diplodocus.org Git - nmh/blob - sbr/fmt_compile.c

   1 /* fmt_compile.c -- "compile" format strings for fmt_scan
   2  *
   3  * This code is Copyright (c) 2002, by the authors of nmh.  See the
   4  * COPYRIGHT file in the root directory of the nmh distribution for
   5  * complete copyright information.
   6  *
   7  * This code compiles the format strings (documented in mh-format(5)) into
   8  * an internal form to be later processed by fmt_scan.c.
   9  *
  10  * What happens here is that the format strings are parsed and an array
  11  * of struct format structures are returned.  Each format structure is
  12  * a single operation interpreted by the the routines in fmt_scan.c.
  13  *
  14  * There is a NOT a one-to-one correspondence between format strings and
  15  * format instructions; some functions have side effects that can result
  16  * in multiple instructions being generated.  The exact list of instructions
  17  * generated by a format string can be seem with the nmh fmttest utility.
  18  *
  19  * A list of format instructions can be found in fmt_compile.h.
  20  *
  21  * If you wish to add a new function, you will need to do the following
  22  * things:
  23  *
  24  * - Add a new instruction to the list of instructions in fmt_compile.h.
  25  *   Note that test instructions (starting with FT_IF_S_NULL) have special
  26  *   handling, so if you are NOT writing a test function then you need
  27  *   to insert it into the list before that _and_ bump all of the
  28  *   following instruction numbers.
  29  *
  30  * - Add the function name to the functable[] array below, and write any
  31  *   special code that your function may require in terms of parsing
  32  *   (it very well may not need anything).
  33  *
  34  * - Add the code in fmt_scan.c to handle your new function.
  35  *
  36  * - Add code to fmttest.c to display your new function.
  37  *
  38  * - Document the new function in mh-format(5).
  39  *
  40  */
  41
  42 #include <h/mh.h>
  43 #include <h/addrsbr.h>
  44 #include <h/tws.h>
  45 #include <h/fmt_scan.h>
  46 #include <h/fmt_compile.h>
  47 #include <h/mts.h>
  48 #include <h/utils.h>
  49 #include "terminal.h"
  50
  51 #ifdef HAVE_SYS_TIME_H
  52 # include <sys/time.h>
  53 #endif
  54 #include <time.h>
  55
  56 /*
  57  * hash table for deciding if a component is "interesting"
  58  */
  59 static struct comp *wantcomp[128];
  60
  61 static struct format *formatvec;        /* array to hold formats */
  62 static struct format *next_fp;          /* next free format slot */
  63 static struct format *fp;               /* current format slot   */
  64 static struct comp *cm;                 /* most recent comp ref  */
  65 static struct ftable *ftbl;             /* most recent func ref  */
  66 static int ncomp;
  67 static int infunction;                  /* function nesting cnt  */
  68
  69 extern struct mailname fmt_mnull;
  70
  71 /* ftable->type (argument type) */
  72 #define TF_COMP    0        /* component expected                 */
  73 #define TF_NUM     1        /* number expected                    */
  74 #define TF_STR     2        /* string expected                    */
  75 #define TF_EXPR    3        /* component or func. expected        */
  76 #define TF_NONE    4        /* no argument                        */
  77 #define TF_MYBOX   5        /* special - get current user's mbox  */
  78 #define TF_NOW     6        /* special - get current unix time    */
  79 #define TF_EXPR_SV 7        /* like expr but save current str reg */
  80 #define TF_NOP     8        /* like expr but no result            */
  81 #define TF_MYNAME  9        /* special - get current name of user */
  82 #define TF_MYHOST  10       /* special - get "local" hostname     */
  83 #define TF_LMBOX   11       /* special - get full local mailbox   */
  84 #define TF_BOLD    12       /* special - enter terminal bold mode */
  85 #define TF_UNDERLN 13       /* special - enter underline mode     */
  86 #define TF_STNDOUT 14       /* special - enter underline mode     */
  87 #define TF_RESET   15       /* special - reset terminal modes     */
  88 #define TF_HASCLR  16       /* special - terminal have color?     */
  89 #define TF_FGCOLR  17       /* special - foreground term color    */
  90 #define TF_BGCOLR  18       /* special - background term color    */
  91
  92 /* ftable->flags */
  93 /* NB that TFL_PUTS is also used to decide whether the test
  94  * in a "%<(function)..." should be a string or numeric one.
  95  */
  96 #define TFL_PUTS   1        /* implicit putstr if top level */
  97 #define TFL_PUTN   2        /* implicit putnum if top level */
  98
  99 /*
 100  * The functable array maps between the text names of format functions and
 101  * the format instructions interpreted by the engine in fmt_scan.c.
 102  *
 103  * The elements of this structure are as follows:
 104  *
 105  * name -   The name of the function as seen in the format string.  This is
 106  *          what maps a particular function name into a format instruction.
 107  * type -   The type of argument this function expects.  Those types are
 108  *          listed above (with the TF_ prefix).  This affects what gets
 109  *          placed in the format instruction (the f_un union).  Also,
 110  *          instructions that require special handling are distinguished
 111  *          here (TF_MYMBOX is one example).
 112  * f_type - The instruction corresponding to this function (from the list
 113  *          in fmt_compile.h).
 114  * extra  - Used by some functions to provide extra data to the compiler.
 115  *          Uses include:
 116  *              - Providing an alternate instruction to combine a load
 117  *                and test operation (see do_if()).
 118  *              - Passed in f_value in the format instruction to provide
 119  *                extra information for the engine (see FT_LV_DAT handling
 120  *                in fmt_scan.c).
 121  *              - Provide a hint as to preprocessing that is required for
 122  *                this instruction (see do_name()).
 123  * flags  - See the definitions for TFL_PUTS & TFL_PUTN above.
