2 * Copyright (c) 2011 Stefano Sabatini
3 * Copyright (c) 2010 S.N. Hemanth Meenakshisundaram
4 * Copyright (c) 2003 Gustavo Sverzut Barbieri <gsbarbieri@yahoo.com.br>
6 * This file is part of FFmpeg.
8 * FFmpeg is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
13 * FFmpeg is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Lesser General Public License for more details.
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with FFmpeg; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
25 * drawtext filter, based on the original vhook/drawtext.c
26 * filter by Gustavo Sverzut Barbieri
33 #include "libavutil/avstring.h"
34 #include "libavutil/bprint.h"
35 #include "libavutil/common.h"
36 #include "libavutil/file.h"
37 #include "libavutil/eval.h"
38 #include "libavutil/opt.h"
39 #include "libavutil/random_seed.h"
40 #include "libavutil/parseutils.h"
41 #include "libavutil/timecode.h"
42 #include "libavutil/tree.h"
43 #include "libavutil/lfg.h"
45 #include "drawutils.h"
51 #include <freetype/config/ftheader.h>
52 #include FT_FREETYPE_H
55 #include <fontconfig/fontconfig.h>
58 static const char *const var_names[] = {
61 "line_h", "lh", ///< line height, same as max_glyph_h
62 "main_h", "h", "H", ///< height of the input video
63 "main_w", "w", "W", ///< width of the input video
64 "max_glyph_a", "ascent", ///< max glyph ascent
65 "max_glyph_d", "descent", ///< min glyph descent
66 "max_glyph_h", ///< max glyph height
67 "max_glyph_w", ///< max glyph width
68 "n", ///< number of frame
70 "t", ///< timestamp expressed in seconds
71 "text_h", "th", ///< height of the rendered text
72 "text_w", "tw", ///< width of the rendered text
79 static const char *const fun2_names[] = {
83 static double drand(void *opaque, double min, double max)
85 return min + (max-min) / UINT_MAX * av_lfg_get(opaque);
88 typedef double (*eval_func2)(void *, double a, double b);
90 static const eval_func2 fun2[] = {
99 VAR_MAIN_H, VAR_h, VAR_H,
100 VAR_MAIN_W, VAR_w, VAR_W,
101 VAR_MAX_GLYPH_A, VAR_ASCENT,
102 VAR_MAX_GLYPH_D, VAR_DESCENT,
116 enum expansion_mode {
123 const AVClass *class;
124 enum expansion_mode exp_mode; ///< expansion mode to use for the text
125 int reinit; ///< tells if the filter is being reinited
126 uint8_t *fontfile; ///< font to be used
127 uint8_t *text; ///< text to be drawn
128 AVBPrint expanded_text; ///< used to contain the expanded text
129 int ft_load_flags; ///< flags used for loading fonts, see FT_LOAD_*
130 FT_Vector *positions; ///< positions for each element in the text
131 size_t nb_positions; ///< number of elements of positions array
132 char *textfile; ///< file with text to be drawn
133 int x; ///< x position to start drawing text
134 int y; ///< y position to start drawing text
135 int max_glyph_w; ///< max glyph width
136 int max_glyph_h; ///< max glyph height
137 int shadowx, shadowy;
138 unsigned int fontsize; ///< font size to use
140 short int draw_box; ///< draw box around text - true or false
141 int use_kerning; ///< font kerning is used - true/false
142 int tabsize; ///< tab size
143 int fix_bounds; ///< do we let it go out of frame bounds - t/f
146 FFDrawColor fontcolor; ///< foreground color
147 FFDrawColor shadowcolor; ///< shadow color
148 FFDrawColor boxcolor; ///< background color
150 FT_Library library; ///< freetype font library handle
151 FT_Face face; ///< freetype font face handle
152 struct AVTreeNode *glyphs; ///< rendered glyphs, stored using the UTF-32 char code
