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
34 #include <sys/types.h>
42 #if CONFIG_LIBFONTCONFIG
43 #include <fontconfig/fontconfig.h>
46 #include "libavutil/avstring.h"
47 #include "libavutil/bprint.h"
48 #include "libavutil/common.h"
49 #include "libavutil/file.h"
50 #include "libavutil/eval.h"
51 #include "libavutil/opt.h"
52 #include "libavutil/random_seed.h"
53 #include "libavutil/parseutils.h"
54 #include "libavutil/timecode.h"
55 #include "libavutil/time_internal.h"
56 #include "libavutil/tree.h"
57 #include "libavutil/lfg.h"
59 #include "drawutils.h"
69 #include FT_FREETYPE_H
73 static const char *const var_names[] = {
76 "line_h", "lh", ///< line height, same as max_glyph_h
77 "main_h", "h", "H", ///< height of the input video
78 "main_w", "w", "W", ///< width of the input video
79 "max_glyph_a", "ascent", ///< max glyph ascent
80 "max_glyph_d", "descent", ///< min glyph descent
81 "max_glyph_h", ///< max glyph height
82 "max_glyph_w", ///< max glyph width
83 "n", ///< number of frame
85 "t", ///< timestamp expressed in seconds
86 "text_h", "th", ///< height of the rendered text
87 "text_w", "tw", ///< width of the rendered text
94 static const char *const fun2_names[] = {
98 static double drand(void *opaque, double min, double max)
100 return min + (max-min) / UINT_MAX * av_lfg_get(opaque);
103 typedef double (*eval_func2)(void *, double a, double b);
105 static const eval_func2 fun2[] = {
114 VAR_MAIN_H, VAR_h, VAR_H,
115 VAR_MAIN_W, VAR_w, VAR_W,
116 VAR_MAX_GLYPH_A, VAR_ASCENT,
117 VAR_MAX_GLYPH_D, VAR_DESCENT,
131 enum expansion_mode {
137 typedef struct DrawTextContext {
138 const AVClass *class;
139 int exp_mode; ///< expansion mode to use for the text
140 int reinit; ///< tells if the filter is being reinited
141 #if CONFIG_LIBFONTCONFIG
142 uint8_t *font; ///< font to be used
144 uint8_t *fontfile; ///< font to be used
145 uint8_t *text; ///< text to be drawn
146 AVBPrint expanded_text; ///< used to contain the expanded text
147 uint8_t *fontcolor_expr; ///< fontcolor expression to evaluate
148 AVBPrint expanded_fontcolor; ///< used to contain the expanded fontcolor spec
149 int ft_load_flags; ///< flags used for loading fonts, see FT_LOAD_*
150 FT_Vector *positions; ///< positions for each element in the text
151 size_t nb_positions; ///< number of elements of positions array
152 char *textfile; ///< file with text to be drawn
153 int x; ///< x position to start drawing text
154 int y; ///< y position to start drawing text
155 int max_glyph_w; ///< max glyph width
156 int max_glyph_h; ///< max glyph height
157 int shadowx, shadowy;
158 int borderw; ///< border width
159 unsigned int fontsize; ///< font size to use
161 short int draw_box; ///< draw box around text - true or false
162 int boxborderw; ///< box border width
163 int use_kerning; ///< font kerning is used - true/false
164 int tabsize; ///< tab size
165 int fix_bounds; ///< do we let it go out of frame bounds - t/f
168 FFDrawColor fontcolor; ///< foreground color
169 FFDrawColor shadowcolor; ///< shadow color
170 FFDrawColor bordercolor; ///< border color
171 FFDrawColor boxcolor; ///< background color
173 FT_Library library; ///< freetype font library handle
174 FT_Face face; ///< freetype font face handle
175 FT_Stroker stroker; ///< freetype stroker handle
176 struct AVTreeNode *glyphs; ///< rendered glyphs, stored using the UTF-32 char code
177 char *x_expr; ///< expression for x position
178 char *y_expr; ///< expression for y position
179 AVExpr *x_pexpr, *y_pexpr; ///< parsed expressions for x and y
180 int64_t basetime; ///< base pts time in the real world for display
181 double var_values[VAR_VARS_NB];
182 AVLFG prng; ///< random
183 char *tc_opt_string; ///< specified timecode option string
184 AVRational tc_rate; ///< frame rate for timecode
185 AVTimecode tc; ///< timecode context
186 int tc24hmax; ///< 1 if timecode is wrapped to 24 hours, 0 otherwise
187 int reload; ///< reload text file for each frame
188 int start_number; ///< starting frame number for n/frame_num var
189 #if CONFIG_LIBFRIBIDI
190 int text_shaping; ///< 1 to shape the text before drawing it
192 AVDictionary *metadata;
195 #define OFFSET(x) offsetof(DrawTextContext, x)
196 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
198 static const AVOption drawtext_options[]= {
199 {"fontfile", "set font file", OFFSET(fontfile), AV_OPT_TYPE_STRING, {.str=NULL}, CHAR_MIN, CHAR_MAX, FLAGS},
