1use std::{
2 collections::BTreeMap,
3 io::{Read, Seek},
4 ops::Range,
5};
6
7use chrono::DateTime;
8use nom::{bytes::streaming, IResult};
9
10use crate::{bbox::to_boxes, values::filter_zero};
11#[allow(deprecated)]
12use crate::{
13 bbox::{
14 find_box, parse_video_tkhd_in_moov, travel_header, IlstBox, KeysBox, MvhdBox, ParseBox,
15 },
16 error::ParsingError,
17 loader::{BufLoader, Load},
18 partial_vec::PartialVec,
19 skip::Seekable,
20 video::TrackInfoTag,
21 EntryValue, FileFormat,
22};
23
24#[deprecated(since = "2.0.0")]
61#[tracing::instrument(skip_all)]
62#[allow(deprecated)]
63pub fn parse_metadata<R: Read + Seek>(reader: R) -> crate::Result<Vec<(String, EntryValue)>> {
64 let mut loader = BufLoader::<Seekable, _>::new(reader);
65 let ff = FileFormat::try_from_load(&mut loader)?;
66 match ff {
67 FileFormat::Jpeg | FileFormat::Heif => {
68 return Err(crate::error::Error::ParseFailed(
69 "can not parse metadata from an image".into(),
70 ));
71 }
72 FileFormat::QuickTime | FileFormat::MP4 => (),
73 FileFormat::Ebml => {
74 return Err(crate::error::Error::ParseFailed(
75 "please use MediaParser to parse *.webm, *.mkv files".into(),
76 ))
77 }
78 };
79
80 let moov_body = extract_moov_body(loader)?;
81
82 let (_, mut entries) = match parse_moov_body(&moov_body) {
83 Ok((remain, Some(entries))) => (remain, entries),
84 Ok((remain, None)) => (remain, Vec::new()),
85 Err(_) => {
86 return Err("invalid moov body".into());
87 }
88 };
89
90 let map: BTreeMap<TrackInfoTag, EntryValue> = convert_video_tags(entries.clone());
91 let mut extras = parse_mvhd_tkhd(&moov_body);
92
93 const CREATIONDATE_KEY: &str = "com.apple.quicktime.creationdate";
94 if map.contains_key(&TrackInfoTag::CreateDate) {
95 extras.remove(&TrackInfoTag::CreateDate);
96 let date = map.get(&TrackInfoTag::CreateDate);
97 if let Some(pos) = entries.iter().position(|x| x.0 == CREATIONDATE_KEY) {
98 if let Some(date) = date {
99 entries[pos] = (CREATIONDATE_KEY.to_string(), date.clone());
100 } else {
101 entries.remove(pos);
102 }
103 }
104 }
105
106 entries.extend(extras.into_iter().map(|(k, v)| match k {
107 TrackInfoTag::ImageWidth => ("width".to_string(), v),
108 TrackInfoTag::ImageHeight => ("height".to_string(), v),
109 TrackInfoTag::DurationMs => (
110 "duration".to_string(),
111 EntryValue::U32(v.as_u64().unwrap() as u32),
113 ),
114 TrackInfoTag::CreateDate => (CREATIONDATE_KEY.to_string(), v),
115 _ => unreachable!(),
116 }));
117
118 if map.contains_key(&TrackInfoTag::GpsIso6709) {
119 const LOCATION_KEY: &str = "com.apple.quicktime.location.ISO6709";
120 if let Some(idx) = entries.iter().position(|(k, _)| k == "udta.©xyz") {
121 entries.remove(idx);
122 entries.push((
123 LOCATION_KEY.to_string(),
124 map.get(&TrackInfoTag::GpsIso6709).unwrap().to_owned(),
125 ));
126 }
127 }
128
129 Ok(entries)
130}
131
132#[tracing::instrument(skip_all)]
133pub(crate) fn parse_isobmff(
134 moov_body: &[u8],
135) -> Result<BTreeMap<TrackInfoTag, EntryValue>, ParsingError> {
136 let (_, entries) = match parse_moov_body(moov_body) {
137 Ok((remain, Some(entries))) => (remain, entries),
138 Ok((remain, None)) => (remain, Vec::new()),
139 Err(_) => {
140 return Err("invalid moov body".into());
141 }
142 };
143
144 let mut entries: BTreeMap<TrackInfoTag, EntryValue> = convert_video_tags(entries);
145 let mut extras = parse_mvhd_tkhd(moov_body);
146 if entries.contains_key(&TrackInfoTag::CreateDate) {
147 extras.remove(&TrackInfoTag::CreateDate);
148 }
149 entries.extend(extras);
150
151 Ok(entries)
