mahbot 0.7.3

//! Container-level removal of AI-provenance metadata from generated image
//! artifacts, applied at the single point where they are written.
//!
//! Platforms read the container metadata a provider leaves on generated bytes to
//! auto-apply an AI label; [`strip_image_provenance`] splices exactly those
//! carriers out (the removal sets are the match arms below) and rewrites
//! nothing else:
//!
//! * Pixels stay bit-identical — nothing is re-encoded, resized, cropped or
//!   dithered. The only rewritten bytes are container framing: the CRC of a
//!   rebuilt `eXIf`, and the WebP RIFF size and `VP8X` feature flags.
//! * Only whole known carriers are dropped, so colour and rendering state
//!   survives by construction. EXIF orientation is rendering state, not
//!   provenance: a non-identity orientation is rebuilt as a payload carrying
//!   that tag alone, so the strip never rotates an image.
//! * Content-sniffed, never extension-based, and fail-open: what this pass
//!   cannot confidently parse is written untouched, and a pass that changes
//!   nothing reports that back, so re-running it is a no-op.
//!
//! Known gaps, both resolved in favour of leaving the artifact untouched: a WebP
//! whose final odd-length chunk omits its pad byte, and a PNG whose chunk CRCs do
//! not match the framing (the walk trusts lengths, not checksums).
//!
//! Deliberately not addressed:
//!
//! 1. The generator-side invisible watermark (e.g. SynthID), which the provider
//!    embeds in the pixels themselves.
//! 2. Disclosure obligations are unaffected — removing this metadata does not
//!    discharge any duty to disclose AI-generated content. Jurisdictional
//!    direction for intent: China prohibits removing or concealing AI labels
//!    *and* providing tools to do so, and India bars intermediaries from
//!    enabling removal/suppression. This strips the metadata on our own
//!    generated artifacts; it is not a general-purpose label-removal feature,
//!    and must not be extended into one.

use image::metadata::Orientation;
use std::borrow::Cow;

// ── Signatures ──────────────────────────────────────────────────────────

/// PNG file signature.
const PNG_SIGNATURE: &[u8; 8] = b"\x89PNG\r\n\x1a\n";

/// JPEG start-of-image marker and sniff signature: SOI plus the next marker's
/// introducer (a bare `FF D8` is not enough to claim a JPEG).
const JPEG_SOI: &[u8; 2] = &[0xFF, 0xD8];

/// RIFF/WEBP container magics (the `RIFF` size field sits between them).
const RIFF_MAGIC: &[u8; 4] = b"RIFF";
const WEBP_MAGIC: &[u8; 4] = b"WEBP";

/// JPEG marker bytes the walk classifies by name.
const JPEG_TEM: u8 = 0x01;
const JPEG_EOI: u8 = 0xD9;
const JPEG_SOS: u8 = 0xDA;
const JPEG_APP1: u8 = 0xE1;
const JPEG_APP11: u8 = 0xEB;
const JPEG_APP13: u8 = 0xED;
const JPEG_COM: u8 = 0xFE;

/// `Exif\0\0` prefix the JPEG/Exif spec puts in front of the TIFF block.
const EXIF_PREFIX: &[u8] = b"Exif\0\0";

/// `VP8X` feature-flag bits for the removed WebP carriers.
const VP8X_FLAG_EXIF: u8 = 0x08;
const VP8X_FLAG_XMP: u8 = 0x04;

// ── Entry point ─────────────────────────────────────────────────────────

/// Strip AI-provenance metadata from generated-artifact bytes.
///
/// Returns `Cow::Borrowed` when the bytes are not a confidently parsed
/// PNG/JPEG/WebP, carry no removal-set carrier, or would come back unchanged.
#[must_use]
pub(crate) fn strip_image_provenance(bytes: &[u8]) -> Cow<'_, [u8]> {
    let stripped = if bytes.starts_with(PNG_SIGNATURE) {
        strip_png(bytes)
    } else if bytes.starts_with(JPEG_SOI) && bytes.get(2) == Some(&0xFF) {
        strip_jpeg(bytes)
    } else if bytes.len() >= 12 && bytes.starts_with(RIFF_MAGIC) && &bytes[8..12] == WEBP_MAGIC {
        strip_webp(bytes)
    } else {
        None
    };

    match stripped {
        // Only bytes that actually changed are owned, so a caller can treat
        // `Cow::Owned` as "this artifact was normalized".
        Some(rewritten) if rewritten != bytes => Cow::Owned(rewritten),
        _ => Cow::Borrowed(bytes),
    }
}

// ── PNG ─────────────────────────────────────────────────────────────────

/// Walk a PNG, dropping the provenance carriers and rebuilding `eXIf` around a
/// preserved non-identity orientation.
///
/// `None` when the walk is not structurally confident (truncated or overrunning
/// chunk, first chunk not `IHDR`, no terminating zero-length `IEND`) or when
/// nothing in the removal set was found.
fn strip_png(bytes: &[u8]) -> Option<Vec<u8>> {
    let mut out = Vec::with_capacity(bytes.len());
    out.extend_from_slice(PNG_SIGNATURE);
    let mut pos = PNG_SIGNATURE.len();
    let mut removed = false;

    loop {
        let length = read_be_u32(bytes, pos)? as usize;
        let chunk_type = read_fourcc(bytes, pos.checked_add(4)?)?;
        let payload_start = pos.checked_add(8)?;
        let payload_end = payload_start.checked_add(length)?;
        let chunk_end = payload_end.checked_add(4)?;
        if chunk_end > bytes.len() {
            return None;
        }
        // The walk is only trusted when the stream opens with IHDR.
        if pos == PNG_SIGNATURE.len() && &chunk_type != b"IHDR" {
            return None;
        }

