]> diplodocus.org Git - nmh/blob - sbr/fmt_compile.c
Bring these changes over from the branch.
[nmh] / sbr / fmt_compile.c
1
2 /*
3 * fmt_compile.c -- "compile" format strings for fmt_scan
4 *
5 * $Id$
6 *
7 * This code is Copyright (c) 2002, by the authors of nmh. See the
8 * COPYRIGHT file in the root directory of the nmh distribution for
9 * complete copyright information.
10 */
11
12 #include <h/mh.h>
13 #include <h/addrsbr.h>
14 #include <h/tws.h>
15 #include <h/fmt_scan.h>
16 #include <h/fmt_compile.h>
17
18 #ifdef TIME_WITH_SYS_TIME
19 # include <sys/time.h>
20 # include <time.h>
21 #else
22 # ifdef TM_IN_SYS_TIME
23 # include <sys/time.h>
24 # else
25 # include <time.h>
26 # endif
27 #endif
28
29 /*
30 * hash table for deciding if a component is "interesting"
31 */
32 struct comp *wantcomp[128];
33
34 static struct format *formatvec; /* array to hold formats */
35 static struct format *next_fp; /* next free format slot */
36 static struct format *fp; /* current format slot */
37 static struct comp *cm; /* most recent comp ref */
38 static struct ftable *ftbl; /* most recent func ref */
39 static int ncomp;
40 static int infunction; /* function nesting cnt */
41
42 extern struct mailname fmt_mnull;
43
44 /* ftable->type (argument type) */
45 #define TF_COMP 0 /* component expected */
46 #define TF_NUM 1 /* number expected */
47 #define TF_STR 2 /* string expected */
48 #define TF_EXPR 3 /* component or func. expected */
49 #define TF_NONE 4 /* no argument */
50 #define TF_MYBOX 5 /* special - get current user's mbox */
51 #define TF_NOW 6 /* special - get current unix time */
52 #define TF_EXPR_SV 7 /* like expr but save current str reg */
53 #define TF_NOP 8 /* like expr but no result */
54
55 /* ftable->flags */
56 #define TFL_PUTS 1 /* implicit putstr if top level */
57 #define TFL_PUTN 2 /* implicit putnum if top level */
58
59 struct ftable {
60 char *name; /* function name */
61 char type; /* argument type */
62 char f_type; /* fmt type */
63 char extra; /* arg. type dependent extra info */
64 char flags;
65 };
66
67 static struct ftable functable[] = {
68 { "nonzero", TF_EXPR, FT_V_NE, FT_IF_V_NE, 0 },
69 { "zero", TF_EXPR, FT_V_EQ, FT_IF_V_EQ, 0 },
70 { "eq", TF_NUM, FT_V_EQ, FT_IF_V_EQ, 0 },
71 { "ne", TF_NUM, FT_V_NE, FT_IF_V_NE, 0 },
72 { "gt", TF_NUM, FT_V_GT, FT_IF_V_GT, 0 },
73 { "null", TF_EXPR, FT_S_NULL, FT_IF_S_NULL, 0 },
74 { "nonnull", TF_EXPR, FT_S_NONNULL, FT_IF_S, 0 },
75 { "match", TF_STR, FT_V_MATCH, FT_IF_MATCH, 0 },
76 { "amatch", TF_STR, FT_V_AMATCH, FT_IF_AMATCH, 0 },
77
78 { "putstr", TF_EXPR, FT_STR, 0, 0 },
79 { "putstrf", TF_EXPR, FT_STRF, 0, 0 },
80 { "putnum", TF_EXPR, FT_NUM, 0, 0 },
81 { "putnumf", TF_EXPR, FT_NUMF, 0, 0 },
82 { "putaddr", TF_STR, FT_PUTADDR, 0, 0 },
83 { "void", TF_NOP, 0, 0, 0 },
84
85 { "comp", TF_COMP, FT_LS_COMP, 0, TFL_PUTS },
