diplodocus.org Git - nmh/blob - sbr/dtimep.l

   1 /* dtimep.l -- parse dates and times.
   2  *
   3  * This exceeds the default table capacities for some old versions
   4  * of lex (and the minimum defaults as specified by POSIX).  The following
   5  * choices meet or exceed the lex defaults for older SunOS4.x, Solaris,
   6  * HPUX, and AIX.
   7  */
   8 %e4000
   9 %p7000
  10 %n2500
  11 %a5000
  12 %{
  13 #include <h/nmh.h>
  14 #include <h/utils.h>
  15 #include <h/tws.h>
  16
  17   /* Since we're looking at a string at a time, don't worry about
  18    *  wrapping to the next buffer.
  19    */
  20 #if YY_FLEX_MAJOR_VERSION == 2 && \
  21     YY_FLEX_MINOR_VERSION == 6 && \
  22     YY_FLEX_SUBMINOR_VERSION == 3
  23 /* https://github.com/westes/flex/issues/162 */
  24 #undef yywrap
  25 #endif
  26 #define yywrap() 1
  27 #define YY_SKIP_YYWRAP
  28
  29 #define YY_NO_INPUT
  30
  31   /* This is the tricky thing that makes this function cool.  We
  32    *  replace the traditional int yylex(void) declaration with our
  33    *  dparsetime() declaration, essentially piggy-backing off the
  34    *  utility of the yylex() function and adding what we need to make
  35    *  the parsing function useful to us.
  36    */
  37 #define YY_DECL struct tws *dparsetime(char *lexstr)
  38
  39   /* yyterminate() is called after the input string is matched to
  40    * completion (actually, when the lexer reaches an EOF).  The only
  41    * thing that really needs to be in this macro function is the
  42    * return call, which must be substituted inline into dparsetime.
  43    */
  44
  45 #define yyterminate() (void)yy_delete_buffer(lexhandle); \
  46   if(!(tw.tw_flags & TW_SUCC)) { \
  47     return (struct tws *)NULL; \
  48   } \
  49   if(tw.tw_year < 1970) \
  50     tw.tw_year += 1900; \
  51   if(tw.tw_year < 1970) \
  52     tw.tw_year += 100; \
  53   return(&tw)
  54
  55 /*
  56  * Table to convert month names to numeric month.  We use the
  57  * fact that the low order 5 bits of the sum of the 2nd & 3rd
  58  * characters of the name is a hash with no collisions for the 12
  59  * valid month names.  (The mask to 5 bits maps any combination of
  60  * upper and lower case into the same hash value).
  61  */
  62 static int month_map[] = {
  63         0,
  64         6,      /* 1 - Jul */
  65         3,      /* 2 - Apr */
  66         5,      /* 3 - Jun */
  67         0,
  68         10,     /* 5 - Nov */
  69         0,
  70         1,      /* 7 - Feb */
  71         11,     /* 8 - Dec */
  72         0,
  73         0,
  74         0,
  75         0,
  76         0,
  77         0,
  78         0,      /*15 - Jan */
  79         0,
  80         0,
  81         0,
  82         2,      /*19 - Mar */
  83         0,
  84         8,      /*21 - Sep */
  85         0,
  86         9,      /*23 - Oct */
  87         0,
  88         0,
  89         4,      /*26 - May */
  90         0,
  91         7       /*28 - Aug */
  92 };
  93
  94 /*
  95  * Lookup table for day-of-week using the same hash trick as for above name-of-
  96  * month table, but using the first and second character, not second and third.
  97  *
  98  * Compute index into table using: (day_name[0] & 7) + (day_name[1] & 4)
  99  */
 100 static int day_map[] = {
 101         0,
 102         0,
 103         0,
 104         6,      /* 3 - Sat */
 105         4,      /* 4 - Thu */
 106         0,
 107         5,      /* 6 - Fri */
 108         0,      /* 7 - Sun */
 109         2,      /* 8 - Tue */
 110         1       /* 9 - Mon */,
 111         0,
 112         3       /*11 - Wed */
 113 };
 114
 115 /* The SET* macros will parse for the appropriate field, and leave the
 116  * cp pointer at the first character after the desired field. Be
 117  * careful with variable-length fields or alpha-num mixes.
