xref: /titanic_52/usr/src/cmd/sgs/lex/common/sub3.c (revision bdfc6d18da790deeec2e0eb09c625902defe2498)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License, Version 1.0 only
6  * (the "License").  You may not use this file except in compliance
7  * with the License.
8  *
9  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10  * or http://www.opensolaris.org/os/licensing.
11  * See the License for the specific language governing permissions
12  * and limitations under the License.
13  *
14  * When distributing Covered Code, include this CDDL HEADER in each
15  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
16  * If applicable, add the following below this CDDL HEADER, with the
17  * fields enclosed by brackets "[]" replaced with your own identifying
18  * information: Portions Copyright [yyyy] [name of copyright owner]
19  *
20  * CDDL HEADER END
21  */
22 /*
23  * Copyright 2005 Sun Microsystems, Inc.
24  * All rights reserved.
25  * Use is subject to license terms.
26  */
27 
28 #pragma ident	"%Z%%M%	%I%	%E% SMI"
29 
30 /*
31  * sub3.c ... ALE enhancement.
32  * Since a typical Asian language has a huge character set, it is not
33  * ideal to index an array by a character code itself, which requires
34  * as large as 2**16 entries per array.
35  * To get arround this problem, we identify a set of characters that
36  * causes the same transition on all states and call it character group.
37  * Every character in a same character group has a unique number called
38  * character group id.  A function yycgid(c) maps the character c (in process
39  * code) to the id.  This mapping is determined by analyzing all regular
40  * expressions in the lex program.
41  *
42  */
43 #include	<stdlib.h>
44 #include	<widec.h>
45 #include	<search.h>
46 #include	"ldefs.c"
47 
48 /*
49  * "lchar" stands for linearized character.  It is a variant of
50  * process code.  AT&T's 16-bit process code has a drawback in which
51  * for three three process code C, D and E where C <= D <= E,
52  * codeset(C)==codeset(E) does not mean codeset(D)==codeset(C).
53  * In other words, four codesets alternates as the magnitude
54  * of character increases.
55  * The lchar representation holds this property:
56  *   If three lchar C', D' and E' have the relationship C' < D' <  E' and
57  *   codeset(C') == codeset(E') then D' is guaranteed to belong to
58  *   the same codeset as C' and E'.
59  * lchar is implemented as 32 bit entities and the function linearize()
60  * that maps a wchar_t to lchar is defined below.  There is no
61  * reverse function for it though.
62  * The 32-bit process code by AT&T, used only for Taiwanese version at the
63  * time of wrting, has no such problem and we use it as it is.
64  */
65 
66 lchar	yycgidtbl[MAXNCG] = {
67 	0, 		/* For ease of computation of the id. */
68 	'\n', 		/* Newline is always special because '.' exclude it. */
69 	0x000000ff, 	/* The upper limit of codeset 0. */
70 	0x20ffffff,	/* The upper limit of codeset 2. */
71 	0x40ffffff	/* The upper limit of codeset 3. */
72 /*	0x60ffffff	   The upper limit of codeset 1. */
73 	/* Above assumes the number of significant bits of wchar_t is <= 24. */
74 };
75 int	ncgidtbl = 5; /* # elements in yycgidtbl. */
76 int	ncg; /* Should set to ncgidtbl*2; this is the largest value yycgid() */
77 		/* returns plus 1. */
78 
79 static void setsymbol(int i);
80 
81 /*
82  * For given 16-bit wchar_t (See NOTE), lchar is computed as illustrated below:
83  *
84  * 	wc: axxxxxxbyyyyyyy
85  *
86  * returns: 0ab0000000000000axxxxxxxbyyyyyyy
87  *
88  * linearize() doesn't do any if compiled with 32-bit wchar_t, use of
89  * which is flagged with LONG_WCHAR_T macro.
90  * NOTE:
91  * The implementation is highly depends on the process code representation.
92  * This function should be modified when 32-bit process code is used.
93  * There is no need to keep 'a' and 'b' bits in the lower half of lchar.
94  * You can actually omit these and squeeze the xxxxxx part one bit right.
95  * We don't do that here just in sake of speed.
96  */
97 lchar
98 linearize(wchar_t wc)
99 {
100 #ifdef LONG_WCHAR_T
101 	return ((lchar)wc); /* Don't do anything. */
102 #else
103 
104 	lchar	prefix;
105 	switch (wc&0x8080) {
106 	case 0x0000: prefix = 0x00000000; break;
107 	case 0x0080: prefix = 0x20000000; break;
108 	case 0x8000: prefix = 0x40000000; break;
109 	case 0x8080: prefix = 0x60000000; break;
110 	}
111 	return (prefix|wc);
112 #endif
113 }
114 
115 /* compare liniear characters pointed to by pc1 and pc2 */
116 int
117 cmplc(const void *arg1, const void *arg2)
118 {
119 	lchar *pc1 = (lchar *)arg1;
120 	lchar *pc2 = (lchar *)arg2;
121 
122 	if (*pc1 > *pc2)
123 		return (1);
124 	else if (*pc1 == *pc2)
125 		return (0);
126 	else
127 		return (-1);
128 }
129 
130 void
131 remch(wchar_t c)
132 {
133 	lchar	lc = linearize(c);
134 
135 	/*
136 	 * User-friendliness consideration:
137 	 * Make sure no EUC chars are used in reg. exp.