        match &chunk_type {
            b"IEND" => {
                if length != 0 {
                    return None;
                }
                out.extend_from_slice(&bytes[pos..chunk_end]);
                // Anything a producer left after IEND is copied verbatim.
                out.extend_from_slice(&bytes[chunk_end..]);
                return removed.then_some(out);
            }
            b"eXIf" => {
                removed = true;
                // A raw TIFF block: keep only a non-identity orientation.
                if let Some(minimal) =
                    minimal_exif_with_orientation(&bytes[payload_start..payload_end])
                {
                    push_png_chunk(&mut out, b"eXIf", &minimal);
                }
            }
            b"tEXt" | b"zTXt" | b"iTXt" | b"caBX" => removed = true,
            // IHDR, PLTE, IDAT, iCCP, gAMA, cHRM, sRGB, tRNS, pHYs, bKGD,
            // acTL/fcTL/fdAT and unknown chunks: verbatim, CRC included.
            _ => out.extend_from_slice(&bytes[pos..chunk_end]),
        }

        pos = chunk_end;
    }
}

/// Append one PNG chunk: big-endian length, four-byte type, payload and the
/// CRC-32 of type+payload (the length field counts the payload only).
fn push_png_chunk(out: &mut Vec<u8>, chunk_type: &[u8], payload: &[u8]) {
    let mut crc = flate2::Crc::new();
    crc.update(chunk_type);
    crc.update(payload);
    let length = u32::try_from(payload.len()).expect("rebuilt metadata fits a chunk length");
    out.extend_from_slice(&length.to_be_bytes());
    out.extend_from_slice(chunk_type);
    out.extend_from_slice(payload);
    out.extend_from_slice(&crc.sum().to_be_bytes());
}

// ── JPEG ────────────────────────────────────────────────────────────────

/// Walk a JPEG marker stream, dropping the provenance segments.
///
/// Only reached for bytes the sniffer already matched as a JPEG. `None` when the
/// framing is not confidently parsed (a stray `0x00` marker outside a scan, an
/// out-of-bounds length field, running out of bytes) or when nothing in the
/// removal set was found.
fn strip_jpeg(bytes: &[u8]) -> Option<Vec<u8>> {
    let mut out = Vec::with_capacity(bytes.len());
    out.extend_from_slice(JPEG_SOI);
    let mut pos = JPEG_SOI.len();
    let mut removed = false;

    loop {
        // Every position outside a scan starts with a marker introducer; 0xFF
        // fill bytes ahead of a marker are legal.
        if bytes.get(pos) != Some(&0xFF) {
            return None;
        }
        let marker_start = pos;
        while bytes.get(pos) == Some(&0xFF) {
            pos += 1;
        }
        let marker = *bytes.get(pos)?;
        pos += 1;
        if marker == 0x00 {
            return None;
        }

        match marker {
            JPEG_EOI => {
                // EOI plus any trailer, verbatim.
                out.extend_from_slice(&bytes[marker_start..]);
                return removed.then_some(out);
            }
            JPEG_TEM | 0xD0..=0xD7 => out.extend_from_slice(&bytes[marker_start..pos]),
            JPEG_SOS => {
                let length = usize::from(read_be_u16(bytes, pos)?);
                if length < 2 {
                    return None;
                }
                let scan_start = pos.checked_add(length)?;
                if scan_start > bytes.len() {
                    return None;
                }
                out.extend_from_slice(&bytes[marker_start..scan_start]);

                // Entropy-coded data: byte stuffing (`FF 00`), restart markers
                // (`FF D0..=FF D7`) and runs of `FF` fill bytes belong to the
                // scan; the scan ends at `FF` followed by any other byte. A
                // progressive JPEG simply re-enters this loop per scan.
                let mut scan_pos = scan_start;
                let scan_end = loop {
                    let i = scan_pos + bytes.get(scan_pos..)?.iter().position(|&b| b == 0xFF)?;
                    match bytes.get(i + 1) {
                        None => return None,
                        Some(&next)
                            if next == 0x00 || next == 0xFF || (0xD0..=0xD7).contains(&next) =>
                        {
                            scan_pos = i + 1;
                        }
                        Some(_) => break i,
                    }
                };
                out.extend_from_slice(&bytes[scan_start..scan_end]);
                pos = scan_end;
            }
            _ => {
                let length = usize::from(read_be_u16(bytes, pos)?);
                if length < 2 {
                    return None;
                }
                let payload_start = pos.checked_add(2)?;
                let payload_end = pos.checked_add(length)?;
                if payload_end > bytes.len() {
                    return None;
                }
                let payload = &bytes[payload_start..payload_end];

                match marker {
                    // APP1 carries only metadata — EXIF, XMP/ExtendedXMP and rare
                    // vendor variants; the rendering-relevant side data of a JPEG
                    // lives elsewhere (JFIF APP0, ICC APP2, Adobe APP14, the
                    // ISO 21496-1 gain map in APP2). A prefixed EXIF block with a
                    // non-identity orientation is rebuilt around that tag alone,
                    // every other APP1 is dropped.
                    JPEG_APP1 => {
                        removed = true;
                        if payload.starts_with(EXIF_PREFIX)
                            && let Some(minimal) = minimal_exif_with_orientation(payload)
                        {
                            push_jpeg_segment(&mut out, marker, &minimal);
                        }
                    }
                    // JUMBF/C2PA only; other APP11 payloads (e.g. JPEG-HDR).
                    JPEG_APP11 if payload.starts_with(b"JP") => removed = true,
                    // The Photoshop IRB carrying IPTC; `Adobe_CM` is kept.
                    JPEG_APP13
                        if payload.starts_with(b"Photoshop 3.0\0")
                            || payload.starts_with(b"Adobe_Photoshop2.5") =>
                    {
                        removed = true;
                    }
                    JPEG_COM => removed = true,
                    // JFIF APP0, ICC APP2, Adobe APP14, DQT/DHT/SOF/DRI/... and
                    // unknown segments: verbatim, length field included.
                    _ => out.extend_from_slice(&bytes[marker_start..payload_end]),
                }