 118
 119  * The SKIP* macros skip over characters of a particular class and
 120  * leave cp at the position of the first character that doesn't match
 121  * that class. Correspondingly, SKIPTO* skips until it reaches a
 122  * character of a particular class.
 123  */
 124
 125 #define INIT()       { cp = yytext;}
 126 #define SETWDAY()    { tw.tw_wday= day_map[(cp[0] & 7) + (cp[1] & 4)]; \
 127                        tw.tw_flags &= ~TW_SDAY; tw.tw_flags |= TW_SEXP; \
 128                        SKIPA(); }
 129 #define SETMON()     { cp++; \
 130                        tw.tw_mon = month_map[(((unsigned char) cp[0]) + ((unsigned char) cp[1])) & 0x1f]; \
 131                        SKIPA(); }
 132 #define SETMON_NUM() { tw.tw_mon = atoi(cp)-1; \
 133                        SKIPD(); }
 134 #define SETYEAR()    { tw.tw_year = atoi(cp); \
 135                        SKIPD(); }
 136 #define SETDAY()     { tw.tw_mday = atoi(cp); \
 137                        tw.tw_flags |= TW_SUCC; \
 138                        SKIPD(); }
 139 #define SETTIME()    { tw.tw_hour = atoi(cp); \
 140                        cp += 2; \
 141                        SKIPTOD(); \
 142                        tw.tw_min = atoi(cp); \
 143                        cp += 2; \
 144                        if(*cp == ':') { \
 145                           tw.tw_sec = atoi(++cp); SKIPD(); } }
 146 #define SETZONE(x)   { tw.tw_zone = ((x)/100)*60+(x)%100; \
 147                        tw.tw_flags |= TW_SZEXP; \
 148                        SKIPD(); }
 149 #define SETDST()     { tw.tw_flags |= TW_DST; }
 150 #define SKIPD()      { while ( isdigit((unsigned char) *cp++) ) ; \
 151                        --cp; }
 152 #define SKIPTOD()    { while ( !isdigit((unsigned char) *cp++) ) ; \
 153                        --cp; }
 154 #define SKIPA()      { while ( isalpha((unsigned char) *cp++) ) ; \
 155                        --cp; }
 156 #define SKIPTOA()    { while ( !isalpha((unsigned char) *cp++) ) ; \
 157                        --cp; }
 158 #define SKIPSP()     { while ( isspace((unsigned char) *cp++) ) ; \
 159                        --cp; }
 160 #define SKIPTOSP()   { while ( !isspace((unsigned char) *cp++) ) ; \
 161                        --cp; }
 162
 163 # ifdef HAVE_SYS_TIME_H
 164 #  include <sys/time.h>
 165 # endif
 166 #include <time.h>
 167
 168 static void
 169 zonehack (struct tws *tw)
 170 {
 171     struct tm *tm;
 172
 173     if (dmktime (tw) == (time_t) -1)
 174         return;
 175
 176     tm = localtime (&tw->tw_clock);
 177     if (tm->tm_isdst) {
 178         tw->tw_flags |= TW_DST;
 179         tw->tw_zone -= 60;
 180     }
 181 }
 182 %}
 183
 184 sun     [Ss]un(day)?
 185 mon     [Mm]on(day)?
 186 tue     [Tt]ue(sday)?
 187 wed     [Ww]ed(nesday)?
 188 thu     [Tt]hu(rsday)?
 189 fri     [Ff]ri(day)?
 190 sat     [Ss]at(urday)?
 191
 192 DAY     {sun}|{mon}|{tue}|{wed}|{thu}|{fri}|{sat}
 193
 194 jan     [Jj]an(uary)?
