]> diplodocus.org Git - nmh/blob - sbr/fmt_compile.c
Just reworded the bit about '%s' being safe not to quote (it's only safe not to
[nmh] / sbr / fmt_compile.c
1
2 /*
3 * fmt_compile.c -- "compile" format strings for fmt_scan
4 *
5 * $Id$
6 */
7
8 #include <h/mh.h>
9 #include <h/addrsbr.h>
10 #include <zotnet/tws/tws.h>
11 #include <h/fmt_scan.h>
12 #include <h/fmt_compile.h>
13
14 #ifdef TIME_WITH_SYS_TIME
15 # include <sys/time.h>
16 # include <time.h>
17 #else
18 # ifdef TM_IN_SYS_TIME
19 # include <sys/time.h>
20 # else
21 # include <time.h>
22 # endif
23 #endif
24
25 /*
26 * hash table for deciding if a component is "interesting"
27 */
28 struct comp *wantcomp[128];
29
30 static struct format *formatvec; /* array to hold formats */
31 static struct format *next_fp; /* next free format slot */
32 static struct format *fp; /* current format slot */
33 static struct comp *cm; /* most recent comp ref */
34 static struct ftable *ftbl; /* most recent func ref */
35 static int ncomp;
36 static int infunction; /* function nesting cnt */
37
38 extern struct mailname fmt_mnull;
39
40 /* ftable->type (argument type) */
41 #define TF_COMP 0 /* component expected */
42 #define TF_NUM 1 /* number expected */
43 #define TF_STR 2 /* string expected */
44 #define TF_EXPR 3 /* component or func. expected */
45 #define TF_NONE 4 /* no argument */
46 #define TF_MYBOX 5 /* special - get current user's mbox */
47 #define TF_NOW 6 /* special - get current unix time */
48 #define TF_EXPR_SV 7 /* like expr but save current str reg */
49 #define TF_NOP 8 /* like expr but no result */
50
51 /* ftable->flags */
52 #define TFL_PUTS 1 /* implicit putstr if top level */
53 #define TFL_PUTN 2 /* implicit putnum if top level */
54
55 struct ftable {
56 char *name; /* function name */
57 char type; /* argument type */
58 char f_type; /* fmt type */
59 char extra; /* arg. type dependent extra info */
60 char flags;
61 };
62
63 static struct ftable functable[] = {
64 { "nonzero", TF_EXPR, FT_V_NE, FT_IF_V_NE, 0 },
65 { "zero", TF_EXPR, FT_V_EQ, FT_IF_V_EQ, 0 },
66 { "eq", TF_NUM, FT_V_EQ, FT_IF_V_EQ, 0 },
67 { "ne", TF_NUM, FT_V_NE, FT_IF_V_NE, 0 },
68 { "gt", TF_NUM, FT_V_GT, FT_IF_V_GT, 0 },
69 { "null", TF_EXPR, FT_S_NULL, FT_IF_S_NULL, 0 },
70 { "nonnull", TF_EXPR, FT_S_NONNULL, FT_IF_S, 0 },
71 { "match", TF_STR, FT_V_MATCH, FT_IF_MATCH, 0 },
72 { "amatch", TF_STR, FT_V_AMATCH, FT_IF_AMATCH, 0 },
73
74 { "putstr", TF_EXPR, FT_STR, 0, 0 },
75 { "putstrf", TF_EXPR, FT_STRF, 0, 0 },
76 { "putnum", TF_EXPR, FT_NUM, 0, 0 },
77 { "putnumf", TF_EXPR, FT_NUMF, 0, 0 },
78 { "putaddr", TF_STR, FT_PUTADDR, 0, 0 },
79 { "void", TF_NOP, 0, 0, 0 },
80
81 { "comp", TF_COMP, FT_LS_COMP, 0, TFL_PUTS },
82 { "lit", TF_STR, FT_LS_LIT, 0, TFL_PUTS },
83 { "getenv", TF_STR, FT_LS_GETENV, 0, TFL_PUTS },
