ddc_dec/lib.rs
1//! ddc-dec — DEX → Java decompiler front-end over `jdc-core`.
2//!
3//! Pipeline per method: DEX code units → CFG → per-block register→IR
4//! lifting → `jdc-core` structuring/conversion → refinement passes →
5//! emission. Class-level rendering mirrors jcdc's `classdec` but reads
6//! DEX metadata (no signatures, no generics).
7
8pub mod cfg;
9pub mod cure_pass;
10pub mod classdec;
11pub mod fwdb;
12
13pub use classdec::{sanitize_fq, ClassOptions};
14pub mod ctx;
15pub mod lift;
16pub mod method;
17pub mod passes;
18pub mod platform;
19mod refscan;
20pub mod ssa_census;
21
22use jdc_core::FxHashMap as HashMap;
23
24use ddc_dex::annotations::{self, EncodedValue};
25
26/// A resolved static-field initializer.
27#[derive(Debug, Clone)]
28pub enum StaticValue {
29 Int(i64),
30 Float(f32),
31 Double(f64),
32 Str(String),
33 Type(String),
34 Boolean(bool),
35 Null,
36 /// (declaring class, field name, field descriptor) — enum constants /
37 /// const field refs. The desc participates in the rename-registry key
38 /// (owner, name, desc) so the render site can consult field_display.
39 Field(String, String, String),
40 Other,
41}
42use ddc_dex::{ClassDef, DexFile};
43use jdc_core::types::{
44 parse_field_descriptor, parse_method_descriptor, JavaType, MethodDescriptor,
45};
46
47// ---------------------------------------------------------------------------
48// Access flags (Dalvik numbering, JVM-compatible subset).
49// ---------------------------------------------------------------------------
50pub mod access {
51 pub const ACC_PUBLIC: u32 = 0x1;
52 pub const ACC_PRIVATE: u32 = 0x2;
53 pub const ACC_PROTECTED: u32 = 0x4;
54 pub const ACC_STATIC: u32 = 0x8;
55 pub const ACC_FINAL: u32 = 0x10;
56 pub const ACC_SYNCHRONIZED: u32 = 0x20;
57 pub const ACC_BRIDGE: u32 = 0x40;
58 pub const ACC_VARARGS: u32 = 0x80;
59 pub const ACC_NATIVE: u32 = 0x100;
60 pub const ACC_INTERFACE: u32 = 0x200;
61 pub const ACC_ABSTRACT: u32 = 0x400;
62 pub const ACC_STRICT: u32 = 0x800;
63 pub const ACC_SYNTHETIC: u32 = 0x1000;
64 pub const ACC_ANNOTATION: u32 = 0x2000;
65 pub const ACC_ENUM: u32 = 0x4000;
66 pub const ACC_CONSTRUCTOR: u32 = 0x1_0000;
67 pub const ACC_DECLARED_SYNCHRONIZED: u32 = 0x2_0000;
68}
69
70/// One field of a pooled class.
71#[derive(Debug, Clone)]
72pub struct PoolField {
73 /// Shared with the dex string table (field names repeat massively
74 /// under R8's single-letter namespace — one Arc per unique name).
75 pub name: std::sync::Arc<str>,
76 /// Field descriptor (`I`, `Ljava/lang/String;`, ...) — shared per
77 /// unique descriptor through the same table.
78 pub desc: std::sync::Arc<str>,
79 pub access: u32,
80 pub is_static: bool,
81}
82
83/// One method of a pooled class.
84#[derive(Debug, Clone)]
85pub struct PoolMethod {
86 /// Shared with the dex string table (the idx IS the dedup key).
87 pub name: std::sync::Arc<str>,
88 /// Method descriptor, shared per proto (a dex's proto table is the
89 /// descriptor dedup layer — 1.9M methods on lark share ~1 proto
90 /// table's worth of unique descriptors).
91 pub desc: std::sync::Arc<str>,
92 pub access: u32,
93 pub code_off: u32,
94 pub debug_info_off: u32,
95 /// Which DEX image the body lives in.
96 pub dex_idx: usize,
97}
98
99impl PoolMethod {
100 pub fn parsed_desc(&self) -> Option<MethodDescriptor> {
101 parse_method_descriptor(&self.desc)
102 }
103 pub fn is_static(&self) -> bool {
104 self.access & access::ACC_STATIC != 0
105 }
106 pub fn is_abstract_or_native(&self) -> bool {
107 self.access & (access::ACC_ABSTRACT | access::ACC_NATIVE) != 0
108 }
109}
110
111/// One pool entry: either already built, or name-registered with its
112/// (dex index, class_def index) locator for on-demand construction.
113enum ClassEntry {
114 Eager(PoolClass),
115 Lazy {
116 at: (usize, usize),
117 pc: std::sync::OnceLock<PoolClass>,
118 },
119}
120
121/// A class materialized from one `class_def_item` (with ids resolved to
122/// names and the class data expanded).
123#[derive(Debug, Clone)]
124pub struct PoolClass {
125 pub name: String,
126 pub access: u32,
127 /// `None` for `java/lang/Object` roots.
128 pub super_name: Option<String>,
129 pub interfaces: Vec<String>,
130 pub source_file: Option<String>,
131 pub static_fields: Vec<PoolField>,
132 pub instance_fields: Vec<PoolField>,
133 pub direct_methods: Vec<PoolMethod>,
134 pub virtual_methods: Vec<PoolMethod>,
135 /// Static initial values aligned with the head of `static_fields`.
