From: Kristian Høgsberg Date: Fri, 25 May 2012 15:59:53 +0000 (-0400) Subject: Import vpxenc from libvpx as a starting point X-Git-Tag: 0.94.90~258 X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=89060c0a8bc78f6b4ccd3853471cfa3dbee6d804;p=platform%2Fupstream%2Fweston.git Import vpxenc from libvpx as a starting point --- diff --git a/Makefile.am b/Makefile.am index 0d8e231..fbc96e6 100644 --- a/Makefile.am +++ b/Makefile.am @@ -1,3 +1,3 @@ -SUBDIRS = shared src clients data protocol tests +SUBDIRS = shared src clients wcap-decode data protocol tests DISTCHECK_CONFIGURE_FLAGS = --disable-setuid-install diff --git a/configure.ac b/configure.ac index d4e596c..19bceef 100644 --- a/configure.ac +++ b/configure.ac @@ -27,6 +27,8 @@ AC_SUBST(DLOPEN_LIBS) PKG_CHECK_MODULES(COMPOSITOR, [wayland-server egl >= 7.10 glesv2 xkbcommon pixman-1]) +PKG_CHECK_MODULES(WCAP, [cairo vpx]) + AC_CHECK_PROG(RSVG_CONVERT, rsvg-convert, rsvg-convert) AM_CONDITIONAL(HAVE_RSVG_CONVERT, test -n "$RSVG_CONVERT") @@ -196,6 +198,7 @@ AC_CONFIG_FILES([Makefile src/Makefile src/xwayland/Makefile clients/Makefile + wcap-decode/Makefile data/Makefile protocol/Makefile tests/Makefile]) diff --git a/wcap-decode/EbmlIDs.h b/wcap-decode/EbmlIDs.h new file mode 100644 index 0000000..3418e36 --- /dev/null +++ b/wcap-decode/EbmlIDs.h @@ -0,0 +1,231 @@ +// Copyright (c) 2010 The WebM project authors. All Rights Reserved. +// +// Use of this source code is governed by a BSD-style license +// that can be found in the LICENSE file in the root of the source +// tree. An additional intellectual property rights grant can be found +// in the file PATENTS. All contributing project authors may +// be found in the AUTHORS file in the root of the source tree. + + +#ifndef MKV_DEFS_HPP +#define MKV_DEFS_HPP 1 + +//Commenting out values not available in webm, but available in matroska + +enum mkv +{ + EBML = 0x1A45DFA3, + EBMLVersion = 0x4286, + EBMLReadVersion = 0x42F7, + EBMLMaxIDLength = 0x42F2, + EBMLMaxSizeLength = 0x42F3, + DocType = 0x4282, + DocTypeVersion = 0x4287, + DocTypeReadVersion = 0x4285, +// CRC_32 = 0xBF, + Void = 0xEC, + SignatureSlot = 0x1B538667, + SignatureAlgo = 0x7E8A, + SignatureHash = 0x7E9A, + SignaturePublicKey = 0x7EA5, + Signature = 0x7EB5, + SignatureElements = 0x7E5B, + SignatureElementList = 0x7E7B, + SignedElement = 0x6532, + //segment + Segment = 0x18538067, + //Meta Seek Information + SeekHead = 0x114D9B74, + Seek = 0x4DBB, + SeekID = 0x53AB, + SeekPosition = 0x53AC, + //Segment Information + Info = 0x1549A966, +// SegmentUID = 0x73A4, +// SegmentFilename = 0x7384, +// PrevUID = 0x3CB923, +// PrevFilename = 0x3C83AB, +// NextUID = 0x3EB923, +// NextFilename = 0x3E83BB, +// SegmentFamily = 0x4444, +// ChapterTranslate = 0x6924, +// ChapterTranslateEditionUID = 0x69FC, +// ChapterTranslateCodec = 0x69BF, +// ChapterTranslateID = 0x69A5, + TimecodeScale = 0x2AD7B1, + Segment_Duration = 0x4489, + DateUTC = 0x4461, +// Title = 0x7BA9, + MuxingApp = 0x4D80, + WritingApp = 0x5741, + //Cluster + Cluster = 0x1F43B675, + Timecode = 0xE7, +// SilentTracks = 0x5854, +// SilentTrackNumber = 0x58D7, +// Position = 0xA7, + PrevSize = 0xAB, + BlockGroup = 0xA0, + Block = 0xA1, +// BlockVirtual = 0xA2, +// BlockAdditions = 0x75A1, +// BlockMore = 0xA6, +// BlockAddID = 0xEE, +// BlockAdditional = 0xA5, + BlockDuration = 0x9B, +// ReferencePriority = 0xFA, + ReferenceBlock = 0xFB, +// ReferenceVirtual = 0xFD, +// CodecState = 0xA4, +// Slices = 0x8E, +// TimeSlice = 0xE8, + LaceNumber = 0xCC, +// FrameNumber = 0xCD, +// BlockAdditionID = 0xCB, +// MkvDelay = 0xCE, +// Cluster_Duration = 0xCF, + SimpleBlock = 0xA3, +// EncryptedBlock = 0xAF, + //Track + Tracks = 0x1654AE6B, + TrackEntry = 0xAE, + TrackNumber = 0xD7, + TrackUID = 0x73C5, + TrackType = 0x83, + FlagEnabled = 0xB9, + FlagDefault = 0x88, + FlagForced = 0x55AA, + FlagLacing = 0x9C, +// MinCache = 0x6DE7, +// MaxCache = 0x6DF8, + DefaultDuration = 0x23E383, +// TrackTimecodeScale = 0x23314F, +// TrackOffset = 0x537F, +// MaxBlockAdditionID = 0x55EE, + Name = 0x536E, + Language = 0x22B59C, + CodecID = 0x86, + CodecPrivate = 0x63A2, + CodecName = 0x258688, +// AttachmentLink = 0x7446, +// CodecSettings = 0x3A9697, +// CodecInfoURL = 0x3B4040, +// CodecDownloadURL = 0x26B240, +// CodecDecodeAll = 0xAA, +// TrackOverlay = 0x6FAB, +// TrackTranslate = 0x6624, +// TrackTranslateEditionUID = 0x66FC, +// TrackTranslateCodec = 0x66BF, +// TrackTranslateTrackID = 0x66A5, + //video + Video = 0xE0, + FlagInterlaced = 0x9A, + StereoMode = 0x53B8, + PixelWidth = 0xB0, + PixelHeight = 0xBA, + PixelCropBottom = 0x54AA, + PixelCropTop = 0x54BB, + PixelCropLeft = 0x54CC, + PixelCropRight = 0x54DD, + DisplayWidth = 0x54B0, + DisplayHeight = 0x54BA, + DisplayUnit = 0x54B2, + AspectRatioType = 0x54B3, +// ColourSpace = 0x2EB524, +// GammaValue = 0x2FB523, + FrameRate = 0x2383E3, + //end video + //audio + Audio = 0xE1, + SamplingFrequency = 0xB5, + OutputSamplingFrequency = 0x78B5, + Channels = 0x9F, +// ChannelPositions = 0x7D7B, + BitDepth = 0x6264, + //end audio + //content encoding +// ContentEncodings = 0x6d80, +// ContentEncoding = 0x6240, +// ContentEncodingOrder = 0x5031, +// ContentEncodingScope = 0x5032, +// ContentEncodingType = 0x5033, +// ContentCompression = 0x5034, +// ContentCompAlgo = 0x4254, +// ContentCompSettings = 0x4255, +// ContentEncryption = 0x5035, +// ContentEncAlgo = 0x47e1, +// ContentEncKeyID = 0x47e2, +// ContentSignature = 0x47e3, +// ContentSigKeyID = 0x47e4, +// ContentSigAlgo = 0x47e5, +// ContentSigHashAlgo = 0x47e6, + //end content encoding + //Cueing Data + Cues = 0x1C53BB6B, + CuePoint = 0xBB, + CueTime = 0xB3, + CueTrackPositions = 0xB7, + CueTrack = 0xF7, + CueClusterPosition = 0xF1, + CueBlockNumber = 0x5378, +// CueCodecState = 0xEA, +// CueReference = 0xDB, +// CueRefTime = 0x96, +// CueRefCluster = 0x97, +// CueRefNumber = 0x535F, +// CueRefCodecState = 0xEB, + //Attachment +// Attachments = 0x1941A469, +// AttachedFile = 0x61A7, +// FileDescription = 0x467E, +// FileName = 0x466E, +// FileMimeType = 0x4660, +// FileData = 0x465C, +// FileUID = 0x46AE, +// FileReferral = 0x4675, + //Chapters +// Chapters = 0x1043A770, +// EditionEntry = 0x45B9, +// EditionUID = 0x45BC, +// EditionFlagHidden = 0x45BD, +// EditionFlagDefault = 0x45DB, +// EditionFlagOrdered = 0x45DD, +// ChapterAtom = 0xB6, +// ChapterUID = 0x73C4, +// ChapterTimeStart = 0x91, +// ChapterTimeEnd = 0x92, +// ChapterFlagHidden = 0x98, +// ChapterFlagEnabled = 0x4598, +// ChapterSegmentUID = 0x6E67, +// ChapterSegmentEditionUID = 0x6EBC, +// ChapterPhysicalEquiv = 0x63C3, +// ChapterTrack = 0x8F, +// ChapterTrackNumber = 0x89, +// ChapterDisplay = 0x80, +// ChapString = 0x85, +// ChapLanguage = 0x437C, +// ChapCountry = 0x437E, +// ChapProcess = 0x6944, +// ChapProcessCodecID = 0x6955, +// ChapProcessPrivate = 0x450D, +// ChapProcessCommand = 0x6911, +// ChapProcessTime = 0x6922, +// ChapProcessData = 0x6933, + //Tagging +// Tags = 0x1254C367, +// Tag = 0x7373, +// Targets = 0x63C0, +// TargetTypeValue = 0x68CA, +// TargetType = 0x63CA, +// Tagging_TrackUID = 0x63C5, +// Tagging_EditionUID = 0x63C9, +// Tagging_ChapterUID = 0x63C4, +// AttachmentUID = 0x63C6, +// SimpleTag = 0x67C8, +// TagName = 0x45A3, +// TagLanguage = 0x447A, +// TagDefault = 0x4484, +// TagString = 0x4487, +// TagBinary = 0x4485, +}; +#endif diff --git a/wcap-decode/EbmlWriter.c b/wcap-decode/EbmlWriter.c new file mode 100644 index 0000000..fbf2c66 --- /dev/null +++ b/wcap-decode/EbmlWriter.c @@ -0,0 +1,171 @@ +// Copyright (c) 2010 The WebM project authors. All Rights Reserved. +// +// Use of this source code is governed by a BSD-style license +// that can be found in the LICENSE file in the root of the source +// tree. An additional intellectual property rights grant can be found +// in the file PATENTS. All contributing project authors may +// be found in the AUTHORS file in the root of the source tree. + + +#include "EbmlWriter.h" +#include +#include +#include +#include +#if defined(_MSC_VER) +#define LITERALU64(n) n +#else +#define LITERALU64(n) n##LLU +#endif + +void Ebml_WriteLen(EbmlGlobal *glob, long long val) +{ + //TODO check and make sure we are not > than 0x0100000000000000LLU + unsigned char size = 8; //size in bytes to output + unsigned long long minVal = LITERALU64(0x00000000000000ff); //mask to compare for byte size + + for (size = 1; size < 8; size ++) + { + if (val < minVal) + break; + + minVal = (minVal << 7); + } + + val |= (LITERALU64(0x000000000000080) << ((size - 1) * 7)); + + Ebml_Serialize(glob, (void *) &val, sizeof(val), size); +} + +void Ebml_WriteString(EbmlGlobal *glob, const char *str) +{ + const size_t size_ = strlen(str); + const unsigned long long size = size_; + Ebml_WriteLen(glob, size); + //TODO: it's not clear from the spec whether the nul terminator + //should be serialized too. For now we omit the null terminator. + Ebml_Write(glob, str, size); +} + +void Ebml_WriteUTF8(EbmlGlobal *glob, const wchar_t *wstr) +{ + const size_t strlen = wcslen(wstr); + + //TODO: it's not clear from the spec whether the nul terminator + //should be serialized too. For now we include it. + const unsigned long long size = strlen; + + Ebml_WriteLen(glob, size); + Ebml_Write(glob, wstr, size); +} + +void Ebml_WriteID(EbmlGlobal *glob, unsigned long class_id) +{ + int len; + + if (class_id >= 0x01000000) + len = 4; + else if (class_id >= 0x00010000) + len = 3; + else if (class_id >= 0x00000100) + len = 2; + else + len = 1; + + Ebml_Serialize(glob, (void *)&class_id, sizeof(class_id), len); +} + +void Ebml_SerializeUnsigned64(EbmlGlobal *glob, unsigned long class_id, uint64_t ui) +{ + unsigned char sizeSerialized = 8 | 0x80; + Ebml_WriteID(glob, class_id); + Ebml_Serialize(glob, &sizeSerialized, sizeof(sizeSerialized), 1); + Ebml_Serialize(glob, &ui, sizeof(ui), 8); +} + +void Ebml_SerializeUnsigned(EbmlGlobal *glob, unsigned long class_id, unsigned long ui) +{ + unsigned char size = 8; //size in bytes to output + unsigned char sizeSerialized = 0; + unsigned long minVal; + + Ebml_WriteID(glob, class_id); + minVal = 0x7fLU; //mask to compare for byte size + + for (size = 1; size < 4; size ++) + { + if (ui < minVal) + { + break; + } + + minVal <<= 7; + } + + sizeSerialized = 0x80 | size; + Ebml_Serialize(glob, &sizeSerialized, sizeof(sizeSerialized), 1); + Ebml_Serialize(glob, &ui, sizeof(ui), size); +} +//TODO: perhaps this is a poor name for this id serializer helper function +void Ebml_SerializeBinary(EbmlGlobal *glob, unsigned long class_id, unsigned long bin) +{ + int size; + for (size=4; size > 1; size--) + { + if (bin & 0x000000ff << ((size-1) * 8)) + break; + } + Ebml_WriteID(glob, class_id); + Ebml_WriteLen(glob, size); + Ebml_WriteID(glob, bin); +} + +void Ebml_SerializeFloat(EbmlGlobal *glob, unsigned long class_id, double d) +{ + unsigned char len = 0x88; + + Ebml_WriteID(glob, class_id); + Ebml_Serialize(glob, &len, sizeof(len), 1); + Ebml_Serialize(glob, &d, sizeof(d), 8); +} + +void Ebml_WriteSigned16(EbmlGlobal *glob, short val) +{ + signed long out = ((val & 0x003FFFFF) | 0x00200000) << 8; + Ebml_Serialize(glob, &out, sizeof(out), 3); +} + +void Ebml_SerializeString(EbmlGlobal *glob, unsigned long class_id, const char *s) +{ + Ebml_WriteID(glob, class_id); + Ebml_WriteString(glob, s); +} + +void Ebml_SerializeUTF8(EbmlGlobal *glob, unsigned long class_id, wchar_t *s) +{ + Ebml_WriteID(glob, class_id); + Ebml_WriteUTF8(glob, s); +} + +void Ebml_SerializeData(EbmlGlobal *glob, unsigned long class_id, unsigned char *data, unsigned long data_length) +{ + Ebml_WriteID(glob, class_id); + Ebml_WriteLen(glob, data_length); + Ebml_Write(glob, data, data_length); +} + +void Ebml_WriteVoid(EbmlGlobal *glob, unsigned long vSize) +{ + unsigned char tmp = 0; + unsigned long i = 0; + + Ebml_WriteID(glob, 0xEC); + Ebml_WriteLen(glob, vSize); + + for (i = 0; i < vSize; i++) + { + Ebml_Write(glob, &tmp, 1); + } +} + +//TODO Serialize Date diff --git a/wcap-decode/EbmlWriter.h b/wcap-decode/EbmlWriter.h new file mode 100644 index 0000000..324c9bc --- /dev/null +++ b/wcap-decode/EbmlWriter.h @@ -0,0 +1,38 @@ +#ifndef EBMLWRITER_HPP +#define EBMLWRITER_HPP + +// Copyright (c) 2010 The WebM project authors. All Rights Reserved. +// +// Use of this source code is governed by a BSD-style license +// that can be found in the LICENSE file in the root of the source +// tree. An additional intellectual property rights grant can be found +// in the file PATENTS. All contributing project authors may +// be found in the AUTHORS file in the root of the source tree. + +//note: you must define write and serialize functions as well as your own EBML_GLOBAL +//These functions MUST be implemented +#include +#include "vpx/vpx_integer.h" + +typedef struct EbmlGlobal EbmlGlobal; +void Ebml_Serialize(EbmlGlobal *glob, const void *, int, unsigned long); +void Ebml_Write(EbmlGlobal *glob, const void *, unsigned long); +///// + + +void Ebml_WriteLen(EbmlGlobal *glob, long long val); +void Ebml_WriteString(EbmlGlobal *glob, const char *str); +void Ebml_WriteUTF8(EbmlGlobal *glob, const wchar_t *wstr); +void Ebml_WriteID(EbmlGlobal *glob, unsigned long class_id); +void Ebml_SerializeUnsigned64(EbmlGlobal *glob, unsigned long class_id, uint64_t ui); +void Ebml_SerializeUnsigned(EbmlGlobal *glob, unsigned long class_id, unsigned long ui); +void Ebml_SerializeBinary(EbmlGlobal *glob, unsigned long class_id, unsigned long ui); +void Ebml_SerializeFloat(EbmlGlobal *glob, unsigned long class_id, double d); +//TODO make this more generic to signed +void Ebml_WriteSigned16(EbmlGlobal *glob, short val); +void Ebml_SerializeString(EbmlGlobal *glob, unsigned long class_id, const char *s); +void Ebml_SerializeUTF8(EbmlGlobal *glob, unsigned long class_id, wchar_t *s); +void Ebml_SerializeData(EbmlGlobal *glob, unsigned long class_id, unsigned char *data, unsigned long data_length); +void Ebml_WriteVoid(EbmlGlobal *glob, unsigned long vSize); +//TODO need date function +#endif diff --git a/wcap-decode/args.c b/wcap-decode/args.c new file mode 100644 index 0000000..37ba778 --- /dev/null +++ b/wcap-decode/args.c @@ -0,0 +1,265 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + + +#include +#include +#include +#include "args.h" + +#ifdef _MSC_VER +#define snprintf _snprintf +#endif + +#if defined(__GNUC__) && __GNUC__ +extern void die(const char *fmt, ...) __attribute__((noreturn)); +#else +extern void die(const char *fmt, ...); +#endif + + +struct arg arg_init(char **argv) +{ + struct arg a; + + a.argv = argv; + a.argv_step = 1; + a.name = NULL; + a.val = NULL; + a.def = NULL; + return a; +} + +int arg_match(struct arg *arg_, const struct arg_def *def, char **argv) +{ + struct arg arg; + + if (!argv[0] || argv[0][0] != '-') + return 0; + + arg = arg_init(argv); + + if (def->short_name + && strlen(arg.argv[0]) == strlen(def->short_name) + 1 + && !strcmp(arg.argv[0] + 1, def->short_name)) + { + + arg.name = arg.argv[0] + 1; + arg.val = def->has_val ? arg.argv[1] : NULL; + arg.argv_step = def->has_val ? 2 : 1; + } + else if (def->long_name) + { + const size_t name_len = strlen(def->long_name); + + if (strlen(arg.argv[0]) >= name_len + 2 + && arg.argv[0][1] == '-' + && !strncmp(arg.argv[0] + 2, def->long_name, name_len) + && (arg.argv[0][name_len+2] == '=' + || arg.argv[0][name_len+2] == '\0')) + { + + arg.name = arg.argv[0] + 2; + arg.val = arg.name[name_len] == '=' ? arg.name + name_len + 1 : NULL; + arg.argv_step = 1; + } + } + + if (arg.name && !arg.val && def->has_val) + die("Error: option %s requires argument.\n", arg.name); + + if (arg.name && arg.val && !def->has_val) + die("Error: option %s requires no argument.