pidgin/pidgin

Port purple_str_to_time to regex and add additional tests
release-2.x.y
2021-09-16, Gary Kramlich
ac6dbd45da27
Parents 4b6b116a586d
Children f7894672b4e3
Port purple_str_to_time to regex and add additional tests

Testing Done:
Ran the unit tests with the normal `./autogen.sh` as well as a setup for fuzzing.

Reviewed at https://reviews.imfreedom.org/r/923/
2 files changed, 195 insertions(+), 137 deletions(-)
  • +12 -0
    libpurple/tests/test_util.c
  • +183 -137
    libpurple/util.c
    • --- a/libpurple/tests/test_util.c Thu Sep 16 21:11:40 2021 -0500
    +++ b/libpurple/tests/test_util.c Thu Sep 16 21:11:59 2021 -0500
    @@ -162,13 +162,21 @@
    time_t timestamp;
    fail_unless(377182200 == purple_str_to_time("19811214T12:50:00", TRUE, NULL, NULL, NULL));
    + fail_unless(377203800 == purple_str_to_time("19811214T12:50:00-06", FALSE, NULL, NULL, NULL));
    fail_unless(1175919261 == purple_str_to_time("20070407T04:14:21", TRUE, NULL, NULL, NULL));
    fail_unless(1282941722 == purple_str_to_time("2010-08-27.204202", TRUE, NULL, NULL, NULL));
    + fail_unless(1175919261 == purple_str_to_time("20070407T04:14:21.3234", TRUE, NULL, NULL, NULL));
    + fail_unless(1175919261 == purple_str_to_time("20070407T04:14:21Z", TRUE, NULL, NULL, NULL));
    + fail_unless(1631512800 == purple_str_to_time("09-13-2021", TRUE, NULL, NULL, NULL));
    timestamp = purple_str_to_time("2010-08-27.134202-0700PDT", FALSE, &tm, &tz_off, &rest);
    fail_unless(1282941722 == timestamp);
    fail_unless((-7 * 60 * 60) == tz_off);
    assert_string_equal("PDT", rest);
    +
    + timestamp = purple_str_to_time("09/13/202115:34:34", TRUE, NULL, NULL, &rest);
    + fail_unless(1631512800 == timestamp);
    + assert_string_equal("15:34:34", rest);
    }
    END_TEST
    @@ -253,6 +261,8 @@
    Suite *
    util_suite(void)
    {
    + purple_util_init();
    +
    Suite *s = suite_create("Utility Functions");
    TCase *tc = tcase_create("Base16");
    @@ -306,5 +316,7 @@
    tcase_add_test(tc, test_uri_escape_for_open);
    suite_add_tcase(s, tc);
    + purple_util_uninit();
    +
    return s;
    }
    --- a/libpurple/util.c Thu Sep 16 21:11:40 2021 -0500
    +++ b/libpurple/util.c Thu Sep 16 21:11:59 2021 -0500
    @@ -84,6 +84,40 @@
    static char *custom_user_dir = NULL;
    static char *user_dir = NULL;
    +static GRegex *str_to_time_regex = NULL;
    +static const gchar *str_to_time_pattern =
    + "^\\s*"
    + "(?P<year>\\d{4})? # look for a leading year\n"
    + "(?:[-\\/]?) # an optional separator of - or /\n"
    + "(?P<month>\\d{2}) # the two digit month\n"
    + "(?:[-\\/]?) # another optional separator of - or /\n"
    + "(?P<day>\\d{2}) # the two digit day\n"
    + "# we now have an optional trailing year or seconds\n"
    + "(?:\n"
    + " (?:[-\\/]?(?P<trailingyear>\\d{4})) # the trailing year may have a - or / separator\n"
    + "|\n"
    + " [T.] # T signifies that this is an iso8601 time\n"
    + " (?:\n"
    + " (?P<hours>\\d{2}) # two digit hour\n"
    + " :? # optional : separator\n"
    + " (?P<minutes>\\d{2}) # two digit minutes\n"
    + " :? # optional : separator\n"
    + " (?P<seconds>\\d{2}) # two digit seconds\n"
    + " (?:\\.(?P<microseconds>\\d+))? # optional microseconds\n"
    + " (?:\n"
    + " (?:\n"
    + " (?P<tzsign>[+-]) # required tz sign\n"
    + " (?P<tzhour>\\d{2}) # required tz hour\n"
    + " (?:\n"
    + " :? # optional : separator\n"
    + " (?<tzminute>\\d{2}))? # optional tz minutes\n"
    + " )\n"
    + " |\n"
    + " (?P<utc>Z) # utc time\n"
    + " )?\n"
    + " )?\n"
    + ")\n"
    + "\\s* # remove trailing whitespace\n";
    PurpleMenuAction *
    purple_menu_action_new(const char *label, PurpleCallback callback, gpointer data,
    @@ -109,8 +143,19 @@
    void
    purple_util_init(void)
    {
    - /* This does nothing right now. It exists for symmetry with
    - * purple_util_uninit() and forwards compatibility. */
    + GError *error = NULL;
    +