 124  */
 125
 126 struct ftable {
 127     char *name;         /* function name                  */
 128     char type;          /* argument type                  */
 129     char f_type;        /* fmt type                       */
 130     char extra;         /* arg. type dependent extra info */
 131     char flags;
 132 };
 133
 134 static struct ftable functable[] = {
 135      { "nonzero",    TF_EXPR,   FT_V_NE,        FT_IF_V_NE,     0 },
 136      { "zero",       TF_EXPR,   FT_V_EQ,        FT_IF_V_EQ,     0 },
 137      { "eq",         TF_NUM,    FT_V_EQ,        FT_IF_V_EQ,     0 },
 138      { "ne",         TF_NUM,    FT_V_NE,        FT_IF_V_NE,     0 },
 139      { "gt",         TF_NUM,    FT_V_GT,        FT_IF_V_GT,     0 },
 140      { "null",       TF_EXPR,   FT_S_NULL,      FT_IF_S_NULL,   0 },
 141      { "nonnull",    TF_EXPR,   FT_S_NONNULL,   FT_IF_S,        0 },
 142      { "match",      TF_STR,    FT_V_MATCH,     FT_IF_MATCH,    0 },
 143      { "amatch",     TF_STR,    FT_V_AMATCH,    FT_IF_AMATCH,   0 },
 144
 145      { "putstr",     TF_EXPR,   FT_STR,         0,              0 },
 146      { "putstrf",    TF_EXPR,   FT_STRF,        0,              0 },
 147      { "putnum",     TF_EXPR,   FT_NUM,         0,              0 },
 148      { "putnumf",    TF_EXPR,   FT_NUMF,        0,              0 },
 149      { "putaddr",    TF_STR,    FT_PUTADDR,     0,              0 },
 150      { "putlit",     TF_EXPR,   FT_STRLIT,      0,              0 },
 151      { "zputlit",    TF_EXPR,   FT_STRLITZ,     0,              0 },
 152      { "void",       TF_NOP,    0,              0,              0 },
 153
 154      { "comp",       TF_COMP,   FT_LS_COMP,     0,              TFL_PUTS },
 155      { "lit",        TF_STR,    FT_LS_LIT,      0,              TFL_PUTS },
 156      { "getenv",     TF_STR,    FT_LS_GETENV,   0,              TFL_PUTS },
 157      { "profile",    TF_STR,    FT_LS_CFIND,    0,              TFL_PUTS },
 158      { "decodecomp", TF_COMP,   FT_LS_DECODECOMP, 0,            TFL_PUTS },
 159      { "decode",     TF_EXPR,   FT_LS_DECODE,   0,              TFL_PUTS },
 160      { "trim",       TF_EXPR,   FT_LS_TRIM,     0,              0 },
 161      { "kilo",       TF_EXPR,   FT_LS_KILO,     0,              TFL_PUTS },
 162      { "kibi",       TF_EXPR,   FT_LS_KIBI,     0,              TFL_PUTS },
 163      { "compval",    TF_COMP,   FT_LV_COMP,     0,              TFL_PUTN },
 164      { "compflag",   TF_COMP,   FT_LV_COMPFLAG, 0,              TFL_PUTN },
 165      { "num",        TF_NUM,    FT_LV_LIT,      0,              TFL_PUTN },
 166      { "msg",        TF_NONE,   FT_LV_DAT,      0,              TFL_PUTN },
 167      { "cur",        TF_NONE,   FT_LV_DAT,      1,              TFL_PUTN },
 168      { "size",       TF_NONE,   FT_LV_DAT,      2,              TFL_PUTN },
 169      { "width",      TF_NONE,   FT_LV_DAT,      3,              TFL_PUTN },
 170      { "unseen",     TF_NONE,   FT_LV_DAT,      4,              TFL_PUTN },
 171      { "dat",        TF_NUM,    FT_LV_DAT,      0,              TFL_PUTN },
 172      { "strlen",     TF_NONE,   FT_LV_STRLEN,   0,              TFL_PUTN },
 173      { "me",         TF_MYBOX,  FT_LS_LIT,      0,              TFL_PUTS },
 174      { "myname",     TF_MYNAME, FT_LS_LIT,      0,              TFL_PUTS },
 175      { "myhost",     TF_MYHOST, FT_LS_LIT,      0,              TFL_PUTS },
 176      { "localmbox",  TF_LMBOX,  FT_LS_LIT,      0,              TFL_PUTS },
 177      { "plus",       TF_NUM,    FT_LV_PLUS_L,   0,              TFL_PUTN },
 178      { "minus",      TF_NUM,    FT_LV_MINUS_L,  0,              TFL_PUTN },
 179      { "multiply",   TF_NUM,    FT_LV_MULTIPLY_L, 0,            TFL_PUTN },
 180      { "divide",     TF_NUM,    FT_LV_DIVIDE_L, 0,              TFL_PUTN },
 181      { "modulo",     TF_NUM,    FT_LV_MODULO_L, 0,              TFL_PUTN },
 182      { "charleft",   TF_NONE,   FT_LV_CHAR_LEFT, 0,             TFL_PUTN },
 183      { "timenow",    TF_NOW,    FT_LV_LIT,      0,              TFL_PUTN },
 184
