add --moov-before-mdat

configure.ac

@@ -31,8 +31,9 @@ AC_TYPE_UINT8_T
 AC_CHECK_TYPES([ptrdiff_t])
 
 AC_SYS_LARGEFILE
+AC_CHECK_TYPES([struct __timeb64],[],[],[[#include <sys/timeb.h>]])
 AC_FUNC_FSEEKO
-AC_CHECK_FUNCS([sigaction gettimeofday nl_langinfo strdup _vscprintf])
+AC_CHECK_FUNCS([sigaction gettimeofday nl_langinfo _vscprintf fseeko64])
 AC_CHECK_FUNC(getopt_long)
 AM_CONDITIONAL([FDK_NO_GETOPT_LONG],[test "$ac_cv_func_getopt_long" != "yes"])
 AC_SEARCH_LIBS([aacEncOpen],[fdk-aac],[],[],[])

src/aacenc.c

@@ -97,7 +97,7 @@ int aacenc_init(HANDLE_AACENCODER *encoder, const aacenc_param_t *params,
     }
     if (aacEncoder_SetParam(*encoder, AACENC_SIGNALING_MODE,
                             params->sbr_signaling) != AACENC_OK) {
-        fprintf(stderr, "ERROR: unsupported transport format\n");
+        fprintf(stderr, "ERROR: failed to set SBR signaling mode\n");
         goto FAIL;
     }
     if (params->adts_crc_check)

src/compat.h

@@ -6,7 +6,10 @@
 #define COMPAT_H
 
 #ifndef HAVE_FSEEKO
-#  if _MSC_VER >= 1400
+#  if HAVE_FSEEKO64
+#    define fseeko fseeko64
+#    define ftello ftello64
+#  elif _MSC_VER >= 1400
 #    define fseeko _fseeki64
 #    define ftello _ftelli64
 #  else

src/compat_win32.c

@@ -31,8 +31,13 @@ int __wgetmainargs(int *, wchar_t ***, wchar_t ***, int, _startupinfo *);
 
 int64_t aacenc_timer(void)
 {
+#if HAVE_STRUCT___TIMEB64
     struct __timeb64 tv;
     _ftime64(&tv);
+#else
+    struct timeb tv;
+    ftime(&tv);
+#endif
     return (int64_t)tv.time * 1000 + tv.millitm;
 }
 

src/m4af.c

@@ -23,6 +23,7 @@
 
 #define m4af_realloc(memory,size) realloc(memory, size)
 #define m4af_free(memory) free(memory)
+#define m4af_max(a,b) ((a)<(b)?(b):(a))
 
 #define M4AF_ATOM_WILD  0xffffffff
 
@@ -79,6 +80,7 @@ struct m4af_ctx_t {
     int64_t modification_time;
     int64_t mdat_pos;
     int64_t mdat_size;
+    int priming_mode;
     int last_error;
 
     m4af_itmf_entry_t *itmf_table;
@@ -148,6 +150,13 @@ int m4af_write(m4af_ctx_t *ctx, const void *data, uint32_t size)
     return rc;
 }
 
+static
+int m4af_write16(m4af_ctx_t *ctx, uint32_t data)
+{
+    data = m4af_htob16(data);
+    return m4af_write(ctx, &data, 2);
+}
+
 static
 int m4af_write32(m4af_ctx_t *ctx, uint32_t data)
 {
@@ -277,6 +286,11 @@ void m4af_set_priming(m4af_ctx_t *ctx, uint32_t track_idx,
     track->padding = padding;
 }
 
