1/*
2 * H.264 encoder
3 *
4 * This file is part of FFmpeg.
5 *
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
10 *
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
14 * Lesser General Public License for more details.
15 *
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19 */
20
21
22#include "libavutil/common.h"
23#include "bitstream.h"
24#include "mpegvideo.h"
25#include "h264data.h"
26
27/**
28 * Write out the provided data into a NAL unit.
29 * @param nal_ref_idc NAL reference IDC
30 * @param nal_unit_type NAL unit payload type
31 * @param dest the target buffer, dst+1 == src is allowed as a special case
32 * @param destsize the length of the dst array
33 * @param b2 the data which should be escaped
34 * @returns pointer to current position in the output buffer or NULL if an error occurred
35 */
36static uint8_t *h264_write_nal_unit(int nal_ref_idc, int nal_unit_type, uint8_t *dest, int *destsize,
37                          PutBitContext *b2)
38{
39    PutBitContext b;
40    int i, destpos, rbsplen, escape_count;
41    uint8_t *rbsp;
42
43    if (nal_unit_type != NAL_END_STREAM)
44        put_bits(b2,1,1); // rbsp_stop_bit
45
46    // Align b2 on a byte boundary
47    align_put_bits(b2);
48    rbsplen = put_bits_count(b2)/8;
49    flush_put_bits(b2);
50    rbsp = b2->buf;
51
52    init_put_bits(&b,dest,*destsize);
53
54    put_bits(&b,16,0);
55    put_bits(&b,16,0x01);
56
57    put_bits(&b,1,0); // forbidden zero bit
58    put_bits(&b,2,nal_ref_idc); // nal_ref_idc
59    put_bits(&b,5,nal_unit_type); // nal_unit_type
60
61    flush_put_bits(&b);
62
63    destpos = 5;
64    escape_count= 0;
65
66    for (i=0; i<rbsplen; i+=2)
67    {
68        if (rbsp[i]) continue;
69        if (i>0 && rbsp[i-1]==0)
70            i--;
71        if (i+2<rbsplen && rbsp[i+1]==0 && rbsp[i+2]<=3)
72        {
73            escape_count++;
74            i+=2;
75        }
76    }
77
78    if(escape_count==0)
79    {
80        if(dest+destpos != rbsp)
81        {
82            memcpy(dest+destpos, rbsp, rbsplen);
83            *destsize -= (rbsplen+destpos);
84        }
85        return dest+rbsplen+destpos;
86    }
87
88    if(rbsplen + escape_count + 1> *destsize)
89    {
90        av_log(NULL, AV_LOG_ERROR, "Destination buffer too small!\n");
91        return NULL;
92    }
93
94    // this should be damn rare (hopefully)
95    for (i = 0 ; i < rbsplen ; i++)
96    {
97        if (i + 2 < rbsplen && (rbsp[i] == 0 && rbsp[i+1] == 0 && rbsp[i+2] < 4))
98        {
99            dest[destpos++] = rbsp[i++];
100            dest[destpos++] = rbsp[i];
101            dest[destpos++] = 0x03; // emulation prevention byte
102        }
103        else
104            dest[destpos++] = rbsp[i];
105    }
106    *destsize -= destpos;
107    return dest+destpos;
108}
109
110static const uint8_t pict_type_to_golomb[7] = {-1, 2, 0, 1, -1, 4, 3};
111
112static const uint8_t intra4x4_cbp_to_golomb[48] = {
113    3, 29, 30, 17, 31, 18, 37,  8, 32, 38, 19,  9, 20, 10, 11,  2,
114   16, 33, 34, 21, 35, 22, 39,  4, 36, 40, 23,  5, 24,  6,  7,  1,
115   41, 42, 43, 25, 44, 26, 46, 12, 45, 47, 27, 13, 28, 14, 15,  0
116};
117
118static const uint8_t inter_cbp_to_golomb[48] = {
119    0,  2,  3,  7,  4,  8, 17, 13,  5, 18,  9, 14, 10, 15, 16, 11,
120    1, 32, 33, 36, 34, 37, 44, 40, 35, 45, 38, 41, 39, 42, 43, 19,
121    6, 24, 25, 20, 26, 21, 46, 28, 27, 47, 22, 29, 23, 30, 31, 12
122};
123
124#define QUANT_SHIFT 22
125
126static const int quant_coeff[52][16] = {
