Files
guara/crates/guara-core/src/scan.rs
T
Lucas RayanandClaude Opus 5.5 beb83fef92
Release / Cria a release (push) Canceled after 0s
Release / CLI (linux-aarch64) (push) Canceled after 0s
Release / CLI (linux-x86_64) (push) Canceled after 0s
Release / CLI (macos-aarch64) (push) Canceled after 0s
Release / CLI (macos-x86_64) (push) Canceled after 0s
Release / CLI (windows-x86_64) (push) Canceled after 0s
Release / App desktop (linux-x86_64) (push) Canceled after 0s
Release / App desktop (macos-aarch64) (push) Canceled after 0s
Release / App desktop (macos-x86_64) (push) Canceled after 0s
Release / App desktop (windows-x86_64) (push) Canceled after 0s
Release / Publica a release (push) Canceled after 0s
Formatação (cargo fmt) e ajustes do clippy para o CI da 1.0.0
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-19 11:41:58 -03:00

538 lines
18 KiB
Rust

//! Orquestra a varredura: layout → metadados dos sistemas de arquivos →
//! carving no espaço livre/desconhecido → fusão sem duplicatas.
use crate::carve::{carve_range, CarveCfg, CarveEnd, CarveEvents, HitAction};
use crate::fs::{self, merge_extents, overlaps, Ctx};
use crate::partition;
use crate::signatures::category_for_ext;
use crate::source::Source;
use crate::types::*;
use serde::{Deserialize, Serialize};
use std::collections::HashMap;
use std::sync::atomic::AtomicU64;
#[derive(Debug, Clone, Serialize, Deserialize, Default)]
pub struct Progress {
pub phase: String,
pub done: u64,
pub total: u64,
}
/// Receptor dos resultados (a interface/CLI implementa).
pub trait Sink {
fn found(&mut self, f: Found);
/// Um achado de carving confirmou um arquivo já listado por metadados.
fn confirm(&mut self, id: u64, confidence: Confidence);
/// Um achado já emitido ganhou dados melhores (ex.: JPEG remontado de fragmentos).
fn replace(&mut self, f: Found) {
self.found(f);
}
fn progress(&mut self, p: Progress);
fn warning(&mut self, w: String);
fn layout(&mut self, l: &DiskLayout);
}
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct RangeState {
pub start: u64,
pub end: u64,
pub next: u64,
}
/// Estado serializável para pausar/retomar.
#[derive(Debug, Clone, Serialize, Deserialize, Default)]
pub struct Checkpoint {
pub fs_done: bool,
pub ranges: Vec<RangeState>,
pub next_id: u64,
pub layout: Option<DiskLayout>,
/// A varredura parou porque o dispositivo parou de responder (desconectado?).
#[serde(default)]
pub io_lost: bool,
}
pub enum Outcome {
Completed,
Interrupted(Checkpoint),
}
pub fn run(
src: &dyn Source,
opts: &ScanOptions,
resume: Option<Checkpoint>,
existing: &[Found],
cancelled: &(dyn Fn() -> bool + Sync),
sink: &mut dyn Sink,
) -> Outcome {
let mut cp = resume.unwrap_or_default();
let next_id = AtomicU64::new(
cp.next_id
.max(existing.iter().map(|f| f.id + 1).max().unwrap_or(1))
.max(1),
);
sink.progress(Progress {
phase: "Lendo tabela de partições".into(),
done: 0,
total: 0,
});
let layout = match cp.layout.clone() {
Some(l) => l,
None => {
let l = partition::analyze(src, opts.mode == ScanMode::Deep);
sink.layout(&l);
for w in &l.warnings {
sink.warning(w.clone());
}
l
}
};
let mut layout = layout;
if opts.skip_primary {
layout.partitions.clear();
layout.gaps.clear();
layout.table = "membro de RAID".into();
}
// Volumes lógicos e outros volumes montados que não estão na tabela do disco.
for p in &opts.extra_parts {
if !layout.partitions.iter().any(|q| q.start == p.start) {
let mut p = p.clone();
p.index = layout.partitions.len();
layout.partitions.push(p);
}
}
cp.layout = Some(layout.clone());
// Índices para fusão: primeiro offset -> (id, confiança), e extents de achados por metadados.
