3 * fmt_compile.c -- "compile" format strings for fmt_scan
5 * This code is Copyright (c) 2002, by the authors of nmh. See the
6 * COPYRIGHT file in the root directory of the nmh distribution for
7 * complete copyright information.
9 * This code compiles the format strings (documented in mh-format(5)) into
10 * an internal form to be later processed by fmt_scan.c.
12 * What happens here is that the format strings are parsed and an array
13 * of struct format structures are returned. Each format structure is
14 * a single operation interpreted by the the routines in fmt_scan.c.
16 * There is a NOT a one-to-one correspondence between format strings and
17 * format instructions; some functions have side effects that can result
18 * in multiple instructions being generated. The exact list of instructions
19 * generated by a format string can be seem with the nmh fmtdump utility.
21 * A list of format instructions can be found in fmt_compile.h.
23 * If you wish to add a new function, you will need to do the following
26 * - Add a new instruction to the list of instructions in fmt_compile.h.
27 * Note that test instructions (starting with FT_IF_S_NULL) have special
28 * handling, so if you are NOT writing a test function then you need
29 * to insert it into the list before that _and_ bump all of the
30 * following instruction numbers.
32 * - Add the function name to the functable[] array below, and write any
33 * special code that your function may require in terms of parsing
34 * (it very well may not need anything).
36 * - Add the code in fmt_scan.c to handle your new function.
38 * - Document the new function in the mh-format(5) man page.
43 #include <h/addrsbr.h>
45 #include <h/fmt_scan.h>
46 #include <h/fmt_compile.h>
49 #ifdef HAVE_SYS_TIME_H
50 # include <sys/time.h>
55 * hash table for deciding if a component is "interesting"
57 struct comp *wantcomp[128];
59 static struct format *formatvec; /* array to hold formats */
60 static struct format *next_fp; /* next free format slot */
61 static struct format *fp; /* current format slot */
62 static struct comp *cm; /* most recent comp ref */
63 static struct ftable *ftbl; /* most recent func ref */
65 static int infunction; /* function nesting cnt */
67 extern struct mailname fmt_mnull;
69 /* ftable->type (argument type) */
70 #define TF_COMP 0 /* component expected */
71 #define TF_NUM 1 /* number expected */
72 #define TF_STR 2 /* string expected */
73 #define TF_EXPR 3 /* component or func. expected */
74 #define TF_NONE 4 /* no argument */
75 #define TF_MYBOX 5 /* special - get current user's mbox */
76 #define TF_NOW 6 /* special - get current unix time */
77 #define TF_EXPR_SV 7 /* like expr but save current str reg */
78 #define TF_NOP 8 /* like expr but no result */
79 #define TF_MYNAME 9 /* special - get current name of user */
80 #define TF_MYHOST 10 /* special - get "local" hostname */
81 #define TF_LMBOX 11 /* special - get full local mailbox */
84 /* NB that TFL_PUTS is also used to decide whether the test
85 * in a "%<(function)..." should be a string or numeric one.
87 #define TFL_PUTS 1 /* implicit putstr if top level */
88 #define TFL_PUTN 2 /* implicit putnum if top level */
91 * The functable array maps between the text names of format functions and
92 * the format instructions interpreted by the engine in fmt_scan.c.
94 * The elements of this structure are as follows:
96 * name - The name of the function as seen in the format string. This is
97 * what maps a particular function name into a format instruction.
98 * type - The type of argument this function expects. Those types are
99 * listed above (with the TF_ prefix). This affects what gets
100 * placed in the format instruction (the f_un union).
101 * f_type - The instruction corresponding to this function (from the list
103 * extra - Used by some functions to provide extra data to the compiler.
105 * - Providing an alternate instruction to combine a load
106 * and test operation (see do_if()).
107 * - Passed in f_value in the format instruction to provide
108 * extra information for the engine (see FT_LV_DAT handling
110 * - Provide a hint as to preprocessing that is required for
111 * this instruction (see do_name()).
112 * flags - See the definitions for TFL_PUTS & TFL_PUTN above.