                pos = payload_end;
            }
        }
    }
}

/// Append one JPEG segment (`FF <marker>` plus a big-endian length that counts
/// itself).
fn push_jpeg_segment(out: &mut Vec<u8>, marker: u8, payload: &[u8]) {
    out.push(0xFF);
    out.push(marker);
    let length = u16::try_from(payload.len() + 2).expect("rebuilt metadata fits a segment length");
    out.extend_from_slice(&length.to_be_bytes());
    out.extend_from_slice(payload);
}

// ── WebP ────────────────────────────────────────────────────────────────

/// Walk a RIFF/WEBP container, dropping the provenance chunks.
///
/// Only reached for bytes the sniffer already matched as a WebP.
///
/// One pass, with the `VP8X` feature flags patched at the end: the flags mirror
/// which carriers survived, and the spec puts `VP8X` ahead of the metadata it
/// describes. `None` when the framing is not confidently parsed or nothing in
/// the removal set was found.
fn strip_webp(bytes: &[u8]) -> Option<Vec<u8>> {
    let riff_end = (read_le_u32(bytes, 4)? as usize).checked_add(8)?;
    if riff_end > bytes.len() {
        return None;
    }

    let mut out = Vec::with_capacity(bytes.len());
    out.extend_from_slice(RIFF_MAGIC);
    out.extend_from_slice(&[0; 4]); // RIFF size, patched below
    out.extend_from_slice(WEBP_MAGIC);

    // Offset of the `VP8X` flags byte in `out`, and the carriers seen.
    let mut flags_at = None;
    let (mut saw_exif, mut saw_xmp, mut saw_c2pa) = (false, false, false);
    let mut kept_exif = false;
    let mut pos = 12;
    while pos < riff_end {
        let (fourcc, payload_start, payload_end, chunk_end) =
            read_webp_chunk(bytes, pos, riff_end)?;
        match &fourcc {
            b"EXIF" => {
                saw_exif = true;
                if let Some(minimal) =
                    minimal_exif_with_orientation(&bytes[payload_start..payload_end])
                {
                    kept_exif = true;
                    push_webp_chunk(&mut out, b"EXIF", &minimal);
                }
            }
            b"XMP " => saw_xmp = true,
            b"C2PA" => saw_c2pa = true,
            b"VP8X" => {
                // Only a `VP8X` that actually carries its flags byte is patched;
                // a zero-length one has nothing to patch and must not be mistaken
                // for the chunk that follows it.
                if payload_end > payload_start {
                    flags_at = Some(out.len() + 8);
                }
                out.extend_from_slice(&bytes[pos..chunk_end]);
            }
            // `VP8 `, `VP8L`, `ALPH`, `ICCP`, `ANIM`, `ANMF`, unknown chunks.
            _ => out.extend_from_slice(&bytes[pos..chunk_end]),
        }
        pos = chunk_end;
    }
    if !(saw_exif || saw_xmp || saw_c2pa) {
        return None;
    }

    if let Some(flags) = flags_at.and_then(|at| out.get_mut(at)) {
        // XMP never survives; EXIF survives only as the rebuilt orientation
        // payload. Every other feature bit is left alone.
        *flags &= !(VP8X_FLAG_EXIF | VP8X_FLAG_XMP);
        if kept_exif {
            *flags |= VP8X_FLAG_EXIF;
        }
    }

    // The declared size covers the rebuilt chunks only: bytes a producer left
    // after the original RIFF end stay an undeclared trailer, exactly as they
    // were, so the framing we emit matches the input's.
    let riff_size = u32::try_from(out.len().checked_sub(8)?).ok()?;
    out[4..8].copy_from_slice(&riff_size.to_le_bytes());
    out.extend_from_slice(&bytes[riff_end..]);
    Some(out)
}

/// Parse the RIFF chunk starting at `pos`, returning its fourcc, payload start,
/// payload end and padded chunk end. `None` on any overrun past `riff_end`.
fn read_webp_chunk(
    bytes: &[u8],
    pos: usize,
    riff_end: usize,
) -> Option<([u8; 4], usize, usize, usize)> {
    let fourcc = read_fourcc(bytes, pos)?;
    let size = read_le_u32(bytes, pos.checked_add(4)?)? as usize;
    let payload_start = pos.checked_add(8)?;
    let payload_end = payload_start.checked_add(size)?;
    let chunk_end = payload_end.checked_add(size & 1)?;
    if chunk_end > riff_end {
        return None;
    }
    Some((fourcc, payload_start, payload_end, chunk_end))
}

/// Append one RIFF chunk: four-byte id, little-endian payload size, payload and
/// the mandatory pad byte when the payload length is odd.
fn push_webp_chunk(out: &mut Vec<u8>, fourcc: &[u8], payload: &[u8]) {
    let size = u32::try_from(payload.len()).expect("rebuilt metadata fits a chunk length");
    out.extend_from_slice(fourcc);
    out.extend_from_slice(&size.to_le_bytes());
    out.extend_from_slice(payload);
    if size & 1 != 0 {
        out.push(0);
    }
}

