Whoops, fix a typo in my last commit.
[mmh] / sbr / dtimep.l
1 /* dtimep.lex exceeds the default table capacities for some old versions
2  * of lex (and the minimum defaults as specified by POSIX).  The following
3  * choices meet or exceed the lex defaults for older SunOS4.x, Solaris,
4  * HPUX, and AIX.
5  */
6 %e4000
7 %p7000
8 %n2500
9 %a5000
10 %{
11 #include <h/nmh.h>
12 #include <h/tws.h>
13
14   /* Since we're looking at a string at a time, don't worry about
15    *  wrapping to the next buffer.
16    */
17 #define yywrap() 1
18 #define YY_SKIP_YYWRAP
19
20 #define YY_NO_INPUT
21
22   /* This is the tricky thing that makes this function cool.  We
23    *  replace the traditional int yylex(void) declaration with our
24    *  dparsetime() declaration, essentially piggy-backing off the
25    *  utility of the yylex() function and adding what we need to make
26    *  the parsing function useful to us.
27    */
28 #define YY_DECL struct tws *dparsetime(char *lexstr)
29
30   /* yyerminate() is called after the input string is matched to
31    * completion (actually, when the lexer reaches an EOF).  The only
32    * thing that really needs to be in this macro function is the
33    * return call, which must be substituted inline into dparsetime.
34    */
35
36 #define yyterminate() (void)yy_delete_buffer(lexhandle); \
37   if(!(tw.tw_flags & TW_SUCC)) { \
38     return (struct tws *)NULL; \
39   } \
40   if(tw.tw_year < 1970) \
41     tw.tw_year += 1900; \
42   if(tw.tw_year < 1970) \
43     tw.tw_year += 100; \
44   return(&tw)
45
46 /*
47  * Patchable flag that says how to interpret NN/NN/NN dates. When
48  * true, we do it European style: DD/MM/YY. When false, we do it
49  * American style: MM/DD/YY.  Of course, these are all non-RFC822
50  * compliant.
51  */
52 int europeandate = 0;
53
54 /*
55  * Table to convert month names to numeric month.  We use the
56  * fact that the low order 5 bits of the sum of the 2nd & 3rd
57  * characters of the name is a hash with no collisions for the 12
58  * valid month names.  (The mask to 5 bits maps any combination of
59  * upper and lower case into the same hash value).
60  */
61 static int month_map[] = {
62         0,
63         6,      /* 1 - Jul */
64         3,      /* 2 - Apr */
65         5,      /* 3 - Jun */
66         0,
67         10,     /* 5 - Nov */
68         0,
69         1,      /* 7 - Feb */
70         11,     /* 8 - Dec */
71         0,
72         0,
73         0,
74         0,
75         0,
76         0,
77         0,      /*15 - Jan */
78         0,
79         0,
80         0,
81         2,      /*19 - Mar */
82         0,
83         8,      /*21 - Sep */
84         0,
85         9,      /*23 - Oct */
86         0,
87         0,
88         4,      /*26 - May */
89         0,
90         7       /*28 - Aug */
91 };
92
93 /*
94  * Lookup table for day-of-week using the same hash trick as for above name-of-
95  * month table, but using the first and second character, not second and third.
96  *
97  * Compute index into table using: (day_name[0] & 7) + (day_name[1] & 4)
98  */
99 static int day_map[] = {
100         0,
101         0,
102         0,
103         6,      /* 3 - Sat */
104         4,      /* 4 - Thu */
105         0,
106         5,      /* 6 - Fri */
107         0,      /* 7 - Sun */
108         2,      /* 8 - Tue */
109         1       /* 9 - Mon */,
110         0,
111         3       /*11 - Wed */
112 };
113
114 /* The SET* macros will parse for the appropriate field, and leave the
115  * cp pointer at the first character after the desired field. Be
116  * careful with variable-length fields or alpha-num mixes.
117
118  * The SKIP* macros skip over characters of a particular class and
119  * leave cp at the position of the first character that doesn't match
120  * that class. Correspondingly, SKIPTO* skips until it reaches a
121  * character of a particular class.