+void m4af_set_priming_mode(m4af_ctx_t *ctx, int mode)
+{
+    ctx->priming_mode = mode;
+}
+
 static
 int m4af_add_sample_entry(m4af_ctx_t *ctx, uint32_t track_idx,
                           uint32_t size, uint32_t delta)
@@ -583,12 +597,13 @@ int m4af_set_iTunSMPB(m4af_ctx_t *ctx)
 }
 
 static
-void m4af_update_box_size(m4af_ctx_t *ctx, int64_t pos)
+uint32_t m4af_update_box_size(m4af_ctx_t *ctx, int64_t pos)
 {
     int64_t current_pos = m4af_tell(ctx);
     m4af_set_pos(ctx, pos);
     m4af_write32(ctx, current_pos - pos);
     m4af_set_pos(ctx, current_pos);
+    return current_pos - pos;
 }
 
 static
@@ -823,12 +838,50 @@ void m4af_write_stsd_box(m4af_ctx_t *ctx, uint32_t track_idx)
     m4af_update_box_size(ctx, pos);
 }
 
+static
+void m4af_write_sbgp_box(m4af_ctx_t *ctx, uint32_t track_idx)
+{
+    m4af_track_t *track = &ctx->track[track_idx];
+    m4af_write(ctx,
+               "\0\0\0\034"  /* size: 28                */
+               "sbgp"        /* type                    */
+               "\0"          /* version                 */
+               "\0\0\0"      /* flags                   */
+               "roll"        /* grouping_type           */
+               "\0\0\0\001"  /* entry_count: 1          */
+               , 20);
+    m4af_write32(ctx, track->num_samples);
+    m4af_write32(ctx, 1);    /* group_description_index */
+}
+
+static
+void m4af_write_sgpd_box(m4af_ctx_t *ctx, uint32_t track_idx)
+{
+    m4af_track_t *track = &ctx->track[track_idx];
+    m4af_write(ctx,
+               "\0\0\0\032"  /* size               */
+               "sgpd"        /* type               */
+               "\001"        /* version            */
+               "\0\0\0"      /* flags              */
+               "roll"        /* grouping_type      */
+               "\0\0\0\002"  /* default_length: 2  */
+               "\0\0\0\001"  /* entry_count: 1     */
+               "\377\377"    /* payload_data: -1   */
+               , 26);
+}
+
 static
 void m4af_write_stbl_box(m4af_ctx_t *ctx, uint32_t track_idx)
 {
+    m4af_track_t *track = &ctx->track[track_idx];
     int64_t pos = m4af_tell(ctx);
     m4af_write(ctx, "\0\0\0\0stbl", 8);
     m4af_write_stsd_box(ctx, track_idx);
+    if ((ctx->priming_mode & M4AF_PRIMING_MODE_EDTS) &&
+        (track->encoder_delay || track->padding)) {
+        m4af_write_sbgp_box(ctx, track_idx);
+        m4af_write_sgpd_box(ctx, track_idx);
+    }
     m4af_write_stts_box(ctx, track_idx);
     m4af_write_stsc_box(ctx, track_idx);
     m4af_write_stsz_box(ctx, track_idx);
@@ -962,6 +1015,42 @@ void m4af_write_mdia_box(m4af_ctx_t *ctx, uint32_t track_idx)
     m4af_update_box_size(ctx, pos);
 }
 