// ── EXIF orientation ────────────────────────────────────────────────────

/// The payload rebuilt to carry nothing but the EXIF orientation, in the same
/// container form as `payload` (the `Exif\0\0` prefix is preserved when present).
/// `None` when `payload` has no parseable non-identity orientation; every other
/// tag is dropped, since orientation is the only one that affects rendering.
///
/// The tag is read with the `image` crate's parser — the one the consumers decode
/// through (`util::exif_orientation`) — so the rule cannot drift from theirs.
fn minimal_exif_with_orientation(payload: &[u8]) -> Option<Vec<u8>> {
    let tiff = payload.strip_prefix(EXIF_PREFIX).unwrap_or(payload);
    let orientation = Orientation::from_exif_chunk(tiff)?;
    if orientation == Orientation::NoTransforms {
        return None;
    }
    let orientation = orientation.to_exif();

    let prefixed = payload.starts_with(EXIF_PREFIX);
    let mut out = Vec::with_capacity(if prefixed { 32 } else { 26 });
    if prefixed {
        out.extend_from_slice(EXIF_PREFIX);
    }
    out.extend_from_slice(b"II\x2A\x00"); // little-endian TIFF magic
    out.extend_from_slice(&8u32.to_le_bytes()); // IFD0 offset
    out.extend_from_slice(&1u16.to_le_bytes()); // one entry
    out.extend_from_slice(&[0x12, 0x01, 0x03, 0x00]); // tag 0x0112, type SHORT
    out.extend_from_slice(&1u32.to_le_bytes()); // count 1
    out.extend_from_slice(&[orientation, 0x00, 0x00, 0x00]); // value
    out.extend_from_slice(&[0x00, 0x00, 0x00, 0x00]); // no next IFD
    Some(out)
}

// ── Primitive reads ─────────────────────────────────────────────────────

/// Four-byte identifier at `offset`.
fn read_fourcc(bytes: &[u8], offset: usize) -> Option<[u8; 4]> {
    let end = offset.checked_add(4)?;
    bytes.get(offset..end)?.try_into().ok()
}

/// Big-endian `u16` at `offset`.
fn read_be_u16(bytes: &[u8], offset: usize) -> Option<u16> {
    let end = offset.checked_add(2)?;
    let raw: [u8; 2] = bytes.get(offset..end)?.try_into().ok()?;
    Some(u16::from_be_bytes(raw))
}

/// Big-endian `u32` at `offset`.
fn read_be_u32(bytes: &[u8], offset: usize) -> Option<u32> {
    let raw: [u8; 4] = bytes.get(offset..offset.checked_add(4)?)?.try_into().ok()?;
    Some(u32::from_be_bytes(raw))
}

/// Little-endian `u32` at `offset`.
fn read_le_u32(bytes: &[u8], offset: usize) -> Option<u32> {
    let raw: [u8; 4] = bytes.get(offset..offset.checked_add(4)?)?.try_into().ok()?;
    Some(u32::from_le_bytes(raw))
}

#[cfg(test)]
mod tests {
    use super::*;
    use image::ImageEncoder as _;

    // ── Fixtures ────────────────────────────────────────────────────────

    /// A 2×2 opaque RGBA image with four distinct pixels.
    fn sample_rgba() -> image::RgbaImage {
        image::RgbaImage::from_raw(
            2,
            2,
            vec![
                255, 0, 0, 255, // red
                0, 255, 0, 255, // green
                0, 0, 255, 255, // blue
                255, 255, 0, 255, // yellow
            ],
        )
        .expect("valid 2x2 RGBA buffer")
    }

    /// The same four pixels without an alpha channel (JPEG/WebP fixtures).
    fn sample_rgb() -> image::RgbImage {
        image::RgbImage::from_raw(
            2,
            2,
            vec![
                255, 0, 0, // red
                0, 255, 0, // green
                0, 0, 255, // blue
                255, 255, 0, // yellow
            ],
        )
        .expect("valid 2x2 RGB buffer")
    }

    /// Decoded RGBA pixels of an image in any supported container.
    fn rgba_pixels(bytes: &[u8]) -> Vec<u8> {
        image::load_from_memory(bytes)
            .expect("fixture must decode")
            .to_rgba8()
            .into_raw()
    }

    /// The rewritten bytes of a fixture that must be normalized.
    fn rewritten(bytes: &[u8]) -> Vec<u8> {
        match strip_image_provenance(bytes) {
            Cow::Owned(out) => out,
            Cow::Borrowed(_) => panic!("a metadata-bearing fixture must be rewritten"),
        }
    }

    fn encode_png(img: &image::RgbaImage) -> Vec<u8> {
        let mut out = Vec::new();
        image::codecs::png::PngEncoder::new(&mut out)
            .write_image(
                img.as_raw(),
                img.width(),
                img.height(),
                image::ExtendedColorType::Rgba8,
            )
            .expect("PNG encode");
        out
    }

    fn encode_jpeg(img: &image::RgbImage) -> Vec<u8> {
        let mut out = Vec::new();
        image::codecs::jpeg::JpegEncoder::new(&mut out)
            .encode_image(&image::DynamicImage::ImageRgb8(img.clone()))
            .expect("JPEG encode");
        out
    }

    fn encode_webp_lossless(img: &image::RgbImage) -> Vec<u8> {
        let mut out = Vec::new();
        image::codecs::webp::WebPEncoder::new_lossless(&mut out)
            .encode(
                img.as_raw(),
                img.width(),
                img.height(),
                image::ExtendedColorType::Rgb8,
            )
            .expect("WebP encode");
        out
    }

