Tvheadend

  int                     scrape_genre; ///< Scrape genre from EIT text fields

    {

      .type   = PT_BOOL,

      .id     = "scrape_genre",

      .name   = N_("Scrape Genre"),

      .desc   = N_("Enable/disable scraping genre from the programme text fields. "

                   "Some broadcasters do not send genre information. "

                   "This allows scraping of genre "

                   "from within the broadcast text fields if supported by the "

                   "configuration file. "

                   "This is less accurate than information a broadcaster could provide "

                   "but is useful when the information is not provided. "

                   "Broadcasters that provide DVB genre information do not require "

                   "this option to be enabled."

                   ),

      .off    = offsetof(epggrab_module_ota_scraper_t, scrape_genre),

      .group  = 2,

    },

typedef struct eit_genre_regex

{

  uint8_t genre;              ///< Genre code from epg.c

  eit_pattern_list_t p_genre; ///< Regex across fields to match this genre.

} eit_genre_regex_t;

  int num_eit_genre_regex;

  eit_genre_regex_t *eit_genres;

/* Genre is handle differently to others in that we build

 * up lists of genres in the event and then afterwards if the

 * list exists we then see if the entire list has changed.

 */

static void

_eit_scrape_genre(const char *str,

                  eit_module_t *eit_mod,

                  eit_event_t *ev)

{

  if (!str || !*str) return;

  if (!eit_mod->num_eit_genre_regex) return;

  char buffer[2048];

  int i = 0;

  for (; i < eit_mod->num_eit_genre_regex; ++i) {

    eit_genre_regex_t *egr = &eit_mod->eit_genres[i];

    if (eit_pattern_apply_list(buffer, sizeof(buffer), str, &egr->p_genre)) {

      if (!ev->genre) ev->genre = calloc(1, sizeof(epg_genre_list_t));

      epg_genre_list_add_by_eit(ev->genre, egr->genre);

    }

  }

}

  if (eit_mod->scrape_genre) {

    /* Genre scraping builds up a list in ev.genre so has no

     * "scraped" value here to check.

     */

    if (ev.title)

      _eit_scrape_genre(lang_str_get(ev.title, ev.default_charset),

                        eit_mod, &ev);

    if (ev.desc)

      _eit_scrape_genre(lang_str_get(ev.desc, ev.default_charset),

                        eit_mod, &ev);

    if (ev.summary)

      _eit_scrape_genre(lang_str_get(ev.summary, ev.default_charset),

                        eit_mod, &ev);

  }

  mod->num_eit_genre_regex = 0;

  free(mod->eit_genres);

}

static void _eit_scrape_load_genre_regex(htsmsg_t *m, eit_module_t *mod)

{

  htsmsg_field_t *f;

  HTSMSG_FOREACH(f, m) {

    htsmsg_t *value;

    if (!strncmp(f->hmf_name, "genre_", 6) && (value = htsmsg_get_list_by_field(f))) {

      int genre_int;

      if (sscanf(f->hmf_name, "genre_%x", &genre_int) == 1) {

        /* One item parsed, this is the hex category code */

        if (genre_int < 0 || genre_int > 255)

          continue;

        ++mod->num_eit_genre_regex;

        mod->eit_genres = realloc(mod->eit_genres,

                                  mod->num_eit_genre_regex * sizeof(eit_genre_regex_t));

        eit_genre_regex_t *egr = &mod->eit_genres[mod->num_eit_genre_regex - 1];

        egr->genre = genre_int;

        eit_pattern_compile_list(&egr->p_genre, value);

      }

    }

  }

  if (mod->scrape_genre) {

    _eit_scrape_load_genre_regex(m, mod);

  }