 185      { "month",      TF_COMP,   FT_LS_MONTH,    FT_PARSEDATE,   TFL_PUTS },
 186      { "lmonth",     TF_COMP,   FT_LS_LMONTH,   FT_PARSEDATE,   TFL_PUTS },
 187      { "tzone",      TF_COMP,   FT_LS_ZONE,     FT_PARSEDATE,   TFL_PUTS },
 188      { "day",        TF_COMP,   FT_LS_DAY,      FT_PARSEDATE,   TFL_PUTS },
 189      { "weekday",    TF_COMP,   FT_LS_WEEKDAY,  FT_PARSEDATE,   TFL_PUTS },
 190      { "tws",        TF_COMP,   FT_LS_822DATE,  FT_PARSEDATE,   TFL_PUTS },
 191      { "sec",        TF_COMP,   FT_LV_SEC,      FT_PARSEDATE,   TFL_PUTN },
 192      { "min",        TF_COMP,   FT_LV_MIN,      FT_PARSEDATE,   TFL_PUTN },
 193      { "hour",       TF_COMP,   FT_LV_HOUR,     FT_PARSEDATE,   TFL_PUTN },
 194      { "mday",       TF_COMP,   FT_LV_MDAY,     FT_PARSEDATE,   TFL_PUTN },
 195      { "mon",        TF_COMP,   FT_LV_MON,      FT_PARSEDATE,   TFL_PUTN },
 196      { "year",       TF_COMP,   FT_LV_YEAR,     FT_PARSEDATE,   TFL_PUTN },
 197      { "yday",       TF_COMP,   FT_LV_YDAY,     FT_PARSEDATE,   TFL_PUTN },
 198      { "wday",       TF_COMP,   FT_LV_WDAY,     FT_PARSEDATE,   TFL_PUTN },
 199      { "zone",       TF_COMP,   FT_LV_ZONE,     FT_PARSEDATE,   TFL_PUTN },
 200      { "clock",      TF_COMP,   FT_LV_CLOCK,    FT_PARSEDATE,   TFL_PUTN },
 201      { "rclock",     TF_COMP,   FT_LV_RCLOCK,   FT_PARSEDATE,   TFL_PUTN },
 202      { "sday",       TF_COMP,   FT_LV_DAYF,     FT_PARSEDATE,   TFL_PUTN },
 203      { "szone",      TF_COMP,   FT_LV_ZONEF,    FT_PARSEDATE,   TFL_PUTN },
 204      { "dst",        TF_COMP,   FT_LV_DST,      FT_PARSEDATE,   TFL_PUTN },
 205      { "pretty",     TF_COMP,   FT_LS_PRETTY,   FT_PARSEDATE,   TFL_PUTS },
 206      { "nodate",     TF_COMP,   FT_LV_COMPFLAG, FT_PARSEDATE,   TFL_PUTN },
 207      { "date2local", TF_COMP,   FT_LOCALDATE,   FT_PARSEDATE,   0 },
 208      { "date2gmt",   TF_COMP,   FT_GMTDATE,     FT_PARSEDATE,   0 },
 209
 210      { "pers",       TF_COMP,   FT_LS_PERS,     FT_PARSEADDR,   TFL_PUTS },
 211      { "mbox",       TF_COMP,   FT_LS_MBOX,     FT_PARSEADDR,   TFL_PUTS },
 212      { "host",       TF_COMP,   FT_LS_HOST,     FT_PARSEADDR,   TFL_PUTS },
 213      { "path",       TF_COMP,   FT_LS_PATH,     FT_PARSEADDR,   TFL_PUTS },
 214      { "gname",      TF_COMP,   FT_LS_GNAME,    FT_PARSEADDR,   TFL_PUTS },
 215      { "note",       TF_COMP,   FT_LS_NOTE,     FT_PARSEADDR,   TFL_PUTS },
 216      { "addr",       TF_COMP,   FT_LS_ADDR,     FT_PARSEADDR,   TFL_PUTS },
 217      { "proper",     TF_COMP,   FT_LS_822ADDR,  FT_PARSEADDR,   TFL_PUTS },
 218      { "type",       TF_COMP,   FT_LV_HOSTTYPE, FT_PARSEADDR,   TFL_PUTN },
 219      { "ingrp",      TF_COMP,   FT_LV_INGRPF,   FT_PARSEADDR,   TFL_PUTN },
 220      { "nohost",     TF_COMP,   FT_LV_NOHOSTF,  FT_PARSEADDR,   TFL_PUTN },
 221      { "formataddr", TF_EXPR_SV,FT_FORMATADDR,  FT_FORMATADDR,  0 },
 222      { "concataddr", TF_EXPR_SV,FT_CONCATADDR,  FT_FORMATADDR,  0 },
 223      { "friendly",   TF_COMP,   FT_LS_FRIENDLY, FT_PARSEADDR,   TFL_PUTS },
 224
 225      { "mymbox",     TF_COMP,   FT_LV_COMPFLAG, FT_MYMBOX,      TFL_PUTN },
 226      { "getmymbox",  TF_COMP,   FT_STR,         FT_GETMYMBOX,   0 },
 227      { "getmyaddr",  TF_COMP,   FT_LS_ADDR,     FT_GETMYADDR,   TFL_PUTS },
 228
 229      { "unquote",    TF_EXPR,   FT_LS_UNQUOTE,  0,              TFL_PUTS },
 230
 231      { "bold",       TF_BOLD,   FT_LS_LIT,      0,              TFL_PUTS },
 232      { "underline",  TF_UNDERLN,FT_LS_LIT,      0,              TFL_PUTS },
 233      { "standout",   TF_STNDOUT,FT_LS_LIT,      0,              TFL_PUTS },
 234      { "resetterm",  TF_RESET,  FT_LS_LIT,      0,              TFL_PUTS },
 235      { "hascolor",   TF_HASCLR, FT_LV_LIT,      0,              0 },
 236      { "fgcolor",    TF_FGCOLR, FT_LS_LIT,      0,              TFL_PUTS },
 237      { "bgcolor",    TF_BGCOLR, FT_LS_LIT,      0,              TFL_PUTS },
 238
 239      { NULL,         0,         0,              0,              0 }
 240 };
 241
 242 /*
 243  * A mapping of color names to terminfo color numbers.
