bump version

Since channel map was not initialized, first channel was copied to every
other channel.

.gitignore

@@ -0,0 +1,22 @@
+*.o
+*~
+Makefile
+Makefile.in
+aclocal.m4
+autom4te.cache
+compile
+config.guess
+config.h
+config.h.in
+config.log
+config.status
+config.sub
+configure
+depcomp
+install-sh
+missing
+missings/.deps/
+src/.deps/
+src/.dirstamp
+stamp-h1
+fdkaac

ChangeLog

@@ -0,0 +1,318 @@
+2013-11-08  nu774 <honeycomb77@gmail.com>
+
+  * update ChangeLog [HEAD]
+
+  * bump version [v0.5.1]
+
+  * fix to use libFDKAAC signaling mode 1
+
+2013-11-05  nu774 <honeycomb77@gmail.com>
+
+  * fix README
+
+2013-11-04  nu774 <honeycomb77@gmail.com>
+
+  * rename README.md -> README
+
+  * Update README -> README.md, generate groff manpage from README.md
+
+  * update ChangeLog
+
+  * update git2changelog to accept non-ascii output
+
+  * add manpage
+
+2013-11-03  nu774 <honeycomb77@gmail.com>
+
+  * fix gcc warnings
+
+  * Merge pull request #7 from rbrito/misc-fixes
+
+2013-11-03  Rogério Brito <rbrito@ime.usp.br>
+
+  * gitignore: Add list of files to ignore.
+
+2013-11-03  nu774 <honeycomb77@gmail.com>
+
+  * update ChangeLog
+
+  * bump version [v0.5.0]
+
+  * add --sbr-ratio to support AACENC_SBR_RATIO appeared on libFDK 3.4.12
+
+  * support 7.1 channel mode added on FDK 3.4.12
+
+2013-10-30  nu774 <honeycomb77@gmail.com>
+
+  * update ChangeLog
+
+  * bump version [v0.4.2]
+
+  * use tell() to obtain data chunk offset
+
+  * rename aacenc_result_t -> aacenc_frame_t, simplify write_sample()
+
+  * prepend 1 sample zero padding in case of SBR and enc_delay is odd
+
+  * cleanup interface of aac_encode_frame()
+
+  * add some copyright notice
+
+2013-10-29  nu774 <honeycomb77@gmail.com>
+
+  * smart padding for better gapless playback
+
+  * fix unused variable warning
+
+  * fix warning: cast size_t as sprintf() arg to int
+
+  * fix vcxproj
+
+  * fix pcm_seek() to inline
+
+2013-10-27  nu774 <honeycomb77@gmail.com>
+
+  * bump version [v0.4.1]
+
+  * add --include-sbr-delay
+
+  * fix help message: show -I as shorthand for --ignorelength
+
+  * remove --sbr-signaling
+
+2013-10-26  nu774 <honeycomb77@gmail.com>
+
+  * re-fix #ifdef cond for lrint() [old]
+
+  * tag mapping: add recorded date and tempo, remove performer->artist
+
+2013-10-25  nu774 <honeycomb77@gmail.com>
+
+  * fix MSVC12 build issue
+
+  * fix build issue on platform where fileno is a naive macro
+
+  * update ChangeLog
+
+  * bump version [v0.4.0]
+
+  * update README
+
+2013-10-24  nu774 <honeycomb77@gmail.com>
+
+  * caf input support
+
+  * refactor pcm io routines
+
+2013-10-23  nu774 <honeycomb77@gmail.com>
+
+  * cleanup metadata handling
+
+  * --tag-from-json: properly support number/total format in json track field
+
+2013-10-22  nu774 <honeycomb77@gmail.com>
+
+  * bump version [v0.3.3]
+
+  * fixed bogus sgpd written on --gapless-mode=1 and 2
+
+2013-10-21  nu774 <honeycomb77@gmail.com>
+
+  * bump version [v0.3.2]
+
+2013-10-20  nu774 <honeycomb77@gmail.com>
+
+  * reimplement int16 conversion as pcm_reader
+
+  * add abstraction layer for pcm reading
+
+  * improve handling of avgBitrate
+
+2013-10-19  nu774 <honeycomb77@gmail.com>
+
+  * update ChangeLog and git2changelog.py
+
+2013-10-18  nu774 <honeycomb77@gmail.com>
+
+  * bump version [v0.3.1]
+
+  * update README
+
+  * set avgBitrate field to zero for 14496-1 compliance
+
+2013-09-07  nu774 <honeycomb77@gmail.com>
+
+  * updated ChangeLog with new git2changelog.py
+
+2013-06-14  nu774 <honeycomb77@gmail.com>
+
+  * add --moov-before-mdat [v0.3.0]
+
+2013-03-04  nu774 <honeycomb77@gmail.com>
+
+  * fix an error message
+
+2013-03-03  nu774 <honeycomb77@gmail.com>
+
+  * bump version [v0.2.0]
+
+  * add --gapless-mode
+
+2013-02-20  nu774 <honeycomb77@gmail.com>
+
+  * simplify __timeb64 condition
+
+  * use fseeko64() on i686-pc-mingw32
+
+2013-02-18  nu774 <honeycomb77@gmail.com>
+
+  * fix build issue on i686-pc-mingw (struct __timeb64 is missing)
+
+2013-02-17  nu774 <honeycomb77@gmail.com>
+
+  * bump version [v0.1.9]
+
+  * fix to accept option -C
+
+2013-02-16  nu774 <honeycomb77@gmail.com>
+
+  * bump version [v0.1.8]
+
+  * refine json metadata importing
+
+  * m4af: duplication check on adding tags
+
+2013-02-15  nu774 <honeycomb77@gmail.com>
+
+  * add --tag-from-json [v0.1.7]
+
+  * fix implicit int variable decl.
+
+  * update m4af
+
+2013-02-03  nu774 <honeycomb77@gmail.com>
+
+  * bump version [v0.1.6]
+
+  * win32: change _wfopen() -> wfsopen()
+
+2013-01-30  nu774 <honeycomb77@gmail.com>
+
+  * update README (add note for character encoding)
+
+2013-01-28  nu774 <honeycomb77@gmail.com>
+
+  * bump version [v0.1.5]
+
+  * gracefully shutdown on signals
+
+2013-01-27  nu774 <honeycomb77@gmail.com>
+
+  * fix MSVC project build issue
+
+2013-01-25  nu774 <honeycomb77@gmail.com>
+
+  * bump version [v0.1.4]
+
+  * add --tag-from-file
+
+2013-01-24  nu774 <honeycomb77@gmail.com>
+
+  * add --silent
+
+2013-01-19  nu774 <honeycomb77@gmail.com>
+
+  * retab
+
+  * bump version [v0.1.3]
+
+  * fix crash on wrong long option, rename --ignore-length to --ignorelength
+
+2013-01-17  nu774 <honeycomb77@gmail.com>
+
+  * bump version [v0.1.2]
+
+  * compat_win32: free argv with atexit()
+
+  * take care of COPYRIGHT-SIGN in UTF-8
+
+2013-01-15  nu774 <honeycomb77@gmail.com>
+
+  * bump version [v0.1.1]
+
+  * fix return type of put_type_entry() to void
+
+  * add ADTS header size(7) to output byte length
+
+2013-01-13  nu774 <honeycomb77@gmail.com>
+
+  * fix mp4 duration & version calcuration [v0.1.0]
+
+2013-01-11  nu774 <honeycomb77@gmail.com>
+
+  * add support for xid
+
+  * support for i686-pc-mingw32 (missing _vscprintf)
+
+2013-01-10  nu774 <honeycomb77@gmail.com>
+
+  * bump version [v0.0.9]
+
+  * rename basename() -> aacenc_basename() and move to compat layer
+
+2013-01-09  nu774 <honeycomb77@gmail.com>
+
+  * add --tag and --long-tag
+
+  * fix corner case of progress display
+
+  * calculate length from file size
+
+  * raw input support
+
+2013-01-08  nu774 <honeycomb77@gmail.com>
+
+  * insert a white space in progress message
+
+2013-01-07  nu774 <honeycomb77@gmail.com>
+
+  * fix typo of bitrate-mode option [v0.0.8]
+
+  * more static inlining (missed on the previous commit) [v0.0.7]
+
+  * check error of fread() and fwrite() [v0.0.6]
+
+  * change inline->static inline to follow C99 semantics (for Clang)
+
+  * explicitly add -lfdk-aac to LDADD in Makefile.am
+
+  * add some files to EXTRA_DIST in Makefile.am
+
+  * fixed a typo in usage message [v0.0.5]
+
+2013-01-06  nu774 <honeycomb77@gmail.com>
+
+  * add MSVC projects
+
+  * add .gitattributes
+
+  * more tweak on configure.ac and Makefile.am (take care of getopt_long) [v0.0.4]
+
+  * use fstat() to test seekability of input file
+
+  * retrieve bitrate for tool tag with aacEncoder_GetParam()
+
+  * output to current working directory by default
+
+2013-01-05  nu774 <honeycomb77@gmail.com>
+
+  * zero clear LIB_INFO before calling aacEncGetLibInfo() [v0.0.3]
+
+  * tweak configure.ac and Makefile.am
+
+  * update version.h [v0.0.2]
+
+  * fixed to clip before converting float to int
+
+  * initial commit [v0.0.1]
+

MSVC/fdkaac.vcxproj

@@ -96,24 +96,33 @@ copy ..\fdk-aac\libSYS\include\machine_type.h include\fdk-aac\ </Command>
   <ItemGroup>
     <ClCompile Include="..\missings\getopt.c" />
     <ClCompile Include="..\src\aacenc.c" />
+    <ClCompile Include="..\src\caf_reader.c" />
     <ClCompile Include="..\src\compat_win32.c" />
+    <ClCompile Include="..\src\extrapolater.c" />
+    <ClCompile Include="..\src\lpc.c" />
     <ClCompile Include="..\src\lpcm.c" />
     <ClCompile Include="..\src\m4af.c" />
     <ClCompile Include="..\src\main.c" />
     <ClCompile Include="..\src\metadata.c" />
     <ClCompile Include="..\src\parson.c" />
+    <ClCompile Include="..\src\pcm_readhelper.c" />
+    <ClCompile Include="..\src\pcm_sint16_converter.c" />
     <ClCompile Include="..\src\progress.c" />
     <ClCompile Include="..\src\wav_reader.c" />
   </ItemGroup>
   <ItemGroup>
     <ClInclude Include="..\missings\getopt.h" />
     <ClInclude Include="..\src\aacenc.h" />
+    <ClInclude Include="..\src\caf_reader.h" />
+    <ClInclude Include="..\src\catypes.h" />
     <ClInclude Include="..\src\compat.h" />
+    <ClInclude Include="..\src\lpc.h" />
     <ClInclude Include="..\src\lpcm.h" />
     <ClInclude Include="..\src\m4af.h" />
     <ClInclude Include="..\src\m4af_endian.h" />
     <ClInclude Include="..\src\metadata.h" />
     <ClInclude Include="..\src\parson.h" />
+    <ClInclude Include="..\src\pcm_reader.h" />
     <ClInclude Include="..\src\progress.h" />
     <ClInclude Include="..\src\wav_reader.h" />
   </ItemGroup>

MSVC/fdkaac.vcxproj.filters

@@ -18,9 +18,18 @@
     <ClCompile Include="..\src\aacenc.c">
       <Filter>Source Files</Filter>
     </ClCompile>
+    <ClCompile Include="..\src\caf_reader.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
     <ClCompile Include="..\src\compat_win32.c">
       <Filter>Source Files</Filter>
     </ClCompile>
+    <ClCompile Include="..\src\extrapolater.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\src\lpc.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
     <ClCompile Include="..\src\lpcm.c">
       <Filter>Source Files</Filter>
     </ClCompile>
@@ -30,6 +39,12 @@
     <ClCompile Include="..\src\main.c">
       <Filter>Source Files</Filter>
     </ClCompile>
+    <ClCompile Include="..\src\pcm_readhelper.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\src\pcm_sint16_converter.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
     <ClCompile Include="..\src\progress.c">
       <Filter>Source Files</Filter>
     </ClCompile>
@@ -44,9 +59,18 @@
     <ClInclude Include="..\src\aacenc.h">
       <Filter>Header Files</Filter>
     </ClInclude>
+    <ClInclude Include="..\src\caf_reader.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="..\src\catypes.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
     <ClInclude Include="..\src\compat.h">
       <Filter>Header Files</Filter>
     </ClInclude>
+    <ClInclude Include="..\src\lpc.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
     <ClInclude Include="..\src\lpcm.h">
       <Filter>Header Files</Filter>
     </ClInclude>
@@ -56,6 +80,9 @@
     <ClInclude Include="..\src\m4af_endian.h">
       <Filter>Header Files</Filter>
     </ClInclude>
+    <ClInclude Include="..\src\pcm_reader.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
     <ClInclude Include="..\src\progress.h">
       <Filter>Header Files</Filter>
     </ClInclude>

Makefile.am

@@ -5,14 +5,21 @@ bin_PROGRAMS = fdkaac
 
 fdkaac_SOURCES = \
     src/aacenc.c               \
+    src/caf_reader.c           \
+    src/extrapolater.c         \
+    src/lpc.c                  \
     src/lpcm.c                 \
     src/m4af.c                 \
     src/main.c                 \
     src/metadata.c             \
     src/parson.c               \
+    src/pcm_readhelper.c       \
+    src/pcm_sint16_converter.c \
     src/progress.c             \
     src/wav_reader.c
 
+dist_man_MANS = man/fdkaac.1
+
 fdkaac_LDADD = \
     @LIBICONV@ -lfdk-aac -lm
 

README

@@ -1,102 +1,332 @@
-==========================================================================
+% FDKAAC(1)
-fdkaac - command line frontend encoder for libfdk-aac
+% nu774 <honeycomb77@gmail.com>
-==========================================================================
+% November, 2013
 
-Prerequisites
+NAME
--------------
+====
-You need libfdk-aac.
-On Posix environment, you will also need GNU gettext (for iconv.m4) and
-GNU autoconf/automake.
 
-How to build on Posix environment
+fdkaac - command line frontend for libfdk-aac encoder
----------------------------------
-First, you need to build libfdk-aac and install on your system.
-Once you have done it, the following will do the task.
-(MinGW build can be done the same way, and doesn't require gettext/iconv)
 
-$ autoreconf -i
+SYNOPSIS
-$ ./configure && make && make install
+========
 
-How to build on MSVC
+**fdkaac** [OPTIONS] [FILE]
---------------------
-First you have to extract libfdk-aac source here, so that directory tree will
-look like the following:
-+- fdk-aac ---+-documentation
-|             +-libAACdec
-|             +-libAACenc
-|             :
-+- m4
-+- missings
-+- MSVC
-+- src
 
-MSVC solution for Visual Studio 2010 is under MSVC directory.
+DESCRIPTION
+===========
 
-Tagging Options
+**fdkaac** reads linear PCM audio in either WAV, raw PCM, or CAF format,
----------------
+and encodes it into either M4A / AAC file.
-Generic tagging options like --tag, --tag-from-file, --long-tag allows you
-to set arbitrary tags.
-Available tags and their fcc (four char code) for --tag and --tag-from-file
-can be found at http://code.google.com/p/mp4v2/wiki/iTunesMetadata
 
-For tags such as Artist where first char of fcc is copyright sign,
+If the input file is "-", data is read from stdin. Likewise, if the
-you can skip first char and just say like --tag="ART:Foo Bar" or
+output file is "-", data is written to stdout if one of streamable AAC
---tag-from-file=lyr:/path/to/your/lyrics.txt
+transport formats are selected by **-f**.
 
-Currently, --tag-from-file just stores file contents into m4a without any
+When CAF input and M4A output is used, tags in CAF file are copied into
-character encoding / line terminater conversion.
+the resulting M4A.
-Therefore, only use UTF-8 (without BOM) when setting text tags by this option.
 
-On the other hand, --tag / --long-tag (and other command line arguments) are
+OPTIONS
-converted from locale character encoding to UTF-8 on Posix environment.
+=======
-On Windows, command line arguments are always treated as Unicode.
 
-Tagging using JSON
+-h, --help
-------------------
+:   Show command help
-With --tag-from-json, fdkaac can read JSON file and set tags from it.
-By default, tags are assumed to be in the root object(dictionary) like this:
 
-{
+-o \<FILE\>
-  "title": "No Expectations",
+:   Output filename.
-  "artist": "The Rolling Stones",
-  "album": "Beggars Banquet",
-  "track": 2
-}
 
-In this case, you can simply specify the filename like:
+-p, --profile \<n\>
---tag-from-json=/path/to/json
+:   Target profile (MPEG4 audio object type, AOT)
 
-If the object containing tags is placed somewhere else, you can optionally
+    2
-specify the path of the object with dotted notation.
+    :   MPEG-4 AAC LC (default)
 
-{
+    5
-  "format" : {
+    :   MPEG-4 HE-AAC (SBR)
-    "filename" : "Middle Curse.flac",
-    "nb_streams" : 1,
-    "format_name" : "flac",
-    "format_long_name" : "raw FLAC",
-    "start_time" : "N/A",
-    "duration" : "216.146667",
-    "size" : "11851007.000000",
-    "bit_rate" : "438628.000000",
-    "tags" : {
-      "ALBUM" : "Scary World Theory",
-      "ARTIST" : "Lali Puna",
-      "DATE" : "2001",
-      "DISCID" : "9208CC0A",
-      "TITLE" : "Middle Curse",
-      "TRACKTOTAL" : "10",
-      "track" : "2"
-    }
-  }
-}
 
-In this example, tags are placed under the object "format.tags".
+    29
-("format" is a child of the root, and "tags" is a child of the "format").
+    :   MPEG-4 HE-AAC v2 (SBR+PS)
-In this case, you can say:
---tag-from-json=/path/to/json?format.tags
 
-For your information, ffprobe of ffmpeg project (or avprobe of libav) can
+    23
-output media information/metadata in json format like this.
+    :   MPEG-4 AAC LD
+
+    39
+    :   MPEG-4 AAC ELD
+
+    129
+    :   MPEG-2 AAC LC
+
+    132
+    :   MPEG-2 HE-AAC (SBR)
+
+    156
+    :   MPEG-2 HE-AAC v2 (SBR+PS)
+
+-b, --bitrate \<n\>
+:   Target bitrate (for CBR)
+
+-m, --bitrate-mode \<n\>
+:   Bitrate configuration mode. Available VBR quality value depends on
+    other parameters such as profile, sample rate, or number of
+    channels.
+
+    0
+    :   CBR (default)
+
+    1-5
+    :   VBR (higher value -\> higher bitrate)
+
+-w, --bandwidth \<n\>
+:   Frequency bandwidth (lowpass cut-off frequency) in Hz. Available on
+    AAC LC only.
+
+-a, --afterburner \<n\>
+:   Configure afterburner mode. When enabled, quality is increased at
+    the expense of additional computational workload.
+
+    0
+    :   Off
+
+    1
+    :   On (default)
+
+-L, --lowdelay-sbr \<n\>
+:   Configure SBR activity on AAC ELD.
+
+    -1
+    :   Use ELD SBR auto configuration
+
+    0
+    :   Disable SBR on ELD (default)
+
+    1
+    :   Enable SBR on ELD
+
+-s, --sbr-ratio \<n\>
+:   Controls activation of downsampled SBR.
+
+    0
+    :   Use lib default (default)
+
+    1
+    :   Use downsampled SBR (default for ELD+SBR)
+
+    2
+    :   Use dual-rate SBR (default for HE-AAC)
+
+    Dual-rate SBR is what is normally used for HE-AAC, where AAC is
+    encoded at half the sample rate of SBR, hence "dual rate". On the
+    other hand, downsampled SBR uses same sample rate for both of AAC
+    and SBR (single rate), therefore downsampled SBR typically consumes
+    more bitrate.
+
+    Downsampled SBR is newly introduced feature in FDK encoder library
+    version 3.4.12. When libfdk-aac in the system doesn't support this,
+    dual-rate SBR will be used. When available, dual-rate SBR is the
+    default for HE-AAC and downsampled SBR is the default for ELD+SBR.
+
+    Note that downsampled HE-AAC is not so common as dual-rate one. When
+    downsampled HE-AAC is selected, **fdkaac** is forced to choose
+    explicit hierarchical SBR signaling, which (at least) iTunes doesn't
+    accept.
+
+-f, --transport-format \<n\>
+:   Transport format. Tagging and gapless playback is only available on
+    M4A. Streaming to stdout is only available on others.
+
+    0
+    :   M4A (default)
+
+    1
+    :   ADIF
+
+    2
+    :   ADTS
+
+    6
+    :   LATM MCP=1
+
+    7
+    :   LATM MCP=0
+
+    10
+    :   LOAS/LATM (LATM within LOAS)
+
+-C, --adts-crc-check
+:   Add CRC protection on ADTS header.
+
+-h, --header-period \<n\>
+:   StreamMuxConfig/PCE repetition period in the transport layer.
+
+-G, --gapless-mode \<n\>
+:   Method to declare amount of encoder delay (and padding) in M4A
+    container. These values are mandatory for proper gapless playback on
+    player side.
+
+    0
+    :   iTunSMPB (default)
+
+    1
+    :   ISO standard (edts and sgpd)
+
+    2
+    :   Both
+
+--include-sbr-delay
+:   When specified, count SBR decoder delay in encoder delay.
+
+    This is not iTunes compatible and will lead to gapless playback
+    issue on LC only decoder, but this is the default behavior of FDK
+    library.
+
+    Whether counting SBR decoder delay in encoder delay or not result in
+    incompatibility in gapless playback. You should pick which one will
+    work for your favorite player.
+
+    However, it's better not to choose SBR at all if you want gapless
+    playback. LC doesn't have such issues.
+
+-I, --ignorelength
+:   Ignore length field of data chunk in input WAV file.
+
+-S, --silent
+:   Don't print progress messages.
+
+--moov-before-mdat
+:   Place moov box before mdat box in M4A container. This option might
+    be important for some hardware players, that are known to refuse
+    moov box placed after mdat box.
+
+-R, --raw
+:   Regard input as raw PCM.
+
+--raw-channels \<n\>
+:   Specify number of channels of raw input (default: 2)
+
+--raw-rate \<n\>
+:   Specify sample rate of raw input (default: 44100)
+
+--raw-format \<spec\>
+:   Specify sample format of raw input (default: "S16L"). **Spec** is as
+    the following (case insensitive):
+
+    1st char -- type of sample
+    :   **S** (igned) | **U** (nsigned) | **F** (loat)
+
+    2nd part (in digits)
+    :   bits per channel
+
+    Last char -- endianness (can be omitted)
+    :   **L** (ittle, default) | **B** (ig)
+
+--title \<string\>
+:   Set title tag.
+
+--artist \<string\>
+:   Set artist tag.
+
+--album \<string\>
+:   Set album tag.
+
+--genre \<string\>
+:   Set genre tag.
+
+--date \<string\>
+:   Set date tag.
+
+--composer \<string\>
+:   Set composer tag.
+
+--grouping \<string\>
+:   Set grouping tag.
+
+--comment \<string\>
+:   Set comment tag.
+
+--album-artist \<string\>
+:   Set album artist tag.
+
+--track \<number[/total]\>
+:   Set track tag, with or without number of total tracks.
+
+--disk \<number[/total]\>
+:   Set disk tag, with or without number of total discs.
+
+--tempo \<n\>
+:   Set tempo (BPM) tag.
+
+--tag \<fcc\>:\<value\>
+:   Set iTunes predefined tag with explicit fourcc key and value. See
+    [https://code.google.com/p/mp4v2/wiki/iTunesMetadata](https://code.google.com/p/mp4v2/wiki/iTunesMetadata)
+    for known predefined keys. You can omit first char of **fcc** when
+    it is the copyright sign.
+
+--tag-from-file \<fcc\>:\<filename\>
+:   Same as --tag, but set content of file as tag value.
+
+--long-tag \<name\>:\<value\>
+:   Set arbitrary tag as iTunes custom metadata. Stored in
+    com.apple.iTunes field.
+
+--tag-from-json \<filename[?dot\_notation]\>
+:   Read tags from JSON. By default, tags are assumed to be direct
+    children of the root object in JSON. Optionally you can specify
+    arbitrary dot notation to locate the object containing tags.
+
+
+EXAMPLES
+========
+
+Encode WAV file into a M4A file. MPEG4 AAC LC, VBR quality 3:
+
+    fdkaac -m3 foo.wav
+
+Encode WAV file into a M4A file. MPEG4 HE-AAC, bitrate 64kbps:
+
+    fdkaac -p5 -b64 foo.wav
+
+Piping from **ffmpeg** (you need version supporting CAF output):
+
+    ffmpeg -i foo.flac -f caf - | fdkaac -b128 - -o foo.m4a
+
+Import tags via json:
+
+    ffprobe -v 0 -of json -show_format foo.flac >foo.json
+
+    flac -dc foo.flac | \
+    fdkaac - -ox.m4a -m2 --import-tag-from-json=foo.json?format.tags
+
+NOTES
+=====
+
+Upto 32bit integer or 64bit floating point format is supported as input.
+However, FDK library is implemented based on fixed point math and only
+supports 16bit integer PCM. Therefore, be wary of clipping. You might
+want to dither/noise shape beforehand when your input has higher
+resolution.
+
+Following channel layouts are supported by the encoder.
+
+1ch
+:   C
+
+2ch
+:   L R
+
+3ch
+:   C L R
+
+4ch
+:   C L R Cs
+
+5ch
+:   C L R Ls Rs
+
+5.1ch
+:   C L R Ls Rs LFE
+
+7.1ch (front)
+:   C Lc Rc L R Ls Rs LFE
+
+7.1ch (rear)
+:   C L R Ls Rs Rls Rrs LFE
 
