vpx[dec|enc]: Extract IVF support from the apps.

- Move IVF reading support into ivfdec.c and ivfdec.h
- Move IVF writing support into ivfenc.c and ivfenc.h
- Removed IVF writing code from the SVC example in favor of ivfenc.

Change-Id: I70adf6240d0320fdd232d8546ed573f0f68dd793

examples.mk

@@ -23,6 +23,7 @@ vpxdec.SRCS                 += md5_utils.c md5_utils.h
 vpxdec.SRCS                 += vpx_ports/vpx_timer.h
 vpxdec.SRCS                 += vpx/vpx_integer.h
 vpxdec.SRCS                 += args.c args.h
+vpxdec.SRCS                 += ivfdec.c ivfdec.h
 vpxdec.SRCS                 += tools_common.c tools_common.h
 vpxdec.SRCS                 += nestegg/halloc/halloc.h
 vpxdec.SRCS                 += nestegg/halloc/src/align.h
@@ -36,6 +37,8 @@ vpxdec.GUID                  = BA5FE66F-38DD-E034-F542-B1578C5FB950
 vpxdec.DESCRIPTION           = Full featured decoder
 UTILS-$(CONFIG_ENCODERS)    += vpxenc.c
 vpxenc.SRCS                 += args.c args.h y4minput.c y4minput.h
+vpxenc.SRCS                 += ivfdec.c ivfdec.h
+vpxenc.SRCS                 += ivfenc.c ivfenc.h
 vpxenc.SRCS                 += tools_common.c tools_common.h
 vpxenc.SRCS                 += webmenc.c webmenc.h
 vpxenc.SRCS                 += vpx_ports/mem_ops.h
@@ -53,18 +56,11 @@ vp8_scalable_patterns.GUID   = 0D6A210B-F482-4D6F-8570-4A9C01ACC88C
 vp8_scalable_patterns.DESCRIPTION = Temporal Scalability Encoder
 UTILS-$(CONFIG_VP9_ENCODER)    += vp9_spatial_scalable_encoder.c
 vp9_spatial_scalable_encoder.SRCS += args.c args.h
+vp9_spatial_scalable_encoder.SRCS += ivfenc.c ivfenc.h
+vp9_spatial_scalable_encoder.SRCS += tools_common.c tools_common.h
 vp9_spatial_scalable_encoder.GUID   = 4A38598D-627D-4505-9C7B-D4020C84100D
 vp9_spatial_scalable_encoder.DESCRIPTION = Spatial Scalable Encoder
 
-# Clean up old ivfenc, ivfdec binaries.
-ifeq ($(CONFIG_MSVS),yes)
-CLEAN-OBJS += $(foreach p,$(VS_PLATFORMS),$(p)/Release/ivfenc.exe)
-CLEAN-OBJS += $(foreach p,$(VS_PLATFORMS),$(p)/Release/ivfdec.exe)
-else
-CLEAN-OBJS += ivfenc{.c.o,.c.d,.dox,.exe,}
-CLEAN-OBJS += ivfdec{.c.o,.c.d,.dox,.exe,}
-endif
-
 # XMA example disabled for now, not used in VP8
 #UTILS-$(CONFIG_DECODERS)    += example_xma.c
 #example_xma.GUID             = A955FC4A-73F1-44F7-135E-30D84D32F022