152}
153
154fn parse_mvhd_tkhd(moov_body: &[u8]) -> BTreeMap<TrackInfoTag, EntryValue> {
155 let mut entries = BTreeMap::new();
156 if let Ok((_, Some(bbox))) = find_box(moov_body, "mvhd") {
157 if let Ok((_, mvhd)) = MvhdBox::parse_box(bbox.data) {
158 entries.insert(TrackInfoTag::DurationMs, mvhd.duration_ms().into());
159
160 entries.insert(
161 TrackInfoTag::CreateDate,
162 EntryValue::Time(mvhd.creation_time()),
163 );
164 }
165 }
166
167 if let Ok(Some(tkhd)) = parse_video_tkhd_in_moov(moov_body) {
168 entries.insert(TrackInfoTag::ImageWidth, tkhd.width.into());
169 entries.insert(TrackInfoTag::ImageHeight, tkhd.height.into());
170 }
171
172 entries
173}
174
175fn convert_video_tags(entries: Vec<(String, EntryValue)>) -> BTreeMap<TrackInfoTag, EntryValue> {
176 entries
177 .into_iter()
178 .filter_map(|(k, v)| {
179 if k == "com.apple.quicktime.creationdate" {
180 v.as_str()
181 .and_then(|s| DateTime::parse_from_str(s, "%+").ok())
182 .map(|t| (TrackInfoTag::CreateDate, EntryValue::Time(t)))
183 } else if k == "com.apple.quicktime.make" {
184 Some((TrackInfoTag::Make, v))
185 } else if k == "com.apple.quicktime.model" {
186 Some((TrackInfoTag::Model, v))
187 } else if k == "com.apple.quicktime.software" {
188 Some((TrackInfoTag::Software, v))
189 } else if k == "com.apple.quicktime.author" {
190 Some((TrackInfoTag::Author, v))
191 } else if k == "com.apple.quicktime.location.ISO6709" {
192 Some((TrackInfoTag::GpsIso6709, v))
193 } else if k == "udta.©xyz" {
194 v.as_u8array()
196 .and_then(parse_udta_gps)
197 .map(|v| (TrackInfoTag::GpsIso6709, EntryValue::Text(v)))
198 } else if k == "udta.auth" {
199 v.as_u8array()
200 .and_then(parse_udta_auth)
201 .map(|v| (TrackInfoTag::Author, EntryValue::Text(v)))
202 } else if k.starts_with("udta.") {
203 let tag = TryInto::<TrackInfoTag>::try_into(k.as_str()).ok();
204 tag.map(|t| (t, v))
205 } else {
206 None
207 }
208 })
209 .collect()
210}
211
212fn parse_udta_gps(data: &[u8]) -> Option<String> {
225 if data.len() <= 4 {
226 tracing::warn!("moov/udta/©xyz body is too small");
227 None
228 } else {
229 let location = data[4..] .iter()
232 .map(|b| *b as char)
233 .collect::<String>();
234 Some(location)
235 }
236}
237
238const ISO_639_2_UND: [u8; 2] = [0x55, 0xc4];
239
240fn parse_udta_auth(data: &[u8]) -> Option<String> {
241 let data = filter_zero(data);
243
244 if data.starts_with(&ISO_639_2_UND) {
247 String::from_utf8(data.into_iter().skip(2).collect()).ok()
248 } else {
249 String::from_utf8(data).ok()
250 }
251}
252
253#[deprecated(since = "2.0.0")]
290pub fn parse_mov_metadata<R: Read + Seek>(reader: R) -> crate::Result<Vec<(String, EntryValue)>> {
291 #[allow(deprecated)]
292 parse_metadata(reader)
293}
294
295#[tracing::instrument(skip_all)]
296fn extract_moov_body<L: Load>(mut loader: L) -> Result<PartialVec, crate::Error> {
297 let moov_body_range = loader.load_and_parse(extract_moov_body_from_buf)?;
298
299 tracing::debug!(?moov_body_range);
300 Ok(PartialVec::from_vec_range(
301 loader.into_vec(),
302 moov_body_range,
303 ))
304}
305
306#[tracing::instrument(skip_all)]
311pub(crate) fn extract_moov_body_from_buf(input: &[u8]) -> Result<Range<usize>, ParsingError> {
312 let remain = input;
314
315 let convert_error = |e: nom::Err<_>, msg: &str| match e {
316 nom::Err::Incomplete(needed) => match needed {
317 nom::Needed::Unknown => ParsingError::Need(1),
318 nom::Needed::Size(n) => ParsingError::Need(n.get()),
319 },
320 nom::Err::Failure(_) | nom::Err::Error(_) => ParsingError::Failed(msg.to_string()),
