audiofp 0.3.9

Pure-Rust audio fingerprinting and identification: Wang, Panako, Haitsma–Kalker, ONNX neural embedder, AudioSeal watermark, and streaming variants. no_std + alloc capable, bytemuck-friendly hash types.
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
//! One-shot audio file decoding via Symphonia.

use std::fs::File;
use std::path::Path;

use symphonia::core::audio::{Audio, AudioBuffer, GenericAudioBufferRef};
use symphonia::core::codecs::audio::AudioDecoderOptions;
use symphonia::core::errors::Error as SymphoniaError;
use symphonia::core::formats::probe::Hint;
use symphonia::core::formats::{FormatOptions, FormatReader, TrackType};
use symphonia::core::io::MediaSourceStream;
use symphonia::core::meta::MetadataOptions;

use crate::dsp::resample::SincResampler;
use crate::error::IoError;
use crate::{AfpError, Result};

/// Resource limits for untrusted-upload decoding.
///
/// Use **both** caps in production: `max_bytes` rejects oversized files
/// before opening the stream, while `max_samples` bounds decoded mono
/// PCM growth (critical for compressed formats where on-disk size does
/// not bound decoded size).
#[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
pub struct DecodeLimits {
    /// Reject when on-disk file size exceeds this many bytes.
    /// `0` disables the byte check.
    pub max_bytes: u64,
    /// Reject when decoded mono PCM would exceed this many samples.
    /// `None` disables the sample check.
    pub max_samples: Option<usize>,
    /// When `true`, any per-packet decode error (corrupt frame, I/O glitch)
    /// becomes a fatal error instead of being silently skipped.
    /// Default: `false` (skip recoverable errors, matching legacy behavior).
    pub integrity_mode: bool,
}

impl DecodeLimits {
    /// Byte-only cap (`max_samples = None`). Prefer [`Self::both`] for
    /// compressed uploads.
    #[must_use]
    pub const fn bytes(max_bytes: u64) -> Self {
        Self {
            max_bytes,
            max_samples: None,
            integrity_mode: false,
        }
    }

    /// Sample-only cap (`max_bytes = 0`).
    #[must_use]
    pub const fn samples(max_samples: usize) -> Self {
        Self {
            max_bytes: 0,
            max_samples: Some(max_samples),
            integrity_mode: false,
        }
    }

    /// Both on-disk and decoded-PCM caps.
    #[must_use]
    pub const fn both(max_bytes: u64, max_samples: usize) -> Self {
        Self {
            max_bytes,
            max_samples: Some(max_samples),
            integrity_mode: false,
        }
    }

    /// Enable integrity mode: any per-packet decode error becomes fatal
    /// instead of being silently skipped.
    ///
    /// # Example
    ///
    /// ```
    /// use audiofp::io::DecodeLimits;
    ///
    /// let limits = DecodeLimits::both(10_000_000, 480_000).strict();
    /// assert!(limits.integrity_mode);
    /// ```
    #[must_use]
    pub const fn strict(mut self) -> Self {
        self.integrity_mode = true;
        self
    }
}

/// Decode an audio file into a mono `f32` buffer at the file's native
/// sample rate.
///
/// Multi-channel files are downmixed to mono by averaging channels per
/// frame. The returned tuple is `(samples, sample_rate_hz)`.
///
/// # Supported formats
///
/// MP3, FLAC, WAV, OGG-Vorbis, AAC-in-MP4, raw PCM — whatever Symphonia's
/// default registries provide with the features enabled in
/// `audiofp`'s `Cargo.toml`. The decoder probes magic bytes too, so
/// extension-less files still work as long as they're a recognised format.
///
/// # Errors
///
/// - [`AfpError::Io`] if the file is missing, the format isn't recognised,
///   or a stream-fatal decode error happens. Recoverable per-packet failures
///   inside Symphonia are silently skipped so a single corrupt block
///   doesn't kill the whole-file decode.
///
/// # Example
///
/// # Security
///
/// This function applies **no resource limits**. A compressed
/// decompression bomb (tiny on-disk, expands to gigabytes of PCM) will
/// succeed and may OOM the process. For untrusted uploads use
/// [`decode_to_mono_limited`] with [`DecodeLimits::both`] so both
/// on-disk size and decoded PCM are bounded.
///
/// ```no_run
/// use audiofp::io::decode_to_mono;
///
/// # fn main() -> Result<(), Box<dyn std::error::Error>> {
/// let (samples, sr) = decode_to_mono("song.flac")?;
/// println!("{} samples at {sr} Hz", samples.len());
/// # Ok(()) }
/// ```
pub fn decode_to_mono<P: AsRef<Path>>(path: P) -> Result<(Vec<f32>, u32)> {
    decode_to_mono_limited(path, DecodeLimits::default())
}