153 char *x_expr; ///< expression for x position
154 char *y_expr; ///< expression for y position
155 AVExpr *x_pexpr, *y_pexpr; ///< parsed expressions for x and y
156 int64_t basetime; ///< base pts time in the real world for display
157 double var_values[VAR_VARS_NB];
158 char *draw_expr; ///< expression for draw
159 AVExpr *draw_pexpr; ///< parsed expression for draw
160 int draw; ///< set to zero to prevent drawing
161 AVLFG prng; ///< random
162 char *tc_opt_string; ///< specified timecode option string
163 AVRational tc_rate; ///< frame rate for timecode
164 AVTimecode tc; ///< timecode context
165 int tc24hmax; ///< 1 if timecode is wrapped to 24 hours, 0 otherwise
166 int reload; ///< reload text file for each frame
167 int start_number; ///< starting frame number for n/frame_num var
168 AVDictionary *metadata;
171 #define OFFSET(x) offsetof(DrawTextContext, x)
172 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
174 static const AVOption drawtext_options[]= {
175 {"fontfile", "set font file", OFFSET(fontfile), AV_OPT_TYPE_STRING, {.str=NULL}, CHAR_MIN, CHAR_MAX, FLAGS},
176 {"text", "set text", OFFSET(text), AV_OPT_TYPE_STRING, {.str=NULL}, CHAR_MIN, CHAR_MAX, FLAGS},
177 {"textfile", "set text file", OFFSET(textfile), AV_OPT_TYPE_STRING, {.str=NULL}, CHAR_MIN, CHAR_MAX, FLAGS},
178 {"fontcolor", "set foreground color", OFFSET(fontcolor.rgba), AV_OPT_TYPE_COLOR, {.str="black"}, CHAR_MIN, CHAR_MAX, FLAGS},
179 {"boxcolor", "set box color", OFFSET(boxcolor.rgba), AV_OPT_TYPE_COLOR, {.str="white"}, CHAR_MIN, CHAR_MAX, FLAGS},
180 {"shadowcolor", "set shadow color", OFFSET(shadowcolor.rgba), AV_OPT_TYPE_COLOR, {.str="black"}, CHAR_MIN, CHAR_MAX, FLAGS},
181 {"box", "set box", OFFSET(draw_box), AV_OPT_TYPE_INT, {.i64=0}, 0, 1 , FLAGS},
182 {"fontsize", "set font size", OFFSET(fontsize), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX , FLAGS},
183 {"x", "set x expression", OFFSET(x_expr), AV_OPT_TYPE_STRING, {.str="0"}, CHAR_MIN, CHAR_MAX, FLAGS},
184 {"y", "set y expression", OFFSET(y_expr), AV_OPT_TYPE_STRING, {.str="0"}, CHAR_MIN, CHAR_MAX, FLAGS},
185 {"shadowx", "set x", OFFSET(shadowx), AV_OPT_TYPE_INT, {.i64=0}, INT_MIN, INT_MAX , FLAGS},
186 {"shadowy", "set y", OFFSET(shadowy), AV_OPT_TYPE_INT, {.i64=0}, INT_MIN, INT_MAX , FLAGS},
187 {"tabsize", "set tab size", OFFSET(tabsize), AV_OPT_TYPE_INT, {.i64=4}, 0, INT_MAX , FLAGS},
188 {"basetime", "set base time", OFFSET(basetime), AV_OPT_TYPE_INT64, {.i64=AV_NOPTS_VALUE}, INT64_MIN, INT64_MAX , FLAGS},
189 {"draw", "if false do not draw", OFFSET(draw_expr), AV_OPT_TYPE_STRING, {.str="1"}, CHAR_MIN, CHAR_MAX, FLAGS},
191 {"expansion", "set the expansion mode", OFFSET(exp_mode), AV_OPT_TYPE_INT, {.i64=EXP_NORMAL}, 0, 2, FLAGS, "expansion"},
192 {"none", "set no expansion", OFFSET(exp_mode), AV_OPT_TYPE_CONST, {.i64=EXP_NONE}, 0, 0, FLAGS, "expansion"},
193 {"normal", "set normal expansion", OFFSET(exp_mode), AV_OPT_TYPE_CONST, {.i64=EXP_NORMAL}, 0, 0, FLAGS, "expansion"},
194 {"strftime", "set strftime expansion (deprecated)", OFFSET(exp_mode), AV_OPT_TYPE_CONST, {.i64=EXP_STRFTIME}, 0, 0, FLAGS, "expansion"},
196 {"timecode", "set initial timecode", OFFSET(tc_opt_string), AV_OPT_TYPE_STRING, {.str=NULL}, CHAR_MIN, CHAR_MAX, FLAGS},
197 {"tc24hmax", "set 24 hours max (timecode only)", OFFSET(tc24hmax), AV_OPT_TYPE_INT, {.i64=0}, 0, 1, FLAGS},
198 {"timecode_rate", "set rate (timecode only)", OFFSET(tc_rate), AV_OPT_TYPE_RATIONAL, {.dbl=0}, 0, INT_MAX, FLAGS},
