History

avformat/dhav: fix backward scanning for get_duration and optimize seeking

The backwards scanning done for incomplete final packets should not assume a specific alignment at the end of the file. Truncated files result in hundreds of thousands of seeks if the final packet does not fall on a specific byte boundary, which can be extremely slow. For example, with HTTP, each backwards seek results in a separate HTTP request. This changes the scanning to check for the end tag 1 byte at a time and buffers the last 1 MiB to avoid additional seek operations.

Justin Ruggles committed 1 year ago

Commit 36ec9211 parent c8b09fb1 file · +39 −15Browse files

Unified · Split

libavformat/dhav.c+39−15
@@ -22,6 +22,7 @@ · 21 unmodified lines
2222
2323 #include <time.h>
2424
25+#include "libavutil/intreadwrite.h"
2526 #include "libavutil/mem.h"
2627 #include "libavutil/parseutils.h"
2728 #include "avio_internal.h"
@@ -232,37 +233,60 @@ static void get_timeinfo(unsigned date, struct tm *timeinfo) · 204 unmodified lines
232233 timeinfo->tm_sec = sec;
233234 }
234235
236+#define MAX_DURATION_BUFFER_SIZE (1024*1024)
237+
235238 static int64_t get_duration(AVFormatContext *s)
236239 {
237− DHAVContext *dhav = s->priv_data;
238240 int64_t start_pos = avio_tell(s->pb);
241+ int64_t end_pos = -1;
239242 int64_t start = 0, end = 0;
240243 struct tm timeinfo;
241− int max_interations = 100000;
244+ uint8_t *end_buffer;
245+ int64_t end_buffer_size;
246+ int64_t end_buffer_pos;
247+ int64_t offset;
248+ unsigned date;
242249
243250 if (!s->pb->seekable)
244251 return 0;
245252
246− avio_seek(s->pb, avio_size(s->pb) - 8, SEEK_SET);
247− while (avio_tell(s->pb) > 12 && max_interations--) {
248− if (avio_rl32(s->pb) == MKTAG('d','h','a','v')) {
249− int64_t seek_back = avio_rl32(s->pb);
253+ if (start_pos + 16 > avio_size(s->pb))
254+ return 0;
250255
251− avio_seek(s->pb, -seek_back, SEEK_CUR);
252− read_chunk(s);
253− get_timeinfo(dhav->date, &timeinfo);
254− end = av_timegm(&timeinfo) * 1000LL;
256+ avio_skip(s->pb, 16);
257+ date = avio_rl32(s->pb);
258+ get_timeinfo(date, &timeinfo);
259+ start = av_timegm(&timeinfo) * 1000LL;
260+
261+ end_buffer_size = FFMIN(MAX_DURATION_BUFFER_SIZE, avio_size(s->pb));
262+ end_buffer = av_malloc(end_buffer_size);
263+ if (!end_buffer) {
264+ avio_seek(s->pb, start_pos, SEEK_SET);
265+ return 0;
266+ }
267+ end_buffer_pos = avio_size(s->pb) - end_buffer_size;
268+ avio_seek(s->pb, end_buffer_pos, SEEK_SET);
269+ avio_read(s->pb, end_buffer, end_buffer_size);
270+
271+ offset = end_buffer_size - 8;
272+ while (offset > 0) {
273+ if (AV_RL32(end_buffer + offset) == MKTAG('d','h','a','v')) {
274+ int64_t seek_back = AV_RL32(end_buffer + offset + 4);
275+ end_pos = end_buffer_pos + offset - seek_back + 8;
255276 break;
256277 } else {
257− avio_seek(s->pb, -12, SEEK_CUR);
278+ offset -= 9;
258279 }
259280 }
260281
261− avio_seek(s->pb, start_pos, SEEK_SET);
282+ if (end_pos < 0 || end_pos + 16 > end_buffer_pos + end_buffer_size) {
283+ avio_seek(s->pb, start_pos, SEEK_SET);
284+ return 0;
285+ }
262286
263− read_chunk(s);
264− get_timeinfo(dhav->date, &timeinfo);
265− start = av_timegm(&timeinfo) * 1000LL;
287+ date = AV_RL32(end_buffer + (end_pos - end_buffer_pos) + 16);
288+ get_timeinfo(date, &timeinfo);
289+ end = av_timegm(&timeinfo) * 1000LL;
266290
267291 avio_seek(s->pb, start_pos, SEEK_SET);
268292

Keyboard shortcuts

General

Show keyboard shortcuts
?
Focus the current search
/

Issues and pull requests

Open filters
f

Issues, pull requests and Inbox

Select next item
j↓
Select previous item
k↑
Open selected item
Enter
Dismiss filters or clear selection
Esc

Repository code

Go to file
t