src/protocol/date.c

   1 /* Parser of HTTP date */
   2
   3 #ifdef HAVE_CONFIG_H
   4 #include "config.h"
   5 #endif
   6
   7 #include <stdlib.h>
   8 #include <string.h>
   9
  10 #ifdef TIME_WITH_SYS_TIME
  11 #ifdef HAVE_SYS_TIME_H
  12 #include <sys/time.h>
  13 #endif
  14 #ifdef HAVE_TIME_H
  15 #include <time.h>
  16 #endif
  17 #else
  18 #if defined(TM_IN_SYS_TIME) && defined(HAVE_SYS_TIME_H)
  19 #include <sys/time.h>
  20 #elif defined(HAVE_TIME_H)
  21 #include <time.h>
  22 #endif
  23 #endif
  24
  25 #include "elinks.h"
  26
  27 #include "protocol/date.h"
  28 #include "util/conv.h"
  29 #include "util/time.h"
  30
  31 /*
  32  * Sun, 06 Nov 1994 08:49:37 GMT  ; RFC 822, updated by RFC 1123
  33  * Sunday, 06-Nov-94 08:49:37 GMT ; RFC 850, obsoleted by RFC 1036
  34  * Sun Nov  6 08:49:37 1994       ; ANSI C's asctime() format
  35  */
  36
  37 int
  38 parse_year(const unsigned char **date_p, unsigned char *end)
  39 {
  40         const unsigned char *date = *date_p;
  41         int year;
  42
  43         if ((end && date + 1 >= end)
  44             || !isdigit(date[0])
  45             || !isdigit(date[1]))
  46                 return -1;
  47
  48         year = (date[0] - '0') * 10 + date[1] - '0';
  49
  50         if ((!end || date + 3 < end)
  51             && isdigit(date[2])
  52             && isdigit(date[3])) {
  53                 /* Four digits date */
  54                 year = year * 10 + date[2] - '0';
  55                 year = year * 10 + date[3] - '0' - 1900;
  56                 date += 4;
  57
  58         } else if (year < 70) {
  59                 /* Assuming the epoch starting at 1.1.1970 so it's already next
  60                  * century. --wget */
  61                 year += 100;
  62                 date += 2;
  63         }
  64
  65         *date_p = date;
  66         return year;
  67 }
  68
  69 int
  70 parse_month(const unsigned char **buf, unsigned char *end)
  71 {
  72         const unsigned char *month = *buf;
  73         int monthnum;
  74
  75         /* Check that the buffer has atleast 3 chars. */
  76         if ((end && month + 2 > end)
  77             || !month[0] || !month[1] || !month[2])
  78                 return -1;
  79
  80         monthnum = month2num(month);
  81
  82         if (monthnum != -1)
  83                 *buf += 3;
  84
  85         return monthnum;
  86 }
  87
  88
  89 /* Return day number. */
  90 int
  91 parse_day(const unsigned char **date_p, unsigned char *end)
  92 {
  93         const unsigned char *date = *date_p;
  94         int day;
  95
  96         if ((end && date >= end) || !isdigit(*date))
  97                 return 32;
  98         day = *date++ - '0';
  99
 100         if ((!end || date < end) && isdigit(*date)) {
 101                 day = day * 10 + *date++ - '0';
 102         }
 103
 104         *date_p = date;
 105         return day;
 106 }
 107
 108 int
 109 parse_time(const unsigned char **time, struct tm *tm, unsigned char *end)
 110 {
 111         unsigned char h1, h2, m1, m2;
 112         const unsigned char *date = *time;
 113
 114 #define check_time(tm) \
 115         ((tm)->tm_hour <= 23 && (tm)->tm_min <= 59 && (tm)->tm_sec <= 59)
 116
 117         /* Eat HH:MM */
 118         if (end && date + 5 > end)
 119                 return 0;
 120
 121         h1 = *date++; if (!isdigit(h1)) return 0;
 122         h2 = *date++; if (!isdigit(h2)) return 0;
 123
 124         if (*date++ != ':') return 0;
 125
 126         m1 = *date++; if (!isdigit(m1)) return 0;
 127         m2 = *date++; if (!isdigit(m2)) return 0;
 128
 129         tm->tm_hour = (h1 - '0') * 10 + h2 - '0';
 130         tm->tm_min  = (m1 - '0') * 10 + m2 - '0';
 131         tm->tm_sec = 0;
 132
 133         /* Eat :SS or [PA]M or nothing */
 134         if (end && date + 2 >= end) {
 135                 *time = date;
 136                 return check_time(tm);
 137         }
 138
 139         if (*date == ':') {
 140                 unsigned char s1, s2;
 141
 142                 date++;
 143
 144                 if (end && date + 2 >= end)
 145                         return 0;
 146
 147                 s1 = *date++; if (!isdigit(s1)) return 0;
 148                 s2 = *date++; if (!isdigit(s2)) return 0;
 149
 150                 tm->tm_sec = (s1 - '0') * 10 + s2 - '0';
 151
 152         } else if (*date == 'A' || *date == 'P') {
 153                 /* Adjust hour from AM/PM. The sequence goes 11:00AM, 12:00PM,
 154                  * 01:00PM ... 11:00PM, 12:00AM, 01:00AM. --wget */
 155                 if (tm->tm_hour == 12)
 156                         tm->tm_hour = 0;
 157
 158                 if (*date++ == 'P')
 159                         tm->tm_hour += 12;
 160
 161                 if (*date++ != 'M')
 162                         return 0;
 163         }
 164
 165         *time = date;
 166
 167         return check_time(tm);
 168 }
 169
 170
 171 static time_t
 172 my_timegm(struct tm *tm)
 173 {
 174         time_t t = 0;
 175
 176         /* Okay, the next part of the code is somehow problematic. Now, we use