/// Decode with explicit on-disk and/or decoded-PCM caps.
///
/// # Errors
///
/// - [`AfpError::InputTooLarge`] if the file exceeds `max_bytes` or
///   decoded mono samples would exceed `max_samples`.
/// - [`AfpError::Io`] for missing/unrecognised/corrupt streams (same as
///   [`decode_to_mono`]).
pub fn decode_to_mono_limited<P: AsRef<Path>>(
    path: P,
    limits: DecodeLimits,
) -> Result<(Vec<f32>, u32)> {
    let path = path.as_ref();
    // Pre-check: don't even open files that are clearly too large.
    // Note: this is best-effort against TOCTOU (file can grow after the
    // stat); `max_samples` is the hard bound on decoded PCM.
    if limits.max_bytes > 0 {
        let meta = std::fs::metadata(path).map_err(|e| AfpError::io_with_path(path, e))?;
        let len = meta.len();
        if len > limits.max_bytes {
            return Err(AfpError::InputTooLarge {
                limit: usize::try_from(limits.max_bytes).unwrap_or(usize::MAX),
                provided: usize::try_from(len).unwrap_or(usize::MAX),
            });
        }
    }
    let file = File::open(path).map_err(|e| AfpError::io_with_path(path, e))?;
    let mss = MediaSourceStream::new(Box::new(file), Default::default());

    let mut hint = Hint::new();
    if let Some(ext) = path.extension().and_then(|e| e.to_str()) {
        hint.with_extension(ext);
    }

    decode_inner(mss, &hint, limits.max_samples, limits.integrity_mode)
}

/// Decode an audio file and resample it to `target_sr` Hz mono `f32`.
///
/// Pass-through (no resample) when the file already matches `target_sr`.
/// Otherwise resamples via [`SincResampler`] at default quality
/// (32-tap Kaiser, β = 8.6). Equivalent to calling [`decode_to_mono`]
/// then [`SincResampler::process`] yourself, but in one step.
///
/// # Errors
///
/// Surfaces every error [`decode_to_mono`] can return; resampling itself
/// cannot fail with the built-in [`SincResampler`].
///
/// # Example
///
/// ```no_run
/// use audiofp::io::decode_to_mono_at;
///
/// # fn main() -> Result<(), Box<dyn std::error::Error>> {
/// // Get audio ready for Wang in one line:
/// let samples = decode_to_mono_at("song.mp3", 8_000)?;
/// # Ok(()) }
/// ```
pub fn decode_to_mono_at<P: AsRef<Path>>(path: P, target_sr: u32) -> Result<Vec<f32>> {
    decode_to_mono_at_limited(path, target_sr, DecodeLimits::default())
}

/// Same as [`decode_to_mono_at`] with full [`DecodeLimits`].
pub fn decode_to_mono_at_limited<P: AsRef<Path>>(
    path: P,
    target_sr: u32,
    limits: DecodeLimits,
) -> Result<Vec<f32>> {
    if target_sr == 0 {
        return Err(AfpError::Config("target sample rate must be > 0".into()));
    }
    let (samples, sr) = decode_to_mono_limited(path, limits)?;
    if sr == target_sr {
        Ok(samples)
    } else {
        let r = SincResampler::new(sr, target_sr);
        Ok(r.process(&samples))
    }
}