ivfdec.c

+/*
+ *  Copyright (c) 2013 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 "./ivfdec.h"
+
+#include <stdio.h>
+#include <stdlib.h>
+
+int file_is_ivf(struct VpxInputContext *input_ctx) {
+  char raw_hdr[32];
+  int is_ivf = 0;
+
+  // TODO(tomfinegan): This can eventually go away, but for now it's required
+  // because the means by which file types are detected differ in vpxdec and
+  // vpxenc.
+  rewind(input_ctx->file);
+
+  if (fread(raw_hdr, 1, 32, input_ctx->file) == 32) {
+    if (raw_hdr[0] == 'D' && raw_hdr[1] == 'K' &&
+        raw_hdr[2] == 'I' && raw_hdr[3] == 'F') {
+      is_ivf = 1;
+
+      if (mem_get_le16(raw_hdr + 4) != 0) {
+        fprintf(stderr, "Error: Unrecognized IVF version! This file may not"
+                " decode properly.");
+      }
+
+      input_ctx->fourcc = mem_get_le32(raw_hdr + 8);
+      input_ctx->width = mem_get_le16(raw_hdr + 12);
+      input_ctx->height = mem_get_le16(raw_hdr + 14);
+      input_ctx->framerate.numerator = mem_get_le32(raw_hdr + 16);
+      input_ctx->framerate.denominator = mem_get_le32(raw_hdr + 20);
+
+      /* Some versions of vpxenc used 1/(2*fps) for the timebase, so
+       * we can guess the framerate using only the timebase in this
+       * case. Other files would require reading ahead to guess the
+       * timebase, like we do for webm.
+       */
+      if (input_ctx->framerate.numerator < 1000) {
+        /* Correct for the factor of 2 applied to the timebase in the
+         * encoder.
+         */
+        if (input_ctx->framerate.numerator & 1)
+          input_ctx->framerate.denominator <<= 1;
+        else
+          input_ctx->framerate.numerator >>= 1;
+      } else {
+        /* Don't know FPS for sure, and don't have readahead code
+         * (yet?), so just default to 30fps.
+         */
+        input_ctx->framerate.numerator = 30;
+        input_ctx->framerate.denominator = 1;
+      }
+    }
+  }
+
+  if (!is_ivf)
+    rewind(input_ctx->file);
+  else
+    input_ctx->detect.position = 4;
+
+  return is_ivf;
+}
+
+int ivf_read_frame(struct VpxInputContext *input_ctx,
+                   uint8_t **buffer,
+                   size_t *bytes_read,
+                   size_t *buffer_size) {
+  char raw_header[IVF_FRAME_HDR_SZ] = {0};
+  size_t frame_size = 0;
+  FILE *infile = input_ctx->file;
+
+  if (input_ctx->file_type != FILE_TYPE_IVF)
+    return 0;
+
+  if (fread(raw_header, IVF_FRAME_HDR_SZ, 1, infile) != 1) {
+    if (!feof(infile))
+      warn("Failed to read frame size\n");
+  } else {
+    frame_size = mem_get_le32(raw_header);
+
+    if (frame_size > 256 * 1024 * 1024) {
+      warn("Read invalid frame size (%u)\n", (unsigned int)frame_size);
+      frame_size = 0;
+    }
+
+    if (frame_size > *buffer_size) {
+      uint8_t *new_buffer = realloc(*buffer, 2 * frame_size);
+
+      if (new_buffer) {
+        *buffer = new_buffer;
+        *buffer_size = 2 * frame_size;
+      } else {
+        warn("Failed to allocate compressed data buffer\n");
+        frame_size = 0;
+      }
+    }
+  }
+
+  if (!feof(infile)) {
+    if (fread(*buffer, 1, frame_size, infile) != frame_size) {
+      warn("Failed to read full frame\n");
+      return 1;
+    }
+
+    *bytes_read = frame_size;
+    return 0;
+  }
+
+  return 1;
+}

ivfdec.h

+/*
+ *  Copyright (c) 2013 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 IVFDEC_H_
+#define IVFDEC_H_
+
+#include "./tools_common.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+int file_is_ivf(struct VpxInputContext *input);
+
+int ivf_read_frame(struct VpxInputContext *input,
+                   uint8_t **buffer,
+                   size_t *bytes_read,
+                   size_t *buffer_size);
+
+#ifdef __cplusplus
+}  /* extern "C" */
+#endif
+
+#endif  /* IVFDEC_H_ */

ivfenc.c

+/*
+ *  Copyright (c) 2013 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 "./ivfenc.h"
+
+#include "./tools_common.h"
+#include "vpx/vpx_encoder.h"
+#include "vpx_ports/mem_ops.h"
+
+void ivf_write_file_header(FILE *outfile,
+                           const struct vpx_codec_enc_cfg *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);            /* four CC */
+  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 */
+
+  (void) fwrite(header, 1, 32, outfile);
+}
+
+void ivf_write_frame_header(FILE *outfile, const struct vpx_codec_cx_pkt *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, (int)pkt->data.frame.sz);
+  mem_put_le32(header + 4, pts & 0xFFFFFFFF);
+  mem_put_le32(header + 8, pts >> 32);
+
+  (void) fwrite(header, 1, 12, outfile);
+}
+
+void ivf_write_frame_size(FILE *outfile, size_t size) {
+  char header[4];
+  mem_put_le32(header, (int)size);
+  (void) fwrite(header, 1, 4, outfile);
+}