-Note that not all tags can be read/written this way.

genman.sh

@@ -0,0 +1,2 @@
+#!/bin/sh
+pandoc -s -f markdown -t man README >fdkaac.1 && mv -f fdkaac.1 man/fdkaac.1

git2changelog.py

@@ -0,0 +1,44 @@
+#!/usr/bin/env python
+
+# Copyright (C) 2013 nu774
+# For conditions of distribution and use, see copyright notice in COPYING
+
+import sys
+import re
+from subprocess import Popen, PIPE
+from itertools import groupby
+from collections import namedtuple
+
+GITLOG_FMT = 'commit %H%nauthor %cn <%ae>%ndate %ad%nsubject %s%nref %d%n%n'
+GITLOG_CMD = ['git','log','--date=short','--format={0}'.format(GITLOG_FMT)]
+
+Commit = namedtuple('Commit', 'commit author date subject ref')
+
+def parse_gitlog(stream):
+    re_decode_ref = re.compile(r'(?<=\()([^,)]+)')
+    re_strip_tag = re.compile(r'^tag: ')
+    commit = dict()
+    for line in stream:
+        fields = line.decode('utf-8').rstrip('\r\n').split(' ', 1)
+        if len(fields) == 2:
+            key, value = fields
+            if key == 'ref':
+                m = re_decode_ref.search(value)
+                if m:
+                    value = ' [{0}]'.format(re_strip_tag.sub('', m.group()))
+                else:
+                    value = ''
+            commit[key] = value
+        elif commit:
+            yield Commit(**commit) 
+            commit = dict()
+
+output=sys.stdout.write
+
+with Popen(GITLOG_CMD, shell=False, stdout=PIPE).stdout as pipe:
+    commits = parse_gitlog(pipe)
+    commits_by_date_author = groupby(commits, key=lambda x: (x.date, x.author))
+    for (date, author), commits in commits_by_date_author:
+        output(u'{0}  {1}\n\n'.format(date, author).encode('utf-8'))
+        for c in commits:
+            output(u'  * {0}{1}\n\n'.format(c.subject, c.ref).encode('utf-8'))

man/fdkaac.1

@@ -0,0 +1,493 @@
+.TH FDKAAC 1 "November, 2013" 
+.SH NAME
+.PP
+fdkaac \- command line frontend for libfdk\-aac encoder
+.SH SYNOPSIS
+.PP
+\f[B]fdkaac\f[] [OPTIONS][FILE]
+.SH DESCRIPTION
+.PP
+\f[B]fdkaac\f[] reads linear PCM audio in either WAV, raw PCM, or CAF
+format, and encodes it into either M4A / AAC file.
+.PP
+If the input file is "\-", data is read from stdin.
+Likewise, if the output file is "\-", data is written to stdout if one
+of streamable AAC transport formats are selected by \f[B]\-f\f[].
+.PP
+When CAF input and M4A output is used, tags in CAF file are copied into
+the resulting M4A.
+.SH OPTIONS
+.TP
+.B \-h, \-\-help
+Show command help
+.RS
+.RE
+.TP
+.B \-o <FILE>
+Output filename.
+.RS
+.RE
+.TP
+.B \-p, \-\-profile <n>
+Target profile (MPEG4 audio object type, AOT)
+.RS
+.TP
+.B 2
+MPEG\-4 AAC LC (default)
+.RS
+.RE
+.TP
+.B 5
+MPEG\-4 HE\-AAC (SBR)
+.RS
+.RE
+.TP
+.B 29
+MPEG\-4 HE\-AAC v2 (SBR+PS)
+.RS
+.RE
+.TP
+.B 23
+MPEG\-4 AAC LD
+.RS
+.RE
+.TP
+.B 39
+MPEG\-4 AAC ELD
+.RS
+.RE
+.TP
+.B 129
+MPEG\-2 AAC LC
+.RS
+.RE
+.TP
+.B 132
+MPEG\-2 HE\-AAC (SBR)
+.RS
+.RE
+.TP
+.B 156
+MPEG\-2 HE\-AAC v2 (SBR+PS)
+.RS
+.RE
+.RE
+.TP
+.B \-b, \-\-bitrate <n>
+Target bitrate (for CBR)
+.RS
+.RE
+.TP
+.B \-m, \-\-bitrate\-mode <n>
+Bitrate configuration mode.
+Available VBR quality value depends on other parameters such as profile,
+sample rate, or number of channels.
+.RS
+.TP
+.B 0
+CBR (default)
+.RS
+.RE
+.TP
+.B 1\-5
+VBR (higher value \-> higher bitrate)
+.RS
+.RE
+.RE
+.TP
+.B \-w, \-\-bandwidth <n>
+Frequency bandwidth (lowpass cut\-off frequency) in Hz.
+Available on AAC LC only.
+.RS
+.RE
+.TP
+.B \-a, \-\-afterburner <n>
+Configure afterburner mode.
+When enabled, quality is increased at the expense of additional
+computational workload.
+.RS
+.TP
+.B 0
+Off
+.RS
+.RE
+.TP
+.B 1
+On (default)
+.RS
+.RE
+.RE
+.TP
+.B \-L, \-\-lowdelay\-sbr <n>
+Configure SBR activity on AAC ELD.
+.RS
+.TP
+.B \-1
+Use ELD SBR auto configuration
+.RS
+.RE
+.TP
+.B 0
+Disable SBR on ELD (default)
+.RS
+.RE
+.TP
+.B 1
+Enable SBR on ELD
+.RS
+.RE
+.RE
+.TP
+.B \-s, \-\-sbr\-ratio <n>
+Controls activation of downsampled SBR.
+.RS
+.TP
+.B 0
+Use lib default (default)
+.RS
+.RE
+.TP
+.B 1
+Use downsampled SBR (default for ELD+SBR)
+.RS
+.RE
+.TP
+.B 2
+Use dual\-rate SBR (default for HE\-AAC)
+.RS
+.RE
+.PP
+Dual\-rate SBR is what is normally used for HE\-AAC, where AAC is
+encoded at half the sample rate of SBR, hence "dual rate".
+On the other hand, downsampled SBR uses same sample rate for both of AAC
+and SBR (single rate), therefore downsampled SBR typically consumes more
+bitrate.
+.PP
+Downsampled SBR is newly introduced feature in FDK encoder library
+version 3.4.12.
+When libfdk\-aac in the system doesn\[aq]t support this, dual\-rate SBR
+will be used.
+When available, dual\-rate SBR is the default for HE\-AAC and
+downsampled SBR is the default for ELD+SBR.
+.PP
+Note that downsampled HE\-AAC is not so common as dual\-rate one.
+When downsampled HE\-AAC is selected, \f[B]fdkaac\f[] is forced to
+choose explicit hierarchical SBR signaling, which (at least) iTunes
+doesn\[aq]t accept.
+.RE
+.TP
+.B \-f, \-\-transport\-format <n>
+Transport format.
+Tagging and gapless playback is only available on M4A.
+Streaming to stdout is only available on others.
+.RS
+.TP
+.B 0
+M4A (default)
+.RS
+.RE
+.TP
+.B 1
+ADIF
+.RS
+.RE
+.TP
+.B 2
+ADTS
+.RS
+.RE
+.TP
+.B 6
+LATM MCP=1
+.RS
+.RE
+.TP
+.B 7
+LATM MCP=0
+.RS
+.RE
+.TP
+.B 10
+LOAS/LATM (LATM within LOAS)
+.RS
+.RE
+.RE
+.TP
+.B \-C, \-\-adts\-crc\-check
+Add CRC protection on ADTS header.
+.RS
+.RE
+.TP
+.B \-h, \-\-header\-period <n>
+StreamMuxConfig/PCE repetition period in the transport layer.
+.RS
+.RE
+.TP
+.B \-G, \-\-gapless\-mode <n>
+Method to declare amount of encoder delay (and padding) in M4A
+container.
+These values are mandatory for proper gapless playback on player side.
+.RS
+.TP
+.B 0
+iTunSMPB (default)
+.RS
+.RE
+.TP
+.B 1
+ISO standard (edts and sgpd)
+.RS
+.RE
+.TP
+.B 2
+Both
+.RS
+.RE
+.RE
+.TP
+.B \-\-include\-sbr\-delay
+When specified, count SBR decoder delay in encoder delay.
+.RS
+.PP
+This is not iTunes compatible and will lead to gapless playback issue on
+LC only decoder, but this is the default behavior of FDK library.
+.PP
+Whether counting SBR decoder delay in encoder delay or not result in
+incompatibility in gapless playback.
+You should pick which one will work for your favorite player.
+.PP
+However, it\[aq]s better not to choose SBR at all if you want gapless
+playback.
+LC doesn\[aq]t have such issues.
+.RE
+.TP
+.B \-I, \-\-ignorelength
+Ignore length field of data chunk in input WAV file.
+.RS
+.RE
+.TP
+.B \-S, \-\-silent
+Don\[aq]t print progress messages.
+.RS
+.RE
+.TP
+.B \-\-moov\-before\-mdat
+Place moov box before mdat box in M4A container.
+This option might be important for some hardware players, that are known
+to refuse moov box placed after mdat box.
+.RS
+.RE
+.TP
+.B \-R, \-\-raw
+Regard input as raw PCM.
+.RS
+.RE
+.TP
+.B \-\-raw\-channels <n>
+Specify number of channels of raw input (default: 2)
+.RS
+.RE
+.TP
+.B \-\-raw\-rate <n>
+Specify sample rate of raw input (default: 44100)
+.RS
+.RE
+.TP
+.B \-\-raw\-format <spec>
+Specify sample format of raw input (default: "S16L").
+\f[B]Spec\f[] is as the following (case insensitive):
+.RS
+.TP
+.B 1st char \-\- type of sample
+\f[B]S\f[] (igned) | \f[B]U\f[] (nsigned) | \f[B]F\f[] (loat)
+.RS
+.RE
+.TP
+.B 2nd part (in digits)
+bits per channel
+.RS
+.RE
+.TP
+.B Last char \-\- endianness (can be omitted)
+\f[B]L\f[] (ittle, default) | \f[B]B\f[] (ig)
+.RS
+.RE
+.RE
+.TP
+.B \-\-title <string>
+Set title tag.
+.RS
+.RE
+.TP
+.B \-\-artist <string>
+Set artist tag.
+.RS
+.RE
+.TP
+.B \-\-album <string>
+Set album tag.
+.RS
+.RE
+.TP
+.B \-\-genre <string>
+Set genre tag.
+.RS
+.RE
+.TP
+.B \-\-date <string>
+Set date tag.
+.RS
+.RE
+.TP
+.B \-\-composer <string>
+Set composer tag.
+.RS
+.RE
+.TP
+.B \-\-grouping <string>
+Set grouping tag.
+.RS
+.RE
+.TP
+.B \-\-comment <string>
+Set comment tag.
+.RS
+.RE
+.TP
+.B \-\-album\-artist <string>
+Set album artist tag.
+.RS
+.RE
+.TP
+.B \-\-track <number[/total]>
+Set track tag, with or without number of total tracks.
+.RS
+.RE
+.TP
+.B \-\-disk <number[/total]>
+Set disk tag, with or without number of total discs.
+.RS
+.RE
+.TP
+.B \-\-tempo <n>
+Set tempo (BPM) tag.
+.RS
+.RE
+.TP
+.B \-\-tag <fcc>:<value>
+Set iTunes predefined tag with explicit fourcc key and value.
+See <https://code.google.com/p/mp4v2/wiki/iTunesMetadata> for known
+predefined keys.
+You can omit first char of \f[B]fcc\f[] when it is the copyright sign.
+.RS
+.RE
+.TP
+.B \-\-tag\-from\-file <fcc>:<filename>
+Same as \-\-tag, but set content of file as tag value.
+.RS
+.RE
+.TP
+.B \-\-long\-tag <name>:<value>
+Set arbitrary tag as iTunes custom metadata.
+Stored in com.apple.iTunes field.
+.RS
+.RE
+.TP
+.B \-\-tag\-from\-json <filename[?dot_notation]>
+Read tags from JSON.
+By default, tags are assumed to be direct children of the root object in
+JSON.
+Optionally you can specify arbitrary dot notation to locate the object
+containing tags.
+.RS
+.RE
+.SH EXAMPLES
+.PP
+Encode WAV file into a M4A file.
+MPEG4 AAC LC, VBR quality 3:
+.IP
+.nf
+\f[C]
+fdkaac\ \-m3\ foo.wav
+\f[]
+.fi
+.PP
+Encode WAV file into a M4A file.
+MPEG4 HE\-AAC, bitrate 64kbps:
+.IP
+.nf
+\f[C]
+fdkaac\ \-p5\ \-b64\ foo.wav
+\f[]
+.fi
+.PP
+Piping from \f[B]ffmpeg\f[] (you need version supporting CAF output):
+.IP
+.nf
+\f[C]
+ffmpeg\ \-i\ foo.flac\ \-f\ caf\ \-\ |\ fdkaac\ \-b128\ \-\ \-o\ foo.m4a
+\f[]
+.fi
+.PP
+Import tags via json:
+.IP
+.nf
+\f[C]
+ffprobe\ \-v\ 0\ \-of\ json\ \-show_format\ foo.flac\ >foo.json
+
+flac\ \-dc\ foo.flac\ |\ \\
+fdkaac\ \-\ \-ox.m4a\ \-m2\ \-\-import\-tag\-from\-json=foo.json?format.tags
+\f[]
+.fi
+.SH NOTES
+.PP
+Upto 32bit integer or 64bit floating point format is supported as input.
+However, FDK library is implemented based on fixed point math and only
+supports 16bit integer PCM.
+Therefore, be wary of clipping.
+You might want to dither/noise shape beforehand when your input has
+higher resolution.
+.PP
+Following channel layouts are supported by the encoder.
+.TP
+.B 1ch
+C
+.RS
+.RE
+.TP
+.B 2ch
+L R
+.RS
+.RE
+.TP
+.B 3ch
+C L R
+.RS
+.RE
+.TP
+.B 4ch
+C L R Cs
+.RS
+.RE
+.TP
+.B 5ch
+C L R Ls Rs
+.RS
+.RE
+.TP
+.B 5.1ch
+C L R Ls Rs LFE
+.RS
+.RE
+.TP
+.B 7.1ch (front)
+C Lc Rc L R Ls Rs LFE
+.RS
+.RE
+.TP
+.B 7.1ch (rear)
+C L R Ls Rs Rls Rrs LFE
+.RS
+.RE
+.SH AUTHORS
+nu774 <honeycomb77@gmail.com>.

src/aacenc.c

@@ -10,8 +10,20 @@
 #endif
 #include <stdio.h>
 #include <stdlib.h>
+#include <string.h>
 #include "aacenc.h"
 
+int aacenc_is_sbr_ratio_available()
+{
+#if AACENCODER_LIB_VL0 < 3 || (AACENCODER_LIB_VL0==3 && AACENCODER_LIB_VL1<4)
+    return 0;
+#else
+    LIB_INFO lib_info;
+    aacenc_get_lib_info(&lib_info);
+    return lib_info.version > 0x03040000;
+#endif
+}
+
 int aacenc_is_sbr_active(const aacenc_param_t *params)
 {
     switch (params->profile) {
@@ -25,15 +37,42 @@ int aacenc_is_sbr_active(const aacenc_param_t *params)
     return 0;
 }
 
+int aacenc_is_dual_rate_sbr(const aacenc_param_t *params)
+{
+    if (params->profile == AOT_PS || params->profile == AOT_MP2_PS)
+        return 1;
+    else if (params->profile == AOT_SBR || params->profile == AOT_MP2_SBR)
+        return params->sbr_ratio == 0 || params->sbr_ratio == 2;
+    else if (params->profile == AOT_ER_AAC_ELD && params->lowdelay_sbr)
+        return params->sbr_ratio == 2;
+    return 0;
+}
+
+void aacenc_get_lib_info(LIB_INFO *info)
+{
+    LIB_INFO *lib_info = 0;
+    lib_info = calloc(FDK_MODULE_LAST, sizeof(LIB_INFO));
+    if (aacEncGetLibInfo(lib_info) == AACENC_OK) {
+        int i;
+        for (i = 0; i < FDK_MODULE_LAST; ++i) {
+            if (lib_info[i].module_id == FDK_AACENC) {
+                memcpy(info, &lib_info[i], sizeof(LIB_INFO));
+                break;
+            }
+        }
+    }
+    free(lib_info);
+}
+
 static
 int aacenc_channel_mode(const pcm_sample_description_t *format)
 {
     uint32_t chanmask = format->channel_mask;
 
-    if (format->channels_per_frame > 6)
+    if (format->channels_per_frame > 8)
         return 0;
     if (!chanmask) {
-        static uint32_t defaults[] = { 0x4, 0x3, 0x7, 0, 0x37, 0x3f };
+        static uint32_t defaults[] = { 0x4, 0x3, 0x7, 0, 0x37, 0x3f, 0, 0x63f };
         chanmask = defaults[format->channels_per_frame - 1];
     }
     switch (chanmask) {
@@ -45,6 +84,10 @@ int aacenc_channel_mode(const pcm_sample_description_t *format)
     case 0x107: return MODE_1_2_1;
     case 0x607: return MODE_1_2_2;
     case 0x60f: return MODE_1_2_2_1;
+#if AACENCODER_LIB_VL0 > 3 || (AACENCODER_LIB_VL0==3 && AACENCODER_LIB_VL1>=4)
+    case 0xff:  return MODE_1_2_2_2_1;
+    case 0x63f: return MODE_7_1_REAR_SURROUND;
+#endif
     }
     return 0;
 }
@@ -55,8 +98,11 @@ int aacenc_init(HANDLE_AACENCODER *encoder, const aacenc_param_t *params,
 {
     int channel_mode;
     int aot;
+    LIB_INFO lib_info;
 
     *encoder = 0;
+    aacenc_get_lib_info(&lib_info);
+
     if ((channel_mode = aacenc_channel_mode(format)) == 0) {
         fprintf(stderr, "ERROR: unsupported channel layout\n");
         goto FAIL;
@@ -82,13 +128,21 @@ int aacenc_init(HANDLE_AACENCODER *encoder, const aacenc_param_t *params,
         fprintf(stderr, "ERROR: unsupported sample rate\n");
         goto FAIL;
     }
-    aacEncoder_SetParam(*encoder, AACENC_CHANNELMODE, channel_mode);
+    if (aacEncoder_SetParam(*encoder, AACENC_CHANNELMODE,
+                            channel_mode) != AACENC_OK) {
+        fprintf(stderr, "ERROR: unsupported channel mode\n");
+        goto FAIL;
+    }
     aacEncoder_SetParam(*encoder, AACENC_BANDWIDTH, params->bandwidth);
     aacEncoder_SetParam(*encoder, AACENC_CHANNELORDER, 1);
     aacEncoder_SetParam(*encoder, AACENC_AFTERBURNER, !!params->afterburner);
 
