1use std::{
2 cmp::{max, min},
3 fmt::{Debug, Display},
4 fs::File,
5 io::{self, Read, Seek},
6 marker::PhantomData,
7 net::TcpStream,
8 ops::Range,
9 path::Path,
10};
11
12use crate::{
13 buffer::Buffers,
14 error::{ParsedError, ParsingError, ParsingErrorState},
15 exif::{parse_exif_iter, TiffHeader},
16 file::Mime,
17 partial_vec::PartialVec,
18 skip::Skip,
19 video::parse_track_info,
20 ExifIter, Seekable, TrackInfo, Unseekable,
21};
22
23pub struct MediaSource<R, S = Seekable> {
44 pub(crate) reader: R,
45 pub(crate) buf: Vec<u8>,
46 pub(crate) mime: Mime,
47 phantom: PhantomData<S>,
48}
49
50impl<R, S: Skip<R>> Debug for MediaSource<R, S> {
51 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
52 f.debug_struct("MediaSource")
53 .field("mime", &self.mime)
55 .field("seekable", &S::debug())
56 .finish_non_exhaustive()
57 }
58}
59
60const HEADER_PARSE_BUF_SIZE: usize = 128;
62
63impl<R: Read, S: Skip<R>> MediaSource<R, S> {
64 #[tracing::instrument(skip(reader))]
65 fn build(mut reader: R) -> crate::Result<Self> {
66 let mut buf = Vec::with_capacity(HEADER_PARSE_BUF_SIZE);
68 reader
69 .by_ref()
70 .take(HEADER_PARSE_BUF_SIZE as u64)
71 .read_to_end(&mut buf)?;
72 let mime: Mime = buf.as_slice().try_into()?;
73 tracing::debug!(?mime);
74 Ok(Self {
75 reader,
76 buf,
77 mime,
78 phantom: PhantomData,
79 })
80 }
81
82 pub fn has_track(&self) -> bool {
83 match self.mime {
84 Mime::Image(_) => false,
85 Mime::Video(_) => true,
86 }
87 }
88
89 pub fn has_exif(&self) -> bool {
90 match self.mime {
91 Mime::Image(_) => true,
92 Mime::Video(_) => false,
93 }
94 }
95}
96
97impl<R: Read + Seek> MediaSource<R, Seekable> {
98 pub fn seekable(reader: R) -> crate::Result<Self> {
99 Self::build(reader)
100 }
101}
102
103impl<R: Read> MediaSource<R, Unseekable> {
104 pub fn unseekable(reader: R) -> crate::Result<Self> {
105 Self::build(reader)
106 }
107}
108
109impl MediaSource<File, Seekable> {
110 pub fn file_path<P: AsRef<Path>>(path: P) -> crate::Result<Self> {
111 Self::seekable(File::open(path)?)
112 }
113
114 pub fn file(file: File) -> crate::Result<Self> {
115 Self::seekable(file)
116 }
117}
118
119impl MediaSource<TcpStream, Unseekable> {
120 pub fn tcp_stream(stream: TcpStream) -> crate::Result<Self> {
121 Self::unseekable(stream)
122 }
123}
124
125pub(crate) const INIT_BUF_SIZE: usize = 4096;
127pub(crate) const MIN_GROW_SIZE: usize = 4096;
128pub(crate) const MAX_ALLOC_SIZE: usize = 1024 * 1024 * 1024;
132
133pub(crate) trait Buf {
134 fn buffer(&self) -> &[u8];
135 fn clear(&mut self);
136
137 fn set_position(&mut self, pos: usize);
138 #[allow(unused)]
139 fn position(&self) -> usize;
140}
141
142#[derive(Debug, Clone)]
143pub(crate) enum ParsingState {
144 TiffHeader(TiffHeader),
145 HeifExifSize(usize),
146}
147
148impl Display for ParsingState {
149 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
150 match self {
151 ParsingState::TiffHeader(h) => Display::fmt(&format!("ParsingState: {h:?})"), f),
152 ParsingState::HeifExifSize(n) => Display::fmt(&format!("ParsingState: {n}"), f),
153 }
154 }
155}
156
157pub(crate) trait BufParser: Buf + Debug {
158 fn fill_buf<R: Read>(&mut self, reader: &mut R, size: usize) -> io::Result<usize>;
159 fn load_and_parse<R: Read, S: Skip<R>, P, O>(
160 &mut self,
161 reader: &mut R,
162 mut parse: P,
163 ) -> Result<O, ParsedError>
164 where
165 P: FnMut(&[u8], Option<ParsingState>) -> Result<O, ParsingErrorState>,
