xref: /titanic_50/usr/src/cmd/sgs/lex/common/sub3.c (revision d2ec54f7875f7e05edd56195adbeb593c947763f)
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 (the "License").
6  * You may not use this file except in compliance with the License.
7  *
8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9  * or http://www.opensolaris.org/os/licensing.
10  * See the License for the specific language governing permissions
11  * and limitations under the License.
12  *
13  * When distributing Covered Code, include this CDDL HEADER in each
14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15  * If applicable, add the following below this CDDL HEADER, with the
16  * fields enclosed by brackets "[]" replaced with your own identifying
17  * information: Portions Copyright [yyyy] [name of copyright owner]
18  *
19  * CDDL HEADER END
20  */
21 /*
22  * Copyright 2007 Sun Microsystems, Inc.  All rights reserved.
23  * Use is subject to license terms.
24  */
25 
26 #pragma ident	"%Z%%M%	%I%	%E% SMI"
27 
28 /*
29  * sub3.c ... ALE enhancement.
30  * Since a typical Asian language has a huge character set, it is not
31  * ideal to index an array by a character code itself, which requires
32  * as large as 2**16 entries per array.
33  * To get arround this problem, we identify a set of characters that
34  * causes the same transition on all states and call it character group.
35  * Every character in a same character group has a unique number called
36  * character group id.  A function yycgid(c) maps the character c (in process
37  * code) to the id.  This mapping is determined by analyzing all regular
38  * expressions in the lex program.
39  *
40  */
41 #include	<stdlib.h>
42 #include	<widec.h>
43 #include	<search.h>
44 #include	"ldefs.h"
45 
46 /*
47  * "lchar" stands for linearized character.  It is a variant of
48  * process code.  AT&T's 16-bit process code has a drawback in which
49  * for three three process code C, D and E where C <= D <= E,
50  * codeset(C)==codeset(E) does not mean codeset(D)==codeset(C).
51  * In other words, four codesets alternates as the magnitude
52  * of character increases.
53  * The lchar representation holds this property:
54  *   If three lchar C', D' and E' have the relationship C' < D' <  E' and
55  *   codeset(C') == codeset(E') then D' is guaranteed to belong to
56  *   the same codeset as C' and E'.
57  * lchar is implemented as 32 bit entities and the function linearize()
58  * that maps a wchar_t to lchar is defined below.  There is no
59  * reverse function for it though.
60  * The 32-bit process code by AT&T, used only for Taiwanese version at the
61  * time of wrting, has no such problem and we use it as it is.
62  */
63 
64 lchar	yycgidtbl[MAXNCG] = {
65 	0, 		/* For ease of computation of the id. */
66 	'\n', 		/* Newline is always special because '.' exclude it. */
67 	0x000000ff, 	/* The upper limit of codeset 0. */
68 	0x20ffffff,	/* The upper limit of codeset 2. */
69 	0x40ffffff	/* The upper limit of codeset 3. */
70 /*	0x60ffffff	   The upper limit of codeset 1. */
71 	/* Above assumes the number of significant bits of wchar_t is <= 24. */
72 };
73 int	ncgidtbl = 5; /* # elements in yycgidtbl. */
74 int	ncg; /* Should set to ncgidtbl*2; this is the largest value yycgid() */
75 		/* returns plus 1. */
76 
77 static void setsymbol(int i);
78 
79 /*
80  * For given 16-bit wchar_t (See NOTE), lchar is computed as illustrated below:
81  *
82  * 	wc: axxxxxxbyyyyyyy
83  *
84  * returns: 0ab0000000000000axxxxxxxbyyyyyyy
85  *
86  * linearize() doesn't do any if compiled with 32-bit wchar_t, use of
87  * which is flagged with LONG_WCHAR_T macro.
88  * NOTE:
89  * The implementation is highly depends on the process code representation.
90  * This function should be modified when 32-bit process code is used.
91  * There is no need to keep 'a' and 'b' bits in the lower half of lchar.
92  * You can actually omit these and squeeze the xxxxxx part one bit right.
93  * We don't do that here just in sake of speed.