116 char *name; /* function name */
117 char type; /* argument type */
118 char f_type; /* fmt type */
119 char extra; /* arg. type dependent extra info */
123 static struct ftable functable[] = {
124 { "nonzero", TF_EXPR, FT_V_NE, FT_IF_V_NE, 0 },
125 { "zero", TF_EXPR, FT_V_EQ, FT_IF_V_EQ, 0 },
126 { "eq", TF_NUM, FT_V_EQ, FT_IF_V_EQ, 0 },
127 { "ne", TF_NUM, FT_V_NE, FT_IF_V_NE, 0 },
128 { "gt", TF_NUM, FT_V_GT, FT_IF_V_GT, 0 },
129 { "null", TF_EXPR, FT_S_NULL, FT_IF_S_NULL, 0 },
130 { "nonnull", TF_EXPR, FT_S_NONNULL, FT_IF_S, 0 },
131 { "match", TF_STR, FT_V_MATCH, FT_IF_MATCH, 0 },
132 { "amatch", TF_STR, FT_V_AMATCH, FT_IF_AMATCH, 0 },
134 { "putstr", TF_EXPR, FT_STR, 0, 0 },
135 { "putstrf", TF_EXPR, FT_STRF, 0, 0 },
136 { "putnum", TF_EXPR, FT_NUM, 0, 0 },
137 { "putnumf", TF_EXPR, FT_NUMF, 0, 0 },
138 { "putaddr", TF_STR, FT_PUTADDR, 0, 0 },
139 { "putlit", TF_STR, FT_STRLIT, 0, 0 },
140 { "void", TF_NOP, 0, 0, 0 },
142 { "comp", TF_COMP, FT_LS_COMP, 0, TFL_PUTS },
143 { "lit", TF_STR, FT_LS_LIT, 0, TFL_PUTS },
144 { "getenv", TF_STR, FT_LS_GETENV, 0, TFL_PUTS },
145 { "profile", TF_STR, FT_LS_CFIND, 0, TFL_PUTS },
146 { "decodecomp", TF_COMP, FT_LS_DECODECOMP, 0, TFL_PUTS },
147 { "decode", TF_EXPR, FT_LS_DECODE, 0, TFL_PUTS },
148 { "trim", TF_EXPR, FT_LS_TRIM, 0, 0 },
149 { "compval", TF_COMP, FT_LV_COMP, 0, TFL_PUTN },
150 { "compflag", TF_COMP, FT_LV_COMPFLAG, 0, TFL_PUTN },
151 { "num", TF_NUM, FT_LV_LIT, 0, TFL_PUTN },
152 { "msg", TF_NONE, FT_LV_DAT, 0, TFL_PUTN },
153 { "cur", TF_NONE, FT_LV_DAT, 1, TFL_PUTN },
154 { "size", TF_NONE, FT_LV_DAT, 2, TFL_PUTN },
155 { "width", TF_NONE, FT_LV_DAT, 3, TFL_PUTN },
156 { "unseen", TF_NONE, FT_LV_DAT, 4, TFL_PUTN },
157 { "dat", TF_NUM, FT_LV_DAT, 0, TFL_PUTN },
158 { "strlen", TF_NONE, FT_LV_STRLEN, 0, TFL_PUTN },
159 { "me", TF_MYBOX, FT_LS_LIT, 0, TFL_PUTS },
160 { "myname", TF_MYNAME, FT_LS_LIT, 0, TFL_PUTS },
161 { "myhost", TF_MYHOST, FT_LS_LIT, 0, TFL_PUTS },
162 { "localmbox", TF_LMBOX, FT_LS_LIT, 0, TFL_PUTS },
163 { "plus", TF_NUM, FT_LV_PLUS_L, 0, TFL_PUTN },
164 { "minus", TF_NUM, FT_LV_MINUS_L, 0, TFL_PUTN },
165 { "divide", TF_NUM, FT_LV_DIVIDE_L, 0, TFL_PUTN },
166 { "modulo", TF_NUM, FT_LV_MODULO_L, 0, TFL_PUTN },
167 { "charleft", TF_NONE, FT_LV_CHAR_LEFT, 0, TFL_PUTN },
168 { "timenow", TF_NOW, FT_LV_LIT, 0, TFL_PUTN },
170 { "month", TF_COMP, FT_LS_MONTH, FT_PARSEDATE, TFL_PUTS },
171 { "lmonth", TF_COMP, FT_LS_LMONTH, FT_PARSEDATE, TFL_PUTS },
172 { "tzone", TF_COMP, FT_LS_ZONE, FT_PARSEDATE, TFL_PUTS },
173 { "day", TF_COMP, FT_LS_DAY, FT_PARSEDATE, TFL_PUTS },
174 { "weekday", TF_COMP, FT_LS_WEEKDAY, FT_PARSEDATE, TFL_PUTS },
175 { "tws", TF_COMP, FT_LS_822DATE, FT_PARSEDATE, TFL_PUTS },
176 { "sec", TF_COMP, FT_LV_SEC, FT_PARSEDATE, TFL_PUTN },
177 { "min", TF_COMP, FT_LV_MIN, FT_PARSEDATE, TFL_PUTN },
178 { "hour", TF_COMP, FT_LV_HOUR, FT_PARSEDATE, TFL_PUTN },