fn decode_inner(
    mss: MediaSourceStream,
    hint: &Hint,
    max_samples: Option<usize>,
    integrity_mode: bool,
) -> Result<(Vec<f32>, u32)> {
    let mut format: Box<dyn FormatReader> = symphonia::default::get_probe()
        .probe(
            hint,
            mss,
            FormatOptions::default(),
            MetadataOptions::default(),
        )
        .map_err(|e| {
            AfpError::Io(IoError::without_path(std::io::Error::other(format!(
                "probe: {e}"
            ))))
        })?;

    let track = format
        .default_track(TrackType::Audio)
        .ok_or_else(|| {
            AfpError::Io(IoError::without_path(std::io::Error::other(
                "no audio track",
            )))
        })?
        .clone();
    let track_id = track.id;

    let audio_params = match track.codec_params.as_ref() {
        Some(symphonia::core::codecs::CodecParameters::Audio(params)) => params,
        _ => {
            return Err(AfpError::Io(IoError::without_path(std::io::Error::other(
                "no audio codec params",
            ))));
        }
    };

    let sample_rate = audio_params.sample_rate.ok_or_else(|| {
        AfpError::Io(IoError::without_path(std::io::Error::other(
            "missing sample rate",
        )))
    })?;

    let codecs = symphonia::default::get_codecs();
    let decoder_factory = codecs
        .get_audio_decoder(audio_params.codec)
        .ok_or_else(|| {
            AfpError::Io(IoError::without_path(std::io::Error::other(
                "unsupported codec",
            )))
        })?;
    let mut decoder = (decoder_factory.factory)(audio_params, &AudioDecoderOptions::default())
        .map_err(|e| {
            AfpError::Io(IoError::without_path(std::io::Error::other(format!(
                "make decoder: {e}"
            ))))
        })?;

    let mut samples: Vec<f32> = Vec::new();
    let mut convert_buf: Option<AudioBuffer<f32>> = None;

    loop {
        let packet = match format.next_packet() {
            Ok(Some(p)) => p,
            Ok(None) => break,
            Err(SymphoniaError::IoError(e)) if e.kind() == std::io::ErrorKind::UnexpectedEof => {
                break;
            }
            Err(SymphoniaError::ResetRequired) => continue,
            Err(e) => {
                return Err(AfpError::Io(IoError::without_path(std::io::Error::other(
                    format!("next_packet: {e}"),
                ))));
            }
        };
        if packet.track_id != track_id {
            continue;
        }

        let decoded: GenericAudioBufferRef = match decoder.decode(&packet) {
            Ok(d) => d,
            // Recoverable per-packet failures: skip unless integrity mode is on.
            Err(SymphoniaError::IoError(e)) => {
                if integrity_mode {
                    return Err(AfpError::Io(IoError::without_path(std::io::Error::other(
                        format!("decode integrity: {e}"),
                    ))));
                }
                continue;
            }
            Err(SymphoniaError::DecodeError(e)) => {
                if integrity_mode {
                    return Err(AfpError::Io(IoError::without_path(std::io::Error::other(
                        format!("decode integrity: {e}"),
                    ))));
                }
                continue;
            }
            Err(e) => {
                return Err(AfpError::Io(IoError::without_path(std::io::Error::other(
                    format!("decode: {e}"),
                ))));
            }
        };

        // Lazily allocate the f32 conversion buffer once the first packet
        // tells us the channel layout / capacity. Reallocate if a later
        // packet decodes to more frames than the current buffer can hold
        // (the first packet's capacity is not guaranteed to bound the rest).
        let needed_cap = decoded.frames().max(decoded.capacity());
        let needs_buf = match &convert_buf {
            None => true,
            Some(buf) => needed_cap > buf.capacity(),
        };
        if needs_buf {
            let spec = decoded.spec().clone();
            convert_buf = Some(AudioBuffer::<f32>::new(spec, needed_cap));
        }
        let buf = convert_buf.as_mut().unwrap();

        // In symphonia 0.6, copy_to requires the destination to have the
        // same frame count as the source. Set it before copying.
        buf.resize_uninit(decoded.frames());
        decoded.copy_to::<f32, _>(buf);

        let n_frames = buf.frames();
        let n_chans = buf.spec().channels().count();