 244  *
 245  * There are two sets of terminal-setting codes: 'setaf/setab' (ANSI) and
 246  * 'setf/setb'.  Different terminals support different capabilities, so
 247  * we provide a mapping for both.  I'm not crazy about putting numbers
 248  * directly in here, but it seems these are well defined by terminfo
 249  * so it should be okay.
 250  */
 251
 252 struct colormap {
 253     char *colorname;    /* Name of color */
 254     int ansinum;        /* The ANSI escape color number */
 255     int nonansinum;     /* The non-ANSI escape color number */
 256 };
 257
 258 static struct colormap colortable[] = {
 259     { "black",          0,      0 },
 260     { "red",            1,      4 },
 261     { "green",          2,      2 },
 262     { "yellow",         3,      6 },
 263     { "blue",           4,      1 },
 264     { "magenta",        5,      5 },
 265     { "cyan",           6,      3 },
 266     { "white",          7,      7 },
 267     { NULL,             0,      0 }
 268 };
 269
 270 /* Hash function for component name.  Deliberately avoids a function
 271  * call.  Is case independent.  Covers interval [0, 126] so never uses
 272  * the last element of wantcomp[]. This function is "pretty good". */
 273 #define CHASH(nm) ( \
 274         (( \
 275             ((nm)[0]) - ((nm)[0] ? ((nm)[1]) : 0) \
 276         ) & 0x1f) + \
 277         ((nm[1]) ? (((nm)[2]) & 0x5f) : 0) \
 278     )
 279
 280 /*
 281  * Find a component in the hash table.
 282  */
 283 #define FINDCOMP(comp,name) do { \
 284         for (comp = wantcomp[CHASH(name)]; \
 285             comp && strcmp(comp->c_name,name); \
 286             comp = comp->c_next) \
 287         ; \
 288     } while (0)
 289
 290 /* Add new component to the hash table */
 291 #define NEWCOMP(cm,name) do { \
 292         NEW0(cm);\
 293         cm->c_name = mh_xstrdup(name);\
 294         cm->c_refcount++;\
 295         ncomp++;\
 296         i = CHASH(name);\
 297         cm->c_next = wantcomp[i];\
 298         wantcomp[i] = cm; \
 299         } while (0)
 300
 301 #define NEW_FP(type,fill,wid) do {\
 302         fp=next_fp++; fp->f_type=(type); fp->f_fill=(fill); fp->f_width=(wid); \
 303         } while (0)
 304
 305 /* Add (possibly new) component to the hash table */
 306 #define ADDC(name) do { \
 307         FINDCOMP(cm, name);\
 308         if (!cm) {\
 309             NEWCOMP(cm,name);\
 310         }\
 311         fp->f_comp = cm; \
 312         fp->f_flags |= FF_COMPREF; \
 313         cm->c_refcount++; \
 314         } while (0)
 315
 316 #define LV(type, value)         do { NEW_FP(type,0,0); fp->f_value = (value); } while (0)
 317 #define LS(type, str)           do { NEW_FP(type,0,0); fp->f_text = getcpy(str); fp->f_flags |= FF_STRALLOC; } while (0)
 318
 319 #define PUTCOMP(comp)           do { NEW_FP(FT_COMP,0,0); ADDC(comp); } while (0)
 320 #define PUTLIT(str)             do { NEW_FP(FT_LIT,0,0); fp->f_text = getcpy(str); fp->f_flags |= FF_STRALLOC; } while (0)
 321 #define PUTC(c)                 do { NEW_FP(FT_CHAR,0,0); fp->f_char = (c); } while (0)
 322
 323 static char *format_string;
 324 static char *usr_fstring;       /* for CERROR */
 325
 326 #define CERROR(str) compile_error (str, cp)
 327
 328 /*
 329  * static prototypes
 330  */
 331 static struct ftable *lookup(char *);
 332 static void compile_error(char *, char *);
 333 static char *compile (char *);
 334 static char *do_spec(char *);
 335 static char *do_name(char *, int);
 336 static char *do_func(char *);
 337 static char *do_expr (char *, int);
 338 static char *do_loop(char *);
 339 static char *do_if(char *);
 340 static void free_component(struct comp *);
 341 static void free_comptable(void);
 342
 343 /*
 344  * Lookup a function name in the functable
 345  */
 346 static struct ftable *
 347 lookup(char *name)
 348 {
 349     struct ftable *t = functable;
 350     char *nm;
 351     char c = *name;
 352
 353     while ((nm = t->name)) {
 354         if (*nm == c && strcmp (nm, name) == 0)
 355             return (ftbl = t);
 356
 357         t++;
 358     }
 359     return NULL;
 360 }
 361
 362
 363 static void
 364 compile_error(char *str, char *cp)
 365 {
 366     int i, errpos, errctx;
 367
 368     errpos = cp - format_string;
 369     errctx = min(errpos, 20);
 370     usr_fstring[errpos] = '\0';
 371
 372     for (i = errpos-errctx; i < errpos; i++) {
 373         if (iscntrl((unsigned char) usr_fstring[i]))
 374             usr_fstring[i] = '_';
 375     }
 376
 377     inform("\"%s\": format compile error - %s",
 378            &usr_fstring[errpos-errctx], str);
 379     adios (NULL, "%*s", errctx+1, "^");
 380 }
 381
 382 /*
 383  * Compile format string "fstring" into format list "fmt".
 384  * Return the number of header components found in the format
 385  * string.