84 { "profile", TF_STR, FT_LS_CFIND, 0, TFL_PUTS },
85 { "decodecomp", TF_COMP, FT_LS_DECODECOMP, 0, TFL_PUTS },
86 { "decode", TF_EXPR, FT_LS_DECODE, 0, TFL_PUTS },
87 { "trim", TF_EXPR, FT_LS_TRIM, 0, 0 },
88 { "compval", TF_COMP, FT_LV_COMP, 0, TFL_PUTN },
89 { "compflag", TF_COMP, FT_LV_COMPFLAG, 0, TFL_PUTN },
90 { "num", TF_NUM, FT_LV_LIT, 0, TFL_PUTN },
91 { "msg", TF_NONE, FT_LV_DAT, 0, TFL_PUTN },
92 { "cur", TF_NONE, FT_LV_DAT, 1, TFL_PUTN },
93 { "size", TF_NONE, FT_LV_DAT, 2, TFL_PUTN },
94 { "width", TF_NONE, FT_LV_DAT, 3, TFL_PUTN },
95 { "unseen", TF_NONE, FT_LV_DAT, 4, TFL_PUTN },
96 { "dat", TF_NUM, FT_LV_DAT, 0, TFL_PUTN },
97 { "strlen", TF_NONE, FT_LV_STRLEN, 0, TFL_PUTN },
98 { "me", TF_MYBOX, FT_LS_LIT, 0, TFL_PUTS },
99 { "plus", TF_NUM, FT_LV_PLUS_L, 0, TFL_PUTN },
100 { "minus", TF_NUM, FT_LV_MINUS_L, 0, TFL_PUTN },
101 { "divide", TF_NUM, FT_LV_DIVIDE_L, 0, TFL_PUTN },
102 { "modulo", TF_NUM, FT_LV_MODULO_L, 0, TFL_PUTN },
103 { "charleft", TF_NONE, FT_LV_CHAR_LEFT, 0, TFL_PUTN },
104 { "timenow", TF_NOW, FT_LV_LIT, 0, TFL_PUTN },
105
106 { "month", TF_COMP, FT_LS_MONTH, FT_PARSEDATE, TFL_PUTS },
107 { "lmonth", TF_COMP, FT_LS_LMONTH, FT_PARSEDATE, TFL_PUTS },
108 { "tzone", TF_COMP, FT_LS_ZONE, FT_PARSEDATE, TFL_PUTS },
109 { "day", TF_COMP, FT_LS_DAY, FT_PARSEDATE, TFL_PUTS },
110 { "weekday", TF_COMP, FT_LS_WEEKDAY, FT_PARSEDATE, TFL_PUTS },
111 { "tws", TF_COMP, FT_LS_822DATE, FT_PARSEDATE, TFL_PUTS },
112 { "sec", TF_COMP, FT_LV_SEC, FT_PARSEDATE, TFL_PUTN },
113 { "min", TF_COMP, FT_LV_MIN, FT_PARSEDATE, TFL_PUTN },
114 { "hour", TF_COMP, FT_LV_HOUR, FT_PARSEDATE, TFL_PUTN },
115 { "mday", TF_COMP, FT_LV_MDAY, FT_PARSEDATE, TFL_PUTN },
116 { "mon", TF_COMP, FT_LV_MON, FT_PARSEDATE, TFL_PUTN },
117 { "year", TF_COMP, FT_LV_YEAR, FT_PARSEDATE, TFL_PUTN },
118 { "yday", TF_COMP, FT_LV_YDAY, FT_PARSEDATE, TFL_PUTN },
119 { "wday", TF_COMP, FT_LV_WDAY, FT_PARSEDATE, TFL_PUTN },
120 { "zone", TF_COMP, FT_LV_ZONE, FT_PARSEDATE, TFL_PUTN },
121 { "clock", TF_COMP, FT_LV_CLOCK, FT_PARSEDATE, TFL_PUTN },
122 { "rclock", TF_COMP, FT_LV_RCLOCK, FT_PARSEDATE, TFL_PUTN },
123 { "sday", TF_COMP, FT_LV_DAYF, FT_PARSEDATE, TFL_PUTN },
124 { "szone", TF_COMP, FT_LV_ZONEF, FT_PARSEDATE, TFL_PUTN },
125 { "dst", TF_COMP, FT_LV_DST, FT_PARSEDATE, TFL_PUTN },
126 { "pretty", TF_COMP, FT_LS_PRETTY, FT_PARSEDATE, TFL_PUTS },
127 { "nodate", TF_COMP, FT_LV_COMPFLAG, FT_PARSEDATE, TFL_PUTN },
128 { "date2local", TF_COMP, FT_LOCALDATE, FT_PARSEDATE, 0 },
129 { "date2gmt", TF_COMP, FT_GMTDATE, FT_PARSEDATE, 0 },
130
131 { "pers", TF_COMP, FT_LS_PERS, FT_PARSEADDR, TFL_PUTS },
132 { "mbox", TF_COMP, FT_LS_MBOX, FT_PARSEADDR, TFL_PUTS },
133 { "host", TF_COMP, FT_LS_HOST, FT_PARSEADDR, TFL_PUTS },
134 { "path", TF_COMP, FT_LS_PATH, FT_PARSEADDR, TFL_PUTS },