199 {"r", "set rate (timecode only)", OFFSET(tc_rate), AV_OPT_TYPE_RATIONAL, {.dbl=0}, 0, INT_MAX, FLAGS},
200 {"rate", "set rate (timecode only)", OFFSET(tc_rate), AV_OPT_TYPE_RATIONAL, {.dbl=0}, 0, INT_MAX, FLAGS},
201 {"reload", "reload text file for each frame", OFFSET(reload), AV_OPT_TYPE_INT, {.i64=0}, 0, 1, FLAGS},
202 {"fix_bounds", "if true, check and fix text coords to avoid clipping", OFFSET(fix_bounds), AV_OPT_TYPE_INT, {.i64=1}, 0, 1, FLAGS},
203 {"start_number", "start frame number for n/frame_num variable", OFFSET(start_number), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, FLAGS},
205 /* FT_LOAD_* flags */
206 { "ft_load_flags", "set font loading flags for libfreetype", OFFSET(ft_load_flags), AV_OPT_TYPE_FLAGS, { .i64 = FT_LOAD_DEFAULT | FT_LOAD_RENDER}, 0, INT_MAX, FLAGS, "ft_load_flags" },
207 { "default", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_DEFAULT }, .flags = FLAGS, .unit = "ft_load_flags" },
208 { "no_scale", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_NO_SCALE }, .flags = FLAGS, .unit = "ft_load_flags" },
209 { "no_hinting", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_NO_HINTING }, .flags = FLAGS, .unit = "ft_load_flags" },
210 { "render", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_RENDER }, .flags = FLAGS, .unit = "ft_load_flags" },
211 { "no_bitmap", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_NO_BITMAP }, .flags = FLAGS, .unit = "ft_load_flags" },
212 { "vertical_layout", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_VERTICAL_LAYOUT }, .flags = FLAGS, .unit = "ft_load_flags" },
213 { "force_autohint", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_FORCE_AUTOHINT }, .flags = FLAGS, .unit = "ft_load_flags" },
214 { "crop_bitmap", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_CROP_BITMAP }, .flags = FLAGS, .unit = "ft_load_flags" },
215 { "pedantic", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_PEDANTIC }, .flags = FLAGS, .unit = "ft_load_flags" },
216 { "ignore_global_advance_width", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_IGNORE_GLOBAL_ADVANCE_WIDTH }, .flags = FLAGS, .unit = "ft_load_flags" },
217 { "no_recurse", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_NO_RECURSE }, .flags = FLAGS, .unit = "ft_load_flags" },
218 { "ignore_transform", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_IGNORE_TRANSFORM }, .flags = FLAGS, .unit = "ft_load_flags" },
219 { "monochrome", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_MONOCHROME }, .flags = FLAGS, .unit = "ft_load_flags" },
220 { "linear_design", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_LINEAR_DESIGN }, .flags = FLAGS, .unit = "ft_load_flags" },
221 { "no_autohint", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_NO_AUTOHINT }, .flags = FLAGS, .unit = "ft_load_flags" },
225 AVFILTER_DEFINE_CLASS(drawtext);
227 #undef __FTERRORS_H__
228 #define FT_ERROR_START_LIST {
229 #define FT_ERRORDEF(e, v, s) { (e), (s) },
230 #define FT_ERROR_END_LIST { 0, NULL } };
236 } static ft_errors[] =
239 #define FT_ERRMSG(e) ft_errors[e].err_msg
244 FT_Bitmap bitmap; ///< array holding bitmaps of font
251 static int glyph_cmp(void *key, const void *b)
253 const Glyph *a = key, *bb = b;
254 int64_t diff = (int64_t)a->code - (int64_t)bb->code;
255 return diff > 0 ? 1 : diff < 0 ? -1 : 0;
259 * Load glyphs corresponding to the UTF-32 codepoint code.