200 {"text", "set text", OFFSET(text), AV_OPT_TYPE_STRING, {.str=NULL}, CHAR_MIN, CHAR_MAX, FLAGS},
201 {"textfile", "set text file", OFFSET(textfile), AV_OPT_TYPE_STRING, {.str=NULL}, CHAR_MIN, CHAR_MAX, FLAGS},
202 {"fontcolor", "set foreground color", OFFSET(fontcolor.rgba), AV_OPT_TYPE_COLOR, {.str="black"}, CHAR_MIN, CHAR_MAX, FLAGS},
203 {"fontcolor_expr", "set foreground color expression", OFFSET(fontcolor_expr), AV_OPT_TYPE_STRING, {.str=""}, CHAR_MIN, CHAR_MAX, FLAGS},
204 {"boxcolor", "set box color", OFFSET(boxcolor.rgba), AV_OPT_TYPE_COLOR, {.str="white"}, CHAR_MIN, CHAR_MAX, FLAGS},
205 {"bordercolor", "set border color", OFFSET(bordercolor.rgba), AV_OPT_TYPE_COLOR, {.str="black"}, CHAR_MIN, CHAR_MAX, FLAGS},
206 {"shadowcolor", "set shadow color", OFFSET(shadowcolor.rgba), AV_OPT_TYPE_COLOR, {.str="black"}, CHAR_MIN, CHAR_MAX, FLAGS},
207 {"box", "set box", OFFSET(draw_box), AV_OPT_TYPE_INT, {.i64=0}, 0, 1 , FLAGS},
208 {"boxborderw", "set box border width", OFFSET(boxborderw), AV_OPT_TYPE_INT, {.i64=0}, INT_MIN, INT_MAX , FLAGS},
209 {"fontsize", "set font size", OFFSET(fontsize), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX , FLAGS},
210 {"x", "set x expression", OFFSET(x_expr), AV_OPT_TYPE_STRING, {.str="0"}, CHAR_MIN, CHAR_MAX, FLAGS},
211 {"y", "set y expression", OFFSET(y_expr), AV_OPT_TYPE_STRING, {.str="0"}, CHAR_MIN, CHAR_MAX, FLAGS},
212 {"shadowx", "set x", OFFSET(shadowx), AV_OPT_TYPE_INT, {.i64=0}, INT_MIN, INT_MAX , FLAGS},
213 {"shadowy", "set y", OFFSET(shadowy), AV_OPT_TYPE_INT, {.i64=0}, INT_MIN, INT_MAX , FLAGS},
214 {"borderw", "set border width", OFFSET(borderw), AV_OPT_TYPE_INT, {.i64=0}, INT_MIN, INT_MAX , FLAGS},
215 {"tabsize", "set tab size", OFFSET(tabsize), AV_OPT_TYPE_INT, {.i64=4}, 0, INT_MAX , FLAGS},
216 {"basetime", "set base time", OFFSET(basetime), AV_OPT_TYPE_INT64, {.i64=AV_NOPTS_VALUE}, INT64_MIN, INT64_MAX , FLAGS},
217 #if CONFIG_LIBFONTCONFIG
218 { "font", "Font name", OFFSET(font), AV_OPT_TYPE_STRING, { .str = "Sans" }, .flags = FLAGS },
221 {"expansion", "set the expansion mode", OFFSET(exp_mode), AV_OPT_TYPE_INT, {.i64=EXP_NORMAL}, 0, 2, FLAGS, "expansion"},
222 {"none", "set no expansion", OFFSET(exp_mode), AV_OPT_TYPE_CONST, {.i64=EXP_NONE}, 0, 0, FLAGS, "expansion"},
223 {"normal", "set normal expansion", OFFSET(exp_mode), AV_OPT_TYPE_CONST, {.i64=EXP_NORMAL}, 0, 0, FLAGS, "expansion"},
224 {"strftime", "set strftime expansion (deprecated)", OFFSET(exp_mode), AV_OPT_TYPE_CONST, {.i64=EXP_STRFTIME}, 0, 0, FLAGS, "expansion"},
226 {"timecode", "set initial timecode", OFFSET(tc_opt_string), AV_OPT_TYPE_STRING, {.str=NULL}, CHAR_MIN, CHAR_MAX, FLAGS},
227 {"tc24hmax", "set 24 hours max (timecode only)", OFFSET(tc24hmax), AV_OPT_TYPE_INT, {.i64=0}, 0, 1, FLAGS},
228 {"timecode_rate", "set rate (timecode only)", OFFSET(tc_rate), AV_OPT_TYPE_RATIONAL, {.dbl=0}, 0, INT_MAX, FLAGS},
229 {"r", "set rate (timecode only)", OFFSET(tc_rate), AV_OPT_TYPE_RATIONAL, {.dbl=0}, 0, INT_MAX, FLAGS},
230 {"rate", "set rate (timecode only)", OFFSET(tc_rate), AV_OPT_TYPE_RATIONAL, {.dbl=0}, 0, INT_MAX, FLAGS},
231 {"reload", "reload text file for each frame", OFFSET(reload), AV_OPT_TYPE_INT, {.i64=0}, 0, 1, FLAGS},
232 {"fix_bounds", "if true, check and fix text coords to avoid clipping", OFFSET(fix_bounds), AV_OPT_TYPE_INT, {.i64=1}, 0, 1, FLAGS},
233 {"start_number", "start frame number for n/frame_num variable", OFFSET(start_number), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, FLAGS},
235 #if CONFIG_LIBFRIBIDI
236 {"text_shaping", "attempt to shape text before drawing", OFFSET(text_shaping), AV_OPT_TYPE_INT, {.i64=1}, 0, 1, FLAGS},
239 /* FT_LOAD_* flags */
240 { "ft_load_flags", "set font loading flags for libfreetype", OFFSET(ft_load_flags), AV_OPT_TYPE_FLAGS, { .i64 = FT_LOAD_DEFAULT }, 0, INT_MAX, FLAGS, "ft_load_flags" },
241 { "default", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_DEFAULT }, .flags = FLAGS, .unit = "ft_load_flags" },
242 { "no_scale", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_NO_SCALE }, .flags = FLAGS, .unit = "ft_load_flags" },
243 { "no_hinting", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_NO_HINTING }, .flags = FLAGS, .unit = "ft_load_flags" },
244 { "render", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_RENDER }, .flags = FLAGS, .unit = "ft_load_flags" },