166 {
167 self.load_and_parse_with_offset::<R, S, _, _>(
168 reader,
169 |data, _, state| parse(data, state),
170 0,
171 )
172 }
173
174 #[tracing::instrument(skip_all)]
175 fn load_and_parse_with_offset<R: Read, S: Skip<R>, P, O>(
176 &mut self,
177 reader: &mut R,
178 mut parse: P,
179 offset: usize,
180 ) -> Result<O, ParsedError>
181 where
182 P: FnMut(&[u8], usize, Option<ParsingState>) -> Result<O, ParsingErrorState>,
183 {
184 if offset >= self.buffer().len() {
185 self.fill_buf(reader, MIN_GROW_SIZE)?;
186 }
187
188 let mut parsing_state: Option<ParsingState> = None;
189 loop {
190 let res = parse(self.buffer(), offset, parsing_state.take());
191 match res {
192 Ok(o) => return Ok(o),
193 Err(es) => {
194 tracing::debug!(?es);
195 parsing_state = es.state;
196
197 match es.err {
198 ParsingError::ClearAndSkip(n) => {
199 self.clear_and_skip::<R, S>(reader, n)?;
200 }
201 ParsingError::Need(i) => {
202 tracing::debug!(need = i, "need more bytes");
203 let to_read = max(i, MIN_GROW_SIZE);
204 let n = self.fill_buf(reader, to_read)?;
207 if n == 0 {
208 return Err(ParsedError::NoEnoughBytes);
209 }
210 tracing::debug!(n, "actual read");
211 }
212 ParsingError::Failed(s) => return Err(ParsedError::Failed(s)),
213 }
214 }
215 }
216 }
217 }
218
219 #[tracing::instrument(skip(reader))]
220 fn clear_and_skip<R: Read, S: Skip<R>>(
221 &mut self,
222 reader: &mut R,
223 n: usize,
224 ) -> Result<(), ParsedError> {
225 tracing::debug!("ClearAndSkip");
226 if n <= self.buffer().len() {
227 tracing::debug!(n, "skip by set_position");
228 self.set_position(n);
229 return Ok(());
230 }
231
232 let skip_n = n - self.buffer().len();
233 tracing::debug!(skip_n, "clear and skip bytes");
234 self.clear();
235
236 let done = S::skip_by_seek(
237 reader,
238 skip_n
239 .try_into()
240 .map_err(|_| ParsedError::Failed("skip too many bytes".into()))?,
241 )?;
242 if !done {
243 tracing::debug!(skip_n, "skip by using our buffer");
244 let mut skipped = 0;
245 while skipped < skip_n {
246 let mut to_skip = skip_n - skipped;
247 to_skip = min(to_skip, MAX_ALLOC_SIZE);
248 let n = self.fill_buf(reader, to_skip)?;
249 skipped += n;
250 if skipped <= skip_n {
251 self.clear();
252 } else {
253 let remain = skipped - skip_n;
254 self.set_position(self.buffer().len() - remain);
255 break;
256 }
257 }
258 } else {
259 tracing::debug!(skip_n, "skip with seek");
260 }
261
262 if self.buffer().is_empty() {
263 self.fill_buf(reader, MIN_GROW_SIZE)?;
264 }
265 Ok(())
266 }
267}
268
269impl BufParser for MediaParser {
270 #[tracing::instrument(skip(self, reader), fields(buf_len=self.buf().len()))]
271 fn fill_buf<R: Read>(&mut self, reader: &mut R, size: usize) -> io::Result<usize> {
272 if size.saturating_add(self.buf().len()) > MAX_ALLOC_SIZE {
273 tracing::error!(?size, "the requested buffer size is too big");
274 return Err(io::ErrorKind::Unsupported.into());
275 }
276 self.buf_mut().reserve_exact(size);
277
278 let n = reader.take(size as u64).read_to_end(self.buf_mut())?;
279 if n == 0 {
280 tracing::error!(buf_len = self.buf().len(), "fill_buf: EOF");
281 return Err(std::io::ErrorKind::UnexpectedEof.into());
282 }
283
284 tracing::debug!(
285 ?size,
286 ?n,
287 buf_len = self.buf().len(),
288 "fill_buf: read bytes"
289 );
290
291 Ok(n)
292 }
293}
294
295impl Buf for MediaParser {
296 fn buffer(&self) -> &[u8] {
297 &self.buf()[self.position..]