261 static int load_glyph(AVFilterContext *ctx, Glyph **glyph_ptr, uint32_t code)
263 DrawTextContext *s = ctx->priv;
265 struct AVTreeNode *node = NULL;
268 /* load glyph into s->face->glyph */
269 if (FT_Load_Char(s->face, code, s->ft_load_flags))
270 return AVERROR(EINVAL);
273 if (!(glyph = av_mallocz(sizeof(*glyph))) ||
274 !(glyph->glyph = av_mallocz(sizeof(*glyph->glyph)))) {
275 ret = AVERROR(ENOMEM);
280 if (FT_Get_Glyph(s->face->glyph, glyph->glyph)) {
281 ret = AVERROR(EINVAL);
285 glyph->bitmap = s->face->glyph->bitmap;
286 glyph->bitmap_left = s->face->glyph->bitmap_left;
287 glyph->bitmap_top = s->face->glyph->bitmap_top;
288 glyph->advance = s->face->glyph->advance.x >> 6;
290 /* measure text height to calculate text_height (or the maximum text height) */
291 FT_Glyph_Get_CBox(*glyph->glyph, ft_glyph_bbox_pixels, &glyph->bbox);
293 /* cache the newly created glyph */
294 if (!(node = av_tree_node_alloc())) {
295 ret = AVERROR(ENOMEM);
298 av_tree_insert(&s->glyphs, glyph, glyph_cmp, &node);
306 av_freep(&glyph->glyph);
312 static int load_font_file(AVFilterContext *ctx, const char *path, int index,
315 DrawTextContext *s = ctx->priv;
318 err = FT_New_Face(s->library, path, index, &s->face);
320 *error = FT_ERRMSG(err);
321 return AVERROR(EINVAL);
326 #if CONFIG_FONTCONFIG
327 static int load_font_fontconfig(AVFilterContext *ctx, const char **error)
329 DrawTextContext *s = ctx->priv;
330 FcConfig *fontconfig;
331 FcPattern *pattern, *fpat;
332 FcResult result = FcResultMatch;
337 fontconfig = FcInitLoadConfigAndFonts();
339 *error = "impossible to init fontconfig\n";
340 return AVERROR(EINVAL);
342 pattern = FcNameParse(s->fontfile ? s->fontfile :
343 (uint8_t *)(intptr_t)"default");
345 *error = "could not parse fontconfig pattern";
346 return AVERROR(EINVAL);
348 if (!FcConfigSubstitute(fontconfig, pattern, FcMatchPattern)) {
349 *error = "could not substitue fontconfig options"; /* very unlikely */
350 return AVERROR(EINVAL);
352 FcDefaultSubstitute(pattern);
353 fpat = FcFontMatch(fontconfig, pattern, &result);
354 if (!fpat || result != FcResultMatch) {
355 *error = "impossible to find a matching font";
356 return AVERROR(EINVAL);
358 if (FcPatternGetString (fpat, FC_FILE, 0, &filename) != FcResultMatch ||
359 FcPatternGetInteger(fpat, FC_INDEX, 0, &index ) != FcResultMatch ||
360 FcPatternGetDouble (fpat, FC_SIZE, 0, &size ) != FcResultMatch) {
361 *error = "impossible to find font information";
362 return AVERROR(EINVAL);
364 av_log(ctx, AV_LOG_INFO, "Using \"%s\"\n", filename);
366 s->fontsize = size + 0.5;
367 err = load_font_file(ctx, filename, index, error);
370 FcPatternDestroy(fpat);
371 FcPatternDestroy(pattern);
372 FcConfigDestroy(fontconfig);
377 static int load_font(AVFilterContext *ctx)
379 DrawTextContext *s = ctx->priv;
381 const char *error = "unknown error\n";
383 /* load the face, and set up the encoding, which is by default UTF-8 */
384 err = load_font_file(ctx, s->fontfile, 0, &error);
387 #if CONFIG_FONTCONFIG
388 err = load_font_fontconfig(ctx, &error);
392 av_log(ctx, AV_LOG_ERROR, "Could not load font \"%s\": %s\n",
397 static int load_textfile(AVFilterContext *ctx)
399 DrawTextContext *s = ctx->priv;
404 if ((err = av_file_map(s->textfile, &textbuf, &textbuf_size, 0, ctx)) < 0) {
405 av_log(ctx, AV_LOG_ERROR,
406 "The text file '%s' could not be read or is empty\n",
411 if (!(s->text = av_realloc(s->text, textbuf_size + 1)))
412 return AVERROR(ENOMEM);
413 memcpy(s->text, textbuf, textbuf_size);
414 s->text[textbuf_size] = 0;
415 av_file_unmap(textbuf, textbuf_size);