    /// Splice extra chunks in right after IHDR (always 25 bytes including the
    /// length, type and CRC).
    fn png_with_extra_chunks(base: &[u8], extra: &[u8]) -> Vec<u8> {
        assert_eq!(&base[12..16], b"IHDR");
        let mut out = base[..33].to_vec();
        out.extend_from_slice(extra);
        out.extend_from_slice(&base[33..]);
        out
    }

    /// A PNG carrying every removal-set chunk plus colour state that must
    /// survive.
    fn png_with_metadata() -> Vec<u8> {
        let base = encode_png(&sample_rgba());
        let mut extra = Vec::new();
        push_png_chunk(&mut extra, b"gAMA", &[0x00, 0x00, 0xB1, 0x8F]);
        push_png_chunk(&mut extra, b"iCCP", b"icc\0\0not-a-zlib-stream");
        push_png_chunk(
            &mut extra,
            b"zTXt",
            b"Raw profile type iptc\0\0not-a-zlib-stream",
        );
        push_png_chunk(
            &mut extra,
            b"iTXt",
            b"XML:com.adobe.xmp\0\x00\x00\0\0<x:xmpmeta/>",
        );
        push_png_chunk(&mut extra, b"caBX", b"jumb\0\x10\0\0c2pa\x11manifest");
        png_with_extra_chunks(&base, &extra)
    }

    /// Chunk types of a PNG, in file order.
    fn png_chunk_types(bytes: &[u8]) -> Vec<[u8; 4]> {
        let mut types = Vec::new();
        let mut pos = 8;
        while pos + 12 <= bytes.len() {
            let length = u32::from_be_bytes(bytes[pos..pos + 4].try_into().unwrap()) as usize;
            types.push(bytes[pos + 4..pos + 8].try_into().unwrap());
            pos += 12 + length;
        }
        types
    }

    /// The `Make` (`0x010F`) value in `exif_with_orientation_and_make_tag`, which
    /// the rebuild must not carry over.
    const MAKE_TAG_VALUE: &[u8] = b"ABC\0";

    /// Whether a rebuilt payload still carries the fixture's `Make` tag value.
    fn carries_make_tag(payload: &[u8]) -> bool {
        payload
            .windows(MAKE_TAG_VALUE.len())
            .any(|w| w == MAKE_TAG_VALUE)
    }

    /// A raw-TIFF EXIF block with an orientation entry plus a `Make` (`0x010F`)
    /// entry that the rebuild must not carry over.
    fn exif_with_orientation_and_make_tag(orientation: u8) -> Vec<u8> {
        let mut out = b"II*\0".to_vec();
        out.extend_from_slice(&8u32.to_le_bytes()); // IFD0 offset
        out.extend_from_slice(&2u16.to_le_bytes()); // two entries
        out.extend_from_slice(&[0x12, 0x01, 0x03, 0x00]); // tag 0x0112, SHORT
        out.extend_from_slice(&[0x01, 0x00, 0x00, 0x00]); // count 1
        out.extend_from_slice(&[orientation, 0x00, 0x00, 0x00]); // value
        out.extend_from_slice(&[0x0F, 0x01, 0x02, 0x00]); // tag 0x010F, ASCII
        out.extend_from_slice(&[0x04, 0x00, 0x00, 0x00]); // count 4
        out.extend_from_slice(MAKE_TAG_VALUE); // inline value
        out.extend_from_slice(&[0x00, 0x00, 0x00, 0x00]); // no next IFD
        out
    }

    /// Splice raw segments in right after SOI.
    fn jpeg_with_segments(base: &[u8], segments: &[Vec<u8>]) -> Vec<u8> {
        assert!(base.starts_with(JPEG_SOI));
        let mut out = base[..2].to_vec();
        for segment in segments {
            out.extend_from_slice(segment);
        }
        out.extend_from_slice(&base[2..]);
        out
    }

    fn jpeg_segment(marker: u8, payload: &[u8]) -> Vec<u8> {
        let mut out = Vec::new();
        push_jpeg_segment(&mut out, marker, payload);
        out
    }

    /// The APPn/COM segments preceding the first scan, as `(marker, payload)`.
    fn jpeg_header_segments(bytes: &[u8]) -> Vec<(u8, Vec<u8>)> {
        let mut segments = Vec::new();
        let mut pos = 2;
        while bytes.get(pos) == Some(&0xFF) {
            while bytes.get(pos) == Some(&0xFF) {
                pos += 1;
            }
            let Some(&marker) = bytes.get(pos) else {
                break;
            };
            pos += 1;
            if marker == JPEG_SOS || marker == JPEG_EOI {
                break;
            }
            if marker == JPEG_TEM || (0xD0..=0xD7).contains(&marker) {
                continue;
            }
            let Some(length) = read_be_u16(bytes, pos) else {
                break;
            };
            let end = pos + usize::from(length);
            segments.push((marker, bytes[pos + 2..end].to_vec()));
            pos = end;
        }
        segments
    }

    fn has_segment(segments: &[(u8, Vec<u8>)], marker: u8, prefix: &[u8]) -> bool {
        segments
            .iter()
            .any(|(m, payload)| *m == marker && payload.starts_with(prefix))
    }

    /// Chunks of a RIFF/WEBP container, as `(fourcc, payload)`.
    fn webp_chunks(bytes: &[u8]) -> Vec<([u8; 4], Vec<u8>)> {
        let mut chunks = Vec::new();
        let riff_end = u32::from_le_bytes(bytes[4..8].try_into().unwrap()) as usize + 8;
        let mut pos = 12;
        while pos < riff_end {
            let fourcc = bytes[pos..pos + 4].try_into().unwrap();
            let size = u32::from_le_bytes(bytes[pos + 4..pos + 8].try_into().unwrap()) as usize;
            chunks.push((fourcc, bytes[pos + 8..pos + 8 + size].to_vec()));
            pos += 8 + size + (size & 1);
        }
        chunks
    }