\n", arg.name); + + if (arg.name + && (arg.val || !def->has_val)) + { + arg.def = def; + *arg_ = arg; + return 1; + } + + return 0; +} + + +const char *arg_next(struct arg *arg) +{ + if (arg->argv[0]) + arg->argv += arg->argv_step; + + return *arg->argv; +} + + +char **argv_dup(int argc, const char **argv) +{ + char **new_argv = malloc((argc + 1) * sizeof(*argv)); + + memcpy(new_argv, argv, argc * sizeof(*argv)); + new_argv[argc] = NULL; + return new_argv; +} + + +void arg_show_usage(FILE *fp, const struct arg_def *const *defs) +{ + char option_text[40] = {0}; + + for (; *defs; defs++) + { + const struct arg_def *def = *defs; + char *short_val = def->has_val ? " " : ""; + char *long_val = def->has_val ? "=" : ""; + + if (def->short_name && def->long_name) + { + char *comma = def->has_val ? "," : ", "; + + snprintf(option_text, 37, "-%s%s%s --%s%6s", + def->short_name, short_val, comma, + def->long_name, long_val); + } + else if (def->short_name) + snprintf(option_text, 37, "-%s%s", + def->short_name, short_val); + else if (def->long_name) + snprintf(option_text, 37, " --%s%s", + def->long_name, long_val); + + fprintf(fp, " %-37s\t%s\n", option_text, def->desc); + + if(def->enums) + { + const struct arg_enum_list *listptr; + + fprintf(fp, " %-37s\t ", ""); + + for(listptr = def->enums; listptr->name; listptr++) + fprintf(fp, "%s%s", listptr->name, + listptr[1].name ? ", " : "\n"); + } + } +} + + +unsigned int arg_parse_uint(const struct arg *arg) +{ + long int rawval; + char *endptr; + + rawval = strtol(arg->val, &endptr, 10); + + if (arg->val[0] != '\0' && endptr[0] == '\0') + { + if (rawval >= 0 && rawval <= UINT_MAX) + return rawval; + + die("Option %s: Value %ld out of range for unsigned int\n", + arg->name, rawval); + } + + die("Option %s: Invalid character '%c'\n", arg->name, *endptr); + return 0; +} + + +int arg_parse_int(const struct arg *arg) +{ + long int rawval; + char *endptr; + + rawval = strtol(arg->val, &endptr, 10); + + if (arg->val[0] != '\0' && endptr[0] == '\0') + { + if (rawval >= INT_MIN && rawval <= INT_MAX) + return rawval; + + die("Option %s: Value %ld out of range for signed int\n", + arg->name, rawval); + } + + die("Option %s: Invalid character '%c'\n", arg->name, *endptr); + return 0; +} + + +struct vpx_rational +{ + int num; /**< fraction numerator */ + int den; /**< fraction denominator */ +}; +struct vpx_rational arg_parse_rational(const struct arg *arg) +{ + long int rawval; + char *endptr; + struct vpx_rational rat; + + /* parse numerator */ + rawval = strtol(arg->val, &endptr, 10); + + if (arg->val[0] != '\0' && endptr[0] == '/') + { + if (rawval >= INT_MIN && rawval <= INT_MAX) + rat.num = rawval; + else die("Option %s: Value %ld out of range for signed int\n", + arg->name, rawval); + } + else die("Option %s: Expected / at '%c'\n", arg->name, *endptr); + + /* parse denominator */ + rawval = strtol(endptr + 1, &endptr, 10); + + if (arg->val[0] != '\0' && endptr[0] == '\0') + { + if (rawval >= INT_MIN && rawval <= INT_MAX) + rat.den = rawval; + else die("Option %s: Value %ld out of range for signed int\n", + arg->name, rawval); + } + else die("Option %s: Invalid character '%c'\n", arg->name, *endptr); + + return rat; +} + + +int arg_parse_enum(const struct arg *arg) +{ + const struct arg_enum_list *listptr; + long int rawval; + char *endptr; + + /* First see if the value can be parsed as a raw value */ + rawval = strtol(arg->val, &endptr, 10); + if (arg->val[0] != '\0' && endptr[0] == '\0') + { + /* Got a raw value, make sure it's valid */ + for(listptr = arg->def->enums; listptr->name; listptr++) + if(listptr->val == rawval) + return rawval; + } + + /* Next see if it can be parsed as a string */ + for(listptr = arg->def->enums; listptr->name; listptr++) + if(!strcmp(arg->val, listptr->name)) + return listptr->val; + + die("Option %s: Invalid value '%s'\n", arg->name, arg->val); + return 0; +} + + +int arg_parse_enum_or_int(const struct arg *arg) +{ + if(arg->def->enums) + return arg_parse_enum(arg); + return arg_parse_int(arg); +} diff --git a/wcap-decode/args.h b/wcap-decode/args.h new file mode 100644 index 0000000..7963fa6 --- /dev/null +++ b/wcap-decode/args.h @@ -0,0 +1,54 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + + +#ifndef ARGS_H +#define ARGS_H +#include + +struct arg +{ + char **argv; + const char *name; + const char *val; + unsigned int argv_step; + const struct arg_def *def; +}; + +struct arg_enum_list +{ + const char *name; + int val; +}; +#define ARG_ENUM_LIST_END {0} + +typedef struct arg_def +{ + const char *short_name; + const char *long_name; + int has_val; + const char *desc; + const struct arg_enum_list *enums; +} arg_def_t; +#define ARG_DEF(s,l,v,d) {s,l,v,d, NULL} +#define ARG_DEF_ENUM(s,l,v,d,e) {s,l,v,d,e} +#define ARG_DEF_LIST_END {0} + +struct arg arg_init(char **argv); +int arg_match(struct arg *arg_, const struct arg_def *def, char **argv); +const char *arg_next(struct arg *arg); +void arg_show_usage(FILE *fp, const struct arg_def *const *defs); +char **argv_dup(int argc, const char **argv); + +unsigned int arg_parse_uint(const struct arg *arg); +int arg_parse_int(const struct arg *arg); +struct vpx_rational arg_parse_rational(const struct arg *arg); +int arg_parse_enum_or_int(const struct arg *arg); +#endif diff --git a/wcap-decode/mem_ops.h b/wcap-decode/mem_ops.h new file mode 100644 index 0000000..0e52368 --- /dev/null +++ b/wcap-decode/mem_ops.h @@ -0,0 +1,234 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + + +/* \file + * \brief Provides portable memory access primitives + * + * This function provides portable primitives for getting and setting of + * signed and unsigned integers in 16, 24, and 32 bit sizes. The operations + * can be performed on unaligned data regardless of hardware support for + * unaligned accesses. + * + * The type used to pass the integral values may be changed by defining + * MEM_VALUE_T with the appropriate type. The type given must be an integral + * numeric type. + * + * The actual functions instantiated have the MEM_VALUE_T type name pasted + * on to the symbol name. This allows the developer to instantiate these + * operations for multiple types within the same translation unit. This is + * of somewhat questionable utility, but the capability exists nonetheless. + * Users not making use of this functionality should call the functions + * without the type name appended, and the preprocessor will take care of + * it. + * + * NOTE: This code is not supported on platforms where char > 1 octet ATM. + */ + +#ifndef MAU_T +/* Minimum Access Unit for this target */ +#define MAU_T unsigned char +#endif + +#ifndef MEM_VALUE_T +#define MEM_VALUE_T int +#endif + +#undef MEM_VALUE_T_SZ_BITS +#define MEM_VALUE_T_SZ_BITS (sizeof(MEM_VALUE_T) << 3) + +#undef mem_ops_wrap_symbol +#define mem_ops_wrap_symbol(fn) mem_ops_wrap_symbol2(fn, MEM_VALUE_T) +#undef mem_ops_wrap_symbol2 +#define mem_ops_wrap_symbol2(fn,typ) mem_ops_wrap_symbol3(fn,typ) +#undef mem_ops_wrap_symbol3 +#define mem_ops_wrap_symbol3(fn,typ) fn##_as_##typ + +/* + * Include aligned access routines + */ +#define INCLUDED_BY_MEM_OPS_H +#include "mem_ops_aligned.h" +#undef INCLUDED_BY_MEM_OPS_H + +#undef mem_get_be16 +#define mem_get_be16 mem_ops_wrap_symbol(mem_get_be16) +static unsigned MEM_VALUE_T mem_get_be16(const void *vmem) +{ + unsigned MEM_VALUE_T val; + const MAU_T *mem = (const MAU_T *)vmem; + + val = mem[0] << 8; + val |= mem[1]; + return val; +} + +#undef mem_get_be24 +#define mem_get_be24 mem_ops_wrap_symbol(mem_get_be24) +static unsigned MEM_VALUE_T mem_get_be24(const void *vmem) +{ + unsigned MEM_VALUE_T val; + const MAU_T *mem = (const MAU_T *)vmem; + + val = mem[0] << 16; + val |= mem[1] << 8; + val |= mem[2]; + return val; +} + +#undef mem_get_be32 +#define mem_get_be32 mem_ops_wrap_symbol(mem_get_be32) +static unsigned MEM_VALUE_T mem_get_be32(const void *vmem) +{ + unsigned MEM_VALUE_T val; + const MAU_T *mem = (const MAU_T *)vmem; + + val = mem[0] << 24; + val |= mem[1] << 16; + val |= mem[2] << 8; + val |= mem[3]; + return val; +} + +#undef mem_get_le16 +#define mem_get_le16 mem_ops_wrap_symbol(mem_get_le16) +static unsigned MEM_VALUE_T mem_get_le16(const void *vmem) +{ + unsigned MEM_VALUE_T val; + const MAU_T *mem = (const MAU_T *)vmem; + + val = mem[1] << 8; + val |= mem[0]; + return val; +} + +#undef mem_get_le24 +#define mem_get_le24 mem_ops_wrap_symbol(mem_get_le24) +static unsigned MEM_VALUE_T mem_get_le24(const void *vmem) +{ + unsigned MEM_VALUE_T val; + const MAU_T *mem = (const MAU_T *)vmem; + + val = mem[2] << 16; + val |= mem[1] << 8; + val |= mem[0]; + return val; +} + +#undef mem_get_le32 +#define mem_get_le32 mem_ops_wrap_symbol(mem_get_le32) +static unsigned MEM_VALUE_T mem_get_le32(const void *vmem) +{ + unsigned MEM_VALUE_T val; + const MAU_T *mem = (const MAU_T *)vmem; + + val = mem[3] << 24; + val |= mem[2] << 16; + val |= mem[1] << 8; + val |= mem[0]; + return val; +} + +#define mem_get_s_generic(end,sz) \ + static signed MEM_VALUE_T mem_get_s##end##sz(const void *vmem) {\ + const MAU_T *mem = (const MAU_T*)vmem;\ + signed MEM_VALUE_T val = mem_get_##end##sz(mem);\ + return (val << (MEM_VALUE_T_SZ_BITS - sz)) >> (MEM_VALUE_T_SZ_BITS - sz);\ + } + +#undef mem_get_sbe16 +#define mem_get_sbe16 mem_ops_wrap_symbol(mem_get_sbe16) +mem_get_s_generic(be, 16); + +#undef mem_get_sbe24 +#define mem_get_sbe24 mem_ops_wrap_symbol(mem_get_sbe24) +mem_get_s_generic(be, 24); + +#undef mem_get_sbe32 +#define mem_get_sbe32 mem_ops_wrap_symbol(mem_get_sbe32) +mem_get_s_generic(be, 32); + +#undef mem_get_sle16 +#define mem_get_sle16 mem_ops_wrap_symbol(mem_get_sle16) +mem_get_s_generic(le, 16); + +#undef mem_get_sle24 +#define mem_get_sle24 mem_ops_wrap_symbol(mem_get_sle24) +mem_get_s_generic(le, 24); + +#undef mem_get_sle32 +#define mem_get_sle32 mem_ops_wrap_symbol(mem_get_sle32) +mem_get_s_generic(le, 32); + +#undef mem_put_be16 +#define mem_put_be16 mem_ops_wrap_symbol(mem_put_be16) +static void mem_put_be16(void *vmem, MEM_VALUE_T val) +{ + MAU_T *mem = (MAU_T *)vmem; + + mem[0] = (val >> 8) & 0xff; + mem[1] = (val >> 0) & 0xff; +} + +#undef mem_put_be24 +#define mem_put_be24 mem_ops_wrap_symbol(mem_put_be24) +static void mem_put_be24(void *vmem, MEM_VALUE_T val) +{ + MAU_T *mem = (MAU_T *)vmem; + + mem[0] = (val >> 16) & 0xff; + mem[1] = (val >> 8) & 0xff; + mem[2] = (val >> 0) & 0xff; +} + +#undef mem_put_be32 +#define mem_put_be32 mem_ops_wrap_symbol(mem_put_be32) +static void mem_put_be32(void *vmem, MEM_VALUE_T val) +{ + MAU_T *mem = (MAU_T *)vmem; + + mem[0] = (val >> 24) & 0xff; + mem[1] = (val >> 16) & 0xff; + mem[2] = (val >> 8) & 0xff; + mem[3] = (val >> 0) & 0xff; +} + +#undef mem_put_le16 +#define mem_put_le16 mem_ops_wrap_symbol(mem_put_le16) +static void mem_put_le16(void *vmem, MEM_VALUE_T val) +{ + MAU_T *mem = (MAU_T *)vmem; + + mem[0] = (val >> 0) & 0xff; + mem[1] = (val >> 8) & 0xff; +} + +#undef mem_put_le24 +#define mem_put_le24 mem_ops_wrap_symbol(mem_put_le24) +static void mem_put_le24(void *vmem, MEM_VALUE_T val) +{ + MAU_T *mem = (MAU_T *)vmem; + + mem[0] = (val >> 0) & 0xff; + mem[1] = (val >> 8) & 0xff; + mem[2] = (val >> 16) & 0xff; +} + +#undef mem_put_le32 +#define mem_put_le32 mem_ops_wrap_symbol(mem_put_le32) +static void mem_put_le32(void *vmem, MEM_VALUE_T val) +{ + MAU_T *mem = (MAU_T *)vmem; + + mem[0] = (val >> 0) & 0xff; + mem[1] = (val >> 8) & 0xff; + mem[2] = (val >> 16) & 0xff; + mem[3] = (val >> 24) & 0xff; +} diff --git a/wcap-decode/mem_ops_aligned.h b/wcap-decode/mem_ops_aligned.h new file mode 100644 index 0000000..0fbba65 --- /dev/null +++ b/wcap-decode/mem_ops_aligned.h @@ -0,0 +1,157 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + + +/* \file + * \brief Provides portable memory access primitives for operating on aligned + * data + * + * This file is split from mem_ops.h for easier maintenance. See mem_ops.h + * for a more detailed description of these primitives. + */ +#ifndef INCLUDED_BY_MEM_OPS_H +#error Include mem_ops.h, not mem_ops_aligned.h directly. +#endif + +/* Architectures that provide instructions for doing this byte swapping + * could redefine these macros. + */ +#define swap_endian_16(val,raw) do {\ + val = ((raw>>8) & 0x00ff) \ + | ((raw<<8) & 0xff00);\ + } while(0) +#define swap_endian_32(val,raw) do {\ + val = ((raw>>24) & 0x000000ff) \ + | ((raw>>8) & 0x0000ff00) \ + | ((raw<<8) & 0x00ff0000) \ + | ((raw<<24) & 0xff000000); \ + } while(0) +#define swap_endian_16_se(val,raw) do {\ + swap_endian_16(val,raw);\ + val = ((val << 16) >> 16);\ + } while(0) +#define swap_endian_32_se(val,raw) swap_endian_32(val,raw) + +#define mem_get_ne_aligned_generic(end,sz) \ + static unsigned MEM_VALUE_T mem_get_##end##sz##_aligned(const void *vmem) {\ + const uint##sz##_t *mem = (const uint##sz##_t *)vmem;\ + return *mem;\ + } + +#define mem_get_sne_aligned_generic(end,sz) \ + static signed MEM_VALUE_T mem_get_s##end##sz##_aligned(const void *vmem) {\ + const int##sz##_t *mem = (const int##sz##_t *)vmem;\ + return *mem;\ + } + +#define mem_get_se_aligned_generic(end,sz) \ + static unsigned MEM_VALUE_T mem_get_##end##sz##_aligned(const void *vmem) {\ + const uint##sz##_t *mem = (const uint##sz##_t *)vmem;\ + unsigned MEM_VALUE_T val, raw = *mem;\ + swap_endian_##sz(val,raw);\ + return val;\ + } + +#define mem_get_sse_aligned_generic(end,sz) \ + static signed MEM_VALUE_T mem_get_s##end##sz##_aligned(const void *vmem) {\ + const int##sz##_t *mem = (const int##sz##_t *)vmem;\ + unsigned MEM_VALUE_T val, raw = *mem;\ + swap_endian_##sz##_se(val,raw);\ + return val;\ + } + +#define mem_put_ne_aligned_generic(end,sz) \ + static void mem_put_##end##sz##_aligned(void *vmem, MEM_VALUE_T val) {\ + uint##sz##_t *mem = (uint##sz##_t *)vmem;\ + *mem = (uint##sz##_t)val;\ + } + +#define mem_put_se_aligned_generic(end,sz) \ + static void mem_put_##end##sz##_aligned(void *vmem, MEM_VALUE_T val) {\ + uint##sz##_t *mem = (uint##sz##_t *)vmem, raw;\ + swap_endian_##sz(raw,val);\ + *mem = (uint##sz##_t)raw;\ + } + +#include "vpx_config.h" +#if CONFIG_BIG_ENDIAN +#define mem_get_be_aligned_generic(sz) mem_get_ne_aligned_generic(be,sz) +#define mem_get_sbe_aligned_generic(sz) mem_get_sne_aligned_generic(be,sz) +#define mem_get_le_aligned_generic(sz) mem_get_se_aligned_generic(le,sz) +#define mem_get_sle_aligned_generic(sz) mem_get_sse_aligned_generic(le,sz) +#define mem_put_be_aligned_generic(sz) mem_put_ne_aligned_generic(be,sz) +#define mem_put_le_aligned_generic(sz) mem_put_se_aligned_generic(le,sz) +#else +#define mem_get_be_aligned_generic(sz) mem_get_se_aligned_generic(be,sz) +#define mem_get_sbe_aligned_generic(sz) mem_get_sse_aligned_generic(be,sz) +#define mem_get_le_aligned_generic(sz) mem_get_ne_aligned_generic(le,sz) +#define mem_get_sle_aligned_generic(sz) mem_get_sne_aligned_generic(le,sz) +#define mem_put_be_aligned_generic(sz) mem_put_se_aligned_generic(be,sz) +#define mem_put_le_aligned_generic(sz) mem_put_ne_aligned_generic(le,sz) +#endif + +#undef mem_get_be16_aligned +#define mem_get_be16_aligned mem_ops_wrap_symbol(mem_get_be16_aligned) +mem_get_be_aligned_generic(16); + +#undef mem_get_be32_aligned +#define mem_get_be32_aligned mem_ops_wrap_symbol(mem_get_be32_aligned) +mem_get_be_aligned_generic(32); + +#undef mem_get_le16_aligned +#define mem_get_le16_aligned mem_ops_wrap_symbol(mem_get_le16_aligned) +mem_get_le_aligned_generic(16); + +#undef mem_get_le32_aligned +#define mem_get_le32_aligned mem_ops_wrap_symbol(mem_get_le32_aligned) +mem_get_le_aligned_generic(32); + +#undef mem_get_sbe16_aligned +#define mem_get_sbe16_aligned mem_ops_wrap_symbol(mem_get_sbe16_aligned) +mem_get_sbe_aligned_generic(16); + +#undef mem_get_sbe32_aligned +#define mem_get_sbe32_aligned mem_ops_wrap_symbol(mem_get_sbe32_aligned) +mem_get_sbe_aligned_generic(32); + +#undef mem_get_sle16_aligned +#define mem_get_sle16_aligned mem_ops_wrap_symbol(mem_get_sle16_aligned) +mem_get_sle_aligned_generic(16); + +#undef mem_get_sle32_aligned +#define mem_get_sle32_aligned mem_ops_wrap_symbol(mem_get_sle32_aligned) +mem_get_sle_aligned_generic(32); + +#undef mem_put_be16_aligned +#define mem_put_be16_aligned