420 static av_cold int init(AVFilterContext *ctx)
423 DrawTextContext *s = ctx->priv;
426 if (!s->fontfile && !CONFIG_FONTCONFIG) {
427 av_log(ctx, AV_LOG_ERROR, "No font filename provided\n");
428 return AVERROR(EINVAL);
433 av_log(ctx, AV_LOG_ERROR,
434 "Both text and text file provided. Please provide only one\n");
435 return AVERROR(EINVAL);
437 if ((err = load_textfile(ctx)) < 0)
441 if (s->reload && !s->textfile)
442 av_log(ctx, AV_LOG_WARNING, "No file to reload\n");
444 if (s->tc_opt_string) {
445 int ret = av_timecode_init_from_string(&s->tc, s->tc_rate,
446 s->tc_opt_string, ctx);
450 s->tc.flags |= AV_TIMECODE_FLAG_24HOURSMAX;
452 s->text = av_strdup("");
456 av_log(ctx, AV_LOG_ERROR,
457 "Either text, a valid file or a timecode must be provided\n");
458 return AVERROR(EINVAL);
461 if ((err = FT_Init_FreeType(&(s->library)))) {
462 av_log(ctx, AV_LOG_ERROR,
463 "Could not load FreeType: %s\n", FT_ERRMSG(err));
464 return AVERROR(EINVAL);
467 err = load_font(ctx);
472 if ((err = FT_Set_Pixel_Sizes(s->face, 0, s->fontsize))) {
473 av_log(ctx, AV_LOG_ERROR, "Could not set font size to %d pixels: %s\n",
474 s->fontsize, FT_ERRMSG(err));
475 return AVERROR(EINVAL);
478 s->use_kerning = FT_HAS_KERNING(s->face);
480 /* load the fallback glyph with code 0 */
481 load_glyph(ctx, NULL, 0);
483 /* set the tabsize in pixels */
484 if ((err = load_glyph(ctx, &glyph, ' ')) < 0) {
485 av_log(ctx, AV_LOG_ERROR, "Could not set tabsize.\n");
488 s->tabsize *= glyph->advance;
490 if (s->exp_mode == EXP_STRFTIME &&
491 (strchr(s->text, '%') || strchr(s->text, '\\')))
492 av_log(ctx, AV_LOG_WARNING, "expansion=strftime is deprecated.\n");
494 av_bprint_init(&s->expanded_text, 0, AV_BPRINT_SIZE_UNLIMITED);
499 static int query_formats(AVFilterContext *ctx)
501 ff_set_common_formats(ctx, ff_draw_supported_pixel_formats(0));
505 static int glyph_enu_free(void *opaque, void *elem)
509 FT_Done_Glyph(*glyph->glyph);
510 av_freep(&glyph->glyph);
515 static av_cold void uninit(AVFilterContext *ctx)
517 DrawTextContext *s = ctx->priv;
519 av_expr_free(s->x_pexpr);
520 av_expr_free(s->y_pexpr);
521 av_expr_free(s->draw_pexpr);
522 s->x_pexpr = s->y_pexpr = s->draw_pexpr = NULL;
523 av_freep(&s->positions);
527 av_tree_enumerate(s->glyphs, NULL, NULL, glyph_enu_free);
528 av_tree_destroy(s->glyphs);
531 FT_Done_Face(s->face);
532 FT_Done_FreeType(s->library);
534 av_bprint_finalize(&s->expanded_text, NULL);
537 static inline int is_newline(uint32_t c)
539 return c == '\n' || c == '\r' || c == '\f' || c == '\v';
542 static int config_input(AVFilterLink *inlink)
544 AVFilterContext *ctx = inlink->dst;
545 DrawTextContext *s = ctx->priv;
548 ff_draw_init(&s->dc, inlink->format, 0);
549 ff_draw_color(&s->dc, &s->fontcolor, s->fontcolor.rgba);
550 ff_draw_color(&s->dc, &s->shadowcolor, s->shadowcolor.rgba);
551 ff_draw_color(&s->dc, &s->boxcolor, s->boxcolor.rgba);
553 s->var_values[VAR_w] = s->var_values[VAR_W] = s->var_values[VAR_MAIN_W] = inlink->w;
554 s->var_values[VAR_h] = s->var_values[VAR_H] = s->var_values[VAR_MAIN_H] = inlink->h;
555 s->var_values[VAR_SAR] = inlink->sample_aspect_ratio.num ? av_q2d(inlink->sample_aspect_ratio) : 1;
556 s->var_values[VAR_DAR] = (double)inlink->w / inlink->h * s->var_values[VAR_SAR];
557 s->var_values[VAR_HSUB] = 1 << s->dc.hsub_max;
558 s->var_values[VAR_VSUB] = 1 << s->dc.vsub_max;
559 s->var_values[VAR_X] = NAN;
560 s->var_values[VAR_Y] = NAN;
561 s->var_values[VAR_T] = NAN;
563 av_lfg_init(&s->prng, av_get_random_seed());
565 av_expr_free(s->x_pexpr);
566 av_expr_free(s->y_pexpr);
567 av_expr_free(s->draw_pexpr);
568 s->x_pexpr = s->y_pexpr = s->draw_pexpr = NULL;