 195 feb     [Ff]eb(ruary)?
 196 mar     [Mm]ar(ch)?
 197 apr     [Aa]pr(il)?
 198 may     [Mm]ay
 199 jun     [Jj]une?
 200 jul     [Jj]uly?
 201 aug     [Aa]ug(ust)?
 202 sep     [Ss]ep(tember)?
 203 oct     [Oo]ct(ober)?
 204 nov     [Nn]ov(ember)?
 205 dec     [Dd]ec(ember)?
 206
 207 MONTH   {jan}|{feb}|{mar}|{apr}|{may}|{jun}|{jul}|{aug}|{sep}|{oct}|{nov}|{dec}
 208
 209 TIME    {D}:{d}{d}(:{d}{d})?
 210
 211      /* The year can either be 2 digits, or 4. However, after
 212         Y2K, we found that some MUA were reporting the year 100, hence
 213         the middle term here. yyterminate() resolves the actual
 214         issues with 2-digit years.
 215      */
 216
 217 YEAR    ({d}{d})|(1{d}{d})|({d}{4})
 218
 219 w       [ \t]*
 220 W       [ \t]+
 221 D       [0-9]?[0-9]
 222 d       [0-9]
 223 nl      [ \t\n()]
 224
 225 %%
 226 %{
 227   /* This section begins the definition of dparsetime().
 228      Put here any local variable definitions and initializations */
 229
 230   YY_BUFFER_STATE lexhandle;
 231
 232   char *cp;
 233   static struct tws tw;
 234
 235   ZERO(&tw);
 236
 237   lexhandle = yy_scan_string(lexstr);
 238 %}
 239
 240 {DAY}","?{W}{MONTH}{W}{D}{W}{TIME}{W}{YEAR}   {
 241                                      INIT();
 242                                      SETWDAY();
 243                                      SKIPTOA();
 244                                      SETMON();
 245                                      SKIPTOD();
 246                                      SETDAY();
 247                                      SKIPTOD();
 248                                      SETTIME();
 249                                      SKIPTOD();
 250                                      SETYEAR();
 251                                      }
 252
 253 {DAY}","?{W}{D}{W}{MONTH}{W}{YEAR}{W}{TIME}   {
 254                                      INIT();
 255                                      SETWDAY();
 256                                      SKIPTOD();
 257                                      SETDAY();
 258                                      SKIPTOA();
 259                                      SETMON();
 260                                      SKIPTOD();
 261                                      SETYEAR();
 262                                      SKIPTOD();
 263                                      SETTIME();
 264                                      }
 265 {D}{W}{MONTH}{W}{YEAR}{W}{TIME}               {
 266                                      INIT();
 267                                      SETDAY();
 268                                      SKIPTOA();
 269                                      SETMON();
 270                                      SKIPTOD();
 271                                      SETYEAR();
 272                                      SKIPTOD();
 273                                      SETTIME();
 274                                      }
 275 {DAY}","?{W}{MONTH}{W}{D}","?{W}{YEAR}","?{W}{TIME} {
 276                                      INIT();
 277                                      SETWDAY();
 278                                      SKIPTOA();
 279                                      SETMON();
 280                                      SKIPTOD();
 281                                      SETDAY();
 282                                      SKIPTOD();
 283                                      SETYEAR();
 284                                      SKIPTOD();
 285                                      SETTIME();
 286                                      }
 287 {DAY}","?{W}{MONTH}{W}{D}","?{W}{YEAR}        {
 288                                      INIT();
 289                                      SETWDAY();
 290                                      SKIPTOA();
 291                                      SETMON();
 292                                      SKIPTOD();
 293                                      SETDAY();
 294                                      SKIPTOD();
 295                                      SETYEAR();
 296                                      }
 297 {MONTH}{W}{D}","?{W}{YEAR}","?{W}{DAY}        {
 298                                      INIT();
 299                                      SETMON();
 300                                      SKIPTOD();
 301                                      SETDAY();
 302                                      SKIPTOD();
