14989f8259e9611d83b5b0beb23a958d37467a1d
[libav.git] / libavfilter / vf_crop.c
1 /*
2 * Copyright (c) 2007 Bobby Bingham
3 *
4 * This file is part of Libav.
5 *
6 * Libav 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 * Libav 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 Libav; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19 */
20
21 /**
22 * @file
23 * video crop filter
24 */
25
26 /* #define DEBUG */
27
28 #include "avfilter.h"
29 #include "formats.h"
30 #include "internal.h"
31 #include "video.h"
32 #include "libavutil/eval.h"
33 #include "libavutil/avstring.h"
34 #include "libavutil/libm.h"
35 #include "libavutil/imgutils.h"
36 #include "libavutil/mathematics.h"
37
38 static const char *const var_names[] = {
39 "E",
40 "PHI",
41 "PI",
42 "in_w", "iw", ///< width of the input video
43 "in_h", "ih", ///< height of the input video
44 "out_w", "ow", ///< width of the cropped video
45 "out_h", "oh", ///< height of the cropped video
46 "x",
47 "y",
48 "n", ///< number of frame
49 "pos", ///< position in the file
50 "t", ///< timestamp expressed in seconds
51 NULL
52 };
53
54 enum var_name {
55 VAR_E,
56 VAR_PHI,
57 VAR_PI,
58 VAR_IN_W, VAR_IW,
59 VAR_IN_H, VAR_IH,
60 VAR_OUT_W, VAR_OW,
61 VAR_OUT_H, VAR_OH,
62 VAR_X,
63 VAR_Y,
64 VAR_N,
65 VAR_POS,
66 VAR_T,
67 VAR_VARS_NB
68 };
69
70 typedef struct {
71 int x; ///< x offset of the non-cropped area with respect to the input area
72 int y; ///< y offset of the non-cropped area with respect to the input area
73 int w; ///< width of the cropped area
74 int h; ///< height of the cropped area
75
76 int max_step[4]; ///< max pixel step for each plane, expressed as a number of bytes
77 int hsub, vsub; ///< chroma subsampling
78 char x_expr[256], y_expr[256], ow_expr[256], oh_expr[256];
79 AVExpr *x_pexpr, *y_pexpr; /* parsed expressions for x and y */
80 double var_values[VAR_VARS_NB];
81 } CropContext;
82
83 static int query_formats(AVFilterContext *ctx)
84 {
85 static const enum PixelFormat pix_fmts[] = {
86 PIX_FMT_RGB48BE, PIX_FMT_RGB48LE,
87 PIX_FMT_BGR48BE, PIX_FMT_BGR48LE,
88 PIX_FMT_ARGB, PIX_FMT_RGBA,
89 PIX_FMT_ABGR, PIX_FMT_BGRA,
90 PIX_FMT_RGB24, PIX_FMT_BGR24,
91 PIX_FMT_RGB565BE, PIX_FMT_RGB565LE,
92 PIX_FMT_RGB555BE, PIX_FMT_RGB555LE,
93 PIX_FMT_BGR565BE, PIX_FMT_BGR565LE,
94 PIX_FMT_BGR555BE, PIX_FMT_BGR555LE,
95 PIX_FMT_GRAY16BE, PIX_FMT_GRAY16LE,
96 PIX_FMT_YUV420P16LE, PIX_FMT_YUV420P16BE,
97 PIX_FMT_YUV422P16LE, PIX_FMT_YUV422P16BE,
98 PIX_FMT_YUV444P16LE, PIX_FMT_YUV444P16BE,
99 PIX_FMT_YUV444P, PIX_FMT_YUV422P,
100 PIX_FMT_YUV420P, PIX_FMT_YUV411P,
101 PIX_FMT_YUV410P, PIX_FMT_YUV440P,
102 PIX_FMT_YUVJ444P, PIX_FMT_YUVJ422P,
103 PIX_FMT_YUVJ420P, PIX_FMT_YUVJ440P,
104 PIX_FMT_YUVA420P,
105 PIX_FMT_RGB8, PIX_FMT_BGR8,
106 PIX_FMT_RGB4_BYTE, PIX_FMT_BGR4_BYTE,
107 PIX_FMT_PAL8, PIX_FMT_GRAY8,
108 PIX_FMT_NONE
109 };
110
111 ff_set_common_formats(ctx, ff_make_format_list(pix_fmts));
112
113 return 0;
114 }
115
116 static av_cold int init(AVFilterContext *ctx, const char *args)
117 {
118 CropContext *crop = ctx->priv;
119
120 av_strlcpy(crop->ow_expr, "iw", sizeof(crop->ow_expr));
121 av_strlcpy(crop->oh_expr, "ih", sizeof(crop->oh_expr));
122 av_strlcpy(crop->x_expr, "(in_w-out_w)/2", sizeof(crop->x_expr));
123 av_strlcpy(crop->y_expr, "(in_h-out_h)/2", sizeof(crop->y_expr));
124
125 if (args)
126 sscanf(args, "%255[^:]:%255[^:]:%255[^:]:%255[^:]", crop->ow_expr, crop->oh_expr, crop->x_expr, crop->y_expr);
127
128 return 0;
129 }
130
131 static av_cold void uninit(AVFilterContext *ctx)
132 {
133 CropContext *crop = ctx->priv;
134
135 av_expr_free(crop->x_pexpr); crop->x_pexpr = NULL;
136 av_expr_free(crop->y_pexpr); crop->y_pexpr = NULL;
137 }
138
139 static inline int normalize_double(int *n, double d)
140 {
141 int ret = 0;
142
143 if (isnan(d)) {
144 ret = AVERROR(EINVAL);
145 } else if (d > INT_MAX || d < INT_MIN) {
146 *n = d > INT_MAX ? INT_MAX : INT_MIN;