        // Defensive: skip packets that report 0 channels (malformed /
        // corrupt). Avoids division by zero and `.plane(0).unwrap()` panic.
        if n_chans == 0 {
            continue;
        }

        // Bound decoded PCM growth before allocating more samples.
        if let Some(limit) = max_samples {
            let next = samples.len().saturating_add(n_frames);
            if next > limit {
                return Err(AfpError::InputTooLarge {
                    limit,
                    provided: next,
                });
            }
        }

        if n_chans == 1 {
            samples.extend_from_slice(&buf.plane(0).unwrap()[..n_frames]);
        } else {
            samples.reserve(n_frames);
            for i in 0..n_frames {
                let mut sum = 0.0_f32;
                for c in 0..n_chans {
                    sum += buf.plane(c).unwrap()[i];
                }
                samples.push(sum / n_chans as f32);
            }
        }
    }

    Ok((samples, sample_rate))
}

#[cfg(test)]
mod tests {
    use super::*;
    use core::f32::consts::PI;
    use std::io::Write;

    fn write_test_wav(channels: u16, sr: u32, len: usize) -> std::path::PathBuf {
        // Counter ensures each test gets a unique path so parallel runs
        // don't clobber each other.
        use std::sync::atomic::{AtomicU64, Ordering};
        static COUNTER: AtomicU64 = AtomicU64::new(0);
        let n = COUNTER.fetch_add(1, Ordering::Relaxed);
        let path = std::env::temp_dir().join(format!(
            "audiofp-decoder-test-{}-{}-{}-{}-{}.wav",
            std::process::id(),
            channels,
            sr,
            len,
            n,
        ));
        let spec = hound::WavSpec {
            channels,
            sample_rate: sr,
            bits_per_sample: 16,
            sample_format: hound::SampleFormat::Int,
        };
        let mut writer = hound::WavWriter::create(&path, spec).unwrap();
        let amp = (i16::MAX as f32) * 0.5;
        for i in 0..len {
            // 440 Hz tone on every channel (mono on every channel for
            // multichannel files = identical channels, downmix is identity).
            let s = libm::sinf(2.0 * PI * 440.0 * i as f32 / sr as f32) * amp;
            for _c in 0..channels {
                writer.write_sample(s as i16).unwrap();
            }
        }
        writer.finalize().unwrap();
        path
    }

    #[test]
    fn open_missing_file_returns_io_error() {
        let res = decode_to_mono("/nonexistent/path/that/does/not/exist.wav");
        match res {
            Err(AfpError::Io(_)) => {}
            other => panic!("expected Io error, got {other:?}"),
        }
    }

    #[test]
    fn round_trip_mono_wav() {
        let path = write_test_wav(1, 8_000, 8_000);
        let result = decode_to_mono(&path);
        std::fs::remove_file(&path).ok();
        let (samples, sr) = result.unwrap();
        assert_eq!(sr, 8_000);
        assert_eq!(samples.len(), 8_000);

        // Spot-check a sample mid-buffer.
        let expected = libm::sinf(2.0 * PI * 440.0 * 100.0 / 8_000.0) * 0.5;
        // 16-bit truncation introduces ~3e-5 error; allow a generous bound.
        assert!(
            (samples[100] - expected).abs() < 0.01,
            "sample[100] = {}, expected ≈ {expected}",
            samples[100]
        );
    }

    #[test]
    fn stereo_wav_downmixes_to_mono() {
        // Both channels are identical so downmix should be the same signal.
        let path = write_test_wav(2, 16_000, 16_000);
        let result = decode_to_mono(&path);
        std::fs::remove_file(&path).ok();
        let (samples, sr) = result.unwrap();
        assert_eq!(sr, 16_000);
        assert_eq!(samples.len(), 16_000);

        let expected = libm::sinf(2.0 * PI * 440.0 * 200.0 / 16_000.0) * 0.5;
        assert!((samples[200] - expected).abs() < 0.01);
    }