    + str_to_time_regex = g_regex_new(str_to_time_pattern,
    + G_REGEX_EXTENDED | G_REGEX_RAW, 0,
    + &error);
    +
    + if(error != NULL) {
    + purple_debug_fatal("util",
    + "Failed to compile the regex for purple_str_to_time"
    + ": %s",
    + error->message ? error->message : "unknown");
    + g_clear_error(&error);
    + }
    }
    void
    @@ -123,6 +168,9 @@
    g_free(user_dir);
    user_dir = NULL;
    +
    + g_regex_unref(str_to_time_regex);
    + str_to_time_regex = NULL;
    }
    /**************************************************************************
    @@ -705,169 +753,167 @@
    purple_str_to_time(const char *timestamp, gboolean utc,
    struct tm *tm, long *tz_off, const char **rest)
    {
    + GMatchInfo *info = NULL;
    + gboolean mktime_with_utc = FALSE, matched = FALSE;
    + gchar *match = NULL;
    struct tm t;
    - const gchar *str;
    - gint year = 0;
    long tzoff = PURPLE_NO_TZ_OFF;
    time_t retval;
    - gboolean mktime_with_utc = FALSE;
    -
    - if (rest != NULL)
    + gint end = 0;
    +
    + if(rest != NULL) {
    *rest = NULL;
    + }
    g_return_val_if_fail(timestamp != NULL, 0);
    memset(&t, 0, sizeof(struct tm));
    - str = timestamp;
    -
    - /* Strip leading whitespace */
    - while (g_ascii_isspace(*str))
    - str++;
    -
    - if (*str == '\0') {
    - if (rest != NULL && *str != '\0')
    - *rest = str;
    -
    - return 0;
    - }
    -
    - if (!g_ascii_isdigit(*str) && *str != '-' && *str != '+') {
    - if (rest != NULL && *str != '\0')
    - *rest = str;
    -
    - return 0;
    - }
    -
    - /* 4 digit year */
    - if (sscanf(str, "%04d", &year) && year >= 1900) {
    - str += 4;
    -
    - if (*str == '-' || *str == '/')
    - str++;
    -
    - t.tm_year = year - 1900;
    - }
    -
    - /* 2 digit month */
    - if (!sscanf(str, "%02d", &t.tm_mon)) {
    - if (rest != NULL && *str != '\0')
    - *rest = str;
    -
    - return 0;
    - }
    -
    - str += 2;
    - t.tm_mon -= 1;
    -
    - if (*str == '-' || *str == '/')
    - str++;
    -
    - /* 2 digit day */
    - if (!sscanf(str, "%02d", &t.tm_mday)) {
    - if (rest != NULL && *str != '\0')
    - *rest = str;
    + matched = g_regex_match(str_to_time_regex, timestamp, 0, &info);
    + if(!matched) {
    + if(rest != NULL) {
    + *rest = timestamp;
    + }
    +
    + g_match_info_free(info);
    return 0;
    }
    - str += 2;
    -
    - /* Grab the year off the end if there's still stuff */
    - if (*str == '/' || *str == '-') {
    - /* But make sure we don't read the year twice */
    - if (year >= 1900) {
    - if (rest != NULL && *str != '\0')
    - *rest = str;
    -
    - return 0;
    + /* Set rest to the end of the match in timestamp. */
    + if(rest != NULL) {
    + if(g_match_info_fetch_pos(info, 0, NULL, &end)) {
    + *rest = timestamp + end;
    }
    -
    - str++;
    -
    - if (!sscanf(str, "%04d", &t.tm_year)) {
    - if (rest != NULL && *str != '\0')
    - *rest = str;
    -
    - return 0;
    + }
    +
    + /* check the year or the trailingyear if year is not set */
    + match = g_match_info_fetch_named(info, "year");
    + if(match != NULL && *match != '\0') {
    + gint year = atoi(match);
    + if(year >= 1900) {
    + t.tm_year = year - 1900;
    + }
    + } else {
    + g_free(match);
    + match = g_match_info_fetch_named(info, "trailingyear");
    + if(match != NULL && *match != '\0') {
    + t.tm_year = atoi(match) - 1900;
    }
    -
    - t.tm_year -= 1900;
    - } else if (*str == 'T' || *str == '.') {
    - str++;
    -
    - /* Continue grabbing the hours/minutes/seconds */
    - if ((sscanf(str, "%02d:%02d:%02d", &t.tm_hour, &t.tm_min, &t.tm_sec) == 3 &&
    - (str += 8)) ||
    - (sscanf(str, "%02d%02d%02d", &t.tm_hour, &t.tm_min, &t.tm_sec) == 3 &&
    - (str += 6)))
    - {
    - gint sign, tzhrs, tzmins;
    -
    - if (*str == '.') {
    - /* Cut off those pesky micro-seconds */
    - do {
    - str++;