135 { "gname", TF_COMP, FT_LS_GNAME, FT_PARSEADDR, TFL_PUTS },
136 { "note", TF_COMP, FT_LS_NOTE, FT_PARSEADDR, TFL_PUTS },
137 { "addr", TF_COMP, FT_LS_ADDR, FT_PARSEADDR, TFL_PUTS },
138 { "proper", TF_COMP, FT_LS_822ADDR, FT_PARSEADDR, TFL_PUTS },
139 { "type", TF_COMP, FT_LV_HOSTTYPE, FT_PARSEADDR, TFL_PUTN },
140 { "ingrp", TF_COMP, FT_LV_INGRPF, FT_PARSEADDR, TFL_PUTN },
141 { "nohost", TF_COMP, FT_LV_NOHOSTF, FT_PARSEADDR, TFL_PUTN },
142 { "formataddr", TF_EXPR_SV,FT_FORMATADDR, FT_FORMATADDR, 0 },
143 { "friendly", TF_COMP, FT_LS_FRIENDLY, FT_PARSEADDR, TFL_PUTS },
144
145 { "mymbox", TF_COMP, FT_LV_COMPFLAG, FT_MYMBOX, TFL_PUTN },
146 { "addtoseq", TF_STR, FT_ADDTOSEQ, 0, 0 },
147
148 { NULL, 0, 0, 0, 0 }
149 };
150
151 /* Add new component to the hash table */
152 #define NEWCOMP(cm,name)\
153 cm = ((struct comp *) calloc(1, sizeof (struct comp)));\
154 cm->c_name = name;\
155 ncomp++;\
156 i = CHASH(name);\
157 cm->c_next = wantcomp[i];\
158 wantcomp[i] = cm;
159
160 #define NEWFMT (next_fp++)
161 #define NEW(type,fill,wid)\
162 fp=NEWFMT; fp->f_type=(type); fp->f_fill=(fill); fp->f_width=(wid);
163
164 /* Add (possibly new) component to the hash table */
165 #define ADDC(name)\
166 FINDCOMP(cm, name);\
167 if (!cm) {\
168 NEWCOMP(cm,name);\
169 }\
170 fp->f_comp = cm;
171
172 #define LV(type, value) NEW(type,0,0); fp->f_value = (value);
173 #define LS(type, str) NEW(type,0,0); fp->f_text = (str);
174
175 #define PUTCOMP(comp) NEW(FT_COMP,0,0); ADDC(comp);
176 #define PUTLIT(str) NEW(FT_LIT,0,0); fp->f_text = (str);
177 #define PUTC(c) NEW(FT_CHAR,0,0); fp->f_char = (c);
178
179 static char *format_string;
180 static char *usr_fstring; /* for CERROR */
181
182 #define CERROR(str) compile_error (str, cp)
183
184 /*
185 * external prototypes
186 */
187 extern char *getusername(void);
188
189 /*
190 * static prototypes
191 */
192 static struct ftable *lookup(char *);
193 static void compile_error(char *, char *);
194 static char *compile (char *);
195 static char *do_spec(char *);
196 static char *do_name(char *, int);
197 static char *do_func(char *);
198 static char *do_expr (char *, int);
199 static char *do_loop(char *);
200 static char *do_if(char *);
201
202
203 static struct ftable *
204 lookup(char *name)
205 {
206 register struct ftable *t = functable;
207 register char *nm;
208 register char c = *name;
209
210 while ((nm = t->name)) {
211 if (*nm == c && strcmp (nm, name) == 0)
212 return (ftbl = t);
213
214 t++;
215 }
216 return (struct ftable *) 0;
217 }
218
219
220 static void
221 compile_error(char *str, char *cp)
222 {
223 int i, errpos, errctx;
224
225 errpos = cp - format_string;
226 errctx = errpos > 20 ? 20 : errpos;
227 usr_fstring[errpos] = '\0';
228
229 for (i = errpos-errctx; i < errpos; i++) {
230 #ifdef LOCALE
231 if (iscntrl(usr_fstring[i]))
232 #else