-    if (aot == AOT_ER_AAC_ELD && params->lowdelay_sbr)
+    aacEncoder_SetParam(*encoder, AACENC_SBR_MODE, params->lowdelay_sbr);
-        aacEncoder_SetParam(*encoder, AACENC_SBR_MODE, 1);
+
+#if AACENCODER_LIB_VL0 > 3 || (AACENCODER_LIB_VL0==3 && AACENCODER_LIB_VL1>=4)
+    if (lib_info.version > 0x03040000)
+        aacEncoder_SetParam(*encoder, AACENC_SBR_RATIO, params->sbr_ratio);
+#endif
 
     if (aacEncoder_SetParam(*encoder, AACENC_TRANSMUX,
                             params->transport_format) != AACENC_OK) {
@@ -124,7 +178,7 @@ FAIL:
 int aac_encode_frame(HANDLE_AACENCODER encoder,
                      const pcm_sample_description_t *format,
                      const int16_t *input, unsigned iframes,
-                     uint8_t **output, uint32_t *olen, uint32_t *osize)
+                     aacenc_frame_t *output)
 {
     uint32_t ilen = iframes * format->channels_per_frame;
     AACENC_BufDesc ibdesc = { 0 }, obdesc = { 0 };
@@ -142,12 +196,14 @@ int aac_encode_frame(HANDLE_AACENCODER encoder,
     unsigned channel_mode, obytes;
 
     channel_mode = aacEncoder_GetParam(encoder, AACENC_CHANNELMODE);
-    obytes = 6144 / 8 * channel_mode + 7;
+    obytes = 6144 / 8 * channel_mode;
-    if (!*output || *osize < obytes) {
+    if (!output->data || output->capacity < obytes) {
-        *osize = obytes;
+        uint8_t *p = realloc(output->data, obytes);
-        *output = realloc(*output, obytes);
+        if (!p) return -1;
+        output->capacity = obytes;
+        output->data = p;
     }
-    obufs[0] = *output;
+    obufs[0] = output->data;
     obuf_sizes[0] = obytes;
 
     iargs.numInSamples = ilen ? ilen : -1; /* -1 for signaling EOF */
@@ -167,6 +223,6 @@ int aac_encode_frame(HANDLE_AACENCODER encoder,
         fprintf(stderr, "ERROR: aacEncEncode() failed\n");
         return -1;
     }
-    *olen = oargs.numOutBytes;
+    output->size = oargs.numOutBytes;
-    return oargs.numInSamples;
+    return oargs.numInSamples / format->channels_per_frame;
 }

src/aacenc.h

@@ -15,6 +15,7 @@
     unsigned bandwidth; \
     unsigned afterburner; \
     unsigned lowdelay_sbr; \
+    unsigned sbr_ratio; \
     unsigned sbr_signaling; \
     unsigned transport_format; \
     unsigned adts_crc_check; \
@@ -24,8 +25,19 @@ typedef struct aacenc_param_t {
     AACENC_PARAMS
 } aacenc_param_t;
 
+typedef struct aacenc_frame_t {
+    uint8_t *data;
+    uint32_t size, capacity;
+} aacenc_frame_t;
+
+int aacenc_is_sbr_ratio_available();
+
 int aacenc_is_sbr_active(const aacenc_param_t *params);
 
+int aacenc_is_dual_rate_sbr(const aacenc_param_t *params);
+
+void aacenc_get_lib_info(LIB_INFO *info);
+
 int aacenc_init(HANDLE_AACENCODER *encoder, const aacenc_param_t *params,
                 const pcm_sample_description_t *format,
                 AACENC_InfoStruct *info);
@@ -33,6 +45,6 @@ int aacenc_init(HANDLE_AACENCODER *encoder, const aacenc_param_t *params,
 int aac_encode_frame(HANDLE_AACENCODER encoder,
                      const pcm_sample_description_t *format,
                      const int16_t *input, unsigned iframes,
-                     uint8_t **output, uint32_t *olen, uint32_t *osize);
+                     aacenc_frame_t *output);
 
 #endif

src/caf_reader.c

@@ -0,0 +1,252 @@
+/* 
+ * Copyright (C) 2013 nu774
+ * For conditions of distribution and use, see copyright notice in COPYING
+ */
+#if HAVE_CONFIG_H
+#  include "config.h"
+#endif
+
+#if HAVE_STDINT_H
+#  include <stdint.h>
+#endif
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdarg.h>
+#include "caf_reader.h"
+#include "m4af.h"
+
+typedef struct caf_reader_t {
+    pcm_reader_vtbl_t *vtbl;
+    pcm_sample_description_t sample_format;
+    int64_t length;
+    int64_t position;
+    int64_t data_offset;
+    pcm_io_context_t io;
+    aacenc_tag_callback_t tag_callback;
+    void *tag_ctx;
+    uint8_t chanmap[8];
+} caf_reader_t;
+
+static const pcm_sample_description_t *caf_get_format(pcm_reader_t *reader)
+{
+    return &((caf_reader_t *)reader)->sample_format;
+}
+
+static int64_t caf_get_length(pcm_reader_t *reader)
+{
+    return ((caf_reader_t *)reader)->length;
+}
+
+static int64_t caf_get_position(pcm_reader_t *reader)
+{
+    return ((caf_reader_t *)reader)->position;
+}
+
+static void caf_teardown(pcm_reader_t **reader)
+{
+    free(*reader);
+    *reader = 0;
+}
+
+static
+uint32_t caf_next_chunk(caf_reader_t *reader, int64_t *chunk_size)
+{
+    uint32_t fcc;
+    if (pcm_scanb(&reader->io, "LQ", &fcc, chunk_size) == 2)
+        return fcc;
+    return 0;
+}
+
+static
+int caf_desc(caf_reader_t *reader, int64_t chunk_size)
+{
+    double mSampleRate;
+    uint32_t mFormatID, mFormatFlags, mBytesPerPacket, mFramesPerPacket,
+             mChannelsPerFrame, mBitsPerChannel;
+    pcm_sample_description_t *desc = &reader->sample_format;
+
+    ENSURE(chunk_size >= 32);
+    TRY_IO(pcm_scanb(&reader->io, "QLLLLLL", &mSampleRate, &mFormatID,
+                     &mFormatFlags, &mBytesPerPacket, &mFramesPerPacket,
+                     &mChannelsPerFrame, &mBitsPerChannel) != 7);
+
+    ENSURE(mFormatID == M4AF_FOURCC('l','p','c','m'));
+    ENSURE(mSampleRate && mBytesPerPacket &&
+           mChannelsPerFrame >= 1 && mChannelsPerFrame <= 8 &&
+           mBitsPerChannel && mFramesPerPacket == 1 &&
+           mBytesPerPacket % mChannelsPerFrame == 0 &&
+           mBytesPerPacket >= mChannelsPerFrame * ((mBitsPerChannel + 7) / 8));
+
+    desc->sample_rate        = mSampleRate;
+    desc->bits_per_channel   = mBitsPerChannel;
+    desc->bytes_per_frame    = mBytesPerPacket;
+    desc->channels_per_frame = mChannelsPerFrame;
+
+    switch (mFormatFlags) {
+    case 0: desc->sample_type = PCM_TYPE_SINT_BE;  break;
+    case 1: desc->sample_type = PCM_TYPE_FLOAT_BE; break;
+    case 2: desc->sample_type = PCM_TYPE_SINT;     break;
+    case 3: desc->sample_type = PCM_TYPE_FLOAT;    break;
+    default: goto FAIL;
+    }
+
+    TRY_IO(pcm_skip(&reader->io, chunk_size - 32));
+    return 0;
+FAIL:
+    return -1;
+}
+
+static
+int caf_info(caf_reader_t *reader, int64_t chunk_size)
+{
+    char *buf, *key, *val, *end;
+    size_t len;
+
+    if (chunk_size < 4 || (buf = malloc(chunk_size)) == 0)
+        return -1;
+    pcm_read(&reader->io, buf, chunk_size);
+    key = buf + 4;
+    end = buf + chunk_size;
+    do {
+        if ((val = key + strlen(key) + 1) < end) {
+            len = strlen(val);
+            if (reader->tag_callback)
+                reader->tag_callback(reader->tag_ctx, key, val, len);
+            key = val + len + 1;
+        }
+    } while (key < end && val < end);
+
+    if (reader->tag_callback)
+        reader->tag_callback(reader->tag_ctx, 0, 0, 0);
+    free(buf);
+    return 0;
+}
+
+static
+int caf_read_frames(pcm_reader_t *preader, void *buffer, unsigned nframes)
+{
+    int rc;
+    unsigned i, j, nbytes;
+    caf_reader_t *reader = (caf_reader_t *)preader;
+    unsigned bpf = reader->sample_format.bytes_per_frame;
+    unsigned nchannels = reader->sample_format.channels_per_frame;
+    unsigned bpc = bpf / nchannels;
+    uint8_t tmp[64]; /* enough room for maximum bpf: 8ch float64 */
+    uint8_t *bp;
+    uint8_t *chanmap = reader->chanmap;
+
+    if (nframes > reader->length - reader->position)
+        nframes = reader->length - reader->position;
+    nbytes = nframes * bpf;
+    if (nbytes) {
+        if ((rc = pcm_read(&reader->io, buffer, nbytes)) < 0)
+            return -1;
+        nframes = rc / bpf;
+        for (bp = buffer, i = 0; i < nframes; ++i, bp += bpf) {
+            memcpy(tmp, bp, bpf);
+            for (j = 0; j < nchannels; ++j)
+                memcpy(bp + bpc * j, tmp + bpc * chanmap[j], bpc);
+        }
+        reader->position += nframes;
+    }
+    if (nframes == 0) {
+        /* fetch info after data chunk */
+        uint32_t fcc;
+        int64_t chunk_size;
+        while ((fcc = caf_next_chunk(reader, &chunk_size)) != 0) {
+            if (fcc == M4AF_FOURCC('i','n','f','o'))
+                TRY_IO(caf_info(reader, chunk_size));
+            else
+                TRY_IO(pcm_skip(&reader->io, chunk_size));
+        }
+    }
+    return nframes;
+FAIL:
+    return 0;
+}
+
+static
+int caf_parse(caf_reader_t *reader, int64_t *data_length)
+{
+    uint32_t fcc;
+    int64_t chunk_size;
+
+    *data_length = 0;
+
+    /* CAFFileHeader */
+    TRY_IO(pcm_read32be(&reader->io, &fcc));
+    ENSURE(fcc == M4AF_FOURCC('c','a','f','f'));
+    TRY_IO(pcm_skip(&reader->io, 4)); /* mFileVersion, mFileFlags */
+
+    while ((fcc = caf_next_chunk(reader, &chunk_size)) != 0) {
+        if (fcc == M4AF_FOURCC('d','e','s','c'))
+            TRY_IO(caf_desc(reader, chunk_size));
+        else if (fcc == M4AF_FOURCC('i','n','f','o'))
+            TRY_IO(caf_info(reader, chunk_size));
+        else if (fcc == M4AF_FOURCC('c','h','a','n')) {
+            ENSURE(reader->sample_format.channels_per_frame);
+            if (apple_chan_chunk(&reader->io, chunk_size,
+                                 &reader->sample_format, reader->chanmap) < 0)
+                goto FAIL;
+        } else if (fcc == M4AF_FOURCC('d','a','t','a')) {
+            TRY_IO(pcm_skip(&reader->io, 4)); /* mEditCount */
+            *data_length = (chunk_size == ~0ULL) ? chunk_size : chunk_size - 4;
+            reader->data_offset = pcm_tell(&reader->io);
+            break;
+        } else
+            TRY_IO(pcm_skip(&reader->io, chunk_size));
+    }
+    ENSURE(reader->sample_format.channels_per_frame);
+    ENSURE(fcc == M4AF_FOURCC('d','a','t','a'));
+    return 0;
+FAIL:
+    return -1;
+}
+
+static pcm_reader_vtbl_t caf_vtable = {
+    caf_get_format,
+    caf_get_length,
+    caf_get_position,
+    caf_read_frames,
+    caf_teardown
+};
+
+pcm_reader_t *caf_open(pcm_io_context_t *io,
+                       aacenc_tag_callback_t tag_callback, void *tag_ctx)
+{
+    caf_reader_t *reader = 0;
+    int64_t data_length;
+    unsigned bpf;
+
+    if ((reader = calloc(1, sizeof(caf_reader_t))) == 0)
+        return 0;
+    memcpy(&reader->io, io, sizeof(pcm_io_context_t));
+    reader->tag_callback = tag_callback;
+    reader->tag_ctx = tag_ctx;
+    memcpy(reader->chanmap, "\000\001\002\003\004\005\006\007", 8);
+
+    if (caf_parse(reader, &data_length) < 0) {
+        free(reader);
+        return 0;
+    }
+    bpf = reader->sample_format.bytes_per_frame;
+
+    /* CAF uses -1 to indicate "unknown size" */
+    if (data_length < 0 || data_length % bpf)
+        reader->length = INT64_MAX;
+    else
+        reader->length = data_length / bpf;
+
+    if (reader->length == INT64_MAX) {
+        if (pcm_seek(&reader->io, 0, SEEK_END) >= 0) {
+            int64_t size = pcm_tell(&reader->io);
+            if (size > 0)
+                reader->length = (size - reader->data_offset) / bpf;
+            pcm_seek(&reader->io, reader->data_offset, SEEK_SET);
+        }
+    }
+    reader->vtbl = &caf_vtable;
+    return (pcm_reader_t *)reader;
+}

src/caf_reader.h

@@ -0,0 +1,15 @@
+/* 
+ * Copyright (C) 2013 nu774
+ * For conditions of distribution and use, see copyright notice in COPYING
+ */
+#ifndef CAF_READER_H
+#define CAF_READER_H
+
+#include "lpcm.h"
+#include "pcm_reader.h"
+#include "metadata.h"
+
+pcm_reader_t *caf_open(pcm_io_context_t *io,
+                       aacenc_tag_callback_t tag_callback, void *tag_ctx);
+
+#endif