    #[expect(
        clippy::trivially_copy_pass_by_ref,
        reason = "call sites pass fourcc literals, which read better by reference"
    )]
    fn webp_chunk<'a>(chunks: &'a [([u8; 4], Vec<u8>)], fourcc: &[u8; 4]) -> Option<&'a [u8]> {
        chunks
            .iter()
            .find(|(id, _)| id == fourcc)
            .map(|(_, payload)| payload.as_slice())
    }

    /// A `VP8X` payload: flags, three reserved bytes, canvas dimensions.
    fn vp8x_payload(flags: u8, width: u32, height: u32) -> Vec<u8> {
        let mut payload = vec![flags, 0, 0, 0];
        payload.extend_from_slice(&(width - 1).to_le_bytes()[..3]);
        payload.extend_from_slice(&(height - 1).to_le_bytes()[..3]);
        payload
    }

    fn write_riff_size(bytes: &mut [u8]) {
        let size = u32::try_from(bytes.len() - 8).expect("fixture fits u32");
        bytes[4..8].copy_from_slice(&size.to_le_bytes());
    }

    /// The `VP8L` payload the lossless encoder emits for `img`.
    fn webp_vp8l(img: &image::RgbImage) -> Vec<u8> {
        webp_chunks(&encode_webp_lossless(img))
            .into_iter()
            .find(|(fourcc, _)| fourcc == b"VP8L")
            .map(|(_, payload)| payload)
            .expect("lossless encoder emits a VP8L chunk")
    }

    /// A RIFF/WEBP fixture: `VP8X` with `flags`, the `VP8L` payload, then `extra`
    /// chunks, with the RIFF size field written.
    fn webp_fixture(
        img: &image::RgbImage,
        vp8l: &[u8],
        flags: u8,
        extra: &[(&[u8], &[u8])],
    ) -> Vec<u8> {
        let mut fixture = Vec::new();
        fixture.extend_from_slice(RIFF_MAGIC);
        fixture.extend_from_slice(&[0; 4]);
        fixture.extend_from_slice(WEBP_MAGIC);
        push_webp_chunk(
            &mut fixture,
            b"VP8X",
            &vp8x_payload(flags, img.width(), img.height()),
        );
        push_webp_chunk(&mut fixture, b"VP8L", vp8l);
        for (fourcc, payload) in extra {
            push_webp_chunk(&mut fixture, fourcc, payload);
        }
        write_riff_size(&mut fixture);
        fixture
    }

    // ── PNG ─────────────────────────────────────────────────────────────

    #[test]
    fn png_drops_metadata_chunks_and_keeps_rendering_state() {
        let base = encode_png(&sample_rgba());
        let fixture = png_with_metadata();
        assert_eq!(rgba_pixels(&fixture), rgba_pixels(&base));

        let out = rewritten(&fixture);

        let types = png_chunk_types(&out);
        assert!(types.contains(b"IHDR"));
        assert!(types.contains(b"IDAT"));
        assert!(types.contains(b"IEND"));
        assert!(types.contains(b"gAMA"));
        assert!(types.contains(b"iCCP"));
        for gone in [b"zTXt", b"iTXt", b"caBX"] {
            assert!(!types.contains(gone), "{gone:?} must be dropped");
        }
        assert_eq!(rgba_pixels(&out), rgba_pixels(&base));
    }

    #[test]
    fn png_keeps_non_identity_orientation_and_drops_the_rest() {
        let base = encode_png(&sample_rgba());
        let mut extra = Vec::new();
        push_png_chunk(&mut extra, b"eXIf", &exif_with_orientation_and_make_tag(6));
        let fixture = png_with_extra_chunks(&base, &extra);
        assert_eq!(rgba_pixels(&fixture), rgba_pixels(&base));

        let out = rewritten(&fixture);
        let types = png_chunk_types(&out);
        assert!(types.contains(b"eXIf"));
        assert_eq!(
            crate::util::exif_orientation(&out).map(Orientation::to_exif),
            Some(6)
        );
        assert_eq!(rgba_pixels(&out), rgba_pixels(&base));

        // Orientation 1 (identity) and an unparsable eXIf are dropped outright.
        for payload in [
            exif_with_orientation_and_make_tag(1),
            b"not a tiff block".to_vec(),
        ] {
            let mut extra = Vec::new();
            push_png_chunk(&mut extra, b"eXIf", &payload);
            let fixture = png_with_extra_chunks(&base, &extra);
            let out = rewritten(&fixture);
            assert!(!png_chunk_types(&out).contains(b"eXIf"));
            assert_eq!(rgba_pixels(&out), rgba_pixels(&base));
        }
    }