 386  */
 387
 388 int
 389 fmt_compile(char *fstring, struct format **fmt, int reset_comptable)
 390 {
 391     char *cp;
 392     size_t i;
 393     static int comptable_initialized = 0;
 394
 395     format_string = mh_xstrdup(fstring);
 396     usr_fstring = fstring;
 397
 398     if (reset_comptable || !comptable_initialized) {
 399         free_comptable();
 400         comptable_initialized = 1;
 401     }
 402
 403     /* it takes at least 4 char to generate one format so we
 404      * allocate a worst-case format array using 1/4 the length
 405      * of the format string.  We actually need twice this much
 406      * to handle both pre-processing (e.g., address parsing) and
 407      * normal processing.
 408      */
 409     i = strlen(fstring)/2 + 1;
 410                 if (i==1) i++;
 411     next_fp = formatvec = mh_xcalloc(i, sizeof *next_fp);
 412     infunction = 0;
 413
 414     cp = compile(format_string);
 415     if (*cp) {
 416         CERROR("extra '%>', '%|' or '%?'");
 417     }
 418     LV(FT_DONE, 0);             /* really done */
 419     *fmt = formatvec;
 420
 421     free(format_string);
 422     return ncomp;
 423 }
 424
 425 static char *
 426 compile (char *sp)
 427 {
 428     char *cp = sp;
 429     int  c;
 430
 431     for (;;) {
 432         sp = cp;
 433         while ((c = *cp) && c != '%')
 434             cp++;
 435         *cp = 0;
 436         switch (cp-sp) {
 437         case 0:
 438             break;
 439         case 1:
 440             PUTC(*sp);
 441             break;
 442         default:
 443             PUTLIT(sp);
 444             break;
 445         }
 446         if (c == 0)
 447             return cp;
 448
 449         switch (c = *++cp) {
 450         case '%':
 451             PUTC (*cp);
 452             cp++;
 453             break;
 454
 455         case '|':
 456         case '>':
 457         case '?':
 458         case ']':
 459             return cp;
 460
 461         case '<':
 462             cp = do_if(++cp);
 463             break;
 464
 465         case '[':       /* ] */
 466             cp = do_loop(++cp);
 467             break;
 468
 469         case ';':       /* comment line */
 470             cp++;
 471             while ((c = *cp++) && c != '\n')
 472                 continue;
 473             break;
 474
 475         default:
 476             cp = do_spec(cp);
 477             break;
 478         }
 479     }
 480 }
 481
 482
 483 /*
 484  * Process functions & components (handle field width here as well
 485  */
 486 static char *
 487 do_spec(char *sp)
 488 {
 489     char *cp = sp;
 490     int c;
 491 #ifndef lint
 492     int ljust = 0;
 493 #endif  /* not lint */
 494     int wid = 0;
 495     char fill = ' ';
 496
 497     c = *cp++;
 498     if (c == '-') {
 499         ljust++;
 500         c = *cp++;
 501     }
 502     if (c == '0') {
 503         fill = c;
 504         c = *cp++;
 505     }
 506     while (isdigit(c)) {
 507         wid = wid*10 + (c - '0');
 508         c = *cp++;
 509     }
 510     if (c == '{') {
 511         cp = do_name(cp, 0);
 512         if (! infunction)
 513             fp->f_type = wid? FT_COMPF : FT_COMP;
 514     }
 515     else if (c == '(') {
 516         cp = do_func(cp);
 517         if (! infunction) {
 518             if (ftbl->flags & TFL_PUTS) {
 519                 LV( wid? FT_STRF : FT_STR, ftbl->extra);
 520             }
 521             else if (ftbl->flags & TFL_PUTN) {
 522                 LV( wid? FT_NUMF : FT_NUM, ftbl->extra);
 523             }
 524         }
 525     }
 526     else {
 527         CERROR("component or function name expected");
 528     }
 529     if (ljust)
 530         wid = -wid;
 531     fp->f_width = wid;
 532     fp->f_fill = fill;
 533
 534     return cp;
 535 }
 536
 537 /*
 538  * Process a component name.  Normally this involves generating an FT_COMP
 539  * instruction for the specified component.  If preprocess is set, then we
 540  * do some extra processing.
 541  */
 542 static char *
 543 do_name(char *sp, int preprocess)
 544 {
 545     char *cp = sp;
 546     int c;
 547     int i;
 548     static int primed = 0;
 549
 550     while (isalnum(c = *cp++) || c == '-' || c == '_')
 551         ;
 552     if (c != '}') {
 553         CERROR("'}' expected");
 554     }
 555     cp[-1] = '\0';
 556     PUTCOMP(sp);
 557     switch (preprocess) {
 558
 559     case FT_PARSEDATE:
 560         if (cm->c_type & CT_ADDR) {
 561             CERROR("component used as both date and address");
 562         }
 563         if (cm->c_tws) {
 564             ZERO(cm->c_tws);
 565         } else {
 566             NEW0(cm->c_tws);
 567         }
 568         fp->f_type = preprocess;
 569         PUTCOMP(sp);
 570         cm->c_type |= CT_DATE;
 571         break;
 572
 573     case FT_MYMBOX:
 574     case FT_GETMYMBOX:
 575     case FT_GETMYADDR:
 576         if (!primed) {
 577             ismymbox(NULL);
 578             primed++;
 579         }
 580         /* FALLTHRU */
 581     case FT_PARSEADDR:
 582         if (cm->c_type & CT_DATE) {
 583             CERROR("component used as both date and address");
 584         }
 585         cm->c_mn = &fmt_mnull;
 586         fp->f_type = preprocess;
 587         PUTCOMP(sp);
 588         cm->c_type |= CT_ADDR;
 589         break;
 590
 591     case FT_FORMATADDR:
 592         if (cm->c_type & CT_DATE) {
 593             CERROR("component used as both date and address");
 594         }
 595         cm->c_type |= CT_ADDR;
 596         break;
 597     }
 598     return cp;
 599 }
 600
 601 /*
 602  * Generate one or more instructions corresponding to the named function.