233 if (usr_fstring[i] < 32)
234 #endif
235 usr_fstring[i] = '_';
236 }
237
238 advise(NULL, "\"%s\": format compile error - %s",
239 &usr_fstring[errpos-errctx], str);
240 adios (NULL, "%*s", errctx+1, "^");
241 }
242
243 /*
244 * Compile format string "fstring" into format list "fmt".
245 * Return the number of header components found in the format
246 * string.
247 */
248
249 int
250 fmt_compile(char *fstring, struct format **fmt)
251 {
252 register char *cp;
253 int i;
254
255 if (format_string)
256 free (format_string);
257 format_string = getcpy (fstring);
258 usr_fstring = fstring;
259
260 /* init the component hash table. */
261 for (i = 0; i < sizeof(wantcomp)/sizeof(wantcomp[0]); i++)
262 wantcomp[i] = 0;
263
264 memset((char *) &fmt_mnull, 0, sizeof(fmt_mnull));
265
266 /* it takes at least 4 char to generate one format so we
267 * allocate a worst-case format array using 1/4 the length
268 * of the format string. We actually need twice this much
269 * to handle both pre-processing (e.g., address parsing) and
270 * normal processing.
271 */
272 i = strlen(fstring)/2 + 1;
273 if (i==1) i++;
274 next_fp = formatvec = (struct format *)calloc ((size_t) i,
275 sizeof(struct format));
276 if (next_fp == NULL)
277 adios (NULL, "unable to allocate format storage");
278
279 ncomp = 0;
280 infunction = 0;
281
282 cp = compile(format_string);
283 if (*cp) {
284 CERROR("extra '%>', '%|' or '%?'");
285 }
286 LV(FT_DONE, 0); /* really done */
287 *fmt = formatvec;
288
289 return (ncomp);
290 }
291
292 static char *
293 compile (char *sp)
294 {
295 register char *cp = sp;
296 register int c;
297
298 for (;;) {
299 sp = cp;
300 while ((c = *cp) && c != '%')
301 cp++;
302 *cp = 0;
303 switch (cp-sp) {
304 case 0:
305 break;
306 case 1:
307 PUTC(*sp);
308 break;
309 default:
310 PUTLIT(sp);
311 break;
312 }
313 if (c == 0)
314 return (cp);
315
316 switch (c = *++cp) {
317 case '%':
318 PUTC (*cp);
319 cp++;
320 break;
321
322 case '|':
323 case '>':
324 case '?':
325 case ']':
326 return (cp);
327
328 case '<':
329 cp = do_if(++cp);
330 break;
331
332 case '[': /* ] */
333 cp = do_loop(++cp);
334 break;
335
336 case ';': /* comment line */
337 cp++;
338 while ((c = *cp++) && c != '\n')
339 continue;
340 break;
341
342 default:
343 cp = do_spec(cp);
344 break;
345 }
346 }
347 }
348
349
350 static char *
351 do_spec(char *sp)
352 {
353 register char *cp = sp;
354 register int c;
355 #ifndef lint
356 register int ljust = 0;
357 #endif /* not lint */
358 register int wid = 0;
359 register char fill = ' ';
360
361 c = *cp++;
362 if (c == '-') {
363 ljust++;
364 c = *cp++;
365 }
366 if (c == '0') {
367 fill = c;
368 c = *cp++;
369 }
370 while (isdigit(c)) {
371 wid = wid*10 + (c - '0');
372 c = *cp++;
373 }
374 if (c == '{') {
375 cp = do_name(cp, 0);
376 if (! infunction)