569 if ((ret = av_expr_parse(&s->x_pexpr, s->x_expr, var_names,
570 NULL, NULL, fun2_names, fun2, 0, ctx)) < 0 ||
571 (ret = av_expr_parse(&s->y_pexpr, s->y_expr, var_names,
572 NULL, NULL, fun2_names, fun2, 0, ctx)) < 0 ||
573 (ret = av_expr_parse(&s->draw_pexpr, s->draw_expr, var_names,
574 NULL, NULL, fun2_names, fun2, 0, ctx)) < 0)
576 return AVERROR(EINVAL);
581 static int command(AVFilterContext *ctx, const char *cmd, const char *arg, char *res, int res_len, int flags)
583 DrawTextContext *s = ctx->priv;
585 if (!strcmp(cmd, "reinit")) {
589 if ((ret = init(ctx)) < 0)
591 return config_input(ctx->inputs[0]);
594 return AVERROR(ENOSYS);
597 static int func_pict_type(AVFilterContext *ctx, AVBPrint *bp,
598 char *fct, unsigned argc, char **argv, int tag)
600 DrawTextContext *s = ctx->priv;
602 av_bprintf(bp, "%c", av_get_picture_type_char(s->var_values[VAR_PICT_TYPE]));
606 static int func_pts(AVFilterContext *ctx, AVBPrint *bp,
607 char *fct, unsigned argc, char **argv, int tag)
609 DrawTextContext *s = ctx->priv;
611 av_bprintf(bp, "%.6f", s->var_values[VAR_T]);
615 static int func_frame_num(AVFilterContext *ctx, AVBPrint *bp,
616 char *fct, unsigned argc, char **argv, int tag)
618 DrawTextContext *s = ctx->priv;
620 av_bprintf(bp, "%d", (int)s->var_values[VAR_N]);
624 static int func_metadata(AVFilterContext *ctx, AVBPrint *bp,
625 char *fct, unsigned argc, char **argv, int tag)
627 DrawTextContext *s = ctx->priv;
628 AVDictionaryEntry *e = av_dict_get(s->metadata, argv[0], NULL, 0);
631 av_bprintf(bp, "%s", e->value);
635 #if !HAVE_LOCALTIME_R
636 static void localtime_r(const time_t *t, struct tm *tm)
642 static int func_strftime(AVFilterContext *ctx, AVBPrint *bp,
643 char *fct, unsigned argc, char **argv, int tag)
645 const char *fmt = argc ? argv[0] : "%Y-%m-%d %H:%M:%S";
651 localtime_r(&now, &tm);
654 av_bprint_strftime(bp, fmt, &tm);
658 static int func_eval_expr(AVFilterContext *ctx, AVBPrint *bp,
659 char *fct, unsigned argc, char **argv, int tag)
661 DrawTextContext *s = ctx->priv;
665 ret = av_expr_parse_and_eval(&res, argv[0], var_names, s->var_values,
666 NULL, NULL, fun2_names, fun2,
669 av_log(ctx, AV_LOG_ERROR,
670 "Expression '%s' for the expr text expansion function is not valid\n",
673 av_bprintf(bp, "%f", res);
678 static const struct drawtext_function {
680 unsigned argc_min, argc_max;
681 int tag; /**< opaque argument to func */
682 int (*func)(AVFilterContext *, AVBPrint *, char *, unsigned, char **, int);
684 { "expr", 1, 1, 0, func_eval_expr },
685 { "e", 1, 1, 0, func_eval_expr },
686 { "pict_type", 0, 0, 0, func_pict_type },
687 { "pts", 0, 0, 0, func_pts },
688 { "gmtime", 0, 1, 'G', func_strftime },
689 { "localtime", 0, 1, 'L', func_strftime },
690 { "frame_num", 0, 0, 0, func_frame_num },
691 { "n", 0, 0, 0, func_frame_num },
692 { "metadata", 1, 1, 0, func_metadata },
695 static int eval_function(AVFilterContext *ctx, AVBPrint *bp, char *fct,
696 unsigned argc, char **argv)
700 for (i = 0; i < FF_ARRAY_ELEMS(functions); i++) {
701 if (strcmp(fct, functions[i].name))
703 if (argc < functions[i].argc_min) {
704 av_log(ctx, AV_LOG_ERROR, "%%{%s} requires at least %d arguments\n",
705 fct, functions[i].argc_min);
706 return AVERROR(EINVAL);
708 if (argc > functions[i].argc_max) {
709 av_log(ctx, AV_LOG_ERROR, "%%{%s} requires at most %d arguments\n",
710 fct, functions[i].argc_max);
711 return AVERROR(EINVAL);
715 if (i >= FF_ARRAY_ELEMS(functions)) {
716 av_log(ctx, AV_LOG_ERROR, "%%{%s} is not known\n", fct);
717 return AVERROR(EINVAL);
719 return functions[i].func(ctx, bp, fct, argc, argv, functions[i].tag);