ivfenc.h

+/*
+ *  Copyright (c) 2013 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 IVFENC_H_
+#define IVFENC_H_
+
+#include "./tools_common.h"
+
+struct vpx_codec_enc_cfg;
+struct vpx_codec_cx_pkt;
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void ivf_write_file_header(FILE *outfile,
+                           const struct vpx_codec_enc_cfg *cfg,
+                           uint32_t fourcc,
+                           int frame_cnt);
+void ivf_write_frame_header(FILE *outfile, const struct vpx_codec_cx_pkt *pkt);
+void ivf_write_frame_size(FILE *outfile, size_t size);
+
+#ifdef __cplusplus
+}  /* extern "C" */
+#endif
+
+#endif  /* IVFENC_H_ */

tools_common.c

@@ -7,10 +7,13 @@
  *  in the file PATENTS.  All contributing project authors may
  *  be found in the AUTHORS file in the root of the source tree.
  */
+
 #include "tools_common.h"
 
 #include <stdarg.h>
+#include <stdio.h>
 #include <stdlib.h>
+#include <string.h>
 
 #if defined(_WIN32) || defined(__OS2__)
 #include <io.h>
@@ -56,3 +59,74 @@ void fatal(const char *fmt, ...) {
 void warn(const char *fmt, ...) {
   LOG_ERROR("Warning");
 }
+
+uint16_t mem_get_le16(const void *data) {
+  uint16_t val;
+  const uint8_t *mem = (const uint8_t*)data;
+
+  val = mem[1] << 8;
+  val |= mem[0];
+  return val;
+}
+
+uint32_t mem_get_le32(const void *data) {
+  uint32_t val;
+  const uint8_t *mem = (const uint8_t*)data;
+
+  val = mem[3] << 24;
+  val |= mem[2] << 16;
+  val |= mem[1] << 8;
+  val |= mem[0];
+  return val;
+}
+
+int read_yuv_frame(struct VpxInputContext *input_ctx, vpx_image_t *yuv_frame) {
+  FILE *f = input_ctx->file;
+  struct FileTypeDetectionBuffer *detect = &input_ctx->detect;
+  int plane = 0;
+  int shortread = 0;
+
+  for (plane = 0; plane < 3; ++plane) {
+    uint8_t *ptr;
+    const int w = (plane ? (1 + yuv_frame->d_w) / 2 : yuv_frame->d_w);
+    const int h = (plane ? (1 + yuv_frame->d_h) / 2 : yuv_frame->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 = yuv_frame->planes[
+            yuv_frame->fmt == VPX_IMG_FMT_YV12 ? VPX_PLANE_V : VPX_PLANE_U];
+        break;
+      case 2:
+        ptr = yuv_frame->planes[
+            yuv_frame->fmt == VPX_IMG_FMT_YV12 ? VPX_PLANE_U : VPX_PLANE_V];
+        break;
+      default:
+        ptr = yuv_frame->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 += yuv_frame->stride[plane];
+    }
+  }
+
+  return shortread;
+}

tools_common.h

@@ -13,6 +13,12 @@
 #include <stdio.h>
 
 #include "./vpx_config.h"
+#include "vpx/vpx_image.h"
+#include "vpx/vpx_integer.h"
+
+#if CONFIG_ENCODERS
+#include "./y4minput.h"
+#endif
 
 #if defined(_MSC_VER)
 /* MSVS doesn't define off_t, and uses _f{seek,tell}i64. */
@@ -52,11 +58,55 @@ typedef long off_t;  /* NOLINT */
 #define PATH_MAX 512
 #endif
 