86 { "lit", TF_STR, FT_LS_LIT, 0, TFL_PUTS },
87 { "getenv", TF_STR, FT_LS_GETENV, 0, TFL_PUTS },
88 { "profile", TF_STR, FT_LS_CFIND, 0, TFL_PUTS },
89 { "decodecomp", TF_COMP, FT_LS_DECODECOMP, 0, TFL_PUTS },
90 { "decode", TF_EXPR, FT_LS_DECODE, 0, TFL_PUTS },
91 { "trim", TF_EXPR, FT_LS_TRIM, 0, 0 },
92 { "compval", TF_COMP, FT_LV_COMP, 0, TFL_PUTN },
93 { "compflag", TF_COMP, FT_LV_COMPFLAG, 0, TFL_PUTN },
94 { "num", TF_NUM, FT_LV_LIT, 0, TFL_PUTN },
95 { "msg", TF_NONE, FT_LV_DAT, 0, TFL_PUTN },
96 { "cur", TF_NONE, FT_LV_DAT, 1, TFL_PUTN },
97 { "size", TF_NONE, FT_LV_DAT, 2, TFL_PUTN },
98 { "width", TF_NONE, FT_LV_DAT, 3, TFL_PUTN },
99 { "unseen", TF_NONE, FT_LV_DAT, 4, TFL_PUTN },
100 { "dat", TF_NUM, FT_LV_DAT, 0, TFL_PUTN },
101 { "strlen", TF_NONE, FT_LV_STRLEN, 0, TFL_PUTN },
102 { "me", TF_MYBOX, FT_LS_LIT, 0, TFL_PUTS },
103 { "plus", TF_NUM, FT_LV_PLUS_L, 0, TFL_PUTN },
104 { "minus", TF_NUM, FT_LV_MINUS_L, 0, TFL_PUTN },
105 { "divide", TF_NUM, FT_LV_DIVIDE_L, 0, TFL_PUTN },
106 { "modulo", TF_NUM, FT_LV_MODULO_L, 0, TFL_PUTN },
107 { "charleft", TF_NONE, FT_LV_CHAR_LEFT, 0, TFL_PUTN },
108 { "timenow", TF_NOW, FT_LV_LIT, 0, TFL_PUTN },
109
110 { "month", TF_COMP, FT_LS_MONTH, FT_PARSEDATE, TFL_PUTS },
111 { "lmonth", TF_COMP, FT_LS_LMONTH, FT_PARSEDATE, TFL_PUTS },
112 { "tzone", TF_COMP, FT_LS_ZONE, FT_PARSEDATE, TFL_PUTS },
113 { "day", TF_COMP, FT_LS_DAY, FT_PARSEDATE, TFL_PUTS },
114 { "weekday", TF_COMP, FT_LS_WEEKDAY, FT_PARSEDATE, TFL_PUTS },
115 { "tws", TF_COMP, FT_LS_822DATE, FT_PARSEDATE, TFL_PUTS },
116 { "sec", TF_COMP, FT_LV_SEC, FT_PARSEDATE, TFL_PUTN },
117 { "min", TF_COMP, FT_LV_MIN, FT_PARSEDATE, TFL_PUTN },
118 { "hour", TF_COMP, FT_LV_HOUR, FT_PARSEDATE, TFL_PUTN },
119 { "mday", TF_COMP, FT_LV_MDAY, FT_PARSEDATE, TFL_PUTN },
120 { "mon", TF_COMP, FT_LV_MON, FT_PARSEDATE, TFL_PUTN },
121 { "year", TF_COMP, FT_LV_YEAR, FT_PARSEDATE, TFL_PUTN },
122 { "yday", TF_COMP, FT_LV_YDAY, FT_PARSEDATE, TFL_PUTN },
123 { "wday", TF_COMP, FT_LV_WDAY, FT_PARSEDATE, TFL_PUTN },
124 { "zone", TF_COMP, FT_LV_ZONE, FT_PARSEDATE, TFL_PUTN },
125 { "clock", TF_COMP, FT_LV_CLOCK, FT_PARSEDATE, TFL_PUTN },
126 { "rclock", TF_COMP, FT_LV_RCLOCK, FT_PARSEDATE, TFL_PUTN },
127 { "sday", TF_COMP, FT_LV_DAYF, FT_PARSEDATE, TFL_PUTN },
128 { "szone", TF_COMP, FT_LV_ZONEF, FT_PARSEDATE, TFL_PUTN },
129 { "dst", TF_COMP, FT_LV_DST, FT_PARSEDATE, TFL_PUTN },
130 { "pretty", TF_COMP, FT_LS_PRETTY, FT_PARSEDATE, TFL_PUTS },
131 { "nodate", TF_COMP, FT_LV_COMPFLAG, FT_PARSEDATE, TFL_PUTN },
132 { "date2local", TF_COMP, FT_LOCALDATE, FT_PARSEDATE, 0 },
133 { "date2gmt", TF_COMP, FT_GMTDATE, FT_PARSEDATE, 0 },
134
135 { "pers", TF_COMP, FT_LS_PERS, FT_PARSEADDR, TFL_PUTS },
136 { "mbox", TF_COMP, FT_LS_MBOX, FT_PARSEADDR, TFL_PUTS },
137 { "host", TF_COMP, FT_LS_HOST, FT_PARSEADDR, TFL_PUTS },