mem_ops_wrap_symbol(mem_put_be16_aligned) +mem_put_be_aligned_generic(16); + +#undef mem_put_be32_aligned +#define mem_put_be32_aligned mem_ops_wrap_symbol(mem_put_be32_aligned) +mem_put_be_aligned_generic(32); + +#undef mem_put_le16_aligned +#define mem_put_le16_aligned mem_ops_wrap_symbol(mem_put_le16_aligned) +mem_put_le_aligned_generic(16); + +#undef mem_put_le32_aligned +#define mem_put_le32_aligned mem_ops_wrap_symbol(mem_put_le32_aligned) +mem_put_le_aligned_generic(32); + +#undef mem_get_ne_aligned_generic +#undef mem_get_se_aligned_generic +#undef mem_get_sne_aligned_generic +#undef mem_get_sse_aligned_generic +#undef mem_put_ne_aligned_generic +#undef mem_put_se_aligned_generic +#undef swap_endian_16 +#undef swap_endian_32 +#undef swap_endian_16_se +#undef swap_endian_32_se diff --git a/wcap-decode/tools_common.c b/wcap-decode/tools_common.c new file mode 100644 index 0000000..6f95028 --- /dev/null +++ b/wcap-decode/tools_common.c @@ -0,0 +1,30 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ +#include +#include "tools_common.h" +#if defined(_WIN32) || defined(__OS2__) +#include +#include + +#ifdef __OS2__ +#define _setmode setmode +#define _fileno fileno +#define _O_BINARY O_BINARY +#endif +#endif + +FILE* set_binary_mode(FILE *stream) +{ + (void)stream; +#if defined(_WIN32) || defined(__OS2__) + _setmode(_fileno(stream), _O_BINARY); +#endif + return stream; +} diff --git a/wcap-decode/tools_common.h b/wcap-decode/tools_common.h new file mode 100644 index 0000000..80c9747 --- /dev/null +++ b/wcap-decode/tools_common.h @@ -0,0 +1,16 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ +#ifndef TOOLS_COMMON_H +#define TOOLS_COMMON_H + +/* Sets a stdio stream into binary mode */ +FILE* set_binary_mode(FILE *stream); + +#endif diff --git a/wcap-decode/vpx_timer.h b/wcap-decode/vpx_timer.h new file mode 100644 index 0000000..d07e086 --- /dev/null +++ b/wcap-decode/vpx_timer.h @@ -0,0 +1,120 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + + +#ifndef VPX_TIMER_H +#define VPX_TIMER_H +#include "vpx/vpx_integer.h" + +#if CONFIG_OS_SUPPORT + +#if defined(_WIN32) +/* + * Win32 specific includes + */ +#ifndef WIN32_LEAN_AND_MEAN +#define WIN32_LEAN_AND_MEAN +#endif +#include +#else +/* + * POSIX specific includes + */ +#include + +/* timersub is not provided by msys at this time. */ +#ifndef timersub +#define timersub(a, b, result) \ + do { \ + (result)->tv_sec = (a)->tv_sec - (b)->tv_sec; \ + (result)->tv_usec = (a)->tv_usec - (b)->tv_usec; \ + if ((result)->tv_usec < 0) { \ + --(result)->tv_sec; \ + (result)->tv_usec += 1000000; \ + } \ + } while (0) +#endif +#endif + + +struct vpx_usec_timer +{ +#if defined(_WIN32) + LARGE_INTEGER begin, end; +#else + struct timeval begin, end; +#endif +}; + + +static void +vpx_usec_timer_start(struct vpx_usec_timer *t) +{ +#if defined(_WIN32) + QueryPerformanceCounter(&t->begin); +#else + gettimeofday(&t->begin, NULL); +#endif +} + + +static void +vpx_usec_timer_mark(struct vpx_usec_timer *t) +{ +#if defined(_WIN32) + QueryPerformanceCounter(&t->end); +#else + gettimeofday(&t->end, NULL); +#endif +} + + +static int64_t +vpx_usec_timer_elapsed(struct vpx_usec_timer *t) +{ +#if defined(_WIN32) + LARGE_INTEGER freq, diff; + + diff.QuadPart = t->end.QuadPart - t->begin.QuadPart; + + QueryPerformanceFrequency(&freq); + return diff.QuadPart * 1000000 / freq.QuadPart; +#else + struct timeval diff; + + timersub(&t->end, &t->begin, &diff); + return diff.tv_sec * 1000000 + diff.tv_usec; +#endif +} + +#else /* CONFIG_OS_SUPPORT = 0*/ + +/* Empty timer functions if CONFIG_OS_SUPPORT = 0 */ +#ifndef timersub +#define timersub(a, b, result) +#endif + +struct vpx_usec_timer +{ + void *dummy; +}; + +static void +vpx_usec_timer_start(struct vpx_usec_timer *t) { } + +static void +vpx_usec_timer_mark(struct vpx_usec_timer *t) { } + +static long +vpx_usec_timer_elapsed(struct vpx_usec_timer *t) { return 0; } + +#endif /* CONFIG_OS_SUPPORT */ + +#endif diff --git a/wcap-decode/vpxenc.c b/wcap-decode/vpxenc.c new file mode 100644 index 0000000..b9aa8e1 --- /dev/null +++ b/wcap-decode/vpxenc.c @@ -0,0 +1,2544 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + + +/* This is a simple program that encodes YV12 files and generates ivf + * files using the new interface. + */ +#if defined(_WIN32) || !CONFIG_OS_SUPPORT +#define USE_POSIX_MMAP 0 +#else +#define USE_POSIX_MMAP 1 +#endif + +#include +#include +#include +#include +#include +#include +#include "vpx/vpx_encoder.h" +#if USE_POSIX_MMAP +#include +#include +#include +#include +#include +#endif +#include "vpx/vp8cx.h" +#include "vpx_ports/mem_ops.h" +#include "vpx_ports/vpx_timer.h" +#include "tools_common.h" +#include "y4minput.h" +#include "libmkv/EbmlWriter.h" +#include "libmkv/EbmlIDs.h" + +/* Need special handling of these functions on Windows */ +#if defined(_MSC_VER) +/* MSVS doesn't define off_t, and uses _f{seek,tell}i64 */ +typedef __int64 off_t; +#define fseeko _fseeki64 +#define ftello _ftelli64 +#elif defined(_WIN32) +/* MinGW defines off_t as long + and uses f{seek,tell}o64/off64_t for large files */ +#define fseeko fseeko64 +#define ftello ftello64 +#define off_t off64_t +#endif + +#if defined(_MSC_VER) +#define LITERALU64(n) n +#else +#define LITERALU64(n) n##LLU +#endif + +/* We should use 32-bit file operations in WebM file format + * when building ARM executable file (.axf) with RVCT */ +#if !CONFIG_OS_SUPPORT +typedef long off_t; +#define fseeko fseek +#define ftello ftell +#endif + +static const char *exec_name; + +static const struct codec_item +{ + char const *name; + const vpx_codec_iface_t *iface; + unsigned int fourcc; +} codecs[] = +{ +#if CONFIG_VP8_ENCODER + {"vp8", &vpx_codec_vp8_cx_algo, 0x30385056}, +#endif +}; + +static void usage_exit(); + +#define LOG_ERROR(label) do \ +{\ + const char *l=label;\ + va_list ap;\ + va_start(ap, fmt);\ + if(l)\ + fprintf(stderr, "%s: ", l);\ + vfprintf(stderr, fmt, ap);\ + fprintf(stderr, "\n");\ + va_end(ap);\ +} while(0) + +void die(const char *fmt, ...) +{ + LOG_ERROR(NULL); + usage_exit(); +} + + +void fatal(const char *fmt, ...) +{ + LOG_ERROR("Fatal"); + exit(EXIT_FAILURE); +} + + +void warn(const char *fmt, ...) +{ + LOG_ERROR("Warning"); +} + + +static void ctx_exit_on_error(vpx_codec_ctx_t *ctx, const char *s, ...) +{ + va_list ap; + + va_start(ap, s); + if (ctx->err) + { + const char *detail = vpx_codec_error_detail(ctx); + + vfprintf(stderr, s, ap); + fprintf(stderr, ": %s\n", vpx_codec_error(ctx)); + + if (detail) + fprintf(stderr, " %s\n", detail); + + exit(EXIT_FAILURE); + } +} + +/* This structure is used to abstract the different ways of handling + * first pass statistics. + */ +typedef struct +{ + vpx_fixed_buf_t buf; + int pass; + FILE *file; + char *buf_ptr; + size_t buf_alloc_sz; +} stats_io_t; + +int stats_open_file(stats_io_t *stats, const char *fpf, int pass) +{ + int res; + + stats->pass = pass; + + if (pass == 0) + { + stats->file = fopen(fpf, "wb"); + stats->buf.sz = 0; + stats->buf.buf = NULL, + res = (stats->file != NULL); + } + else + { +#if 0 +#elif USE_POSIX_MMAP + struct stat stat_buf; + int fd; + + fd = open(fpf, O_RDONLY); + stats->file = fdopen(fd, "rb"); + fstat(fd, &stat_buf); + stats->buf.sz = stat_buf.st_size; + stats->buf.buf = mmap(NULL, stats->buf.sz, PROT_READ, MAP_PRIVATE, + fd, 0); + res = (stats->buf.buf != NULL); +#else + size_t nbytes; + + stats->file = fopen(fpf, "rb"); + + if (fseek(stats->file, 0, SEEK_END)) + fatal("First-pass stats file must be seekable!"); + + stats->buf.sz = stats->buf_alloc_sz = ftell(stats->file); + rewind(stats->file); + + stats->buf.buf = malloc(stats->buf_alloc_sz); + + if (!stats->buf.buf) + fatal("Failed to allocate first-pass stats buffer (%lu bytes)", + (unsigned long)stats->buf_alloc_sz); + + nbytes = fread(stats->buf.buf, 1, stats->buf.sz, stats->file); + res = (nbytes == stats->buf.sz); +#endif + } + + return res; +} + +int stats_open_mem(stats_io_t *stats, int pass) +{ + int res; + stats->pass = pass; + + if (!pass) + { + stats->buf.sz = 0; + stats->buf_alloc_sz = 64 * 1024; + stats->buf.buf = malloc(stats->buf_alloc_sz); + } + + stats->buf_ptr = stats->buf.buf; + res = (stats->buf.buf != NULL); + return res; +} + + +void stats_close(stats_io_t *stats, int last_pass) +{ + if (stats->file) + { + if (stats->pass == last_pass) + { +#if 0 +#elif USE_POSIX_MMAP + munmap(stats->buf.buf, stats->buf.sz); +#else + free(stats->buf.buf); +#endif + } + + fclose(stats->file); + stats->file = NULL; + } + else + { + if (stats->pass == last_pass) + free(stats->buf.buf); + } +} + +void stats_write(stats_io_t *stats, const void *pkt, size_t len) +{ + if (stats->file) + { + if(fwrite(pkt, 1, len, stats->file)); + } + else + { + if (stats->buf.sz + len > stats->buf_alloc_sz) + { + size_t new_sz = stats->buf_alloc_sz + 64 * 1024; + char *new_ptr = realloc(stats->buf.buf, new_sz); + + if (new_ptr) + { + stats->buf_ptr = new_ptr + (stats->buf_ptr - (char *)stats->buf.buf); + stats->buf.buf = new_ptr; + stats->buf_alloc_sz = new_sz; + } + else + fatal("Failed to realloc firstpass stats buffer."); + } + + memcpy(stats->buf_ptr, pkt, len); + stats->buf.sz += len; + stats->buf_ptr += len; + } +} + +vpx_fixed_buf_t stats_get(stats_io_t *stats) +{ + return stats->buf; +} + +/* Stereo 3D packed frame format */ +typedef enum stereo_format +{ + STEREO_FORMAT_MONO = 0, + STEREO_FORMAT_LEFT_RIGHT = 1, + STEREO_FORMAT_BOTTOM_TOP = 2, + STEREO_FORMAT_TOP_BOTTOM = 3, + STEREO_FORMAT_RIGHT_LEFT = 11 +} stereo_format_t; + +enum video_file_type +{ + FILE_TYPE_RAW, + FILE_TYPE_IVF, + FILE_TYPE_Y4M +}; + +struct detect_buffer { + char buf[4]; + size_t buf_read; + size_t position; +}; + + +struct input_state +{ + char *fn; + FILE *file; + y4m_input y4m; + struct detect_buffer detect; + enum video_file_type file_type; + unsigned int w; + unsigned int h; + struct vpx_rational framerate; + int use_i420; +}; + + +#define IVF_FRAME_HDR_SZ (4+8) /* 4 byte size + 8 byte timestamp */ +static int read_frame(struct input_state *input, vpx_image_t *img) +{ + FILE *f = input->file; + enum video_file_type file_type = input->file_type; + y4m_input *y4m = &input->y4m; + struct detect_buffer *detect = &input->detect; + int plane = 0; + int shortread = 0; + + if (file_type == FILE_TYPE_Y4M) + { + if (y4m_input_fetch_frame(y4m, f, img) < 1) + return 0; + } + else + { + if (file_type == FILE_TYPE_IVF) + { + char junk[IVF_FRAME_HDR_SZ]; + + /* Skip the frame header. We know how big the frame should be. See + * write_ivf_frame_header() for documentation on the frame header + * layout. + */ + if(fread(junk, 1, IVF_FRAME_HDR_SZ, f)); + } + + for (plane = 0; plane < 3; plane++) + { + unsigned char *ptr; + int w = (plane ? (1 + img->d_w) / 2 : img->d_w); + int h = (plane ? (1 + img->d_h) / 2 : img->d_h); + int r; + + /* Determine the correct plane based on the image format. The for-loop + * always counts in Y,U,V order, but this may not match the order of + * the data on disk. + */ + switch (plane) + { + case 1: + ptr = img->planes[img->fmt==VPX_IMG_FMT_YV12? VPX_PLANE_V : VPX_PLANE_U]; + break; + case 2: + ptr = img->planes[img->fmt==VPX_IMG_FMT_YV12?VPX_PLANE_U : VPX_PLANE_V]; + break; + default: + ptr = img->planes[plane]; + } + + for (r = 0; r < h; r++) + { + size_t needed = w; + size_t buf_position = 0; + const size_t left = detect->buf_read - detect->position; + if (left > 0) + { + const size_t more = (left < needed) ? left : needed; + memcpy(ptr, detect->buf + detect->position, more); + buf_position = more; + needed -= more; + detect->position += more; + } + if (needed > 0) + { + shortread |= (fread(ptr + buf_position, 1, needed, f) < needed); + } + + ptr += img->stride[plane]; + } + } + } + + return !shortread; +} + + +unsigned int file_is_y4m(FILE *infile, + y4m_input *y4m, + char detect[4]) +{ + if(memcmp(detect, "YUV4", 4) == 0) + { + return 1; + } + return 0; +} + +#define IVF_FILE_HDR_SZ (32) +unsigned int file_is_ivf(struct input_state *input, + unsigned int *fourcc) +{ + char raw_hdr[IVF_FILE_HDR_SZ]; + int is_ivf = 0; + FILE *infile = input->file; + unsigned int *width = &input->w; + unsigned int *height = &input->h; + struct detect_buffer *detect = &input->detect; + + if(memcmp(detect->buf, "DKIF", 4) != 0) + return 0; + + /* See write_ivf_file_header() for more documentation on the file header + * layout. + */ + if (fread(raw_hdr + 4, 1, IVF_FILE_HDR_SZ - 4, infile) + == IVF_FILE_HDR_SZ - 4) + { + { + is_ivf = 1; + + if (mem_get_le16(raw_hdr + 4) != 0) + warn("Unrecognized IVF version! This file may not decode " + "properly."); + + *fourcc = mem_get_le32(raw_hdr + 8); + } + } + + if (is_ivf) + { + *width = mem_get_le16(raw_hdr + 12); + *height = mem_get_le16(raw_hdr + 14); + detect->position = 4; + } + + return is_ivf; +} + + +static void write_ivf_file_header(FILE *outfile, + const vpx_codec_enc_cfg_t *cfg, + unsigned int fourcc, + int frame_cnt) +{ + char header[32]; + + if (cfg->g_pass != VPX_RC_ONE_PASS && cfg->g_pass != VPX_RC_LAST_PASS) + return; + + header[0] = 'D'; + header[1] = 'K'; + header[2] = 'I'; + header[3] = 'F'; + mem_put_le16(header + 4, 0); /* version */ + mem_put_le16(header + 6, 32); /* headersize */ + mem_put_le32(header + 8, fourcc); /* headersize */ + mem_put_le16(header + 12, cfg->g_w); /* width */ + mem_put_le16(header + 14, cfg->g_h); /* height */ + mem_put_le32(header + 16, cfg->g_timebase.den); /* rate */ + mem_put_le32(header + 20, cfg->g_timebase.num); /* scale */ + mem_put_le32(header + 24, frame_cnt); /* length */ + mem_put_le32(header + 28, 0); /* unused */ + + if(fwrite(header, 1, 32, outfile)); +} + + +static void write_ivf_frame_header(FILE *outfile, + const vpx_codec_cx_pkt_t *pkt) +{ + char header[12]; + vpx_codec_pts_t pts; + + if (pkt->kind != VPX_CODEC_CX_FRAME_PKT) + return; + + pts = pkt->data.frame.pts; + mem_put_le32(header, pkt->data.frame.sz); + mem_put_le32(header + 4, pts & 0xFFFFFFFF); + mem_put_le32(header + 8, pts >> 32); + + if(fwrite(header, 1, 12, outfile)); +} + +static void write_ivf_frame_size(FILE *outfile, size_t size) +{ + char header[4]; + mem_put_le32(header, size); + fwrite(header, 1, 4, outfile); +} + + +typedef off_t EbmlLoc; + + +struct cue_entry +{ + unsigned int time; + uint64_t loc; +}; + + +struct EbmlGlobal +{ + int debug; + + FILE *stream; + int64_t last_pts_ms; + vpx_rational_t framerate; + + /* These pointers are to the start of an element */ + off_t position_reference; + off_t seek_info_pos; + off_t segment_info_pos; + off_t track_pos; + off_t cue_pos; + off_t cluster_pos; + + /* This pointer is to a specific element to be serialized */ + off_t track_id_pos; + + /* These pointers are to the size field of the element */ + EbmlLoc startSegment; + EbmlLoc startCluster; + + uint32_t cluster_timecode; + int cluster_open; + + struct cue_entry *cue_list; + unsigned int cues; + +}; + + +void Ebml_Write(EbmlGlobal *glob, const void *buffer_in, unsigned long len) +{ + if(fwrite(buffer_in, 1, len, glob->stream)); +} + +#define WRITE_BUFFER(s) \ +for(i = len-1; i>=0; i--)\ +{ \ + x = *(const s *)buffer_in >> (i * CHAR_BIT); \ + Ebml_Write(glob, &x, 1); \ +} +void Ebml_Serialize(EbmlGlobal *glob, const void *buffer_in, int buffer_size, unsigned long len) +{ + char x; + int i; + + /* buffer_size: + * 1 - int8_t; + * 2 - int16_t; + * 3 - int32_t; + * 4 - int64_t; + */ + switch (buffer_size) + { + case 1: + WRITE_BUFFER(int8_t) + break; + case 2: + WRITE_BUFFER(int16_t) + break; + case 4: + WRITE_BUFFER(int32_t) + break; + case 8: + WRITE_BUFFER(int64_t) + break; + default: + break; + } +} +#undef WRITE_BUFFER + +/* Need a fixed size serializer for the track ID. libmkv provides a 64 bit + * one, but not a 32 bit one. + */ +static void Ebml_SerializeUnsigned32(EbmlGlobal *glob, unsigned long class_id, uint64_t ui) +{ + unsigned char sizeSerialized = 4 | 0x80; + Ebml_WriteID(glob, class_id); + Ebml_Serialize(glob, &sizeSerialized, sizeof(sizeSerialized), 1); + Ebml_Serialize(glob, &ui, sizeof(ui), 