    #[test]
    fn decode_to_mono_at_resamples() {
        let path = write_test_wav(1, 16_000, 16_000); // 1 sec @ 16 kHz
        let result = decode_to_mono_at(&path, 8_000);
        std::fs::remove_file(&path).ok();
        let samples = result.unwrap();
        // 16k → 8k means roughly half as many samples.
        assert!(
            (samples.len() as i64 - 8_000).abs() < 16,
            "resampled len = {}",
            samples.len()
        );
    }

    #[test]
    fn decode_to_mono_at_passthrough_when_rates_match() {
        let path = write_test_wav(1, 8_000, 4_000);
        let result = decode_to_mono_at(&path, 8_000);
        std::fs::remove_file(&path).ok();
        let samples = result.unwrap();
        assert_eq!(samples.len(), 4_000);
    }

    #[test]
    fn unknown_extension_still_decodes() {
        // Symphonia probes magic bytes too, so an extensionless file still
        // works as long as it's a recognised format.
        let path = write_test_wav(1, 8_000, 4_000);
        let renamed = path.with_extension("");
        std::fs::rename(&path, &renamed).unwrap();

        let result = decode_to_mono(&renamed);
        std::fs::remove_file(&renamed).ok();

        // Use of ? syntax via match: succeed or report.
        let (samples, sr) = match result {
            Ok(v) => v,
            Err(e) => panic!("decode without extension failed: {e}"),
        };
        assert_eq!(sr, 8_000);
        assert_eq!(samples.len(), 4_000);
    }

    /// Ensure the public APIs don't hold onto the file handle past
    /// successful decode (otherwise removing the file would fail on
    /// Windows; on Unix it would leak a descriptor).
    #[test]
    fn temp_file_can_be_deleted_after_decode() {
        let path = write_test_wav(1, 8_000, 1_000);
        decode_to_mono(&path).unwrap();
        // Should not error out.
        std::fs::remove_file(&path).unwrap();
    }

    /// Dummy `Write` ensures the unused-import pruner doesn't strip
    /// `std::io::Write` if a future test needs in-memory writers.
    #[allow(dead_code)]
    fn _writer_witness<W: Write>(_w: W) {}

    fn write_test_wav_float(channels: u16, sr: u32, len: usize) -> std::path::PathBuf {
        use std::sync::atomic::{AtomicU64, Ordering};
        static COUNTER: AtomicU64 = AtomicU64::new(0);
        let n = COUNTER.fetch_add(1, Ordering::Relaxed);
        let path = std::env::temp_dir().join(format!(
            "audiofp-decoder-float-{}-{}-{}-{}.wav",
            std::process::id(),
            channels,
            sr,
            n,
        ));
        let spec = hound::WavSpec {
            channels,
            sample_rate: sr,
            bits_per_sample: 32,
            sample_format: hound::SampleFormat::Float,
        };
        let mut writer = hound::WavWriter::create(&path, spec).unwrap();
        for i in 0..len {
            let s = libm::sinf(2.0 * PI * 440.0 * i as f32 / sr as f32) * 0.5;
            for _c in 0..channels {
                writer.write_sample(s).unwrap();
            }
        }
        writer.finalize().unwrap();
        path
    }

    #[test]
    fn float_wav_decodes_with_higher_precision() {
        let path = write_test_wav_float(1, 16_000, 4_000);
        let result = decode_to_mono(&path);
        std::fs::remove_file(&path).ok();
        let (samples, sr) = result.unwrap();
        assert_eq!(sr, 16_000);
        assert_eq!(samples.len(), 4_000);
        // 32-bit float should give near-exact reconstruction.
        let expected = libm::sinf(2.0 * PI * 440.0 * 100.0 / 16_000.0) * 0.5;
        assert!(
            (samples[100] - expected).abs() < 1e-6,
            "sample[100] = {}, expected {expected}",
            samples[100]
        );
    }

    #[test]
    fn high_sample_rate_preserved() {
        let path = write_test_wav(1, 48_000, 4_800);
        let result = decode_to_mono(&path);
        std::fs::remove_file(&path).ok();
        let (samples, sr) = result.unwrap();
        assert_eq!(sr, 48_000);
        assert_eq!(samples.len(), 4_800);
    }