94  */
95 lchar
96 linearize(wchar_t wc)
97 {
98 #ifdef LONG_WCHAR_T
99 	return ((lchar)wc); /* Don't do anything. */
100 #else
101 
102 	lchar	prefix;
103 	switch (wc&0x8080) {
104 	case 0x0000: prefix = 0x00000000; break;
105 	case 0x0080: prefix = 0x20000000; break;
106 	case 0x8000: prefix = 0x40000000; break;
107 	case 0x8080: prefix = 0x60000000; break;
108 	}
109 	return (prefix|wc);
110 #endif
111 }
112 
113 /* compare liniear characters pointed to by pc1 and pc2 */
114 int
115 cmplc(const void *arg1, const void *arg2)
116 {
117 	lchar *pc1 = (lchar *)arg1;
118 	lchar *pc2 = (lchar *)arg2;
119 
120 	if (*pc1 > *pc2)
121 		return (1);
122 	else if (*pc1 == *pc2)
123 		return (0);
124 	else
125 		return (-1);
126 }
127 
128 void
129 remch(wchar_t c)
130 {
131 	lchar	lc = linearize(c);
132 
133 	/*
134 	 * User-friendliness consideration:
135 	 * Make sure no EUC chars are used in reg. exp.
136 	 */
137 	if (!handleeuc) {
138 		if (!isascii(c))
139 			if (iswprint(c))
140 				warning(
141 "Non-ASCII character '%wc' in pattern; use -w or -e lex option.", c);
142 			else warning(
143 "Non-ASCII character of value %#x in pattern; use -w or -e lex option.", c);
144 		/* In any case, we don't need to construct ncgidtbl[]. */
145 		return;
146 	}
147 
148 	lsearch(&lc, yycgidtbl,
149 	    (size_t *)&ncgidtbl, sizeof (lchar), cmplc);
150 }
151 
152 void
153 sortcgidtbl(void)
154 {
155 	if (!handleeuc)
156 		return;
157 	qsort(yycgidtbl, ncgidtbl, sizeof (lchar), cmplc);
158 }
159 
160 /*
161  * int yycgid(wchar_t c)
162  *	Takes c and returns its character group id, determind by the
163  *	following algorithm.  The program also uses the binary search
164  *	algorithm, generalized from Knuth (6.2.1) Algorithm B.
165  *
166  *	This function computes the "character group id" based on
167  *	a table yycgidtbl of which each lchar entry is pre-sorted
168  *	in ascending sequence  The number of valid entries is given
169  *	by YYNCGIDTBL.  There is no duplicate entries in yycgidtbl.
170  *		const int YYNCGIDTBL;
171  *		lchar	yycgidtbl[YYNCGIDTBL];
172  *
173  *	yycgidtbl[0] is guaranteed to have zero.
174  *
175  *	For given c, yycgid(c) returns:
176  *		2*i	iff yycgidtbl[i] == lc
177  *		2*i+1	iff yycgidtbl[i] < lc < yycgidtbl[i+1]
178  *		YYNCGIDTBL*2-1
179  *			iff yycgidtbl[YYNCGIDTBL-1] < lc
180  *	where lc=linearize(c).
181  *
182  *	Some interesting properties.:
183  *	1.  For any c, 0 <= yycgid(c) <= 2*YYNCGIDTBL-1
184  *	2.  yycgid(c) == 0  iff  c == 0.
185  *	3.  For any wchar_t c and d, if linearize(c) < linearize(d) then
186  *	    yycgid(c) <= yycgid(d).
187  *	4.  For any wchar_t c and d, if yycgid(c) < yycgid(d) then
188  *	    linearize(c) < linearize(d).
189  */
190 #define	YYNCGIDTBL ncgidtbl
191 
192 int
193 yycgid(wchar_t c)
194 {
195 	int first = 0;
196 	int last = YYNCGIDTBL - 1;
197 	lchar lc;
198 
199 	/*
200 	 * In ASCII compat. mode, each character forms a "group" and the
201 	 * group-id is itself...
202 	 */
203 	if (!handleeuc)
204 		return (c);
205 
206 	lc = linearize(c);
207 
208 	/* An exceptional case: yycgidtbl[YYNCGIDTBL-1] < lc */
209 	if (yycgidtbl[YYNCGIDTBL - 1] < lc)
210 		return (YYNCGIDTBL*2 - 1);
211 
212 	while (last >= 0) {
213 		int i = (first+last)/2;
214 		if (lc == yycgidtbl[i])
215 			return (2*i);	/* lc exactly matches an element. */
216 		else if (yycgidtbl[i] < lc) {
217 			if (lc < yycgidtbl[i+1]) {
218 				/* lc is in between two elements */
219 				return (2*i+1);
220 			}
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",
259 				    i, c, right[i]);
260 			}
261 #endif
262 		} else if ((c == RCCL) || (c == RNCCL)) {
263 			CHR cc, *s;
264 			int j;
265 			CHR ccltoken[CCLSIZE];
266 			CHR *ccp;
267 			int m;
268 			/*
269 			 * This node represetns a character class RE [ccccc]
270 			 * s points to the string of characters that forms
271 			 * the class and/or a special prefix notation
272 			 * <RANGE>XY which corresponds to the RE X-Y,
273 			 * characters in the range of X and Y.  Here,
274 			 * X <= Y is guranteed.