 303                                      SETYEAR();
 304                                      SKIPTOA();
 305                                      SETWDAY();
 306                                      }
 307 {MONTH}{W}{D}","?{W}{YEAR}                    {
 308                                      INIT();
 309                                      SETMON();
 310                                      SKIPTOD();
 311                                      SETDAY();
 312                                      SKIPTOD();
 313                                      SETYEAR();
 314                                      }
 315 {D}("-"|"/"){D}("-"|"/"){YEAR}{W}{TIME}       {
 316                                      INIT();
 317                                      /* MM/DD/YY */
 318                                      SETMON_NUM();
 319                                      SKIPTOD();
 320                                      SETDAY();
 321                                      SKIPTOD();
 322                                      SETYEAR();
 323                                      SKIPTOD();
 324                                      SETTIME();
 325                                      }
 326 {D}("-"|"/"){D}("-"|"/"){YEAR}                {
 327                                      INIT();
 328                                      /* MM/DD/YY */
 329                                      SETMON_NUM();
 330                                      SKIPTOD();
 331                                      SETDAY();
 332                                      SKIPTOD();
 333                                      SETYEAR();
 334                                      }
 335
 336 "[Aa][Mm]"
 337 "[Pp][Mm]"                           tw.tw_hour += 12;
 338
 339 "+"{D}{d}{d}                                  {
 340                                     INIT();
 341                                     SKIPTOD();
 342                                     SETZONE(atoi(cp));
 343                                     zonehack (&tw);
 344                                     yyterminate();
 345                                     }
 346 "-"{D}{d}{d}                                  {
 347                                     INIT();
 348                                     SKIPTOD();
 349                                     SETZONE(-atoi(cp));
 350                                     zonehack (&tw);
 351                                     yyterminate();
 352
 353                                     }
 354 {nl}("ut"|"UT")                     INIT(); SETZONE(0); yyterminate();
 355 {nl}("gmt"|"GMT")                   INIT(); SETZONE(0); yyterminate();
 356 {nl}("est"|"EST")                   INIT(); SETZONE(-500); yyterminate();
 357 {nl}("edt"|"EDT")                   { INIT(); SETDST(); SETZONE(-500);
 358                                       yyterminate(); }
 359 {nl}("cst"|"CST")                   INIT(); SETZONE(-600); yyterminate();
 360 {nl}("cdt"|"CDT")                   { INIT(); SETDST(); SETZONE(-600);
 361                                       yyterminate(); }
 362 {nl}("mst"|"MST")                   INIT(); SETZONE(-700); yyterminate();
 363 {nl}("mdt"|"MDT")                   { INIT(); SETDST(); SETZONE(-700);
 364                                       yyterminate(); }
 365 {nl}("pst"|"PST")                   INIT(); SETZONE(-800); yyterminate();
 366 {nl}("pdt"|"PDT")                   { INIT(); SETDST(); SETZONE(-800);
 367                                       yyterminate(); }
 368 {nl}("nst"|"NST")                   INIT(); SETZONE(-330); yyterminate();
 369 {nl}("ast"|"AST")                   INIT(); SETZONE(-400); yyterminate();
 370 {nl}("adt"|"ADT")                   { INIT(); SETDST(); SETZONE(-400);
 371                                       yyterminate(); }
 372 {nl}("hst"|"HST")                   INIT(); SETZONE(-1000); yyterminate();
 373 {nl}("hdt"|"HDT")                   { INIT(); SETDST(); SETZONE(-1000);
 374                                       yyterminate(); }
 375 .|\n
 376
 377 %%
 378 /* This is a portable way to squash a warning about the yyunput()
 379  * function being static but never used. It costs us a tiny amount
 380  * of extra code in the binary but the other options are:
 381  *  "%option nounput" which is flex-specific
 382  *  makefile hackery just to compile dtimep.c with different flags
 383  */
 384 void dtimep_yyunput(int c)
 385 {
 386     unput(c);
 387 }