179 { "mday", TF_COMP, FT_LV_MDAY, FT_PARSEDATE, TFL_PUTN },
180 { "mon", TF_COMP, FT_LV_MON, FT_PARSEDATE, TFL_PUTN },
181 { "year", TF_COMP, FT_LV_YEAR, FT_PARSEDATE, TFL_PUTN },
182 { "yday", TF_COMP, FT_LV_YDAY, FT_PARSEDATE, TFL_PUTN },
183 { "wday", TF_COMP, FT_LV_WDAY, FT_PARSEDATE, TFL_PUTN },
184 { "zone", TF_COMP, FT_LV_ZONE, FT_PARSEDATE, TFL_PUTN },
185 { "clock", TF_COMP, FT_LV_CLOCK, FT_PARSEDATE, TFL_PUTN },
186 { "rclock", TF_COMP, FT_LV_RCLOCK, FT_PARSEDATE, TFL_PUTN },
187 { "sday", TF_COMP, FT_LV_DAYF, FT_PARSEDATE, TFL_PUTN },
188 { "szone", TF_COMP, FT_LV_ZONEF, FT_PARSEDATE, TFL_PUTN },
189 { "dst", TF_COMP, FT_LV_DST, FT_PARSEDATE, TFL_PUTN },
190 { "pretty", TF_COMP, FT_LS_PRETTY, FT_PARSEDATE, TFL_PUTS },
191 { "nodate", TF_COMP, FT_LV_COMPFLAG, FT_PARSEDATE, TFL_PUTN },
192 { "date2local", TF_COMP, FT_LOCALDATE, FT_PARSEDATE, 0 },
193 { "date2gmt", TF_COMP, FT_GMTDATE, FT_PARSEDATE, 0 },
195 { "pers", TF_COMP, FT_LS_PERS, FT_PARSEADDR, TFL_PUTS },
196 { "mbox", TF_COMP, FT_LS_MBOX, FT_PARSEADDR, TFL_PUTS },
197 { "host", TF_COMP, FT_LS_HOST, FT_PARSEADDR, TFL_PUTS },
198 { "path", TF_COMP, FT_LS_PATH, FT_PARSEADDR, TFL_PUTS },
199 { "gname", TF_COMP, FT_LS_GNAME, FT_PARSEADDR, TFL_PUTS },
200 { "note", TF_COMP, FT_LS_NOTE, FT_PARSEADDR, TFL_PUTS },
201 { "addr", TF_COMP, FT_LS_ADDR, FT_PARSEADDR, TFL_PUTS },
202 { "proper", TF_COMP, FT_LS_822ADDR, FT_PARSEADDR, TFL_PUTS },
203 { "type", TF_COMP, FT_LV_HOSTTYPE, FT_PARSEADDR, TFL_PUTN },
204 { "ingrp", TF_COMP, FT_LV_INGRPF, FT_PARSEADDR, TFL_PUTN },
205 { "nohost", TF_COMP, FT_LV_NOHOSTF, FT_PARSEADDR, TFL_PUTN },
206 { "formataddr", TF_EXPR_SV,FT_FORMATADDR, FT_FORMATADDR, 0 },
207 { "concataddr", TF_EXPR_SV,FT_CONCATADDR, FT_FORMATADDR, 0 },
208 { "friendly", TF_COMP, FT_LS_FRIENDLY, FT_PARSEADDR, TFL_PUTS },
210 { "mymbox", TF_COMP, FT_LV_COMPFLAG, FT_MYMBOX, TFL_PUTN },
211 { "addtoseq", TF_STR, FT_ADDTOSEQ, 0, 0 },
213 { "unquote", TF_EXPR, FT_LS_UNQUOTE, 0, TFL_PUTS},
218 /* Add new component to the hash table */
219 #define NEWCOMP(cm,name) do { \
220 cm = ((struct comp *) calloc(1, sizeof (struct comp)));\
224 cm->c_next = wantcomp[i];\
228 #define NEWFMT (next_fp++)
229 #define NEW(type,fill,wid) do {\
230 fp=NEWFMT; fp->f_type=(type); fp->f_fill=(fill); fp->f_width=(wid); \
233 /* Add (possibly new) component to the hash table */
234 #define ADDC(name) do { \
242 #define LV(type, value) do { NEW(type,0,0); fp->f_value = (value); } while (0)
243 #define LS(type, str) do { NEW(type,0,0); fp->f_text = (str); } while (0)
245 #define PUTCOMP(comp) do { NEW(FT_COMP,0,0); ADDC(comp); } while (0)
246 #define PUTLIT(str) do { NEW(FT_LIT,0,0); fp->f_text = (str); } while (0)
247 #define PUTC(c) do { NEW(FT_CHAR,0,0); fp->f_char = (c); } while (0)
249 static char *format_string;
250 static unsigned char *usr_fstring; /* for CERROR */