src/catypes.h

@@ -0,0 +1,241 @@
+#if !defined(__CoreAudioTypes_h__)
+#define __CoreAudioTypes_h__
+
+enum { kVariableLengthArray = 1 };
+
+typedef uint32_t AudioChannelLabel;
+typedef uint32_t AudioChannelLayoutTag;
+
+struct AudioChannelDescription
+{
+    AudioChannelLabel mChannelLabel;
+    uint32_t          mChannelFlags;
+    float             mCoordinates[3];
+};
+typedef struct AudioChannelDescription AudioChannelDescription;
+
+struct AudioChannelLayout
+{
+    AudioChannelLayoutTag       mChannelLayoutTag;
+    uint32_t                    mChannelBitmap;
+    uint32_t                    mNumberChannelDescriptions;
+    AudioChannelDescription     mChannelDescriptions[kVariableLengthArray];
+};
+typedef struct AudioChannelLayout AudioChannelLayout;
+
+enum
+{
+    kAudioChannelLabel_Unknown                  = 0xFFFFFFFF,   // unknown or unspecified other use
+    kAudioChannelLabel_Unused                   = 0,            // channel is present, but has no intended use or destination
+    kAudioChannelLabel_UseCoordinates           = 100,          // channel is described by the mCoordinates fields.
+
+    kAudioChannelLabel_Left                     = 1,
+    kAudioChannelLabel_Right                    = 2,
+    kAudioChannelLabel_Center                   = 3,
+    kAudioChannelLabel_LFEScreen                = 4,
+    kAudioChannelLabel_LeftSurround             = 5,            // WAVE: "Back Left"
+    kAudioChannelLabel_RightSurround            = 6,            // WAVE: "Back Right"
+    kAudioChannelLabel_LeftCenter               = 7,
+    kAudioChannelLabel_RightCenter              = 8,
+    kAudioChannelLabel_CenterSurround           = 9,            // WAVE: "Back Center" or plain "Rear Surround"
+    kAudioChannelLabel_LeftSurroundDirect       = 10,           // WAVE: "Side Left"
+    kAudioChannelLabel_RightSurroundDirect      = 11,           // WAVE: "Side Right"
+    kAudioChannelLabel_TopCenterSurround        = 12,
+    kAudioChannelLabel_VerticalHeightLeft       = 13,           // WAVE: "Top Front Left"
+    kAudioChannelLabel_VerticalHeightCenter     = 14,           // WAVE: "Top Front Center"
+    kAudioChannelLabel_VerticalHeightRight      = 15,           // WAVE: "Top Front Right"
+
+    kAudioChannelLabel_TopBackLeft              = 16,
+    kAudioChannelLabel_TopBackCenter            = 17,
+    kAudioChannelLabel_TopBackRight             = 18,
+
+    kAudioChannelLabel_RearSurroundLeft         = 33,
+    kAudioChannelLabel_RearSurroundRight        = 34,
+    kAudioChannelLabel_LeftWide                 = 35,
+    kAudioChannelLabel_RightWide                = 36,
+    kAudioChannelLabel_LFE2                     = 37,
+    kAudioChannelLabel_LeftTotal                = 38,           // matrix encoded 4 channels
+    kAudioChannelLabel_RightTotal               = 39,           // matrix encoded 4 channels
+    kAudioChannelLabel_HearingImpaired          = 40,
+    kAudioChannelLabel_Narration                = 41,
+    kAudioChannelLabel_Mono                     = 42,
+    kAudioChannelLabel_DialogCentricMix         = 43,
+
+    kAudioChannelLabel_CenterSurroundDirect     = 44,           // back center, non diffuse
+    
+    kAudioChannelLabel_Haptic                   = 45,
+
+    // first order ambisonic channels
+    kAudioChannelLabel_Ambisonic_W              = 200,
+    kAudioChannelLabel_Ambisonic_X              = 201,
+    kAudioChannelLabel_Ambisonic_Y              = 202,
+    kAudioChannelLabel_Ambisonic_Z              = 203,
+
+    // Mid/Side Recording
+    kAudioChannelLabel_MS_Mid                   = 204,
+    kAudioChannelLabel_MS_Side                  = 205,
+
+    // X-Y Recording
+    kAudioChannelLabel_XY_X                     = 206,
+    kAudioChannelLabel_XY_Y                     = 207,
+
+    // other
+    kAudioChannelLabel_HeadphonesLeft           = 301,
+    kAudioChannelLabel_HeadphonesRight          = 302,
+    kAudioChannelLabel_ClickTrack               = 304,
+    kAudioChannelLabel_ForeignLanguage          = 305,
+
+    // generic discrete channel
+    kAudioChannelLabel_Discrete                 = 400,
+
+    // numbered discrete channel
+    kAudioChannelLabel_Discrete_0               = (1L<<16) | 0,
+    kAudioChannelLabel_Discrete_1               = (1L<<16) | 1,
+    kAudioChannelLabel_Discrete_2               = (1L<<16) | 2,
+    kAudioChannelLabel_Discrete_3               = (1L<<16) | 3,
+    kAudioChannelLabel_Discrete_4               = (1L<<16) | 4,
+    kAudioChannelLabel_Discrete_5               = (1L<<16) | 5,
+    kAudioChannelLabel_Discrete_6               = (1L<<16) | 6,
+    kAudioChannelLabel_Discrete_7               = (1L<<16) | 7,
+    kAudioChannelLabel_Discrete_8               = (1L<<16) | 8,
+    kAudioChannelLabel_Discrete_9               = (1L<<16) | 9,
+    kAudioChannelLabel_Discrete_10              = (1L<<16) | 10,
+    kAudioChannelLabel_Discrete_11              = (1L<<16) | 11,
+    kAudioChannelLabel_Discrete_12              = (1L<<16) | 12,
+    kAudioChannelLabel_Discrete_13              = (1L<<16) | 13,
+    kAudioChannelLabel_Discrete_14              = (1L<<16) | 14,
+    kAudioChannelLabel_Discrete_15              = (1L<<16) | 15,
+    kAudioChannelLabel_Discrete_65535           = (1L<<16) | 65535
+};
+
+#define AudioChannelLayoutTag_GetNumberOfChannels(layoutTag) \
+    ((uint32_t)((layoutTag) & 0x0000FFFF))
+
+enum
+{
+    kAudioChannelLayoutTag_UseChannelDescriptions   = (0L<<16) | 0,     // use the array of AudioChannelDescriptions to define the mapping.
+    kAudioChannelLayoutTag_UseChannelBitmap         = (1L<<16) | 0,     // use the bitmap to define the mapping.
+
+    kAudioChannelLayoutTag_Mono                     = (100L<<16) | 1,   // a standard mono stream
+    kAudioChannelLayoutTag_Stereo                   = (101L<<16) | 2,   // a standard stereo stream (L R) - implied playback
+    kAudioChannelLayoutTag_StereoHeadphones         = (102L<<16) | 2,   // a standard stereo stream (L R) - implied headphone playbac
+    kAudioChannelLayoutTag_MatrixStereo             = (103L<<16) | 2,   // a matrix encoded stereo stream (Lt, Rt)
+    kAudioChannelLayoutTag_MidSide                  = (104L<<16) | 2,   // mid/side recording
+    kAudioChannelLayoutTag_XY                       = (105L<<16) | 2,   // coincident mic pair (often 2 figure 8's)
+    kAudioChannelLayoutTag_Binaural                 = (106L<<16) | 2,   // binaural stereo (left, right)
+    kAudioChannelLayoutTag_Ambisonic_B_Format       = (107L<<16) | 4,   // W, X, Y, Z
+
+    kAudioChannelLayoutTag_Quadraphonic             = (108L<<16) | 4,   // front left, front right, back left, back right
+
+    kAudioChannelLayoutTag_Pentagonal               = (109L<<16) | 5,   // left, right, rear left, rear right, center
+
+    kAudioChannelLayoutTag_Hexagonal                = (110L<<16) | 6,   // left, right, rear left, rear right, center, rear
+
+    kAudioChannelLayoutTag_Octagonal                = (111L<<16) | 8,   // front left, front right, rear left, rear right,
+                                                                        // front center, rear center, side left, side right
+
+    kAudioChannelLayoutTag_Cube                     = (112L<<16) | 8,   // left, right, rear left, rear right
+                                                                        // top left, top right, top rear left, top rear right
+
+    //  MPEG defined layouts
+    kAudioChannelLayoutTag_MPEG_1_0                 = kAudioChannelLayoutTag_Mono,          //  C
+    kAudioChannelLayoutTag_MPEG_2_0                 = kAudioChannelLayoutTag_Stereo,        //  L R
+    kAudioChannelLayoutTag_MPEG_3_0_A               = (113L<<16) | 3,                       //  L R C
+    kAudioChannelLayoutTag_MPEG_3_0_B               = (114L<<16) | 3,                       //  C L R
+    kAudioChannelLayoutTag_MPEG_4_0_A               = (115L<<16) | 4,                       //  L R C Cs
+    kAudioChannelLayoutTag_MPEG_4_0_B               = (116L<<16) | 4,                       //  C L R Cs
+    kAudioChannelLayoutTag_MPEG_5_0_A               = (117L<<16) | 5,                       //  L R C Ls Rs
+    kAudioChannelLayoutTag_MPEG_5_0_B               = (118L<<16) | 5,                       //  L R Ls Rs C
+    kAudioChannelLayoutTag_MPEG_5_0_C               = (119L<<16) | 5,                       //  L C R Ls Rs
+    kAudioChannelLayoutTag_MPEG_5_0_D               = (120L<<16) | 5,                       //  C L R Ls Rs
+    kAudioChannelLayoutTag_MPEG_5_1_A               = (121L<<16) | 6,                       //  L R C LFE Ls Rs
+    kAudioChannelLayoutTag_MPEG_5_1_B               = (122L<<16) | 6,                       //  L R Ls Rs C LFE
+    kAudioChannelLayoutTag_MPEG_5_1_C               = (123L<<16) | 6,                       //  L C R Ls Rs LFE
+    kAudioChannelLayoutTag_MPEG_5_1_D               = (124L<<16) | 6,                       //  C L R Ls Rs LFE
+    kAudioChannelLayoutTag_MPEG_6_1_A               = (125L<<16) | 7,                       //  L R C LFE Ls Rs Cs
+    kAudioChannelLayoutTag_MPEG_7_1_A               = (126L<<16) | 8,                       //  L R C LFE Ls Rs Lc Rc
+    kAudioChannelLayoutTag_MPEG_7_1_B               = (127L<<16) | 8,                       //  C Lc Rc L R Ls Rs LFE    (doc: IS-13818-7 MPEG2-AAC Table 3.1)
+    kAudioChannelLayoutTag_MPEG_7_1_C               = (128L<<16) | 8,                       //  L R C LFE Ls Rs Rls Rrs
+    kAudioChannelLayoutTag_Emagic_Default_7_1       = (129L<<16) | 8,                       //  L R Ls Rs C LFE Lc Rc
+    kAudioChannelLayoutTag_SMPTE_DTV                = (130L<<16) | 8,                       //  L R C LFE Ls Rs Lt Rt
+                                                                                            //      (kAudioChannelLayoutTag_ITU_5_1 plus a matrix encoded stereo mix)
+
+    //  ITU defined layouts
+    kAudioChannelLayoutTag_ITU_1_0                  = kAudioChannelLayoutTag_Mono,          //  C
+    kAudioChannelLayoutTag_ITU_2_0                  = kAudioChannelLayoutTag_Stereo,        //  L R
+
+    kAudioChannelLayoutTag_ITU_2_1                  = (131L<<16) | 3,                       //  L R Cs
+    kAudioChannelLayoutTag_ITU_2_2                  = (132L<<16) | 4,                       //  L R Ls Rs
+    kAudioChannelLayoutTag_ITU_3_0                  = kAudioChannelLayoutTag_MPEG_3_0_A,    //  L R C
+    kAudioChannelLayoutTag_ITU_3_1                  = kAudioChannelLayoutTag_MPEG_4_0_A,    //  L R C Cs
+
+    kAudioChannelLayoutTag_ITU_3_2                  = kAudioChannelLayoutTag_MPEG_5_0_A,    //  L R C Ls Rs
+    kAudioChannelLayoutTag_ITU_3_2_1                = kAudioChannelLayoutTag_MPEG_5_1_A,    //  L R C LFE Ls Rs
+    kAudioChannelLayoutTag_ITU_3_4_1                = kAudioChannelLayoutTag_MPEG_7_1_C,    //  L R C LFE Ls Rs Rls Rrs
+
+    // DVD defined layouts
+    kAudioChannelLayoutTag_DVD_0                    = kAudioChannelLayoutTag_Mono,          // C (mono)
+    kAudioChannelLayoutTag_DVD_1                    = kAudioChannelLayoutTag_Stereo,        // L R
+    kAudioChannelLayoutTag_DVD_2                    = kAudioChannelLayoutTag_ITU_2_1,       // L R Cs
+    kAudioChannelLayoutTag_DVD_3                    = kAudioChannelLayoutTag_ITU_2_2,       // L R Ls Rs
+    kAudioChannelLayoutTag_DVD_4                    = (133L<<16) | 3,                       // L R LFE
+    kAudioChannelLayoutTag_DVD_5                    = (134L<<16) | 4,                       // L R LFE Cs
+    kAudioChannelLayoutTag_DVD_6                    = (135L<<16) | 5,                       // L R LFE Ls Rs
+    kAudioChannelLayoutTag_DVD_7                    = kAudioChannelLayoutTag_MPEG_3_0_A,    // L R C
+    kAudioChannelLayoutTag_DVD_8                    = kAudioChannelLayoutTag_MPEG_4_0_A,    // L R C Cs
+    kAudioChannelLayoutTag_DVD_9                    = kAudioChannelLayoutTag_MPEG_5_0_A,    // L R C Ls Rs
+    kAudioChannelLayoutTag_DVD_10                   = (136L<<16) | 4,                       // L R C LFE
+    kAudioChannelLayoutTag_DVD_11                   = (137L<<16) | 5,                       // L R C LFE Cs
+    kAudioChannelLayoutTag_DVD_12                   = kAudioChannelLayoutTag_MPEG_5_1_A,    // L R C LFE Ls Rs
+    // 13 through 17 are duplicates of 8 through 12.
+    kAudioChannelLayoutTag_DVD_13                   = kAudioChannelLayoutTag_DVD_8,         // L R C Cs
+    kAudioChannelLayoutTag_DVD_14                   = kAudioChannelLayoutTag_DVD_9,         // L R C Ls Rs
+    kAudioChannelLayoutTag_DVD_15                   = kAudioChannelLayoutTag_DVD_10,        // L R C LFE
+    kAudioChannelLayoutTag_DVD_16                   = kAudioChannelLayoutTag_DVD_11,        // L R C LFE Cs
+    kAudioChannelLayoutTag_DVD_17                   = kAudioChannelLayoutTag_DVD_12,        // L R C LFE Ls Rs
+    kAudioChannelLayoutTag_DVD_18                   = (138L<<16) | 5,                       // L R Ls Rs LFE
+    kAudioChannelLayoutTag_DVD_19                   = kAudioChannelLayoutTag_MPEG_5_0_B,    // L R Ls Rs C
+    kAudioChannelLayoutTag_DVD_20                   = kAudioChannelLayoutTag_MPEG_5_1_B,    // L R Ls Rs C LFE
+
+    // These layouts are recommended for AudioUnit usage
+        // These are the symmetrical layouts
+    kAudioChannelLayoutTag_AudioUnit_4              = kAudioChannelLayoutTag_Quadraphonic,
+    kAudioChannelLayoutTag_AudioUnit_5              = kAudioChannelLayoutTag_Pentagonal,
+    kAudioChannelLayoutTag_AudioUnit_6              = kAudioChannelLayoutTag_Hexagonal,
+    kAudioChannelLayoutTag_AudioUnit_8              = kAudioChannelLayoutTag_Octagonal,
+        // These are the surround-based layouts
+    kAudioChannelLayoutTag_AudioUnit_5_0            = kAudioChannelLayoutTag_MPEG_5_0_B,    // L R Ls Rs C
+    kAudioChannelLayoutTag_AudioUnit_6_0            = (139L<<16) | 6,                       // L R Ls Rs C Cs
+    kAudioChannelLayoutTag_AudioUnit_7_0            = (140L<<16) | 7,                       // L R Ls Rs C Rls Rrs
+    kAudioChannelLayoutTag_AudioUnit_7_0_Front      = (148L<<16) | 7,                       // L R Ls Rs C Lc Rc
+    kAudioChannelLayoutTag_AudioUnit_5_1            = kAudioChannelLayoutTag_MPEG_5_1_A,    // L R C LFE Ls Rs
+    kAudioChannelLayoutTag_AudioUnit_6_1            = kAudioChannelLayoutTag_MPEG_6_1_A,    // L R C LFE Ls Rs Cs
+    kAudioChannelLayoutTag_AudioUnit_7_1            = kAudioChannelLayoutTag_MPEG_7_1_C,    // L R C LFE Ls Rs Rls Rrs
+    kAudioChannelLayoutTag_AudioUnit_7_1_Front      = kAudioChannelLayoutTag_MPEG_7_1_A,    // L R C LFE Ls Rs Lc Rc
+
+    kAudioChannelLayoutTag_AAC_3_0                  = kAudioChannelLayoutTag_MPEG_3_0_B,    // C L R
+    kAudioChannelLayoutTag_AAC_Quadraphonic         = kAudioChannelLayoutTag_Quadraphonic,  // L R Ls Rs
+    kAudioChannelLayoutTag_AAC_4_0                  = kAudioChannelLayoutTag_MPEG_4_0_B,    // C L R Cs
+    kAudioChannelLayoutTag_AAC_5_0                  = kAudioChannelLayoutTag_MPEG_5_0_D,    // C L R Ls Rs
+    kAudioChannelLayoutTag_AAC_5_1                  = kAudioChannelLayoutTag_MPEG_5_1_D,    // C L R Ls Rs Lfe
+    kAudioChannelLayoutTag_AAC_6_0                  = (141L<<16) | 6,                       // C L R Ls Rs Cs
+    kAudioChannelLayoutTag_AAC_6_1                  = (142L<<16) | 7,                       // C L R Ls Rs Cs Lfe
+    kAudioChannelLayoutTag_AAC_7_0                  = (143L<<16) | 7,                       // C L R Ls Rs Rls Rrs
+    kAudioChannelLayoutTag_AAC_7_1                  = kAudioChannelLayoutTag_MPEG_7_1_B,    // C Lc Rc L R Ls Rs Lfe
+    kAudioChannelLayoutTag_AAC_Octagonal            = (144L<<16) | 8,                       // C L R Ls Rs Rls Rrs Cs
+
+    kAudioChannelLayoutTag_TMH_10_2_std             = (145L<<16) | 16,                      // L R C Vhc Lsd Rsd Ls Rs Vhl Vhr Lw Rw Csd Cs LFE1 LFE2
+    kAudioChannelLayoutTag_TMH_10_2_full            = (146L<<16) | 21,                      // TMH_10_2_std plus: Lc Rc HI VI Haptic
+
+    kAudioChannelLayoutTag_AC3_1_0_1                = (149L<<16) | 2,                       // C LFE
+    kAudioChannelLayoutTag_AC3_3_0                  = (150L<<16) | 3,                       // L C R
+    kAudioChannelLayoutTag_AC3_3_1                  = (151L<<16) | 4,                       // L C R Cs
+    kAudioChannelLayoutTag_AC3_3_0_1                = (152L<<16) | 4,                       // L C R LFE
+    kAudioChannelLayoutTag_AC3_2_1_1                = (153L<<16) | 4,                       // L R Cs LFE
+    kAudioChannelLayoutTag_AC3_3_1_1                = (154L<<16) | 5,                       // L C R Cs LFE
+
+    kAudioChannelLayoutTag_DiscreteInOrder          = (147L<<16) | 0,                       // needs to be ORed with the actual number of channels  
+    kAudioChannelLayoutTag_Unknown                  = 0xFFFF0000                            // needs to be ORed with the actual number of channels  
+};
+
+#endif

src/extrapolater.c

@@ -0,0 +1,208 @@
+/* 
+ * Copyright (C) 2013 nu774
+ * For conditions of distribution and use, see copyright notice in COPYING
+ */
+#if HAVE_CONFIG_H
+#  include "config.h"
+#endif
+#if HAVE_STDINT_H
+#  include <stdint.h>
+#endif
+#include <stdlib.h>
+#include <string.h>
+#include <assert.h>
+#include "pcm_reader.h"
+#include "lpc.h"
+
+typedef int16_t sample_t;
+
+typedef struct buffer_t {
+    sample_t *data;
+    unsigned count;    /* count in frames */
+    unsigned capacity; /* size in bytes   */
+} buffer_t;
+
+typedef struct extrapolater_t {
+    pcm_reader_vtbl_t *vtbl;
+    pcm_reader_t *src;
+    pcm_sample_description_t format;
+    buffer_t buffer[2];
+    unsigned nbuffer;
+    int (*process)(struct extrapolater_t *, void *, unsigned);
+} extrapolater_t;
+
+#define LPC_ORDER 32
+
+static inline pcm_reader_t *get_source(pcm_reader_t *reader)
+{
+    return ((extrapolater_t *)reader)->src;
+}
+
+static const
+pcm_sample_description_t *get_format(pcm_reader_t *reader)
+{
+    return pcm_get_format(get_source(reader));
+}
+
+static int64_t get_length(pcm_reader_t *reader)
+{
+    return pcm_get_length(get_source(reader));
+}
+
+static int64_t get_position(pcm_reader_t *reader)
+{
+    return pcm_get_position(get_source(reader));
+}
+
+static int realloc_buffer(buffer_t *bp, size_t size)
+{
+    if (bp->capacity < size) {
+        void *p = realloc(bp->data, size);
+        if (!p) return -1;
+        bp->data = p;
+        bp->capacity = size;
+    }
+    return 0;
+}
+
+static void reverse_buffer(sample_t *data, unsigned nframes, unsigned nchannels)
+{
+    unsigned i = 0, j = nchannels * (nframes - 1), n;
+
+    for (; i < j; i += nchannels, j -= nchannels) {
+        for (n = 0; n < nchannels; ++n) {
+            sample_t tmp = data[i + n];
+            data[i + n] = data[j + n];
+            data[j + n] = tmp;
+        }
+    }
+}
+
+static int fetch(extrapolater_t *self, unsigned nframes)
+{
+    const pcm_sample_description_t *sfmt = pcm_get_format(self->src);
+    buffer_t *bp = &self->buffer[self->nbuffer];
+    int rc = 0;
+
+    if (realloc_buffer(bp, nframes * sfmt->bytes_per_frame) == 0) {
+        rc = pcm_read_frames(self->src, bp->data, nframes);
+        bp->count = rc > 0 ? rc : 0;
+    }
+    if (rc > 0)
+        self->nbuffer ^= 1;
+    return bp->count;
+}
+
+static int extrapolate(extrapolater_t *self, const buffer_t *bp,
+                        void *dst, unsigned nframes)
+{
+    const pcm_sample_description_t *sfmt = pcm_get_format(self->src);
+    unsigned i, n = sfmt->channels_per_frame;
+    float lpc[LPC_ORDER];
+
+    for (i = 0; i < n; ++i) {
+        vorbis_lpc_from_data(bp->data + i, lpc, bp->count, LPC_ORDER, n);
+        vorbis_lpc_predict(lpc, &bp->data[i + n * (bp->count - LPC_ORDER)],
+                           LPC_ORDER, (sample_t*)dst + i, nframes, n);
+    }
+    return nframes;
+}
+
+static int process1(extrapolater_t *self, void *buffer, unsigned nframes);
+static int process2(extrapolater_t *self, void *buffer, unsigned nframes);
+static int process3(extrapolater_t *self, void *buffer, unsigned nframes);
+
+static int process0(extrapolater_t *self, void *buffer, unsigned nframes)
+{
+    const pcm_sample_description_t *sfmt = pcm_get_format(self->src);
+    unsigned nchannels = sfmt->channels_per_frame;
+    buffer_t *bp = &self->buffer[self->nbuffer];
+
+    if (fetch(self, nframes) < 2 * LPC_ORDER)
+        memset(buffer, 0, nframes * sfmt->bytes_per_frame);
+    else {
+        reverse_buffer(bp->data, bp->count, nchannels);
+        extrapolate(self, bp, buffer, nframes);
+        reverse_buffer(buffer, nframes, nchannels);
+        reverse_buffer(bp->data, bp->count, nchannels);
+    }
+    self->process = bp->count ? process1 : process2;
+    return nframes;
+}
+
+static int process1(extrapolater_t *self, void *buffer, unsigned nframes)
+{
+    const pcm_sample_description_t *sfmt = pcm_get_format(self->src);
+    buffer_t *bp = &self->buffer[self->nbuffer ^ 1];
+
+    assert(bp->count <= nframes);
+    memcpy(buffer, bp->data, bp->count * sfmt->bytes_per_frame); 
+    if (!fetch(self, nframes))
+        self->process = process2;
+    return bp->count;
+}
+
+static int process2(extrapolater_t *self, void *buffer, unsigned nframes)
+{
+    const pcm_sample_description_t *sfmt = pcm_get_format(self->src);
+    buffer_t *bp = &self->buffer[self->nbuffer];
+    buffer_t *bbp = &self->buffer[self->nbuffer ^ 1];
+
+    if (bp->count < 2 * LPC_ORDER) {
+        size_t total = bp->count + bbp->count;
+        if (bbp->count &&
+            realloc_buffer(bbp, total * sfmt->bytes_per_frame) == 0)
+        {
+            memcpy(bbp->data + bbp->count * sfmt->channels_per_frame,
+                   bp->data, bp->count * sfmt->bytes_per_frame);
+            bbp->count = total;
+            bp->count = 0;
+            bp = bbp;
+            self->nbuffer ^= 1;
+        }
+    }
+    self->process = process3;
+
+    if (bp->count >= 2 * LPC_ORDER)
+        extrapolate(self, bp, buffer, nframes);
+    else
+        memset(buffer, 0, nframes * sfmt->bytes_per_frame);
+    return nframes;
+}
+
+static int process3(extrapolater_t *self, void *buffer, unsigned nframes)
+{
+    return 0;
+}
+
+static int read_frames(pcm_reader_t *reader, void *buffer, unsigned nframes)
+{
+    extrapolater_t *self = (extrapolater_t *)reader;
+    return self->process(self, buffer, nframes);
+}
+
+static void teardown(pcm_reader_t **reader)
+{
+    extrapolater_t *self = (extrapolater_t *)*reader;
+    pcm_teardown(&self->src);
+    free(self->buffer[0].data);
+    free(self->buffer[1].data);
+    free(self);
+    *reader = 0;
+}
+
+static pcm_reader_vtbl_t my_vtable = {
+    get_format, get_length, get_position, read_frames, teardown
+};
+
+pcm_reader_t *extrapolater_open(pcm_reader_t *reader)
+{
+    extrapolater_t *self = 0;
+
+    if ((self = calloc(1, sizeof(extrapolater_t))) == 0)
+        return 0;
+    self->src = reader;
+    self->vtbl = &my_vtable;
+    self->process = process0;
+    return (pcm_reader_t *)self;
+}

src/lpc.c

@@ -0,0 +1,169 @@
+/********************************************************************
+ *                                                                  *
+ * THIS FILE IS PART OF THE OggVorbis SOFTWARE CODEC SOURCE CODE.   *
+ * USE, DISTRIBUTION AND REPRODUCTION OF THIS LIBRARY SOURCE IS     *
+ * GOVERNED BY A BSD-STYLE SOURCE LICENSE INCLUDED WITH THIS SOURCE *
+ * IN 'COPYING'. PLEASE READ THESE TERMS BEFORE DISTRIBUTING.       *
+ *                                                                  *
+ * THE OggVorbis SOURCE CODE IS (C) COPYRIGHT 1994-2009             *
+ * by the Xiph.Org Foundation http://www.xiph.org/                  *
+ *                                                                  *
+ ********************************************************************
+
+  function: LPC low level routines
+  last mod: $Id: lpc.c 16227 2009-07-08 06:58:46Z xiphmont $
+
+ ********************************************************************/
+
+/* Some of these routines (autocorrelator, LPC coefficient estimator)
+   are derived from code written by Jutta Degener and Carsten Bormann;
+   thus we include their copyright below.  The entirety of this file
+   is freely redistributable on the condition that both of these
+   copyright notices are preserved without modification.  */
+
+/* Preserved Copyright: *********************************************/
+
+/* Copyright 1992, 1993, 1994 by Jutta Degener and Carsten Bormann,
+Technische Universita"t Berlin
+
+Any use of this software is permitted provided that this notice is not
+removed and that neither the authors nor the Technische Universita"t
+Berlin are deemed to have made any representations as to the
+suitability of this software for any purpose nor are held responsible
+for any defects of this software. THERE IS ABSOLUTELY NO WARRANTY FOR
+THIS SOFTWARE.
+
+As a matter of courtesy, the authors request to be informed about uses
+this software has found, about bugs in this software, and about any
+improvements that may be of general interest.
+
+Berlin, 28.11.1994
+Jutta Degener
+Carsten Bormann
+
+*********************************************************************/
+
+#ifdef HAVE_CONFIG_H
+# include "config.h"
+#endif
+
+#if HAVE_STDINT_H
+# include <stdint.h>
+#endif
+
+#include <stdlib.h>
+#include <string.h>
+#include <math.h>
+#include "lpc.h"
+#include "lpcm.h"
+
+/* Autocorrelation LPC coeff generation algorithm invented by
+   N. Levinson in 1947, modified by J. Durbin in 1959. */
+
+/* Input : n elements of time doamin data
+   Output: m lpc coefficients, excitation energy */
+
+float vorbis_lpc_from_data(short *data,float *lpci,int n,int m,int stride){
+  double *aut=malloc(sizeof(*aut)*(m+1));
+  double *lpc=malloc(sizeof(*lpc)*(m));
+  double error;
+  double epsilon;
+  int i,j;
+
+  /* autocorrelation, p+1 lag coefficients */
+  j=m+1;
+  while(j--){
+    double d=0; /* double needed for accumulator depth */
+    for(i=j;i<n;i++)d+=(double)data[i*stride]*data[(i-j)*stride]/1073741824.0;
+    aut[j]=d;
+  }
+
+  /* Generate lpc coefficients from autocorr values */
+
+  /* set our noise floor to about -100dB */
+  error=aut[0] * (1. + 1e-10);
+  epsilon=1e-9*aut[0]+1e-10;
+
+  for(i=0;i<m;i++){
+    double r= -aut[i+1];
+
+    if(error<epsilon){
+      memset(lpc+i,0,(m-i)*sizeof(*lpc));
+      goto done;
+    }
+
+    /* Sum up this iteration's reflection coefficient; note that in
+       Vorbis we don't save it.  If anyone wants to recycle this code
+       and needs reflection coefficients, save the results of 'r' from
+       each iteration. */
+
+    for(j=0;j<i;j++)r-=lpc[j]*aut[i-j];
+    r/=error;
+
+    /* Update LPC coefficients and total error */
+
+    lpc[i]=r;
+    for(j=0;j<i/2;j++){
+      double tmp=lpc[j];
+
+      lpc[j]+=r*lpc[i-1-j];
+      lpc[i-1-j]+=r*tmp;
+    }
+    if(i&1)lpc[j]+=lpc[j]*r;
+
+    error*=1.-r*r;
+
+  }
+
+ done:
+
+  /* slightly damp the filter */
+  {
+    double g = .99;
+    double damp = g;
+    for(j=0;j<m;j++){
+      lpc[j]*=damp;
+      damp*=g;
+    }
+  }
+
+  for(j=0;j<m;j++)lpci[j]=(float)lpc[j];
+
+  /* we need the error value to know how big an impulse to hit the
+     filter with later */
+
+  free(aut);
+  free(lpc);
+  return error;
+}
+
+void vorbis_lpc_predict(float *coeff,short *prime,int m,
+                        short *data,long n,int stride){
+
+  /* in: coeff[0...m-1] LPC coefficients
+         prime[0...m-1] initial values (allocated size of n+m-1)
+    out: data[0...n-1] data samples */
+
+  long i,j,o,p;
+  float y;
+  float *work=malloc(sizeof(*work)*(m+n));
+
+  if(!prime)
+    for(i=0;i<m;i++)
+      work[i]=0.f;
+  else
+    for(i=0;i<m;i++)
+      work[i]=prime[i*stride]/32768.0f;
+
+  for(i=0;i<n;i++){
+    y=0;
+    o=i;
+    p=m;
+    for(j=0;j<m;j++)
+      y-=work[o++]*coeff[--p];
+
+    work[o]=y;
+    data[i*stride]=lrint(pcm_clip(y*32768.0,-32768.0,32767.0));
+  }
+  free(work);
+}

src/lpc.h

@@ -0,0 +1,27 @@
+/********************************************************************
+ *                                                                  *
+ * THIS FILE IS PART OF THE OggVorbis SOFTWARE CODEC SOURCE CODE.   *
+ * USE, DISTRIBUTION AND REPRODUCTION OF THIS LIBRARY SOURCE IS     *
+ * GOVERNED BY A BSD-STYLE SOURCE LICENSE INCLUDED WITH THIS SOURCE *
+ * IN 'COPYING'. PLEASE READ THESE TERMS BEFORE DISTRIBUTING.       *
+ *                                                                  *
+ * THE OggVorbis SOURCE CODE IS (C) COPYRIGHT 1994-2007             *
+ * by the Xiph.Org Foundation http://www.xiph.org/                  *
+ *                                                                  *
+ ********************************************************************
+
+  function: LPC low level routines
+  last mod: $Id: lpc.h 16037 2009-05-26 21:10:58Z xiphmont $
+
+ ********************************************************************/
+
+#ifndef _V_LPC_H_
+#define _V_LPC_H_
+
+/* simple linear scale LPC code */
+extern float vorbis_lpc_from_data(short *data,float *lpc,int n,int m,int stride);
+
+extern void vorbis_lpc_predict(float *coeff,short *prime,int m,
+                               short *data,long n,int stride);
+
+#endif

src/lpcm.c

@@ -13,36 +13,6 @@
 #include "lpcm.h"
 #include "m4af_endian.h"
 