245 { "no_bitmap", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_NO_BITMAP }, .flags = FLAGS, .unit = "ft_load_flags" },
246 { "vertical_layout", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_VERTICAL_LAYOUT }, .flags = FLAGS, .unit = "ft_load_flags" },
247 { "force_autohint", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_FORCE_AUTOHINT }, .flags = FLAGS, .unit = "ft_load_flags" },
248 { "crop_bitmap", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_CROP_BITMAP }, .flags = FLAGS, .unit = "ft_load_flags" },
249 { "pedantic", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_PEDANTIC }, .flags = FLAGS, .unit = "ft_load_flags" },
250 { "ignore_global_advance_width", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_IGNORE_GLOBAL_ADVANCE_WIDTH }, .flags = FLAGS, .unit = "ft_load_flags" },
251 { "no_recurse", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_NO_RECURSE }, .flags = FLAGS, .unit = "ft_load_flags" },
252 { "ignore_transform", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_IGNORE_TRANSFORM }, .flags = FLAGS, .unit = "ft_load_flags" },
253 { "monochrome", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_MONOCHROME }, .flags = FLAGS, .unit = "ft_load_flags" },
254 { "linear_design", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_LINEAR_DESIGN }, .flags = FLAGS, .unit = "ft_load_flags" },
255 { "no_autohint", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = FT_LOAD_NO_AUTOHINT }, .flags = FLAGS, .unit = "ft_load_flags" },
259 AVFILTER_DEFINE_CLASS(drawtext);
261 #undef __FTERRORS_H__
262 #define FT_ERROR_START_LIST {
263 #define FT_ERRORDEF(e, v, s) { (e), (s) },
264 #define FT_ERROR_END_LIST { 0, NULL } };
266 static const struct ft_error
273 #define FT_ERRMSG(e) ft_errors[e].err_msg
275 typedef struct Glyph {
277 FT_Glyph border_glyph;
279 FT_Bitmap bitmap; ///< array holding bitmaps of font
280 FT_Bitmap border_bitmap; ///< array holding bitmaps of font border
287 static int glyph_cmp(void *key, const void *b)
289 const Glyph *a = key, *bb = b;
290 int64_t diff = (int64_t)a->code - (int64_t)bb->code;
291 return diff > 0 ? 1 : diff < 0 ? -1 : 0;
295 * Load glyphs corresponding to the UTF-32 codepoint code.
297 static int load_glyph(AVFilterContext *ctx, Glyph **glyph_ptr, uint32_t code)
299 DrawTextContext *s = ctx->priv;
300 FT_BitmapGlyph bitmapglyph;
302 struct AVTreeNode *node = NULL;
305 /* load glyph into s->face->glyph */
306 if (FT_Load_Char(s->face, code, s->ft_load_flags))
307 return AVERROR(EINVAL);
309 glyph = av_mallocz(sizeof(*glyph));
311 ret = AVERROR(ENOMEM);
316 if (FT_Get_Glyph(s->face->glyph, &glyph->glyph)) {
317 ret = AVERROR(EINVAL);
321 glyph->border_glyph = glyph->glyph;
322 if (FT_Glyph_StrokeBorder(&glyph->border_glyph, s->stroker, 0, 0) ||
323 FT_Glyph_To_Bitmap(&glyph->border_glyph, FT_RENDER_MODE_NORMAL, 0, 1)) {
324 ret = AVERROR_EXTERNAL;
327 bitmapglyph = (FT_BitmapGlyph) glyph->border_glyph;
328 glyph->border_bitmap = bitmapglyph->bitmap;
330 if (FT_Glyph_To_Bitmap(&glyph->glyph, FT_RENDER_MODE_NORMAL, 0, 1)) {
331 ret = AVERROR_EXTERNAL;
334 bitmapglyph = (FT_BitmapGlyph) glyph->glyph;
336 glyph->bitmap = bitmapglyph->bitmap;
337 glyph->bitmap_left = bitmapglyph->left;
338 glyph->bitmap_top = bitmapglyph->top;
339 glyph->advance = s->face->glyph->advance.x >> 6;
341 /* measure text height to calculate text_height (or the maximum text height) */
342 FT_Glyph_Get_CBox(glyph->glyph, ft_glyph_bbox_pixels, &glyph->bbox);
344 /* cache the newly created glyph */
345 if (!(node = av_tree_node_alloc())) {
346 ret = AVERROR(ENOMEM);
349 av_tree_insert(&s->glyphs, glyph, glyph_cmp, &node);
357 av_freep(&glyph->glyph);
364 static int load_font_file(AVFilterContext *ctx, const char *path, int index)
366 DrawTextContext *s = ctx->priv;
369 err = FT_New_Face(s->library, path, index, &s->face);
371 av_log(ctx, AV_LOG_ERROR, "Could not load font \"%s\": %s\n",
372 s->fontfile, FT_ERRMSG(err));
373 return AVERROR(EINVAL);
378 #if CONFIG_LIBFONTCONFIG
379 static int load_font_fontconfig(AVFilterContext *ctx)
381 DrawTextContext *s = ctx->priv;
382 FcConfig *fontconfig;
383 FcPattern *pat, *best;
384 FcResult result = FcResultMatch;
388 int err = AVERROR(ENOENT);
390 fontconfig = FcInitLoadConfigAndFonts();
392 av_log(ctx, AV_LOG_ERROR, "impossible to init fontconfig\n");
393 return AVERROR_UNKNOWN;
395 pat = FcNameParse(s->fontfile ? s->fontfile :
396 (uint8_t *)(intptr_t)"default");