138 	 */
139 	if (!handleeuc) {
140 		if (!isascii(c))
141 		    if (iswprint(c))
142 			warning(
143 "Non-ASCII character '%wc' in pattern; use -w or -e lex option.", c);
144 		    else warning(
145 "Non-ASCII character of value %#x in pattern; use -w or -e lex option.", c);
146 		/* In any case, we don't need to construct ncgidtbl[]. */
147 		return;
148 	}
149 
150 	lsearch(&lc, yycgidtbl,
151 		(size_t *)&ncgidtbl, sizeof (lchar), cmplc);
152 }
153 
154 void
155 sortcgidtbl(void)
156 {
157 	if (!handleeuc)
158 		return;
159 	qsort(yycgidtbl, ncgidtbl, sizeof (lchar), cmplc);
160 }
161 
162 /*
163  * int yycgid(wchar_t c)
164  *	Takes c and returns its character group id, determind by the
165  *	following algorithm.  The program also uses the binary search
166  *	algorithm, generalized from Knuth (6.2.1) Algorithm B.
167  *
168  *	This function computes the "character group id" based on
169  *	a table yycgidtbl of which each lchar entry is pre-sorted
170  *	in ascending sequence  The number of valid entries is given
171  *	by YYNCGIDTBL.  There is no duplicate entries in yycgidtbl.
172  *		const int YYNCGIDTBL;
173  *		lchar	yycgidtbl[YYNCGIDTBL];
174  *
175  *	yycgidtbl[0] is guaranteed to have zero.
176  *
177  *	For given c, yycgid(c) returns:
178  *		2*i	iff yycgidtbl[i] == lc
179  *		2*i+1	iff yycgidtbl[i] < lc < yycgidtbl[i+1]
180  *		YYNCGIDTBL*2-1
181  *			iff yycgidtbl[YYNCGIDTBL-1] < lc
182  *	where lc=linearize(c).
183  *
184  *	Some interesting properties.:
185  *	1.  For any c, 0 <= yycgid(c) <= 2*YYNCGIDTBL-1
186  *	2.  yycgid(c) == 0  iff  c == 0.
187  *	3.  For any wchar_t c and d, if linearize(c) < linearize(d) then
188  *	    yycgid(c) <= yycgid(d).
189  *	4.  For any wchar_t c and d, if yycgid(c) < yycgid(d) then
190  *	    linearize(c) < linearize(d).
191  */
192 #define	YYNCGIDTBL ncgidtbl
193 
194 int
195 yycgid(wchar_t c)
196 {
197 	int first = 0;
198 	int last = YYNCGIDTBL - 1;
199 	lchar lc;
200 
201 	/*
202 	 * In ASCII compat. mode, each character forms a "group" and the
203 	 * group-id is itself...
204 	 */
205 	if (!handleeuc)
206 		return (c);
207 
208 	lc = linearize(c);
209 
210 	/* An exceptional case: yycgidtbl[YYNCGIDTBL-1] < lc */
211 	if (yycgidtbl[YYNCGIDTBL - 1] < lc)
212 		return (YYNCGIDTBL*2 - 1);
213 
214 	while (last >= 0) {
215 		int i = (first+last)/2;
216 		if (lc == yycgidtbl[i])
217 		    return (2*i);	/* lc  exactly matches an element. */
218 		else if (yycgidtbl[i] < lc) {
219 			if (lc < yycgidtbl[i+1])
220 			    return (2*i+1); /* lc is in between two elements. */
221 			else
222 			    first = i + 1;
223 		} else
224 		    last = i - 1;
225 	}
226 	error(
227 	"system error in yycgid():binary search failed for c=0x%04x\n", c);
228 	return (0);
229 }
230 
231 /*
232  * repbycgid --- replaces each character in the parsing tree by its
233  * character group id.   This, however, should be called even in
234  * the ASCII compat. mode to process DOT nodes and to call cclinter()
235  * for the DOT and CCL nodes.