    #[test]
    fn decode_to_mono_at_handles_upsample() {
        let path = write_test_wav(1, 8_000, 4_000);
        let result = decode_to_mono_at(&path, 16_000);
        std::fs::remove_file(&path).ok();
        let samples = result.unwrap();
        // 8k → 16k should give roughly 2× samples.
        assert!(
            (samples.len() as i64 - 8_000).abs() < 16,
            "upsampled len = {}",
            samples.len()
        );
    }

    #[test]
    fn capped_rejects_oversized_file_with_input_too_large() {
        let path = write_test_wav(1, 8_000, 8_000);
        let meta_len = std::fs::metadata(&path).unwrap().len();
        assert!(meta_len > 100, "expected a non-trivial wav, got {meta_len}");
        let err = decode_to_mono_limited(&path, DecodeLimits::bytes(100)).unwrap_err();
        std::fs::remove_file(&path).ok();
        match err {
            AfpError::InputTooLarge { limit, provided } => {
                assert_eq!(limit, 100);
                assert_eq!(provided, usize::try_from(meta_len).unwrap());
            }
            other => panic!("expected InputTooLarge, got {other:?}"),
        }
    }

    #[test]
    fn capped_accepts_file_under_byte_limit() {
        let path = write_test_wav(1, 8_000, 1_000);
        let meta_len = std::fs::metadata(&path).unwrap().len();
        let (samples, sr) = decode_to_mono_limited(&path, DecodeLimits::bytes(meta_len)).unwrap();
        std::fs::remove_file(&path).ok();
        assert_eq!(sr, 8_000);
        assert_eq!(samples.len(), 1_000);
    }

    #[test]
    fn limited_rejects_when_decoded_samples_exceed_cap() {
        let path = write_test_wav(1, 8_000, 4_000);
        let err = decode_to_mono_limited(&path, DecodeLimits::samples(100)).unwrap_err();
        std::fs::remove_file(&path).ok();
        assert!(
            matches!(err, AfpError::InputTooLarge { limit: 100, .. }),
            "got {err:?}"
        );
    }

    #[test]
    fn limited_both_caps_small_file_ok() {
        let path = write_test_wav(1, 8_000, 500);
        let meta_len = std::fs::metadata(&path).unwrap().len();
        let (samples, sr) =
            decode_to_mono_limited(&path, DecodeLimits::both(meta_len, 500)).unwrap();
        std::fs::remove_file(&path).ok();
        assert_eq!(sr, 8_000);
        assert_eq!(samples.len(), 500);
    }

    // -- integrity mode tests --

    /// Create a WAV file and corrupt some bytes in the data section.
    /// WAV header is 44 bytes for standard PCM; corrupting bytes well
    /// past that ensures we hit the data region, not the header.
    fn write_corrupt_wav(sr: u32, len: usize) -> std::path::PathBuf {
        use std::sync::atomic::{AtomicU64, Ordering};
        static COUNTER: AtomicU64 = AtomicU64::new(0);
        let n = COUNTER.fetch_add(1, Ordering::Relaxed);
        let path = std::env::temp_dir().join(format!(
            "audiofp-decoder-corrupt-{}-{}-{}.wav",
            std::process::id(),
            sr,
            n,
        ));
        let spec = hound::WavSpec {
            channels: 1,
            sample_rate: sr,
            bits_per_sample: 16,
            sample_format: hound::SampleFormat::Int,
        };
        let mut writer = hound::WavWriter::create(&path, spec).unwrap();
        let amp = (i16::MAX as f32) * 0.5;
        for i in 0..len {
            let s = libm::sinf(2.0 * PI * 440.0 * i as f32 / sr as f32) * amp;
            writer.write_sample(s as i16).unwrap();
        }
        writer.finalize().unwrap();

        // Corrupt a few bytes in the middle of the data section.
        // For a 16-bit mono WAV, data starts at byte 44. Corrupt a
        // chunk in the middle of the file.
        let file_len = std::fs::metadata(&path).unwrap().len() as usize;
        let mut bytes = std::fs::read(&path).unwrap();
        // Overwrite bytes near the middle with invalid patterns.
        let mid = file_len / 2;
        for i in 0..core::cmp::min(64, file_len - mid) {
            bytes[mid + i] = 0xFF;
        }
        std::fs::write(&path, &bytes).unwrap();
        path
    }