    // ── JPEG ────────────────────────────────────────────────────────────

    #[test]
    fn jpeg_drops_metadata_segments_and_keeps_rendering_state() {
        let base = encode_jpeg(&sample_rgb());
        let fixture = jpeg_with_segments(
            &base,
            &[
                jpeg_segment(0xE1, b"http://ns.adobe.com/xap/1.0/\0<x:xmpmeta/>"),
                jpeg_segment(0xED, b"Photoshop 3.0\0iptc"),
                jpeg_segment(0xEB, b"JP\0\x01\0\0jumb-c2pa"),
                jpeg_segment(JPEG_COM, b"generated by a model"),
                jpeg_segment(0xE2, b"ICC_PROFILE\0\x01\x01fake-icc-bytes"),
                // Standard 12-byte Adobe APP14 (transform 1 = YCbCr).
                jpeg_segment(0xEE, b"Adobe\x00\x64\x00\x00\x00\x00\x01"),
            ],
        );
        assert_eq!(rgba_pixels(&fixture), rgba_pixels(&base));

        let out = rewritten(&fixture);

        let segments = jpeg_header_segments(&out);
        assert!(has_segment(&segments, 0xE0, b"JFIF\0"));
        assert!(has_segment(&segments, 0xE2, b"ICC_PROFILE\0"));
        assert!(has_segment(&segments, 0xEE, b"Adobe"));
        assert!(!has_segment(
            &segments,
            0xE1,
            b"http://ns.adobe.com/xap/1.0/"
        ));
        assert!(!has_segment(&segments, 0xED, b"Photoshop 3.0\0"));
        assert!(!has_segment(&segments, 0xEB, b"JP"));
        assert!(!segments.iter().any(|(marker, _)| *marker == JPEG_COM));
        assert_eq!(rgba_pixels(&out), rgba_pixels(&base));

        // A truncated stream cannot be walked confidently: pass through.
        let truncated = &base[..base.len() / 2];
        assert_eq!(strip_image_provenance(truncated), truncated);
    }

    #[test]
    fn jpeg_keeps_non_identity_orientation_and_drops_the_rest() {
        let base = encode_jpeg(&sample_rgb());
        let mut exif = EXIF_PREFIX.to_vec();
        exif.extend_from_slice(&exif_with_orientation_and_make_tag(6));
        let fixture = jpeg_with_segments(&base, &[jpeg_segment(JPEG_APP1, &exif)]);

        let out = rewritten(&fixture);
        let segments = jpeg_header_segments(&out);
        assert_eq!(
            crate::util::exif_orientation(&out).map(Orientation::to_exif),
            Some(6)
        );
        let app1: Vec<_> = segments
            .iter()
            .filter(|(marker, _)| *marker == JPEG_APP1)
            .collect();
        assert_eq!(app1.len(), 1);
        // Only the orientation tag survives, in the prefixed container form.
        assert!(app1[0].1.starts_with(EXIF_PREFIX));
        assert!(!carries_make_tag(&app1[0].1));

        // Orientation 1 leaves nothing to preserve: the APP1 is gone.
        let mut exif = EXIF_PREFIX.to_vec();
        exif.extend_from_slice(&exif_with_orientation_and_make_tag(1));
        let fixture = jpeg_with_segments(&base, &[jpeg_segment(JPEG_APP1, &exif)]);
        let out = rewritten(&fixture);
        assert!(
            !jpeg_header_segments(&out)
                .iter()
                .any(|(marker, _)| *marker == JPEG_APP1)
        );
        assert_eq!(rgba_pixels(&out), rgba_pixels(&base));
    }

    #[test]
    fn jpeg_copies_entropy_data_and_scans_verbatim() {
        // A progressive-style skeleton (two scans, restart markers, `FF 00`
        // stuffing, an `FF FF` fill run, a trailer after EOI) that never decodes:
        // the walk is asserted byte-for-byte, which is the strongest statement
        // that entropy-coded data survives — and the comment flanking each scan
        // only disappears if the walk really re-enters the marker loop there.
        let soi = JPEG_SOI.to_vec();
        let head_com = jpeg_segment(JPEG_COM, b"generated by a model");
        let dqt = jpeg_segment(0xDB, b"quant tables");
        let sos_1 = jpeg_segment(JPEG_SOS, b"scan-1 header");
        let scan_1 = vec![
            0x01, 0xFF, 0x00, 0x02, 0xFF, 0xFF, 0x00, 0x03, 0xFF, 0xD0, 0x04, 0xFF, 0xD3, 0x05,
        ];
        let mid_com = jpeg_segment(JPEG_COM, b"comment between scans");
        let dht = jpeg_segment(0xC4, b"huffman tables");
        let sos_2 = jpeg_segment(JPEG_SOS, b"scan-2 header");
        let scan_2 = vec![0x06, 0xFF, 0x00, 0x07, 0xFF, 0x00];
        let tail = vec![0xFF, 0xD9];
        let trailer = b"TRAILER".to_vec();

        let parts: [&[u8]; 11] = [
            &soi, &head_com, &dqt, &sos_1, &scan_1, &mid_com, &dht, &sos_2, &scan_2, &tail,
            &trailer,
        ];
        let kept: [&[u8]; 9] = [
            &soi, &dqt, &sos_1, &scan_1, &dht, &sos_2, &scan_2, &tail, &trailer,
        ];
        let fixture = parts.concat();
        let expected = kept.concat();

        let out = rewritten(&fixture);
        assert_eq!(out, expected);
    }

    // ── WebP ────────────────────────────────────────────────────────────

    #[test]
    fn webp_drops_metadata_chunks_and_patches_vp8x_flags() {
        let img = sample_rgb();
        let encoded = encode_webp_lossless(&img);
        let vp8l = webp_vp8l(&img);

        // ICC + EXIF + XMP flags set: the two dropped carriers must clear,
        // ICC must survive.
        let fixture = webp_fixture(
            &img,
            &vp8l,
            0x2C,
            &[
                (b"EXIF", &exif_with_orientation_and_make_tag(1)),
                (b"XMP ", b"<x:xmpmeta/>"),
                (b"C2PA", b"jumb\0\x10\0\0c2pa"),
            ],
        );
        assert_eq!(rgba_pixels(&fixture), rgba_pixels(&encoded));

        let out = rewritten(&fixture);

        let chunks = webp_chunks(&out);
        assert_eq!(webp_chunk(&chunks, b"VP8L"), Some(vp8l.as_slice()));
        assert_eq!(webp_chunk(&chunks, b"EXIF"), None);
        assert_eq!(webp_chunk(&chunks, b"XMP "), None);
        assert_eq!(webp_chunk(&chunks, b"C2PA"), None);
        assert_eq!(webp_chunk(&chunks, b"VP8X").unwrap()[0], 0x20);
        assert_eq!(
            out.len() - 8,
            u32::from_le_bytes(out[4..8].try_into().unwrap()) as usize
        );
        assert_eq!(rgba_pixels(&out), rgba_pixels(&encoded));