122  */
123
124 #define INIT()       { cp = yytext;} 
125 #define SETWDAY()    { tw.tw_wday= day_map[(cp[0] & 7) + (cp[1] & 4)]; \
126                        tw.tw_flags &= ~TW_SDAY; tw.tw_flags |= TW_SEXP; \
127                        SKIPA(); }
128 #define SETMON()     { cp++; \
129                        tw.tw_mon = month_map[(cp[0] + cp[1]) & 0x1f]; \
130                        SKIPA(); }
131 #define SETMON_NUM() { tw.tw_mon = atoi(cp)-1; \
132                        SKIPD(); }
133 #define SETYEAR()    { tw.tw_year = atoi(cp); \
134                        SKIPD(); }
135 #define SETDAY()     { tw.tw_mday = atoi(cp); \
136                        tw.tw_flags |= TW_YES; \
137                        SKIPD(); }
138 #define SETTIME()    { tw.tw_hour = atoi(cp); \
139                        cp += 2; \
140                        SKIPTOD(); \
141                        tw.tw_min = atoi(cp); \
142                        cp += 2; \
143                        if(*cp == ':') { \
144                           tw.tw_sec = atoi(++cp); SKIPD(); } }
145 #define SETZONE(x)   { tw.tw_zone = ((x)/100)*60+(x)%100; \
146                        tw.tw_flags |= TW_SZEXP; \
147                        SKIPD(); }
148 #define SETDST()     { tw.tw_flags |= TW_DST; }
149 #define SKIPD()      { while ( isdigit(*cp++) ) ; \
150                        --cp; }
151 #define SKIPTOD()    { while ( !isdigit(*cp++) ) ; \
152                        --cp; }
153 #define SKIPA()      { while ( isalpha(*cp++) ) ; \
154                        --cp; }
155 #define SKIPTOA()    { while ( !isalpha(*cp++) ) ; \
156                        --cp; }
157 #define SKIPSP()     { while ( isspace(*cp++) ) ; \
158                        --cp; }
159 #define SKIPTOSP()   { while ( !isspace(*cp++) ) ; \
160                        --cp; }
161
162 #ifdef ADJUST_NUMERIC_ONLY_TZ_OFFSETS_WRT_DST
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     register 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 #endif  /* ADJUST_NUMERIC_ONLY_TZ_OFFSETS_WRT_DST */
183 %}
184
185 sun     ([Ss]un(day)?)
186 mon     ([Mm]on(day)?)
187 tue     ([Tt]ue(sday)?)
188 wed     ([Ww]ed(nesday)?)
189 thu     ([Tt]hu(rsday)?)
190 fri     ([Ff]ri(day)?)
191 sat     ([Ss]at(urday)?)
192
193 DAY     ({sun}|{mon}|{tue}|{wed}|{thu}|{fri}|{sat})
194
195 jan     ([Jj]an(uary)?)
196 feb     ([Ff]eb(ruary)?)
197 mar     ([Mm]ar(ch)?)
198 apr     ([Aa]pr(il)?)
199 may     ([Mm]ay)
200 jun     ([Jj]un(e)?)
201 jul     ([Jj]ul(y)?)
202 aug     ([Aa]ug(ust)?)
203 sep     ([Ss]ep(tember)?)
204 oct     ([Oo]ct(ober)?)
205 nov     ([Nn]ov(ember)?)
206 dec     ([Dd]ec(ember)?)
207
208 MONTH   ({jan}|{feb}|{mar}|{apr}|{may}|{jun}|{jul}|{aug}|{sep}|{oct}|{nov}|{dec})
209
210 TIME    ({D}:{d}{d}(:{d}{d})?)
211
212      /* The year can either be 2 digits, or 4. However, after
213         Y2K, we found that some MUA were reporting the year 100, hence
214         the middle term here. yyterminate() resolves the actual
215         issues with 2-digit years.
216      */
217
218 YEAR    (({d}{d})|(1{d}{d})|({d}{4}))
219
220 w       ([ \t]*)
221 W       ([ \t]+)
222 D       ([0-9]?[0-9])
223 d       [0-9]
224 nl      [ \t\n()]
225
226 %%
227 %{
228   /* This section begins the definition of dparsetime().
229      Put here any local variable definitions and initializations */
230   
231   YY_BUFFER_STATE lexhandle;
232
233   register unsigned char *cp;
234   static struct tws tw; 
235
236   memset(&tw,0,sizeof(struct tws));
237
238   lexhandle = yy_scan_string(lexstr);
239 %}
240
241 {DAY}","?{W}{MONTH}{W}{D}{W}{TIME}{W}{YEAR}   {
242                                      INIT();
243                                      SETWDAY();
244                                      SKIPTOA();
245                                      SETMON();
246                                      SKIPTOD();
247                                      SETDAY();
248                                      SKIPTOD();
249                                      SETTIME();
250                                      SKIPTOD();
251                                      SETYEAR();
252                                      }
253
254 {DAY}","?{W}{D}{W}{MONTH}{W}{YEAR}{W}{TIME}   {
255                                      INIT();
256                                      SETWDAY();
257                                      SKIPTOD();
258                                      SETDAY();
259                                      SKIPTOA();
260                                      SETMON();
261                                      SKIPTOD();
262                                      SETYEAR();
263                                      SKIPTOD();
264                                      SETTIME();
265                                      }
266 {D}{W}{MONTH}{W}{YEAR}{W}{TIME}               {
267                                      INIT();
268                                      SETDAY();
269                                      SKIPTOA();
270                                      SETMON();
271                                      SKIPTOD();
272                                      SETYEAR();
273                                      SKIPTOD();
274                                      SETTIME();
275                                      }
276 {DAY}","?{W}{MONTH}{W}{D}","?{W}{YEAR}","?{W}{TIME} {
277                                      INIT();
278                                      SETWDAY();
279                                      SKIPTOA();
280                                      SETMON();
281                                      SKIPTOD();
282                                      SETDAY();
283                                      SKIPTOD();
284                                      SETYEAR();
285                                      SKIPTOD();
286                                      SETTIME();
287                                      }
288 {DAY}","?{W}{MONTH}{W}{D}","?{W}{YEAR}        {
289                                      INIT();
290                                      SETWDAY();