138 { "path", TF_COMP, FT_LS_PATH, FT_PARSEADDR, TFL_PUTS },
139 { "gname", TF_COMP, FT_LS_GNAME, FT_PARSEADDR, TFL_PUTS },
140 { "note", TF_COMP, FT_LS_NOTE, FT_PARSEADDR, TFL_PUTS },
141 { "addr", TF_COMP, FT_LS_ADDR, FT_PARSEADDR, TFL_PUTS },
142 { "proper", TF_COMP, FT_LS_822ADDR, FT_PARSEADDR, TFL_PUTS },
143 { "type", TF_COMP, FT_LV_HOSTTYPE, FT_PARSEADDR, TFL_PUTN },
144 { "ingrp", TF_COMP, FT_LV_INGRPF, FT_PARSEADDR, TFL_PUTN },
145 { "nohost", TF_COMP, FT_LV_NOHOSTF, FT_PARSEADDR, TFL_PUTN },
146 { "formataddr", TF_EXPR_SV,FT_FORMATADDR, FT_FORMATADDR, 0 },
147 { "friendly", TF_COMP, FT_LS_FRIENDLY, FT_PARSEADDR, TFL_PUTS },
148
149 { "mymbox", TF_COMP, FT_LV_COMPFLAG, FT_MYMBOX, TFL_PUTN },
150 { "addtoseq", TF_STR, FT_ADDTOSEQ, 0, 0 },
151
152 { NULL, 0, 0, 0, 0 }
153 };
154
155 /* Add new component to the hash table */
156 #define NEWCOMP(cm,name)\
157 cm = ((struct comp *) calloc(1, sizeof (struct comp)));\
158 cm->c_name = name;\
159 ncomp++;\
160 i = CHASH(name);\
161 cm->c_next = wantcomp[i];\
162 wantcomp[i] = cm;
163
164 #define NEWFMT (next_fp++)
165 #define NEW(type,fill,wid)\
166 fp=NEWFMT; fp->f_type=(type); fp->f_fill=(fill); fp->f_width=(wid);
167
168 /* Add (possibly new) component to the hash table */
169 #define ADDC(name)\
170 FINDCOMP(cm, name);\
171 if (!cm) {\
172 NEWCOMP(cm,name);\
173 }\
174 fp->f_comp = cm;
175
176 #define LV(type, value) NEW(type,0,0); fp->f_value = (value);
177 #define LS(type, str) NEW(type,0,0); fp->f_text = (str);
178
179 #define PUTCOMP(comp) NEW(FT_COMP,0,0); ADDC(comp);
180 #define PUTLIT(str) NEW(FT_LIT,0,0); fp->f_text = (str);
181 #define PUTC(c) NEW(FT_CHAR,0,0); fp->f_char = (c);
182
183 static char *format_string;
184 static char *usr_fstring; /* for CERROR */
185
186 #define CERROR(str) compile_error (str, cp)
187
188 /*
189 * external prototypes
190 */
191 extern char *getusername(void);
192
193 /*
194 * static prototypes
195 */
196 static struct ftable *lookup(char *);
197 static void compile_error(char *, char *);
198 static char *compile (char *);
199 static char *do_spec(char *);
200 static char *do_name(char *, int);
201 static char *do_func(char *);
202 static char *do_expr (char *, int);
203 static char *do_loop(char *);
204 static char *do_if(char *);
205
206
207 static struct ftable *
208 lookup(char *name)
209 {
210 register struct ftable *t = functable;
211 register char *nm;
212 register char c = *name;
213
214 while ((nm = t->name)) {
215 if (*nm == c && strcmp (nm, name) == 0)
216 return (ftbl = t);
217
218 t++;
219 }
220 return (struct ftable *) 0;
221 }
222
223
224 static void
225 compile_error(char *str, char *cp)
226 {
227 int i, errpos, errctx;
228
229 errpos = cp - format_string;
230 errctx = errpos > 20 ? 20 : errpos;
231 usr_fstring[errpos] = '\0';
232
233 for (i = errpos-errctx; i < errpos; i++) {