4); +} + + +static void +Ebml_StartSubElement(EbmlGlobal *glob, EbmlLoc *ebmlLoc, + unsigned long class_id) +{ + //todo this is always taking 8 bytes, this may need later optimization + //this is a key that says length unknown + uint64_t unknownLen = LITERALU64(0x01FFFFFFFFFFFFFF); + + Ebml_WriteID(glob, class_id); + *ebmlLoc = ftello(glob->stream); + Ebml_Serialize(glob, &unknownLen, sizeof(unknownLen), 8); +} + +static void +Ebml_EndSubElement(EbmlGlobal *glob, EbmlLoc *ebmlLoc) +{ + off_t pos; + uint64_t size; + + /* Save the current stream pointer */ + pos = ftello(glob->stream); + + /* Calculate the size of this element */ + size = pos - *ebmlLoc - 8; + size |= LITERALU64(0x0100000000000000); + + /* Seek back to the beginning of the element and write the new size */ + fseeko(glob->stream, *ebmlLoc, SEEK_SET); + Ebml_Serialize(glob, &size, sizeof(size), 8); + + /* Reset the stream pointer */ + fseeko(glob->stream, pos, SEEK_SET); +} + + +static void +write_webm_seek_element(EbmlGlobal *ebml, unsigned long id, off_t pos) +{ + uint64_t offset = pos - ebml->position_reference; + EbmlLoc start; + Ebml_StartSubElement(ebml, &start, Seek); + Ebml_SerializeBinary(ebml, SeekID, id); + Ebml_SerializeUnsigned64(ebml, SeekPosition, offset); + Ebml_EndSubElement(ebml, &start); +} + + +static void +write_webm_seek_info(EbmlGlobal *ebml) +{ + + off_t pos; + + /* Save the current stream pointer */ + pos = ftello(ebml->stream); + + if(ebml->seek_info_pos) + fseeko(ebml->stream, ebml->seek_info_pos, SEEK_SET); + else + ebml->seek_info_pos = pos; + + { + EbmlLoc start; + + Ebml_StartSubElement(ebml, &start, SeekHead); + write_webm_seek_element(ebml, Tracks, ebml->track_pos); + write_webm_seek_element(ebml, Cues, ebml->cue_pos); + write_webm_seek_element(ebml, Info, ebml->segment_info_pos); + Ebml_EndSubElement(ebml, &start); + } + { + //segment info + EbmlLoc startInfo; + uint64_t frame_time; + char version_string[64]; + + /* Assemble version string */ + if(ebml->debug) + strcpy(version_string, "vpxenc"); + else + { + strcpy(version_string, "vpxenc "); + strncat(version_string, + vpx_codec_version_str(), + sizeof(version_string) - 1 - strlen(version_string)); + } + + frame_time = (uint64_t)1000 * ebml->framerate.den + / ebml->framerate.num; + ebml->segment_info_pos = ftello(ebml->stream); + Ebml_StartSubElement(ebml, &startInfo, Info); + Ebml_SerializeUnsigned(ebml, TimecodeScale, 1000000); + Ebml_SerializeFloat(ebml, Segment_Duration, + ebml->last_pts_ms + frame_time); + Ebml_SerializeString(ebml, 0x4D80, version_string); + Ebml_SerializeString(ebml, 0x5741, version_string); + Ebml_EndSubElement(ebml, &startInfo); + } +} + + +static void +write_webm_file_header(EbmlGlobal *glob, + const vpx_codec_enc_cfg_t *cfg, + const struct vpx_rational *fps, + stereo_format_t stereo_fmt) +{ + { + EbmlLoc start; + Ebml_StartSubElement(glob, &start, EBML); + Ebml_SerializeUnsigned(glob, EBMLVersion, 1); + Ebml_SerializeUnsigned(glob, EBMLReadVersion, 1); //EBML Read Version + Ebml_SerializeUnsigned(glob, EBMLMaxIDLength, 4); //EBML Max ID Length + Ebml_SerializeUnsigned(glob, EBMLMaxSizeLength, 8); //EBML Max Size Length + Ebml_SerializeString(glob, DocType, "webm"); //Doc Type + Ebml_SerializeUnsigned(glob, DocTypeVersion, 2); //Doc Type Version + Ebml_SerializeUnsigned(glob, DocTypeReadVersion, 2); //Doc Type Read Version + Ebml_EndSubElement(glob, &start); + } + { + Ebml_StartSubElement(glob, &glob->startSegment, Segment); //segment + glob->position_reference = ftello(glob->stream); + glob->framerate = *fps; + write_webm_seek_info(glob); + + { + EbmlLoc trackStart; + glob->track_pos = ftello(glob->stream); + Ebml_StartSubElement(glob, &trackStart, Tracks); + { + unsigned int trackNumber = 1; + uint64_t trackID = 0; + + EbmlLoc start; + Ebml_StartSubElement(glob, &start, TrackEntry); + Ebml_SerializeUnsigned(glob, TrackNumber, trackNumber); + glob->track_id_pos = ftello(glob->stream); + Ebml_SerializeUnsigned32(glob, TrackUID, trackID); + Ebml_SerializeUnsigned(glob, TrackType, 1); //video is always 1 + Ebml_SerializeString(glob, CodecID, "V_VP8"); + { + unsigned int pixelWidth = cfg->g_w; + unsigned int pixelHeight = cfg->g_h; + float frameRate = (float)fps->num/(float)fps->den; + + EbmlLoc videoStart; + Ebml_StartSubElement(glob, &videoStart, Video); + Ebml_SerializeUnsigned(glob, PixelWidth, pixelWidth); + Ebml_SerializeUnsigned(glob, PixelHeight, pixelHeight); + Ebml_SerializeUnsigned(glob, StereoMode, stereo_fmt); + Ebml_SerializeFloat(glob, FrameRate, frameRate); + Ebml_EndSubElement(glob, &videoStart); //Video + } + Ebml_EndSubElement(glob, &start); //Track Entry + } + Ebml_EndSubElement(glob, &trackStart); + } + // segment element is open + } +} + + +static void +write_webm_block(EbmlGlobal *glob, + const vpx_codec_enc_cfg_t *cfg, + const vpx_codec_cx_pkt_t *pkt) +{ + unsigned long block_length; + unsigned char track_number; + unsigned short block_timecode = 0; + unsigned char flags; + int64_t pts_ms; + int start_cluster = 0, is_keyframe; + + /* Calculate the PTS of this frame in milliseconds */ + pts_ms = pkt->data.frame.pts * 1000 + * (uint64_t)cfg->g_timebase.num / (uint64_t)cfg->g_timebase.den; + if(pts_ms <= glob->last_pts_ms) + pts_ms = glob->last_pts_ms + 1; + glob->last_pts_ms = pts_ms; + + /* Calculate the relative time of this block */ + if(pts_ms - glob->cluster_timecode > SHRT_MAX) + start_cluster = 1; + else + block_timecode = pts_ms - glob->cluster_timecode; + + is_keyframe = (pkt->data.frame.flags & VPX_FRAME_IS_KEY); + if(start_cluster || is_keyframe) + { + if(glob->cluster_open) + Ebml_EndSubElement(glob, &glob->startCluster); + + /* Open the new cluster */ + block_timecode = 0; + glob->cluster_open = 1; + glob->cluster_timecode = pts_ms; + glob->cluster_pos = ftello(glob->stream); + Ebml_StartSubElement(glob, &glob->startCluster, Cluster); //cluster + Ebml_SerializeUnsigned(glob, Timecode, glob->cluster_timecode); + + /* Save a cue point if this is a keyframe. */ + if(is_keyframe) + { + struct cue_entry *cue, *new_cue_list; + + new_cue_list = realloc(glob->cue_list, + (glob->cues+1) * sizeof(struct cue_entry)); + if(new_cue_list) + glob->cue_list = new_cue_list; + else + fatal("Failed to realloc cue list."); + + cue = &glob->cue_list[glob->cues]; + cue->time = glob->cluster_timecode; + cue->loc = glob->cluster_pos; + glob->cues++; + } + } + + /* Write the Simple Block */ + Ebml_WriteID(glob, SimpleBlock); + + block_length = pkt->data.frame.sz + 4; + block_length |= 0x10000000; + Ebml_Serialize(glob, &block_length, sizeof(block_length), 4); + + track_number = 1; + track_number |= 0x80; + Ebml_Write(glob, &track_number, 1); + + Ebml_Serialize(glob, &block_timecode, sizeof(block_timecode), 2); + + flags = 0; + if(is_keyframe) + flags |= 0x80; + if(pkt->data.frame.flags & VPX_FRAME_IS_INVISIBLE) + flags |= 0x08; + Ebml_Write(glob, &flags, 1); + + Ebml_Write(glob, pkt->data.frame.buf, pkt->data.frame.sz); +} + + +static void +write_webm_file_footer(EbmlGlobal *glob, long hash) +{ + + if(glob->cluster_open) + Ebml_EndSubElement(glob, &glob->startCluster); + + { + EbmlLoc start; + unsigned int i; + + glob->cue_pos = ftello(glob->stream); + Ebml_StartSubElement(glob, &start, Cues); + for(i=0; icues; i++) + { + struct cue_entry *cue = &glob->cue_list[i]; + EbmlLoc start; + + Ebml_StartSubElement(glob, &start, CuePoint); + { + EbmlLoc start; + + Ebml_SerializeUnsigned(glob, CueTime, cue->time); + + Ebml_StartSubElement(glob, &start, CueTrackPositions); + Ebml_SerializeUnsigned(glob, CueTrack, 1); + Ebml_SerializeUnsigned64(glob, CueClusterPosition, + cue->loc - glob->position_reference); + //Ebml_SerializeUnsigned(glob, CueBlockNumber, cue->blockNumber); + Ebml_EndSubElement(glob, &start); + } + Ebml_EndSubElement(glob, &start); + } + Ebml_EndSubElement(glob, &start); + } + + Ebml_EndSubElement(glob, &glob->startSegment); + + /* Patch up the seek info block */ + write_webm_seek_info(glob); + + /* Patch up the track id */ + fseeko(glob->stream, glob->track_id_pos, SEEK_SET); + Ebml_SerializeUnsigned32(glob, TrackUID, glob->debug ? 0xDEADBEEF : hash); + + fseeko(glob->stream, 0, SEEK_END); +} + + +/* Murmur hash derived from public domain reference implementation at + * http://sites.google.com/site/murmurhash/ + */ +static unsigned int murmur ( const void * key, int len, unsigned int seed ) +{ + const unsigned int m = 0x5bd1e995; + const int r = 24; + + unsigned int h = seed ^ len; + + const unsigned char * data = (const unsigned char *)key; + + while(len >= 4) + { + unsigned int k; + + k = data[0]; + k |= data[1] << 8; + k |= data[2] << 16; + k |= data[3] << 24; + + k *= m; + k ^= k >> r; + k *= m; + + h *= m; + h ^= k; + + data += 4; + len -= 4; + } + + switch(len) + { + case 3: h ^= data[2] << 16; + case 2: h ^= data[1] << 8; + case 1: h ^= data[0]; + h *= m; + }; + + h ^= h >> 13; + h *= m; + h ^= h >> 15; + + return h; +} + +#include "math.h" + +static double vp8_mse2psnr(double Samples, double Peak, double Mse) +{ + double psnr; + + if ((double)Mse > 0.0) + psnr = 10.0 * log10(Peak * Peak * Samples / Mse); + else + psnr = 60; // Limit to prevent / 0 + + if (psnr > 60) + psnr = 60; + + return psnr; +} + + +#include "args.h" +static const arg_def_t debugmode = ARG_DEF("D", "debug", 0, + "Debug mode (makes output deterministic)"); +static const arg_def_t outputfile = ARG_DEF("o", "output", 1, + "Output filename"); +static const arg_def_t use_yv12 = ARG_DEF(NULL, "yv12", 0, + "Input file is YV12 "); +static const arg_def_t use_i420 = ARG_DEF(NULL, "i420", 0, + "Input file is I420 (default)"); +static const arg_def_t codecarg = ARG_DEF(NULL, "codec", 1, + "Codec to use"); +static const arg_def_t passes = ARG_DEF("p", "passes", 1, + "Number of passes (1/2)"); +static const arg_def_t pass_arg = ARG_DEF(NULL, "pass", 1, + "Pass to execute (1/2)"); +static const arg_def_t fpf_name = ARG_DEF(NULL, "fpf", 1, + "First pass statistics file name"); +static const arg_def_t limit = ARG_DEF(NULL, "limit", 1, + "Stop encoding after n input frames"); +static const arg_def_t deadline = ARG_DEF("d", "deadline", 1, + "Deadline per frame (usec)"); +static const arg_def_t best_dl = ARG_DEF(NULL, "best", 0, + "Use Best Quality Deadline"); +static const arg_def_t good_dl = ARG_DEF(NULL, "good", 0, + "Use Good Quality Deadline"); +static const arg_def_t rt_dl = ARG_DEF(NULL, "rt", 0, + "Use Realtime Quality Deadline"); +static const arg_def_t verbosearg = ARG_DEF("v", "verbose", 0, + "Show encoder parameters"); +static const arg_def_t psnrarg = ARG_DEF(NULL, "psnr", 0, + "Show PSNR in status line"); +static const arg_def_t framerate = ARG_DEF(NULL, "fps", 1, + "Stream frame rate (rate/scale)"); +static const arg_def_t use_ivf = ARG_DEF(NULL, "ivf", 0, + "Output IVF (default is WebM)"); +static const arg_def_t out_part = ARG_DEF("P", "output-partitions", 0, + "Makes encoder output partitions. Requires IVF output!"); +static const arg_def_t q_hist_n = ARG_DEF(NULL, "q-hist", 1, + "Show quantizer histogram (n-buckets)"); +static const arg_def_t rate_hist_n = ARG_DEF(NULL, "rate-hist", 1, + "Show rate histogram (n-buckets)"); +static const arg_def_t *main_args[] = +{ + &debugmode, + &outputfile, &codecarg, &passes, &pass_arg, &fpf_name, &limit, &deadline, + &best_dl, &good_dl, &rt_dl, + &verbosearg, &psnrarg, &use_ivf, &out_part, &q_hist_n, &rate_hist_n, + NULL +}; + +static const arg_def_t usage = ARG_DEF("u", "usage", 1, + "Usage profile number to use"); +static const arg_def_t threads = ARG_DEF("t", "threads", 1, + "Max number of threads to use"); +static const arg_def_t profile = ARG_DEF(NULL, "profile", 1, + "Bitstream profile number to use"); +static const arg_def_t width = ARG_DEF("w", "width", 1, + "Frame width"); +static const arg_def_t height = ARG_DEF("h", "height", 1, + "Frame height"); +static const struct arg_enum_list stereo_mode_enum[] = { + {"mono" , STEREO_FORMAT_MONO}, + {"left-right", STEREO_FORMAT_LEFT_RIGHT}, + {"bottom-top", STEREO_FORMAT_BOTTOM_TOP}, + {"top-bottom", STEREO_FORMAT_TOP_BOTTOM}, + {"right-left", STEREO_FORMAT_RIGHT_LEFT}, + {NULL, 0} +}; +static const arg_def_t stereo_mode = ARG_DEF_ENUM(NULL, "stereo-mode", 1, + "Stereo 3D video format", stereo_mode_enum); +static const arg_def_t timebase = ARG_DEF(NULL, "timebase", 1, + "Output timestamp precision (fractional seconds)"); +static const arg_def_t error_resilient = ARG_DEF(NULL, "error-resilient", 1, + "Enable error resiliency features"); +static const arg_def_t lag_in_frames = ARG_DEF(NULL, "lag-in-frames", 1, + "Max number of frames to lag"); + +static const arg_def_t *global_args[] = +{ + &use_yv12, &use_i420, &usage, &threads, &profile, + &width, &height, &stereo_mode, &timebase, &framerate, &error_resilient, + &lag_in_frames, NULL +}; + +static const arg_def_t dropframe_thresh = ARG_DEF(NULL, "drop-frame", 1, + "Temporal resampling threshold (buf %)"); +static const arg_def_t resize_allowed = ARG_DEF(NULL, "resize-allowed", 1, + "Spatial resampling enabled (bool)"); +static const arg_def_t resize_up_thresh = ARG_DEF(NULL, "resize-up", 1, + "Upscale threshold (buf %)"); +static const arg_def_t resize_down_thresh = ARG_DEF(NULL, "resize-down", 1, + "Downscale threshold (buf %)"); +static const struct arg_enum_list end_usage_enum[] = { + {"vbr", VPX_VBR}, + {"cbr", VPX_CBR}, + {"cq", VPX_CQ}, + {NULL, 0} +}; +static const arg_def_t end_usage = ARG_DEF_ENUM(NULL, "end-usage", 1, + "Rate control mode", end_usage_enum); +static const arg_def_t target_bitrate = ARG_DEF(NULL, "target-bitrate", 1, + "Bitrate (kbps)"); +static const arg_def_t min_quantizer = ARG_DEF(NULL, "min-q", 1, + "Minimum (best) quantizer"); +static const arg_def_t max_quantizer = ARG_DEF(NULL, "max-q", 1, + "Maximum (worst) quantizer"); +static const arg_def_t undershoot_pct = ARG_DEF(NULL, "undershoot-pct", 1, + "Datarate undershoot (min) target (%)"); +static const arg_def_t overshoot_pct = ARG_DEF(NULL, "overshoot-pct", 1, + "Datarate overshoot (max) target (%)"); +static const arg_def_t buf_sz = ARG_DEF(NULL, "buf-sz", 1, + "Client buffer size (ms)"); +static const arg_def_t buf_initial_sz = ARG_DEF(NULL, "buf-initial-sz", 1, + "Client initial buffer size (ms)"); +static const arg_def_t buf_optimal_sz = ARG_DEF(NULL, "buf-optimal-sz", 1, + "Client optimal buffer size (ms)"); +static const arg_def_t *rc_args[] = +{ + &dropframe_thresh, &resize_allowed, &resize_up_thresh, &resize_down_thresh, + &end_usage, &target_bitrate, &min_quantizer, &max_quantizer, + &undershoot_pct, &overshoot_pct, &buf_sz, &buf_initial_sz, &buf_optimal_sz, + NULL +}; + + +static const arg_def_t bias_pct = ARG_DEF(NULL, "bias-pct", 1, + "CBR/VBR bias (0=CBR, 100=VBR)"); +static const arg_def_t minsection_pct = ARG_DEF(NULL, "minsection-pct", 1, + "GOP min bitrate (% of target)"); +static const arg_def_t maxsection_pct = ARG_DEF(NULL, "maxsection-pct", 1, + "GOP max bitrate (% of target)"); +static const arg_def_t *rc_twopass_args[] = +{ + &bias_pct, &minsection_pct, &maxsection_pct, NULL +}; + + +static const arg_def_t kf_min_dist = ARG_DEF(NULL, "kf-min-dist", 1, + "Minimum keyframe interval (frames)"); +static const arg_def_t kf_max_dist = ARG_DEF(NULL, "kf-max-dist", 1, + "Maximum keyframe interval (frames)"); +static const arg_def_t kf_disabled = ARG_DEF(NULL, "disable-kf", 0, + "Disable keyframe placement"); +static const arg_def_t *kf_args[] = +{ + &kf_min_dist, &kf_max_dist, &kf_disabled, NULL +}; + + +#if CONFIG_VP8_ENCODER +static const arg_def_t noise_sens = ARG_DEF(NULL, "noise-sensitivity", 1, + "Noise sensitivity (frames to blur)"); +static const arg_def_t sharpness = ARG_DEF(NULL, "sharpness", 1, + "Filter sharpness (0-7)"); +static const arg_def_t static_thresh = ARG_DEF(NULL, "static-thresh", 1, + "Motion detection threshold"); +#endif + +#if CONFIG_VP8_ENCODER +static const arg_def_t cpu_used = ARG_DEF(NULL, "cpu-used", 1, + "CPU Used (-16..16)"); +#endif + + +#if CONFIG_VP8_ENCODER +static