+#define IVF_FRAME_HDR_SZ (4 + 8)  /* 4 byte size + 8 byte timestamp */
+#define IVF_FILE_HDR_SZ 32
+
+#define RAW_FRAME_HDR_SZ sizeof(uint32_t)
+
 #define VP8_FOURCC (0x30385056)
 #define VP9_FOURCC (0x30395056)
 #define VP8_FOURCC_MASK (0x00385056)
 #define VP9_FOURCC_MASK (0x00395056)
 
+enum VideoFileType {
+  FILE_TYPE_RAW,
+  FILE_TYPE_IVF,
+  FILE_TYPE_Y4M,
+  FILE_TYPE_WEBM
+};
+
+struct FileTypeDetectionBuffer {
+  char buf[4];
+  size_t buf_read;
+  size_t position;
+};
+
+struct VpxRational {
+  int numerator;
+  int denominator;
+};
+
+struct VpxInputContext {
+  const char *filename;
+  FILE *file;
+  off_t length;
+  struct FileTypeDetectionBuffer detect;
+  enum VideoFileType file_type;
+  unsigned int width;
+  unsigned int height;
+  int use_i420;
+  int only_i420;
+  unsigned int fourcc;
+  struct VpxRational framerate;
+#if CONFIG_ENCODERS
+  y4m_input y4m;
+#endif
+};
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
 /* Sets a stdio stream into binary mode */
 FILE *set_binary_mode(FILE *stream);
 
@@ -67,4 +117,13 @@ void warn(const char *fmt, ...);
 /* The tool including this file must define usage_exit() */
 void usage_exit();
 
+uint16_t mem_get_le16(const void *data);
+uint32_t mem_get_le32(const void *data);
+
+int read_yuv_frame(struct VpxInputContext *input_ctx, vpx_image_t *yuv_frame);
+
+#ifdef __cplusplus
+}  /* extern "C" */
+#endif
+
 #endif  // TOOLS_COMMON_H_

vp9_spatial_scalable_encoder.c

@@ -19,12 +19,12 @@
 #include <string.h>
 #include <time.h>
 #include "./args.h"
+#include "./ivfenc.h"
+#include "./tools_common.h"
 #include "vpx/svc_context.h"
 #include "vpx/vp8cx.h"
 #include "vpx/vpx_encoder.h"
 
-#define VP90_FOURCC 0x30395056
-
 static const struct arg_enum_list encoding_mode_enum[] = {
   {"i", INTER_LAYER_PREDICTION_I},
   {"alt-ip", ALT_INTER_LAYER_PREDICTION_IP},
@@ -77,25 +77,13 @@ static const uint32_t default_kf_dist = 100;
 static const int default_use_dummy_frame = 1;
 
 typedef struct {
-  char *input_filename;
   char *output_filename;
   uint32_t frames_to_code;
   uint32_t frames_to_skip;
+  struct VpxInputContext input_ctx;
 } AppInput;
 
-static void mem_put_le16(char *mem, uint32_t val) {
-  mem[0] = val;
-  mem[1] = val >> 8;
-}
-
-static void mem_put_le32(char *mem, uint32_t val) {
-  mem[0] = val;
-  mem[1] = val >> 8;
-  mem[2] = val >> 16;
-  mem[3] = val >> 24;
-}
-
-static void usage(const char *exec_name) {
+void usage_exit(const char *exec_name) {
   fprintf(stderr, "Usage: %s <options> input_filename output_filename\n",
           exec_name);
   fprintf(stderr, "Options:\n");
@@ -103,15 +91,6 @@ static void usage(const char *exec_name) {
   exit(EXIT_FAILURE);
 }
 
-void die(const char *fmt, ...) {
-  va_list ap;
-
-  va_start(ap, fmt);
-  vfprintf(stderr, fmt, ap);
-  if (fmt[strlen(fmt) - 1] != '\n') printf("\n");
-  exit(EXIT_FAILURE);
-}
-
 static void die_codec(vpx_codec_ctx_t *ctx, const char *s) {
   const char *detail = vpx_codec_error_detail(ctx);
 
@@ -120,83 +99,12 @@ static void die_codec(vpx_codec_ctx_t *ctx, const char *s) {
   exit(EXIT_FAILURE);
 }
 