    - } while (*str >= '0' && *str <= '9');
    + }
    + g_free(match);
    +
    + /* get the month */
    + match = g_match_info_fetch_named(info, "month");
    + if(match != NULL && *match != '\0') {
    + t.tm_mon = atoi(match) - 1;
    + }
    + g_free(match);
    +
    + /* get the day */
    + match = g_match_info_fetch_named(info, "day");
    + if(match != NULL && *match != '\0') {
    + t.tm_mday = atoi(match);
    + }
    + g_free(match);
    +
    + /* get the hour */
    + match = g_match_info_fetch_named(info, "hours");
    + if(match != NULL && *match != '\0') {
    + t.tm_hour = atoi(match);
    + }
    + g_free(match);
    +
    + /* get the minutes */
    + match = g_match_info_fetch_named(info, "minutes");
    + if(match != NULL && *match != '\0') {
    + t.tm_min = atoi(match);
    + }
    + g_free(match);
    +
    + /* get the seconds */
    + match = g_match_info_fetch_named(info, "seconds");
    + if(match != NULL && *match != '\0') {
    + t.tm_sec = atoi(match);
    + }
    + g_free(match);
    +
    + /* check if this is utc time */
    + match = g_match_info_fetch_named(info, "utc");
    + if(match != NULL && *match != '\0') {
    + tzoff = 0;
    + g_free(match);
    + } else {
    + /* free match if it was just an empty string */
    + g_free(match);
    +
    + /* if match is null, check if we have tzsign which is required */
    + match = g_match_info_fetch_named(info, "tzsign");
    + if(match != NULL) {
    + gint tzsign = -1, tzhour = 0, tzminute = 0;
    +
    + /* Figure out if we're ahead or behind of utc, assuming ahead by
    + * default.
    + */
    + if(*match == '+') {
    + tzsign = -1;
    }
    -
    - sign = (*str == '+') ? 1 : -1;
    -
    - /* Process the timezone */
    - if (*str == '+' || *str == '-') {
    - str++;
    -
    - if (((sscanf(str, "%02d:%02d", &tzhrs, &tzmins) == 2 && (str += 5)) ||
    - (sscanf(str, "%02d%02d", &tzhrs, &tzmins) == 2 && (str += 4))))
    - {
    - mktime_with_utc = TRUE;
    - tzoff = tzhrs * 60 * 60 + tzmins * 60;
    - tzoff *= sign;
    - }
    - } else if (*str == 'Z') {
    - /* 'Z' = Zulu = UTC */
    - str++;
    + g_free(match);
    +
    + /* get the tz hour */
    + match = g_match_info_fetch_named(info, "tzhour");
    + if(match != NULL && *match != '\0') {
    + tzhour = atoi(match);
    + }
    + g_free(match);
    +
    + /* get the tz minute */
    + match = g_match_info_fetch_named(info, "tzminute");
    + if(match != NULL && *match != '\0') {
    + tzminute = atoi(match);
    + }
    + g_free(match);
    +
    + /* we need at least either an hour or minute to offset the timezone */
    + if(tzhour > 0 || tzminute > 0) {
    mktime_with_utc = TRUE;
    - tzoff = 0;
    - }
    -
    - if (!mktime_with_utc)
    - {
    - /* No timezone specified. */
    -
    - if (utc) {
    - mktime_with_utc = TRUE;
    - tzoff = 0;
    - } else {
    - /* Local Time */
    - t.tm_isdst = -1;
    - }
    + tzoff = tzsign * ((tzhour * 3600) + (tzminute * 60));
    }
    }
    }
    - if (rest != NULL && *str != '\0') {
    - /* Strip trailing whitespace */
    - while (g_ascii_isspace(*str))
    - str++;
    -
    - if (*str != '\0')
    - *rest = str;
    + /* If we have a time, figure out if we need to adjust our tz offset. */
    + if(t.tm_hour > 0 || t.tm_min > 0 || t.tm_sec > 0) {
    + if(!mktime_with_utc) {
    + if(utc) {
    + mktime_with_utc = TRUE;
    + tzoff = 0;
    + } else {
    + /* Local Time */
    + t.tm_isdst = -1;
    + }
    + }
    }
    - if (mktime_with_utc)
    + if(mktime_with_utc) {
    retval = mktime_utc(&t);
    - else
    + } else {
    retval = mktime(&t);
    -
    - if (tm != NULL)
    + }
    +
    + if(tm != NULL) {
    *tm = t;
    -
    - if (tzoff != PURPLE_NO_TZ_OFF)
    + }
    +
    + if(tzoff != PURPLE_NO_TZ_OFF) {
    retval -= tzoff;
    -
    - if (tz_off != NULL)
    + }
    +
    + if(tz_off != NULL) {
    *tz_off = tzoff;
    + }
    +
    + g_match_info_free(info);
    return retval;
    }