    #[test]
    fn strict_builder_sets_integrity_mode() {
        let limits = DecodeLimits::default().strict();
        assert!(limits.integrity_mode);

        let limits2 = DecodeLimits::both(1_000_000, 480_000).strict();
        assert!(limits2.integrity_mode);
        assert_eq!(limits2.max_bytes, 1_000_000);
        assert_eq!(limits2.max_samples, Some(480_000));
    }

    #[test]
    fn default_mode_skips_corrupt_packets() {
        // With default (non-strict) limits, corrupted WAV data packets
        // should be skipped and decoding should succeed (possibly with
        // fewer samples, but no error).
        let path = write_corrupt_wav(8_000, 8_000);
        let result = decode_to_mono_limited(&path, DecodeLimits::default());
        std::fs::remove_file(&path).ok();
        // WAV is a simple container: symphonia may or may not report a
        // per-packet decode error for corrupted PCM (it might just decode
        // the bytes as garbage audio). Either outcome (Ok or recoverable
        // skip) is acceptable for the default mode — the key invariant is
        // that it does NOT return an Io error with "decode integrity".
        match result {
            Ok(_) => {} // fine — corrupt PCM was decoded as-is or skipped
            Err(AfpError::Io(ref e)) if e.source.to_string().contains("decode integrity") => {
                panic!("default mode should NOT fail with integrity error: {e}");
            }
            Err(_) => {} // other errors (probe failure, etc.) are acceptable
        }
    }

    #[test]
    fn integrity_mode_fails_on_corrupt_packets() {
        // With integrity_mode=true, if Symphonia reports a per-packet
        // decode/IO error, the decode should fail.
        let path = write_corrupt_wav(8_000, 8_000);
        let limits = DecodeLimits::default().strict();
        let result = decode_to_mono_limited(&path, limits);
        std::fs::remove_file(&path).ok();
        // WAV PCM corruption may not always trigger a Symphonia DecodeError
        // (symphonia might just decode the garbage bytes). So this test
        // verifies the contract: IF an error is returned, it must be the
        // integrity error. If it succeeds, that's also fine (means
        // symphonia didn't detect corruption in the PCM stream).
        match result {
            Ok(_) => {
                // Symphonia decoded garbage as valid PCM — acceptable for
                // raw PCM WAV since there's no checksum. The integrity
                // check only fires when Symphonia itself raises an error.
            }
            Err(AfpError::Io(ref e)) if e.source.to_string().contains("decode integrity") => {
                // This is exactly what we want when corruption IS detected.
            }
            Err(other) => {
                // Other errors (e.g. probe failure if header was hit) are ok.
                let _ = other;
            }
        }
    }

    /// A more reliable test: corrupt the WAV header's format chunk to
    /// trigger a guaranteed codec-level error.
    #[test]
    fn integrity_mode_rejects_mangled_format() {
        let path = write_test_wav(1, 8_000, 8_000);
        // Mangle the "fmt " chunk by changing bits_per_sample (offset 34-35
        // in a standard WAV) to an absurd value.
        let mut bytes = std::fs::read(&path).unwrap();
        // Verify this is a RIFF WAV with "fmt " at offset 12.
        assert_eq!(&bytes[0..4], b"RIFF");
        assert_eq!(&bytes[8..12], b"WAVE");
        // bits_per_sample is at byte 34 in a standard 16-byte fmt chunk.
        // Set it to 0 to trigger a codec init failure.
        bytes[34] = 0;
        bytes[35] = 0;
        std::fs::write(&path, &bytes).unwrap();

        // With default mode — should fail with a codec/probe error, not
        // "decode integrity" since the codec can't even initialize.
        let result = decode_to_mono_limited(&path, DecodeLimits::default());
        assert!(result.is_err(), "mangled format should fail");

        std::fs::remove_file(&path).ok();
    }
}