Update heif_enc.cc

examples/heif_enc.cc

@@ -96,6 +96,7 @@ uint32_t sequence_repetitions = 1;
 std::string vmt_metadata_file;
 bool binary_metadata_track = false;
 std::string metadata_track_uri = "vmt:metadata";
+const uint32_t BAD_VMT_TIMESTAMP = 0xFFFFFFFE;
 
 int quality = 50;
 bool lossless = false;
@@ -2189,6 +2190,44 @@ static std::vector<uint8_t> hex_to_binary(const std::string& line)
 }
 
 
+uint32_t parse_vmt_timestamp(const std::string& vmt_time)
+{
+  std::regex pattern(R"(-?((\d*):)?(\d\d):(\d\d)(\.(\d*))?)");
+  std::smatch match;
+
+  if (!std::regex_match(vmt_time, match, pattern)) {
+    return 0; // no match
+  }
+
+  std::string hh = match[2]; // optional
+  std::string mm = match[3];
+  std::string ss = match[4];
+  std::string fs = match[6]; // optional
+
+  if (vmt_time.find('-') != std::string::npos) {
+    return 0; // negative time not supported
+  }
+
+  uint32_t ms = 0;
+
+  if (fs != "") {
+    if (fs.length() == 3) {
+      ms = std::stoi(fs);
+    }
+    else {
+      return BAD_VMT_TIMESTAMP; // invalid
+    }
+  }
+
+  uint32_t ts = ((hh != "" ? std::stoi(hh) : 0) * 3600 * 1000 +
+                 std::stoi(mm) * 60 * 1000 +
+                 std::stoi(ss) * 1000 +
+                 ms);
+
+  return ts;
+}
+
+
 int encode_vmt_metadata_track(heif_context* context, heif_track* visual_track,
                               const std::string& track_uri, bool binary)
 {
@@ -2205,7 +2244,7 @@ int encode_vmt_metadata_track(heif_context* context, heif_track* visual_track,
 
   std::ifstream istr(vmt_metadata_file.c_str());
 
-  std::regex pattern(R"((\d\d):(\d\d):(\d\d).(\d\d\d) -->$)");
+  std::regex pattern(R"(^\s*(-?(\d|:|\.)*)\s*-->\s*(-?(\d|:|\.)*)?)");
 
   static std::vector<uint8_t> prev_metadata;
   static std::optional<uint32_t> prev_ts;
@@ -2219,15 +2258,10 @@ int encode_vmt_metadata_track(heif_context* context, heif_track* visual_track,
       continue;
     }
 
-    std::string hh = match[1];
-    std::string mm = match[2];
-    std::string ss = match[3];
-    std::string mil = match[4];
+    std::string cue_start = match[1];
+    std::string cue_end = match[3]; // == "" for unbounded cues
 
-    uint32_t ts = (std::stoi(hh) * 3600 * 1000 +
-                   std::stoi(mm) * 60 * 1000 +
-                   std::stoi(ss) * 1000 +
-                   std::stoi(mil));
+    uint32_t ts = parse_vmt_timestamp(cue_start);
 
     std::vector<uint8_t> concat;
 
@@ -2255,14 +2289,26 @@ int encode_vmt_metadata_track(heif_context* context, heif_track* visual_track,
       concat.push_back(0);
     }
 
-    if (prev_ts) {
-      heif_raw_sequence_sample_set_data(sample, (const uint8_t*)prev_metadata.data(), prev_metadata.size());
-      heif_raw_sequence_sample_set_duration(sample, ts - *prev_ts);
-      heif_track_add_raw_sequence_sample(track, sample);
-    }
+    if (ts != BAD_VMT_TIMESTAMP) {
 
-    prev_ts = ts;
-    prev_metadata = concat;
+      if (ts > *prev_ts) {
+        heif_raw_sequence_sample_set_data(sample, (const uint8_t*)prev_metadata.data(), prev_metadata.size());
+        heif_raw_sequence_sample_set_duration(sample, ts - *prev_ts);
+        heif_track_add_raw_sequence_sample(track, sample);
+      }
+      else if (ts == *prev_ts) {
+        concat.insert(concat.begin(), prev_metadata.begin(), prev_metadata.end());
+      }
+      else {
+        std::cerr << "Bad VMT timestamp order: " << cue_start << "\n";
+      }
+
+      prev_ts = ts;
+      prev_metadata = concat;
+    }
+    else {
+      std::cerr << "Bad VMT timestamp: " << cue_start << "\n";
+    }
   }
 
   // --- flush last metadata packet