-#ifdef _MSC_VER
-#  define inline __inline
-#  ifdef _M_IX86
-inline int lrint(double x)
-{
-    int n;
-    _asm {
-        fld x
-        fistp n
-    }
-    return n;
-}
-#  else
-#    include <emmintrin.h>
-inline int lrint(double x)
-{
-    return _mm_cvtsd_si32(_mm_load_sd(&x));
-}
-#  endif
-#endif
-
-static
-inline double pcm_clip(double n, double min_value, double max_value)
-{
-    if (n < min_value)
-        return min_value;
-    else if (n > max_value)
-        return max_value;
-    return n;
-}
 static
 inline float pcm_i2f(int32_t n)
 {
@@ -187,14 +157,14 @@ inline int16_t pcm_f64be_to_s16(int64_t n)
 
 int pcm_convert_to_native_sint16(const pcm_sample_description_t *format,
                                  const void *input, uint32_t nframes,
-                                 int16_t **result, uint32_t *osize)
+                                 int16_t *result)
 {
 #define CONVERT(type, conv) \
     do { \
         unsigned i; \
         type *ip = (type *)input; \
         for (i = 0; i < count; ++i) { \
-            (*result)[i] = conv(ip[i]); \
+            result[i] = conv(ip[i]); \
         } \
     } while(0)
 
@@ -204,7 +174,7 @@ int pcm_convert_to_native_sint16(const pcm_sample_description_t *format,
         uint8_t *ip = (uint8_t *)input; \
         bytes_per_channel = PCM_BYTES_PER_CHANNEL(format); \
         for (i = 0; i < count; ++i) { \
-            (*result)[i] = conv(ip); \
+            result[i] = conv(ip); \
             ip += bytes_per_channel; \
         } \
     } while(0)
@@ -212,11 +182,6 @@ int pcm_convert_to_native_sint16(const pcm_sample_description_t *format,
     uint32_t count = nframes * format->channels_per_frame;
     if (!count)
         return 0;
-    if (!*result || *osize < count) {
-        *osize  = count;
-        *result = realloc(*result, count * sizeof(int16_t));
-    }
-
     switch (PCM_BYTES_PER_CHANNEL(format) | format->sample_type<<4) {
     case 1 | PCM_TYPE_SINT<<4:
         CONVERT(int8_t, pcm_s8_to_s16); break;

src/lpcm.h

@@ -31,7 +31,37 @@ typedef struct pcm_sample_description_t {
 #define PCM_BYTES_PER_CHANNEL(desc) \
     ((desc)->bytes_per_frame / (desc)->channels_per_frame)
 
+#if defined(_MSC_VER) && _MSC_VER < 1800
+#  ifdef _M_IX86
+static inline int lrint(double x)
+{
+    int n;
+    _asm {
+        fld x
+        fistp n
+    }
+    return n;
+}
+#  else
+#    include <emmintrin.h>
+static inline int lrint(double x)
+{
+    return _mm_cvtsd_si32(_mm_load_sd(&x));
+}
+#  endif
+#endif
+
+static
+inline double pcm_clip(double n, double min_value, double max_value)
+{
+    if (n < min_value)
+        return min_value;
+    else if (n > max_value)
+        return max_value;
+    return n;
+}
+
 int pcm_convert_to_native_sint16(const pcm_sample_description_t *format,
                                  const void *input, uint32_t nframes,
-                                 int16_t **result, uint32_t *osize);
+                                 int16_t *result);
 #endif

src/m4af.c

@@ -53,6 +53,7 @@ typedef struct m4af_track_t {
     uint32_t bufferSizeDB;
     uint32_t maxBitrate;
     uint32_t avgBitrate;
+    int is_vbr;
 
     m4af_sample_entry_t *sample_table;
     uint32_t num_samples;
@@ -278,6 +279,12 @@ DONE:
     return ctx->last_error;
 }
 
+void m4af_set_vbr_mode(m4af_ctx_t *ctx, uint32_t track_idx, int is_vbr)
+{
+    m4af_track_t *track = &ctx->track[track_idx];
+    track->is_vbr = is_vbr;
+}
+
 void m4af_set_priming(m4af_ctx_t *ctx, uint32_t track_idx,
                       uint32_t encoder_delay, uint32_t padding)
 {
@@ -771,7 +778,7 @@ void m4af_write_esds_box(m4af_ctx_t *ctx, uint32_t track_idx)
                , 2);
     m4af_write24(ctx, track->bufferSizeDB);
     m4af_write32(ctx, track->maxBitrate);
-    m4af_write32(ctx, track->avgBitrate);
+    m4af_write32(ctx, track->is_vbr ? 0: track->avgBitrate);
     /* DecoderSpecificInfo */
     m4af_write_descriptor(ctx, 5, track->decSpecificInfoSize);
     m4af_write(ctx, track->decSpecificInfo, track->decSpecificInfoSize);
@@ -857,17 +864,15 @@ void m4af_write_sbgp_box(m4af_ctx_t *ctx, uint32_t track_idx)
 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               */
+               "\0\0\0\026"  /* size: 22           */
                "sgpd"        /* type               */
-               "\001"        /* version            */
+               "\0"          /* 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);
+               , 22);
 }
 
 static
@@ -1044,7 +1049,6 @@ void m4af_write_elst_box(m4af_ctx_t *ctx, uint32_t track_idx)
 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);

src/m4af.h

@@ -97,6 +97,8 @@ int m4af_write_sample(m4af_ctx_t *ctx, uint32_t track_idx, const void *data,
 int m4af_set_decoder_specific_info(m4af_ctx_t *ctx, uint32_t track_idx,
                                    uint8_t *data, uint32_t size);
 
+void m4af_set_vbr_mode(m4af_ctx_t *ctx, uint32_t track_idx, int is_vbr);
+
 void m4af_set_priming(m4af_ctx_t *ctx, uint32_t track_idx,
                       uint32_t encoder_delay, uint32_t padding);
 

src/main.c

@@ -34,6 +34,7 @@
 #endif
 #include "compat.h"
 #include "wav_reader.h"
+#include "caf_reader.h"
 #include "aacenc.h"
 #include "m4af.h"
 #include "progress.h"
@@ -53,7 +54,8 @@ static void handle_signals(void)
 {
     int i, sigs[] = { SIGINT, SIGHUP, SIGTERM };
     for (i = 0; i < sizeof(sigs)/sizeof(sigs[0]); ++i) {
-        struct sigaction sa = { 0 };
+        struct sigaction sa;
+        memset(&sa, 0, sizeof sa);
         sa.sa_handler = signal_handler;
         sa.sa_flags |= SA_RESTART;
         sigaction(sigs[i], &sa, 0);
@@ -131,11 +133,14 @@ PROGNAME " %s\n"
 " -a, --afterburner <n>         Afterburner\n"
 "                                 0: Off\n"
 "                                 1: On(default)\n"
-" -L, --lowdelay-sbr            Enable ELD-SBR (AAC ELD only)\n"
+" -L, --lowdelay-sbr <-1|0|1>   Configure SBR activity on AAC ELD\n"
-" -s, --sbr-signaling <n>       SBR signaling mode\n"
+"                                -1: Use ELD SBR auto configurator\n"
-"                                 0: Implicit, backward compatible(default)\n"
+"                                 0: Disable SBR on ELD (default)\n"
-"                                 1: Explicit SBR and implicit PS\n"
+"                                 1: Enable SBR on ELD\n"
-"                                 2: Explicit hierarchical signaling\n"
+" -s, --sbr-ratio <0|1|2>       Controls activation of downsampled SBR\n"
+"                                 0: Use lib default (default)\n"
+"                                 1: downsampled SBR (default for ELD+SBR)\n"
+"                                 2: dual-rate SBR (default for HE-AAC)\n"
 " -f, --transport-format <n>    Transport format\n"
 "                                 0: RAW (default, muxed into M4A)\n"
 "                                 1: ADIF\n"
@@ -152,7 +157,10 @@ PROGNAME " %s\n"
 "                                 0: iTunSMPB (default)\n"
 "                                 1: ISO standard (edts + sgpd)\n"
 "                                 2: Both\n"
-" --ignorelength                Ignore length of WAV header\n"
+" --include-sbr-delay           Count SBR decoder delay in encoder delay\n"
+"                               This is not iTunes compatible, but is default\n"
+"                               behavior of FDK library.\n"
+" -I, --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"
@@ -203,8 +211,11 @@ typedef struct aacenc_param_ex_t {
     AACENC_PARAMS
 
     char *input_filename;
+    FILE *input_fp;
     char *output_filename;
+    FILE *output_fp;
     unsigned gapless_mode;
+    unsigned include_sbr_delay;
     unsigned ignore_length;
     int silent;
     int moov_before_mdat;
@@ -214,7 +225,9 @@ typedef struct aacenc_param_ex_t {
     unsigned raw_rate;
     const char *raw_format;
 
-    aacenc_tag_param_t tags;
+    aacenc_tag_store_t tags;
+    aacenc_tag_store_t source_tags;
+    aacenc_translate_generic_text_tag_ctx_t source_tag_ctx;
 
     char *json_filename;
 } aacenc_param_ex_t;
@@ -223,8 +236,9 @@ static
 int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
 {
     int ch;
-    unsigned n;
+    int n;
 
+#define OPT_INCLUDE_SBR_DELAY    M4AF_FOURCC('s','d','l','y')
 #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')
@@ -241,13 +255,14 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
         { "bitrate-mode",     required_argument, 0, 'm' },
         { "bandwidth",        required_argument, 0, 'w' },
         { "afterburner",      required_argument, 0, 'a' },
-        { "lowdelay-sbr",     no_argument,       0, 'L' },
+        { "lowdelay-sbr",     required_argument, 0, 'L' },
-        { "sbr-signaling",    required_argument, 0, 's' },
+        { "sbr-ratio",        required_argument, 0, 's' },
         { "transport-format", required_argument, 0, 'f' },
         { "adts-crc-check",   no_argument,       0, 'C' },
         { "header-period",    required_argument, 0, 'P' },
 
         { "gapless-mode",     required_argument, 0, 'G' },
+        { "include-sbr-delay", no_argument,      0, OPT_INCLUDE_SBR_DELAY  },
         { "ignorelength",     no_argument,       0, 'I' },
         { "silent",           no_argument,       0, 'S' },
         { "moov-before-mdat", no_argument,       0, OPT_MOOV_BEFORE_MDAT   },
@@ -319,14 +334,18 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
             params->afterburner = n;
             break;
         case 'L':
-            params->lowdelay_sbr = 1;
+            if (sscanf(optarg, "%d", &n) != 1 || n < -1 || n > 1) {
+                fprintf(stderr, "invalid arg for lowdelay-sbr\n");
+                return -1;
+            }
+            params->lowdelay_sbr = n;
             break;
         case 's':
             if (sscanf(optarg, "%u", &n) != 1 || n > 2) {
-                fprintf(stderr, "invalid arg for sbr-signaling\n");
+                fprintf(stderr, "invalid arg for sbr-ratio\n");
                 return -1;
             }
-            params->sbr_signaling = n;
+            params->sbr_ratio = n;
             break;
         case 'f':
             if (sscanf(optarg, "%u", &n) != 1) {
@@ -355,6 +374,9 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
             }
             params->gapless_mode = n;
             break;
+        case OPT_INCLUDE_SBR_DELAY:
+            params->include_sbr_delay = 1;
+            break;
         case 'I':
             params->ignore_length = 1;
             break;
@@ -396,7 +418,7 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
         case M4AF_TAG_TRACK:
         case M4AF_TAG_DISK:
         case M4AF_TAG_TEMPO:
-            aacenc_param_add_itmf_entry(&params->tags, ch, 0, optarg,
+            aacenc_add_tag_to_store(&params->tags, ch, 0, optarg,
                                     strlen(optarg), 0);
             break;
         case OPT_SHORT_TAG:
@@ -430,7 +452,7 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
                     for (; *optarg; ++optarg)
                         fcc = ((fcc << 8) | (*optarg & 0xff));
                 }
-                aacenc_param_add_itmf_entry(&params->tags, fcc, optarg,
+                aacenc_add_tag_to_store(&params->tags, fcc, optarg,
                                         val, strlen(val),
                                         ch == OPT_SHORT_TAG_FILE);
             }
@@ -470,16 +492,15 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
 };
 
 static
-int write_sample(FILE *ofp, m4af_ctx_t *m4af,
+int write_sample(FILE *ofp, m4af_ctx_t *m4af, aacenc_frame_t *frame)
-                 const void *data, uint32_t size, uint32_t duration)
 {
     if (!m4af) {
-        fwrite(data, 1, size, ofp);
+        fwrite(frame->data, 1, frame->size, ofp);
         if (ferror(ofp)) {
             fprintf(stderr, "ERROR: fwrite(): %s\n", strerror(errno));
             return -1;
         }
-    } else if (m4af_write_sample(m4af, 0, data, size, duration) < 0) {
+    } else if (m4af_write_sample(m4af, 0, frame->data, frame->size, 0) < 0) {
         fprintf(stderr, "ERROR: failed to write m4a sample\n");
         return -1;
     }
@@ -487,60 +508,76 @@ int write_sample(FILE *ofp, m4af_ctx_t *m4af,
 }
 
 static
-int encode(wav_reader_t *wavf, HANDLE_AACENCODER encoder,
+int encode(aacenc_param_ex_t *params, pcm_reader_t *reader,
-           uint32_t frame_length, FILE *ofp, m4af_ctx_t *m4af,
+           HANDLE_AACENCODER encoder, uint32_t frame_length, 
-           int show_progress)
+           m4af_ctx_t *m4af)
 {
-    uint8_t *ibuf = 0;
+    int16_t *ibuf = 0, *ip;
-    int16_t *pcmbuf = 0;
+    aacenc_frame_t obuf[2] = {{ 0 }}, *obp;
-    uint32_t pcmsize = 0;
+    unsigned flip = 0;
-    uint8_t *obuf = 0;
-    uint32_t olen;
-    uint32_t osize = 0;
     int nread = 1;
-    int consumed;
     int rc = -1;
-    int frames_written = 0;
+    int remaining, consumed;
+    int frames_written = 0, encoded = 0;
     aacenc_progress_t progress = { 0 };
-    const pcm_sample_description_t *format = wav_get_format(wavf);
+    const pcm_sample_description_t *fmt = pcm_get_format(reader);
 
-    ibuf = malloc(frame_length * format->bytes_per_frame);
+    ibuf = malloc(frame_length * fmt->bytes_per_frame);
-    aacenc_progress_init(&progress, wav_get_length(wavf), format->sample_rate);
+    aacenc_progress_init(&progress, pcm_get_length(reader), fmt->sample_rate);
-    do {
+
+    for (;;) {
+        /*
+         * Since we delay the write, we cannot just exit loop when interrupted.
+         * Instead, we regard it as EOF.
+         */
         if (g_interrupted)
             nread = 0;
-        else if (nread) {
+        if (nread > 0) {
-            if ((nread = wav_read_frames(wavf, ibuf, frame_length)) < 0) {
+            if ((nread = pcm_read_frames(reader, ibuf, frame_length)) < 0) {
                 fprintf(stderr, "ERROR: read failed\n");
                 goto END;
-            } else if (nread > 0) {
-                if (pcm_convert_to_native_sint16(format, ibuf, nread,
-                                                 &pcmbuf, &pcmsize) < 0) {
-                    fprintf(stderr, "ERROR: unsupported sample format\n");
-                    goto END;
             }
+            if (!params->silent)
+                aacenc_progress_update(&progress, pcm_get_position(reader),
+                                       fmt->sample_rate * 2);
         }
-            if (show_progress)
+        ip = ibuf;
-                aacenc_progress_update(&progress, wav_get_position(wavf),
+        remaining = nread;
-                                       format->sample_rate * 2);
+        do {
-        }
+            obp = &obuf[flip];
-        if ((consumed = aac_encode_frame(encoder, format, pcmbuf, nread,
+            consumed = aac_encode_frame(encoder, fmt, ip, remaining, obp);
-                                         &obuf, &olen, &osize)) < 0)
+            if (consumed < 0) goto END;
-            goto END;
+            if (consumed == 0 && obp->size == 0) goto DONE;
-        if (olen > 0) {
+            if (obp->size == 0) break;
-            if (write_sample(ofp, m4af, obuf, olen, frame_length) < 0)
+
+            remaining -= consumed;
+            ip += consumed * fmt->channels_per_frame;
+            flip ^= 1;
+            /*
+             * As we pad 1 frame at beginning and ending by our extrapolator,
+             * we want to drop them.
+             * We delay output by 1 frame by double buffering, and discard
+             * second frame and final frame from the encoder.
+             * Since sbr_header is included in the first frame (in case of
+             * SBR), we cannot discard first frame. So we pick second instead.
+             */
+            ++encoded;
+            if (encoded == 1 || encoded == 3)
+                continue;
+            obp = &obuf[flip];
+            if (write_sample(params->output_fp, m4af, obp) < 0)
                 goto END;
             ++frames_written;
+        } while (remaining > 0);
     }
-    } while (nread > 0 || olen > 0);
+DONE:
-
+    if (!params->silent)
-    if (show_progress)
+        aacenc_progress_finish(&progress, pcm_get_position(reader));
-        aacenc_progress_finish(&progress, wav_get_position(wavf));
     rc = frames_written;
 END:
     if (ibuf) free(ibuf);
-    if (pcmbuf) free(pcmbuf);
+    if (obuf[0].data) free(obuf[0].data);
-    if (obuf) free(obuf);
+    if (obuf[1].data) free(obuf[1].data);
     return rc;
 }
 
@@ -550,19 +587,11 @@ void put_tool_tag(m4af_ctx_t *m4af, const aacenc_param_ex_t *params,
 {
     char tool_info[256];
     char *p = tool_info;
-    LIB_INFO *lib_info = 0;
+    LIB_INFO lib_info;
 
     p += sprintf(p, PROGNAME " %s, ", fdkaac_version);
-
+    aacenc_get_lib_info(&lib_info);
-    lib_info = calloc(FDK_MODULE_LAST, sizeof(LIB_INFO));
+    p += sprintf(p, "libfdk-aac %s, ", lib_info.versionStr);
-    if (aacEncGetLibInfo(lib_info) == AACENC_OK) {
-        int i;
-        for (i = 0; i < FDK_MODULE_LAST; ++i)
-            if (lib_info[i].module_id == FDK_AACENC)
-                break;
-        p += sprintf(p, "libfdk-aac %s, ", lib_info[i].versionStr);
-    }
-    free(lib_info);
     if (params->bitrate_mode)
         sprintf(p, "VBR mode %d", params->bitrate_mode);
     else
@@ -577,13 +606,18 @@ int finalize_m4a(m4af_ctx_t *m4af, const aacenc_param_ex_t *params,
                  HANDLE_AACENCODER encoder)
 {
     unsigned i;
-    aacenc_tag_entry_t *tag = params->tags.tag_table;
+    aacenc_tag_entry_t *tag;
+    
+    tag = params->source_tags.tag_table;
+    for (i = 0; i < params->source_tags.tag_count; ++i, ++tag)
+        aacenc_write_tag_entry(m4af, tag);
 
     if (params->json_filename)
-        aacenc_put_tags_from_json(m4af, params->json_filename);
+        aacenc_write_tags_from_json(m4af, params->json_filename);
 
+    tag = params->tags.tag_table;
     for (i = 0; i < params->tags.tag_count; ++i, ++tag)
-        aacenc_put_tag_entry(m4af, tag);
+        aacenc_write_tag_entry(m4af, tag);
 
     put_tool_tag(m4af, params, encoder);
 
@@ -609,7 +643,7 @@ char *generate_output_filename(const char *filename, const char *ext)
         const char *ext_org = strrchr(base, '.');
         if (ext_org) ilen = ext_org - base;
         p = malloc(ilen + ext_len + 1);
-        sprintf(p, "%.*s%s", ilen, base, ext);
+        sprintf(p, "%.*s%s", (int)ilen, base, ext);
     }
     return p;
 }
@@ -650,25 +684,95 @@ int parse_raw_spec(const char *spec, pcm_sample_description_t *desc)
     return 0;
 }
 