398 av_log(ctx, AV_LOG_ERROR, "could not parse fontconfig pat");
399 return AVERROR(EINVAL);
402 FcPatternAddString(pat, FC_FAMILY, s->font);
404 FcPatternAddDouble(pat, FC_SIZE, (double)s->fontsize);
406 FcDefaultSubstitute(pat);
408 if (!FcConfigSubstitute(fontconfig, pat, FcMatchPattern)) {
409 av_log(ctx, AV_LOG_ERROR, "could not substitue fontconfig options"); /* very unlikely */
410 FcPatternDestroy(pat);
411 return AVERROR(ENOMEM);
414 best = FcFontMatch(fontconfig, pat, &result);
415 FcPatternDestroy(pat);
417 if (!best || result != FcResultMatch) {
418 av_log(ctx, AV_LOG_ERROR,
419 "Cannot find a valid font for the family %s\n",
425 FcPatternGetInteger(best, FC_INDEX, 0, &index ) != FcResultMatch ||
426 FcPatternGetDouble (best, FC_SIZE, 0, &size ) != FcResultMatch) {
427 av_log(ctx, AV_LOG_ERROR, "impossible to find font information");
428 return AVERROR(EINVAL);
431 if (FcPatternGetString(best, FC_FILE, 0, &filename) != FcResultMatch) {
432 av_log(ctx, AV_LOG_ERROR, "No file path for %s\n",
437 av_log(ctx, AV_LOG_INFO, "Using \"%s\"\n", filename);
439 s->fontsize = size + 0.5;
441 err = load_font_file(ctx, filename, index);
444 FcConfigDestroy(fontconfig);
446 FcPatternDestroy(best);
451 static int load_font(AVFilterContext *ctx)
453 DrawTextContext *s = ctx->priv;
456 /* load the face, and set up the encoding, which is by default UTF-8 */
457 err = load_font_file(ctx, s->fontfile, 0);
460 #if CONFIG_LIBFONTCONFIG
461 err = load_font_fontconfig(ctx);
468 static int load_textfile(AVFilterContext *ctx)
470 DrawTextContext *s = ctx->priv;
476 if ((err = av_file_map(s->textfile, &textbuf, &textbuf_size, 0, ctx)) < 0) {
477 av_log(ctx, AV_LOG_ERROR,
478 "The text file '%s' could not be read or is empty\n",
483 if (textbuf_size > SIZE_MAX - 1 || !(tmp = av_realloc(s->text, textbuf_size + 1))) {
484 av_file_unmap(textbuf, textbuf_size);
485 return AVERROR(ENOMEM);
488 memcpy(s->text, textbuf, textbuf_size);
489 s->text[textbuf_size] = 0;
490 av_file_unmap(textbuf, textbuf_size);
495 static inline int is_newline(uint32_t c)
497 return c == '\n' || c == '\r' || c == '\f' || c == '\v';
500 #if CONFIG_LIBFRIBIDI
501 static int shape_text(AVFilterContext *ctx)
503 DrawTextContext *s = ctx->priv;
505 int ret = AVERROR(ENOMEM);
506 static const FriBidiFlags flags = FRIBIDI_FLAGS_DEFAULT |
507 FRIBIDI_FLAGS_ARABIC;
508 FriBidiChar *unicodestr = NULL;
510 FriBidiParType direction = FRIBIDI_PAR_LTR;
511 FriBidiStrIndex line_start = 0;
512 FriBidiStrIndex line_end = 0;
513 FriBidiLevel *embedding_levels = NULL;
514 FriBidiArabicProp *ar_props = NULL;
515 FriBidiCharType *bidi_types = NULL;
518 len = strlen(s->text);
519 if (!(unicodestr = av_malloc_array(len, sizeof(*unicodestr)))) {
522 len = fribidi_charset_to_unicode(FRIBIDI_CHAR_SET_UTF8,
523 s->text, len, unicodestr);
525 bidi_types = av_malloc_array(len, sizeof(*bidi_types));
530 fribidi_get_bidi_types(unicodestr, len, bidi_types);
532 embedding_levels = av_malloc_array(len, sizeof(*embedding_levels));
533 if (!embedding_levels) {
537 if (!fribidi_get_par_embedding_levels(bidi_types, len, &direction,
542 ar_props = av_malloc_array(len, sizeof(*ar_props));
547 fribidi_get_joining_types(unicodestr, len, ar_props);
548 fribidi_join_arabic(bidi_types, len, embedding_levels, ar_props);
549 fribidi_shape(flags, embedding_levels, len, ar_props, unicodestr);
551 for (line_end = 0, line_start = 0; line_end < len; line_end++) {
552 if (is_newline(unicodestr[line_end]) || line_end == len - 1) {
553 if (!fribidi_reorder_line(flags, bidi_types,
554 line_end - line_start + 1, line_start,
555 direction, embedding_levels, unicodestr,
559 line_start = line_end + 1;
563 /* Remove zero-width fill chars put in by libfribidi */
564 for (i = 0, j = 0; i < len; i++)
565 if (unicodestr[i] != FRIBIDI_CHAR_FILL)
566 unicodestr[j++] = unicodestr[i];
569 if (!(tmp = av_realloc(s->text, (len * 4 + 1) * sizeof(*s->text)))) {
570 /* Use len * 4, as a unicode character can be up to 4 bytes in UTF-8 */
575 len = fribidi_unicode_to_charset(FRIBIDI_CHAR_SET_UTF8,
576 unicodestr, len, s->text);
581 av_free(embedding_levels);
588 static av_cold int init(AVFilterContext *ctx)
591 DrawTextContext *s = ctx->priv;
594 if (!s->fontfile && !CONFIG_LIBFONTCONFIG) {
595 av_log(ctx, AV_LOG_ERROR, "No font filename provided\n");
596 return AVERROR(EINVAL);
601 av_log(ctx, AV_LOG_ERROR,
602 "Both text and text file provided. Please provide only one\n");
