It looks like SSA can have more that one block per packet, newline separated.

So look for those, and concatenate the lines, so we don't strip the control
information from the first one only. Not sure if that's per spec, but I've
got one sample with those, and it's from the VLC test streams, so it must
be more than just an oddball.

src/ffmpeg2theora.c

@@ -322,62 +322,79 @@ static int is_supported_subtitle_stream(ff2theora this, int idx)
 static char *get_raw_text_from_ssa(const char *ssa)
 {
   int n,intag,inescape;
+  char *multiblock = NULL, *realloced_mb;
   char *allocated;
   const char *dialogue, *ptr, *tag_start;
 
   if (!ssa) return NULL;
-  dialogue=strstr(ssa, "Dialogue:");
-  if (!dialogue) return NULL;
-
-  ptr = dialogue;
-  for (n=0;n<9;++n) {
-    ptr=strchr(ptr,',');
-    if (!ptr) return NULL;
-    ++ptr;
-  }
-  dialogue = ptr;
-  allocated = strdup(dialogue);
-
-  /* find all "{...}" tags - the following must work for UTF-8 */
-  intag=inescape=0;
-  n=0;
-  for (ptr=dialogue; *ptr; ++ptr) {
-    if (*ptr=='{') {
-      if (intag==0) tag_start = ptr;
-      ++intag;
+
+  /* turns out some SSA packets have several blocks, each on a single line, so loop */
+  while (ssa) {
+    dialogue=strstr(ssa, "Dialogue:");
+    if (!dialogue) break;
+
+    ptr = dialogue;
+    for (n=0;n<9;++n) {
+      ptr=strchr(ptr,',');
+      if (!ptr) return NULL;
+      ++ptr;
     }
-    else if (*ptr=='}') {
-      --intag;
-      if (intag == 0) {
-        /* tag parsing - none for now */
+    dialogue = ptr;
+    allocated = strdup(dialogue);
+
+    /* find all "{...}" tags - the following must work for UTF-8 */
+    intag=inescape=0;
+    n=0;
+    for (ptr=dialogue; *ptr && *ptr!='\n'; ++ptr) {
+      if (*ptr=='{') {
+        if (intag==0) tag_start = ptr;
+        ++intag;
       }
-    }
-    else if (!intag) {
-      if (inescape) {
-        if (*ptr == 'N' || *ptr == 'n')
-          allocated[n++] = '\n';
-        else if (*ptr == 'h')
-          allocated[n++] = ' ';
-        inescape=0;
+      else if (*ptr=='}') {
+        --intag;
+        if (intag == 0) {
+          /* tag parsing - none for now */
+        }
       }
-      else {
-        if (*ptr=='\\') {
-          inescape=1;
+      else if (!intag) {
+        if (inescape) {
+          if (*ptr == 'N' || *ptr == 'n')
+            allocated[n++] = '\n';
+          else if (*ptr == 'h')
+            allocated[n++] = ' ';
+          inescape=0;
         }
         else {
-          allocated[n++]=*ptr;
+          if (*ptr=='\\') {
+            inescape=1;
+          }
+          else {
+            allocated[n++]=*ptr;
+          }
         }
       }
     }
+    allocated[n]=0;
+
+    /* skip over what we read */
+    ssa = ptr;
+
+    /* remove any trailing newlines (also \r characters) */
+    n = strlen(allocated);
+    while (n>0 && (allocated[n-1]=='\n' || allocated[n-1]=='\r'))
+      allocated[--n]=0;
+
+    /* add this new block */
+    realloced_mb = (char*)realloc(multiblock, (multiblock?strlen(multiblock):0) + strlen(allocated) + 2); /* \n + 0 */
+    if (realloced_mb) {
+      if (multiblock) strcat(realloced_mb, "\n"); else strcpy(realloced_mb, "");
+      strcat(realloced_mb, allocated);
+      multiblock = realloced_mb;
+      free(allocated);
+    }
   }
-  allocated[n]=0;
-
-  /* remove any trailing newlines (also \r characters) */
-  n = strlen(allocated);
-  while (n>0 && (allocated[n-1]=='\n' || allocated[n-1]=='\r'))
-    allocated[--n]=0;
 
-  return allocated;
+  return multiblock;
 }
 
 static const float get_ssa_time(const char *p)