let mut starts: HashMap<u64, (u64, Confidence)> = HashMap::new();
let mut fs_extents: Vec<Extent> = vec![];
for f in existing {
index_found(f, &mut starts, &mut fs_extents);
}
if !cp.fs_done {
let mut live_all: Vec<Extent> = vec![];
let mut free_all: Vec<Extent> = vec![];
let mut have_free_for: Vec<(u64, u64)> = vec![];
for p in &layout.partitions {
if p.encrypted.is_some() {
continue;
}
let phase = format!("Lendo metadados {} (partição {})", p.fs.name(), p.index + 1);
sink.progress(Progress {
phase: phase.clone(),
done: 0,
total: 0,
});
let mut emit_buf: Vec<Found> = vec![];
let res = {
let progress_phase = phase.clone();
let prog_cell = std::sync::Mutex::new((0u64, 0u64));
let progress = |d: u64, t: u64| {
*prog_cell.lock().unwrap() = (d, t);
};
let mut emit = |f: Found| emit_buf.push(f);
let mut ctx = Ctx {
src,
next_id: &next_id,
cancelled,
progress: &progress,
emit: &mut emit,
};
let r = match p.fs {
FsKind::Ntfs => Some(fs::ntfs::scan(&mut ctx, p.start)),
FsKind::Fat12 | FsKind::Fat16 | FsKind::Fat32 => {
Some(fs::fat::scan(&mut ctx, p.start, p.fs))
}
FsKind::Exfat => Some(fs::exfat::scan(&mut ctx, p.start)),
FsKind::Ext => Some(fs::ext::scan(&mut ctx, p.start, p.len)),
FsKind::HfsPlus => Some(fs::hfsplus::scan(&mut ctx, p.start, p.len)),
FsKind::Btrfs => Some(fs::btrfs::scan(&mut ctx, p.start, p.len)),
FsKind::Xfs => Some(fs::xfs::scan(&mut ctx, p.start, p.len)),
FsKind::Apfs => Some(fs::apfs::scan(&mut ctx, p.start, p.len)),
_ => None,
};
let _ = progress_phase;
r
};
for f in emit_buf {
index_found(&f, &mut starts, &mut fs_extents);
sink.found(f);
}
if let Some(r) = res {
for w in r.warnings {
sink.warning(w);
}
live_all.extend(r.live);
if let Some(fr) = r.free {
free_all.extend(fr);
have_free_for.push((p.start, p.start + p.len));
}
}
if cancelled() {
cp.next_id = next_id.load(std::sync::atomic::Ordering::Relaxed);
return Outcome::Interrupted(cp);
}
}
cp.fs_done = true;
cp.ranges = if opts.mode == ScanMode::Deep {
build_ranges(
&layout,
&live_all,
&free_all,
&have_free_for,
opts,
src.size(),
)
} else {
vec![]
};
}
fs_extents = merge_extents(fs_extents);
let total: u64 = cp.ranges.iter().map(|r| r.end - r.start).sum();
let done: u64 = cp
.ranges
.iter()
.map(|r| r.next.saturating_sub(r.start).min(r.end - r.start))
.sum();
let cfg = CarveCfg {
max_size: opts.max_carve_size,
categories: &opts.categories,
};
let mut handler = Handler {
src,
sink,
starts,
fs_extents,
next_id: &next_id,
done,
total,
index: crate::defrag::ClusterIndex::new(src.size()),
pending: vec![],
intact_mp4: vec![],
refs: opts
.references
.iter()
.filter_map(|p| {
let s = crate::source::open_path(std::path::Path::new(p)).ok()?;
crate::mp4rebuild::config_from_mp4(s.as_ref(), 0, s.size())
})
.collect(),
carved: vec![],
};
for ri in 0..cp.ranges.len() {
let (start, end) = (cp.ranges[ri].start, cp.ranges[ri].end);
let mut next = cp.ranges[ri].next;
if next >= end {
continue;
}
let end_state = carve_range(src, start, end, &mut next, &cfg, cancelled, &mut handler);
cp.ranges[ri].next = next;
if end_state != CarveEnd::Done {
cp.io_lost = end_state == CarveEnd::IoLost;
cp.next_id = next_id.load(std::sync::atomic::Ordering::Relaxed);
return Outcome::Interrupted(cp);
}
}
handler.rebuild_pending();
Outcome::Completed
}
/// Arquivo que parece quebrado: tenta recompor depois de varrer o disco todo.