377 fp->f_type = wid? FT_COMPF : FT_COMP;
378 }
379 else if (c == '(') {
380 cp = do_func(cp);
381 if (! infunction) {
382 if (ftbl->flags & TFL_PUTS) {
383 LV( wid? FT_STRF : FT_STR, ftbl->extra);
384 }
385 else if (ftbl->flags & TFL_PUTN) {
386 LV( wid? FT_NUMF : FT_NUM, ftbl->extra);
387 }
388 }
389 }
390 else {
391 CERROR("component or function name expected");
392 }
393 if (ljust)
394 wid = -wid;
395 fp->f_width = wid;
396 fp->f_fill = fill;
397
398 return (cp);
399 }
400
401 static char *
402 do_name(char *sp, int preprocess)
403 {
404 register char *cp = sp;
405 register int c;
406 register int i;
407 static int primed = 0;
408
409 while (isalnum(c = *cp++) || c == '-' || c == '_')
410 ;
411 if (c != '}') {
412 CERROR("'}' expected");
413 }
414 cp[-1] = '\0';
415 PUTCOMP(sp);
416 switch (preprocess) {
417
418 case FT_PARSEDATE:
419 if (cm->c_type & CT_ADDR) {
420 CERROR("component used as both date and address");
421 }
422 if (! (cm->c_type & CT_DATE)) {
423 cm->c_tws = (struct tws *)
424 calloc((size_t) 1, sizeof(*cm->c_tws));
425 fp->f_type = preprocess;
426 PUTCOMP(sp);
427 cm->c_type |= CT_DATE;
428 }
429 break;
430
431 case FT_MYMBOX:
432 if (!primed) {
433 ismymbox ((struct mailname *) 0);
434 primed++;
435 }
436 cm->c_type |= CT_MYMBOX;
437 /* fall through */
438 case FT_PARSEADDR:
439 if (cm->c_type & CT_DATE) {
440 CERROR("component used as both date and address");
441 }
442 if (! (cm->c_type & CT_ADDRPARSE)) {
443 cm->c_mn = &fmt_mnull;
444 fp->f_type = preprocess;
445 PUTCOMP(sp);
446 cm->c_type |= (CT_ADDR | CT_ADDRPARSE);
447 }
448 break;
449
450 case FT_FORMATADDR:
451 if (cm->c_type & CT_DATE) {
452 CERROR("component used as both date and address");
453 }
454 cm->c_type |= CT_ADDR;
455 break;
456 }
457 return (cp);
458 }
459
460 static char *
461 do_func(char *sp)
462 {
463 register char *cp = sp;
464 register int c;
465 register struct ftable *t;
466 register int n;
467 int mflag; /* minus sign in NUM */
468
469 infunction++;
470
471 while (isalnum(c = *cp++))
472 ;
473 if (c != '(' && c != '{' && c != ' ' && c != ')') {
474 CERROR("'(', '{', ' ' or ')' expected");
475 }
476 cp[-1] = '\0';
477 if ((t = lookup (sp)) == 0) {
478 CERROR("unknown function");
479 }
480 if (isspace(c))
481 c = *cp++;
482
483 switch (t->type) {
484
485 case TF_COMP:
486 if (c != '{') {
487 CERROR("component name expected");
488 }
489 cp = do_name(cp, t->extra);
490 fp->f_type = t->f_type;
491 c = *cp++;
492 break;
493
494 case TF_NUM:
495 if ((mflag = (c == '-')))
496 c = *cp++;
497 n = 0;
498 while (isdigit(c)) {
499 n = n*10 + (c - '0');
500 c = *cp++;
501 }
502 if (mflag)
503 n = (-n);
504 LV(t->f_type,n);
505 break;
506
507 case TF_STR:
508 sp = cp - 1;
509 while (c && c != ')')
510 c = *cp++;
511 cp[-1] = '\0';
512 LS(t->f_type,sp);
513 break;
514
515 case TF_NONE:
516 LV(t->f_type,t->extra);
