Add some of the missing stuff to xmltv and EIT. I think the only significant issue...

diff --git a/src/epg.c b/src/epg.c
index d459702eaf4daf311b027f557b1969c0c5ab3c9e..ca59845334fb4334594365ed9cb976d16c68dac9 100644 (file)
--- a/src/epg.c
+++ b/src/epg.c
@@ -959,9 +959,14 @@ int epg_episode_fuzzy_match
   ( epg_episode_t *episode, const char *uri, const char *title,
     const char *summary, const char *description )
 {
+  // TODO: this is pretty noddy and likely to fail!
+  //       hence the reason I don't recommend mixing external grabbers and EIT
   if ( !episode ) return 0;
   if ( uri && episode->_.uri && !strcmp(episode->_.uri, uri) ) return 1;
-  // TODO: how to match title/summary/description
+  if ( title && episode->title && (strstr(title, episode->title) || strstr(episode->title, title)) ) return 1;
+  // TODO: could we do fuzzy string matching on the description/summary
+  //     : there are a few algorithms that might work, but some early testing
+  //     : suggested it wasn't clear cut enough to make sensible decisions.
   return 0;
 }
 

diff --git a/src/epggrab/eit.c b/src/epggrab/eit.c
index 6bbcb167a745305e942fda3dd1dfad0a5099431b..2a828792cab8950012b350fdf04e3701e5b2579d 100644 (file)
--- a/src/epggrab/eit.c
+++ b/src/epggrab/eit.c
@@ -26,11 +26,12 @@
  * Module Setup
  * ***********************************************************************/
 
-static void _eit_episode_uri 
-  ( char *uri, const char *title, const char *summary )
+static const char *
+longest_string ( const char *a, const char *b )
 {
-  // TODO: do something better
-  snprintf(uri, 1023, "%s::%s", title, summary);
+  if (!a) return b;
+  if (!b) return a;
+  if (strlen(a) - strlen(b) >= 0) return a;
 }
 
 // called from dvb_tables.c
@@ -44,10 +45,11 @@ void eit_callback ( channel_t *ch, int id, time_t start, time_t stop,
   epg_episode_t *ee;
   const char *summary     = NULL;
   const char *description = NULL;
-  char uri[1024];
+  char *uri;
 
   /* Ignore */
   if (!ch || !ch->ch_name || !ch->ch_name[0]) return;
+  if (!title) return;
 
   /* Disabled? */
   if (!epggrab_eitenabled) return;
@@ -56,11 +58,16 @@ void eit_callback ( channel_t *ch, int id, time_t start, time_t stop,
   ebc  = epg_broadcast_find_by_time(ch, start, stop, 1, &save);
   if (!ebc) return;
 
-  /* TODO: Determine summary */
-  summary = desc;
+  /* Determine summary */
+  description = summary = desc;
+  description = longest_string(description, extitem);
+  description = longest_string(description, extdesc);
+  description = longest_string(description, exttext);
+  if (summary == description) description = NULL;
+  // TODO: is this correct?
 
   /* Create episode URI */
-  _eit_episode_uri(uri, title, summary);
+  uri = md5sum(longest_string(title, longest_string(description, summary)));
 
   /* Create/Replace episode */
   if ( !ebc->episode ||
@@ -83,6 +90,7 @@ void eit_callback ( channel_t *ch, int id, time_t start, time_t stop,
     /* Update */
     save |= epg_broadcast_set_episode(ebc, ee);
   }
+  free(uri);
 }
 
 /* ************************************************************************

diff --git a/src/epggrab/xmltv.c b/src/epggrab/xmltv.c
index 4823a2a02baded37063c00add4596428c0ccce99..78b5671edb5f52c4c330e95b1c32cd44ca7bda18 100644 (file)
--- a/src/epggrab/xmltv.c
+++ b/src/epggrab/xmltv.c
@@ -51,25 +51,6 @@ static epggrab_channel_t *_xmltv_channel_find
  * Parsing
  * *************************************************************************/
 