147 ret = AVERROR(EINVAL);
148 } else
149 *n = round(d);
150
151 return ret;
152 }
153
154 static int config_input(AVFilterLink *link)
155 {
156 AVFilterContext *ctx = link->dst;
157 CropContext *crop = ctx->priv;
158 const AVPixFmtDescriptor *pix_desc = &av_pix_fmt_descriptors[link->format];
159 int ret;
160 const char *expr;
161 double res;
162
163 crop->var_values[VAR_E] = M_E;
164 crop->var_values[VAR_PHI] = M_PHI;
165 crop->var_values[VAR_PI] = M_PI;
166 crop->var_values[VAR_IN_W] = crop->var_values[VAR_IW] = ctx->inputs[0]->w;
167 crop->var_values[VAR_IN_H] = crop->var_values[VAR_IH] = ctx->inputs[0]->h;
168 crop->var_values[VAR_X] = NAN;
169 crop->var_values[VAR_Y] = NAN;
170 crop->var_values[VAR_OUT_W] = crop->var_values[VAR_OW] = NAN;
171 crop->var_values[VAR_OUT_H] = crop->var_values[VAR_OH] = NAN;
172 crop->var_values[VAR_N] = 0;
173 crop->var_values[VAR_T] = NAN;
174 crop->var_values[VAR_POS] = NAN;
175
176 av_image_fill_max_pixsteps(crop->max_step, NULL, pix_desc);
177 crop->hsub = av_pix_fmt_descriptors[link->format].log2_chroma_w;
178 crop->vsub = av_pix_fmt_descriptors[link->format].log2_chroma_h;
179
180 if ((ret = av_expr_parse_and_eval(&res, (expr = crop->ow_expr),
181 var_names, crop->var_values,
182 NULL, NULL, NULL, NULL, NULL, 0, ctx)) < 0) goto fail_expr;
183 crop->var_values[VAR_OUT_W] = crop->var_values[VAR_OW] = res;
184 if ((ret = av_expr_parse_and_eval(&res, (expr = crop->oh_expr),
185 var_names, crop->var_values,
186 NULL, NULL, NULL, NULL, NULL, 0, ctx)) < 0) goto fail_expr;
187 crop->var_values[VAR_OUT_H] = crop->var_values[VAR_OH] = res;
188 /* evaluate again ow as it may depend on oh */
189 if ((ret = av_expr_parse_and_eval(&res, (expr = crop->ow_expr),
190 var_names, crop->var_values,
191 NULL, NULL, NULL, NULL, NULL, 0, ctx)) < 0) goto fail_expr;
192 crop->var_values[VAR_OUT_W] = crop->var_values[VAR_OW] = res;
193 if (normalize_double(&crop->w, crop->var_values[VAR_OUT_W]) < 0 ||
194 normalize_double(&crop->h, crop->var_values[VAR_OUT_H]) < 0) {
195 av_log(ctx, AV_LOG_ERROR,
196 "Too big value or invalid expression for out_w/ow or out_h/oh. "
197 "Maybe the expression for out_w:'%s' or for out_h:'%s' is self-referencing.\n",
198 crop->ow_expr, crop->oh_expr);
199 return AVERROR(EINVAL);
200 }
201 crop->w &= ~((1 << crop->hsub) - 1);
202 crop->h &= ~((1 << crop->vsub) - 1);
203
204 if ((ret = av_expr_parse(&crop->x_pexpr, crop->x_expr, var_names,
205 NULL, NULL, NULL, NULL, 0, ctx)) < 0 ||
206 (ret = av_expr_parse(&crop->y_pexpr, crop->y_expr, var_names,
207 NULL, NULL, NULL, NULL, 0, ctx)) < 0)
208 return AVERROR(EINVAL);
209
210 av_log(ctx, AV_LOG_VERBOSE, "w:%d h:%d -> w:%d h:%d\n",
211 link->w, link->h, crop->w, crop->h);
212
213 if (crop->w <= 0 || crop->h <= 0 ||
214 crop->w > link->w || crop->h > link->h) {
215 av_log(ctx, AV_LOG_ERROR,
216 "Invalid too big or non positive size for width '%d' or height '%d'\n",
217 crop->w, crop->h);
218 return AVERROR(EINVAL);
219 }
220
221 /* set default, required in the case the first computed value for x/y is NAN */
222 crop->x = (link->w - crop->w) / 2;
223 crop->y = (link->h - crop->h) / 2;
224 crop->x &= ~((1 << crop->hsub) - 1);
225 crop->y &= ~((1 << crop->vsub) - 1);
226 return 0;
227
228 fail_expr:
229 av_log(NULL, AV_LOG_ERROR, "Error when evaluating the expression '%s'\n", expr);
230 return ret;
231 }
232
233 static int config_output(AVFilterLink *link)
234 {
235 CropContext *crop = link->src->priv;
236
237 link->w = crop->w;
238 link->h = crop->h;
239
240 return 0;
241 }
242
243 static void start_frame(AVFilterLink *link, AVFilterBufferRef *picref)
244 {
245 AVFilterContext *ctx = link->dst;
246 CropContext *crop = ctx->priv;
247 AVFilterBufferRef *ref2;
248 int i;
249
250 ref2 = avfilter_ref_buffer(picref, ~0);
251 ref2->video->w = crop->w;
252 ref2->video->h = crop->h;
253
254 crop->var_values[VAR_T] = picref->pts == AV_NOPTS_VALUE ?