const arg_def_t token_parts = ARG_DEF(NULL, "token-parts", 1, + "Number of token partitions to use, log2"); +static const arg_def_t auto_altref = ARG_DEF(NULL, "auto-alt-ref", 1, + "Enable automatic alt reference frames"); +static const arg_def_t arnr_maxframes = ARG_DEF(NULL, "arnr-maxframes", 1, + "AltRef Max Frames"); +static const arg_def_t arnr_strength = ARG_DEF(NULL, "arnr-strength", 1, + "AltRef Strength"); +static const arg_def_t arnr_type = ARG_DEF(NULL, "arnr-type", 1, + "AltRef Type"); +static const struct arg_enum_list tuning_enum[] = { + {"psnr", VP8_TUNE_PSNR}, + {"ssim", VP8_TUNE_SSIM}, + {NULL, 0} +}; +static const arg_def_t tune_ssim = ARG_DEF_ENUM(NULL, "tune", 1, + "Material to favor", tuning_enum); +static const arg_def_t cq_level = ARG_DEF(NULL, "cq-level", 1, + "Constrained Quality Level"); +static const arg_def_t max_intra_rate_pct = ARG_DEF(NULL, "max-intra-rate", 1, + "Max I-frame bitrate (pct)"); + +static const arg_def_t *vp8_args[] = +{ + &cpu_used, &auto_altref, &noise_sens, &sharpness, &static_thresh, + &token_parts, &arnr_maxframes, &arnr_strength, &arnr_type, + &tune_ssim, &cq_level, &max_intra_rate_pct, NULL +}; +static const int vp8_arg_ctrl_map[] = +{ + VP8E_SET_CPUUSED, VP8E_SET_ENABLEAUTOALTREF, + VP8E_SET_NOISE_SENSITIVITY, VP8E_SET_SHARPNESS, VP8E_SET_STATIC_THRESHOLD, + VP8E_SET_TOKEN_PARTITIONS, + VP8E_SET_ARNR_MAXFRAMES, VP8E_SET_ARNR_STRENGTH , VP8E_SET_ARNR_TYPE, + VP8E_SET_TUNING, VP8E_SET_CQ_LEVEL, VP8E_SET_MAX_INTRA_BITRATE_PCT, 0 +}; +#endif + +static const arg_def_t *no_args[] = { NULL }; + +static void usage_exit() +{ + int i; + + fprintf(stderr, "Usage: %s -o dst_filename src_filename \n", + exec_name); + + fprintf(stderr, "\nOptions:\n"); + arg_show_usage(stdout, main_args); + fprintf(stderr, "\nEncoder Global Options:\n"); + arg_show_usage(stdout, global_args); + fprintf(stderr, "\nRate Control Options:\n"); + arg_show_usage(stdout, rc_args); + fprintf(stderr, "\nTwopass Rate Control Options:\n"); + arg_show_usage(stdout, rc_twopass_args); + fprintf(stderr, "\nKeyframe Placement Options:\n"); + arg_show_usage(stdout, kf_args); +#if CONFIG_VP8_ENCODER + fprintf(stderr, "\nVP8 Specific Options:\n"); + arg_show_usage(stdout, vp8_args); +#endif + fprintf(stderr, "\nStream timebase (--timebase):\n" + " The desired precision of timestamps in the output, expressed\n" + " in fractional seconds. Default is 1/1000.\n"); + fprintf(stderr, "\n" + "Included encoders:\n" + "\n"); + + for (i = 0; i < sizeof(codecs) / sizeof(codecs[0]); i++) + fprintf(stderr, " %-6s - %s\n", + codecs[i].name, + vpx_codec_iface_name(codecs[i].iface)); + + exit(EXIT_FAILURE); +} + + +#define HIST_BAR_MAX 40 +struct hist_bucket +{ + int low, high, count; +}; + + +static int merge_hist_buckets(struct hist_bucket *bucket, + int *buckets_, + int max_buckets) +{ + int small_bucket = 0, merge_bucket = INT_MAX, big_bucket=0; + int buckets = *buckets_; + int i; + + /* Find the extrema for this list of buckets */ + big_bucket = small_bucket = 0; + for(i=0; i < buckets; i++) + { + if(bucket[i].count < bucket[small_bucket].count) + small_bucket = i; + if(bucket[i].count > bucket[big_bucket].count) + big_bucket = i; + } + + /* If we have too many buckets, merge the smallest with an adjacent + * bucket. + */ + while(buckets > max_buckets) + { + int last_bucket = buckets - 1; + + // merge the small bucket with an adjacent one. + if(small_bucket == 0) + merge_bucket = 1; + else if(small_bucket == last_bucket) + merge_bucket = last_bucket - 1; + else if(bucket[small_bucket - 1].count < bucket[small_bucket + 1].count) + merge_bucket = small_bucket - 1; + else + merge_bucket = small_bucket + 1; + + assert(abs(merge_bucket - small_bucket) <= 1); + assert(small_bucket < buckets); + assert(big_bucket < buckets); + assert(merge_bucket < buckets); + + if(merge_bucket < small_bucket) + { + bucket[merge_bucket].high = bucket[small_bucket].high; + bucket[merge_bucket].count += bucket[small_bucket].count; + } + else + { + bucket[small_bucket].high = bucket[merge_bucket].high; + bucket[small_bucket].count += bucket[merge_bucket].count; + merge_bucket = small_bucket; + } + + assert(bucket[merge_bucket].low != bucket[merge_bucket].high); + + buckets--; + + /* Remove the merge_bucket from the list, and find the new small + * and big buckets while we're at it + */ + big_bucket = small_bucket = 0; + for(i=0; i < buckets; i++) + { + if(i > merge_bucket) + bucket[i] = bucket[i+1]; + + if(bucket[i].count < bucket[small_bucket].count) + small_bucket = i; + if(bucket[i].count > bucket[big_bucket].count) + big_bucket = i; + } + + } + + *buckets_ = buckets; + return bucket[big_bucket].count; +} + + +static void show_histogram(const struct hist_bucket *bucket, + int buckets, + int total, + int scale) +{ + const char *pat1, *pat2; + int i; + + switch((int)(log(bucket[buckets-1].high)/log(10))+1) + { + case 1: + case 2: + pat1 = "%4d %2s: "; + pat2 = "%4d-%2d: "; + break; + case 3: + pat1 = "%5d %3s: "; + pat2 = "%5d-%3d: "; + break; + case 4: + pat1 = "%6d %4s: "; + pat2 = "%6d-%4d: "; + break; + case 5: + pat1 = "%7d %5s: "; + pat2 = "%7d-%5d: "; + break; + case 6: + pat1 = "%8d %6s: "; + pat2 = "%8d-%6d: "; + break; + case 7: + pat1 = "%9d %7s: "; + pat2 = "%9d-%7d: "; + break; + default: + pat1 = "%12d %10s: "; + pat2 = "%12d-%10d: "; + break; + } + + for(i=0; isamples = cfg->rc_buf_sz * 5 / 4 * fps->num / fps->den / 1000; + + // prevent division by zero + if (hist->samples == 0) + hist->samples=1; + + hist->pts = calloc(hist->samples, sizeof(*hist->pts)); + hist->sz = calloc(hist->samples, sizeof(*hist->sz)); + for(i=0; ibucket[i].low = INT_MAX; + hist->bucket[i].high = 0; + hist->bucket[i].count = 0; + } +} + + +static void destroy_rate_histogram(struct rate_hist *hist) +{ + free(hist->pts); + free(hist->sz); +} + + +static void update_rate_histogram(struct rate_hist *hist, + const vpx_codec_enc_cfg_t *cfg, + const vpx_codec_cx_pkt_t *pkt) +{ + int i, idx; + int64_t now, then, sum_sz = 0, avg_bitrate; + + now = pkt->data.frame.pts * 1000 + * (uint64_t)cfg->g_timebase.num / (uint64_t)cfg->g_timebase.den; + + idx = hist->frames++ % hist->samples; + hist->pts[idx] = now; + hist->sz[idx] = pkt->data.frame.sz; + + if(now < cfg->rc_buf_initial_sz) + return; + + then = now; + + /* Sum the size over the past rc_buf_sz ms */ + for(i = hist->frames; i > 0 && hist->frames - i < hist->samples; i--) + { + int i_idx = (i-1) % hist->samples; + + then = hist->pts[i_idx]; + if(now - then > cfg->rc_buf_sz) + break; + sum_sz += hist->sz[i_idx]; + } + + if (now == then) + return; + + avg_bitrate = sum_sz * 8 * 1000 / (now - then); + idx = avg_bitrate * (RATE_BINS/2) / (cfg->rc_target_bitrate * 1000); + if(idx < 0) + idx = 0; + if(idx > RATE_BINS-1) + idx = RATE_BINS-1; + if(hist->bucket[idx].low > avg_bitrate) + hist->bucket[idx].low = avg_bitrate; + if(hist->bucket[idx].high < avg_bitrate) + hist->bucket[idx].high = avg_bitrate; + hist->bucket[idx].count++; + hist->total++; +} + + +static void show_rate_histogram(struct rate_hist *hist, + const vpx_codec_enc_cfg_t *cfg, + int max_buckets) +{ + int i, scale; + int buckets = 0; + + for(i = 0; i < RATE_BINS; i++) + { + if(hist->bucket[i].low == INT_MAX) + continue; + hist->bucket[buckets++] = hist->bucket[i]; + } + + fprintf(stderr, "\nRate (over %dms window):\n", cfg->rc_buf_sz); + scale = merge_hist_buckets(hist->bucket, &buckets, max_buckets); + show_histogram(hist->bucket, buckets, hist->total, scale); +} + +#define NELEMENTS(x) (sizeof(x)/sizeof(x[0])) +#define ARG_CTRL_CNT_MAX NELEMENTS(vp8_arg_ctrl_map) + + +/* Configuration elements common to all streams */ +struct global_config +{ + const struct codec_item *codec; + int passes; + int pass; + int usage; + int deadline; + int use_i420; + int verbose; + int limit; + int show_psnr; + int have_framerate; + struct vpx_rational framerate; + int out_part; + int debug; + int show_q_hist_buckets; + int show_rate_hist_buckets; +}; + + +/* Per-stream configuration */ +struct stream_config +{ + struct vpx_codec_enc_cfg cfg; + const char *out_fn; + const char *stats_fn; + stereo_format_t stereo_fmt; + int arg_ctrls[ARG_CTRL_CNT_MAX][2]; + int arg_ctrl_cnt; + int write_webm; + int have_kf_max_dist; +}; + + +struct stream_state +{ + int index; + struct stream_state *next; + struct stream_config config; + FILE *file; + struct rate_hist rate_hist; + EbmlGlobal ebml; + uint32_t hash; + uint64_t psnr_sse_total; + uint64_t psnr_samples_total; + double psnr_totals[4]; + int psnr_count; + int counts[64]; + vpx_codec_ctx_t encoder; + unsigned int frames_out; + uint64_t cx_time; + size_t nbytes; + stats_io_t stats; +}; + + +void validate_positive_rational(const char *msg, + struct vpx_rational *rat) +{ + if (rat->den < 0) + { + rat->num *= -1; + rat->den *= -1; + } + + if (rat->num < 0) + die("Error: %s must be positive\n", msg); + + if (!rat->den) + die("Error: %s has zero denominator\n", msg); +} + + +static void parse_global_config(struct global_config *global, char **argv) +{ + char **argi, **argj; + struct arg arg; + + /* Initialize default parameters */ + memset(global, 0, sizeof(*global)); + global->codec = codecs; + global->passes = 1; + global->use_i420 = 1; + + for (argi = argj = argv; (*argj = *argi); argi += arg.argv_step) + { + arg.argv_step = 1; + + if (arg_match(&arg, &codecarg, argi)) + { + int j, k = -1; + + for (j = 0; j < sizeof(codecs) / sizeof(codecs[0]); j++) + if (!strcmp(codecs[j].name, arg.val)) + k = j; + + if (k >= 0) + global->codec = codecs + k; + else + die("Error: Unrecognized argument (%s) to --codec\n", + arg.val); + + } + else if (arg_match(&arg, &passes, argi)) + { + global->passes = arg_parse_uint(&arg); + + if (global->passes < 1 || global->passes > 2) + die("Error: Invalid number of passes (%d)\n", global->passes); + } + else if (arg_match(&arg, &pass_arg, argi)) + { + global->pass = arg_parse_uint(&arg); + + if (global->pass < 1 || global->pass > 2) + die("Error: Invalid pass selected (%d)\n", + global->pass); + } + else if (arg_match(&arg, &usage, argi)) + global->usage = arg_parse_uint(&arg); + else if (arg_match(&arg, &deadline, argi)) + global->deadline = arg_parse_uint(&arg); + else if (arg_match(&arg, &best_dl, argi)) + global->deadline = VPX_DL_BEST_QUALITY; + else if (arg_match(&arg, &good_dl, argi)) + global->deadline = VPX_DL_GOOD_QUALITY; + else if (arg_match(&arg, &rt_dl, argi)) + global->deadline = VPX_DL_REALTIME; + else if (arg_match(&arg, &use_yv12, argi)) + global->use_i420 = 0; + else if (arg_match(&arg, &use_i420, argi)) + global->use_i420 = 1; + else if (arg_match(&arg, &verbosearg, argi)) + global->verbose = 1; + else if (arg_match(&arg, &limit, argi)) + global->limit = arg_parse_uint(&arg); + else if (arg_match(&arg, &psnrarg, argi)) + global->show_psnr = 1; + else if (arg_match(&arg, &framerate, argi)) + { + global->framerate = arg_parse_rational(&arg); + validate_positive_rational(arg.name, &global->framerate); + global->have_framerate = 1; + } + else if (arg_match(&arg,&out_part, argi)) + global->out_part = 1; + else if (arg_match(&arg, &debugmode, argi)) + global->debug = 1; + else if (arg_match(&arg, &q_hist_n, argi)) + global->show_q_hist_buckets = arg_parse_uint(&arg); + else if (arg_match(&arg, &rate_hist_n, argi)) + global->show_rate_hist_buckets = arg_parse_uint(&arg); + else + argj++; + } + + /* Validate global config */ + + if (global->pass) + { + /* DWIM: Assume the user meant passes=2 if pass=2 is specified */ + if (global->pass > global->passes) + { + warn("Assuming --pass=%d implies --passes=%d\n", + global->pass, global->pass); + global->passes = global->pass; + } + } +} + + +void open_input_file(struct input_state *input) +{ + unsigned int fourcc; + + /* Parse certain options from the input file, if possible */ + input->file = strcmp(input->fn, "-") ? fopen(input->fn, "rb") + : set_binary_mode(stdin); + + if (!input->file) + fatal("Failed to open input file"); + + /* For RAW input sources, these bytes will applied on the first frame + * in read_frame(). + */ + input->detect.buf_read = fread(input->detect.buf, 1, 4, input->file); + input->detect.position = 0; + + if (input->detect.buf_read == 4 + && file_is_y4m(input->file, &input->y4m, input->detect.buf)) + { + if (y4m_input_open(&input->y4m, input->file, input->detect.buf, 4) >= 0) + { + input->file_type = FILE_TYPE_Y4M; + input->w = input->y4m.pic_w; + input->h = input->y4m.pic_h; + input->framerate.num = input->y4m.fps_n; + input->framerate.den = input->y4m.fps_d; + input->use_i420 = 0; + } + else + fatal("Unsupported Y4M stream."); + } + else if (input->detect.buf_read == 4 && file_is_ivf(input, &fourcc)) + { + input->file_type = FILE_TYPE_IVF; + switch (fourcc) + { + case 0x32315659: + input->use_i420 = 0; + break; + case 0x30323449: + input->use_i420 = 1; + break; + default: + fatal("Unsupported fourcc (%08x) in IVF", fourcc); + } + } + else + { + input->file_type = FILE_TYPE_RAW; + } +} + + +static void close_input_file(struct input_state *input) +{ + fclose(input->file); + if (input->file_type == FILE_TYPE_Y4M) + y4m_input_close(&input->y4m); +} + +static struct stream_state *new_stream(struct global_config *global, + struct stream_state *prev) +{ + struct stream_state *stream; + + stream = calloc(1, sizeof(*stream)); + if(!stream) + fatal("Failed to allocate new stream."); + if(prev) + { + memcpy(stream, prev, sizeof(*stream)); + stream->index++; + prev->next = stream; + } + else + { + vpx_codec_err_t res; + + /* Populate encoder configuration */ + res = vpx_codec_enc_config_default(global->codec->iface, + &stream->config.cfg, + global->usage); + if (res) + fatal("Failed to get config: %s\n", vpx_codec_err_to_string(res)); + + /* Change the default timebase to a high enough value so that the + * encoder will always create strictly increasing timestamps. + */ + stream->config.cfg.g_timebase.den = 1000; + + /* Never use the library's default resolution, require it be parsed + * from the file or set on the command line. + */ + stream->config.cfg.g_w = 0; + stream->config.cfg.g_h = 0; + + /* Initialize remaining stream parameters */ + stream->config.stereo_fmt = STEREO_FORMAT_MONO; + stream->config.write_webm = 1; + stream->ebml.last_pts_ms = -1; + + /* Allows removal of the application version from the EBML tags */ + stream->ebml.debug = global->debug; + } + + /* Output files must be specified for each stream */ + stream->config.out_fn = NULL; + + stream->next = NULL; + return stream; +} + + +static int parse_stream_params(struct global_config *global, + struct stream_state *stream, + char **argv) +{ + char **argi, **argj; + struct arg arg; + static const arg_def_t **ctrl_args = no_args; + static const int *ctrl_args_map = NULL; + struct stream_config *config = &stream->config; + int eos_mark_found = 0; + + /* Handle codec specific options */ + if (global->codec->iface == &vpx_codec_vp8_cx_algo) + { + ctrl_args = vp8_args; + ctrl_args_map = vp8_arg_ctrl_map; + } + + for (argi = argj = argv; (*argj = *argi); argi += arg.argv_step) + { + arg.argv_step = 1; + + /* Once we've found an end-of-stream marker (--) we want to continue + * shifting arguments but not consuming them. + */ + if (eos_mark_found) + { + argj++; + continue; + } + else if (!strcmp(*argj, "--")) + { + eos_mark_found = 1; + continue; + } + + if (0); + else if (arg_match(&arg, &outputfile, argi)) + config->out_fn = arg.val; + else if (arg_match(&arg, &fpf_name, argi)) + config->stats_fn = arg.val; + else if (arg_match(&arg, &use_ivf, argi)) + config->write_webm = 0; + else if (arg_match(&arg, &threads, argi)) + config->cfg.g_threads = arg_parse_uint(&arg); + else if (arg_match(&arg, &profile, argi)) + config->cfg.g_profile = arg_parse_uint(&arg); + else if (arg_match(&arg, &width, argi)) + config->cfg.g_w = arg_parse_uint(&arg); + else if (arg_match(&arg, &height, argi)) + config->cfg.g_h = arg_parse_uint(&arg); + else if (arg_match(&arg, &stereo_mode, argi)) + config->stereo_fmt = arg_parse_enum_or_int(&arg); + else if (arg_match(&arg, &timebase, argi)) + { + config->cfg.g_timebase = arg_parse_rational(&arg); + validate_positive_rational(arg.name, &config->cfg.g_timebase); + } + else if (arg_match(&arg, &error_resilient, argi)) + config->cfg.g_error_resilient = arg_parse_uint(&arg); + else if (arg_match(&arg, &lag_in_frames, argi)) + config->cfg.g_lag_in_frames = arg_parse_uint(&arg); + else if (arg_match(&arg, &dropframe_thresh, argi)) + config->cfg.rc_dropframe_thresh = arg_parse_uint(&arg); + else if (arg_match(&arg, &resize_allowed, argi)) + config->cfg.rc_resize_allowed = arg_parse_uint(&arg); + else if (arg_match(&arg, &resize_up_thresh, argi)) + config->cfg.rc_resize_up_thresh = arg_parse_uint(&arg); + else if (arg_match(&arg, &resize_down_thresh, argi)) + config->cfg.rc_resize_down_thresh = arg_parse_uint(&arg); + else if (arg_match(&arg, &end_usage, argi)) + config->cfg.rc_end_usage = arg_parse_enum_or_int(&arg); + else if (arg_match(&arg, &target_bitrate, argi)) + config->cfg.rc_target_bitrate = arg_parse_uint(&arg); + else if (arg_match(&arg, &min_quantizer, argi)) + config->cfg.rc_min_quantizer = arg_parse_uint(&arg); + else if (arg_match(&arg, &max_quantizer, argi)) + config->cfg.rc_max_quantizer = arg_parse_uint(&arg); + else if (arg_match(&arg, &undershoot_pct, argi)) + config->cfg.rc_undershoot_pct = arg_parse_uint(&arg); + else if (arg_match(&arg, &overshoot_pct, argi)) + config->cfg.rc_overshoot_pct = arg_parse_uint(&arg); + else if (arg_match(&arg, &buf_sz, argi)) + config->cfg.rc_buf_sz = arg_parse_uint(&arg); + else if (arg_match(&arg, &buf_initial_sz, argi)) + config->cfg.rc_buf_initial_sz = arg_parse_uint(&arg); + else if (arg_match(&arg, &buf_optimal_sz, argi)) + config->cfg.rc_buf_optimal_sz = arg_parse_uint(&arg); + else if (arg_match(&arg, &bias_pct, argi)) + { + config->cfg.rc_2pass_vbr_bias_pct = arg_parse_uint(&arg); + + if (global->passes < 2) + warn("option %s ignored in one-pass mode.