298 }
299
300 fn clear(&mut self) {
301 self.buf_mut().clear();
302 }
303
304 fn set_position(&mut self, pos: usize) {
305 self.position = pos;
306 }
307
308 fn position(&self) -> usize {
309 self.position
310 }
311}
312
313pub trait ParseOutput<R, S>: Sized {
314 fn parse(parser: &mut MediaParser, ms: MediaSource<R, S>) -> crate::Result<Self>;
315}
316
317impl<R: Read, S: Skip<R>> ParseOutput<R, S> for ExifIter {
318 fn parse(parser: &mut MediaParser, mut ms: MediaSource<R, S>) -> crate::Result<Self> {
319 if !ms.has_exif() {
320 return Err(crate::Error::ParseFailed("no Exif data here".into()));
321 }
322 parse_exif_iter::<R, S>(parser, ms.mime.unwrap_image(), &mut ms.reader)
323 }
324}
325
326impl<R: Read, S: Skip<R>> ParseOutput<R, S> for TrackInfo {
327 fn parse(parser: &mut MediaParser, mut ms: MediaSource<R, S>) -> crate::Result<Self> {
328 if !ms.has_track() {
329 return Err(crate::Error::ParseFailed("no track info here".into()));
330 }
331 let out = parser.load_and_parse::<R, S, _, _>(ms.reader.by_ref(), |data, _| {
332 parse_track_info(data, ms.mime.unwrap_video())
333 .map_err(|e| ParsingErrorState::new(e, None))
334 })?;
335 Ok(out)
336 }
337}
338
339pub struct MediaParser {
390 bb: Buffers,
391 buf: Option<Vec<u8>>,
392 position: usize,
393}
394
395impl Debug for MediaParser {
396 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
397 f.debug_struct("MediaParser")
398 .field("buffers", &self.bb)
399 .field("buf len", &self.buf.as_ref().map(|x| x.len()))
400 .field("position", &self.position)
401 .finish_non_exhaustive()
402 }
403}
404
405impl Default for MediaParser {
406 fn default() -> Self {
407 Self {
408 bb: Buffers::new(),
409 buf: None,
410 position: 0,
411 }
412 }
413}
414
415pub(crate) trait ShareBuf {
416 fn share_buf(&mut self, range: Range<usize>) -> PartialVec;
417}
418
419impl ShareBuf for MediaParser {
420 fn share_buf(&mut self, mut range: Range<usize>) -> PartialVec {
421 let buf = self.buf.take().unwrap();
422 let vec = self.bb.release_to_share(buf);
423 range.start += self.position;
424 range.end += self.position;
425 PartialVec::new(vec, range)
426 }
427}
428
429impl MediaParser {
430 pub fn new() -> Self {
431 Self::default()
432 }
433
434 pub fn parse<R: Read, S, O: ParseOutput<R, S>>(
454 &mut self,
455 mut ms: MediaSource<R, S>,
456 ) -> crate::Result<O> {
457 self.reset();
458 self.acquire_buf();
459
460 self.buf_mut().append(&mut ms.buf);
461 let res = self.do_parse(ms);
462
463 self.reset();
464 res
465 }
466
467 fn do_parse<R: Read, S, O: ParseOutput<R, S>>(
468 &mut self,
469 mut ms: MediaSource<R, S>,
470 ) -> Result<O, crate::Error> {
471 self.fill_buf(&mut ms.reader, INIT_BUF_SIZE)?;
472 let res = ParseOutput::parse(self, ms)?;
473 Ok(res)
474 }
475
476 fn reset(&mut self) {
477 if let Some(buf) = self.buf.take() {
479 self.bb.release(buf);
480 }
481
482 self.set_position(0);
484 }
485
486 pub(crate) fn buf(&self) -> &Vec<u8> {
487 match self.buf.as_ref() {
488 Some(b) => b,
489 None => panic!("no buf here"),
490 }
491 }
492
493 fn buf_mut(&mut self) -> &mut Vec<u8> {
494 match self.buf.as_mut() {
495 Some(b) => b,
496 None => panic!("no buf here"),
497 }
498 }
499
500 fn acquire_buf(&mut self) {
501 assert!(self.buf.is_none());
502 self.buf = Some(self.bb.acquire());
503 }
504}
505
506#[cfg(test)]
507mod tests {
508 use std::sync::{LazyLock, Mutex, MutexGuard};
509
510 use super::*;
511 use test_case::case;
512
513 enum TrackExif {
514 Track,
515 Exif,
516 NoData,
517 Invalid,
518 }
519 use TrackExif::*;
520
521 static PARSER: LazyLock<Mutex<MediaParser>> = LazyLock::new(|| Mutex::new(MediaParser::new()));
522 fn parser() -> MutexGuard<'static, MediaParser> {
523 PARSER.lock().unwrap()
524 }
525
526 #[case("3gp_640x360.3gp", Track)]
527 #[case("broken.jpg", Exif)]
528 #[case("compatible-brands-fail.heic", Invalid)]