+static
+void m4af_write_elst_box(m4af_ctx_t *ctx, uint32_t track_idx)
+{
+    m4af_track_t *track = &ctx->track[track_idx];
+    uint8_t version;
+    int64_t duration = track->duration - track->encoder_delay - track->padding;
+    int64_t pos = m4af_tell(ctx);
+    duration = (double)duration / track->timescale * ctx->timescale + .5;
+    version  = (duration > UINT32_MAX);
+
+    m4af_write(ctx, "\0\0\0\0elst", 8);
+    m4af_write(ctx, &version, 1);
+    m4af_write(ctx, "\0\0\0", 3);  /* flags          */
+    m4af_write32(ctx, 1);          /* entry_count: 1 */
+    if (version) {
+        m4af_write64(ctx, duration);
+        m4af_write64(ctx, track->encoder_delay);
+    } else {
+        m4af_write32(ctx, duration);
+        m4af_write32(ctx, track->encoder_delay);
+    }
+    m4af_write16(ctx, 1);    /* media_rate_integer  */
+    m4af_write16(ctx, 0);    /* media_rate_fraction */
+    m4af_update_box_size(ctx, pos);
+}
+
+static
+void m4af_write_edts_box(m4af_ctx_t *ctx, uint32_t track_idx)
+{
+    m4af_track_t *track = &ctx->track[track_idx];
+    int64_t pos = m4af_tell(ctx);
+    m4af_write(ctx, "\0\0\0\0edts", 8);
+    m4af_write_elst_box(ctx, track_idx);
+    m4af_update_box_size(ctx, pos);
+}
+
 static
 void m4af_write_tkhd_box(m4af_ctx_t *ctx, uint32_t track_idx)
 {
@@ -1015,9 +1104,13 @@ void m4af_write_tkhd_box(m4af_ctx_t *ctx, uint32_t track_idx)
 static
 void m4af_write_trak_box(m4af_ctx_t *ctx, uint32_t track_idx)
 {
+    m4af_track_t *track = &ctx->track[track_idx];
     int64_t pos = m4af_tell(ctx);
     m4af_write(ctx, "\0\0\0\0trak", 8);
     m4af_write_tkhd_box(ctx, track_idx);
+    if ((ctx->priming_mode & M4AF_PRIMING_MODE_EDTS) &&
+        (track->encoder_delay || track->padding))
+        m4af_write_edts_box(ctx, track_idx);
     m4af_write_mdia_box(ctx, track_idx);
     m4af_update_box_size(ctx, pos);
 }
@@ -1182,7 +1275,7 @@ void m4af_write_udta_box(m4af_ctx_t *ctx)
 }
 
 static
-void m4af_write_moov_box(m4af_ctx_t *ctx)
+uint32_t m4af_write_moov_box(m4af_ctx_t *ctx)
 {
     unsigned i;
     int64_t pos = m4af_tell(ctx);
@@ -1192,7 +1285,7 @@ void m4af_write_moov_box(m4af_ctx_t *ctx)
         m4af_write_trak_box(ctx, i);
     if (ctx->num_tags)
         m4af_write_udta_box(ctx);
-    m4af_update_box_size(ctx, pos);
+    return m4af_update_box_size(ctx, pos);
 }
 
 static
@@ -1210,10 +1303,35 @@ void m4af_finalize_mdat(m4af_ctx_t *ctx)
     m4af_set_pos(ctx, ctx->mdat_pos + ctx->mdat_size);
 }
 
-int m4af_finalize(m4af_ctx_t *ctx)
+static
+void m4af_shift_mdat_pos(m4af_ctx_t *ctx, uint32_t offset)
+{
+    unsigned i, j;
+    int64_t begin, end;
+    char buf[8192];
+
+    end = ctx->mdat_pos + ctx->mdat_size;
+    for (; (begin = m4af_max(ctx->mdat_pos, end - 8192)) < end; end = begin) {
+        m4af_set_pos(ctx, begin);
+        ctx->io.read(ctx->io_cookie, buf, end - begin);
+        m4af_set_pos(ctx, begin + offset);
+        m4af_write(ctx, buf, end - begin);
+    }
+    for (i = 0; i < ctx->num_tracks; ++i)
+        for (j = 0; j < ctx->track[i].num_chunks; ++j)
+            ctx->track[i].chunk_table[j].offset += offset;
+    ctx->mdat_pos += offset;
+    m4af_set_pos(ctx, ctx->mdat_pos - 16);
+    m4af_write_free_box(ctx, 0);
+    m4af_write(ctx, "\0\0\0\0mdat", 8);
+    m4af_finalize_mdat(ctx);
+}
+
+int m4af_finalize(m4af_ctx_t *ctx, int optimize)
 {
     unsigned i;
     m4af_track_t *track;
+    uint32_t moov_size;
 