603 return AVERROR(EINVAL);
605 if ((err = load_textfile(ctx)) < 0)
609 #if CONFIG_LIBFRIBIDI
611 if ((err = shape_text(ctx)) < 0)
615 if (s->reload && !s->textfile)
616 av_log(ctx, AV_LOG_WARNING, "No file to reload\n");
618 if (s->tc_opt_string) {
619 int ret = av_timecode_init_from_string(&s->tc, s->tc_rate,
620 s->tc_opt_string, ctx);
624 s->tc.flags |= AV_TIMECODE_FLAG_24HOURSMAX;
626 s->text = av_strdup("");
630 av_log(ctx, AV_LOG_ERROR,
631 "Either text, a valid file or a timecode must be provided\n");
632 return AVERROR(EINVAL);
635 if ((err = FT_Init_FreeType(&(s->library)))) {
636 av_log(ctx, AV_LOG_ERROR,
637 "Could not load FreeType: %s\n", FT_ERRMSG(err));
638 return AVERROR(EINVAL);
641 err = load_font(ctx);
646 if ((err = FT_Set_Pixel_Sizes(s->face, 0, s->fontsize))) {
647 av_log(ctx, AV_LOG_ERROR, "Could not set font size to %d pixels: %s\n",
648 s->fontsize, FT_ERRMSG(err));
649 return AVERROR(EINVAL);
653 if (FT_Stroker_New(s->library, &s->stroker)) {
654 av_log(ctx, AV_LOG_ERROR, "Coult not init FT stroker\n");
655 return AVERROR_EXTERNAL;
657 FT_Stroker_Set(s->stroker, s->borderw << 6, FT_STROKER_LINECAP_ROUND,
658 FT_STROKER_LINEJOIN_ROUND, 0);
661 s->use_kerning = FT_HAS_KERNING(s->face);
663 /* load the fallback glyph with code 0 */
664 load_glyph(ctx, NULL, 0);
666 /* set the tabsize in pixels */
667 if ((err = load_glyph(ctx, &glyph, ' ')) < 0) {
668 av_log(ctx, AV_LOG_ERROR, "Could not set tabsize.\n");
671 s->tabsize *= glyph->advance;
673 if (s->exp_mode == EXP_STRFTIME &&
674 (strchr(s->text, '%') || strchr(s->text, '\\')))
675 av_log(ctx, AV_LOG_WARNING, "expansion=strftime is deprecated.\n");
677 av_bprint_init(&s->expanded_text, 0, AV_BPRINT_SIZE_UNLIMITED);
678 av_bprint_init(&s->expanded_fontcolor, 0, AV_BPRINT_SIZE_UNLIMITED);
683 static int query_formats(AVFilterContext *ctx)
685 return ff_set_common_formats(ctx, ff_draw_supported_pixel_formats(0));
688 static int glyph_enu_free(void *opaque, void *elem)
692 FT_Done_Glyph(glyph->glyph);
693 FT_Done_Glyph(glyph->border_glyph);
698 static av_cold void uninit(AVFilterContext *ctx)
700 DrawTextContext *s = ctx->priv;
702 av_expr_free(s->x_pexpr);
703 av_expr_free(s->y_pexpr);
704 s->x_pexpr = s->y_pexpr = NULL;
705 av_freep(&s->positions);
709 av_tree_enumerate(s->glyphs, NULL, NULL, glyph_enu_free);
710 av_tree_destroy(s->glyphs);
713 FT_Done_Face(s->face);
714 FT_Stroker_Done(s->stroker);
715 FT_Done_FreeType(s->library);
717 av_bprint_finalize(&s->expanded_text, NULL);
718 av_bprint_finalize(&s->expanded_fontcolor, NULL);
721 static int config_input(AVFilterLink *inlink)
723 AVFilterContext *ctx = inlink->dst;
724 DrawTextContext *s = ctx->priv;
727 ff_draw_init(&s->dc, inlink->format, 0);
728 ff_draw_color(&s->dc, &s->fontcolor, s->fontcolor.rgba);
729 ff_draw_color(&s->dc, &s->shadowcolor, s->shadowcolor.rgba);
730 ff_draw_color(&s->dc, &s->bordercolor, s->bordercolor.rgba);
731 ff_draw_color(&s->dc, &s->boxcolor, s->boxcolor.rgba);
733 s->var_values[VAR_w] = s->var_values[VAR_W] = s->var_values[VAR_MAIN_W] = inlink->w;
734 s->var_values[VAR_h] = s->var_values[VAR_H] = s->var_values[VAR_MAIN_H] = inlink->h;
735 s->var_values[VAR_SAR] = inlink->sample_aspect_ratio.num ? av_q2d(inlink->sample_aspect_ratio) : 1;
736 s->var_values[VAR_DAR] = (double)inlink->w / inlink->h * s->var_values[VAR_SAR];
737 s->var_values[VAR_HSUB] = 1 << s->dc.hsub_max;
738 s->var_values[VAR_VSUB] = 1 << s->dc.vsub_max;
739 s->var_values[VAR_X] = NAN;
740 s->var_values[VAR_Y] = NAN;
741 s->var_values[VAR_T] = NAN;
743 av_lfg_init(&s->prng, av_get_random_seed());
745 av_expr_free(s->x_pexpr);
746 av_expr_free(s->y_pexpr);
747 s->x_pexpr = s->y_pexpr = NULL;
749 if ((ret = av_expr_parse(&s->x_pexpr, s->x_expr, var_names,
750 NULL, NULL, fun2_names, fun2, 0, ctx)) < 0 ||
751 (ret = av_expr_parse(&s->y_pexpr, s->y_expr, var_names,
752 NULL, NULL, fun2_names, fun2, 0, ctx)) < 0)
754 return AVERROR(EINVAL);
759 static int command(AVFilterContext *ctx, const char *cmd, const char *arg, char *res, int res_len, int flags)
761 DrawTextContext *s = ctx->priv;
763 if (!strcmp(cmd, "reinit")) {
767 if ((ret = av_set_options_string(ctx, arg, "=", ":")) < 0)
769 if ((ret = init(ctx)) < 0)
771 return config_input(ctx->inputs[0]);
774 return AVERROR(ENOSYS);
777 static int func_pict_type(AVFilterContext *ctx, AVBPrint *bp,
778 char *fct, unsigned argc, char **argv, int tag)