236  */
237 void
238 repbycgid(void)
239 {
240 	int i, c;
241 
242 	for (i = 0; i < tptr; ++i) {
243 		c = name[i];
244 		if (!ISOPERATOR(c)) {
245 		/* If not an operator, it must be a char.  */
246 		    name[i] = yycgid((wchar_t)c); /* So replace it. */
247 #ifdef DEBUG
248 		    if (debug) {
249 			    printf("name[%d]:'%c'->%d;\n", i, c, name[i]);
250 		    }
251 #endif
252 		} else if (c == RSTR) {
253 			c = right[i];
254 			right[i] = yycgid((wchar_t)c);
255 #ifdef DEBUG
256 		    if (debug) {
257 			    printf(
258 			    "name[%d].right:'%c'->%d;\n", i, c, right[i]);
259 		    }
260 #endif
261 		} else if ((c == RCCL) || (c == RNCCL)) {
262 			CHR cc, *s;
263 			int j;
264 			CHR ccltoken[CCLSIZE];
265 			CHR *ccp;
266 			int m;
267 			/*
268 			 * This node represetns a character class RE [ccccc]
269 			 * s points to the string of characters that forms
270 			 * the class and/or a special prefix notation
271 			 * <RANGE>XY which corresponds to the RE X-Y,
272 			 * characters in the range of X and Y.  Here,
273 			 * X <= Y is guranteed.
274 			 * We transform these characters into a string
275 			 * of sorted character group ids.
276 			 *
277 			 * There is another mechanism of packing tables
278 			 * that is inherited from the ASCII lex.  Call of
279 			 * cclinter() is required for this packing.
280 			 * This used to be done as yylex() reads the lex
281 			 * rules but we have to do this here because the
282 			 * transition table is made to work on the char-group
283 			 * ids and the mapping cannot be determined until
284 			 * the entire file is read.
285 			 */
286 #ifdef DEBUG
287 			if (debug) {
288 				printf("name[%d]:R[N]CCL of \"", i);
289 				strpt(left[i]);
290 				printf(" -> {");
291 			}
292 #endif
293 			/* Prepare symbol[] for cclinter(). */
294 			for (j = 0; j < ncg; ++j)
295 				symbol[j] = FALSE;
296 
297 			s = (CHR *) left[i];
298 			while (cc = *s++) {
299 				if (cc == RANGE) {
300 					int	low, high, i;
301 					/*
302 					 * Special form: <RANGE>XY
303 					 * This means the range X-Y.
304 					 * We mark all symbols[]
305 					 * elements for yycgid(X) thru
306 					 * yycgid(Y), inclusively.
307 					 */
308 					low = yycgid(*s++);
309 					high = yycgid(*s++);
310 					for (i = low; i <= high; ++i)
311 					    setsymbol(i);
312 				} else {
313 					setsymbol(yycgid(cc));
314 				}
315 			}
316 
317 			/* Now make a transformed string of cgids. */
318 			s = ccptr;
319 			m = 0;
320 			for (j = 0; j < ncg; ++j)
321 			    if (symbol[j]) {
322 				    ccltoken[m++] = (CHR)j;
323 #ifdef DEBUG
324 				    if (debug) printf("%d, ", j);
325 #endif
326 			    }
327 
328 #ifdef DEBUG
329 			if (debug) printf("}\n");
330 #endif
331 			ccltoken[m] = 0;
332 			ccp = ccl;
333 			while (ccp < ccptr && scomp(ccltoken, ccp) != 0)
334 				ccp++;
335 			if (ccp < ccptr) {  /* character class found in ccl */
336 			    left[i] = (int)ccp;
337 			} else { /* not in ccl, add it */
338 			    left[i] = (int)ccptr;
339 			    scopy(ccltoken, ccptr);
340 			    ccptr += slength(ccltoken) + 1;
341 			    if (ccptr > ccl + CCLSIZE)
342 				error("Too many large character classes");
343 			}
344 			cclinter(c == RCCL);
345 		} else if (c == DOT) {
346 			if (psave == 0) { /* First DOT node. */
347 				int j, nlid;
348 				/*
349 				 * Make symbol[k]=TRUE for all k
350 				 *  except k == yycgid('\n').
351 				 */
352 				nlid = yycgid('\n');
353 				psave = ccptr;
354 				for (j = 1; j < ncg; ++j) {
355 					if (j == nlid) {
356 						symbol[j] = FALSE;
357 					} else {
358 						symbol[j] = TRUE;
359 						*ccptr++ = (CHR) j;
360 					}
361 				}
362 				*ccptr++ = 0;
363 				if (ccptr > ccl + CCLSIZE)
364 				    error("Too many large character classes");
365 			}
366 			/* Mimic mn1(RCCL,psave)... */
367 			name[i] = RCCL;
368 			left[i] = (int)psave;
369 			cclinter(1);
370 		}
371 	}
372 #ifdef DEBUG
373 	if (debug) {
374 		printf("treedump after repbycgid().\n");
375 		treedump();
376 	}
377 #endif
378 }
379 
380 static void
381 setsymbol(int i)
382 {
383 	if (i > sizeof (symbol))
384 		error("setsymbol: (SYSERR) %d out of range", i);
385 	symbol[i] = TRUE;
386 }
387