 603  * The different type of function arguments are handled here.
 604  */
 605 static char *
 606 do_func(char *sp)
 607 {
 608     char *cp = sp;
 609     int c;
 610     struct ftable *t;
 611     int n;
 612     int mflag;          /* minus sign in NUM */
 613
 614     infunction++;
 615
 616     while (isalnum(c = *cp++))
 617         ;
 618     if (c != '(' && c != '{' && c != ' ' && c != ')') {
 619         CERROR("'(', '{', ' ' or ')' expected");
 620     }
 621     cp[-1] = '\0';
 622     if ((t = lookup (sp)) == 0) {
 623         CERROR("unknown function");
 624     }
 625     if (isspace(c))
 626         c = *cp++;
 627
 628     switch (t->type) {
 629
 630     case TF_COMP:
 631         if (c != '{') {
 632             CERROR("component name expected");
 633         }
 634         cp = do_name(cp, t->extra);
 635         fp->f_type = t->f_type;
 636         c = *cp++;
 637         break;
 638
 639     case TF_NUM:
 640         if ((mflag = (c == '-')))
 641             c = *cp++;
 642         n = 0;
 643         while (isdigit(c)) {
 644             n = n*10 + (c - '0');
 645             c = *cp++;
 646         }
 647         if (mflag)
 648             n = (-n);
 649         LV(t->f_type,n);
 650         break;
 651
 652     case TF_STR:
 653         sp = cp - 1;
 654         while (c && c != ')')
 655             c = *cp++;
 656         cp[-1] = '\0';
 657         LS(t->f_type,sp);
 658         break;
 659
 660     case TF_NONE:
 661         LV(t->f_type,t->extra);
 662         break;
 663
 664     case TF_MYBOX:
 665         LS(t->f_type, getusername());
 666         break;
 667
 668     case TF_MYNAME:
 669         LS(t->f_type, getfullname());
 670         break;
 671
 672     case TF_MYHOST:
 673         LS(t->f_type, LocalName(0));
 674         break;
 675
 676     case TF_LMBOX:
 677         LS(t->f_type, getlocalmbox());
 678         break;
 679
 680     case TF_BOLD:
 681         LS(t->f_type, get_term_stringcap("bold"));
 682         break;
 683
 684     case TF_UNDERLN:
 685         LS(t->f_type, get_term_stringcap("smul"));
 686         break;
 687
 688     case TF_STNDOUT:
 689         LS(t->f_type, get_term_stringcap("smso"));
 690         break;
 691
 692     case TF_RESET:
 693         LS(t->f_type, get_term_stringcap("sgr0"));
 694         break;
 695
 696     case TF_HASCLR:
 697         LV(t->f_type, get_term_numcap("colors") > 1);
 698         break;
 699
 700     case TF_FGCOLR:
 701     case TF_BGCOLR: {
 702         struct colormap *cmap = colortable;
 703         char *code;
 704
 705         sp = cp - 1;
 706         while (c && c != ')')
 707             c = *cp++;
 708         cp[-1] = '\0';
 709
 710         while (cmap->colorname != NULL) {
 711             if (strcasecmp(sp, cmap->colorname) == 0)
 712                 break;
 713             cmap++;
 714         }
 715
 716         if (cmap->colorname == NULL) {
 717             CERROR("Unknown color name");
 718             break;
 719         }
 720
 721         code = get_term_stringparm(t->type == TF_FGCOLR ? "setaf" : "setab",
 722                                    cmap->ansinum, 0);
 723
 724         /*
 725          * If this doesn't have anything, try falling back to setf/setb
 726          */
 727
 728         if (! code)
 729             code = get_term_stringparm(t->type == TF_FGCOLR ? "setf" : "setb",
 730                                        cmap->nonansinum, 0);
 731
 732         LS(t->f_type, code);
 733         break;
 734     }
 735
 736     case TF_NOW:
 737         LV(t->f_type, time(NULL));
 738         break;
 739
 740     case TF_EXPR_SV:
 741         LV(FT_SAVESTR, 0);
 742         /* FALLTHRU */
 743     case TF_EXPR:
 744         *--cp = c;
 745         cp = do_expr(cp, t->extra);
 746         LV(t->f_type, 0);
 747         c = *cp++;
 748         ftbl = t;
 749         break;
 750
 751     case TF_NOP:
 752         *--cp = c;
 753         cp = do_expr(cp, t->extra);
 754         c = *cp++;
 755         ftbl = t;
 756         break;
 757     }
 758     if (c != ')') {
 759         CERROR("')' expected");
 760     }
 761     --infunction;
 762     return cp;
 763 }
 764
 765 /*
 766  * Handle an expression as an argument.  Basically we call one of do_name(),
 767  * do_func(), or do_if()
 768  */
 769 static char *
 770 do_expr (char *sp, int preprocess)
 771 {
 772     char *cp = sp;
 773     int  c;
 774
 775     if ((c = *cp++) == '{') {
 776         cp = do_name (cp, preprocess);
 777         fp->f_type = FT_LS_COMP;
 778     } else if (c == '(') {
 779         cp = do_func (cp);
 780     } else if (c == ')') {
 781         return --cp;
 782     } else if (c == '%' && *cp == '<') {
 783         cp = do_if (cp+1);
 784     } else {
 785         CERROR ("'(', '{', '%<' or ')' expected");
 786     }
 787     return cp;
 788 }
 789
 790 /*
 791  * I am guessing this was for some kind of loop statement, which would have
 792  * looked like %[ .... %].  It looks like the way this would have worked
 793  * is that the format engine would have seen that FT_DONE had a 1 in the
 794  * f_un.f_un_value and then decided whether or not to continue the loop.