722 static int expand_function(AVFilterContext *ctx, AVBPrint *bp, char **rtext)
724 const char *text = *rtext;
725 char *argv[16] = { NULL };
726 unsigned argc = 0, i;
730 av_log(ctx, AV_LOG_ERROR, "Stray %% near '%s'\n", text);
731 return AVERROR(EINVAL);
735 if (!(argv[argc++] = av_get_token(&text, ":}"))) {
736 ret = AVERROR(ENOMEM);
740 av_log(ctx, AV_LOG_ERROR, "Unterminated %%{} near '%s'\n", *rtext);
741 ret = AVERROR(EINVAL);
744 if (argc == FF_ARRAY_ELEMS(argv))
745 av_freep(&argv[--argc]); /* error will be caught later */
751 if ((ret = eval_function(ctx, bp, argv[0], argc - 1, argv + 1)) < 0)
754 *rtext = (char *)text + 1;
757 for (i = 0; i < argc; i++)
762 static int expand_text(AVFilterContext *ctx)
764 DrawTextContext *s = ctx->priv;
765 char *text = s->text;
766 AVBPrint *bp = &s->expanded_text;
771 if (*text == '\\' && text[1]) {
772 av_bprint_chars(bp, text[1], 1);
774 } else if (*text == '%') {
776 if ((ret = expand_function(ctx, bp, &text)) < 0)
779 av_bprint_chars(bp, *text, 1);
783 if (!av_bprint_is_complete(bp))
784 return AVERROR(ENOMEM);
788 static int draw_glyphs(DrawTextContext *s, AVFrame *frame,
789 int width, int height, const uint8_t rgbcolor[4], FFDrawColor *color, int x, int y)
791 char *text = s->expanded_text.str;
797 for (i = 0, p = text; *p; i++) {
799 GET_UTF8(code, *p++, continue;);
801 /* skip new line chars, just go to new line */
802 if (code == '\n' || code == '\r' || code == '\t')
806 glyph = av_tree_find(s->glyphs, &dummy, (void *)glyph_cmp, NULL);
808 if (glyph->bitmap.pixel_mode != FT_PIXEL_MODE_MONO &&
809 glyph->bitmap.pixel_mode != FT_PIXEL_MODE_GRAY)
810 return AVERROR(EINVAL);
812 x1 = s->positions[i].x+s->x+x;
813 y1 = s->positions[i].y+s->y+y;
815 ff_blend_mask(&s->dc, color,
816 frame->data, frame->linesize, width, height,
817 glyph->bitmap.buffer, glyph->bitmap.pitch,
818 glyph->bitmap.width, glyph->bitmap.rows,
819 glyph->bitmap.pixel_mode == FT_PIXEL_MODE_MONO ? 0 : 3,
826 static int draw_text(AVFilterContext *ctx, AVFrame *frame,
827 int width, int height)
829 DrawTextContext *s = ctx->priv;
830 AVFilterLink *inlink = ctx->inputs[0];
832 uint32_t code = 0, prev_code = 0;
833 int x = 0, y = 0, i = 0, ret;
834 int max_text_line_w = 0, len;
836 char *text = s->text;
838 int y_min = 32000, y_max = -32000;
839 int x_min = 32000, x_max = -32000;
841 Glyph *glyph = NULL, *prev_glyph = NULL;
844 time_t now = time(0);
846 AVBPrint *bp = &s->expanded_text;
850 if(s->basetime != AV_NOPTS_VALUE)
851 now= frame->pts*av_q2d(ctx->inputs[0]->time_base) + s->basetime/1000000;
853 switch (s->exp_mode) {
855 av_bprintf(bp, "%s", s->text);
858 if ((ret = expand_text(ctx)) < 0)
862 localtime_r(&now, <ime);
863 av_bprint_strftime(bp, s->text, <ime);
867 if (s->tc_opt_string) {
868 char tcbuf[AV_TIMECODE_STR_SIZE];
869 av_timecode_make_string(&s->tc, tcbuf, inlink->frame_count);
871 av_bprintf(bp, "%s%s", s->text, tcbuf);
874 if (!av_bprint_is_complete(bp))
875 return AVERROR(ENOMEM);
876 text = s->expanded_text.str;
877 if ((len = s->expanded_text.len) > s->nb_positions) {
879 av_realloc(s->positions, len*sizeof(*s->positions))))
880 return AVERROR(ENOMEM);
881 s->nb_positions = len;
887 /* load and cache glyphs */
888 for (i = 0, p = text; *p; i++) {
889 GET_UTF8(code, *p++, continue;);
893 glyph = av_tree_find(s->glyphs, &dummy, glyph_cmp, NULL);
895 load_glyph(ctx, &glyph, code);
898 y_min = FFMIN(glyph->bbox.yMin, y_min);
899 y_max = FFMAX(glyph->bbox.yMax, y_max);
900 x_min = FFMIN(glyph->bbox.xMin, x_min);
901 x_max = FFMAX(glyph->bbox.xMax, x_max);