\n", arg.name); + } + else if (arg_match(&arg, &minsection_pct, argi)) + { + config->cfg.rc_2pass_vbr_minsection_pct = arg_parse_uint(&arg); + + if (global->passes < 2) + warn("option %s ignored in one-pass mode.\n", arg.name); + } + else if (arg_match(&arg, &maxsection_pct, argi)) + { + config->cfg.rc_2pass_vbr_maxsection_pct = arg_parse_uint(&arg); + + if (global->passes < 2) + warn("option %s ignored in one-pass mode.\n", arg.name); + } + else if (arg_match(&arg, &kf_min_dist, argi)) + config->cfg.kf_min_dist = arg_parse_uint(&arg); + else if (arg_match(&arg, &kf_max_dist, argi)) + { + config->cfg.kf_max_dist = arg_parse_uint(&arg); + config->have_kf_max_dist = 1; + } + else if (arg_match(&arg, &kf_disabled, argi)) + config->cfg.kf_mode = VPX_KF_DISABLED; + else + { + int i, match = 0; + + for (i = 0; ctrl_args[i]; i++) + { + if (arg_match(&arg, ctrl_args[i], argi)) + { + int j; + match = 1; + + /* Point either to the next free element or the first + * instance of this control. + */ + for(j=0; jarg_ctrl_cnt; j++) + if(config->arg_ctrls[j][0] == ctrl_args_map[i]) + break; + + /* Update/insert */ + assert(j < ARG_CTRL_CNT_MAX); + if (j < ARG_CTRL_CNT_MAX) + { + config->arg_ctrls[j][0] = ctrl_args_map[i]; + config->arg_ctrls[j][1] = arg_parse_enum_or_int(&arg); + if(j == config->arg_ctrl_cnt) + config->arg_ctrl_cnt++; + } + + } + } + + if (!match) + argj++; + } + } + + return eos_mark_found; +} + + +#define FOREACH_STREAM(func)\ +do\ +{\ + struct stream_state *stream;\ +\ + for(stream = streams; stream; stream = stream->next)\ + func;\ +}while(0) + + +static void validate_stream_config(struct stream_state *stream) +{ + struct stream_state *streami; + + if(!stream->config.cfg.g_w || !stream->config.cfg.g_h) + fatal("Stream %d: Specify stream dimensions with --width (-w) " + " and --height (-h)", stream->index); + + for(streami = stream; streami; streami = streami->next) + { + /* All streams require output files */ + if(!streami->config.out_fn) + fatal("Stream %d: Output file is required (specify with -o)", + streami->index); + + /* Check for two streams outputting to the same file */ + if(streami != stream) + { + const char *a = stream->config.out_fn; + const char *b = streami->config.out_fn; + if(!strcmp(a,b) && strcmp(a, "/dev/null") && strcmp(a, ":nul")) + fatal("Stream %d: duplicate output file (from stream %d)", + streami->index, stream->index); + } + + /* Check for two streams sharing a stats file. */ + if(streami != stream) + { + const char *a = stream->config.stats_fn; + const char *b = streami->config.stats_fn; + if(a && b && !strcmp(a,b)) + fatal("Stream %d: duplicate stats file (from stream %d)", + streami->index, stream->index); + } + } +} + + +static void set_stream_dimensions(struct stream_state *stream, + unsigned int w, + unsigned int h) +{ + if ((stream->config.cfg.g_w && stream->config.cfg.g_w != w) + ||(stream->config.cfg.g_h && stream->config.cfg.g_h != h)) + fatal("Stream %d: Resizing not yet supported", stream->index); + stream->config.cfg.g_w = w; + stream->config.cfg.g_h = h; +} + + +static void set_default_kf_interval(struct stream_state *stream, + struct global_config *global) +{ + /* Use a max keyframe interval of 5 seconds, if none was + * specified on the command line. + */ + if (!stream->config.have_kf_max_dist) + { + double framerate = (double)global->framerate.num/global->framerate.den; + if (framerate > 0.0) + stream->config.cfg.kf_max_dist = 5.0*framerate; + } +} + + +static void show_stream_config(struct stream_state *stream, + struct global_config *global, + struct input_state *input) +{ + +#define SHOW(field) \ + fprintf(stderr, " %-28s = %d\n", #field, stream->config.cfg.field) + + if(stream->index == 0) + { + fprintf(stderr, "Codec: %s\n", + vpx_codec_iface_name(global->codec->iface)); + fprintf(stderr, "Source file: %s Format: %s\n", input->fn, + input->use_i420 ? "I420" : "YV12"); + } + if(stream->next || stream->index) + fprintf(stderr, "\nStream Index: %d\n", stream->index); + fprintf(stderr, "Destination file: %s\n", stream->config.out_fn); + fprintf(stderr, "Encoder parameters:\n"); + + SHOW(g_usage); + SHOW(g_threads); + SHOW(g_profile); + SHOW(g_w); + SHOW(g_h); + SHOW(g_timebase.num); + SHOW(g_timebase.den); + SHOW(g_error_resilient); + SHOW(g_pass); + SHOW(g_lag_in_frames); + SHOW(rc_dropframe_thresh); + SHOW(rc_resize_allowed); + SHOW(rc_resize_up_thresh); + SHOW(rc_resize_down_thresh); + SHOW(rc_end_usage); + SHOW(rc_target_bitrate); + SHOW(rc_min_quantizer); + SHOW(rc_max_quantizer); + SHOW(rc_undershoot_pct); + SHOW(rc_overshoot_pct); + SHOW(rc_buf_sz); + SHOW(rc_buf_initial_sz); + SHOW(rc_buf_optimal_sz); + SHOW(rc_2pass_vbr_bias_pct); + SHOW(rc_2pass_vbr_minsection_pct); + SHOW(rc_2pass_vbr_maxsection_pct); + SHOW(kf_mode); + SHOW(kf_min_dist); + SHOW(kf_max_dist); +} + + +static void open_output_file(struct stream_state *stream, + struct global_config *global) +{ + const char *fn = stream->config.out_fn; + + stream->file = strcmp(fn, "-") ? fopen(fn, "wb") : set_binary_mode(stdout); + + if (!stream->file) + fatal("Failed to open output file"); + + if(stream->config.write_webm && fseek(stream->file, 0, SEEK_CUR)) + fatal("WebM output to pipes not supported."); + + if(stream->config.write_webm) + { + stream->ebml.stream = stream->file; + write_webm_file_header(&stream->ebml, &stream->config.cfg, + &global->framerate, + stream->config.stereo_fmt); + } + else + write_ivf_file_header(stream->file, &stream->config.cfg, + global->codec->fourcc, 0); +} + + +static void close_output_file(struct stream_state *stream, + unsigned int fourcc) +{ + if(stream->config.write_webm) + { + write_webm_file_footer(&stream->ebml, stream->hash); + free(stream->ebml.cue_list); + stream->ebml.cue_list = NULL; + } + else + { + if (!fseek(stream->file, 0, SEEK_SET)) + write_ivf_file_header(stream->file, &stream->config.cfg, + fourcc, + stream->frames_out); + } + + fclose(stream->file); +} + + +static void setup_pass(struct stream_state *stream, + struct global_config *global, + int pass) +{ + if (stream->config.stats_fn) + { + if (!stats_open_file(&stream->stats, stream->config.stats_fn, + pass)) + fatal("Failed to open statistics store"); + } + else + { + if (!stats_open_mem(&stream->stats, pass)) + fatal("Failed to open statistics store"); + } + + stream->config.cfg.g_pass = global->passes == 2 + ? pass ? VPX_RC_LAST_PASS : VPX_RC_FIRST_PASS + : VPX_RC_ONE_PASS; + if (pass) + stream->config.cfg.rc_twopass_stats_in = stats_get(&stream->stats); + + stream->cx_time = 0; + stream->nbytes = 0; + stream->frames_out = 0; +} + + +static void initialize_encoder(struct stream_state *stream, + struct global_config *global) +{ + int i; + int flags = 0; + + flags |= global->show_psnr ? VPX_CODEC_USE_PSNR : 0; + flags |= global->out_part ? VPX_CODEC_USE_OUTPUT_PARTITION : 0; + + /* Construct Encoder Context */ + vpx_codec_enc_init(&stream->encoder, global->codec->iface, + &stream->config.cfg, flags); + ctx_exit_on_error(&stream->encoder, "Failed to initialize encoder"); + + /* Note that we bypass the vpx_codec_control wrapper macro because + * we're being clever to store the control IDs in an array. Real + * applications will want to make use of the enumerations directly + */ + for (i = 0; i < stream->config.arg_ctrl_cnt; i++) + { + int ctrl = stream->config.arg_ctrls[i][0]; + int value = stream->config.arg_ctrls[i][1]; + if (vpx_codec_control_(&stream->encoder, ctrl, value)) + fprintf(stderr, "Error: Tried to set control %d = %d\n", + ctrl, value); + + ctx_exit_on_error(&stream->encoder, "Failed to control codec"); + } +} + + +static void encode_frame(struct stream_state *stream, + struct global_config *global, + struct vpx_image *img, + unsigned int frames_in) +{ + vpx_codec_pts_t frame_start, next_frame_start; + struct vpx_codec_enc_cfg *cfg = &stream->config.cfg; + struct vpx_usec_timer timer; + + frame_start = (cfg->g_timebase.den * (int64_t)(frames_in - 1) + * global->framerate.den) + / cfg->g_timebase.num / global->framerate.num; + next_frame_start = (cfg->g_timebase.den * (int64_t)(frames_in) + * global->framerate.den) + / cfg->g_timebase.num / global->framerate.num; + vpx_usec_timer_start(&timer); + vpx_codec_encode(&stream->encoder, img, frame_start, + next_frame_start - frame_start, + 0, global->deadline); + vpx_usec_timer_mark(&timer); + stream->cx_time += vpx_usec_timer_elapsed(&timer); + ctx_exit_on_error(&stream->encoder, "Stream %d: Failed to encode frame", + stream->index); +} + + +static void update_quantizer_histogram(struct stream_state *stream) +{ + if(stream->config.cfg.g_pass != VPX_RC_FIRST_PASS) + { + int q; + + vpx_codec_control(&stream->encoder, VP8E_GET_LAST_QUANTIZER_64, &q); + ctx_exit_on_error(&stream->encoder, "Failed to read quantizer"); + stream->counts[q]++; + } +} + + +static void get_cx_data(struct stream_state *stream, + struct global_config *global, + int *got_data) +{ + const vpx_codec_cx_pkt_t *pkt; + const struct vpx_codec_enc_cfg *cfg = &stream->config.cfg; + vpx_codec_iter_t iter = NULL; + + while ((pkt = vpx_codec_get_cx_data(&stream->encoder, &iter))) + { + static size_t fsize = 0; + static off_t ivf_header_pos = 0; + + *got_data = 1; + + switch (pkt->kind) + { + case VPX_CODEC_CX_FRAME_PKT: + if (!(pkt->data.frame.flags & VPX_FRAME_IS_FRAGMENT)) + { + stream->frames_out++; + } + fprintf(stderr, " %6luF", + (unsigned long)pkt->data.frame.sz); + + update_rate_histogram(&stream->rate_hist, cfg, pkt); + if(stream->config.write_webm) + { + /* Update the hash */ + if(!stream->ebml.debug) + stream->hash = murmur(pkt->data.frame.buf, + pkt->data.frame.sz, stream->hash); + + write_webm_block(&stream->ebml, cfg, pkt); + } + else + { + if (pkt->data.frame.partition_id <= 0) + { + ivf_header_pos = ftello(stream->file); + fsize = pkt->data.frame.sz; + + write_ivf_frame_header(stream->file, pkt); + } + else + { + fsize += pkt->data.frame.sz; + + if (!(pkt->data.frame.flags & VPX_FRAME_IS_FRAGMENT)) + { + off_t currpos = ftello(stream->file); + fseeko(stream->file, ivf_header_pos, SEEK_SET); + write_ivf_frame_size(stream->file, fsize); + fseeko(stream->file, currpos, SEEK_SET); + } + } + + fwrite(pkt->data.frame.buf, 1, + pkt->data.frame.sz, stream->file); + } + stream->nbytes += pkt->data.raw.sz; + break; + case VPX_CODEC_STATS_PKT: + stream->frames_out++; + fprintf(stderr, " %6luS", + (unsigned long)pkt->data.twopass_stats.sz); + stats_write(&stream->stats, + pkt->data.twopass_stats.buf, + pkt->data.twopass_stats.sz); + stream->nbytes += pkt->data.raw.sz; + break; + case VPX_CODEC_PSNR_PKT: + + if (global->show_psnr) + { + int i; + + stream->psnr_sse_total += pkt->data.psnr.sse[0]; + stream->psnr_samples_total += pkt->data.psnr.samples[0]; + for (i = 0; i < 4; i++) + { + fprintf(stderr, "%.3lf ", pkt->data.psnr.psnr[i]); + stream->psnr_totals[i] += pkt->data.psnr.psnr[i]; + } + stream->psnr_count++; + } + + break; + default: + break; + } + } +} + + +static void show_psnr(struct stream_state *stream) +{ + int i; + double ovpsnr; + + if (!stream->psnr_count) + return; + + fprintf(stderr, "Stream %d PSNR (Overall/Avg/Y/U/V)", stream->index); + ovpsnr = vp8_mse2psnr(stream->psnr_samples_total, 255.0, + stream->psnr_sse_total); + fprintf(stderr, " %.3lf", ovpsnr); + + for (i = 0; i < 4; i++) + { + fprintf(stderr, " %.3lf", stream->psnr_totals[i]/stream->psnr_count); + } + fprintf(stderr, "\n"); +} + + +float usec_to_fps(uint64_t usec, unsigned int frames) +{ + return usec > 0 ? (float)frames * 1000000.0 / (float)usec : 0; +} + + +int main(int argc, const char **argv_) +{ + int pass; + vpx_image_t raw; + int frame_avail, got_data; + + struct input_state input = {0}; + struct global_config global; + struct stream_state *streams = NULL; + char **argv, **argi; + unsigned long cx_time = 0; + int stream_cnt = 0; + + exec_name = argv_[0]; + + if (argc < 3) + usage_exit(); + + /* Setup default input stream settings */ + input.framerate.num = 30; + input.framerate.den = 1; + input.use_i420 = 1; + + /* First parse the global configuration values, because we want to apply + * other parameters on top of the default configuration provided by the + * codec. + */ + argv = argv_dup(argc - 1, argv_ + 1); + parse_global_config(&global, argv); + + { + /* Now parse each stream's parameters. Using a local scope here + * due to the use of 'stream' as loop variable in FOREACH_STREAM + * loops + */ + struct stream_state *stream = NULL; + + do + { + stream = new_stream(&global, stream); + stream_cnt++; + if(!streams) + streams = stream; + } while(parse_stream_params(&global, stream, argv)); + } + + /* Check for unrecognized options */ + for (argi = argv; *argi; argi++) + if (argi[0][0] == '-' && argi[0][1]) + die("Error: Unrecognized option %s\n", *argi); + + /* Handle non-option arguments */ + input.fn = argv[0]; + + if (!input.fn) + usage_exit(); + + for (pass = global.pass ? global.pass - 1 : 0; pass < global.passes; pass++) + { + int frames_in = 0; + + open_input_file(&input); + + /* If the input file doesn't specify its w/h (raw files), try to get + * the data from the first stream's configuration. + */ + if(!input.w || !input.h) + FOREACH_STREAM({ + if(stream->config.cfg.g_w && stream->config.cfg.g_h) + { + input.w = stream->config.cfg.g_w; + input.h = stream->config.cfg.g_h; + break; + } + }); + + /* Update stream configurations from the input file's parameters */ + FOREACH_STREAM(set_stream_dimensions(stream, input.w, input.h)); + FOREACH_STREAM(validate_stream_config(stream)); + + /* Ensure that --passes and --pass are consistent. If --pass is set and + * --passes=2, ensure --fpf was set. + */ + if (global.pass && global.passes == 2) + FOREACH_STREAM({ + if(!stream->config.stats_fn) + die("Stream %d: Must specify --fpf when --pass=%d" + " and --passes=2\n", stream->index, global.pass); + }); + + + /* Use the frame rate from the file only if none was specified + * on the command-line. + */ + if (!global.have_framerate) + global.framerate = input.framerate; + + FOREACH_STREAM(set_default_kf_interval(stream, &global)); + + /* Show configuration */ + if (global.verbose && pass == 0) + FOREACH_STREAM(show_stream_config(stream, &global, &input)); + + if(pass == (global.pass ? global.pass - 1 : 0)) { + if (input.file_type == FILE_TYPE_Y4M) + /*The Y4M reader does its own allocation. + Just initialize this here to avoid problems if we never read any + frames.