127    { 419430, 258111, 419430, 258111, 258111, 167772, 258111, 167772, 419430, 258111, 419430, 258111, 258111, 167772, 258111, 167772,},
128    { 381300, 239675, 381300, 239675, 239675, 149131, 239675, 149131, 381300, 239675, 381300, 239675, 239675, 149131, 239675, 149131,},
129    { 322639, 209715, 322639, 209715, 209715, 134218, 209715, 134218, 322639, 209715, 322639, 209715, 209715, 134218, 209715, 134218,},
130    { 299593, 186414, 299593, 186414, 186414, 116711, 186414, 116711, 299593, 186414, 299593, 186414, 186414, 116711, 186414, 116711,},
131    { 262144, 167772, 262144, 167772, 167772, 107374, 167772, 107374, 262144, 167772, 262144, 167772, 167772, 107374, 167772, 107374,},
132    { 233017, 145889, 233017, 145889, 145889,  92564, 145889,  92564, 233017, 145889, 233017, 145889, 145889,  92564, 145889,  92564,},
133    { 209715, 129056, 209715, 129056, 129056,  83886, 129056,  83886, 209715, 129056, 209715, 129056, 129056,  83886, 129056,  83886,},
134    { 190650, 119837, 190650, 119837, 119837,  74565, 119837,  74565, 190650, 119837, 190650, 119837, 119837,  74565, 119837,  74565,},
135    { 161319, 104858, 161319, 104858, 104858,  67109, 104858,  67109, 161319, 104858, 161319, 104858, 104858,  67109, 104858,  67109,},
136    { 149797,  93207, 149797,  93207,  93207,  58356,  93207,  58356, 149797,  93207, 149797,  93207,  93207,  58356,  93207,  58356,},
137    { 131072,  83886, 131072,  83886,  83886,  53687,  83886,  53687, 131072,  83886, 131072,  83886,  83886,  53687,  83886,  53687,},
138    { 116508,  72944, 116508,  72944,  72944,  46282,  72944,  46282, 116508,  72944, 116508,  72944,  72944,  46282,  72944,  46282,},
139    { 104858,  64528, 104858,  64528,  64528,  41943,  64528,  41943, 104858,  64528, 104858,  64528,  64528,  41943,  64528,  41943,},
140    {  95325,  59919,  95325,  59919,  59919,  37283,  59919,  37283,  95325,  59919,  95325,  59919,  59919,  37283,  59919,  37283,},
141    {  80660,  52429,  80660,  52429,  52429,  33554,  52429,  33554,  80660,  52429,  80660,  52429,  52429,  33554,  52429,  33554,},
142    {  74898,  46603,  74898,  46603,  46603,  29178,  46603,  29178,  74898,  46603,  74898,  46603,  46603,  29178,  46603,  29178,},
143    {  65536,  41943,  65536,  41943,  41943,  26844,  41943,  26844,  65536,  41943,  65536,  41943,  41943,  26844,  41943,  26844,},
144    {  58254,  36472,  58254,  36472,  36472,  23141,  36472,  23141,  58254,  36472,  58254,  36472,  36472,  23141,  36472,  23141,},
145    {  52429,  32264,  52429,  32264,  32264,  20972,  32264,  20972,  52429,  32264,  52429,  32264,  32264,  20972,  32264,  20972,},
146    {  47663,  29959,  47663,  29959,  29959,  18641,  29959,  18641,  47663,  29959,  47663,  29959,  29959,  18641,  29959,  18641,},
147    {  40330,  26214,  40330,  26214,  26214,  16777,  26214,  16777,  40330,  26214,  40330,  26214,  26214,  16777,  26214,  16777,},
148    {  37449,  23302,  37449,  23302,  23302,  14589,  23302,  14589,  37449,  23302,  37449,  23302,  23302,  14589,  23302,  14589,},
149    {  32768,  20972,  32768,  20972,  20972,  13422,  20972,  13422,  32768,  20972,  32768,  20972,  20972,  13422,  20972,  13422,},
150    {  29127,  18236,  29127,  18236,  18236,  11570,  18236,  11570,  29127,  18236,  29127,  18236,  18236,  11570,  18236,  11570,},
151    {  26214,  16132,  26214,  16132,  16132,  10486,  16132,  10486,  26214,  16132,  26214,  16132,  16132,  10486,  16132,  10486,},