234 #ifdef LOCALE
235 if (iscntrl(usr_fstring[i]))
236 #else
237 if (usr_fstring[i] < 32)
238 #endif
239 usr_fstring[i] = '_';
240 }
241
242 advise(NULL, "\"%s\": format compile error - %s",
243 &usr_fstring[errpos-errctx], str);
244 adios (NULL, "%*s", errctx+1, "^");
245 }
246
247 /*
248 * Compile format string "fstring" into format list "fmt".
249 * Return the number of header components found in the format
250 * string.
251 */
252
253 int
254 fmt_compile(char *fstring, struct format **fmt)
255 {
256 register char *cp;
257 int i;
258
259 if (format_string)
260 free (format_string);
261 format_string = getcpy (fstring);
262 usr_fstring = fstring;
263
264 /* init the component hash table. */
265 for (i = 0; i < sizeof(wantcomp)/sizeof(wantcomp[0]); i++)
266 wantcomp[i] = 0;
267
268 memset((char *) &fmt_mnull, 0, sizeof(fmt_mnull));
269
270 /* it takes at least 4 char to generate one format so we
271 * allocate a worst-case format array using 1/4 the length
272 * of the format string. We actually need twice this much
273 * to handle both pre-processing (e.g., address parsing) and
274 * normal processing.
275 */
276 i = strlen(fstring)/2 + 1;
277 if (i==1) i++;
278 next_fp = formatvec = (struct format *)calloc ((size_t) i,
279 sizeof(struct format));
280 if (next_fp == NULL)
281 adios (NULL, "unable to allocate format storage");
282
283 ncomp = 0;
284 infunction = 0;
285
286 cp = compile(format_string);
287 if (*cp) {
288 CERROR("extra '%>', '%|' or '%?'");
289 }
290 LV(FT_DONE, 0); /* really done */
291 *fmt = formatvec;
292
293 return (ncomp);
294 }
295
296 static char *
297 compile (char *sp)
298 {
299 register char *cp = sp;
300 register int c;
301
302 for (;;) {
303 sp = cp;
304 while ((c = *cp) && c != '%')
305 cp++;
306 *cp = 0;
307 switch (cp-sp) {
308 case 0:
309 break;
310 case 1:
311 PUTC(*sp);
312 break;
313 default:
314 PUTLIT(sp);
315 break;
316 }
317 if (c == 0)
318 return (cp);
319
320 switch (c = *++cp) {
321 case '%':
322 PUTC (*cp);
323 cp++;
324 break;
325
326 case '|':
327 case '>':
328 case '?':
329 case ']':
330 return (cp);
331
332 case '<':
333 cp = do_if(++cp);
334 break;
335
336 case '[': /* ] */
337 cp = do_loop(++cp);
338 break;
339
340 case ';': /* comment line */
341 cp++;
342 while ((c = *cp++) && c != '\n')
343 continue;
344 break;
345
346 default:
347 cp = do_spec(cp);
348 break;
349 }
350 }
351 }
352
353
354 static char *
355 do_spec(char *sp)
356 {
357 register char *cp = sp;
358 register int c;
359 #ifndef lint
360 register int ljust = 0;
361 #endif /* not lint */
362 register int wid = 0;
363 register char fill = ' ';
364
365 c = *cp++;
366 if (c == '-') {
367 ljust++;
368 c = *cp++;
369 }
370 if (c == '0') {
371 fill = c;
372 c = *cp++;
373 }
374 while (isdigit(c)) {
375 wid = wid*10 + (c - '0');
376 c = *cp++;
377 }
378 if (c == '{') {
379 cp = do_name(cp, 0);