275 			 * We transform these characters into a string
276 			 * of sorted character group ids.
277 			 *
278 			 * There is another mechanism of packing tables
279 			 * that is inherited from the ASCII lex.  Call of
280 			 * cclinter() is required for this packing.
281 			 * This used to be done as yylex() reads the lex
282 			 * rules but we have to do this here because the
283 			 * transition table is made to work on the char-group
284 			 * ids and the mapping cannot be determined until
285 			 * the entire file is read.
286 			 */
287 #ifdef DEBUG
288 			if (debug) {
289 				printf("name[%d]:R[N]CCL of \"", i);
290 				strpt(left[i]);
291 				printf(" -> {");
292 			}
293 #endif
294 			/* Prepare symbol[] for cclinter(). */
295 			for (j = 0; j < ncg; ++j)
296 				symbol[j] = FALSE;
297 
298 			s = (CHR *) left[i];
299 			while (cc = *s++) {
300 				if (cc == RANGE) {
301 					int	low, high, i;
302 					/*
303 					 * Special form: <RANGE>XY
304 					 * This means the range X-Y.
305 					 * We mark all symbols[]
306 					 * elements for yycgid(X) thru
307 					 * yycgid(Y), inclusively.
308 					 */
309 					low = yycgid(*s++);
310 					high = yycgid(*s++);
311 					for (i = low; i <= high; ++i)
312 						setsymbol(i);
313 				} else {
314 					setsymbol(yycgid(cc));
315 				}
316 			}
317 
318 			/* Now make a transformed string of cgids. */
319 			s = ccptr;
320 			m = 0;
321 			for (j = 0; j < ncg; ++j)
322 				if (symbol[j]) {
323 					ccltoken[m++] = (CHR)j;
324 #ifdef DEBUG
325 					if (debug) printf("%d, ", j);
326 #endif
327 				}
328 
329 #ifdef DEBUG
330 			if (debug) printf("}\n");
331 #endif
332 			ccltoken[m] = 0;
333 			ccp = ccl;
334 			while (ccp < ccptr && scomp(ccltoken, ccp) != 0)
335 				ccp++;
336 			if (ccp < ccptr) {  /* character class found in ccl */
337 				left[i] = (int)ccp;
338 			} else { /* not in ccl, add it */
339 				left[i] = (int)ccptr;
340 				scopy(ccltoken, ccptr);
341 				ccptr += slength(ccltoken) + 1;
342 				if (ccptr > ccl + CCLSIZE)
343 					error(
344 					"Too many large character classes");
345 			}
346 			cclinter(c == RCCL);
347 		} else if (c == DOT) {
348 			if (psave == 0) { /* First DOT node. */
349 				int j, nlid;
350 				/*
351 				 * Make symbol[k]=TRUE for all k
352 				 *  except k == yycgid('\n').
353 				 */
354 				nlid = yycgid('\n');
355 				psave = ccptr;
356 				for (j = 1; j < ncg; ++j) {
357 					if (j == nlid) {
358 						symbol[j] = FALSE;
359 					} else {
360 						symbol[j] = TRUE;
361 						*ccptr++ = (CHR) j;
362 					}
363 				}
364 				*ccptr++ = 0;
365 				if (ccptr > ccl + CCLSIZE)
366 					error(
367 					"Too many large character classes");
368 			}
369 			/* Mimic mn1(RCCL,psave)... */
370 			name[i] = RCCL;
371 			left[i] = (int)psave;
372 			cclinter(1);
373 		}
374 	}
375 #ifdef DEBUG
376 	if (debug) {
377 		printf("treedump after repbycgid().\n");
378 		treedump();
379 	}
380 #endif
381 }
382 
383 static void
384 setsymbol(int i)
385 {
386 	if (i > sizeof (symbol))
387 		error("setsymbol: (SYSERR) %d out of range", i);
388 	symbol[i] = TRUE;
389 }
390