252 #define CERROR(str) compile_error (str, cp)
257 static struct ftable *lookup(char *);
258 static void compile_error(char *, char *);
259 static char *compile (char *);
260 static char *do_spec(char *);
261 static char *do_name(char *, int);
262 static char *do_func(char *);
263 static char *do_expr (char *, int);
264 static char *do_loop(char *);
265 static char *do_if(char *);
269 * Lookup a function name in the functable
271 static struct ftable *
274 register struct ftable *t = functable;
276 register char c = *name;
278 while ((nm = t->name)) {
279 if (*nm == c && strcmp (nm, name) == 0)
284 return (struct ftable *) 0;
289 compile_error(char *str, char *cp)
291 int i, errpos, errctx;
293 errpos = cp - format_string;
294 errctx = errpos > 20 ? 20 : errpos;
295 usr_fstring[errpos] = '\0';
297 for (i = errpos-errctx; i < errpos; i++) {
299 if (iscntrl(usr_fstring[i]))
301 if (usr_fstring[i] < 32)
303 usr_fstring[i] = '_';
306 advise(NULL, "\"%s\": format compile error - %s",
307 &usr_fstring[errpos-errctx], str);
308 adios (NULL, "%*s", errctx+1, "^");
312 * Compile format string "fstring" into format list "fmt".
313 * Return the number of header components found in the format
318 fmt_compile(char *fstring, struct format **fmt)
324 free (format_string);
325 format_string = getcpy (fstring);
326 usr_fstring = fstring;
328 /* init the component hash table. */
329 for (i = 0; i < sizeof(wantcomp)/sizeof(wantcomp[0]); i++)
332 memset((char *) &fmt_mnull, 0, sizeof(fmt_mnull));
334 /* it takes at least 4 char to generate one format so we
335 * allocate a worst-case format array using 1/4 the length
336 * of the format string. We actually need twice this much
337 * to handle both pre-processing (e.g., address parsing) and
340 i = strlen(fstring)/2 + 1;
342 next_fp = formatvec = (struct format *)calloc ((size_t) i,
343 sizeof(struct format));
345 adios (NULL, "unable to allocate format storage");
350 cp = compile(format_string);
352 CERROR("extra '%>', '%|' or '%?'");
354 LV(FT_DONE, 0); /* really done */
363 register char *cp = sp;
368 while ((c = *cp) && c != '%')
404 case ';': /* comment line */
406 while ((c = *cp++) && c != '\n')
419 * Process functions & components (handle field width here as well
424 register char *cp = sp;
427 register int ljust = 0;
428 #endif /* not lint */
429 register int wid = 0;
430 register char fill = ' ';
442 wid = wid*10 + (c - '0');
448 fp->f_type = wid? FT_COMPF : FT_COMP;
453 if (ftbl->flags & TFL_PUTS) {
454 LV( wid? FT_STRF : FT_STR, ftbl->extra);
456 else if (ftbl->flags & TFL_PUTN) {
457 LV( wid? FT_NUMF : FT_NUM, ftbl->extra);
462 CERROR("component or function name expected");
473 * Process a component name. Normally this involves generating an FT_COMP
474 * instruction for the specified component. If preprocess is set, then we
475 * do some extra processing.