529 #[case("compatible-brands-fail.mov", Invalid)]
530 #[case("compatible-brands.heic", NoData)]
531 #[case("compatible-brands.mov", NoData)]
532 #[case("embedded-in-heic.mov", Track)]
533 #[case("exif.heic", Exif)]
534 #[case("exif.jpg", Exif)]
535 #[case("exif-no-tz.jpg", Exif)]
536 #[case("fujifilm_x_t1_01.raf.meta", Exif)]
537 #[case("meta.mov", Track)]
538 #[case("meta.mp4", Track)]
539 #[case("mka.mka", Track)]
540 #[case("mkv_640x360.mkv", Track)]
541 #[case("exif-one-entry.heic", Exif)]
542 #[case("no-exif.jpg", NoData)]
543 #[case("tif.tif", Exif)]
544 #[case("ramdisk.img", Invalid)]
545 #[case("webm_480.webm", Track)]
546 fn parse_media(path: &str, te: TrackExif) {
547 let mut parser = parser();
548 let ms = MediaSource::file_path(Path::new("testdata").join(path));
549 match te {
550 Track => {
551 let ms = ms.unwrap();
552 assert!(ms.has_track());
554 let _: TrackInfo = parser.parse(ms).unwrap();
555 }
556 Exif => {
557 let ms = ms.unwrap();
558 assert!(ms.has_exif());
560 let mut it: ExifIter = parser.parse(ms).unwrap();
561 let _ = it.parse_gps_info();
562
563 if path.contains("one-entry") {
564 assert!(it.next().is_some());
565 assert!(it.next().is_none());
566
567 let exif: crate::Exif = it.clone_and_rewind().into();
568 assert!(exif.get(ExifTag::Orientation).is_some());
569 } else {
570 let _: crate::Exif = it.clone_and_rewind().into();
571 }
572 }
573 NoData => {
574 let ms = ms.unwrap();
575 if ms.has_exif() {
577 let res: Result<ExifIter, _> = parser.parse(ms);
578 res.unwrap_err();
579 } else if ms.has_track() {
580 let res: Result<TrackInfo, _> = parser.parse(ms);
581 res.unwrap_err();
582 }
583 }
584 Invalid => {
585 ms.unwrap_err();
586 }
587 }
588 }
589
590 use crate::testkit::open_sample;
591 use crate::{EntryValue, ExifTag, TrackInfoTag};
592 use chrono::DateTime;
593 use test_case::test_case;
594
595 use crate::video::TrackInfoTag::*;
596
597 #[test_case("mkv_640x360.mkv", ImageWidth, 640_u32.into())]
598 #[test_case("mkv_640x360.mkv", ImageHeight, 360_u32.into())]
599 #[test_case("mkv_640x360.mkv", DurationMs, 13346_u64.into())]
600 #[test_case("mkv_640x360.mkv", CreateDate, DateTime::parse_from_str("2008-08-08T08:08:08Z", "%+").unwrap().into())]
601 #[test_case("meta.mov", Make, "Apple".into())]
602 #[test_case("meta.mov", Model, "iPhone X".into())]
603 #[test_case("meta.mov", GpsIso6709, "+27.1281+100.2508+000.000/".into())]
604 #[test_case("meta.mp4", ImageWidth, 1920_u32.into())]
605 #[test_case("meta.mp4", ImageHeight, 1080_u32.into())]
606 #[test_case("meta.mp4", DurationMs, 1063_u64.into())]
607 #[test_case("meta.mp4", GpsIso6709, "+27.2939+112.6932/".into())]
608 #[test_case("meta.mp4", CreateDate, DateTime::parse_from_str("2024-02-03T07:05:38Z", "%+").unwrap().into())]
609 #[test_case("udta.auth.mp4", Author, "ReplayKitRecording".into(); "udta author")]
610 #[test_case("auth.mov", Author, "ReplayKitRecording".into(); "mov author")]
611 fn parse_track_info(path: &str, tag: TrackInfoTag, v: EntryValue) {
612 let mut parser = parser();
613
614 let mf = MediaSource::file(open_sample(path).unwrap()).unwrap();
615 let info: TrackInfo = parser.parse(mf).unwrap();
616 assert_eq!(info.get(tag).unwrap(), &v);
617
618 let mf = MediaSource::unseekable(open_sample(path).unwrap()).unwrap();
619 let info: TrackInfo = parser.parse(mf).unwrap();
620 assert_eq!(info.get(tag).unwrap(), &v);
621 }
622
623 #[test_case("crash_moov-trak")]
624 #[test_case("crash_skip_large")]
625 #[test_case("crash_add_large")]
626 fn parse_track_crash(path: &str) {
627 let mut parser = parser();
628
629 let mf = MediaSource::file(open_sample(path).unwrap()).unwrap();
630 let _: TrackInfo = parser.parse(mf).unwrap_or_default();
631
632 let mf = MediaSource::unseekable(open_sample(path).unwrap()).unwrap();
633 let _: TrackInfo = parser.parse(mf).unwrap_or_default();
634 }
635}