152    {  23831,  14980,  23831,  14980,  14980,   9321,  14980,   9321,  23831,  14980,  23831,  14980,  14980,   9321,  14980,   9321,},
153    {  20165,  13107,  20165,  13107,  13107,   8389,  13107,   8389,  20165,  13107,  20165,  13107,  13107,   8389,  13107,   8389,},
154    {  18725,  11651,  18725,  11651,  11651,   7294,  11651,   7294,  18725,  11651,  18725,  11651,  11651,   7294,  11651,   7294,},
155    {  16384,  10486,  16384,  10486,  10486,   6711,  10486,   6711,  16384,  10486,  16384,  10486,  10486,   6711,  10486,   6711,},
156    {  14564,   9118,  14564,   9118,   9118,   5785,   9118,   5785,  14564,   9118,  14564,   9118,   9118,   5785,   9118,   5785,},
157    {  13107,   8066,  13107,   8066,   8066,   5243,   8066,   5243,  13107,   8066,  13107,   8066,   8066,   5243,   8066,   5243,},
158    {  11916,   7490,  11916,   7490,   7490,   4660,   7490,   4660,  11916,   7490,  11916,   7490,   7490,   4660,   7490,   4660,},
159    {  10082,   6554,  10082,   6554,   6554,   4194,   6554,   4194,  10082,   6554,  10082,   6554,   6554,   4194,   6554,   4194,},
160    {   9362,   5825,   9362,   5825,   5825,   3647,   5825,   3647,   9362,   5825,   9362,   5825,   5825,   3647,   5825,   3647,},
161    {   8192,   5243,   8192,   5243,   5243,   3355,   5243,   3355,   8192,   5243,   8192,   5243,   5243,   3355,   5243,   3355,},
162    {   7282,   4559,   7282,   4559,   4559,   2893,   4559,   2893,   7282,   4559,   7282,   4559,   4559,   2893,   4559,   2893,},
163    {   6554,   4033,   6554,   4033,   4033,   2621,   4033,   2621,   6554,   4033,   6554,   4033,   4033,   2621,   4033,   2621,},
164    {   5958,   3745,   5958,   3745,   3745,   2330,   3745,   2330,   5958,   3745,   5958,   3745,   3745,   2330,   3745,   2330,},
165    {   5041,   3277,   5041,   3277,   3277,   2097,   3277,   2097,   5041,   3277,   5041,   3277,   3277,   2097,   3277,   2097,},
166    {   4681,   2913,   4681,   2913,   2913,   1824,   2913,   1824,   4681,   2913,   4681,   2913,   2913,   1824,   2913,   1824,},
167    {   4096,   2621,   4096,   2621,   2621,   1678,   2621,   1678,   4096,   2621,   4096,   2621,   2621,   1678,   2621,   1678,},
168    {   3641,   2280,   3641,   2280,   2280,   1446,   2280,   1446,   3641,   2280,   3641,   2280,   2280,   1446,   2280,   1446,},
169    {   3277,   2016,   3277,   2016,   2016,   1311,   2016,   1311,   3277,   2016,   3277,   2016,   2016,   1311,   2016,   1311,},
170    {   2979,   1872,   2979,   1872,   1872,   1165,   1872,   1165,   2979,   1872,   2979,   1872,   1872,   1165,   1872,   1165,},
171    {   2521,   1638,   2521,   1638,   1638,   1049,   1638,   1049,   2521,   1638,   2521,   1638,   1638,   1049,   1638,   1049,},
172    {   2341,   1456,   2341,   1456,   1456,    912,   1456,    912,   2341,   1456,   2341,   1456,   1456,    912,   1456,    912,},
173    {   2048,   1311,   2048,   1311,   1311,    839,   1311,    839,   2048,   1311,   2048,   1311,   1311,    839,   1311,    839,},
174    {   1820,   1140,   1820,   1140,   1140,    723,   1140,    723,   1820,   1140,   1820,   1140,   1140,    723,   1140,    723,},
175    {   1638,   1008,   1638,   1008,   1008,    655,   1008,    655,   1638,   1008,   1638,   1008,   1008,    655,   1008,    655,},
176    {   1489,    936,   1489,    936,    936,    583,    936,    583,   1489,    936,   1489,    936,    936,    583,    936,    583,},