478 do_name(char *sp, int preprocess)
480 register char *cp = sp;
483 static int primed = 0;
485 while (isalnum(c = *cp++) || c == '-' || c == '_')
488 CERROR("'}' expected");
492 switch (preprocess) {
495 if (cm->c_type & CT_ADDR) {
496 CERROR("component used as both date and address");
498 cm->c_tws = (struct tws *)
499 calloc((size_t) 1, sizeof(*cm->c_tws));
500 fp->f_type = preprocess;
502 cm->c_type |= CT_DATE;
507 ismymbox ((struct mailname *) 0);
512 if (cm->c_type & CT_DATE) {
513 CERROR("component used as both date and address");
515 cm->c_mn = &fmt_mnull;
516 fp->f_type = preprocess;
518 cm->c_type |= CT_ADDR;
522 if (cm->c_type & CT_DATE) {
523 CERROR("component used as both date and address");
525 cm->c_type |= CT_ADDR;
532 * Generate one or more instructions corresponding to the named function.
533 * The different type of function arguments are handled here.
538 register char *cp = sp;
540 register struct ftable *t;
542 int mflag; /* minus sign in NUM */
546 while (isalnum(c = *cp++))
548 if (c != '(' && c != '{' && c != ' ' && c != ')') {
549 CERROR("'(', '{', ' ' or ')' expected");
552 if ((t = lookup (sp)) == 0) {
553 CERROR("unknown function");
562 CERROR("component name expected");
564 cp = do_name(cp, t->extra);
565 fp->f_type = t->f_type;
570 if ((mflag = (c == '-')))
574 n = n*10 + (c - '0');
584 while (c && c != ')')
591 LV(t->f_type,t->extra);
595 LS(t->f_type, getusername());
599 LS(t->f_type, getfullname());
603 LS(t->f_type, LocalName(0));
607 LS(t->f_type, getlocalmbox());
611 LV(t->f_type, time((time_t *) 0));
619 cp = do_expr(cp, t->extra);
627 cp = do_expr(cp, t->extra);
633 CERROR("')' expected");
640 * Handle an expression as an argument. Basically we call one of do_name(),
641 * do_func(), or do_if()
644 do_expr (char *sp, int preprocess)
646 register char *cp = sp;
649 if ((c = *cp++) == '{') {
650 cp = do_name (cp, preprocess);
651 fp->f_type = FT_LS_COMP;
652 } else if (c == '(') {
654 } else if (c == ')') {
656 } else if (c == '%' && *cp == '<') {
659 CERROR ("'(', '{', '%<' or ')' expected");
665 * I am guessing this was for some kind of loop statement, which would have
666 * looked like %[ .... %]. It looks like the way this would have worked
667 * is that the format engine would have seen that FT_DONE had a 1 in the
668 * f_un.f_un_value and then decided whether or not to continue the loop.
669 * There is no support for this in the format engine, so right now if
670 * you try using it you will reach the FT_DONE and simply stop. I'm leaving
671 * this here in case someone wants to continue the work.
676 register char *cp = sp;
677 struct format *floop;
682 CERROR ("']' expected");
684 LV(FT_DONE, 1); /* not yet done */
686 fp->f_skip = floop - fp; /* skip backwards */
692 * Handle an if-elsif-endif statement. Note here that the branching
693 * is handled by the f_skip member of the struct format (which is really
694 * just f_width overloaded). This number controls how far to move forward
695 * (or back) in the format instruction array.
700 register char *cp = sp;
701 register struct format *fexpr,
702 *fif = (struct format *)NULL;
703 register int c = '<';
706 if (c == '<') { /* doing an IF */
707 if ((c = *cp++) == '{') /*}*/{
709 fp->f_type = FT_LS_COMP;
714 /* see if we can merge the load and the "if" */
715 if (ftbl->f_type >= IF_FUNCS)
716 fp->f_type = ftbl->extra;
718 /* Put out a string test or a value test depending
719 * on what this function's return type is.
721 if (ftbl->flags & TFL_PUTS) {
729 CERROR("'(' or '{' expected"); /*}*/
733 fexpr = fp; /* loc of [ELS]IF */
734 cp = compile (cp); /* compile IF TRUE stmts */
736 fif->f_skip = next_fp - fif;
738 if ((c = *cp++) == '|') { /* the last ELSE */
740 fif = fp; /* loc of GOTO */
741 fexpr->f_skip = next_fp - fexpr;
743 fexpr = (struct format *)NULL;/* no extra ENDIF */
745 cp = compile (cp); /* compile ELSE stmts */
746 fif->f_skip = next_fp - fif;
749 else if (c == '?') { /* another ELSIF */
751 fif = fp; /* loc of GOTO */
752 fexpr->f_skip = next_fp - fexpr;
754 c = '<'; /* impersonate an IF */
761 CERROR("'>' expected.");
764 if (fexpr) /* IF ... [ELSIF ...] ENDIF */
765 fexpr->f_skip = next_fp - fexpr;