[db,library] Add support for parsing lyrics and storing them in DB

src/db.c

@@ -777,6 +777,7 @@ free_mfi(struct media_file_info *mfi, int content_only)
   free(mfi->composer_sort);
   free(mfi->album_artist_sort);
   free(mfi->virtual_path);
+  free(mfi->lyrics);
 
   if (!content_only)
     free(mfi);

src/db.h

@@ -248,6 +248,7 @@ struct media_file_info {
   char *composer_sort;
 
   uint32_t scan_kind; /* Identifies the library_source that created/updates this item */
+  char *lyrics;
 };
 
 #define mfi_offsetof(field) offsetof(struct media_file_info, field)

src/db_init.c

@@ -98,7 +98,8 @@
   "   composer_sort      VARCHAR(1024) DEFAULT NULL COLLATE DAAP,"	\
   "   channels           INTEGER DEFAULT 0,"		\
   "   usermark           INTEGER DEFAULT 0,"		\
-  "   scan_kind          INTEGER DEFAULT 0"		\
+  "   scan_kind          INTEGER DEFAULT 0,"		\
+  "   lyrics             TEXT DEFAULT NULL COLLATE DAAP"		\
   ");"
 
 #define T_PL					\

src/db_init.h

@@ -26,7 +26,7 @@
  * is a major upgrade. In other words minor version upgrades permit downgrading
  * the server after the database was upgraded. */
 #define SCHEMA_VERSION_MAJOR 22
-#define SCHEMA_VERSION_MINOR 0
+#define SCHEMA_VERSION_MINOR 1
 
 int
 db_init_indices(sqlite3 *hdl);

src/db_upgrade.c

@@ -1227,6 +1227,26 @@ static const struct db_upgrade_query db_upgrade_v2200_queries[] =
     { U_v2200_SCVER_MINOR,    "set schema_version_minor to 00" },
   };
 
+/* ---------------------------- 22.00 -> 22.01 ------------------------------ */
+
+#define U_v2201_ALTER_FILES_ADD_LYRICS \
+  "ALTER TABLE files ADD COLUMN lyrics TEXT DEFAULT NULL COLLATE DAAP;"
+
+#define U_v2201_SCVER_MAJOR                    \
+  "UPDATE admin SET value = '22' WHERE key = 'schema_version_major';"
+#define U_v2201_SCVER_MINOR                    \
+  "UPDATE admin SET value = '01' WHERE key = 'schema_version_minor';"
+
+static const struct db_upgrade_query db_upgrade_v2201_queries[] =
+  {
+    { U_v2201_ALTER_FILES_ADD_LYRICS, "alter table files add column lyrics" },
+
+    { U_v2201_SCVER_MAJOR,    "set schema_version_major to 22" },
+    { U_v2201_SCVER_MINOR,    "set schema_version_minor to 01" },
+  };
+
+
+
 /* -------------------------- Main upgrade handler -------------------------- */
 
 int
@@ -1437,6 +1457,12 @@ db_upgrade(sqlite3 *hdl, int db_ver)
       if (ret < 0)
 	return -1;
 
+      /* FALLTHROUGH */
+
+    case 2200:
+      ret = db_generic_upgrade(hdl, db_upgrade_v2201_queries, ARRAY_SIZE(db_upgrade_v2201_queries));
+      if (ret < 0)
+	return -1;
 
       /* Last case statement is the only one that ends with a break statement! */
       break;

src/library/filescanner_ffmpeg.c

@@ -49,6 +49,8 @@ struct metadata_map {
 // Used for passing errors to DPRINTF (can't count on av_err2str being present)
 static char errbuf[64];
 
+static int lyricsindex = -1;
+
 static inline char *
 err2str(int errnum)
 {
@@ -186,6 +188,7 @@ static const struct metadata_map md_map_generic[] =
     { "artist-sort",  0, mfi_offsetof(artist_sort),        NULL },
     { "album-sort",   0, mfi_offsetof(album_sort),         NULL },
     { "compilation",  1, mfi_offsetof(compilation),        NULL },
+    { "lyrics",       0, mfi_offsetof(lyrics),             NULL },
 