903 s->max_glyph_h = y_max - y_min;
904 s->max_glyph_w = x_max - x_min;
906 /* compute and save position for each glyph */
908 for (i = 0, p = text; *p; i++) {
909 GET_UTF8(code, *p++, continue;);
911 /* skip the \n in the sequence \r\n */
912 if (prev_code == '\r' && code == '\n')
916 if (is_newline(code)) {
918 max_text_line_w = FFMAX(max_text_line_w, x);
927 glyph = av_tree_find(s->glyphs, &dummy, glyph_cmp, NULL);
930 if (s->use_kerning && prev_glyph && glyph->code) {
931 FT_Get_Kerning(s->face, prev_glyph->code, glyph->code,
932 ft_kerning_default, &delta);
937 s->positions[i].x = x + glyph->bitmap_left;
938 s->positions[i].y = y - glyph->bitmap_top + y_max;
939 if (code == '\t') x = (x / s->tabsize + 1)*s->tabsize;
940 else x += glyph->advance;
943 max_text_line_w = FFMAX(x, max_text_line_w);
945 s->var_values[VAR_TW] = s->var_values[VAR_TEXT_W] = max_text_line_w;
946 s->var_values[VAR_TH] = s->var_values[VAR_TEXT_H] = y + s->max_glyph_h;
948 s->var_values[VAR_MAX_GLYPH_W] = s->max_glyph_w;
949 s->var_values[VAR_MAX_GLYPH_H] = s->max_glyph_h;
950 s->var_values[VAR_MAX_GLYPH_A] = s->var_values[VAR_ASCENT ] = y_max;
951 s->var_values[VAR_MAX_GLYPH_D] = s->var_values[VAR_DESCENT] = y_min;
953 s->var_values[VAR_LINE_H] = s->var_values[VAR_LH] = s->max_glyph_h;
955 s->x = s->var_values[VAR_X] = av_expr_eval(s->x_pexpr, s->var_values, &s->prng);
956 s->y = s->var_values[VAR_Y] = av_expr_eval(s->y_pexpr, s->var_values, &s->prng);
957 s->x = s->var_values[VAR_X] = av_expr_eval(s->x_pexpr, s->var_values, &s->prng);
958 s->draw = av_expr_eval(s->draw_pexpr, s->var_values, &s->prng);
963 box_w = FFMIN(width - 1 , max_text_line_w);
964 box_h = FFMIN(height - 1, y + s->max_glyph_h);
968 ff_blend_rectangle(&s->dc, &s->boxcolor,
969 frame->data, frame->linesize, width, height,
970 s->x, s->y, box_w, box_h);
972 if (s->shadowx || s->shadowy) {
973 if ((ret = draw_glyphs(s, frame, width, height, s->shadowcolor.rgba,
974 &s->shadowcolor, s->shadowx, s->shadowy)) < 0)
978 if ((ret = draw_glyphs(s, frame, width, height, s->fontcolor.rgba,
979 &s->fontcolor, 0, 0)) < 0)
985 static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
987 AVFilterContext *ctx = inlink->dst;
988 AVFilterLink *outlink = ctx->outputs[0];
989 DrawTextContext *s = ctx->priv;
993 if ((ret = load_textfile(ctx)) < 0)
996 s->var_values[VAR_N] = inlink->frame_count+s->start_number;
997 s->var_values[VAR_T] = frame->pts == AV_NOPTS_VALUE ?
998 NAN : frame->pts * av_q2d(inlink->time_base);
1000 s->var_values[VAR_PICT_TYPE] = frame->pict_type;
1001 s->metadata = av_frame_get_metadata(frame);
1003 draw_text(ctx, frame, frame->width, frame->height);
1005 av_log(ctx, AV_LOG_DEBUG, "n:%d t:%f text_w:%d text_h:%d x:%d y:%d\n",
1006 (int)s->var_values[VAR_N], s->var_values[VAR_T],
1007 (int)s->var_values[VAR_TEXT_W], (int)s->var_values[VAR_TEXT_H],
1010 return ff_filter_frame(outlink, frame);
1013 static const AVFilterPad avfilter_vf_drawtext_inputs[] = {
1016 .type = AVMEDIA_TYPE_VIDEO,
1017 .get_video_buffer = ff_null_get_video_buffer,
1018 .filter_frame = filter_frame,
1019 .config_props = config_input,
1020 .needs_writable = 1,
1025 static const AVFilterPad avfilter_vf_drawtext_outputs[] = {
1028 .type = AVMEDIA_TYPE_VIDEO,
1033 AVFilter avfilter_vf_drawtext = {
1035 .description = NULL_IF_CONFIG_SMALL("Draw text on top of video frames using libfreetype library."),
1036 .priv_size = sizeof(DrawTextContext),
1037 .priv_class = &drawtext_class,
1040 .query_formats = query_formats,
1042 .inputs = avfilter_vf_drawtext_inputs,
1043 .outputs = avfilter_vf_drawtext_outputs,
1044 .process_command = command,