 177          * own code for calculating the number of seconds from 1.1.1970,
 178          * brought here by SC from w3m. I don't like it a lot, but it's 100%
 179          * portable, it's faster and it's shorter. --pasky */
 180 #if 0
 181 #ifdef HAVE_TIMEGM
 182         t = timegm(tm);
 183 #else
 184         /* Since mktime thinks we have localtime, we need a wrapper
 185          * to handle GMT. */
 186         /* FIXME: It was reported that it doesn't work somewhere :/. */
 187         {
 188                 unsigned char *tz = getenv("TZ");
 189
 190                 if (tz && *tz) {
 191                         /* Temporary disable timezone in-place. */
 192                         unsigned char tmp = *tz;
 193
 194                         *tz = '\0';
 195                         tzset();
 196
 197                         t = mktime(tm);
 198
 199                         *tz = tmp;
 200                         tzset();
 201
 202                 } else {
 203                         /* Already GMT, cool! */
 204                         t = mktime(tm);
 205                 }
 206         }
 207 #endif
 208 #else
 209         /* Following code was borrowed from w3m, and its developers probably
 210          * borrowed it from somewhere else as well, altough they didn't bother
 211          * to mention that. */ /* Actually, same code appears to be in lynx as
 212          * well.. oh well. :) */
 213         /* See README.timegm for explanation about how this works. */
 214         tm->tm_mon -= 2;
 215         if (tm->tm_mon < 0) {
 216                 tm->tm_mon += 12;
 217                 tm->tm_year--;
 218         }
 219         tm->tm_mon *= 153; tm->tm_mon += 2;
 220         tm->tm_year -= 68;
 221
 222         tm->tm_mday += tm->tm_year * 1461 / 4;
 223         tm->tm_mday += ((tm->tm_mon / 5) - 672);
 224
 225         t = ((tm->tm_mday * 60 * 60 * 24) +
 226              (tm->tm_hour * 60 * 60) +
 227              (tm->tm_min * 60) +
 228              tm->tm_sec);
 229 #endif
 230
 231         if (t == (time_t) -1) return 0;
 232
 233         return t;
 234 }
 235
 236
 237 time_t
 238 parse_date(unsigned char **date_pos, unsigned char *end,
 239            int update_pos, int skip_week_day)
 240 {
 241 #define skip_time_sep(date, end) \
 242         do { \
 243                 const unsigned char *start = (date); \
 244                 while ((!(end) || (date) < (end)) \
 245                         && (*(date) == ' ' || *(date) == '-')) \
 246                         (date)++; \
 247                 if (date == start) return 0; \
 248         } while (0)
 249
 250         struct tm tm;
 251         const unsigned char *date = (const unsigned char *) *date_pos;
 252
 253         if (!date) return 0;
 254
 255         if (skip_week_day) {
 256                 /* Skip day-of-week */
 257                 for (; (!end || date < end) && *date != ' '; date++)
 258                         if (!*date) return 0;
 259
 260                 /* As pasky said who cares if we allow '-'s here? */
 261                 skip_time_sep(date, end);
 262         }
 263
 264         if (isdigit(*date)) {
 265                 /* RFC 1036 / RFC 1123 */
 266
 267                 /* Eat day */
 268
 269                 /* date++; */
 270                 tm.tm_mday = parse_day(&date, end);
 271                 if (tm.tm_mday > 31) return 0;
 272
 273                 skip_time_sep(date, end);
 274
 275                 /* Eat month */
 276
 277                 tm.tm_mon = parse_month(&date, end);
 278                 if (tm.tm_mon < 0) return 0;
 279
 280                 skip_time_sep(date, end);
 281
 282                 /* Eat year */
 283
 284                 tm.tm_year = parse_year(&date, end);
 285                 if (tm.tm_year < 0) return 0;
 286
 287                 skip_time_sep(date, end);
 288
 289                 /* Eat time */
 290
 291                 if (!parse_time(&date, &tm, end)) return 0;
 292
 293         } else {
 294                 /* ANSI C's asctime() format */
 295
 296                 /* Eat month */
 297
 298                 tm.tm_mon = parse_month(&date, end);
 299                 if (tm.tm_mon < 0) return 0;
 300
 301                 /* I know, we shouldn't allow '-', but who cares ;). --pasky */
 302                 skip_time_sep(date, end);
 303
 304                 /* Eat day */
 305
 306                 tm.tm_mday = parse_day(&date, end);
 307                 if (tm.tm_mday > 31) return 0;
 308
 309                 skip_time_sep(date, end);
 310
 311                 /* Eat time */
 312
 313                 if (!parse_time(&date, &tm, end)) return 0;
 314
 315                 skip_time_sep(date, end);
 316
 317                 /* Eat year */
 318
 319                 tm.tm_year = parse_year(&date, end);
 320                 if (tm.tm_year < 0) return 0;
 321         }
 322 #undef skip_time_sep
 323
 324         if (update_pos)
 325                 *date_pos = (unsigned char *) date;
 326
 327         return (time_t) my_timegm(&tm);
 328 }