        // A `VP8X` advertising carriers the container does not hold: the output's
        // flags describe the output, so both bits clear.
        let fixture = webp_fixture(&img, &vp8l, 0x0C, &[(b"C2PA", b"jumb\0\x10\0\0c2pa")]);
        let out = rewritten(&fixture);
        assert_eq!(webp_chunk(&webp_chunks(&out), b"VP8X").unwrap()[0], 0x00);
    }

    #[test]
    fn webp_keeps_non_identity_orientation_and_sets_the_flag() {
        let img = sample_rgb();
        let encoded = encode_webp_lossless(&img);
        let vp8l = webp_vp8l(&img);

        // ICC + XMP flags set and the EXIF flag unset even though an `EXIF` chunk
        // is present: keeping the orientation must also set the flag.
        let fixture = webp_fixture(
            &img,
            &vp8l,
            0x24,
            &[
                (b"EXIF", &exif_with_orientation_and_make_tag(6)),
                (b"XMP ", b"<x:xmpmeta/>"),
            ],
        );
        assert_eq!(rgba_pixels(&fixture), rgba_pixels(&encoded));

        let out = rewritten(&fixture);

        let chunks = webp_chunks(&out);
        assert_eq!(
            crate::util::exif_orientation(&out).map(Orientation::to_exif),
            Some(6)
        );
        // Only the orientation tag survives, in the chunk's raw-TIFF form.
        let exif = webp_chunk(&chunks, b"EXIF").unwrap();
        assert!(!exif.starts_with(EXIF_PREFIX));
        assert!(!carries_make_tag(exif));
        assert_eq!(webp_chunk(&chunks, b"XMP "), None);
        assert_eq!(webp_chunk(&chunks, b"VP8X").unwrap()[0], 0x28);
        assert_eq!(rgba_pixels(&out), rgba_pixels(&encoded));
    }

    #[test]
    fn webp_leaves_a_vp8x_without_a_flags_byte_unpatched() {
        // A zero-length `VP8X` has no flags byte to patch, and the chunk that
        // follows it must not be mistaken for one (which silently corrupts its
        // fourcc).
        let img = sample_rgb();
        let vp8l = webp_vp8l(&img);
        let mut fixture = Vec::new();
        fixture.extend_from_slice(RIFF_MAGIC);
        fixture.extend_from_slice(&[0; 4]);
        fixture.extend_from_slice(WEBP_MAGIC);
        push_webp_chunk(&mut fixture, b"VP8X", &[]);
        push_webp_chunk(&mut fixture, b"VP8L", &vp8l);
        push_webp_chunk(&mut fixture, b"XMP ", b"<x:xmpmeta/>");
        write_riff_size(&mut fixture);

        let out = rewritten(&fixture);
        let chunks = webp_chunks(&out);
        assert_eq!(webp_chunk(&chunks, b"VP8L"), Some(vp8l.as_slice()));
        assert_eq!(webp_chunk(&chunks, b"XMP "), None);
    }

    #[test]
    fn webp_leaves_a_trailer_outside_the_declared_size() {
        let img = sample_rgb();
        let encoded = encode_webp_lossless(&img);
        let mut fixture = webp_fixture(&img, &webp_vp8l(&img), 0x20, &[(b"XMP ", b"<x:xmpmeta/>")]);
        let trailer = b"TRAILER";
        fixture.extend_from_slice(trailer);

        let out = rewritten(&fixture);

        // The trailer stays undeclared, exactly as the input framed it.
        assert!(out.ends_with(trailer));
        let declared = u32::from_le_bytes(out[4..8].try_into().unwrap()) as usize;
        assert_eq!(declared, out.len() - 8 - trailer.len());
        assert_eq!(rgba_pixels(&out), rgba_pixels(&encoded));
    }

    // ── Pass-through and idempotency ────────────────────────────────────

    #[test]
    fn non_images_and_unparsable_payloads_pass_through_unchanged() {
        let png = encode_png(&sample_rgba());
        let mp4 = b"\0\0\0\x18ftypmp42\0\0\0\0mp42isom avc1".as_slice();
        let cases: [&[u8]; 3] = [
            b"<svg xmlns=\"http://www.w3.org/2000/svg\"><rect/></svg>",
            mp4,
            &png[..20], // truncated PNG: signature only
        ];
        for bytes in cases {
            let stripped = strip_image_provenance(bytes);
            assert!(matches!(stripped, Cow::Borrowed(_)));
            assert_eq!(stripped.as_ref(), bytes);
        }

        // A second pass has nothing left to remove and says so — including for a
        // rebuilt orientation payload, which the first pass re-emits byte-for-byte.
        let mut extra = Vec::new();
        push_png_chunk(&mut extra, b"eXIf", &exif_with_orientation_and_make_tag(6));
        for fixture in [
            png_with_metadata(),
            png_with_extra_chunks(&encode_png(&sample_rgba()), &extra),
        ] {
            let stripped = rewritten(&fixture);
            assert!(matches!(
                strip_image_provenance(&stripped),
                Cow::Borrowed(_)
            ));
        }
    }
}