+static pcm_io_vtbl_t pcm_io_vtbl = {
+    read_callback, seek_callback, tell_callback
+};
+static pcm_io_vtbl_t pcm_io_vtbl_noseek = { read_callback, 0, tell_callback };
+
+static
+pcm_reader_t *open_input(aacenc_param_ex_t *params)
+{
+    pcm_io_context_t io = { 0 };
+    pcm_reader_t *reader = 0;
+    struct stat stb = { 0 };
+
+    if ((params->input_fp = aacenc_fopen(params->input_filename, "rb")) == 0) {
+        aacenc_fprintf(stderr, "ERROR: %s: %s\n", params->input_filename,
+                       strerror(errno));
+        goto END;
+    }
+    io.cookie = params->input_fp;
+    if (fstat(fileno(params->input_fp), &stb) == 0
+            && (stb.st_mode & S_IFMT) == S_IFREG)
+        io.vtbl = &pcm_io_vtbl;
+    else
+        io.vtbl = &pcm_io_vtbl_noseek;
+
+    if (params->is_raw) {
+        int bytes_per_channel;
+        pcm_sample_description_t desc = { 0 };
+        if (parse_raw_spec(params->raw_format, &desc) < 0) {
+            fprintf(stderr, "ERROR: invalid raw-format spec\n");
+            goto END;
+        }
+        desc.sample_rate = params->raw_rate;
+        desc.channels_per_frame = params->raw_channels;
+        bytes_per_channel = (desc.bits_per_channel + 7) / 8;
+        desc.bytes_per_frame = params->raw_channels * bytes_per_channel;
+        if ((reader = raw_open(&io, &desc)) == 0) {
+            fprintf(stderr, "ERROR: failed to open raw input\n");
+            goto END;
+        }
+    } else {
+        int c;
+        ungetc(c = getc(params->input_fp), params->input_fp);
+
+        switch (c) {
+        case 'R':
+            if ((reader = wav_open(&io, params->ignore_length)) == 0) {
+                fprintf(stderr, "ERROR: broken / unsupported input file\n");
+                goto END;
+            }
+            break;
+        case 'c':
+            params->source_tag_ctx.add = aacenc_add_tag_entry_to_store;
+            params->source_tag_ctx.add_ctx = &params->source_tags;
+            if ((reader = caf_open(&io,
+                                   aacenc_translate_generic_text_tag,
+                                   &params->source_tag_ctx)) == 0) {
+                fprintf(stderr, "ERROR: broken / unsupported input file\n");
+                goto END;
+            }
+            break;
+        default:
+            fprintf(stderr, "ERROR: unsupported input file\n");
+            goto END;
+        }
+    }
+    if ((reader = pcm_open_sint16_converter(reader)) != 0)
+        reader = extrapolater_open(reader);
+    return reader;
+END:
+    return 0;
+}
+
 int main(int argc, char **argv)
 {
-    wav_io_context_t wav_io = { read_callback, seek_callback, tell_callback };
+    static m4af_io_callbacks_t m4af_io = {
-    m4af_io_callbacks_t
+        read_callback, 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;
-    FILE *ifp = 0;
-    FILE *ofp = 0;
     char *output_filename = 0;
-    wav_reader_t *wavf = 0;
+    pcm_reader_t *reader = 0;
     HANDLE_AACENCODER encoder = 0;
     AACENC_InfoStruct aacinfo = { 0 };
     m4af_ctx_t *m4af = 0;
     const pcm_sample_description_t *sample_format;
-    int downsampled_timescale = 0;
     int frame_count = 0;
-    struct stat stb = { 0 };
+    int sbr_mode = 0;
+    unsigned scale_shift = 0;
 
     setlocale(LC_CTYPE, "");
     setbuf(stderr, 0);
@@ -676,37 +780,23 @@ int main(int argc, char **argv)
     if (parse_options(argc, argv, &params) < 0)
         return 1;
 
-    if ((ifp = aacenc_fopen(params.input_filename, "rb")) == 0) {
+    if ((reader = open_input(&params)) == 0)
-        aacenc_fprintf(stderr, "ERROR: %s: %s\n", params.input_filename,
-                       strerror(errno));
         goto END;
+
+    sample_format = pcm_get_format(reader);
+
+    sbr_mode = aacenc_is_sbr_active((aacenc_param_t*)&params);
+    if (sbr_mode && !aacenc_is_sbr_ratio_available()) {
+        fprintf(stderr, "WARNING: Only dual-rate SBR is available "
+                        "for this version\n");
+        params.sbr_ratio = 2;
     }
-    if (fstat(fileno(ifp), &stb) == 0 && (stb.st_mode & S_IFMT) != S_IFREG) {
+    scale_shift = aacenc_is_dual_rate_sbr((aacenc_param_t*)&params);
-        wav_io.seek = 0;
+    params.sbr_signaling =
-        wav_io.tell = 0;
+        (params.transport_format == TT_MP4_LOAS) ? 2 :
-    }
+        (params.transport_format == TT_MP4_RAW)  ? 1 : 0;
-    if (!params.is_raw) {
+    if (sbr_mode && !scale_shift)
-        if ((wavf = wav_open(&wav_io, ifp, params.ignore_length)) == 0) {
+        params.sbr_signaling = 2;
-            fprintf(stderr, "ERROR: broken / unsupported input file\n");
-            goto END;
-        }
-    } else {
-        int bytes_per_channel;
-        pcm_sample_description_t desc = { 0 };
-        if (parse_raw_spec(params.raw_format, &desc) < 0) {
-            fprintf(stderr, "ERROR: invalid raw-format spec\n");
-            goto END;
-        }
-        desc.sample_rate = params.raw_rate;
-        desc.channels_per_frame = params.raw_channels;
-        bytes_per_channel = (desc.bits_per_channel + 7) / 8;
-        desc.bytes_per_frame = params.raw_channels * bytes_per_channel;
-        if ((wavf = raw_open(&wav_io, ifp, &desc)) == 0) {
-            fprintf(stderr, "ERROR: failed to open raw input\n");
-            goto END;
-        }
-    }
-    sample_format = wav_get_format(wavf);
 
     if (aacenc_init(&encoder, (aacenc_param_t*)&params, sample_format,
                     &aacinfo) < 0)
@@ -718,52 +808,68 @@ int main(int argc, char **argv)
         params.output_filename = output_filename;
     }
 
-    if ((ofp = aacenc_fopen(params.output_filename, "wb+")) == 0) {
+    if ((params.output_fp = aacenc_fopen(params.output_filename, "wb+")) == 0) {
         aacenc_fprintf(stderr, "ERROR: %s: %s\n", params.output_filename,
                        strerror(errno));
         goto END;
     }
     handle_signals();
+
     if (!params.transport_format) {
         uint32_t scale;
         unsigned framelen = aacinfo.frameLength;
-        int sbr_mode = aacenc_is_sbr_active((aacenc_param_t*)&params);
+        scale = sample_format->sample_rate >> scale_shift;
-        int sig_mode = aacEncoder_GetParam(encoder, AACENC_SIGNALING_MODE);
+        if ((m4af = m4af_create(M4AF_CODEC_MP4A, scale, &m4af_io,
-        if (sbr_mode && !sig_mode)
+                                params.output_fp)) < 0)
-            downsampled_timescale = 1;
-        scale = sample_format->sample_rate >> downsampled_timescale;
-        if ((m4af = m4af_create(M4AF_CODEC_MP4A, scale, &m4af_io, ofp)) < 0)
             goto END;
-        m4af_set_decoder_specific_info(m4af, 0, aacinfo.confBuf,
+        m4af_set_decoder_specific_info(m4af, 0,
-                                       aacinfo.confSize);
+                                       aacinfo.confBuf, aacinfo.confSize);
         m4af_set_fixed_frame_duration(m4af, 0,
-                                      framelen >> downsampled_timescale);
+                                      framelen >> scale_shift);
+        m4af_set_vbr_mode(m4af, 0, params.bitrate_mode);
         m4af_set_priming_mode(m4af, params.gapless_mode + 1);
         m4af_begin_write(m4af);
     }
-    frame_count = encode(wavf, encoder, aacinfo.frameLength, ofp, m4af,
+    if (scale_shift && (aacinfo.encoderDelay & 1)) {
-                         !params.silent);
+        /*
+         * Since odd delay cannot be exactly expressed in downsampled scale,
+         * we push one zero frame to the encoder here, to make delay even
+         */
+        int16_t zero[8] = { 0 };
+        aacenc_frame_t frame = { 0 };
+        aac_encode_frame(encoder, sample_format, zero, 1, &frame);
+        free(frame.data);
+    }
+    frame_count = encode(&params, reader, encoder, aacinfo.frameLength, m4af);
     if (frame_count < 0)
         goto END;
     if (m4af) {
         uint32_t delay = aacinfo.encoderDelay;
-        int64_t frames_read = wav_get_position(wavf);
+        uint32_t padding;
-        uint32_t padding = frame_count * aacinfo.frameLength
+        int64_t frames_read = pcm_get_position(reader);
-                            - frames_read - aacinfo.encoderDelay;
+
-        m4af_set_priming(m4af, 0, delay >> downsampled_timescale,
+        if (sbr_mode && params.profile != AOT_ER_AAC_ELD &&
-                         padding >> downsampled_timescale);
+            !params.include_sbr_delay)
+            delay -= 481 << scale_shift;
+        if (scale_shift && (delay & 1))
+            ++delay;
+        padding = frame_count * aacinfo.frameLength - frames_read - delay;
+        m4af_set_priming(m4af, 0, delay >> scale_shift, padding >> scale_shift);
         if (finalize_m4a(m4af, &params, encoder) < 0)
             goto END;
     }
     result = 0;
 END:
-    if (wavf) wav_teardown(&wavf);
+    if (reader) pcm_teardown(&reader);
-    if (ifp) fclose(ifp);
+    if (params.input_fp) fclose(params.input_fp);
     if (m4af) m4af_teardown(&m4af);
-    if (ofp) fclose(ofp);
+    if (params.output_fp) fclose(params.output_fp);
     if (encoder) aacEncClose(&encoder);
     if (output_filename) free(output_filename);
-    if (params.tags.tag_table) free(params.tags.tag_table);
+    if (params.tags.tag_table)
+        aacenc_free_tag_store(&params.tags);
+    if (params.source_tags.tag_table)
+        aacenc_free_tag_store(&params.source_tags);
 
     return result;
 }

src/metadata.c

@@ -1,3 +1,7 @@
+/* 
+ * Copyright (C) 2013 nu774
+ * For conditions of distribution and use, see copyright notice in COPYING
+ */
 #if HAVE_CONFIG_H
 #  include "config.h"
 #endif
@@ -50,12 +54,14 @@ static tag_key_mapping_t tag_mapping_table[] = {
     { "copyright",                  M4AF_TAG_COPYRIGHT           },
     { "date",                       M4AF_TAG_DATE                },
     { "disc",                       M4AF_TAG_DISK                },
-    { "disctotal",                  TAG_TOTAL_DISCS              },
     { "discnumber",                 M4AF_TAG_DISK                },
+    { "disctotal",                  TAG_TOTAL_DISCS              },
     { "genre",                      M4AF_TAG_GENRE               },
     { "grouping",                   M4AF_TAG_GROUPING            },
     { "itunescompilation",          M4AF_TAG_COMPILATION         },
     { "lyrics",                     M4AF_TAG_LYRICS              },
+    { "tempo",                      M4AF_TAG_TEMPO               },
+    { "recordeddate",               M4AF_TAG_DATE                },
     { "title",                      M4AF_TAG_TITLE               },
     { "titlesort",                  M4AF_FOURCC('s','o','n','m') },
     { "titlesortorder",             M4AF_FOURCC('s','o','n','m') },
@@ -95,6 +101,7 @@ uint32_t get_tag_fcc_from_name(const char *name)
     return ent ? ent->fcc : 0;
 }
 
+static
 char *aacenc_load_tag_from_file(const char *path, uint32_t *data_size)
 {
     FILE *fp = 0;
@@ -121,48 +128,152 @@ END:
     return data;
 }
 
-void aacenc_param_add_itmf_entry(aacenc_tag_param_t *params, uint32_t tag,
+static
+int aacenc_is_string_tag(uint32_t tag)
+{
+    switch (tag) {
+    case M4AF_TAG_TITLE:
+    case M4AF_TAG_ARTIST:
+    case M4AF_TAG_ALBUM:
+    case M4AF_TAG_GENRE:
+    case M4AF_TAG_DATE:
+    case M4AF_TAG_COMPOSER:
+    case M4AF_TAG_GROUPING:
+    case M4AF_TAG_COMMENT:
+    case M4AF_TAG_LYRICS:
+    case M4AF_TAG_TOOL:
+    case M4AF_TAG_ALBUM_ARTIST:
+    case M4AF_TAG_DESCRIPTION:
+    case M4AF_TAG_LONG_DESCRIPTION:
+    case M4AF_TAG_COPYRIGHT:
+    case M4AF_FOURCC('a','p','I','D'):
+    case M4AF_FOURCC('c','a','t','g'):
+    case M4AF_FOURCC('k','e','y','w'):
+    case M4AF_FOURCC('p','u','r','d'):
+    case M4AF_FOURCC('p','u','r','l'):
+    case M4AF_FOURCC('s','o','a','a'):
+    case M4AF_FOURCC('s','o','a','l'):
+    case M4AF_FOURCC('s','o','a','r'):
+    case M4AF_FOURCC('s','o','c','o'):
+    case M4AF_FOURCC('s','o','n','m'):
+    case M4AF_FOURCC('s','o','s','n'):
+    case M4AF_FOURCC('t','v','e','n'):
+    case M4AF_FOURCC('t','v','n','n'):
+    case M4AF_FOURCC('t','v','s','h'):
+    case M4AF_FOURCC('x','i','d',' '):
+    case M4AF_FOURCC('\xa9','e','n','c'):
+    case M4AF_FOURCC('\xa9','s','t','3'):
+    case M4AF_FOURCC('-','-','-','-'):
+        return 1;
+    }
+    return 0;
+}
+
+void aacenc_add_tag_to_store(aacenc_tag_store_t *store, uint32_t tag,
                              const char *key, const char *value,
                              uint32_t size, int is_file_name)
 {
-    aacenc_tag_entry_t *entry;
+    aacenc_tag_entry_t entry = { 0 };
+    char *dp = 0;
 
     if (!is_file_name && !size)
         return;
-    if (params->tag_count == params->tag_table_capacity) {
+    entry.tag = tag;
-        unsigned newsize = params->tag_table_capacity;
-        newsize = newsize ? newsize * 2 : 1;
-        params->tag_table =
-            realloc(params->tag_table, newsize * sizeof(aacenc_tag_entry_t));
-        params->tag_table_capacity = newsize;
-    }
-    entry = params->tag_table + params->tag_count;
-    entry->tag = tag;
     if (tag == M4AF_FOURCC('-','-','-','-'))
-        entry->name = key;
+        entry.name = (char *)key;
-    entry->data = value;
+
-    entry->data_size = size;
+    if (is_file_name) {
-    entry->is_file_name = is_file_name;
+        entry.data = dp = aacenc_load_tag_from_file(value, &size);
-    params->tag_count++;
+        entry.data_size = size;
+    } else if (aacenc_is_string_tag(tag)) {
+        entry.data = dp = aacenc_to_utf8(value);
+        entry.data_size = strlen(entry.data);
+    } else {
+        entry.data = (char *)value;
+        entry.data_size = size;
+    }
+    aacenc_add_tag_entry_to_store(store, &entry);
+    free(dp);
+}
+
+void aacenc_add_tag_entry_to_store(void *ctx, const aacenc_tag_entry_t *tag)
+{
+    aacenc_tag_store_t *store = ctx;
+    aacenc_tag_entry_t *entry;
+    if (store->tag_count == store->tag_table_capacity) {
+        unsigned newsize = store->tag_table_capacity;
+        newsize = newsize ? newsize * 2 : 1;
+        store->tag_table =
+            realloc(store->tag_table, newsize * sizeof(aacenc_tag_entry_t));
+        store->tag_table_capacity = newsize;
+    }
+    entry = store->tag_table + store->tag_count;
+    entry->tag  = tag->tag;
+    entry->data_size = tag->data_size;
+    entry->name = tag->name ? strdup(tag->name) : 0;
+    entry->data = malloc(tag->data_size + 1);
+    memcpy(entry->data, tag->data, tag->data_size);
+    entry->data[tag->data_size] = 0;
+    store->tag_count++;
 }
 
 static
-void tag_put_number_pair(m4af_ctx_t *m4af, uint32_t fcc,
+void tag_put_number_pair(aacenc_translate_generic_text_tag_ctx_t *ctx,
-                         const char *snumber, const char *stotal)
+                         uint32_t fcc, unsigned number, unsigned total)
 {
-    unsigned number = 0, total = 0;
     char buf[128];
     aacenc_tag_entry_t entry = { 0 };
 
-    if (snumber) sscanf(snumber, "%u", &number);
-    if (stotal)  sscanf(stotal,  "%u", &total);
     if (number) {
         if (total) sprintf(buf, "%u/%u", number, total);
         else       sprintf(buf, "%u",    number);
         entry.tag = fcc;
         entry.data = buf;
         entry.data_size = strlen(buf);
-        aacenc_put_tag_entry(m4af, &entry);
+        ctx->add(ctx->add_ctx, &entry);
+    }
+}
+
+void aacenc_translate_generic_text_tag(void *pctx, const char *key,
+                                       const char *val, uint32_t size)
+{
+    aacenc_translate_generic_text_tag_ctx_t *ctx = pctx;
+    aacenc_tag_entry_t entry = { 0 };
+    uint32_t fcc;
+    /*
+     * Since track/disc number pair (number and total) can be stored within
+     * either single tag or separately, we cannot instantly translate
+     * them in one-to-one manner.
+     * Instead, we keep and store them until all tags are processed,
+     * then finally submit.
+     */
+    if (!key) {
+        /* use null key as flushing signal */
+        tag_put_number_pair(ctx, M4AF_TAG_TRACK, ctx->track, ctx->track_total);
+        tag_put_number_pair(ctx, M4AF_TAG_DISK,  ctx->disc,  ctx->disc_total);
+        return;
+    }
+    if (!val || !size)
+        return;
+    if ((fcc = get_tag_fcc_from_name(key)) == 0)
+        return;
+
+    switch (fcc) {
+    case TAG_TOTAL_DISCS:
+        sscanf(val, "%d", &ctx->disc_total); break;
+    case TAG_TOTAL_TRACKS:
+        sscanf(val, "%d", &ctx->track_total); break;
+    case M4AF_TAG_DISK:
+        sscanf(val, "%d/%d", &ctx->disc, &ctx->disc_total); break;
+    case M4AF_TAG_TRACK:
+        sscanf(val, "%d/%d", &ctx->track, &ctx->track_total); break;
+    default:
+        {
+            entry.tag = fcc;
+            entry.data = (char *)val;
+            entry.data_size = (size == ~0U) ? strlen(val) : size;
+            ctx->add(ctx->add_ctx, &entry);
+        }
     }
 }
 
@@ -188,7 +299,7 @@ const char *aacenc_json_object_get_string(JSON_Object *obj, const char *key,
     return val;
 }
 
-void aacenc_put_tags_from_json(m4af_ctx_t *m4af, const char *json_filename)
+void aacenc_write_tags_from_json(m4af_ctx_t *m4af, const char *json_filename)
 {
     char *data = 0;
     JSON_Value *json = 0;
@@ -197,11 +308,10 @@ void aacenc_put_tags_from_json(m4af_ctx_t *m4af, const char *json_filename)
     uint32_t data_size;
     char *json_dot_path;
     char *filename = 0;
-    char *disc = 0;
+    aacenc_translate_generic_text_tag_ctx_t ctx = { 0 };
-    char *track = 0;
+
-    char *total_discs = 0;
+    ctx.add = aacenc_write_tag_entry;
-    char *total_tracks = 0;
+    ctx.add_ctx = m4af;
-    aacenc_tag_entry_t entry = { 0 };
 
     filename = strdup(json_filename);
     if ((json_dot_path = strchr(filename, '?')) != 0)
@@ -226,59 +336,36 @@ void aacenc_put_tags_from_json(m4af_ctx_t *m4af, const char *json_filename)
         char buf[256];
         const char *key = json_object_get_name(root, i);
         const char *val = aacenc_json_object_get_string(root, key, buf);
-        uint32_t fcc = get_tag_fcc_from_name(key);
+        if (val) aacenc_translate_generic_text_tag(&ctx, key, val, ~0U);
-        if (!val || !fcc)
-            continue;
-
-        switch (fcc) {
-        case TAG_TOTAL_DISCS:
-            total_discs = realloc(total_discs, strlen(val) + 1);
-            strcpy(total_discs, val);
-            break;
-        case TAG_TOTAL_TRACKS:
-            total_tracks = realloc(total_tracks, strlen(val) + 1);
-            strcpy(total_tracks, val);
-            break;
-        case M4AF_TAG_DISK:
-            disc = realloc(disc, strlen(val) + 1);
-            strcpy(disc, val);
-            break;
-        case M4AF_TAG_TRACK:
-            track = realloc(track, strlen(val) + 1);
-            strcpy(track, val);
-            break;
-        default:
-            {
-                entry.tag = fcc;
-                entry.data = val;
-                entry.data_size = strlen(val);
-                aacenc_put_tag_entry(m4af, &entry);
     }
-        }
+    aacenc_translate_generic_text_tag(&ctx, 0, 0, 0);
-    }
-    tag_put_number_pair(m4af, M4AF_TAG_TRACK, track, total_tracks);
-    tag_put_number_pair(m4af, M4AF_TAG_DISK,  disc,  total_discs);
 DONE:
-    if (track) free(track);
-    if (disc) free(disc);
-    if (total_tracks) free(total_tracks);
-    if (total_discs) free(total_discs);
     if (data) free(data);
     if (filename) free(filename);
     if (json) json_value_free(json);
 }
 
-void aacenc_put_tag_entry(m4af_ctx_t *m4af, const aacenc_tag_entry_t *tag)
+void aacenc_free_tag_store(aacenc_tag_store_t *store)
 {
+    if (store->tag_table) {
+        unsigned i;
+        for (i = 0; i < store->tag_count; ++i) {
+            aacenc_tag_entry_t *ent = &store->tag_table[i];
+            free(ent->name);
+            free(ent->data);
+        }
+        free(store->tag_table);
+        store->tag_table = 0;
+        store->tag_count = 0;
+    }
+}
+
+void aacenc_write_tag_entry(void *ctx, const aacenc_tag_entry_t *tag)
+{
+    m4af_ctx_t *m4af = ctx;
     unsigned m, n = 0;
     const char *data = tag->data;
-    uint32_t data_size = tag->data_size;
-    char *file_contents = 0;
 