-static int read_frame(FILE *f, vpx_image_t *img) {
-  size_t nbytes;
-  int res = 1;
-  int plane;
-
-  for (plane = 0; plane < 3; ++plane) {
-    uint8_t *ptr;
-    const int w = (plane ? (1 + img->d_w) / 2 : img->d_w);
-    const int h = (plane ? (1 + img->d_h) / 2 : img->d_h);
-    int r;
-
-    switch (plane) {
-      case 1:
-        ptr = img->planes[VPX_PLANE_U];
-        break;
-      case 2:
-        ptr = img->planes[VPX_PLANE_V];
-        break;
-      default:
-        ptr = img->planes[plane];
-    }
-    for (r = 0; r < h; ++r) {
-      const int to_read = w;
-
-      nbytes = fread(ptr, 1, to_read, f);
-      if (nbytes != to_read) {
-        res = 0;
-        if (nbytes > 0)
-          printf("Warning: Read partial frame. Check your width & height!\n");
-        break;
-      }
-      ptr += img->stride[plane];
-    }
-    if (!res) break;
-  }
-  return res;
-}
-
 static int create_dummy_frame(vpx_image_t *img) {
   const size_t buf_size = img->w * img->h * 3 / 2;
   memset(img->planes[0], 129, buf_size);
   return 1;
 }
 
-static void write_ivf_file_header(FILE *outfile,
-                                  uint32_t width, uint32_t height,
-                                  int timebase_num, int timebase_den,
-                                  int frame_cnt) {
-  char header[32];
-
-  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, VP90_FOURCC);   /* fourcc */
-  mem_put_le16(header + 12, width);        /* width */
-  mem_put_le16(header + 14, height);       /* height */
-  mem_put_le32(header + 16, timebase_den); /* rate */
-  mem_put_le32(header + 20, timebase_num); /* scale */
-  mem_put_le32(header + 24, frame_cnt);    /* length */
-  mem_put_le32(header + 28, 0);            /* unused */
-
-  (void)fwrite(header, 1, 32, outfile);
-}
-
-static void write_ivf_frame_header(FILE *outfile, vpx_codec_pts_t pts,
-                                   size_t sz) {
-  char header[12];
-  mem_put_le32(header, (uint32_t)sz);
-  mem_put_le32(header + 4, pts & 0xFFFFFFFF);
-  mem_put_le32(header + 8, pts >> 32);
-
-  (void)fwrite(header, 1, 12, outfile);
-}
-
 static void parse_command_line(int argc, const char **argv_,
                                AppInput *app_input, SvcContext *svc_ctx,
                                vpx_codec_enc_cfg_t *enc_cfg) {
@@ -272,9 +180,9 @@ static void parse_command_line(int argc, const char **argv_,
       die("Error: Unrecognized option %s\n", *argi);
 
   if (argv[0] == NULL || argv[1] == 0) {
-    usage(argv_[0]);
+    usage_exit(argv_[0]);
   }
-  app_input->input_filename = argv[0];
+  app_input->input_ctx.filename = argv[0];
   app_input->output_filename = argv[1];
   free(argv);
 
@@ -298,7 +206,7 @@ static void parse_command_line(int argc, const char **argv_,
 
 int main(int argc, const char **argv) {
   AppInput app_input = {0};
-  FILE *infile, *outfile;
+  FILE *outfile;
   vpx_codec_ctx_t codec;
   vpx_codec_enc_cfg_t enc_cfg;
   SvcContext svc_ctx;
@@ -308,6 +216,8 @@ int main(int argc, const char **argv) {
   vpx_codec_err_t res;
   int pts = 0;            /* PTS starts at 0 */
   int frame_duration = 1; /* 1 timebase tick per frame */
+  vpx_codec_cx_pkt_t packet = {0};
+  packet.kind = VPX_CODEC_CX_FRAME_PKT;
 
   memset(&svc_ctx, 0, sizeof(svc_ctx));
   svc_ctx.log_print = 1;
@@ -317,8 +227,8 @@ int main(int argc, const char **argv) {
   if (!vpx_img_alloc(&raw, VPX_IMG_FMT_I420, enc_cfg.g_w, enc_cfg.g_h, 32))
     die("Failed to allocate image %dx%d\n", enc_cfg.g_w, enc_cfg.g_h);
 
-  if (!(infile = fopen(app_input.