*/ + memset(&raw, 0, sizeof(raw)); + else + vpx_img_alloc(&raw, + input.use_i420 ? VPX_IMG_FMT_I420 + : VPX_IMG_FMT_YV12, + input.w, input.h, 32); + + FOREACH_STREAM(init_rate_histogram(&stream->rate_hist, + &stream->config.cfg, + &global.framerate)); + } + + FOREACH_STREAM(open_output_file(stream, &global)); + FOREACH_STREAM(setup_pass(stream, &global, pass)); + FOREACH_STREAM(initialize_encoder(stream, &global)); + + frame_avail = 1; + got_data = 0; + + while (frame_avail || got_data) + { + struct vpx_usec_timer timer; + + if (!global.limit || frames_in < global.limit) + { + frame_avail = read_frame(&input, &raw); + + if (frame_avail) + frames_in++; + + if(stream_cnt == 1) + fprintf(stderr, + "\rPass %d/%d frame %4d/%-4d %7"PRId64"B \033[K", + pass + 1, global.passes, frames_in, + streams->frames_out, (int64_t)streams->nbytes); + else + fprintf(stderr, + "\rPass %d/%d frame %4d %7lu %s (%.2f fps)\033[K", + pass + 1, global.passes, frames_in, + cx_time > 9999999 ? cx_time / 1000 : cx_time, + cx_time > 9999999 ? "ms" : "us", + usec_to_fps(cx_time, frames_in)); + + } + else + frame_avail = 0; + + vpx_usec_timer_start(&timer); + FOREACH_STREAM(encode_frame(stream, &global, + frame_avail ? &raw : NULL, + frames_in)); + vpx_usec_timer_mark(&timer); + cx_time += vpx_usec_timer_elapsed(&timer); + + FOREACH_STREAM(update_quantizer_histogram(stream)); + + got_data = 0; + FOREACH_STREAM(get_cx_data(stream, &global, &got_data)); + + fflush(stdout); + } + + if(stream_cnt > 1) + fprintf(stderr, "\n"); + + FOREACH_STREAM(fprintf( + stderr, + "\rPass %d/%d frame %4d/%-4d %7"PRId64"B %7lub/f %7"PRId64"b/s" + " %7"PRId64" %s (%.2f fps)\033[K\n", pass + 1, + global.passes, frames_in, stream->frames_out, (int64_t)stream->nbytes, + frames_in ? (unsigned long)(stream->nbytes * 8 / frames_in) : 0, + frames_in ? (int64_t)stream->nbytes * 8 + * (int64_t)global.framerate.num / global.framerate.den + / frames_in + : 0, + stream->cx_time > 9999999 ? stream->cx_time / 1000 : stream->cx_time, + stream->cx_time > 9999999 ? "ms" : "us", + usec_to_fps(stream->cx_time, frames_in)); + ); + + if (global.show_psnr) + FOREACH_STREAM(show_psnr(stream)); + + FOREACH_STREAM(vpx_codec_destroy(&stream->encoder)); + + close_input_file(&input); + + FOREACH_STREAM(close_output_file(stream, global.codec->fourcc)); + + FOREACH_STREAM(stats_close(&stream->stats, global.passes-1)); + + if (global.pass) + break; + } + + if (global.show_q_hist_buckets) + FOREACH_STREAM(show_q_histogram(stream->counts, + global.show_q_hist_buckets)); + + if (global.show_rate_hist_buckets) + FOREACH_STREAM(show_rate_histogram(&stream->rate_hist, + &stream->config.cfg, + global.show_rate_hist_buckets)); + FOREACH_STREAM(destroy_rate_histogram(&stream->rate_hist)); + + vpx_img_free(&raw); + free(argv); + free(streams); + return EXIT_SUCCESS; +} diff --git a/wcap-decode/wcap-decode.c b/wcap-decode/wcap-decode.c new file mode 100644 index 0000000..4f1d870 --- /dev/null +++ b/wcap-decode/wcap-decode.c @@ -0,0 +1,218 @@ +/* + * Copyright © 2012 Intel Corporation + * + * Permission to use, copy, modify, distribute, and sell this software and + * its documentation for any purpose is hereby granted without fee, provided + * that the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of the copyright holders not be used in + * advertising or publicity pertaining to distribution of the software + * without specific, written prior permission. The copyright holders make + * no representations about the suitability of this software for any + * purpose. It is provided "as is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER + * RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF + * CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN + * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +struct wcap_header { + uint32_t width, height; +}; + +struct wcap_frame_header { + uint32_t msecs; + uint32_t nrects; +}; + +struct wcap_rectangle { + int32_t x1, y1, x2, y2; +}; + +struct wcap_decoder { + int fd; + size_t size; + void *map, *p, *end; + uint32_t *frame; + int width, height; +}; + +static void +wcap_decoder_decode_rectangle(struct wcap_decoder *decoder, + struct wcap_rectangle *rect) +{ + uint32_t v, *p = decoder->p, *d; + int width = rect->x2 - rect->x1, height = rect->y2 - rect->y1; + int x, i, j, k, l, count = width * height; + + d = decoder->frame + (rect->y2 - 1) * decoder->width; + x = rect->x1; + i = 0; + while (i < count) { + v = *p++; + l = v >> 24; + if (l < 0xe0) { + j = l + 1; + } else { + j = 1 << (l - 0xe0 + 7); + } + + for (k = 0; k < j; k++) { + d[x] = (d[x] + v) | 0xff000000; + x++; + if (x == rect->x2) { + x = rect->x1; + d -= decoder->width; + } + } + i += j; + } + + if (i != count) + printf("rle encoding longer than expected (%d expected %d)\n", + i, count); + + decoder->p = p; +} + +static int +wcap_decoder_get_frame(struct wcap_decoder *decoder) +{ + struct wcap_rectangle *rects; + struct wcap_frame_header *header; + uint32_t *s; + uint32_t i; + int width, height; + + if (decoder->p == decoder->end) + return 0; + + header = decoder->p; + + rects = (void *) (header + 1); + decoder->p = (uint32_t *) (rects + header->nrects); + for (i = 0; i < header->nrects; i++) { + width = rects[i].x2 - rects[i].x1; + height = rects[i].y2 - rects[i].y1; + wcap_decoder_decode_rectangle(decoder, &rects[i]); + } + + return 1; +} + +struct wcap_decoder * +wcap_decoder_create(const char *filename) +{ + struct wcap_decoder *decoder; + struct wcap_header *header; + int frame_size; + struct stat buf; + + decoder = malloc(sizeof *decoder); + if (decoder == NULL) + return NULL; + + decoder->fd = open(filename, O_RDONLY); + if (decoder->fd == -1) + return NULL; + + fstat(decoder->fd, &buf); + decoder->size = buf.st_size; + decoder->map = mmap(NULL, decoder->size, + PROT_READ, MAP_PRIVATE, decoder->fd, 0); + + header = decoder->map; + decoder->width = header->width; + decoder->height = header->height; + decoder->p = header + 1; + decoder->end = decoder->map + decoder->size; + + frame_size = header->width * header->height * 4; + decoder->frame = malloc(frame_size); + memset(decoder->frame, 0, frame_size); + + return decoder; +} + +void +wcap_decoder_destroy(struct wcap_decoder *decoder) +{ + munmap(decoder->map, decoder->size); + free(decoder->frame); + free(decoder); +} + +static void +write_png(struct wcap_decoder *decoder, const char *filename) +{ + cairo_surface_t *surface; + + surface = cairo_image_surface_create_for_data((unsigned char *) decoder->frame, + CAIRO_FORMAT_ARGB32, + decoder->width, + decoder->height, + decoder->width * 4); + cairo_surface_write_to_png(surface, filename); + cairo_surface_destroy(surface); +} + +static void +write_all_pngs(struct wcap_decoder *decoder, int frame) +{ + char filename[200]; + int i; + + i = 0; + while (wcap_decoder_get_frame(decoder)) { + if (i == frame || frame == -1) { + snprintf(filename, sizeof filename, + "wcap-frame-%d.png", i); + write_png(decoder, filename); + printf("wrote %s\n", filename); + } + i++; + } +} + +int main(int argc, char *argv[]) +{ + struct wcap_decoder *decoder; + int i, output_frame; + + if (argc != 2 && argc != 3) { + fprintf(stderr, "usage: wcap-decode WCAP_FILE [FRAME]\n"); + return 1; + } + + decoder = wcap_decoder_create(argv[1]); + output_frame = -1; + if (argc == 3) + output_frame = strtol(argv[2], NULL, 0); + + if (0) + write_all_pngs(decoder, -1); + else + write_webm(decoder); + + printf("wcap file: size %dx%d, %d frames\n", + decoder->width, decoder->height, i); + + wcap_decoder_destroy(decoder); +} diff --git a/wcap-decode/y4minput.c b/wcap-decode/y4minput.c new file mode 100644 index 0000000..dd51421 --- /dev/null +++ b/wcap-decode/y4minput.c @@ -0,0 +1,871 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + * + * Based on code from the OggTheora software codec source code, + * Copyright (C) 2002-2010 The Xiph.Org Foundation and contributors. + */ +#include +#include +#include "y4minput.h" + +static int y4m_parse_tags(y4m_input *_y4m,char *_tags){ + int got_w; + int got_h; + int got_fps; + int got_interlace; + int got_par; + int got_chroma; + char *p; + char *q; + got_w=got_h=got_fps=got_interlace=got_par=got_chroma=0; + for(p=_tags;;p=q){ + /*Skip any leading spaces.*/ + while(*p==' ')p++; + /*If that's all we have, stop.*/ + if(p[0]=='\0')break; + /*Find the end of this tag.*/ + for(q=p+1;*q!='\0'&&*q!=' ';q++); + /*Process the tag.*/ + switch(p[0]){ + case 'W':{ + if(sscanf(p+1,"%d",&_y4m->pic_w)!=1)return -1; + got_w=1; + }break; + case 'H':{ + if(sscanf(p+1,"%d",&_y4m->pic_h)!=1)return -1; + got_h=1; + }break; + case 'F':{ + if(sscanf(p+1,"%d:%d",&_y4m->fps_n,&_y4m->fps_d)!=2){ + return -1; + } + got_fps=1; + }break; + case 'I':{ + _y4m->interlace=p[1]; + got_interlace=1; + }break; + case 'A':{ + if(sscanf(p+1,"%d:%d",&_y4m->par_n,&_y4m->par_d)!=2){ + return -1; + } + got_par=1; + }break; + case 'C':{ + if(q-p>16)return -1; + memcpy(_y4m->chroma_type,p+1,q-p-1); + _y4m->chroma_type[q-p-1]='\0'; + got_chroma=1; + }break; + /*Ignore unknown tags.*/ + } + } + if(!got_w||!got_h||!got_fps)return -1; + if(!got_interlace)_y4m->interlace='?'; + if(!got_par)_y4m->par_n=_y4m->par_d=0; + /*Chroma-type is not specified in older files, e.g., those generated by + mplayer.*/ + if(!got_chroma)strcpy(_y4m->chroma_type,"420"); + return 0; +} + + + +/*All anti-aliasing filters in the following conversion functions are based on + one of two window functions: + The 6-tap Lanczos window (for down-sampling and shifts): + sinc(\pi*t)*sinc(\pi*t/3), |t|<3 (sinc(t)==sin(t)/t) + 0, |t|>=3 + The 4-tap Mitchell window (for up-sampling): + 7|t|^3-12|t|^2+16/3, |t|<1 + -(7/3)|x|^3+12|x|^2-20|x|+32/3, |t|<2 + 0, |t|>=2 + The number of taps is intentionally kept small to reduce computational + overhead and limit ringing. + + The taps from these filters are scaled so that their sum is 1, and the result + is scaled by 128 and rounded to integers to create a filter whose + intermediate values fit inside 16 bits. + Coefficients are rounded in such a way as to ensure their sum is still 128, + which is usually equivalent to normal rounding. + + Conversions which require both horizontal and vertical filtering could + have these steps pipelined, for less memory consumption and better cache + performance, but we do them separately for simplicity.*/ + +#define OC_MINI(_a,_b) ((_a)>(_b)?(_b):(_a)) +#define OC_MAXI(_a,_b) ((_a)<(_b)?(_b):(_a)) +#define OC_CLAMPI(_a,_b,_c) (OC_MAXI(_a,OC_MINI(_b,_c))) + +/*420jpeg chroma samples are sited like: + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + + 420mpeg2 chroma samples are sited like: + Y-------Y-------Y-------Y------- + | | | | + BR | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + Y-------Y-------Y-------Y------- + | | | | + BR | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + + We use a resampling filter to shift the site locations one quarter pixel (at + the chroma plane's resolution) to the right. + The 4:2:2 modes look exactly the same, except there are twice as many chroma + lines, and they are vertically co-sited with the luma samples in both the + mpeg2 and jpeg cases (thus requiring no vertical resampling).*/ +static void y4m_42xmpeg2_42xjpeg_helper(unsigned char *_dst, + const unsigned char *_src,int _c_w,int _c_h){ + int y; + int x; + for(y=0;y<_c_h;y++){ + /*Filter: [4 -17 114 35 -9 1]/128, derived from a 6-tap Lanczos + window.*/ + for(x=0;x>7,255); + } + for(;x<_c_w-3;x++){ + _dst[x]=(unsigned char)OC_CLAMPI(0,(4*_src[x-2]-17*_src[x-1]+ + 114*_src[x]+35*_src[x+1]-9*_src[x+2]+_src[x+3]+64)>>7,255); + } + for(;x<_c_w;x++){ + _dst[x]=(unsigned char)OC_CLAMPI(0,(4*_src[x-2]-17*_src[x-1]+ + 114*_src[x]+35*_src[OC_MINI(x+1,_c_w-1)]-9*_src[OC_MINI(x+2,_c_w-1)]+ + _src[_c_w-1]+64)>>7,255); + } + _dst+=_c_w; + _src+=_c_w; + } +} + +/*Handles both 422 and 420mpeg2 to 422jpeg and 420jpeg, respectively.*/ +static void y4m_convert_42xmpeg2_42xjpeg(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ + int c_w; + int c_h; + int c_sz; + int pli; + /*Skip past the luma data.*/ + _dst+=_y4m->pic_w*_y4m->pic_h; + /*Compute the size of each chroma plane.*/ + c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h; + c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v; + c_sz=c_w*c_h; + for(pli=1;pli<3;pli++){ + y4m_42xmpeg2_42xjpeg_helper(_dst,_aux,c_w,c_h); + _dst+=c_sz; + _aux+=c_sz; + } +} + +/*This format is only used for interlaced content, but is included for + completeness. + + 420jpeg chroma samples are sited like: + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + + 420paldv chroma samples are sited like: + YR------Y-------YR------Y------- + | | | | + | | | | + | | | | + YB------Y-------YB------Y------- + | | | | + | | | | + | | | | + YR------Y-------YR------Y------- + | | | | + | | | | + | | | | + YB------Y-------YB------Y------- + | | | | + | | | | + | | | | + + We use a resampling filter to shift the site locations one quarter pixel (at + the chroma plane's resolution) to the right. + Then we use another filter to move the C_r location down one quarter pixel, + and the C_b location up one quarter pixel.*/ +static void y4m_convert_42xpaldv_42xjpeg(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ + unsigned char *tmp; + int c_w; + int c_h; + int c_sz; + int pli; + int y; + int x; + /*Skip past the luma data.*/ + _dst+=_y4m->pic_w*_y4m->pic_h; + /*Compute the size of each chroma plane.*/ + c_w=(_y4m->pic_w+1)/2; + c_h=(_y4m->pic_h+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h; + c_sz=c_w*c_h; + tmp=_aux+2*c_sz; + for(pli=1;pli<3;pli++){ + /*First do the horizontal re-sampling. + This is the same as the mpeg2 case, except that after the horizontal + case, we need to apply a second vertical filter.*/ + y4m_42xmpeg2_42xjpeg_helper(tmp,_aux,c_w,c_h); + _aux+=c_sz; + switch(pli){ + case 1:{ + /*Slide C_b up a quarter-pel. + This is the same filter used above, but in the other order.*/ + for(x=0;x>7,255); + } + for(;y>7,255); + } + for(;y>7,255); + } + _dst++; + tmp++; + } + _dst+=c_sz-c_w; + tmp-=c_w; + }break; + case 2:{ + /*Slide C_r down a quarter-pel. + This is the same as the horizontal filter.*/ + for(x=0;x>7,255); + } + for(;y>7,255); + } + for(;y>7,255); + } + _dst++; + tmp++; + } + }break; + } + /*For actual interlaced material, this would have to be done separately on + each field, and the shift amounts would be different. + C_r moves down 1/8, C_b up 3/8 in the top field, and C_r moves down 3/8, + C_b up 1/8 in the bottom field. + The corresponding filters would be: + Down 1/8 (reverse order for up): [3 -11 125 15 -4 0]/128 + Down 3/8 (reverse order for up): [4 -19 98 56 -13 2]/128*/ + } +} + +/*Perform vertical filtering to reduce a single plane from 4:2:2 to 4:2:0. + This is used as a helper by several converation routines.*/ +static void y4m_422jpeg_420jpeg_helper(unsigned char *_dst, + const unsigned char *_src,int _c_w,int _c_h){ + int y; + int x; + /*Filter: [3 -17 78 78 -17 3]/128, derived from a 6-tap Lanczos window.