struct Pending {
found: Found,
hit_len: u64,
mp4: bool,
}
struct Handler<'a> {
src: &'a dyn Source,
sink: &'a mut dyn Sink,
starts: HashMap<u64, (u64, Confidence)>,
fs_extents: Vec<Extent>,
next_id: &'a AtomicU64,
done: u64,
total: u64,
index: crate::defrag::ClusterIndex,
pending: Vec<Pending>,
/// Vídeos íntegros achados no carving (início, tamanho): fontes de SPS/PPS.
intact_mp4: Vec<(u64, u64)>,
refs: Vec<crate::mp4rebuild::MediaConfig>,
/// Extents de arquivos carved já emitidos (não podem servir de fragmento para outro).
carved: Vec<Extent>,
}
impl Handler<'_> {
fn rebuild_mp4(&mut self, p: Pending, off: u64) {
// Referências: as informadas pelo usuário e os vídeos íntegros mais próximos no disco.
let mut refs = self.refs.clone();
let mut near = self.intact_mp4.clone();
near.sort_by_key(|&(o, _)| o.abs_diff(off));
for (o, l) in near.into_iter().take(6) {
if let Some(c) = crate::mp4rebuild::config_from_mp4(self.src, o, l) {
refs.push(c);
}
}
let Some(r) = crate::mp4rebuild::rebuild(self.src, off, p.hit_len, &refs) else {
return;
};
let Ok(pre) = crate::source::read_vec(self.src, off, r.pre_len as usize) else {
return;
};
let payload_len = r.payload.1 - r.payload.0;
let mut f = p.found;
let hdr = r.mdat_header();
f.size = pre.len() as u64 + hdr.len() as u64 + payload_len + r.moov.len() as u64;
f.data = DataLoc::Parts {
parts: vec![
Part::Bytes { bytes: pre },
Part::Bytes { bytes: hdr },
Part::Run {
offset: r.payload.0,
len: payload_len,
},
Part::Bytes { bytes: r.moov },
],
};
f.ext = "mp4".into();
f.name = format!("arquivo_{:06}.mp4", f.id);
f.confidence = Confidence::Medium;
f.note = Some(format!(
"índice (moov) reconstruído: {} quadros {}x{} a {:.1} fps, sem áudio; ajuste de tempo aproximado",
r.frames, r.width, r.height, r.fps
));
self.carved.push(Extent::new(r.payload.0, payload_len));
self.sink.replace(f);
}
fn rebuild_pending(&mut self) {
let pend = std::mem::take(&mut self.pending);
for p in pend {
let off = p.found.first_offset().unwrap_or(0);
if p.mp4 {
self.rebuild_mp4(p, off);
continue;
}
let mut used = self.fs_extents.clone();
used.extend(self.carved.iter().copied());
// O próprio começo do arquivo não conta como "ocupado" para ele mesmo.
used.retain(|e| !(e.offset == off));
let pool = crate::defrag::Pool {
index: &self.index,
used: &used,
};
if let crate::defrag::Analysis::Rebuilt { extents, size } =
crate::defrag::analyze_jpeg(self.src, off, p.hit_len, &pool, true)
{
let mut f = p.found;
f.size = size;
f.data = DataLoc::Extents {
extents: extents.clone(),
};
f.confidence = Confidence::Medium;
f.note = Some(format!(
"remontado a partir de {} fragmentos separados no disco; confira a imagem",
extents.len()
));
self.carved.extend(extents);
self.sink.replace(f);
}
}
}
}
impl CarveEvents for Handler<'_> {
fn chunk(&mut self, abs: u64, data: &[u8]) {
self.index.scan_chunk(abs, data);
}
fn advance(&mut self, n: u64) {
self.done += n;
self.sink.progress(Progress {
phase: "Varredura profunda por assinatura".into(),
done: self.done.min(self.total),
total: self.total,
});
}
fn hit(&mut self, off: u64, hit: &crate::signatures::Hit) -> HitAction {
if let Some(&(id, conf)) = self.starts.get(&off) {
if hit.confidence > conf {
self.sink.confirm(id, hit.confidence);
self.starts.insert(off, (id, hit.confidence));
}
return HitAction::Skip;
}
if overlaps(&self.fs_extents, off, 1) {
return HitAction::Continue; // dentro de arquivo já listado (ex.: miniatura embutida)
}
let mut hit = hit.clone();
let mut skip_len = None;
let mut pending_len = None;
if hit.ext == "jpg" {
let pool = crate::defrag::Pool {
index: &self.index,
used: &[],
};