780 DrawTextContext *s = ctx->priv;
782 av_bprintf(bp, "%c", av_get_picture_type_char(s->var_values[VAR_PICT_TYPE]));
786 static int func_pts(AVFilterContext *ctx, AVBPrint *bp,
787 char *fct, unsigned argc, char **argv, int tag)
789 DrawTextContext *s = ctx->priv;
791 double pts = s->var_values[VAR_T];
794 fmt = argc >= 1 ? argv[0] : "flt";
797 if ((ret = av_parse_time(&delta, argv[1], 1)) < 0) {
798 av_log(ctx, AV_LOG_ERROR, "Invalid delta '%s'\n", argv[1]);
801 pts += (double)delta / AV_TIME_BASE;
803 if (!strcmp(fmt, "flt")) {
804 av_bprintf(bp, "%.6f", s->var_values[VAR_T]);
805 } else if (!strcmp(fmt, "hms")) {
807 av_bprintf(bp, " ??:??:??.???");
809 int64_t ms = round(pts * 1000);
815 av_bprintf(bp, "%c%02d:%02d:%02d.%03d", sign,
816 (int)(ms / (60 * 60 * 1000)),
817 (int)(ms / (60 * 1000)) % 60,
818 (int)(ms / 1000) % 60,
822 av_log(ctx, AV_LOG_ERROR, "Invalid format '%s'\n", fmt);
823 return AVERROR(EINVAL);
828 static int func_frame_num(AVFilterContext *ctx, AVBPrint *bp,
829 char *fct, unsigned argc, char **argv, int tag)
831 DrawTextContext *s = ctx->priv;
833 av_bprintf(bp, "%d", (int)s->var_values[VAR_N]);
837 static int func_metadata(AVFilterContext *ctx, AVBPrint *bp,
838 char *fct, unsigned argc, char **argv, int tag)
840 DrawTextContext *s = ctx->priv;
841 AVDictionaryEntry *e = av_dict_get(s->metadata, argv[0], NULL, 0);
844 av_bprintf(bp, "%s", e->value);
848 static int func_strftime(AVFilterContext *ctx, AVBPrint *bp,
849 char *fct, unsigned argc, char **argv, int tag)
851 const char *fmt = argc ? argv[0] : "%Y-%m-%d %H:%M:%S";
857 localtime_r(&now, &tm);
859 tm = *gmtime_r(&now, &tm);
860 av_bprint_strftime(bp, fmt, &tm);
864 static int func_eval_expr(AVFilterContext *ctx, AVBPrint *bp,
865 char *fct, unsigned argc, char **argv, int tag)
867 DrawTextContext *s = ctx->priv;
871 ret = av_expr_parse_and_eval(&res, argv[0], var_names, s->var_values,
872 NULL, NULL, fun2_names, fun2,
875 av_log(ctx, AV_LOG_ERROR,
876 "Expression '%s' for the expr text expansion function is not valid\n",
879 av_bprintf(bp, "%f", res);
884 static int func_eval_expr_int_format(AVFilterContext *ctx, AVBPrint *bp,
885 char *fct, unsigned argc, char **argv, int tag)
887 DrawTextContext *s = ctx->priv;
891 unsigned int positions = 0;
892 char fmt_str[30] = "%";
895 * argv[0] expression to be converted to `int`
896 * argv[1] format: 'x', 'X', 'd' or 'u'
897 * argv[2] positions printed (optional)
900 ret = av_expr_parse_and_eval(&res, argv[0], var_names, s->var_values,
901 NULL, NULL, fun2_names, fun2,
904 av_log(ctx, AV_LOG_ERROR,
905 "Expression '%s' for the expr text expansion function is not valid\n",
910 if (!strchr("xXdu", argv[1][0])) {
911 av_log(ctx, AV_LOG_ERROR, "Invalid format '%c' specified,"
912 " allowed values: 'x', 'X', 'd', 'u'\n", argv[1][0]);
913 return AVERROR(EINVAL);
917 ret = sscanf(argv[2], "%u", &positions);
919 av_log(ctx, AV_LOG_ERROR, "expr_int_format(): Invalid number of positions"
920 " to print: '%s'\n", argv[2]);
921 return AVERROR(EINVAL);
925 feclearexcept(FE_ALL_EXCEPT);
927 if ((ret = fetestexcept(FE_INVALID|FE_OVERFLOW|FE_UNDERFLOW))) {
928 av_log(ctx, AV_LOG_ERROR, "Conversion of floating-point result to int failed. Control register: 0x%08x. Conversion result: %d\n", ret, intval);
929 return AVERROR(EINVAL);
933 av_strlcatf(fmt_str, sizeof(fmt_str), "0%u", positions);
934 av_strlcatf(fmt_str, sizeof(fmt_str), "%c", argv[1][0]);
936 av_log(ctx, AV_LOG_DEBUG, "Formatting value %f (expr '%s') with spec '%s'\n",
937 res, argv[0], fmt_str);
939 av_bprintf(bp, fmt_str, intval);
944 static const struct drawtext_function {
946 unsigned argc_min, argc_max;
947 int tag; /**< opaque argument to func */
948 int (*func)(AVFilterContext *, AVBPrint *, char *, unsigned, char **, int);
950 { "expr", 1, 1, 0, func_eval_expr },
951 { "e", 1, 1, 0, func_eval_expr },
952 { "expr_int_format", 2, 3, 0, func_eval_expr_int_format },
953 { "eif", 2, 3, 0, func_eval_expr_int_format },
954 { "pict_type", 0, 0, 0, func_pict_type },
955 { "pts", 0, 2, 0, func_pts },
956 { "gmtime", 0, 1, 'G', func_strftime },
957 { "localtime", 0, 1, 'L', func_strftime },
958 { "frame_num", 0, 0, 0, func_frame_num },
959 { "n", 0, 0, 0, func_frame_num },
960 { "metadata", 1, 1, 0, func_metadata },
963 static int eval_function(AVFilterContext *ctx, AVBPrint *bp, char *fct,
964 unsigned argc, char **argv)