     // ALAC sort tags
     { "sort_name",           0, mfi_offsetof(title_sort),         NULL },
@@ -277,43 +280,41 @@ static const struct metadata_map md_map_id3[] =
     { NULL,                  0, 0,                                   NULL }
   };
 
-
+/* TODO: If the md_map was sorted, the search would be O(log N) instead of O(N) */
 static int
-extract_metadata_core(struct media_file_info *mfi, AVDictionary *md, const struct metadata_map *md_map)
+match_metadata(const char *key, const struct metadata_map *md_map)
 {
-  AVDictionaryEntry *mdt;
-  char **strval;
-  uint32_t *intval;
-  int mdcount;
   int i;
-  int ret;
 
-#if 0
-  /* Dump all the metadata reported by ffmpeg */
-  mdt = NULL;
-  while ((mdt = av_dict_get(md, "", mdt, AV_DICT_IGNORE_SUFFIX)) != NULL)
-    DPRINTF(E_DBG, L_SCAN, " -> %s = %s\n", mdt->key, mdt->value);
-#endif
-
-  mdcount = 0;
-
-  /* Extract actual metadata */
   for (i = 0; md_map[i].key != NULL; i++)
     {
-      mdt = av_dict_get(md, md_map[i].key, NULL, 0);
-      if (mdt == NULL)
-	continue;
+      if (strcmp(key, md_map[i].key) == 0)
+	return i;
+    }
+  return -1;
+}
+
+
+static int
+iterate_metadata(struct media_file_info *mfi, const AVDictionaryEntry *mdt, const struct metadata_map *md_map)
+{
+  char **strval;
+  uint32_t *intval;
+  int i;
 
   if ((mdt->value == NULL) || (strlen(mdt->value) == 0))
-	continue;
+    return 0;
+
+  if (strncmp(mdt->key, "lyrics-", sizeof("lyrics-") - 1) == 0)
+    i = lyricsindex;
+  else
+    i = match_metadata(mdt->key, md_map);
+
+  if (i == -1)
+    return 0;
 
   if (md_map[i].handler_function)
-	{
-	  mdcount += md_map[i].handler_function(mfi, mdt->value);
-	  continue;
-	}
-
-      mdcount++;
+    return md_map[i].handler_function(mfi, mdt->value);
 
   if (!md_map[i].as_int)
     {
@@ -328,11 +329,43 @@ extract_metadata_core(struct media_file_info *mfi, AVDictionary *md, const struc
 
       if (*intval == 0)
 	{
-	      ret = safe_atou32(mdt->value, intval);
-	      if (ret < 0)
-		continue;
+          if (safe_atou32(mdt->value, intval) < 0)
+	    return 1; /* Should probably be 0 */
 	}
     }
+  return 1;
+}
+
+static int
+extract_metadata_core(struct media_file_info *mfi, AVDictionary *md, const struct metadata_map *md_map)
+{
+  const AVDictionaryEntry *mdt;
+  int mdcount;
+
+#if 0
+  /* Dump all the metadata reported by ffmpeg */
+  mdt = NULL;
+  while ((mdt = av_dict_get(md, "", mdt, AV_DICT_IGNORE_SUFFIX)) != NULL)
+    DPRINTF(E_DBG, L_SCAN, " -> %s = %s\n", mdt->key, mdt->value);
+#endif
+
+  mdcount = 0;
+
+  /* Cache this search once to avoid doing it per metadata key */
+  if (lyricsindex == -1)
+    lyricsindex = match_metadata("lyrics", md_map);
+
+  /* Extract lyrics if any found.
+     FFMPEG creates a metadata key that's embedding the language code in it, like 'lyrics-eng' or 'lyrics-XXX'
+     So it's not possible to query this key directly except by brute-forcing all languages.
+     Instead, we are reversing the metadata searching algorithm to query all metadata key that FFMPEG fetched
+     and matching them against our own map. This is the most efficient method to search it without having 2 pass
+     on the KV store */
+  mdt = av_dict_iterate(md, NULL);
+  while (mdt != NULL)
+    {
+      mdcount += iterate_metadata(mfi, mdt, md_map);
+      mdt = av_dict_iterate(md, mdt);
     }
 
   return mdcount;