380 if (! infunction)
381 fp->f_type = wid? FT_COMPF : FT_COMP;
382 }
383 else if (c == '(') {
384 cp = do_func(cp);
385 if (! infunction) {
386 if (ftbl->flags & TFL_PUTS) {
387 LV( wid? FT_STRF : FT_STR, ftbl->extra);
388 }
389 else if (ftbl->flags & TFL_PUTN) {
390 LV( wid? FT_NUMF : FT_NUM, ftbl->extra);
391 }
392 }
393 }
394 else {
395 CERROR("component or function name expected");
396 }
397 if (ljust)
398 wid = -wid;
399 fp->f_width = wid;
400 fp->f_fill = fill;
401
402 return (cp);
403 }
404
405 static char *
406 do_name(char *sp, int preprocess)
407 {
408 register char *cp = sp;
409 register int c;
410 register int i;
411 static int primed = 0;
412
413 while (isalnum(c = *cp++) || c == '-' || c == '_')
414 ;
415 if (c != '}') {
416 CERROR("'}' expected");
417 }
418 cp[-1] = '\0';
419 PUTCOMP(sp);
420 switch (preprocess) {
421
422 case FT_PARSEDATE:
423 if (cm->c_type & CT_ADDR) {
424 CERROR("component used as both date and address");
425 }
426 if (! (cm->c_type & CT_DATE)) {
427 cm->c_tws = (struct tws *)
428 calloc((size_t) 1, sizeof(*cm->c_tws));
429 fp->f_type = preprocess;
430 PUTCOMP(sp);
431 cm->c_type |= CT_DATE;
432 }
433 break;
434
435 case FT_MYMBOX:
436 if (!primed) {
437 ismymbox ((struct mailname *) 0);
438 primed++;
439 }
440 cm->c_type |= CT_MYMBOX;
441 /* fall through */
442 case FT_PARSEADDR:
443 if (cm->c_type & CT_DATE) {
444 CERROR("component used as both date and address");
445 }
446 if (! (cm->c_type & CT_ADDRPARSE)) {
447 cm->c_mn = &fmt_mnull;
448 fp->f_type = preprocess;
449 PUTCOMP(sp);
450 cm->c_type |= (CT_ADDR | CT_ADDRPARSE);
451 }
452 break;
453
454 case FT_FORMATADDR:
455 if (cm->c_type & CT_DATE) {
456 CERROR("component used as both date and address");
457 }
458 cm->c_type |= CT_ADDR;
459 break;
460 }
461 return (cp);
462 }
463
464 static char *
465 do_func(char *sp)
466 {
467 register char *cp = sp;
468 register int c;
469 register struct ftable *t;
470 register int n;
471 int mflag; /* minus sign in NUM */
472
473 infunction++;
474
475 while (isalnum(c = *cp++))
476 ;
477 if (c != '(' && c != '{' && c != ' ' && c != ')') {
478 CERROR("'(', '{', ' ' or ')' expected");
479 }
480 cp[-1] = '\0';
481 if ((t = lookup (sp)) == 0) {
482 CERROR("unknown function");
483 }
484 if (isspace(c))
485 c = *cp++;
486
487 switch (t->type) {
488
489 case TF_COMP:
490 if (c != '{') {
491 CERROR("component name expected");
492 }
493 cp = do_name(cp, t->extra);
494 fp->f_type = t->f_type;
495 c = *cp++;
496 break;
497
498 case TF_NUM:
499 if ((mflag = (c == '-')))
500 c = *cp++;
501 n = 0;
502 while (isdigit(c)) {
503 n = n*10 + (c - '0');
504 c = *cp++;
505 }
506 if (mflag)
507 n = (-n);
508 LV(t->f_type,n);
509 break;
510
511 case TF_STR:
512 sp = cp - 1;
513 while (c && c != ')')
514 c = *cp++;
515 cp[-1] = '\0';
516 LS(t->f_type,sp);
517 break;
518
519 case TF_NONE:
520 LV(t->f_type,t->extra);
521 break;
522
523 case TF_MYBOX:
524 LS(t->f_type, getusername());
525 break;
526
527 case TF_NOW:
528 LV(t->f_type, time((time_t *) 0));
529 break;
530
531 case TF_EXPR_SV:
532 LV(FT_SAVESTR, 0);
533 /* fall through */
534 case TF_EXPR:
535 *--cp = c;
536 cp = do_expr(cp, t->extra);
537 LV(t->f_type, 0);
538 c = *cp++;
539 ftbl = t;
540 break;
541
542 case TF_NOP:
543 *--cp = c;
544 cp = do_expr(cp, t->extra);
545 c = *cp++;
546 ftbl = t;
547 break;
548 }
549 if (c != ')') {
550 CERROR("')' expected");
551 }
552 --infunction;
553 return (cp);
554 }
555
556 static char *
557 do_expr (char *sp, int preprocess)
558 {
559 register char *cp = sp;
560 register int c;
561
562 if ((c = *cp++) == '{') {
563 cp = do_name (cp, preprocess);
564 fp->f_type = FT_LS_COMP;
565 } else if (c == '(') {
566 cp = do_func (cp);
567 } else if (c == ')') {
568 return (--cp);
569 } else if (c == '%' && *cp == '<') {
570 cp = do_if (cp+1);
571 } else {
572 CERROR ("'(', '{', '%<' or ')' expected");
573 }
574 return (cp);
575 }
576
577 static char *
578 do_loop(char *sp)
579 {
580 register char *cp = sp;
581 struct format *floop;
582
583 floop = next_fp;
584 cp = compile (cp);
585 if (*cp++ != ']')
586 CERROR ("']' expected");
587
588 LV(FT_DONE, 1); /* not yet done */
589 LV(FT_GOTO, 0);
590 fp->f_skip = floop - fp; /* skip backwards */
591
592 return cp;
593 }
594
595 static char *
596 do_if(char *sp)
597 {
598 register char *cp = sp;
599 register struct format *fexpr,
600 *fif = (struct format *)NULL;
601 register int c = '<';
602
603 for (;;) {
604 if (c == '<') { /* doing an IF */
605 if ((c = *cp++) == '{') /*}*/{
606 cp = do_name(cp, 0);
607 fp->f_type = FT_LS_COMP;
608 LV (FT_IF_S, 0);
609 }
610 else if (c == '(') {
611 cp = do_func(cp);
612 /* see if we can merge the load and the "if" */
613 if (ftbl->f_type >= IF_FUNCS)
614 fp->f_type = ftbl->extra;
615 else {
616 LV (FT_IF_V_NE, 0);
617 }
618 }
619 else {
620 CERROR("'(' or '{' expected"); /*}*/
621 }
622 }
623
624 fexpr = fp; /* loc of [ELS]IF */
625 cp = compile (cp); /* compile IF TRUE stmts */
626 if (fif)
627 fif->f_skip = next_fp - fif;
628
629 if ((c = *cp++) == '|') { /* the last ELSE */
630 LV(FT_GOTO, 0);
631 fif = fp; /* loc of GOTO */
632 fexpr->f_skip = next_fp - fexpr;
633
634 fexpr = (struct format *)NULL;/* no extra ENDIF */
635
636 cp = compile (cp); /* compile ELSE stmts */
637 fif->f_skip = next_fp - fif;
638 c = *cp++;
639 }
640 else if (c == '?') { /* another ELSIF */
641 LV(FT_GOTO, 0);
642 fif = fp; /* loc of GOTO */
643 fexpr->f_skip = next_fp - fexpr;
644
645 c = '<'; /* impersonate an IF */
646 continue;
647 }
648 break;
649 }
650
651 if (c != '>') {
652 CERROR("'>' expected.");
653 }
654
655 if (fexpr) /* IF ... [ELSIF ...] ENDIF */
656 fexpr->f_skip = next_fp - fexpr;
657
658 return (cp);
659 }