255 NAN : picref->pts * av_q2d(link->time_base);
256 crop->var_values[VAR_POS] = picref->pos == -1 ? NAN : picref->pos;
257 crop->var_values[VAR_X] = av_expr_eval(crop->x_pexpr, crop->var_values, NULL);
258 crop->var_values[VAR_Y] = av_expr_eval(crop->y_pexpr, crop->var_values, NULL);
259 crop->var_values[VAR_X] = av_expr_eval(crop->x_pexpr, crop->var_values, NULL);
260
261 normalize_double(&crop->x, crop->var_values[VAR_X]);
262 normalize_double(&crop->y, crop->var_values[VAR_Y]);
263
264 if (crop->x < 0) crop->x = 0;
265 if (crop->y < 0) crop->y = 0;
266 if ((unsigned)crop->x + (unsigned)crop->w > link->w) crop->x = link->w - crop->w;
267 if ((unsigned)crop->y + (unsigned)crop->h > link->h) crop->y = link->h - crop->h;
268 crop->x &= ~((1 << crop->hsub) - 1);
269 crop->y &= ~((1 << crop->vsub) - 1);
270
271 av_dlog(ctx, "n:%d t:%f x:%d y:%d x+w:%d y+h:%d\n",
272 (int)crop->var_values[VAR_N], crop->var_values[VAR_T], crop->x,
273 crop->y, crop->x+crop->w, crop->y+crop->h);
274
275 ref2->data[0] += crop->y * ref2->linesize[0];
276 ref2->data[0] += crop->x * crop->max_step[0];
277
278 if (!(av_pix_fmt_descriptors[link->format].flags & PIX_FMT_PAL ||
279 av_pix_fmt_descriptors[link->format].flags & PIX_FMT_PSEUDOPAL)) {
280 for (i = 1; i < 3; i ++) {
281 if (ref2->data[i]) {
282 ref2->data[i] += (crop->y >> crop->vsub) * ref2->linesize[i];
283 ref2->data[i] += (crop->x * crop->max_step[i]) >> crop->hsub;
284 }
285 }
286 }
287
288 /* alpha plane */
289 if (ref2->data[3]) {
290 ref2->data[3] += crop->y * ref2->linesize[3];
291 ref2->data[3] += crop->x * crop->max_step[3];
292 }
293
294 ff_start_frame(link->dst->outputs[0], ref2);
295 }
296
297 static void draw_slice(AVFilterLink *link, int y, int h, int slice_dir)
298 {
299 AVFilterContext *ctx = link->dst;
300 CropContext *crop = ctx->priv;
301
302 if (y >= crop->y + crop->h || y + h <= crop->y)
303 return;
304
305 if (y < crop->y) {
306 h -= crop->y - y;
307 y = crop->y;
308 }
309 if (y + h > crop->y + crop->h)
310 h = crop->y + crop->h - y;
311
312 ff_draw_slice(ctx->outputs[0], y - crop->y, h, slice_dir);
313 }
314
315 static void end_frame(AVFilterLink *link)
316 {
317 CropContext *crop = link->dst->priv;
318
319 crop->var_values[VAR_N] += 1.0;
320 ff_end_frame(link->dst->outputs[0]);
321 }
322
323 AVFilter avfilter_vf_crop = {
324 .name = "crop",
325 .description = NULL_IF_CONFIG_SMALL("Crop the input video to width:height:x:y."),
326
327 .priv_size = sizeof(CropContext),
328
329 .query_formats = query_formats,
330 .init = init,
331 .uninit = uninit,
332
333 .inputs = (const AVFilterPad[]) {{ .name = "default",
334 .type = AVMEDIA_TYPE_VIDEO,
335 .start_frame = start_frame,
336 .draw_slice = draw_slice,
337 .end_frame = end_frame,
338 .get_video_buffer = ff_null_get_video_buffer,
339 .config_props = config_input, },
340 { .name = NULL}},
341 .outputs = (const AVFilterPad[]) {{ .name = "default",
342 .type = AVMEDIA_TYPE_VIDEO,
343 .config_props = config_output, },
344 { .name = NULL}},
345 };