-    if (tag->is_file_name) {
-        data = file_contents = aacenc_load_tag_from_file(tag->data, &data_size);
-        if (!data) return;
-    }
     switch (tag->tag) {
     case M4AF_TAG_TRACK:
         if (sscanf(data, "%u/%u", &m, &n) >= 1)
@@ -334,59 +421,22 @@ void aacenc_put_tag_entry(m4af_ctx_t *m4af, const aacenc_tag_entry_t *tag)
                 data_type = M4AF_PNG;
             if (data_type)
                 m4af_add_itmf_short_tag(m4af, tag->tag, data_type,
-                                        data, data_size);
+                                        data, tag->data_size);
             break;
         }
     case M4AF_FOURCC('-','-','-','-'):
         {
-            char *u8 = aacenc_to_utf8(data);
+            m4af_add_itmf_long_tag(m4af, tag->name, data);
-            m4af_add_itmf_long_tag(m4af, tag->name, u8);
-            free(u8);
-            break;
-        }
-    case M4AF_TAG_TITLE:
-    case M4AF_TAG_ARTIST:
-    case M4AF_TAG_ALBUM:
-    case M4AF_TAG_GENRE:
-    case M4AF_TAG_DATE:
-    case M4AF_TAG_COMPOSER:
-    case M4AF_TAG_GROUPING:
-    case M4AF_TAG_COMMENT:
-    case M4AF_TAG_LYRICS:
-    case M4AF_TAG_TOOL:
-    case M4AF_TAG_ALBUM_ARTIST:
-    case M4AF_TAG_DESCRIPTION:
-    case M4AF_TAG_LONG_DESCRIPTION:
-    case M4AF_TAG_COPYRIGHT:
-    case M4AF_FOURCC('a','p','I','D'):
-    case M4AF_FOURCC('c','a','t','g'):
-    case M4AF_FOURCC('k','e','y','w'):
-    case M4AF_FOURCC('p','u','r','d'):
-    case M4AF_FOURCC('p','u','r','l'):
-    case M4AF_FOURCC('s','o','a','a'):
-    case M4AF_FOURCC('s','o','a','l'):
-    case M4AF_FOURCC('s','o','a','r'):
-    case M4AF_FOURCC('s','o','c','o'):
-    case M4AF_FOURCC('s','o','n','m'):
-    case M4AF_FOURCC('s','o','s','n'):
-    case M4AF_FOURCC('t','v','e','n'):
-    case M4AF_FOURCC('t','v','n','n'):
-    case M4AF_FOURCC('t','v','s','h'):
-    case M4AF_FOURCC('x','i','d',' '):
-    case M4AF_FOURCC('\xa9','e','n','c'):
-    case M4AF_FOURCC('\xa9','s','t','3'):
-        {
-            char *u8 = aacenc_to_utf8(data);
-            m4af_add_itmf_string_tag(m4af, tag->tag, u8);
-            free(u8);
             break;
         }
     default:
+        if (aacenc_is_string_tag(tag->tag))
+            m4af_add_itmf_string_tag(m4af, tag->tag, data);
+        else
             fprintf(stderr, "WARNING: unknown/unsupported tag: %c%c%c%c\n",
                     tag->tag >> 24, (tag->tag >> 16) & 0xff,
                     (tag->tag >> 8) & 0xff, tag->tag & 0xff);
     }
-    if (file_contents) free(file_contents);
 }
 
 

src/metadata.h

@@ -1,28 +1,48 @@
+/* 
+ * Copyright (C) 2013 nu774
+ * For conditions of distribution and use, see copyright notice in COPYING
+ */
 #ifndef METADATA_H
 #define METADATA_H
 
+#include "m4af.h"
+
 typedef struct aacenc_tag_entry_t {
     uint32_t tag;
-    const char *name;
+    char    *name;
-    const char *data;
+    char    *data;
     uint32_t data_size;
-    int         is_file_name;
 } aacenc_tag_entry_t;
 
-typedef struct aacenc_tag_param_t {
+typedef struct aacenc_tag_store_t {
     aacenc_tag_entry_t *tag_table;
     unsigned tag_count;
     unsigned tag_table_capacity;
-} aacenc_tag_param_t;
+} aacenc_tag_store_t;
 
-char *aacenc_load_tag_from_file(const char *path, uint32_t *data_size);
+typedef struct aacenc_translate_generic_text_tag_ctx_t {
+    unsigned track, track_total, disc, disc_total;
+    void   (*add)(void *, const aacenc_tag_entry_t *);
+    void    *add_ctx;
+} aacenc_translate_generic_text_tag_ctx_t;
 
-void aacenc_param_add_itmf_entry(aacenc_tag_param_t *params, uint32_t tag,
+typedef void (*aacenc_tag_callback_t)(void *ctx, const char *key,
+                                      const char *value, uint32_t size);
+
+void aacenc_translate_generic_text_tag(void *ctx, const char *key,
+                                       const char *val, uint32_t size);
+
+
+void aacenc_add_tag_to_store(aacenc_tag_store_t *store, uint32_t tag,
                              const char *key, const char *value,
                              uint32_t size, int is_file_name);
 
-void aacenc_put_tags_from_json(m4af_ctx_t *m4af, const char *json_filename);
+void aacenc_add_tag_entry_to_store(void *store, const aacenc_tag_entry_t *tag);
 
-void aacenc_put_tag_entry(m4af_ctx_t *m4af, const aacenc_tag_entry_t *tag);
+void aacenc_free_tag_store(aacenc_tag_store_t *store);
+
+void aacenc_write_tags_from_json(m4af_ctx_t *m4af, const char *json_filename);
+
+void aacenc_write_tag_entry(void *m4af, const aacenc_tag_entry_t *tag);
 
 #endif

src/pcm_reader.h

@@ -0,0 +1,118 @@
+/* 
+ * Copyright (C) 2013 nu774
+ * For conditions of distribution and use, see copyright notice in COPYING
+ */
+#ifndef PCM_READER_H
+#define PCM_READER_H
+
+#include "lpcm.h"
+
+typedef struct pcm_reader_t pcm_reader_t;
+
+typedef struct pcm_reader_vtbl_t {
+    const pcm_sample_description_t *(*get_format)(pcm_reader_t *);
+    int64_t (*get_length)(pcm_reader_t *);
+    int64_t (*get_position)(pcm_reader_t *);
+    int (*read_frames)(pcm_reader_t *, void *, unsigned);
+    void (*teardown)(pcm_reader_t **);
+} pcm_reader_vtbl_t;
+
+struct pcm_reader_t {
+    pcm_reader_vtbl_t *vtbl;
+};
+
+typedef int (*pcm_read_callback)(void *cookie, void *data, uint32_t count);
+typedef int (*pcm_seek_callback)(void *cookie, int64_t off, int whence);
+typedef int64_t (*pcm_tell_callback)(void *cookie);
+
+typedef struct pcm_io_vtbl_t {
+    pcm_read_callback read;
+    pcm_seek_callback seek;
+    pcm_tell_callback tell;
+} pcm_io_vtbl_t;
+
+typedef struct pcm_io_context_t {
+    pcm_io_vtbl_t *vtbl;
+    void *cookie;
+} pcm_io_context_t;
+
+static inline
+const pcm_sample_description_t *pcm_get_format(pcm_reader_t *r)
+{
+    return r->vtbl->get_format(r);
+}
+
+static inline
+int64_t pcm_get_length(pcm_reader_t *r)
+{
+    return r->vtbl->get_length(r);
+}
+
+static inline
+int64_t pcm_get_position(pcm_reader_t *r)
+{
+    return r->vtbl->get_position(r);
+}
+
+static inline
+int64_t pcm_read_frames(pcm_reader_t *r, void *data, unsigned nframes)
+{
+    return r->vtbl->read_frames(r, data, nframes);
+}
+
+static inline
+void pcm_teardown(pcm_reader_t **r)
+{
+    (*r)->vtbl->teardown(r);
+}
+
+static inline
+uint32_t bitcount(uint32_t bits)
+{
+    bits = (bits & 0x55555555) + (bits >> 1 & 0x55555555);
+    bits = (bits & 0x33333333) + (bits >> 2 & 0x33333333);
+    bits = (bits & 0x0f0f0f0f) + (bits >> 4 & 0x0f0f0f0f);
+    bits = (bits & 0x00ff00ff) + (bits >> 8 & 0x00ff00ff);
+    return (bits & 0x0000ffff) + (bits >>16 & 0x0000ffff);
+}
+
+#define TRY_IO(expr) \
+    do { \
+        if ((expr)) goto FAIL; \
+    } while (0)
+
+#define ENSURE(expr) \
+    do { \
+        if (!(expr)) goto FAIL;\
+    } while (0)
+
+int pcm_read(pcm_io_context_t *io, void *buffer, uint32_t size);
+int pcm_skip(pcm_io_context_t *io, int64_t count);
+
+static inline int pcm_seek(pcm_io_context_t *io, int64_t off, int whence)
+{
+    return io->vtbl->seek ? io->vtbl->seek(io->cookie, off, whence) : -1;
+}
+
+static inline int64_t pcm_tell(pcm_io_context_t *io)
+{
+    return io->vtbl->tell ? io->vtbl->tell(io->cookie) : -1;
+}
+
+int pcm_read16le(pcm_io_context_t *io, uint16_t *value);
+int pcm_read16be(pcm_io_context_t *io, uint16_t *value);
+int pcm_read32le(pcm_io_context_t *io, uint32_t *value);
+int pcm_read32be(pcm_io_context_t *io, uint32_t *value);
+int pcm_read64le(pcm_io_context_t *io, uint64_t *value);
+int pcm_read64be(pcm_io_context_t *io, uint64_t *value);
+int pcm_scanl(pcm_io_context_t *io, const char *fmt, ...);
+int pcm_scanb(pcm_io_context_t *io, const char *fmt, ...);
+
+int apple_chan_chunk(pcm_io_context_t *io, uint32_t chunk_size,
+                     pcm_sample_description_t *fmt, uint8_t *mapping);
+
+pcm_reader_t *pcm_open_sint16_converter(pcm_reader_t *reader);
+
+pcm_reader_t *extrapolater_open(pcm_reader_t *reader);
+
+#endif

src/pcm_readhelper.c

@@ -0,0 +1,356 @@
+/* 
+ * Copyright (C) 2013 nu774
+ * For conditions of distribution and use, see copyright notice in COPYING
+ */
+
+#if HAVE_CONFIG_H
+#  include "config.h"
+#endif
+
+#if HAVE_STDINT_H
+#  include <stdint.h>
+#endif
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdarg.h>
+#include "pcm_reader.h"
+#include "m4af_endian.h"
+#include "catypes.h"
+
+int pcm_read(pcm_io_context_t *io, void *buffer, uint32_t size)
+{
+    int rc;
+    uint32_t count = 0;
+
+    do {
+        rc = io->vtbl->read(io->cookie, buffer, size - count);
+        if (rc > 0)
+            count += rc;
+    } while (rc > 0 && count < size);
+    return count > 0 ? count : rc;
+}
+
+int pcm_skip(pcm_io_context_t *io, int64_t count)
+{
+    char buff[8192];
+    int rc;
+    pcm_io_vtbl_t *vp = io->vtbl;
+
+    if (count == 0 || pcm_seek(io, count, SEEK_CUR) >= 0)
+        return 0;
+    do {
+        if ((rc = vp->read(io->cookie, buff, count > 8192 ? 8192 : count)) > 0)
+            count -= rc;
+    } while (rc > 0 && count > 0);
+
+    return count == 0 ? 0 : -1;
+}
+
+int pcm_read16le(pcm_io_context_t *io, uint16_t *value)
+{
+    if (pcm_read(io, value, 2) == 2) {
+        *value = m4af_ltoh16(*value);
+        return 0;
+    }
+    return -1;
+}
+
+int pcm_read16be(pcm_io_context_t *io, uint16_t *value)
+{
+    if (pcm_read(io, value, 2) == 2) {
+        *value = m4af_btoh16(*value);
+        return 0;
+    }
+    return -1;
+}
+
+int pcm_read32le(pcm_io_context_t *io, uint32_t *value)
+{
+    if (pcm_read(io, value, 4) == 4) {
+        *value = m4af_ltoh32(*value);
+        return 0;
+    }
+    return -1;
+}
+
+int pcm_read32be(pcm_io_context_t *io, uint32_t *value)
+{
+    if (pcm_read(io, value, 4) == 4) {
+        *value = m4af_btoh32(*value);
+        return 0;
+    }
+    return -1;
+}
+
+int pcm_read64le(pcm_io_context_t *io, uint64_t *value)
+{
+    if (pcm_read(io, value, 8) == 8) {
+        *value = m4af_ltoh64(*value);
+        return 0;
+    }
+    return -1;
+}
+
+int pcm_read64be(pcm_io_context_t *io, uint64_t *value)
+{
+    if (pcm_read(io, value, 8) == 8) {
+        *value = m4af_btoh64(*value);
+        return 0;
+    }
+    return -1;
+}
+
+int pcm_scanl(pcm_io_context_t *io, const char *fmt, ...)
+{
+    int c, count = 0;
+    va_list ap;
+
+    va_start(ap, fmt);
+    while ((c = *fmt++)) {
+        switch (c) {
+        case 'S':
+            TRY_IO(pcm_read16le(io, va_arg(ap, uint16_t*)));
+            ++count;
+            break;
+        case 'L':
+            TRY_IO(pcm_read32le(io, va_arg(ap, uint32_t*)));
+            ++count;
+            break;
+        case 'Q':
+            TRY_IO(pcm_read64le(io, va_arg(ap, uint64_t*)));
+            ++count;
+            break;
+        }
+    }
+FAIL:
+    va_end(ap);
+    return count;
+}
+
+int pcm_scanb(pcm_io_context_t *io, const char *fmt, ...)
+{
+    int c, count = 0;
+    va_list ap;
+
+    va_start(ap, fmt);
+    while ((c = *fmt++)) {
+        switch (c) {
+        case 'S':
+            TRY_IO(pcm_read16be(io, va_arg(ap, uint16_t*)));
+            ++count;
+            break;
+        case 'L':
+            TRY_IO(pcm_read32be(io, va_arg(ap, uint32_t*)));
+            ++count;
+            break;
+        case 'Q':
+            TRY_IO(pcm_read64be(io, va_arg(ap, uint64_t*)));
+            ++count;
+            break;
+        }
+    }
+FAIL:
+    va_end(ap);
+    return count;
+}
+
+static
+int channel_compare(const void *a, const void *b)
+{
+    return (*(const uint8_t **)a)[0] - (*(const uint8_t **)b)[0];
+}
+
+void apple_translate_channel_labels(uint8_t *channels, unsigned n)
+{
+    unsigned i;
+    char *has_side = strpbrk((char*)channels, "\x0A\x0B");
+
+    for (i = 0; i < n; ++i) {
+        switch (channels[i]) {
+        case kAudioChannelLabel_LeftSurround:
+        case kAudioChannelLabel_RightSurround:
+            if (!has_side) channels[i] += 5; // map to SL/SR
+            break;
+        case kAudioChannelLabel_RearSurroundLeft:
+        case kAudioChannelLabel_RearSurroundRight:
+            if (!has_side) channels[i] -= 28; // map to BL/BR
+            break;
+        case kAudioChannelLabel_Mono:
+            channels[i] = kAudioChannelLabel_Center;
+            break;
+        }
+    }
+}
+
+int apple_chan_chunk(pcm_io_context_t *io, uint32_t chunk_size,
+                     pcm_sample_description_t *fmt, uint8_t *mapping)
+{
+    /*
+     * Although FDK encoder supports upto 5.1ch, we handle upto
+     * 8 channels here.
+     */
+    uint32_t i, mChannelLayoutTag, mChannelBitmap, mNumberChannelDescriptions;
+    uint32_t mask = 0;
+    const uint32_t nchannels = fmt->channels_per_frame;
+    uint8_t channels[9] = { 0 };
+    uint8_t *index[8] = { 0 };
+    const char *layout = 0;
+
+    ENSURE(chunk_size >= 12);
+    TRY_IO(pcm_scanb(io, "LLL", &mChannelLayoutTag, &mChannelBitmap,
+                     &mNumberChannelDescriptions) != 3);
+
+    switch (mChannelLayoutTag) {
+    case kAudioChannelLayoutTag_UseChannelBitmap:
+        ENSURE(bitcount(mask) == nchannels);
+        TRY_IO(pcm_skip(io, chunk_size - 12));
+        fmt->channel_mask = mChannelBitmap;
+        for (i = 0; i < nchannels; ++i)
+            mapping[i] = i;
+        return 0;
+    case kAudioChannelLayoutTag_UseChannelDescriptions:
+        ENSURE(mNumberChannelDescriptions == nchannels);
+        ENSURE(chunk_size >= 12 + nchannels * 20);
+        for (i = 0; i < mNumberChannelDescriptions; ++i) {
+            uint32_t mChannelLabel;
+            TRY_IO(pcm_read32be(io, &mChannelLabel));
+            ENSURE(mChannelLabel && mChannelLabel <= 0xff);
+            channels[i] = mChannelLabel;
+            TRY_IO(pcm_skip(io, 16));
+        }
+        TRY_IO(pcm_skip(io, chunk_size - 12 - nchannels * 20));
+        apple_translate_channel_labels(channels, nchannels);
+        for (i = 0; i < nchannels; ++i)
+            if (channels[i] > kAudioChannelLabel_TopBackLeft)
+                goto FAIL;
+        break;
+    default:
+        ENSURE((mChannelLayoutTag & 0xffff) == nchannels);
+        TRY_IO(pcm_skip(io, chunk_size - 12));
+
+        switch (mChannelLayoutTag) {
+        /* 1ch */
+        case kAudioChannelLayoutTag_Mono:
+            layout = "\x03"; break;
+        /* 1.1ch */
+        case kAudioChannelLayoutTag_AC3_1_0_1:
+            layout = "\x03\x04"; break;
+        /* 2ch */
+        case kAudioChannelLayoutTag_Stereo:
+        case kAudioChannelLayoutTag_MatrixStereo:
+        case kAudioChannelLayoutTag_Binaural:
+            layout = "\x01\x02"; break;
+        /* 2.1ch */
+        case kAudioChannelLayoutTag_DVD_4:
+            layout = "\x01\x02\x04"; break;
+        /* 3ch */
+        case kAudioChannelLayoutTag_MPEG_3_0_A:
+            layout = "\x01\x02\x03"; break;
+        case kAudioChannelLayoutTag_AC3_3_0:
+            layout = "\x01\x03\x02"; break;
+        case kAudioChannelLayoutTag_MPEG_3_0_B:
+            layout = "\x03\x01\x02"; break;
+        case kAudioChannelLayoutTag_ITU_2_1:
+            layout = "\x01\x02\x09"; break;
+        /* 3.1ch */
+        case kAudioChannelLayoutTag_DVD_10:
+            layout = "\x01\x02\x03\x04"; break;
+        case kAudioChannelLayoutTag_AC3_3_0_1:
+            layout = "\x01\x03\x02\x04"; break;
+        case kAudioChannelLayoutTag_DVD_5:
+            layout = "\x01\x02\x04\x09"; break;
+        case kAudioChannelLayoutTag_AC3_2_1_1:
+            layout = "\x01\x02\x09\x04"; break;
+        /* 4ch */
+        case kAudioChannelLayoutTag_Quadraphonic:
+        case kAudioChannelLayoutTag_ITU_2_2:
+            layout = "\x01\x02\x0A\x0B"; break;
+        case kAudioChannelLayoutTag_MPEG_4_0_A:
+            layout = "\x01\x02\x03\x09"; break;
+        case kAudioChannelLayoutTag_MPEG_4_0_B:
+            layout = "\x03\x01\x02\x09"; break;
+        case kAudioChannelLayoutTag_AC3_3_1:
+            layout = "\x01\x03\x02\x09"; break;
+        /* 4.1ch */
+        case kAudioChannelLayoutTag_DVD_6:
+            layout = "\x01\x02\x04\x0A\x0B"; break;
+        case kAudioChannelLayoutTag_DVD_18:
+            layout = "\x01\x02\x0A\x0B\x04"; break;
+        case kAudioChannelLayoutTag_DVD_11:
+            layout = "\x01\x02\x03\x04\x09"; break;
+        case kAudioChannelLayoutTag_AC3_3_1_1:
+            layout = "\x01\x03\x02\x09\x04"; break;
+        /* 5ch */
+        case kAudioChannelLayoutTag_MPEG_5_0_A:
+            layout = "\x01\x02\x03\x0A\x0B"; break;
+        case kAudioChannelLayoutTag_Pentagonal:
+        case kAudioChannelLayoutTag_MPEG_5_0_B:
+            layout = "\x01\x02\x0A\x0B\x03"; break;
+        case kAudioChannelLayoutTag_MPEG_5_0_C:
+            layout = "\x01\x03\x02\x0A\x0B"; break;
+        case kAudioChannelLayoutTag_MPEG_5_0_D:
+            layout = "\x03\x01\x02\x0A\x0B"; break;
+        /* 5.1ch */
+        case kAudioChannelLayoutTag_MPEG_5_1_A:
+            layout = "\x01\x02\x03\x04\x0A\x0B"; break;
+        case kAudioChannelLayoutTag_MPEG_5_1_B:
+            layout = "\x01\x02\x0A\x0B\x03\x04"; break;
+        case kAudioChannelLayoutTag_MPEG_5_1_C:
+            layout = "\x01\x03\x02\x0A\x0B\x04"; break;
+        case kAudioChannelLayoutTag_MPEG_5_1_D:
+            layout = "\x03\x01\x02\x0A\x0B\x04"; break;
+        /* 6ch */
+        case kAudioChannelLayoutTag_Hexagonal:
+        case kAudioChannelLayoutTag_AudioUnit_6_0:
+            layout = "\x01\x02\x0A\x0B\x03\x09"; break;
+        case kAudioChannelLayoutTag_AAC_6_0:
+            layout = "\x03\x01\x02\x0A\x0B\x09"; break;
+        /* 6.1ch */
+        case kAudioChannelLayoutTag_MPEG_6_1_A:
+            layout = "\x01\x02\x03\x04\x0A\x0B\x09"; break;
+        case kAudioChannelLayoutTag_AAC_6_1:
+            layout = "\x03\x01\x02\x0A\x0B\x09\x04"; break;
+        /* 7ch */
+        case kAudioChannelLayoutTag_AudioUnit_7_0:
+            layout = "\x01\x02\x0A\x0B\x03\x05\x06"; break;
+        case kAudioChannelLayoutTag_AudioUnit_7_0_Front:
+            layout = "\x01\x02\x0A\x0B\x03\x07\x08"; break;
+        case kAudioChannelLayoutTag_AAC_7_0:
+            layout = "\x03\x01\x02\x0A\x0B\x05\x06"; break;
+        /* 7.1ch */
+        case kAudioChannelLayoutTag_MPEG_7_1_A:
+            layout = "\x01\x02\x03\x04\x0A\x0B\x07\x08"; break;
+        case kAudioChannelLayoutTag_MPEG_7_1_B:
+            layout = "\x03\x07\x08\x01\x02\x05\x06\x04"; break;
+        case kAudioChannelLayoutTag_MPEG_7_1_C:
+            layout = "\x01\x02\x03\x04\x0A\x0B\x05\x06"; break;
+        case kAudioChannelLayoutTag_Emagic_Default_7_1:
+            layout = "\x01\x02\x0A\x0B\x03\x04\x07\x08"; break;
+        /* 8ch */
+        case kAudioChannelLayoutTag_Octagonal:
+            layout = "\x01\x02\x05\x06\x03\x09\x0A\x0B"; break;
+        case kAudioChannelLayoutTag_AAC_Octagonal:
+            layout = "\x03\x01\x02\x0A\x0B\x05\x06\x09"; break;
+        default:
+            goto FAIL;
+        }
+        strcpy((char*)channels, layout);
+    }
+
+    for (i = 0; i < nchannels; ++i)
+        mask |= 1 << (channels[i] - 1);
+    fmt->channel_mask = mask;
+    ENSURE(bitcount(mask) == nchannels);
+
+    for (i = 0; i < nchannels; ++i) 
+        index[i] = channels + i;
+    qsort(index, nchannels, sizeof(char*), channel_compare);
+    for (i = 0; i < nchannels; ++i) 
+        mapping[i] = index[i] - channels;
+
+    return 0;
+FAIL:
+    return -1;
+}