177    {   1260,    819,   1260,    819,    819,    524,    819,    524,   1260,    819,   1260,    819,    819,    524,    819,    524,},
178    {   1170,    728,   1170,    728,    728,    456,    728,    456,   1170,    728,   1170,    728,    728,    456,    728,    456,},
179};
180
181//FIXME need to check that this does not overflow signed 32 bit for low qp, I am not sure, it's very close
182//FIXME check that gcc inlines this (and optimizes intra & separate_dc stuff away)
183static inline int quantize_c(DCTELEM *block, uint8_t *scantable, int qscale,
184                             int intra, int separate_dc)
185{
186    int i;
187    const int * const quant_3Btable = quant_coeff[qscale];
188    const int bias = intra ? (1 << QUANT_SHIFT) / 3 : (1 << QUANT_SHIFT) / 6;
189    const unsigned int threshold1 = (1 << QUANT_SHIFT) - bias - 1;
190    const unsigned int threshold2 = (threshold1 << 1);
191    int last_non_zero;
192
193    if (separate_dc) {
194        if (qscale <= 18) {
195            //avoid overflows
196            const int dc_bias = intra ? (1 << (QUANT_SHIFT - 2)) / 3 : (1 << (QUANT_SHIFT - 2)) / 6;
197            const unsigned int dc_threshold1 = (1 << (QUANT_SHIFT - 2)) - dc_bias - 1;
198            const unsigned int dc_threshold2 = (dc_threshold1 << 1);
199
200            int level = block[0]*quant_coeff[qscale+18][0];
201            if (((unsigned)(level + dc_threshold1)) > dc_threshold2) {
202                if (level > 0) {
203                    level = (dc_bias + level) >> (QUANT_SHIFT - 2);
204                    block[0] = level;
205                } else {
206                    level = (dc_bias - level) >> (QUANT_SHIFT - 2);
207                    block[0] = -level;
208                }
209//                last_non_zero = i;
210            } else {
211                block[0] = 0;
212            }
213        } else {
214            const int dc_bias = intra ? (1 << (QUANT_SHIFT + 1)) / 3 : (1 << (QUANT_SHIFT + 1)) / 6;
215            const unsigned int dc_threshold1 = (1 << (QUANT_SHIFT + 1)) - dc_bias - 1;
216            const unsigned int dc_threshold2 = (dc_threshold1 << 1);
217
218            int level = block[0]*quant_table[0];
219            if (((unsigned)(level + dc_threshold1)) > dc_threshold2) {
220                if (level > 0) {
221                    level = (dc_bias + level) >> (QUANT_SHIFT + 1);
222                    block[0] = level;
223                } else {
224                    level = (dc_bias - level) >> (QUANT_SHIFT + 1);
225                    block[0] = -level;
226                }
227//                last_non_zero = i;
228            } else {
229                block[0] = 0;
230            }
231        }
232        last_non_zero = 0;
233        i = 1;
234    } else {
235        last_non_zero = -1;
236        i = 0;
237    }
238
239    for (; i < 16; i++) {
240        const int j = scantable[i];
241        int level = block[j]*quant_table[j];
242
243//        if (   bias+level >= (1 << (QMAT_SHIFT - 3))
244//            || bias-level >= (1 << (QMAT_SHIFT - 3))) {
245        if (((unsigned)(level + threshold1)) > threshold2) {
246            if (level > 0) {
247                level = (bias + level) >> QUANT_SHIFT;
248                block[j] = level;
249            } else {
250                level = (bias - level) >> QUANT_SHIFT;
251                block[j] = -level;
252            }
253            last_non_zero = i;
254        } else {
255            block[j] = 0;
256        }
257    }
258
259    return last_non_zero;
260}
261