 795  * There is no support for this in the format engine, so right now if
 796  * you try using it you will reach the FT_DONE and simply stop.  I'm leaving
 797  * this here in case someone wants to continue the work.
 798  *
 799  * Okay, got some more information on this from John L. Romine!  From an
 800  * email he sent to the nmh-workers mailing list on December 2, 2010, he
 801  * explains it so:
 802  *
 803  *    In this case (scan, formatsbr) it has to do with an extension to
 804  *    the mh-format syntax to allow for looping.
 805  *
 806  *    The scan format is processed once for each message.  Those #ifdef
 807  *    JLR changes allowed for the top part of the format file to be
 808  *    processed once, then a second, looping part to be processed
 809  *    once per message.  As I recall, there were new mh-format escape
 810  *    sequences to delimit the loop.  This would have allowed for things
 811  *    like per-format column headings in the scan output.
 812  *
 813  *    Since existing format files didn't include the scan listing
 814  *    header (it was hard-coded in scan.c) it would not have been
 815  *    backward-compatible.  All existing format files (including any
 816  *    local ones) would have needed to be changed to include the format
 817  *    codes for a header.  The practice at the time was not to introduce
 818  *    incompatible changes in a minor release, and I never managed to
 819  *    put out a newer major release.
 820  *
 821  * I can see how this would work, and I suspect part of the motivation was
 822  * because the format compiler routines (at the time) couldn't really be
 823  * called multiple times on the same message because the memory management
 824  * was so lousy.  That's been reworked and things are now a lot cleaner,
 825  * so I suspect if we're going to allow a format string to be used for the
 826  * scan header it might be simpler to have a separate format string just
 827  * for the header.  But I'll leave this code in for now just in case we
 828  * decide that we want some kind of looping support.
 829  */
 830 static char *
 831 do_loop(char *sp)
 832 {
 833     char *cp = sp;
 834     struct format *floop;
 835
 836     floop = next_fp;
 837     cp = compile (cp);
 838     if (*cp++ != ']')
 839         CERROR ("']' expected");
 840
 841     LV(FT_DONE, 1);             /* not yet done */
 842     LV(FT_GOTO, 0);
 843     fp->f_skip = floop - fp;    /* skip backwards */
 844
 845     return cp;
 846 }
 847
 848 /*
 849  * Handle an if-elsif-endif statement.  Note here that the branching
 850  * is handled by the f_skip member of the struct format (which is really
 851  * just f_width overloaded).  This number controls how far to move forward
 852  * (or back) in the format instruction array.
 853  */
 854 static char *
 855 do_if(char *sp)
 856 {
 857     char *cp = sp;
 858     struct format *fexpr,
 859         *fif = NULL;
 860     int c = '<';
 861
 862     for (;;) {
 863         if (c == '<') {                 /* doing an IF */
 864             if ((c = *cp++) == '{') /*}*/{
 865                 cp = do_name(cp, 0);
 866                 fp->f_type = FT_LS_COMP;
 867                 LV (FT_IF_S, 0);
 868             }
 869             else if (c == '(') {
 870                 cp = do_func(cp);
 871                 /* see if we can merge the load and the "if" */
 872                 if (ftbl->f_type >= IF_FUNCS)
 873                     fp->f_type = ftbl->extra;
 874                 else {
 875                     /* Put out a string test or a value test depending
 876                      * on what this function's return type is.
 877                      */
 878                     if (ftbl->flags & TFL_PUTS) {
 879                         LV (FT_IF_S, 0);
 880                     } else {
 881                         LV (FT_IF_V_NE, 0);
 882                     }
 883                 }
 884             }
 885             else {
 886                 CERROR("'(' or '{' expected");  /*}*/
 887             }
 888         }
 889
 890         fexpr = fp;                     /* loc of [ELS]IF */
 891         cp = compile (cp);              /* compile IF TRUE stmts */
 892         if (fif)
 893             fif->f_skip = next_fp - fif;
 894
 895         if ((c = *cp++) == '|') {       /* the last ELSE */
 896             LV(FT_GOTO, 0);
 897             fif = fp;                   /* loc of GOTO */
 898             fexpr->f_skip = next_fp - fexpr;
 899
 900             fexpr = NULL;               /* no extra ENDIF */
 901
 902             cp = compile (cp);          /* compile ELSE stmts */
 903             fif->f_skip = next_fp - fif;
 904             c = *cp++;
 905         }
 906         else if (c == '?') {            /* another ELSIF */
 907             LV(FT_GOTO, 0);
 908             fif = fp;                   /* loc of GOTO */
 909             fexpr->f_skip = next_fp - fexpr;
 910
 911             c = '<';                    /* impersonate an IF */
 912             continue;
 913         }
 914         break;
 915     }
 916
 917     if (c != '>') {
 918         CERROR("'>' expected.");
 919     }
 920
 921     if (fexpr)                          /* IF ... [ELSIF ...] ENDIF */
 922         fexpr->f_skip = next_fp - fexpr;
 923
 924     return cp;
 925 }
 926
 927 /*
 928  * Free a set of format instructions.
 929  *
 930  * What we do here is:
 931  *
 932  * - Iterate through the list of format instructions, freeing any references
 933  *   to allocated memory in each instruction.
 934  * - Free component references.
 935  * - If requested, reset the component hash table; that will also free any
 936  *   references to components stored there.