321 };
322
323 let mut to_skip = 0;
324 let mut skipped = 0;
325 let (remain, header) = travel_header(remain, |h, remain| {
326 tracing::debug!(?h.box_type, ?h.box_size, "Got");
327 if h.box_type == "moov" {
328 skipped += h.header_size;
330 false
331 } else if (remain.len() as u64) < h.body_size() {
332 to_skip = h.body_size() as usize - remain.len();
334 false
335 } else {
336 skipped += h.box_size as usize;
338 true
339 }
340 })
341 .map_err(|e| convert_error(e, "search atom moov failed"))?;
342
343 if to_skip > 0 {
344 return Err(ParsingError::ClearAndSkip(
345 to_skip
346 .checked_add(input.len())
347 .ok_or_else(|| ParsingError::Failed("to_skip is too big".into()))?,
348 ));
349 }
350
351 let size: usize = header.body_size().try_into().expect("must fit");
352 let (_, body) =
353 streaming::take(size)(remain).map_err(|e| convert_error(e, "moov is too small"))?;
354
355 Ok(skipped..skipped + body.len())
356}
357
358type EntriesResult<'a> = IResult<&'a [u8], Option<Vec<(String, EntryValue)>>>;
359
360#[tracing::instrument(skip(input))]
361fn parse_moov_body(input: &[u8]) -> EntriesResult<'_> {
362 tracing::debug!("parse_moov_body");
363
364 let mut entries = parse_meta(input).unwrap_or_default();
365
366 if let Ok((_, Some(udta))) = find_box(input, "udta") {
367 tracing::debug!("udta");
368 if let Ok(boxes) = to_boxes(udta.body_data()) {
369 for entry in boxes.iter() {
370 tracing::debug!(?entry, "udta entry");
371 entries.push((
372 format!("udta.{}", entry.box_type()),
373 EntryValue::U8Array(Vec::from(entry.body_data())),
374 ));
375 }
376 }
377 }
378
379 Ok((input, Some(entries)))
380}
381
382fn parse_meta(input: &[u8]) -> Option<Vec<(String, EntryValue)>> {
383 let (_, Some(meta)) = find_box(input, "meta").ok()? else {
384 return None;
385 };
386
387 let (_, Some(keys)) = find_box(meta.body_data(), "keys").ok()? else {
388 return None;
389 };
390
391 let (_, Some(ilst)) = find_box(meta.body_data(), "ilst").ok()? else {
392 return None;
393 };
394
395 let (_, keys) = KeysBox::parse_box(keys.data).ok()?;
396 let (_, ilst) = IlstBox::parse_box(ilst.data).ok()?;
397
398 let entries = keys
399 .entries
400 .into_iter()
401 .map(|k| k.key)
402 .zip(ilst.items.into_iter().map(|v| v.value))
403 .collect::<Vec<_>>();
404
405 Some(entries)
406}
407
408#[allow(dead_code)]
413fn tz_iso_8601_to_rfc3339(s: String) -> String {
414 use regex::Regex;
415
416 let ss = s.trim();
417 let re = Regex::new(r"([+-][0-9][0-9])([0-9][0-9])?$").unwrap();
419
420 if let Some((offset, tz)) = re.captures(ss).map(|caps| {
421 (
422 caps.get(1).unwrap().start(),
424 format!(
425 "{}:{}",
426 caps.get(1).map_or("00", |m| m.as_str()),
427 caps.get(2).map_or("00", |m| m.as_str())
428 ),
429 )
430 }) {
431 let s1 = &ss.as_bytes()[..offset]; let s2 = tz.as_bytes();
433 s1.iter().chain(s2.iter()).map(|x| *x as char).collect()
434 } else {
435 s
436 }
437}
438
439#[cfg(test)]
440#[allow(deprecated)]
441mod tests {
442 use super::*;
443 use crate::testkit::*;
444 use test_case::test_case;
445
446 #[test_case("meta.mov")]
447 fn mov_parse(path: &str) {
448 let reader = open_sample(path).unwrap();
449 let entries = parse_metadata(reader).unwrap();
450 assert_eq!(
451 entries
452 .iter()
453 .map(|x| format!("{x:?}"))
454 .collect::<Vec<_>>()
455 .join("\n"),
456 "(\"com.apple.quicktime.make\", Text(\"Apple\"))
457(\"com.apple.quicktime.model\", Text(\"iPhone X\"))
458(\"com.apple.quicktime.software\", Text(\"12.1.2\"))