src/pcm_sint16_converter.c

@@ -0,0 +1,94 @@
+/* 
+ * Copyright (C) 2013 nu774
+ * For conditions of distribution and use, see copyright notice in COPYING
+ */
+#if HAVE_CONFIG_H
+#  include "config.h"
+#endif
+#include <stdlib.h>
+#include <string.h>
+#if HAVE_STDINT_H
+#  include <stdint.h>
+#endif
+#include "pcm_reader.h"
+
+typedef struct pcm_sint16_converter_t {
+    pcm_reader_vtbl_t *vtbl;
+    pcm_reader_t *src;
+    pcm_sample_description_t format;
+    void *pivot;
+    unsigned capacity;
+} pcm_sint16_converter_t;
+
+static inline pcm_reader_t *get_source(pcm_reader_t *reader)
+{
+    return ((pcm_sint16_converter_t *)reader)->src;
+}
+
+static const
+pcm_sample_description_t *get_format(pcm_reader_t *reader)
+{
+    return &((pcm_sint16_converter_t *)reader)->format;
+}
+
+static int64_t get_length(pcm_reader_t *reader)
+{
+    return pcm_get_length(get_source(reader));
+}
+
+static int64_t get_position(pcm_reader_t *reader)
+{
+    return pcm_get_position(get_source(reader));
+}
+
+static int read_frames(pcm_reader_t *reader, void *buffer, unsigned nframes)
+{
+    pcm_sint16_converter_t *self = (pcm_sint16_converter_t *)reader;
+    const pcm_sample_description_t *sfmt = pcm_get_format(self->src);
+    unsigned bytes = nframes * sfmt->bytes_per_frame;
+
+    if (self->capacity < bytes) {
+        void *p = realloc(self->pivot, bytes);
+        if (!p) return -1;
+        self->pivot = p;
+        self->capacity = bytes;
+    }
+    nframes = pcm_read_frames(self->src, self->pivot, nframes);
+    if (pcm_convert_to_native_sint16(sfmt, self->pivot, nframes, buffer) < 0)
+        return -1;
+    return nframes;
+}
+
+static void teardown(pcm_reader_t **reader)
+{
+    pcm_sint16_converter_t *self = (pcm_sint16_converter_t *)*reader;
+    pcm_teardown(&self->src);
+    free(self->pivot);
+    free(self);
+    *reader = 0;
+}
+
+static pcm_reader_vtbl_t my_vtable = {
+    get_format, get_length, get_position, read_frames, teardown
+};
+
+pcm_reader_t *pcm_open_sint16_converter(pcm_reader_t *reader)
+{
+    pcm_sint16_converter_t *self = 0;
+    pcm_sample_description_t *fmt;
+
+    if ((self = calloc(1, sizeof(pcm_sint16_converter_t))) == 0)
+        return 0;
+    self->src = reader;
+    self->vtbl = &my_vtable;
+    memcpy(&self->format, pcm_get_format(reader), sizeof(self->format));
+    fmt = &self->format;
+#if WORDS_BIGENDIAN
+    fmt->sample_type = PCM_TYPE_SINT_BE;
+#else
+    fmt->sample_type = PCM_TYPE_SINT;
+#endif
+    fmt->bits_per_channel = 16;
+    fmt->bytes_per_frame = 2 * fmt->channels_per_frame;
+    return (pcm_reader_t *)self;
+}

src/wav_reader.c

@@ -15,35 +15,18 @@
 #include <string.h>
 #include <stdarg.h>
 #include "wav_reader.h"
-#include "m4af_endian.h"
 
 #define RIFF_FOURCC(a,b,c,d) ((a)|((b)<<8)|((c)<<16)|((d)<<24))
 
-#define TRY_IO(expr) \
+typedef struct wav_reader_t {
-    do { \
+    pcm_reader_vtbl_t *vtbl;
-        if (expr) \
-            goto FAIL; \
-    } while (0)
-
-#define ASSERT_FORMAT(ctx, expr) \
-    do { \
-        if (!expr) { \
-            if (!ctx->last_error) \
-                ctx->last_error = WAV_INVALID_FORMAT; \
-            goto FAIL;\
-        } \
-    } while (0)
-
-struct wav_reader_t {
     pcm_sample_description_t sample_format;
     int64_t length;
     int64_t position;
     int32_t data_offset;
     int ignore_length;
-    int last_error;
+    pcm_io_context_t io;
-    wav_io_context_t io;
+} wav_reader_t;
-    void *io_cookie;
-};
 
 static const uint8_t WAV_GUID_PCM[] = {
     1, 0, 0, 0, 0, 0, 0x10, 0, 0x80, 0, 0, 0xaa, 0, 0x38, 0x9b, 0x71
@@ -52,177 +35,46 @@ static const uint8_t WAV_GUID_FLOAT[] = {
     3, 0, 0, 0, 0, 0, 0x10, 0, 0x80, 0, 0, 0xaa, 0, 0x38, 0x9b, 0x71
 };
 
-const pcm_sample_description_t *wav_get_format(wav_reader_t *reader)
+static const pcm_sample_description_t *wav_get_format(pcm_reader_t *reader)
 {
-    return &reader->sample_format;
+    return &((wav_reader_t *)reader)->sample_format;
 }
 
-int64_t wav_get_length(wav_reader_t *reader)
+static int64_t wav_get_length(pcm_reader_t *reader)
 {
-    return reader->length;
+    return ((wav_reader_t *)reader)->length;
 }
 
-int64_t wav_get_position(wav_reader_t *reader)
+static int64_t wav_get_position(pcm_reader_t *reader)
 {
-    return reader->position;
+    return ((wav_reader_t *)reader)->position;
 }
 
-void wav_teardown(wav_reader_t **reader)
+static void wav_teardown(pcm_reader_t **reader)
 {
     free(*reader);
     *reader = 0;
 }
 
-static
-int riff_read(wav_reader_t *reader, void *buffer, uint32_t size)
-{
-    int rc;
-    uint32_t count = 0;
-
-    if (reader->last_error)
-        return -1;
-    do {
-        rc = reader->io.read(reader->io_cookie, buffer, size - count);
-        if (rc > 0)
-            count += rc;
-        else if (rc < 0)
-            reader->last_error = WAV_IO_ERROR;
-    } while (rc > 0 && count < size);
-    return count > 0 ? count : rc;
-}
-
-static
-int riff_skip(wav_reader_t *reader, int64_t count)
-{
-    char buff[8192];
-    int rc;
-
-    if (reader->last_error)
-        return -1;
-    if (count == 0)
-        return 0;
-    if (reader->io.seek &&
-        reader->io.seek(reader->io_cookie, count, SEEK_CUR) >= 0)
-        return 0;
-
-    do {
-        if ((rc = riff_read(reader, buff, count > 8192 ? 8192 : count)) > 0)
-            count -= rc;
-    } while (rc > 0 && count > 0);
-
-    if (count > 0)
-        reader->last_error = WAV_IO_ERROR;
-    return reader->last_error ? -1 : 0;
-}
-
-static
-int riff_seek(wav_reader_t *reader, int64_t off, int whence)
-{
-    int rc;
-    if (reader->last_error)
-        return -1;
-    if (!reader->io.seek)
-        goto FAIL;
-    if ((rc = reader->io.seek(reader->io_cookie, off, whence)) < 0)
-        goto FAIL;
-    return 0;
-FAIL:
-    reader->last_error = WAV_IO_ERROR;
-    return -1;
-}
-
-static
-int64_t riff_tell(wav_reader_t *reader)
-{
-    int64_t off;
-
-    if (reader->last_error || !reader->io.tell)
-        return -1;
-    off = reader->io.tell(reader->io_cookie);
-    if (off < 0) {
-        reader->last_error = WAV_IO_ERROR;
-        return -1;
-    }
-    return off;
-}
-
-static
-int riff_read16(wav_reader_t *reader, uint16_t *value)
-{
-    TRY_IO(riff_read(reader, value, 2) != 2);
-    *value = m4af_ltoh16(*value);
-    return 0;
-FAIL:
-    return -1;
-}
-
-static
-int riff_read32(wav_reader_t *reader, uint32_t *value)
-{
-    TRY_IO(riff_read(reader, value, 4) != 4);
-    *value = m4af_ltoh32(*value);
-    return 0;
-FAIL:
-    return -1;
-}
-
-static
-int riff_read64(wav_reader_t *reader, uint64_t *value)
-{
-    TRY_IO(riff_read(reader, value, 8) != 8);
-    *value = m4af_ltoh64(*value);
-    return 0;
-FAIL:
-    return -1;
-}
-
-static
-int riff_scan(wav_reader_t *reader, const char *fmt, ...)
-{
-    int c, count = 0;
-    va_list ap;
-
-    va_start(ap, fmt);
-    while ((c = *fmt++)) {
-        switch (c) {
-        case 'S':
-            TRY_IO(riff_read16(reader, va_arg(ap, uint16_t*)));
-            ++count;
-            break;
-        case 'L':
-            TRY_IO(riff_read32(reader, va_arg(ap, uint32_t*)));
-            ++count;
-            break;
-        case 'Q':
-            TRY_IO(riff_read64(reader, va_arg(ap, uint64_t*)));
-            ++count;
-            break;
-        }
-    }
-FAIL:
-    va_end(ap);
-    return count;
-}
-
 static
 uint32_t riff_next_chunk(wav_reader_t *reader, uint32_t *chunk_size)
 {
     uint32_t fcc;
-    if (riff_scan(reader, "LL", &fcc, chunk_size) == 2)
+    return (pcm_scanl(&reader->io, "LL", &fcc, chunk_size) == 2) ? fcc : 0;
-        return fcc;
-    return 0;
 }
 
-int wav_read_frames(wav_reader_t *reader, void *buffer, unsigned nframes)
+static
+int wav_read_frames(pcm_reader_t *preader, void *buffer, unsigned nframes)
 {
     int rc;
     unsigned nbytes;
+    wav_reader_t *reader = (wav_reader_t *)preader;
 
     if (!reader->ignore_length && nframes > reader->length - reader->position)
         nframes = reader->length - reader->position;
     nbytes = nframes * reader->sample_format.bytes_per_frame;
     if (nbytes) {
-        if ((rc = riff_read(reader, buffer, nbytes)) < 0)
+        if ((rc = pcm_read(&reader->io, buffer, nbytes)) < 0)
             return -1;
         nframes = rc / reader->sample_format.bytes_per_frame;
         reader->position += nframes;
@@ -237,12 +89,10 @@ int riff_ds64(wav_reader_t *reader, int64_t *length)
     uint64_t riff_size, sample_count;
 
     fcc = riff_next_chunk(reader, &chunk_size);
-    ASSERT_FORMAT(reader,
+    ENSURE(fcc == RIFF_FOURCC('d','s','6','4') && chunk_size >= 28);
-                  fcc == RIFF_FOURCC('d','s','6','4') && chunk_size >= 28);
+    TRY_IO(pcm_scanl(&reader->io, "QQQL",
-    TRY_IO(riff_scan(reader, "QQQL",
                      &riff_size, length, &sample_count, &table_size) != 4);
-    TRY_IO(riff_skip(reader, (chunk_size - 27) & ~1));
+    TRY_IO(pcm_skip(&reader->io, (chunk_size - 27) & ~1));
-    reader->data_offset += (chunk_size + 9) & ~1;
 FAIL:
     return -1;
 }
@@ -256,41 +106,34 @@ int wav_fmt(wav_reader_t *reader, uint32_t size)
     uint8_t guid[16];
     int is_float = 0;
 
-    ASSERT_FORMAT(reader, size >= 16);
+    ENSURE(size >= 16);
-    TRY_IO(riff_scan(reader, "SSLLSS", &wFormatTag, &nChannels,
+    TRY_IO(pcm_scanl(&reader->io, "SSLLSS", &wFormatTag, &nChannels,
                      &nSamplesPerSec, &nAvgBytesPerSec, &nBlockAlign,
                      &wBitsPerSample) != 6);
     wValidBitsPerSample = wBitsPerSample;
 
-    if (wFormatTag != 1 && wFormatTag != 3 && wFormatTag != 0xfffe) {
+    ENSURE(wFormatTag == 1 || wFormatTag == 3 || wFormatTag == 0xfffe);
-        reader->last_error = WAV_UNSUPPORTED_FORMAT;
+    ENSURE(nChannels && nSamplesPerSec && nAvgBytesPerSec &&
-        goto FAIL;
-    }
-    ASSERT_FORMAT(reader,
-                  nChannels && nSamplesPerSec && nAvgBytesPerSec &&
            nBlockAlign && wBitsPerSample && !(wBitsPerSample & 7) &&
            nBlockAlign == nChannels * wBitsPerSample / 8);
+
     if (wFormatTag == 3)
         is_float = 1;
 
     if (wFormatTag != 0xfffe)
-        TRY_IO(riff_skip(reader, (size - 15) & ~1));
+        TRY_IO(pcm_skip(&reader->io, (size - 15) & ~1));
     else {
-        ASSERT_FORMAT(reader, size >= 40);
+        ENSURE(size >= 40);
-        TRY_IO(riff_scan(reader, "SSL",
+        TRY_IO(pcm_scanl(&reader->io, "SSL",
                          &cbSize, &wValidBitsPerSample, &dwChannelMask) != 3);
-        TRY_IO(riff_read(reader, guid, 16) != 16);
+        TRY_IO(pcm_read(&reader->io, guid, 16) != 16);
 
         if (memcmp(guid, WAV_GUID_FLOAT, 16) == 0)
             is_float = 1;
-        else if (memcmp(guid, WAV_GUID_PCM, 16) != 0) {
+        else if (memcmp(guid, WAV_GUID_PCM, 16) != 0)
-            reader->last_error = WAV_UNSUPPORTED_FORMAT;
             goto FAIL;
-        }
+        ENSURE(wValidBitsPerSample && wValidBitsPerSample <= wBitsPerSample);
-        ASSERT_FORMAT(reader,
+        TRY_IO(pcm_skip(&reader->io, (size - 39) & ~1));
-                      wValidBitsPerSample &&
-                      wValidBitsPerSample <= wBitsPerSample);
-        TRY_IO(riff_skip(reader, (size - 39) & ~1));
     }
     reader->sample_format.sample_rate = nSamplesPerSec;
     reader->sample_format.bits_per_channel = wValidBitsPerSample;
@@ -315,13 +158,10 @@ int wav_parse(wav_reader_t *reader, int64_t *data_length)
 
     *data_length = 0;
     container = riff_next_chunk(reader, &chunk_size);
-    if (container != RIFF_FOURCC('R','I','F','F') &&
+    ENSURE(container == RIFF_FOURCC('R','I','F','F') ||
-        container != RIFF_FOURCC('R','F','6','4'))
+           container == RIFF_FOURCC('R','F','6','4'));
-        goto FAIL;
+    TRY_IO(pcm_read32le(&reader->io, &fcc));
-    TRY_IO(riff_read32(reader, &fcc));
+    ENSURE(fcc == RIFF_FOURCC('W','A','V','E'));
-    if (fcc != RIFF_FOURCC('W','A','V','E'))
-        goto FAIL;
-    reader->data_offset = 12;
 
     if (container == RIFF_FOURCC('R','F','6','4'))
         riff_ds64(reader, data_length);
@@ -332,12 +172,11 @@ int wav_parse(wav_reader_t *reader, int64_t *data_length)
         } else if (fcc == RIFF_FOURCC('d','a','t','a')) {
             if (container == RIFF_FOURCC('R','I','F','F'))
                 *data_length = chunk_size;
-            reader->data_offset += 8;
+            reader->data_offset = pcm_tell(&reader->io);
             break;
         } else {
-            TRY_IO(riff_skip(reader, (chunk_size + 1) & ~1));
+            TRY_IO(pcm_skip(&reader->io, (chunk_size + 1) & ~1));
         }
-        reader->data_offset += (chunk_size + 9) & ~1;
     }
     if (fcc == RIFF_FOURCC('d','a','t','a'))
         return 0;
@@ -345,8 +184,15 @@ FAIL:
     return -1;
 }
 
-wav_reader_t *wav_open(wav_io_context_t *io_ctx, void *io_cookie,
+static pcm_reader_vtbl_t wav_vtable = {
-                       int ignore_length)
+    wav_get_format,
+    wav_get_length,
+    wav_get_position,
+    wav_read_frames,
+    wav_teardown
+};
+
+pcm_reader_t *wav_open(pcm_io_context_t *io, int ignore_length)
 {
     wav_reader_t *reader = 0;
     int64_t data_length;
@@ -354,8 +200,7 @@ wav_reader_t *wav_open(wav_io_context_t *io_ctx, void *io_cookie,
 
     if ((reader = calloc(1, sizeof(wav_reader_t))) == 0)
         return 0;
-    memcpy(&reader->io, io_ctx, sizeof(wav_io_context_t));
+    memcpy(&reader->io, io, sizeof(pcm_io_context_t));
-    reader->io_cookie = io_cookie;
     reader->ignore_length = ignore_length;
     if (wav_parse(reader, &data_length) < 0) {
         free(reader);
@@ -367,37 +212,34 @@ wav_reader_t *wav_open(wav_io_context_t *io_ctx, void *io_cookie,
     else
         reader->length = data_length / bpf;
 
-    if (reader->length == INT64_MAX && reader->io.seek && reader->io.tell) {
+    if (reader->length == INT64_MAX) {
-        if (reader->io.seek(reader->io_cookie, 0, SEEK_END) >= 0) {
+        if (pcm_seek(&reader->io, 0, SEEK_END) >= 0) {
-            int64_t size = reader->io.tell(reader->io_cookie);
+            int64_t size = pcm_tell(&reader->io);
             if (size > 0)
                 reader->length = (size - reader->data_offset) / bpf;
-            reader->io.seek(reader->io_cookie, reader->data_offset, SEEK_SET);
+            pcm_seek(&reader->io, reader->data_offset, SEEK_SET);
         }
     }
-    return reader;
+    reader->vtbl = &wav_vtable;
+    return (pcm_reader_t *)reader;
 }
 
-wav_reader_t *raw_open(wav_io_context_t *io_ctx, void *io_cookie,
+pcm_reader_t *raw_open(pcm_io_context_t *io,
                        const pcm_sample_description_t *desc)
 {
     wav_reader_t *reader = 0;
 
     if ((reader = calloc(1, sizeof(wav_reader_t))) == 0)
         return 0;
-    memcpy(&reader->io, io_ctx, sizeof(wav_io_context_t));
+    memcpy(&reader->io, io, sizeof(pcm_io_context_t));
     memcpy(&reader->sample_format, desc, sizeof(pcm_sample_description_t));
-    reader->io_cookie = io_cookie;
+    if (pcm_seek(&reader->io, 0, SEEK_END) >= 0) {
-    if (io_ctx->seek && io_ctx->tell) {
+        int64_t size = pcm_tell(&reader->io);
-        if (reader->io.seek(reader->io_cookie, 0, SEEK_END) >= 0) {
-            int64_t size = reader->io.tell(reader->io_cookie);
         if (size > 0)
-                reader->length = size / desc->bytes_per_frame;
+            reader->length = size / reader->sample_format.bytes_per_frame;
-            reader->io.seek(reader->io_cookie, reader->data_offset, SEEK_SET);
+        pcm_seek(&reader->io, 0, SEEK_SET);
-        }
     } else
         reader->length = INT64_MAX;
-    return reader;
+    reader->vtbl = &wav_vtable;
+    return (pcm_reader_t *)reader;
 }
-
-

src/wav_reader.h

@@ -6,34 +6,10 @@
 #define WAV_READER_H
 
 #include "lpcm.h"
+#include "pcm_reader.h"
 
-enum wav_error_code {
+pcm_reader_t *wav_open(pcm_io_context_t *io, int ignore_length);
-    WAV_IO_ERROR = 1,
+pcm_reader_t *raw_open(pcm_io_context_t *io,
-    WAV_NO_MEMORY,
-    WAV_INVALID_FORMAT,
-    WAV_UNSUPPORTED_FORMAT
-};
-
-typedef int (*wav_read_callback)(void *cookie, void *data, uint32_t size);
-typedef int (*wav_seek_callback)(void *cookie, int64_t off, int whence);
-typedef int64_t (*wav_tell_callback)(void *cookie);
-
-typedef struct wav_io_context_t {
-    wav_read_callback read;
-    wav_seek_callback seek;
-    wav_tell_callback tell;
-} wav_io_context_t;
-
-typedef struct wav_reader_t wav_reader_t;
-
-wav_reader_t *wav_open(wav_io_context_t *io_ctx, void *io_cookie,
-                       int ignore_length);
-wav_reader_t *raw_open(wav_io_context_t *io_ctx, void *io_cookie,
                        const pcm_sample_description_t *desc);
-const pcm_sample_description_t *wav_get_format(wav_reader_t *reader);
-int wav_read_frames(wav_reader_t *reader, void *buffer, unsigned nframes);
-int64_t wav_get_length(wav_reader_t *reader);
-int64_t wav_get_position(wav_reader_t *reader);
-void wav_teardown(wav_reader_t **reader);
 
 #endif

version.h

@@ -1,4 +1,4 @@
 #ifndef VERSION_H
 #define VERSION_H
-const char *fdkaac_version = "0.3.0";
+const char *fdkaac_version = "0.5.2";
 #endif