968 for (i = 0; i < FF_ARRAY_ELEMS(functions); i++) {
969 if (strcmp(fct, functions[i].name))
971 if (argc < functions[i].argc_min) {
972 av_log(ctx, AV_LOG_ERROR, "%%{%s} requires at least %d arguments\n",
973 fct, functions[i].argc_min);
974 return AVERROR(EINVAL);
976 if (argc > functions[i].argc_max) {
977 av_log(ctx, AV_LOG_ERROR, "%%{%s} requires at most %d arguments\n",
978 fct, functions[i].argc_max);
979 return AVERROR(EINVAL);
983 if (i >= FF_ARRAY_ELEMS(functions)) {
984 av_log(ctx, AV_LOG_ERROR, "%%{%s} is not known\n", fct);
985 return AVERROR(EINVAL);
987 return functions[i].func(ctx, bp, fct, argc, argv, functions[i].tag);
990 static int expand_function(AVFilterContext *ctx, AVBPrint *bp, char **rtext)
992 const char *text = *rtext;
993 char *argv[16] = { NULL };
994 unsigned argc = 0, i;
998 av_log(ctx, AV_LOG_ERROR, "Stray %% near '%s'\n", text);
999 return AVERROR(EINVAL);
1003 if (!(argv[argc++] = av_get_token(&text, ":}"))) {
1004 ret = AVERROR(ENOMEM);
1008 av_log(ctx, AV_LOG_ERROR, "Unterminated %%{} near '%s'\n", *rtext);
1009 ret = AVERROR(EINVAL);
1012 if (argc == FF_ARRAY_ELEMS(argv))
1013 av_freep(&argv[--argc]); /* error will be caught later */
1019 if ((ret = eval_function(ctx, bp, argv[0], argc - 1, argv + 1)) < 0)
1022 *rtext = (char *)text + 1;
1025 for (i = 0; i < argc; i++)
1030 static int expand_text(AVFilterContext *ctx, char *text, AVBPrint *bp)
1034 av_bprint_clear(bp);
1036 if (*text == '\\' && text[1]) {
1037 av_bprint_chars(bp, text[1], 1);
1039 } else if (*text == '%') {
1041 if ((ret = expand_function(ctx, bp, &text)) < 0)
1044 av_bprint_chars(bp, *text, 1);
1048 if (!av_bprint_is_complete(bp))
1049 return AVERROR(ENOMEM);
1053 static int draw_glyphs(DrawTextContext *s, AVFrame *frame,
1054 int width, int height,
1055 FFDrawColor *color, int x, int y, int borderw)
1057 char *text = s->expanded_text.str;
1061 Glyph *glyph = NULL;
1063 for (i = 0, p = text; *p; i++) {
1065 Glyph dummy = { 0 };
1066 GET_UTF8(code, *p++, continue;);
1068 /* skip new line chars, just go to new line */
1069 if (code == '\n' || code == '\r' || code == '\t')
1073 glyph = av_tree_find(s->glyphs, &dummy, (void *)glyph_cmp, NULL);
1075 bitmap = borderw ? glyph->border_bitmap : glyph->bitmap;
1077 if (glyph->bitmap.pixel_mode != FT_PIXEL_MODE_MONO &&
1078 glyph->bitmap.pixel_mode != FT_PIXEL_MODE_GRAY)
1079 return AVERROR(EINVAL);
1081 x1 = s->positions[i].x+s->x+x - borderw;
1082 y1 = s->positions[i].y+s->y+y - borderw;
1084 ff_blend_mask(&s->dc, color,
1085 frame->data, frame->linesize, width, height,
1086 bitmap.buffer, bitmap.pitch,
1087 bitmap.width, bitmap.rows,
1088 bitmap.pixel_mode == FT_PIXEL_MODE_MONO ? 0 : 3,
1095 static int draw_text(AVFilterContext *ctx, AVFrame *frame,
1096 int width, int height)
1098 DrawTextContext *s = ctx->priv;
1099 AVFilterLink *inlink = ctx->inputs[0];
1101 uint32_t code = 0, prev_code = 0;
1102 int x = 0, y = 0, i = 0, ret;
1103 int max_text_line_w = 0, len;
1107 int y_min = 32000, y_max = -32000;
1108 int x_min = 32000, x_max = -32000;
1110 Glyph *glyph = NULL, *prev_glyph = NULL;
1111 Glyph dummy = { 0 };
1113 time_t now = time(0);
1115 AVBPrint *bp = &s->expanded_text;
1117 av_bprint_clear(bp);
1119 if(s->basetime != AV_NOPTS_VALUE)
1120 now= frame->pts*av_q2d(ctx->inputs[0]->time_base) + s->basetime/1000000;
1122 switch (s->exp_mode) {
1124 av_bprintf(bp, "%s", s->text);
1127 if ((ret = expand_text(ctx, s->text, &s->expanded_text)) < 0)
1131 localtime_r(&now, <ime);
1132 av_bprint_strftime(bp, s->text, <ime);
1136 if (s->tc_opt_string) {
1137 char tcbuf[AV_TIMECODE_STR_SIZE];
1138 av_timecode_make_string(&s->tc, tcbuf, inlink->frame_count);
1139 av_bprint_clear(bp);
1140 av_bprintf(bp, "%s%s", s->text, tcbuf);
1143 if (!av_bprint_is_complete(bp))
1144 return AVERROR(ENOMEM);
1145 text = s->expanded_text.str;
1146 if ((len = s->expanded_text.len) > s->nb_positions) {
1147 if (!(s->positions =
1148 av_realloc(s->positions, len*sizeof(*s->positions))))
1149 return AVERROR(ENOMEM);
1150 s->nb_positions = len;
1153 if (s->fontcolor_expr[0]) {
1154 /* If expression is set, evaluate and replace the static value */
1155 av_bprint_clear(&s->expanded_fontcolor);
1156 if ((ret = expand_text(ctx, s->fontcolor_expr, &s->expanded_fontcolor)) < 0)
1158 if (!av_bprint_is_complete(&s->expanded_fontcolor))
1159 return AVERROR(ENOMEM);
1160 av_log(s, AV_LOG_DEBUG, "Evaluated fontcolor is '%s'\n", s->expanded_fontcolor.str);