517 break;
518
519 case TF_MYBOX:
520 LS(t->f_type, getusername());
521 break;
522
523 case TF_NOW:
524 LV(t->f_type, time((time_t *) 0));
525 break;
526
527 case TF_EXPR_SV:
528 LV(FT_SAVESTR, 0);
529 /* fall through */
530 case TF_EXPR:
531 *--cp = c;
532 cp = do_expr(cp, t->extra);
533 LV(t->f_type, 0);
534 c = *cp++;
535 ftbl = t;
536 break;
537
538 case TF_NOP:
539 *--cp = c;
540 cp = do_expr(cp, t->extra);
541 c = *cp++;
542 ftbl = t;
543 break;
544 }
545 if (c != ')') {
546 CERROR("')' expected");
547 }
548 --infunction;
549 return (cp);
550 }
551
552 static char *
553 do_expr (char *sp, int preprocess)
554 {
555 register char *cp = sp;
556 register int c;
557
558 if ((c = *cp++) == '{') {
559 cp = do_name (cp, preprocess);
560 fp->f_type = FT_LS_COMP;
561 } else if (c == '(') {
562 cp = do_func (cp);
563 } else if (c == ')') {
564 return (--cp);
565 } else if (c == '%' && *cp == '<') {
566 cp = do_if (cp+1);
567 } else {
568 CERROR ("'(', '{', '%<' or ')' expected");
569 }
570 return (cp);
571 }
572
573 static char *
574 do_loop(char *sp)
575 {
576 register char *cp = sp;
577 struct format *floop;
578
579 floop = next_fp;
580 cp = compile (cp);
581 if (*cp++ != ']')
582 CERROR ("']' expected");
583
584 LV(FT_DONE, 1); /* not yet done */
585 LV(FT_GOTO, 0);
586 fp->f_skip = floop - fp; /* skip backwards */
587
588 return cp;
589 }
590
591 static char *
592 do_if(char *sp)
593 {
594 register char *cp = sp;
595 register struct format *fexpr,
596 *fif = (struct format *)NULL;
597 register int c = '<';
598
599 for (;;) {
600 if (c == '<') { /* doing an IF */
601 if ((c = *cp++) == '{') /*}*/{
602 cp = do_name(cp, 0);
603 fp->f_type = FT_LS_COMP;
604 LV (FT_IF_S, 0);
605 }
606 else if (c == '(') {
607 cp = do_func(cp);
608 /* see if we can merge the load and the "if" */
609 if (ftbl->f_type >= IF_FUNCS)
610 fp->f_type = ftbl->extra;
611 else {
612 LV (FT_IF_V_NE, 0);
613 }
614 }
615 else {
616 CERROR("'(' or '{' expected"); /*}*/
617 }
618 }
619
620 fexpr = fp; /* loc of [ELS]IF */
621 cp = compile (cp); /* compile IF TRUE stmts */
622 if (fif)
623 fif->f_skip = next_fp - fif;
624
625 if ((c = *cp++) == '|') { /* the last ELSE */
626 LV(FT_GOTO, 0);
627 fif = fp; /* loc of GOTO */
628 fexpr->f_skip = next_fp - fexpr;
629
630 fexpr = (struct format *)NULL;/* no extra ENDIF */
631
632 cp = compile (cp); /* compile ELSE stmts */
633 fif->f_skip = next_fp - fif;
634 c = *cp++;
635 }
636 else if (c == '?') { /* another ELSIF */
637 LV(FT_GOTO, 0);
638 fif = fp; /* loc of GOTO */
639 fexpr->f_skip = next_fp - fexpr;
640
641 c = '<'; /* impersonate an IF */
642 continue;
643 }
644 break;
645 }
646
647 if (c != '>') {
648 CERROR("'>' expected.");
649 }
650
651 if (fexpr) /* IF ... [ELSIF ...] ENDIF */
652 fexpr->f_skip = next_fp - fexpr;
653
654 return (cp);
655 }