291                                      SKIPTOA();
292                                      SETMON();
293                                      SKIPTOD();
294                                      SETDAY();
295                                      SKIPTOD();
296                                      SETYEAR();
297                                      }
298 {MONTH}{W}{D}","?{W}{YEAR}","?{W}{DAY}        {
299                                      INIT();
300                                      SETMON();
301                                      SKIPTOD();
302                                      SETDAY();
303                                      SKIPTOD();
304                                      SETYEAR();
305                                      SKIPTOA();
306                                      SETWDAY();
307                                      }
308 {MONTH}{W}{D}","?{W}{YEAR}                    {
309                                      INIT();
310                                      SETMON();
311                                      SKIPTOD();
312                                      SETDAY();
313                                      SKIPTOD();
314                                      SETYEAR();
315                                      }
316 {D}("-"|"/"){D}("-"|"/"){YEAR}{W}{TIME}       {
317                                      INIT();
318                                      if(europeandate) {
319                                        /* DD/MM/YY */
320                                      SETDAY();
321                                      SKIPTOD();
322                                      SETMON_NUM();
323                                      } else {
324                                        /* MM/DD/YY */
325                                      SETMON_NUM();
326                                      SKIPTOD();
327                                      SETDAY();
328                                      }
329                                      SKIPTOD();
330                                      SETYEAR();
331                                      SKIPTOD();
332                                      SETTIME();
333                                      }
334 {D}("-"|"/"){D}("-"|"/"){YEAR}                {
335                                      INIT();
336                                      if(europeandate) {
337                                        /* DD/MM/YY */
338                                      SETDAY();
339                                      SKIPTOD();
340                                      SETMON_NUM();
341                                      } else {
342                                        /* MM/DD/YY */
343                                      SETMON_NUM();
344                                      SKIPTOD();
345                                      SETDAY();
346                                      }
347                                      SKIPTOD();
348                                      SETYEAR();
349                                      }
350
351 "[Aa][Mm]"
352 "[Pp][Mm]"                           tw.tw_hour += 12;
353
354 "+"{D}{d}{d}                                  {
355                                     INIT();
356                                     SKIPTOD();
357                                     SETZONE(atoi(cp));
358 #ifdef  ADJUST_NUMERIC_ONLY_TZ_OFFSETS_WRT_DST
359                                     zonehack (&tw);
360 #endif  /* ADJUST_NUMERIC_ONLY_TZ_OFFSETS_WRT_DST */
361                                     yyterminate();
362                                     }
363 "-"{D}{d}{d}                                  {
364                                     INIT();
365                                     SKIPTOD();
366                                     SETZONE(-atoi(cp));
367 #ifdef  ADJUST_NUMERIC_ONLY_TZ_OFFSETS_WRT_DST
368                                     zonehack (&tw);
369 #endif  /* ADJUST_NUMERIC_ONLY_TZ_OFFSETS_WRT_DST */
370                                     yyterminate();
371                                     
372                                     }
373 {nl}("ut"|"UT")                     INIT(); SETZONE(0); yyterminate();
374 {nl}("gmt"|"GMT")                   INIT(); SETZONE(0); yyterminate();
375 {nl}("est"|"EST")                   INIT(); SETZONE(-500); yyterminate();
376 {nl}("edt"|"EDT")                   { INIT(); SETDST(); SETZONE(-500);
377                                       yyterminate(); }
378 {nl}("cst"|"CST")                   INIT(); SETZONE(-600); yyterminate();
379 {nl}("cdt"|"CDT")                   { INIT(); SETDST(); SETZONE(-600);
380                                       yyterminate(); }
381 {nl}("mst"|"MST")                   INIT(); SETZONE(-700); yyterminate();
382 {nl}("mdt"|"MDT")                   { INIT(); SETDST(); SETZONE(-700);
383                                       yyterminate(); }
384 {nl}("pst"|"PST")                   INIT(); SETZONE(-800); yyterminate();
385 {nl}("pdt"|"PDT")                   { INIT(); SETDST(); SETZONE(-800);
386                                       yyterminate(); }
387 {nl}("nst"|"NST")                   INIT(); SETZONE(-330); yyterminate();
388 {nl}("ast"|"AST")                   INIT(); SETZONE(-400); yyterminate();
389 {nl}("adt"|"ADT")                   { INIT(); SETDST(); SETZONE(-400);
390                                       yyterminate(); }
391 {nl}("hst"|"HST")                   INIT(); SETZONE(-1000); yyterminate();
392 {nl}("hdt"|"HDT")                   { INIT(); SETDST(); SETZONE(-1000);
393                                       yyterminate(); }
394 .|\n
395
396 %%
397 /* This is a portable way to squash a warning about the yyunput()
398  * function being static but never used. It costs us a tiny amount
399  * of extra code in the binary but the other options are:
400  *  "%option nounput" which is flex-specific
401  *  makefile hackery just to compile dtimep.c with different flags
402  */
403 void dtimep_yyunput(int c)
404 {
405     unput(c);
406 }
407