459(\"com.apple.quicktime.location.ISO6709\", Text(\"+27.1281+100.2508+000.000/\"))
460(\"com.apple.quicktime.creationdate\", Time(2019-02-12T15:27:12+08:00))
461(\"duration\", U32(500))
462(\"width\", U32(720))
463(\"height\", U32(1280))"
464 );
465 }
466
467 #[test_case("meta.mov")]
468 fn mov_extract_mov(path: &str) {
469 let _ = tracing_subscriber::fmt().with_test_writer().try_init();
470
471 let buf = read_sample(path).unwrap();
472 tracing::info!(bytes = buf.len(), "File size.");
473 let range = extract_moov_body_from_buf(&buf).unwrap();
474 let (_, entries) = parse_moov_body(&buf[range]).unwrap();
475 assert_eq!(
476 entries
477 .unwrap()
478 .iter()
479 .map(|x| format!("{x:?}"))
480 .collect::<Vec<_>>()
481 .join("\n"),
482 "(\"com.apple.quicktime.make\", Text(\"Apple\"))
483(\"com.apple.quicktime.model\", Text(\"iPhone X\"))
484(\"com.apple.quicktime.software\", Text(\"12.1.2\"))
485(\"com.apple.quicktime.location.ISO6709\", Text(\"+27.1281+100.2508+000.000/\"))
486(\"com.apple.quicktime.creationdate\", Text(\"2019-02-12T15:27:12+08:00\"))"
487 );
488 }
489
490 #[test_case("meta.mp4")]
491 fn parse_mp4(path: &str) {
492 let _ = tracing_subscriber::fmt().with_test_writer().try_init();
493
494 let entries = parse_metadata(open_sample(path).unwrap()).unwrap();
495 assert_eq!(
496 entries
497 .iter()
498 .map(|x| format!("{x:?}"))
499 .collect::<Vec<_>>()
500 .join("\n"),
501 "(\"com.apple.quicktime.creationdate\", Time(2024-02-03T07:05:38+00:00))
502(\"duration\", U32(1063))
503(\"width\", U32(1920))
504(\"height\", U32(1080))
505(\"com.apple.quicktime.location.ISO6709\", Text(\"+27.2939+112.6932/\"))"
506 );
507 }
508
509 #[test_case("embedded-in-heic.mov")]
510 fn parse_embedded_mov(path: &str) {
511 let _ = tracing_subscriber::fmt().with_test_writer().try_init();
512
513 let entries = parse_mov_metadata(open_sample(path).unwrap()).unwrap();
514 assert_eq!(
515 entries
516 .iter()
517 .map(|x| format!("{x:?}"))
518 .collect::<Vec<_>>()
519 .join("\n"),
520 "(\"com.apple.quicktime.location.accuracy.horizontal\", Text(\"14.235563\"))
521(\"com.apple.quicktime.live-photo.auto\", U8(1))
522(\"com.apple.quicktime.content.identifier\", Text(\"DA1A7EE8-0925-4C9F-9266-DDA3F0BB80F0\"))
523(\"com.apple.quicktime.live-photo.vitality-score\", F32(0.93884003))
524(\"com.apple.quicktime.live-photo.vitality-scoring-version\", I64(4))
525(\"com.apple.quicktime.location.ISO6709\", Text(\"+22.5797+113.9380+028.396/\"))
526(\"com.apple.quicktime.make\", Text(\"Apple\"))
527(\"com.apple.quicktime.model\", Text(\"iPhone 15 Pro\"))
528(\"com.apple.quicktime.software\", Text(\"17.1\"))
529(\"com.apple.quicktime.creationdate\", Time(2023-11-02T19:58:34+08:00))
530(\"duration\", U32(2795))
531(\"width\", U32(1920))
532(\"height\", U32(1440))"
533 );
534 }
535
536 #[test]
537 fn test_iso_8601_tz_to_rfc3339() {
538 let _ = tracing_subscriber::fmt().with_test_writer().try_init();
539
540 let s = "2023-11-02T19:58:34+08".to_string();
541 assert_eq!(tz_iso_8601_to_rfc3339(s), "2023-11-02T19:58:34+08:00");
542
543 let s = "2023-11-02T19:58:34+0800".to_string();
544 assert_eq!(tz_iso_8601_to_rfc3339(s), "2023-11-02T19:58:34+08:00");
545
546 let s = "2023-11-02T19:58:34+08:00".to_string();
547 assert_eq!(tz_iso_8601_to_rfc3339(s), "2023-11-02T19:58:34+08:00");
548
549 let s = "2023-11-02T19:58:34Z".to_string();
550 assert_eq!(tz_iso_8601_to_rfc3339(s), "2023-11-02T19:58:34Z");
551
552 let s = "2023-11-02T19:58:34".to_string();
553 assert_eq!(tz_iso_8601_to_rfc3339(s), "2023-11-02T19:58:34");
554 }
555}