     for (i = 0; i < ctx->num_tracks; ++i) {
         track = ctx->track + i;
@@ -1227,9 +1345,19 @@ int m4af_finalize(m4af_ctx_t *ctx)
         }
         m4af_flush_chunk(ctx, i);
     }
-    if (ctx->track[0].encoder_delay || ctx->track[0].padding)
+    track = ctx->track;
+    if ((ctx->priming_mode & M4AF_PRIMING_MODE_ITUNSMPB) &&
+        (track->encoder_delay || track->padding))
         m4af_set_iTunSMPB(ctx);
     m4af_finalize_mdat(ctx);
+    moov_size = m4af_write_moov_box(ctx);
+    if (optimize) {
+        int64_t pos;
+        m4af_shift_mdat_pos(ctx, moov_size + 1024);
+        m4af_set_pos(ctx, 32);
         m4af_write_moov_box(ctx);
+        pos = m4af_tell(ctx);
+        m4af_write_free_box(ctx, ctx->mdat_pos - pos - 24);
+    }
     return ctx->last_error;
 }

src/m4af.h

@@ -52,6 +52,12 @@ enum m4af_codec_type {
     M4AF_CODEC_TEXT = M4AF_FOURCC('t','e','x','t'),
 };
 
+enum m4af_priming_mode {
+    M4AF_PRIMING_MODE_ITUNSMPB = 1,
+    M4AF_PRIMING_MODE_EDTS = 2,
+    M4AF_PRIMING_MODE_BOTH = 3
+};
+
 typedef int (*m4af_read_callback)(void *cookie, void *buffer, uint32_t size);
 typedef int (*m4af_write_callback)(void *cookie, const void *data,
                                    uint32_t size);
@@ -81,7 +87,7 @@ m4af_ctx_t *m4af_create(uint32_t codec, uint32_t timescale,
 
 int m4af_begin_write(m4af_ctx_t *ctx);
 
-int m4af_finalize(m4af_ctx_t *ctx);
+int m4af_finalize(m4af_ctx_t *ctx, int optimize);
 
 void m4af_teardown(m4af_ctx_t **ctx);
 
@@ -94,6 +100,8 @@ int m4af_set_decoder_specific_info(m4af_ctx_t *ctx, uint32_t track_idx,
 void m4af_set_priming(m4af_ctx_t *ctx, uint32_t track_idx,
                       uint32_t encoder_delay, uint32_t padding);
 
+void m4af_set_priming_mode(m4af_ctx_t *ctx, int mode);
+
 void m4af_set_fixed_frame_duration(m4af_ctx_t *ctx, uint32_t track_idx,
                                    uint32_t length);
 

src/main.c

@@ -148,8 +148,13 @@ PROGNAME " %s\n"
 "                               transport layer\n"
 "\n"
 " -o <filename>                 Output filename\n"
+" -G, --gapless-mode <n>        Encoder delay signaling for gapless playback\n"
+"                                 0: iTunSMPB (default)\n"
+"                                 1: ISO standard (edts + sgpd)\n"
+"                                 2: Both\n"
 " --ignorelength                Ignore length of WAV header\n"
 " -S, --silent                  Don't print progress messages\n"
+" --moov-before-mdat            Place moov box before mdat box on m4a output\n"
 "\n"
 "Options for raw (headerless) input:\n"
 " -R, --raw                     Treat input as raw (by default WAV is\n"
@@ -199,8 +204,10 @@ typedef struct aacenc_param_ex_t {
 
     char *input_filename;
     char *output_filename;
+    unsigned gapless_mode;
     unsigned ignore_length;
     int silent;
+    int moov_before_mdat;
 
     int is_raw;
     unsigned raw_channels;
@@ -218,6 +225,7 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
     int ch;
     unsigned n;
 
