> ::= <name>
<r_name> ::= <name>
<g_name> ::= <name>
<literal string>::= '[<char>]*'
<number> ::= <digit> [ <digit> ]*
S ::= <s_name> || <literal string>
G ::= <g_name> || <literal string>
<declaration> ::= strings ( [<s_name>]* ) ||
integers ( [<i_name>]* ) ||
booleans ( [<b_name>]* ) ||
routines ( [<r_name>]* ) ||
externals ( [<r_name>]* ) ||
groupings ( [<g_name>]* )
<r_definition> ::= define <r_name> as C
<plus_or_minus> ::= + || -
<g_definition> ::= define <g_name> G [ <plus_or_minus> G ]*
AE ::= (AE) ||
AE + AE || AE - AE || AE * AE || AE / AE || - AE ||
maxint || minint || cursor || limit || size ||
sizeof <s_name> || <i_name> || <number>
<i_command> ::= $ <i_name> = AE ||
$ <i_name> += AE || $ <i_name> -= AE ||
$ <i_name> *= AE || $ <i_name> /= AE ||
$ <i_name> == AE || $ <i_name> != AE ||
$ <i_name> > AE || $ <i_name> >= AE ||
$ <i_name> < AE || $ <i_name> <= AE ||
<s_command> ::= $ <s_name> C
C ::= ( [C]* ) ||
<i_command> || <s_command> || C or C || C and C ||
not C || test C || try C || do C || fail C ||
goto C || gopast C || repeat C || loop AE C ||
atleast AE C || S || = S || insert S || attach S ||
<- S || delete || hop AE || next ||
=> <s_name> || [ || ] || -> <s_name> ||
setmark <i_name> || tomark AE || atmark AE ||
tolimit || atlimit || setlimit C for C ||
backwards C || reverse C || substring ||
among ( [<literal string> [<r_name>] || (C)]* ) ||
set <b_name> || unset <b_name> || <b_name> ||
<r_name> || <g_name> || non [-] <g_name> ||
true || false || ?
P ::= [P]* || <declaration> ||
<r_definition> || <g_definition> ||
backwardmode ( P )
<program> ::= P
synonyms: <+ for insert
</PRE>
</DL>
</TR>
</TABLE>
</BODY>
</HTML>
|