 937  *
 938  */
 939
 940 void
 941 fmt_free(struct format *fmt, int reset_comptable)
 942 {
 943     struct format *fp = fmt;
 944
 945     if (fp) {
 946         while (! (fp->f_type == FT_DONE && fp->f_value == 0)) {
 947             if (fp->f_flags & FF_STRALLOC)
 948                 free(fp->f_text);
 949             if (fp->f_flags & FF_COMPREF)
 950                 free_component(fp->f_comp);
 951             fp++;
 952         }
 953         free(fmt);
 954     }
 955
 956     if (reset_comptable)
 957         free_comptable();
 958 }
 959
 960 /*
 961  * Free just the text strings from all of the component hash table entries
 962  */
 963
 964 void
 965 fmt_freecomptext(void)
 966 {
 967     unsigned int i;
 968     struct comp *cm;
 969
 970     for (i = 0; i < DIM(wantcomp); i++)
 971         for (cm = wantcomp[i]; cm; cm = cm->c_next) {
 972             free(cm->c_text);
 973             cm->c_text = NULL;
 974         }
 975 }
 976
 977 /*
 978  * Find a component in our hash table.  This is just a public interface to
 979  * the FINDCOMP macro, so we don't have to expose our hash table.
 980  */
 981
 982 struct comp *
 983 fmt_findcomp(char *component)
 984 {
 985     struct comp *cm;
 986
 987     FINDCOMP(cm, component);
 988
 989     return cm;
 990 }
 991
 992 /*
 993  * Like fmt_findcomp, but case-insensitive.
 994  */
 995
 996 struct comp *
 997 fmt_findcasecomp(char *component)
 998 {
 999     struct comp *cm;
1000
1001     for (cm = wantcomp[CHASH(component)]; cm; cm = cm->c_next)
1002         if (strcasecmp(component, FENDNULL(cm->c_name)) == 0)
1003             break;
1004
1005     return cm;
1006 }
1007
1008 /*
1009  * Add an entry to the component hash table
1010  *
1011  * Returns true if the component was added, 0 if it already existed.
1012  *
1013  */
1014
1015 int
1016 fmt_addcompentry(char *component)
1017 {
1018     struct comp *cm;
1019     int i;
1020
1021     FINDCOMP(cm, component);
1022
1023     if (cm)
1024         return 0;
1025
1026     NEWCOMP(cm, component);
1027
1028     /*
1029      * ncomp is really meant for fmt_compile() and this function is
1030      * meant to be used outside of it.  So decrement it just to be safe
1031      * (internal callers should be using NEWCOMP()).
1032      */
1033
1034     ncomp--;
1035
1036     return 1;
1037 }
1038
1039 /*
1040  * Add a string to a component hash table entry.
1041  *
1042  * Note the special handling for components marked with CT_ADDR.  The comments
1043  * in fmt_scan.h explain this in more detail.
1044  */
1045
1046 int
1047 fmt_addcomptext(char *component, char *text)
1048 {
1049     int i, found = 0, bucket = CHASH(component);
1050     struct comp *cptr = wantcomp[bucket];
1051     char *cp;
1052
1053     while (cptr) {
1054         if (strcasecmp(component, FENDNULL(cptr->c_name)) == 0) {
1055             found++;
1056             if (! cptr->c_text) {
1057                 cptr->c_text = getcpy(text);
1058             } else {
1059                 i = strlen(cp = cptr->c_text) - 1;
1060                 if (cp[i] == '\n') {
1061                     if (cptr->c_type & CT_ADDR) {
1062                         cp[i] = '\0';
1063                         cp = add(",\n\t", cp);
1064                     } else {
1065                         cp = add("\t", cp);
1066                     }
1067                 }
1068                 cptr->c_text = add(text, cp);
1069             }
1070         }
1071         cptr = cptr->c_next;
1072     }
1073
1074     return found ? bucket : -1;
1075 }
1076
1077 /*
1078  * Append text to a component we've already found.  See notes in fmt_scan.h
1079  * for more information.
1080  */
1081
1082 void
1083 fmt_appendcomp(int bucket, char *component, char *text)
1084 {
1085     struct comp *cptr;
1086
1087     if (bucket != -1) {
1088         for (cptr = wantcomp[bucket]; cptr; cptr = cptr->c_next)
1089             if (strcasecmp(component, FENDNULL(cptr->c_name)) == 0)
1090                 cptr->c_text = add(text, cptr->c_text);
1091     }
1092 }
1093
1094 /*
1095  * Iterate over our component hash table
1096  */
1097
1098 struct comp *
1099 fmt_nextcomp(struct comp *comp, unsigned int *bucket)
1100 {
1101     if (comp == NULL)
1102         *bucket = 0;
1103     else
1104         comp = comp->c_next;
1105
1106     while (comp == NULL && *bucket < DIM(wantcomp)) {
1107         comp = wantcomp[(*bucket)++];
1108     }
1109
1110     return comp;
1111 }
1112
1113 /*
1114  * Free and reset our component hash table
1115  */
1116
1117 static void
1118 free_comptable(void)
1119 {
1120     unsigned int i;
1121     struct comp *cm, *cm2;
1122
1123     for (i = 0; i < DIM(wantcomp); i++) {
1124         cm = wantcomp[i];
1125         while (cm != NULL) {
1126             cm2 = cm->c_next;
1127             free_component(cm);
1128             cm = cm2;
1129         }
1130         wantcomp[i] = 0;
1131     }
1132
1133     ncomp = 0;
1134 }
1135
1136 /*
1137  * Decrement the reference count of a component structure.  If it reaches
1138  * zero, free it
1139  */
1140
1141 static void
1142 free_component(struct comp *cm)
1143 {
1144     if (--cm->c_refcount <= 0) {
1145         /* Shouldn't ever be NULL, but just in case ... */
1146         free(cm->c_name);
1147         free(cm->c_text);
1148         if (cm->c_type & CT_DATE)
1149             free(cm->c_tws);
1150         if (cm->c_type & CT_ADDR && cm->c_mn && cm->c_mn != &fmt_mnull)
1151             mnfree(cm->c_mn);
1152         free(cm);
1153     }
1154 }