+#define OPT_MOOV_BEFORE_MDAT     M4AF_FOURCC('m','o','o','v')
 #define OPT_RAW_CHANNELS         M4AF_FOURCC('r','c','h','n')
 #define OPT_RAW_RATE             M4AF_FOURCC('r','r','a','t')
 #define OPT_RAW_FORMAT           M4AF_FOURCC('r','f','m','t')
@@ -239,8 +247,10 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
         { "adts-crc-check",   no_argument,       0, 'C' },
         { "header-period",    required_argument, 0, 'P' },
 
+        { "gapless-mode",     required_argument, 0, 'G' },
         { "ignorelength",     no_argument,       0, 'I' },
         { "silent",           no_argument,       0, 'S' },
+        { "moov-before-mdat", no_argument,       0, OPT_MOOV_BEFORE_MDAT   },
 
         { "raw",              no_argument,       0, 'R' },
         { "raw-channels",     required_argument, 0, OPT_RAW_CHANNELS       },
@@ -268,7 +278,7 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
     params->afterburner = 1;
 
     aacenc_getmainargs(&argc, &argv);
-    while ((ch = getopt_long(argc, argv, "hp:b:m:w:a:Ls:f:CP:Io:SR",
+    while ((ch = getopt_long(argc, argv, "hp:b:m:w:a:Ls:f:CP:G:Io:SR",
                              long_options, 0)) != EOF) {
         switch (ch) {
         case 'h':
@@ -338,12 +348,22 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
         case 'o':
             params->output_filename = optarg;
             break;
+        case 'G':
+            if (sscanf(optarg, "%u", &n) != 1 || n > 2) {
+                fprintf(stderr, "invalid arg for gapless-mode\n");
+                return -1;
+            }
+            params->gapless_mode = n;
+            break;
         case 'I':
             params->ignore_length = 1;
             break;
         case 'S':
             params->silent = 1;
             break;
+        case OPT_MOOV_BEFORE_MDAT:
+            params->moov_before_mdat = 1;
+            break;
         case 'R':
             params->is_raw = 1;
             break;
@@ -567,7 +587,7 @@ int finalize_m4a(m4af_ctx_t *m4af, const aacenc_param_ex_t *params,
 
     put_tool_tag(m4af, params, encoder);
 
-    if (m4af_finalize(m4af) < 0) {
+    if (m4af_finalize(m4af, params->moov_before_mdat) < 0) {
         fprintf(stderr, "ERROR: failed to finalize m4a\n");
         return -1;
     }
@@ -634,7 +654,7 @@ int main(int argc, char **argv)
 {
     wav_io_context_t wav_io = { read_callback, seek_callback, tell_callback };
     m4af_io_callbacks_t
-        m4af_io = { 0, write_callback, seek_callback, tell_callback };
+        m4af_io = { read_callback, write_callback, seek_callback, tell_callback };
     aacenc_param_ex_t params = { 0 };
 
     int result = 2;
@@ -698,7 +718,7 @@ int main(int argc, char **argv)
         params.output_filename = output_filename;
     }
 
-    if ((ofp = aacenc_fopen(params.output_filename, "wb")) == 0) {
+    if ((ofp = aacenc_fopen(params.output_filename, "wb+")) == 0) {
         aacenc_fprintf(stderr, "ERROR: %s: %s\n", params.output_filename,
                        strerror(errno));
         goto END;
@@ -718,6 +738,7 @@ int main(int argc, char **argv)
                                        aacinfo.confSize);
         m4af_set_fixed_frame_duration(m4af, 0,
                                       framelen >> downsampled_timescale);
+        m4af_set_priming_mode(m4af, params.gapless_mode + 1);
         m4af_begin_write(m4af);
     }
     frame_count = encode(wavf, encoder, aacinfo.frameLength, ofp, m4af,

version.h

@@ -1,4 +1,4 @@
 #ifndef VERSION_H
 #define VERSION_H
-const char *fdkaac_version = "0.1.9";
+const char *fdkaac_version = "0.3.0";
 #endif