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