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