match crate::defrag::analyze_jpeg(self.src, off, hit.len, &pool, false) {
crate::defrag::Analysis::Intact { len } => {
hit.len = len;
hit.confidence = Confidence::High;
}
crate::defrag::Analysis::Broken { good_len } => {
pending_len = Some(hit.len);
hit.len = good_len;
hit.confidence = Confidence::Low;
skip_len = Some(good_len);
}
crate::defrag::Analysis::Unsupported | crate::defrag::Analysis::Rebuilt { .. } => {}
}
}
let mut pending_mp4 = false;
if matches!(hit.ext, "mp4" | "mov" | "m4v" | "3gp") {
if hit.confidence == Confidence::High {
self.intact_mp4.push((off, hit.len));
} else {
pending_mp4 = true;
pending_len = Some(hit.len);
}
}
let hit = &hit;
let id = self
.next_id
.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
let f = Found {
id,
name: format!("arquivo_{id:06}.{}", hit.ext),
dir: None,
ext: hit.ext.to_string(),
category: category_for_ext(hit.ext),
size: hit.len,
origin: Origin::Carved,
confidence: hit.confidence,
data: DataLoc::Extents {
extents: vec![Extent::new(off, hit.len)],
},
modified: None,
created: None,
accessed: None,
taken: None,
camera: None,
px: None,
gps: false,
sig: None,
note: if hit.confidence == Confidence::Low {
Some("sem rodapé/estrutura completa: pode estar truncado".into())
} else {
None
},
};
self.starts.insert(off, (id, hit.confidence));
self.carved.push(Extent::new(off, hit.len));
if let Some(hl) = pending_len {
let mut pf = f.clone();
pf.note = Some(if pending_mp4 {
"vídeo sem índice (moov)".into()
} else {
"JPEG interrompido ou fragmentado".into()
});
self.pending.push(Pending {
found: pf,
hit_len: hl,
mp4: pending_mp4,
});
}
self.sink.found(f);
match skip_len {
Some(n) => HitAction::SkipLen(n),
None => HitAction::Skip,
}
}
}
fn index_found(f: &Found, starts: &mut HashMap<u64, (u64, Confidence)>, ex: &mut Vec<Extent>) {
if let DataLoc::Extents { extents } = &f.data {
if let Some(o) = f.first_offset() {
starts.insert(o, (f.id, f.confidence));
}
ex.extend(extents.iter().filter(|e| !e.sparse).copied());
}
}
fn build_ranges(
layout: &DiskLayout,
live: &[Extent],
free: &[Extent],
have_free_for: &[(u64, u64)],
opts: &ScanOptions,
disk_size: u64,
) -> Vec<RangeState> {
let live = merge_extents(live.to_vec());
let free = merge_extents(free.to_vec());
let mut out: Vec<Extent> = vec![];
for p in &layout.partitions {
if p.encrypted.is_some() || opts.covered.contains(&p.start) {
continue;
}
let whole = Extent::new(p.start, p.len.min(disk_size.saturating_sub(p.start)));
if opts.include_allocated {
out.push(whole);
} else if have_free_for
.iter()
.any(|&(s, e)| s == p.start && e == p.start + p.len)
{
out.extend(clip(&free, whole));
} else if p.fs.has_metadata_scanner() {
out.extend(subtract(whole, &live));
} else {
out.push(whole);
}
}
for g in &layout.gaps {
out.push(Extent::new(g.start, g.len));
}
if layout.partitions.is_empty() && layout.gaps.is_empty() {
out.push(Extent::new(0, disk_size));
}
merge_extents(out)
.into_iter()
.filter(|e| e.len >= 4096)
.map(|e| RangeState {
start: e.offset,
end: e.end(),
next: e.offset,
})
.collect()
}
fn clip(sorted: &[Extent], window: Extent) -> Vec<Extent> {
sorted
.iter()
.filter_map(|e| {
let s = e.offset.max(window.offset);
let t = e.end().min(window.end());
(t > s).then(|| Extent::new(s, t - s))
})
.collect()
}
fn subtract(window: Extent, sorted_live: &[Extent]) -> Vec<Extent> {
let mut out = vec![];
let mut cur = window.offset;
let i = sorted_live.partition_point(|e| e.end() <= window.offset);
for e in &sorted_live[i..] {
if e.offset >= window.end() {
break;
}
if e.offset > cur {
out.push(Extent::new(cur, e.offset - cur));
}
cur = cur.max(e.end());
}
if cur < window.end() {
out.push(Extent::new(cur, window.end() - cur));
}
out
}