-/**
- * Hash the description to get a URI for episode
- *
- * Note: converts to an ASCII format
- * TODO: move somewhere else
- */
-static const char *_xmltv_hash ( const char *str )
-{
-  size_t i;
-  static char ret[(MD5_DIGEST_LENGTH*2)+1];
-  static unsigned char md5[MD5_DIGEST_LENGTH];
-  (void)MD5((const unsigned char*)str, strlen(str), md5);
-  for ( i = 0; i < MD5_DIGEST_LENGTH; i++ ) {
-    sprintf(&ret[i*2], "%02X", md5[i]);
-  }
-  ret[MD5_DIGEST_LENGTH*2] = '\0';
-  return ret;
-}
-
 /**
  *
  */
@@ -181,8 +162,8 @@ xmltv_ns_get_parse_num(const char *s, int *ap, int *bp)
 
 
  out:
-  if(ap) *ap = a;
-  if(bp) *bp = b;
+  if(ap) *ap = a + 1;
+  if(bp) *bp = b + 1;
   return s;
 }
 
@@ -191,15 +172,11 @@ xmltv_ns_get_parse_num(const char *s, int *ap, int *bp)
 
 static void
 parse_xmltv_ns_episode  
-(const char *s, int *season, int *episode, int *part)
+(const char *s, int *sn, int *sc, int *en, int *ec, int *pn, int *pc)
 {
-  s = xmltv_ns_get_parse_num(s, season, NULL);
-  s = xmltv_ns_get_parse_num(s, episode, NULL);
-  xmltv_ns_get_parse_num(s, part, NULL);
-
-  *season += 1;
-  *episode += 1;
-  *part += 1;
+  s = xmltv_ns_get_parse_num(s, sn, sc);
+  s = xmltv_ns_get_parse_num(s, en, ec);
+  xmltv_ns_get_parse_num(s, pn, pc);
 }
 
 /**
@@ -207,7 +184,7 @@ parse_xmltv_ns_episode
  */
 static void
 get_episode_info
-(htsmsg_t *tags, const char **screen, int *season, int *episode, int *part)
+(htsmsg_t *tags, const char **screen, int *sn, int *sc, int *en, int *ec, int *pn, int *pc)
 {
   htsmsg_field_t *f;
   htsmsg_t *c, *a;
@@ -223,7 +200,7 @@ get_episode_info
     
     if(!strcmp(sys, "onscreen")) *screen = cdata;
     else if(!strcmp(sys, "xmltv_ns"))
-      parse_xmltv_ns_episode(cdata, season, episode, part);
+      parse_xmltv_ns_episode(cdata, sn, sc, en, ec, pn, pc);
   }
 }
 
@@ -237,54 +214,50 @@ _xmltv_parse_programme_tags(channel_t *ch, htsmsg_t *tags,
   int save = 0, save2 = 0;
   epg_episode_t *ee;
   epg_broadcast_t *ebc;
-  int sn = 0, en = 0, pn = 0;
+  int sn = 0, sc = 0, en = 0, ec = 0, pn = 0, pc = 0;
   const char *onscreen = NULL;
-  const char *uri = NULL;
+  char *uri;
   const char *title = htsmsg_xml_get_cdata_str(tags, "title");
   const char *desc  = htsmsg_xml_get_cdata_str(tags, "desc");
-  //const char *category = htsmsg_xml_get_cdata_str(tags, "category");
-  get_episode_info(tags, &onscreen, &sn, &en, &pn);
+  const char *category[2];
+  get_episode_info(tags, &onscreen, &sn, &sc, &en, &ec, &pn, &pc);
+
+  /* Ignore */
+  if (!title) return 0;
+
+  /* Build episode */
+  uri = md5sum(desc ?: title);
+  ee  = epg_episode_find_by_uri(uri, 1, &save);
+  free(uri);
+  if (!ee) return 0;
+  stats->episodes.total++;
+  if (save) stats->episodes.created++;
+
+  category[0] = htsmsg_xml_get_cdata_str(tags, "category");
+  category[1] = NULL;
+  if (title)     save |= epg_episode_set_title(ee, title);
+  if (desc)      save |= epg_episode_set_description(ee, desc);
+  if (*category) save |= epg_episode_set_genre_str(ee, category);
+  if (pn)        save |= epg_episode_set_part(ee, pn, pc);
+  if (en)        save |= epg_episode_set_number(ee, en);
+  if (save) stats->episodes.modified++;
+
+  // TODO: need to handle season numbering!
+  // TODO: need to handle onscreen numbering
+  //if (onscreen) save |= epg_episode_set_onscreen(ee, onscreen);
 
   /* Create/Find broadcast */
   ebc = epg_broadcast_find_by_time(ch, start, stop, 1, &save2);
   if ( ebc != NULL ) {
     stats->broadcasts.total++;
     if (save2) stats->broadcasts.created++;
-    //save2 |= epg_broadcast_set_episode(ebc, ee);
+    save2 |= epg_broadcast_set_episode(ebc, ee);
     if (save2) stats->broadcasts.modified++;
   }
-  // TODO: this needs reworking
-
-  // TODO:
-  //   do we attempt to make an episode URI from info?
-  //   OR
-  //   do we first try to find the broadcast and do a fuzzy check on params?
-  //   OR
-  //   do we do something else?
-
-  /* Generate a URI */
-  if ( desc ) 
-    uri = _xmltv_hash(desc);
-  if ( !uri ) return 0;
-  
-  /* Find episode */
-  ee = epg_episode_find_by_uri(uri, 1, &save);
-  stats->episodes.total++;
-  if (save) stats->episodes.created++;
-  if (title) save |= epg_episode_set_title(ee, title);
-  if (desc)  save |= epg_episode_set_description(ee, desc);
-  //if (category) save |= epg_episode_set_genre_str(ee, category);
-  //if (onscreen) save |= epg_episode_set_onscreen(ee, onscreen);
-  if (pn) save |= epg_episode_set_part(ee, pn, 0);
-  if (en) save |= epg_episode_set_number(ee, en);
-  // TODO: how can we handle season number?
-  if (save) stats->episodes.modified++;
-
   
   return save | save2;
 }
 
-
 /**
  * Parse a <programme> tag from xmltv
  */

diff --git a/src/tvheadend.h b/src/tvheadend.h
index 91211deb8821e8cd07fd76a92f87ee76e580c396..ae10afd137f71cc67d62b96a8aa8781f10c80744 100644 (file)
--- a/src/tvheadend.h
+++ b/src/tvheadend.h
@@ -449,4 +449,6 @@ void sbuf_put_be16(sbuf_t *sb, uint16_t u16);
 
 void sbuf_put_byte(sbuf_t *sb, uint8_t u8);
 
+char *md5sum ( const char *str );
+
 #endif /* TV_HEAD_H */

diff --git a/src/utils.c b/src/utils.c
index 5b792d6653e8fc2fb04334920621191fe2728171..391597b1efa7b42828a0566d50b395dd376fbe05 100644 (file)
--- a/src/utils.c
+++ b/src/utils.c
@@ -20,8 +20,10 @@
 #include <limits.h>
 #include <string.h>
 #include <assert.h>
+#include <openssl/md5.h>
 #include "tvheadend.h"
 
+
 /**
  * CRC32 
  */
@@ -306,3 +308,17 @@ sbuf_cut(sbuf_t *sb, int off)
   sb->sb_ptr = sb->sb_ptr - off;
   memmove(sb->sb_data, sb->sb_data + off, sb->sb_ptr);
 }
+
+char *
+md5sum ( const char *str )
+{
+  int i;
+  static unsigned char md5[MD5_DIGEST_LENGTH];
+  char *ret = malloc((MD5_DIGEST_LENGTH * 2) + 1);
+  MD5((const unsigned char*)str, strlen(str), md5);
+  for ( i = 0; i < MD5_DIGEST_LENGTH; i++ ) {
+    sprintf(&ret[i*2], "%02X", md5[i]);
+  }
+  ret[MD5_DIGEST_LENGTH*2] = '\0';
+  return ret;
+}