1161 ret = av_parse_color(s->fontcolor.rgba, s->expanded_fontcolor.str, -1, s);
1164 ff_draw_color(&s->dc, &s->fontcolor, s->fontcolor.rgba);
1170 /* load and cache glyphs */
1171 for (i = 0, p = text; *p; i++) {
1172 GET_UTF8(code, *p++, continue;);
1176 glyph = av_tree_find(s->glyphs, &dummy, glyph_cmp, NULL);
1178 load_glyph(ctx, &glyph, code);
1181 y_min = FFMIN(glyph->bbox.yMin, y_min);
1182 y_max = FFMAX(glyph->bbox.yMax, y_max);
1183 x_min = FFMIN(glyph->bbox.xMin, x_min);
1184 x_max = FFMAX(glyph->bbox.xMax, x_max);
1186 s->max_glyph_h = y_max - y_min;
1187 s->max_glyph_w = x_max - x_min;
1189 /* compute and save position for each glyph */
1191 for (i = 0, p = text; *p; i++) {
1192 GET_UTF8(code, *p++, continue;);
1194 /* skip the \n in the sequence \r\n */
1195 if (prev_code == '\r' && code == '\n')
1199 if (is_newline(code)) {
1201 max_text_line_w = FFMAX(max_text_line_w, x);
1202 y += s->max_glyph_h;
1210 glyph = av_tree_find(s->glyphs, &dummy, glyph_cmp, NULL);
1213 if (s->use_kerning && prev_glyph && glyph->code) {
1214 FT_Get_Kerning(s->face, prev_glyph->code, glyph->code,
1215 ft_kerning_default, &delta);
1220 s->positions[i].x = x + glyph->bitmap_left;
1221 s->positions[i].y = y - glyph->bitmap_top + y_max;
1222 if (code == '\t') x = (x / s->tabsize + 1)*s->tabsize;
1223 else x += glyph->advance;
1226 max_text_line_w = FFMAX(x, max_text_line_w);
1228 s->var_values[VAR_TW] = s->var_values[VAR_TEXT_W] = max_text_line_w;
1229 s->var_values[VAR_TH] = s->var_values[VAR_TEXT_H] = y + s->max_glyph_h;
1231 s->var_values[VAR_MAX_GLYPH_W] = s->max_glyph_w;
1232 s->var_values[VAR_MAX_GLYPH_H] = s->max_glyph_h;
1233 s->var_values[VAR_MAX_GLYPH_A] = s->var_values[VAR_ASCENT ] = y_max;
1234 s->var_values[VAR_MAX_GLYPH_D] = s->var_values[VAR_DESCENT] = y_min;
1236 s->var_values[VAR_LINE_H] = s->var_values[VAR_LH] = s->max_glyph_h;
1238 s->x = s->var_values[VAR_X] = av_expr_eval(s->x_pexpr, s->var_values, &s->prng);
1239 s->y = s->var_values[VAR_Y] = av_expr_eval(s->y_pexpr, s->var_values, &s->prng);
1240 s->x = s->var_values[VAR_X] = av_expr_eval(s->x_pexpr, s->var_values, &s->prng);
1242 box_w = FFMIN(width - 1 , max_text_line_w);
1243 box_h = FFMIN(height - 1, y + s->max_glyph_h);
1247 ff_blend_rectangle(&s->dc, &s->boxcolor,
1248 frame->data, frame->linesize, width, height,
1249 s->x - s->boxborderw, s->y - s->boxborderw,
1250 box_w + s->boxborderw * 2, box_h + s->boxborderw * 2);
1252 if (s->shadowx || s->shadowy) {
1253 if ((ret = draw_glyphs(s, frame, width, height,
1254 &s->shadowcolor, s->shadowx, s->shadowy, 0)) < 0)
1259 if ((ret = draw_glyphs(s, frame, width, height,
1260 &s->bordercolor, 0, 0, s->borderw)) < 0)
1263 if ((ret = draw_glyphs(s, frame, width, height,
1264 &s->fontcolor, 0, 0, 0)) < 0)
1270 static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
1272 AVFilterContext *ctx = inlink->dst;
1273 AVFilterLink *outlink = ctx->outputs[0];
1274 DrawTextContext *s = ctx->priv;
1278 if ((ret = load_textfile(ctx)) < 0)
1280 #if CONFIG_LIBFRIBIDI
1281 if (s->text_shaping)
1282 if ((ret = shape_text(ctx)) < 0)
1287 s->var_values[VAR_N] = inlink->frame_count+s->start_number;
1288 s->var_values[VAR_T] = frame->pts == AV_NOPTS_VALUE ?
1289 NAN : frame->pts * av_q2d(inlink->time_base);
1291 s->var_values[VAR_PICT_TYPE] = frame->pict_type;
1292 s->metadata = av_frame_get_metadata(frame);
1294 draw_text(ctx, frame, frame->width, frame->height);
1296 av_log(ctx, AV_LOG_DEBUG, "n:%d t:%f text_w:%d text_h:%d x:%d y:%d\n",
1297 (int)s->var_values[VAR_N], s->var_values[VAR_T],
1298 (int)s->var_values[VAR_TEXT_W], (int)s->var_values[VAR_TEXT_H],
1301 return ff_filter_frame(outlink, frame);
1304 static const AVFilterPad avfilter_vf_drawtext_inputs[] = {
1307 .type = AVMEDIA_TYPE_VIDEO,
1308 .filter_frame = filter_frame,
1309 .config_props = config_input,
1310 .needs_writable = 1,
1315 static const AVFilterPad avfilter_vf_drawtext_outputs[] = {
1318 .type = AVMEDIA_TYPE_VIDEO,
1323 AVFilter ff_vf_drawtext = {
1325 .description = NULL_IF_CONFIG_SMALL("Draw text on top of video frames using libfreetype library."),
1326 .priv_size = sizeof(DrawTextContext),
1327 .priv_class = &drawtext_class,
1330 .query_formats = query_formats,
1331 .inputs = avfilter_vf_drawtext_inputs,
1332 .outputs = avfilter_vf_drawtext_outputs,
1333 .process_command = command,
1334 .flags = AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC,