[gforth] / gforth / regexp.fs  

gforth: gforth/regexp.fs


1 : pazsan 1.1 \ Regexp compile
2 :    
3 : anton 1.13 \ Copyright (C) 2005,2006,2007,2008 Free Software Foundation, Inc.
4 : anton 1.3
5 :     \ This file is part of Gforth.
6 :    
7 :     \ Gforth is free software; you can redistribute it and/or
8 :     \ modify it under the terms of the GNU General Public License
9 : anton 1.10 \ as published by the Free Software Foundation, either version 3
10 : anton 1.3 \ of the License, or (at your option) any later version.
11 :    
12 :     \ This program is distributed in the hope that it will be useful,
13 :     \ but WITHOUT ANY WARRANTY; without even the implied warranty of
14 :     \ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 :     \ GNU General Public License for more details.
16 :    
17 :     \ You should have received a copy of the GNU General Public License
18 : anton 1.10 \ along with this program. If not, see http://www.gnu.org/licenses/.
19 : anton 1.3
20 : pazsan 1.1 \ The idea of the parser is the following:
21 :     \ As long as there's a match, continue
22 :     \ On a mismatch, LEAVE.
23 :     \ Insert appropriate control structures on alternative branches
24 :     \ Keep the old pointer (backtracking) on the stack
25 :     \ I try to keep the syntax as close to a real regexp system as possible
26 :     \ All regexp stuff is compiled into one function as forward branching
27 :     \ state machine
28 :    
29 : pazsan 1.2 \ special control structure
30 : pazsan 1.1
31 : pazsan 1.8 : FORK ( compilation -- orig ; run-time f -- ) \ gforth
32 :     \G AHEAD-like control structure: calls the code after JOIN.
33 : pazsan 1.1 POSTPONE call >mark ; immediate restrict
34 : pazsan 1.8 : JOIN ( orig -- ) \ gforth
35 :     \G THEN-like control structure for FORK
36 :     postpone THEN ; immediate restrict
37 : pazsan 1.1
38 :     \ Charclasses
39 :    
40 :     : +bit ( addr n -- ) + 1 swap c! ;
41 :     : -bit ( addr n -- ) + 0 swap c! ;
42 :     : @+ ( addr -- n addr' ) dup @ swap cell+ ;
43 :    
44 :     0 Value cur-class
45 : pazsan 1.8 : charclass ( -- ) \ regexp-cg
46 :     \G Create a charclass
47 :     Create here dup to cur-class $100 dup allot erase ;
48 :     : +char ( char -- ) \ regexp-cg
49 :     \G add a char to the current charclass
50 :     cur-class swap +bit ;
51 :     : -char ( char -- ) \ regexp-cg
52 :     \G remove a char from the current charclass
53 :     cur-class swap -bit ;
54 :     : ..char ( start end -- ) \ regexp-cg
55 :     \G add a range of chars to the current charclass
56 :     1+ swap ?DO I +char LOOP ;
57 : pazsan 1.1 : or! ( n addr -- ) dup @ rot or swap ! ;
58 :     : and! ( n addr -- ) dup @ rot and swap ! ;
59 : pazsan 1.8 : +class ( class -- ) \ regexp-cg
60 :     \G union of charclass @var{class} and the current charclass
61 :     $100 0 ?DO @+ swap
62 :     cur-class I + or! cell +LOOP drop ;
63 :     : -class ( class -- ) \ regexp-cg
64 :     \G subtract the charclass @var{class} from the current charclass
65 :     $100 0 ?DO @+ swap invert
66 :     cur-class I + and! cell +LOOP drop ;
67 : pazsan 1.1
68 :     : char? ( addr class -- addr' flag )
69 :     >r count r> + c@ ;
70 :    
71 :     \ Charclass tests
72 :    
73 : pazsan 1.8 : c? ( addr class -- ) \ regexp-pattern
74 :     \G check @var{addr} for membership in charclass @var{class}
75 :     ]] char? 0= ?LEAVE [[ ; immediate
76 :     : -c? ( addr class -- ) \ regexp-pattern
77 :     \G check @var{addr} for not membership in charclass @var{class}
78 :     ]] char? ?LEAVE [[ ; immediate
79 : pazsan 1.1
80 :     charclass digit '0 '9 ..char
81 :     charclass blanks 0 bl ..char
82 :     \ bl +char #tab +char #cr +char #lf +char ctrl L +char
83 :     charclass letter 'a 'z ..char 'A 'Z ..char
84 :     charclass any 0 $FF ..char #lf -char
85 :    
86 : pazsan 1.8 : \d ( addr -- addr' ) \ regexp-pattern
87 :     \G check for digit
88 :     ]] digit c? [[ ; immediate
89 :     : \s ( addr -- addr' ) \ regexp-pattern
90 :     \G check for blanks
91 :     ]] blanks c? [[ ; immediate
92 :     : .? ( addr -- addr' ) \ regexp-pattern
93 :     \G check for any single charachter
94 :     ]] any c? [[ ; immediate
95 :     : -\d ( addr -- addr' ) \ regexp-pattern
96 :     \G check for not digit
97 :     ]] digit -c? [[ ; immediate
98 :     : -\s ( addr -- addr' ) \ regexp-pattern
99 :     \G check for not blank
100 :     ]] blanks -c? [[ ; immediate
101 :     : ` ( "char" -- ) \ regexp-pattern
102 :     \G check for particular char
103 : pazsan 1.1 ]] count [[ char ]] Literal <> ?LEAVE [[ ; immediate
104 : pazsan 1.17 : -` ( "char" -- ) \ regexp-pattern
105 :     \G check for particular char
106 :     ]] count [[ char ]] Literal = ?LEAVE [[ ; immediate
107 : pazsan 1.1
108 :     \ loop stack
109 :    
110 :     Variable loops $40 3 * cells allot
111 :     : 3@ ( addr -- a b c ) dup >r 2 cells + @ r> 2@ ;
112 :     : 3! ( a b c addr -- ) dup >r 2! r> 2 cells + ! ;
113 :     : loops> ( -- addr ) -3 loops +! loops @+ swap cells + 3@ ;
114 :     : >loops ( addr -- ) loops @+ swap cells + 3! 3 loops +! ;
115 :     : BEGIN, ( -- ) ]] BEGIN [[ >loops ;
116 : pazsan 1.23 : DONE, ( -- ) loops @ IF loops> ]] DONE [[ ELSE ." no done left!" cr THEN ;
117 : pazsan 1.1
118 :     \ variables
119 :    
120 :     Variable vars &18 cells allot
121 :     Variable varstack 9 cells allot
122 :     Variable varsmax
123 :     : >var ( -- addr ) vars @+ swap 2* cells +
124 :     vars @ varstack @+ swap cells + !
125 :     1 vars +! 1 varstack +! ;
126 :     : var> ( -- addr ) -1 varstack +!
127 :     varstack @+ swap cells + @
128 :     1+ 2* cells vars + ;
129 : pazsan 1.26 Variable greed-counts 9 cells allot \ no more than 9 nested greedy loops
130 :     : greed' ( -- addr ) greed-counts dup @ + ;
131 : pazsan 1.1
132 :     \ start end
133 :    
134 :     0 Value end$
135 :     0 Value start$
136 :     : !end ( addr u -- addr ) over + to end$ dup to start$ ;
137 : pazsan 1.12 : end-rex? ( addr -- addr flag ) dup end$ u< ;
138 :     : start-rex? ( addr -- addr flag ) dup start$ u> ;
139 : pazsan 1.1 : ?end ( addr -- addr ) ]] dup end$ u> ?LEAVE [[ ; immediate
140 : pazsan 1.20 : rest$ ( addr -- addr addr u ) dup end$ over - ;
141 : pazsan 1.1
142 :     \ start and end
143 :    
144 : pazsan 1.8 : \^ ( addr -- addr ) \ regexp-pattern
145 :     \G check for string start
146 : pazsan 1.12 ]] start-rex? ?LEAVE [[ ; immediate
147 : pazsan 1.8 : \$ ( addr -- addr ) \ regexp-pattern
148 :     \G check for string end
149 : pazsan 1.12 ]] end-rex? ?LEAVE [[ ; immediate
150 : pazsan 1.1
151 : pazsan 1.17 \ A word for string comparison
152 :    
153 : pazsan 1.22 : (str=?) ( addr1 addr u -- addr2 )
154 : pazsan 1.20 dup >r 2>r rest$ r@ umin 2r> compare IF rdrop true ELSE r> + false THEN ;
155 : pazsan 1.22 : str=? ( addr1 addr u -- addr2 ) ]] (str=?) ?LEAVE [[ ; immediate
156 :     : ,=" ( addr u -- ) tuck dup ]] rest$ Literal umin SLiteral compare ?LEAVE Literal + [[ ;
157 : pazsan 1.17 : =" ( <string>" -- ) \ regexp-pattern
158 :     \G check for string
159 :     '" parse ,=" ; immediate
160 :    
161 : pazsan 1.1 \ regexp block
162 :    
163 :     \ FORK/JOIN are like AHEAD THEN, but producing a call on AHEAD
164 :     \ instead of a jump.
165 :    
166 : pazsan 1.8 : (( ( addr u -- ) \ regexp-pattern
167 :     \G start regexp block
168 : pazsan 1.26 vars off varsmax off loops off greed-counts off
169 : pazsan 1.1 ]] FORK AHEAD BUT JOIN !end [[ BEGIN, ; immediate
170 : pazsan 1.8 : )) ( -- addr f ) \ regexp-pattern
171 :     \G end regexp block
172 : pazsan 1.19 ]] ?end drop true ;S [[
173 :     DONE, ]] drop false ;S THEN [[ ; immediate
174 : pazsan 1.1
175 :     \ greedy loops
176 :    
177 :     \ Idea: scan as many characters as possible, try the rest of the pattern
178 :     \ and then back off one pattern at a time
179 :    
180 :     : drops ( n -- ) 1+ cells sp@ + sp! ;
181 :    
182 : pazsan 1.8 : {** ( addr -- addr addr ) \ regexp-pattern
183 :     \G greedy zero-or-more pattern
184 : pazsan 1.26 cell greed-counts +!
185 :     greed' ]] Literal off BEGIN dup [[ BEGIN, ; immediate
186 : pazsan 1.8 ' {** Alias {++ ( addr -- addr addr ) \ regexp-pattern
187 :     \G greedy one-or-more pattern
188 :     immediate
189 :     : n*} ( sys n -- ) \ regexp-pattern
190 :     \G At least @var{n} pattern
191 : pazsan 1.26 >r greed' ]] 1 Literal +! end-rex? 0= UNTIL dup [[ DONE, ]] drop [[
192 :     r@ greed' ]] Literal @ 1+ Literal U+DO FORK BUT [[
193 : pazsan 1.19 ]] IF I' I - [[ r@ 1- ]] Literal + drops true UNLOOP ;S THEN LOOP [[
194 : pazsan 1.1 r@ IF r@ ]] Literal drops [[ THEN
195 : pazsan 1.26 rdrop ]] dup LEAVE JOIN [[
196 :     cell negate greed-counts +! ; immediate
197 : pazsan 1.8 : **} ( sys -- ) \ regexp-pattern
198 :     \G end of greedy zero-or-more pattern
199 :     0 postpone n*} ; immediate
200 :     : ++} ( sys -- ) \ regexp-pattern
201 :     \G end of greedy zero-or-more pattern
202 :     1 postpone n*} ; immediate
203 : pazsan 1.1
204 :     \ non-greedy loops
205 :    
206 :     \ Idea: Try to match rest of the regexp, and if that fails, try match
207 :     \ first expr and then try again rest of regexp.
208 :    
209 : pazsan 1.8 : {+ ( addr -- addr addr ) \ regexp-pattern
210 :     \G non-greedy one-or-more pattern
211 : pazsan 1.1 ]] BEGIN [[ BEGIN, ; immediate
212 : pazsan 1.8 : {* ( addr -- addr addr ) \ regexp-pattern
213 :     \G non-greedy zero-or-more pattern
214 : pazsan 1.19 ]] {+ dup FORK BUT IF drop true ;S THEN [[ ; immediate
215 : pazsan 1.8 : *} ( addr addr' -- addr' ) \ regexp-pattern
216 :     \G end of non-greedy zero-or-more pattern
217 : pazsan 1.1 ]] dup end$ u> UNTIL [[
218 : pazsan 1.19 DONE, ]] drop false ;S JOIN [[ ; immediate
219 : pazsan 1.8 : +} ( addr addr' -- addr' ) \ regexp-pattern
220 :     \G end of non-greedy one-or-more pattern
221 : pazsan 1.19 ]] dup FORK BUT IF drop true ;S [[
222 : pazsan 1.23 DONE, ]] drop dup LEAVE [[ BEGIN, ]] THEN *} [[ ; immediate
223 : pazsan 1.1
224 : pazsan 1.8 : // ( -- ) \ regexp-pattern
225 :     \G search for string
226 :     ]] {* 1+ *} [[ ; immediate
227 : pazsan 1.1
228 :     \ alternatives
229 :    
230 :     \ idea: try to match one alternative and then the rest of regexp.
231 :     \ if that fails, jump back to second alternative
232 :    
233 : pazsan 1.24 : THENs ( sys -- ) BEGIN dup WHILE ]] THEN [[ REPEAT drop ;
234 : pazsan 1.1
235 : pazsan 1.8 : {{ ( addr -- addr addr ) \ regexp-pattern
236 :     \G Start of alternatives
237 : pazsan 1.24 0 ]] dup dup FORK IF 2drop true ;S BUT JOIN [[ vars @ ; immediate
238 : pazsan 1.8 : || ( addr addr -- addr addr ) \ regexp-pattern
239 :     \G separator between alternatives
240 : pazsan 1.24 vars @ varsmax @ max varsmax ! vars !
241 :     ]] AHEAD BUT THEN drop [[
242 :     ]] dup dup FORK IF 2drop true ;S BUT JOIN [[ vars @ ; immediate
243 : pazsan 1.21 : }} ( addr addr -- addr ) \ regexp-pattern
244 : pazsan 1.8 \G end of alternatives
245 : pazsan 1.24 vars @ varsmax @ max vars ! drop
246 : pazsan 1.25 ]] AHEAD BUT THEN drop LEAVE [[ THENs ; immediate
247 : pazsan 1.1
248 :     \ match variables
249 :    
250 : pazsan 1.8 : \( ( addr -- addr ) \ regexp-pattern
251 :     \G start of matching variable; variables are referred as \\1--9
252 :     ]] dup [[
253 : pazsan 1.1 >var ]] ALiteral ! [[ ; immediate
254 : pazsan 1.8 : \) ( addr -- addr ) \ regexp-pattern
255 :     \G end of matching variable
256 :     ]] dup [[
257 : pazsan 1.1 var> ]] ALiteral ! [[ ; immediate
258 : pazsan 1.8 : \0 ( -- addr u ) \ regexp-pattern
259 :     \G the whole string
260 :     start$ end$ over - ;
261 : pazsan 1.1 : \: ( i -- )
262 :     Create 2* 1+ cells vars + ,
263 :     DOES> ( -- addr u ) @ 2@ tuck - ;
264 :     : \:s ( n -- ) 0 ?DO I \: LOOP ;
265 :     9 \:s \1 \2 \3 \4 \5 \6 \7 \8 \9
266 : pazsan 1.6
267 :     \ replacements, needs string.fs
268 :    
269 :     require string.fs
270 :    
271 :     0 Value >>ptr
272 :     0 Value <<ptr
273 :     Variable >>string
274 : pazsan 1.17 : s>> ( addr -- addr ) \ regexp-replace
275 : pazsan 1.8 \G Start replace pattern region
276 :     dup to >>ptr ;
277 :     : << ( run-addr addr u -- run-addr ) \ regexp-replace
278 :     \G Replace string from start of replace pattern region with
279 :     \G @var{addr} @var{u}
280 : pazsan 1.6 <<ptr >>ptr over - >>string $+!
281 :     >>string $+! dup to <<ptr ;
282 : pazsan 1.8 : <<" ( "string<">" -- ) \ regexp-replace
283 :     \G Replace string from start of replace pattern region with
284 :     \G @var{string}
285 :     '" parse postpone SLiteral postpone << ; immediate
286 : pazsan 1.6 : >>string@ ( -- addr u )
287 : pazsan 1.16 >>string $@ ;
288 :     : >>string0 ( addr u -- addr u ) s" " >>string $!
289 :     0 to >>ptr over to <<ptr ;
290 : pazsan 1.6 : >>next ( -- addr u ) <<ptr end$ over - ;
291 : pazsan 1.14 : >>rest ( -- ) >>next >>string $+! ;
292 :     : s// ( addr u -- ptr )
293 : pazsan 1.8 \G start search/replace loop
294 : pazsan 1.17 ]] >>string0 (( // s>> [[ ; immediate
295 :     : >> ( addr -- addr )
296 :     ]] <<ptr >>ptr u> ?LEAVE ?end [[ ; immediate
297 :     : //s ( ptr -- )
298 :     \G search end
299 :     ]] )) drop >>rest >>string@ [[ ; immediate
300 : pazsan 1.15 : //o ( ptr addr u -- addr' u' )
301 : pazsan 1.14 \G end search/replace single loop
302 : pazsan 1.17 ]] << //s [[ ; immediate
303 : pazsan 1.14 : //g ( ptr addr u -- addr' u' )
304 :     \G end search/replace all loop
305 : pazsan 1.17 ]] << LEAVE //s [[ ; immediate

CVS Admin

Powered by ViewCVS 1.0-dev
(Powered by ViewCVS)

ViewCVS and CVS Help