*/ + for(x=0;x<_c_w;x++){ + for(y=0;y>1)*_c_w]=OC_CLAMPI(0,(64*_src[0] + +78*_src[OC_MINI(1,_c_h-1)*_c_w] + -17*_src[OC_MINI(2,_c_h-1)*_c_w] + +3*_src[OC_MINI(3,_c_h-1)*_c_w]+64)>>7,255); + } + for(;y<_c_h-3;y+=2){ + _dst[(y>>1)*_c_w]=OC_CLAMPI(0,(3*(_src[(y-2)*_c_w]+_src[(y+3)*_c_w]) + -17*(_src[(y-1)*_c_w]+_src[(y+2)*_c_w]) + +78*(_src[y*_c_w]+_src[(y+1)*_c_w])+64)>>7,255); + } + for(;y<_c_h;y+=2){ + _dst[(y>>1)*_c_w]=OC_CLAMPI(0,(3*(_src[(y-2)*_c_w] + +_src[(_c_h-1)*_c_w])-17*(_src[(y-1)*_c_w] + +_src[OC_MINI(y+2,_c_h-1)*_c_w]) + +78*(_src[y*_c_w]+_src[OC_MINI(y+1,_c_h-1)*_c_w])+64)>>7,255); + } + _src++; + _dst++; + } +} + +/*420jpeg chroma samples are sited like: + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + + 422jpeg chroma samples are sited like: + Y---BR--Y-------Y---BR--Y------- + | | | | + | | | | + | | | | + Y---BR--Y-------Y---BR--Y------- + | | | | + | | | | + | | | | + Y---BR--Y-------Y---BR--Y------- + | | | | + | | | | + | | | | + Y---BR--Y-------Y---BR--Y------- + | | | | + | | | | + | | | | + + We use a resampling filter to decimate the chroma planes by two in the + vertical direction.*/ +static void y4m_convert_422jpeg_420jpeg(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ + int c_w; + int c_h; + int c_sz; + int dst_c_w; + int dst_c_h; + int dst_c_sz; + int pli; + /*Skip past the luma data.*/ + _dst+=_y4m->pic_w*_y4m->pic_h; + /*Compute the size of each chroma plane.*/ + c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h; + c_h=_y4m->pic_h; + dst_c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h; + dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v; + c_sz=c_w*c_h; + dst_c_sz=dst_c_w*dst_c_h; + for(pli=1;pli<3;pli++){ + y4m_422jpeg_420jpeg_helper(_dst,_aux,c_w,c_h); + _aux+=c_sz; + _dst+=dst_c_sz; + } +} + +/*420jpeg chroma samples are sited like: + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + + 422 chroma samples are sited like: + YBR-----Y-------YBR-----Y------- + | | | | + | | | | + | | | | + YBR-----Y-------YBR-----Y------- + | | | | + | | | | + | | | | + YBR-----Y-------YBR-----Y------- + | | | | + | | | | + | | | | + YBR-----Y-------YBR-----Y------- + | | | | + | | | | + | | | | + + We use a resampling filter to shift the original site locations one quarter + pixel (at the original chroma resolution) to the right. + Then we use a second resampling filter to decimate the chroma planes by two + in the vertical direction.*/ +static void y4m_convert_422_420jpeg(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ + unsigned char *tmp; + int c_w; + int c_h; + int c_sz; + int dst_c_h; + int dst_c_sz; + int pli; + /*Skip past the luma data.*/ + _dst+=_y4m->pic_w*_y4m->pic_h; + /*Compute the size of each chroma plane.*/ + c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h; + c_h=_y4m->pic_h; + dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v; + c_sz=c_w*c_h; + dst_c_sz=c_w*dst_c_h; + tmp=_aux+2*c_sz; + for(pli=1;pli<3;pli++){ + /*In reality, the horizontal and vertical steps could be pipelined, for + less memory consumption and better cache performance, but we do them + separately for simplicity.*/ + /*First do horizontal filtering (convert to 422jpeg)*/ + y4m_42xmpeg2_42xjpeg_helper(tmp,_aux,c_w,c_h); + /*Now do the vertical filtering.*/ + y4m_422jpeg_420jpeg_helper(_dst,tmp,c_w,c_h); + _aux+=c_sz; + _dst+=dst_c_sz; + } +} + +/*420jpeg chroma samples are sited like: + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + + 411 chroma samples are sited like: + YBR-----Y-------Y-------Y------- + | | | | + | | | | + | | | | + YBR-----Y-------Y-------Y------- + | | | | + | | | | + | | | | + YBR-----Y-------Y-------Y------- + | | | | + | | | | + | | | | + YBR-----Y-------Y-------Y------- + | | | | + | | | | + | | | | + + We use a filter to resample at site locations one eighth pixel (at the source + chroma plane's horizontal resolution) and five eighths of a pixel to the + right. + Then we use another filter to decimate the planes by 2 in the vertical + direction.*/ +static void y4m_convert_411_420jpeg(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ + unsigned char *tmp; + int c_w; + int c_h; + int c_sz; + int dst_c_w; + int dst_c_h; + int dst_c_sz; + int tmp_sz; + int pli; + int y; + int x; + /*Skip past the luma data.*/ + _dst+=_y4m->pic_w*_y4m->pic_h; + /*Compute the size of each chroma plane.*/ + c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h; + c_h=_y4m->pic_h; + dst_c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h; + dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v; + c_sz=c_w*c_h; + dst_c_sz=dst_c_w*dst_c_h; + tmp_sz=dst_c_w*c_h; + tmp=_aux+2*c_sz; + for(pli=1;pli<3;pli++){ + /*In reality, the horizontal and vertical steps could be pipelined, for + less memory consumption and better cache performance, but we do them + separately for simplicity.*/ + /*First do horizontal filtering (convert to 422jpeg)*/ + for(y=0;y>7,255); + tmp[x<<1|1]=(unsigned char)OC_CLAMPI(0,(47*_aux[0] + +86*_aux[OC_MINI(1,c_w-1)]-5*_aux[OC_MINI(2,c_w-1)]+64)>>7,255); + } + for(;x>7,255); + tmp[x<<1|1]=(unsigned char)OC_CLAMPI(0,(-3*_aux[x-1]+50*_aux[x] + +86*_aux[x+1]-5*_aux[x+2]+64)>>7,255); + } + for(;x>7,255); + if((x<<1|1)>7,255); + } + } + tmp+=dst_c_w; + _aux+=c_w; + } + tmp-=tmp_sz; + /*Now do the vertical filtering.*/ + y4m_422jpeg_420jpeg_helper(_dst,tmp,dst_c_w,c_h); + _dst+=dst_c_sz; + } +} + +/*Convert 444 to 420jpeg.*/ +static void y4m_convert_444_420jpeg(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ + unsigned char *tmp; + int c_w; + int c_h; + int c_sz; + int dst_c_w; + int dst_c_h; + int dst_c_sz; + int tmp_sz; + int pli; + int y; + int x; + /*Skip past the luma data.*/ + _dst+=_y4m->pic_w*_y4m->pic_h; + /*Compute the size of each chroma plane.*/ + c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h; + c_h=_y4m->pic_h; + dst_c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h; + dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v; + c_sz=c_w*c_h; + dst_c_sz=dst_c_w*dst_c_h; + tmp_sz=dst_c_w*c_h; + tmp=_aux+2*c_sz; + for(pli=1;pli<3;pli++){ + /*Filter: [3 -17 78 78 -17 3]/128, derived from a 6-tap Lanczos window.*/ + for(y=0;y>1]=OC_CLAMPI(0,(64*_aux[0]+78*_aux[OC_MINI(1,c_w-1)] + -17*_aux[OC_MINI(2,c_w-1)] + +3*_aux[OC_MINI(3,c_w-1)]+64)>>7,255); + } + for(;x>1]=OC_CLAMPI(0,(3*(_aux[x-2]+_aux[x+3]) + -17*(_aux[x-1]+_aux[x+2])+78*(_aux[x]+_aux[x+1])+64)>>7,255); + } + for(;x>1]=OC_CLAMPI(0,(3*(_aux[x-2]+_aux[c_w-1])- + 17*(_aux[x-1]+_aux[OC_MINI(x+2,c_w-1)])+ + 78*(_aux[x]+_aux[OC_MINI(x+1,c_w-1)])+64)>>7,255); + } + tmp+=dst_c_w; + _aux+=c_w; + } + tmp-=tmp_sz; + /*Now do the vertical filtering.*/ + y4m_422jpeg_420jpeg_helper(_dst,tmp,dst_c_w,c_h); + _dst+=dst_c_sz; + } +} + +/*The image is padded with empty chroma components at 4:2:0.*/ +static void y4m_convert_mono_420jpeg(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ + int c_sz; + _dst+=_y4m->pic_w*_y4m->pic_h; + c_sz=((_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h)* + ((_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v); + memset(_dst,128,c_sz*2); +} + +/*No conversion function needed.*/ +static void y4m_convert_null(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ +} + +int y4m_input_open(y4m_input *_y4m,FILE *_fin,char *_skip,int _nskip){ + char buffer[80]; + int ret; + int i; + /*Read until newline, or 80 cols, whichever happens first.*/ + for(i=0;i<79;i++){ + if(_nskip>0){ + buffer[i]=*_skip++; + _nskip--; + } + else{ + ret=fread(buffer+i,1,1,_fin); + if(ret<1)return -1; + } + if(buffer[i]=='\n')break; + } + /*We skipped too much header data.*/ + if(_nskip>0)return -1; + if(i==79){ + fprintf(stderr,"Error parsing header; not a YUV2MPEG2 file?\n"); + return -1; + } + buffer[i]='\0'; + if(memcmp(buffer,"YUV4MPEG",8)){ + fprintf(stderr,"Incomplete magic for YUV4MPEG file.\n"); + return -1; + } + if(buffer[8]!='2'){ + fprintf(stderr,"Incorrect YUV input file version; YUV4MPEG2 required.\n"); + } + ret=y4m_parse_tags(_y4m,buffer+5); + if(ret<0){ + fprintf(stderr,"Error parsing YUV4MPEG2 header.\n"); + return ret; + } + if(_y4m->interlace=='?'){ + fprintf(stderr,"Warning: Input video interlacing format unknown; " + "assuming progressive scan.\n"); + } + else if(_y4m->interlace!='p'){ + fprintf(stderr,"Input video is interlaced; " + "Only progressive scan handled.\n"); + return -1; + } + if(strcmp(_y4m->chroma_type,"420")==0|| + strcmp(_y4m->chroma_type,"420jpeg")==0){ + _y4m->src_c_dec_h=_y4m->dst_c_dec_h=_y4m->src_c_dec_v=_y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h + +2*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2); + /*Natively supported: no conversion required.*/ + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=0; + _y4m->convert=y4m_convert_null; + } + else if(strcmp(_y4m->chroma_type,"420mpeg2")==0){ + _y4m->src_c_dec_h=_y4m->dst_c_dec_h=_y4m->src_c_dec_v=_y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*Chroma filter required: read into the aux buf first.*/ + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz= + 2*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2); + _y4m->convert=y4m_convert_42xmpeg2_42xjpeg; + } + else if(strcmp(_y4m->chroma_type,"420paldv")==0){ + _y4m->src_c_dec_h=_y4m->dst_c_dec_h=_y4m->src_c_dec_v=_y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*Chroma filter required: read into the aux buf first. + We need to make two filter passes, so we need some extra space in the + aux buffer.*/ + _y4m->aux_buf_sz=3*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2); + _y4m->aux_buf_read_sz=2*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2); + _y4m->convert=y4m_convert_42xpaldv_42xjpeg; + } + else if(strcmp(_y4m->chroma_type,"422jpeg")==0){ + _y4m->src_c_dec_h=_y4m->dst_c_dec_h=2; + _y4m->src_c_dec_v=1; + _y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*Chroma filter required: read into the aux buf first.*/ + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=2*((_y4m->pic_w+1)/2)*_y4m->pic_h; + _y4m->convert=y4m_convert_422jpeg_420jpeg; + } + else if(strcmp(_y4m->chroma_type,"422")==0){ + _y4m->src_c_dec_h=_y4m->dst_c_dec_h=2; + _y4m->src_c_dec_v=1; + _y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*Chroma filter required: read into the aux buf first. + We need to make two filter passes, so we need some extra space in the + aux buffer.*/ + _y4m->aux_buf_read_sz=2*((_y4m->pic_w+1)/2)*_y4m->pic_h; + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz+((_y4m->pic_w+1)/2)*_y4m->pic_h; + _y4m->convert=y4m_convert_422_420jpeg; + } + else if(strcmp(_y4m->chroma_type,"411")==0){ + _y4m->src_c_dec_h=4; + _y4m->dst_c_dec_h=2; + _y4m->src_c_dec_v=1; + _y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*Chroma filter required: read into the aux buf first. + We need to make two filter passes, so we need some extra space in the + aux buffer.*/ + _y4m->aux_buf_read_sz=2*((_y4m->pic_w+3)/4)*_y4m->pic_h; + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz+((_y4m->pic_w+1)/2)*_y4m->pic_h; + _y4m->convert=y4m_convert_411_420jpeg; + } + else if(strcmp(_y4m->chroma_type,"444")==0){ + _y4m->src_c_dec_h=1; + _y4m->dst_c_dec_h=2; + _y4m->src_c_dec_v=1; + _y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*Chroma filter required: read into the aux buf first. + We need to make two filter passes, so we need some extra space in the + aux buffer.*/ + _y4m->aux_buf_read_sz=2*_y4m->pic_w*_y4m->pic_h; + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz+((_y4m->pic_w+1)/2)*_y4m->pic_h; + _y4m->convert=y4m_convert_444_420jpeg; + } + else if(strcmp(_y4m->chroma_type,"444alpha")==0){ + _y4m->src_c_dec_h=1; + _y4m->dst_c_dec_h=2; + _y4m->src_c_dec_v=1; + _y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*Chroma filter required: read into the aux buf first. + We need to make two filter passes, so we need some extra space in the + aux buffer. + The extra plane also gets read into the aux buf. + It will be discarded.*/ + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=3*_y4m->pic_w*_y4m->pic_h; + _y4m->convert=y4m_convert_444_420jpeg; + } + else if(strcmp(_y4m->chroma_type,"mono")==0){ + _y4m->src_c_dec_h=_y4m->src_c_dec_v=0; + _y4m->dst_c_dec_h=_y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*No extra space required, but we need to clear the chroma planes.*/ + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=0; + _y4m->convert=y4m_convert_mono_420jpeg; + } + else{ + fprintf(stderr,"Unknown chroma sampling type: %s\n",_y4m->chroma_type); + return -1; + } + /*The size of the final frame buffers is always computed from the + destination chroma decimation type.*/ + _y4m->dst_buf_sz=_y4m->pic_w*_y4m->pic_h + +2*((_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h)* + ((_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v); + _y4m->dst_buf=(unsigned char *)malloc(_y4m->dst_buf_sz); + _y4m->aux_buf=(unsigned char *)malloc(_y4m->aux_buf_sz); + return 0; +} + +void y4m_input_close(y4m_input *_y4m){ + free(_y4m->dst_buf); + free(_y4m->aux_buf); +} + +int y4m_input_fetch_frame(y4m_input *_y4m,FILE *_fin,vpx_image_t *_img){ + char frame[6]; + int pic_sz; + int c_w; + int c_h; + int c_sz; + int ret; + /*Read and skip the frame header.*/ + ret=fread(frame,1,6,_fin); + if(ret<6)return 0; + if(memcmp(frame,"FRAME",5)){ + fprintf(stderr,"Loss of framing in Y4M input data\n"); + return -1; + } + if(frame[5]!='\n'){ + char c; + int j; + for(j=0;j<79&&fread(&c,1,1,_fin)&&c!='\n';j++); + if(j==79){ + fprintf(stderr,"Error parsing Y4M frame header\n"); + return -1; + } + } + /*Read the frame data that needs no conversion.*/ + if(fread(_y4m->dst_buf,1,_y4m->dst_buf_read_sz,_fin)!=_y4m->dst_buf_read_sz){ + fprintf(stderr,"Error reading Y4M frame data.\n"); + return -1; + } + /*Read the frame data that does need conversion.*/ + if(fread(_y4m->aux_buf,1,_y4m->aux_buf_read_sz,_fin)!=_y4m->aux_buf_read_sz){ + fprintf(stderr,"Error reading Y4M frame data.\n"); + return -1; + } + /*Now convert the just read frame.*/ + (*_y4m->convert)(_y4m,_y4m->dst_buf,_y4m->aux_buf); + /*Fill in the frame buffer pointers. + We don't use vpx_img_wrap() because it forces padding for odd picture + sizes, which would require a separate fread call for every row.*/ + memset(_img,0,sizeof(*_img)); + /*Y4M has the planes in Y'CbCr order, which libvpx calls Y, U, and V.*/ + _img->fmt=IMG_FMT_I420; + _img->w=_img->d_w=_y4m->pic_w; + _img->h=_img->d_h=_y4m->pic_h; + /*This is hard-coded to 4:2:0 for now, as that's all VP8 supports.*/ + _img->x_chroma_shift=1; + _img->y_chroma_shift=1; + _img->bps=12; + /*Set up the buffer pointers.*/ + pic_sz=_y4m->pic_w*_y4m->pic_h; + c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h; + c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v; + c_sz=c_w*c_h; + _img->stride[PLANE_Y]=_y4m->pic_w; + _img->stride[PLANE_U]=_img->stride[PLANE_V]=c_w; + _img->planes[PLANE_Y]=_y4m->dst_buf; + _img->planes[PLANE_U]=_y4m->dst_buf+pic_sz; + _img->planes[PLANE_V]=_y4m->dst_buf+pic_sz+c_sz; + return 1; +} diff --git a/wcap-decode/y4minput.h b/wcap-decode/y4minput.h new file mode 100644 index 0000000..1a01bcd --- /dev/null +++ b/wcap-decode/y4minput.h @@ -0,0 +1,60 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + * + * Based on code from the OggTheora software codec source code, + * Copyright (C) 2002-2010 The Xiph.Org Foundation and contributors. + */ +#if !defined(_y4minput_H) +# define _y4minput_H (1) +# include +# include "vpx/vpx_image.h" + + + +typedef struct y4m_input y4m_input; + + + +/*The function used to perform chroma conversion.*/ +typedef void (*y4m_convert_func)(y4m_input *_y4m, + unsigned char *_dst,unsigned char *_src); + + + +struct y4m_input{ + int pic_w; + int pic_h; + int fps_n; + int fps_d; + int par_n; + int par_d; + char interlace; + int src_c_dec_h; + int src_c_dec_v; + int dst_c_dec_h; + int dst_c_dec_v; + char chroma_type[16]; + /*The size of each converted frame buffer.*/ + size_t dst_buf_sz; + /*The amount to read directly into the converted frame buffer.*/ + size_t dst_buf_read_sz; + /*The size of the auxilliary buffer.*/ + size_t aux_buf_sz; + /*The amount to read into the auxilliary buffer.*/ + size_t aux_buf_read_sz; + y4m_convert_func convert; + unsigned char *dst_buf; + unsigned char *aux_buf; +}; + +int y4m_input_open(y4m_input *_y4m,FILE *_fin,char *_skip,int _nskip); +void y4m_input_close(y4m_input *_y4m); +int y4m_input_fetch_frame(y4m_input *_y4m,FILE *_fin,vpx_image_t *img); + +#endif