136 pub static_values: Vec<StaticValue>,
137 /// Nesting evidence with annotation ids resolved to names.
138 pub nesting: ResolvedNesting,
139 /// Which DEX image the class body lives in (provenance header).
140 pub dex_idx: usize,
141}
142
143/// Nesting evidence, names resolved at pooling time.
144#[derive(Debug, Clone, Default)]
145pub struct ResolvedNesting {
146 pub enclosing_class: Option<String>,
147 /// (class internal name, method name) of the enclosing method.
148 pub enclosing_method: Option<(String, String)>,
149 pub member_classes: Vec<String>,
150}
151
152impl PoolClass {
153 pub fn is_interface(&self) -> bool {
154 self.access & access::ACC_INTERFACE != 0
155 }
156
157 pub fn is_enum(&self) -> bool {
158 self.access & access::ACC_ENUM != 0
159 }
160
161 pub fn is_synthetic(&self) -> bool {
162 self.access & access::ACC_SYNTHETIC != 0
163 }
164
165 pub fn is_static_nested(&self) -> bool {
166 self.access & access::ACC_STATIC != 0
167 }
168
169 /// All methods in declaration order.
170 pub fn all_methods(&self) -> impl Iterator<Item = &PoolMethod> {
171 self.direct_methods
172 .iter()
173 .chain(self.virtual_methods.iter())
174 }
175
176 /// Find a method by name + descriptor.
177 pub fn find_method(&self, name: &str, desc: &str) -> Option<&PoolMethod> {
178 self.all_methods()
179 .find(|m| &*m.name == name && &*m.desc == desc)
180 }
181
182 /// Constructors `<init>` matching `arity` descriptor arguments.
183 pub fn ctors_by_arity(&self, arity: usize) -> Vec<&PoolMethod> {
184 self.all_methods()
185 .filter(|m| &*m.name == "<init>")
186 .filter(|m| {
187 m.parsed_desc()
188 .map(|d| d.args.len() == arity)
189 .unwrap_or(false)
190 })
191 .collect()
192 }
193
194 pub fn field_flags_of(&self, name: &str) -> Option<u32> {
195 self.static_fields
196 .iter()
197 .chain(self.instance_fields.iter())
198 .find(|f| f.name.as_ref() == name)
199 .map(|f| f.access)
200 }
201}
202
203/// Multi-DEX class pool: classes from all images, first definition wins.
204/// One facade-map entry: the public base plus, for PARTIALLY covered
205/// parts, the exact static members the base declares (None = the whole
206/// part rewrites; Some = per-member gate). Single map = one hash probe
207/// per static reference (two maps measured +2.4s on weibo).
208type FacadeTarget = (
209 String,
210 Option<
211 std::sync::Arc<(
212 jdc_core::FxHashSet<(String, String)>,
213 jdc_core::FxHashSet<String>,
214 )>,
215 >,
216);
217
218pub struct DexPool {
219 /// Images as Arcs: `dex()` hands out snapshots whose borrows stay
220 /// valid for the Arc's lifetime (decompile workers hold them), while
221 /// `release_images` empties the image data through exclusive access
222 /// when all of an image's classes are done (full-decompile driver
223 /// only — progressive callers keep images whole).
224 dexes: Vec<std::sync::Arc<DexFile>>,
225 /// Human label per image ("weibo!classes.dex") for the provenance
226 /// header; defaults to "dex N" until the CLI names the inputs.
227 pub dex_labels: Vec<String>,
228 /// Name → class entry. Eager pools materialize everything at add
229 /// time (full decompile, tests); lazy pools register names only and
230 /// materialize on first `get` (progressive `getclass` — one class's
231 /// PoolClass costs annotation reads, 145k of them cost ~0.2s).
232 classes: HashMap<String, ClassEntry>,
233 /// Class names in insertion order (stable for output).
234 pub order: Vec<String>,
235 /// Per-image remaining class count once `arm_retirement` fires.
236 retire_counts: std::sync::Mutex<Vec<u64>>,
237 retire_armed: std::sync::atomic::AtomicBool,
238 /// name → outer (computed once; `$` heuristic + dalvik annotations).
239 outers: std::sync::OnceLock<HashMap<String, Option<String>>>,
240 /// outer → direct children (computed once).
241 children: std::sync::OnceLock<HashMap<String, Vec<String>>>,
242 /// Per-image hot-reference interning (parallel to `dexes`).
243 ref_caches: Vec<DexRefCache>,
244 /// Synthetic-static accessor code snapshots keyed by (dex_idx,
245 /// code_off). `inline_accessors` reads accessor bodies ACROSS images
246 /// during worker runs; image retirement can release those bytes
247 /// mid-run, making the inline decision — and with it var numbering
248 /// downstream — dependent on worker completion interleaving
249 /// (observed: reqable enum-constant args flipping `var0`/`var1`
250 /// between identical runs; DDC_NORETIRE=1 stable). The full-decompile
251 /// driver fills this BEFORE `arm_retirement`; progressive/lazy pools
252 /// never retire and fall through to live reads.
253 accessor_code: std::sync::OnceLock<AccessorSnapshots>,
254 /// Cached package → direct-class simple names (import-collision gate).
255 pkg_simples:
256 std::sync::OnceLock<jdc_core::FxHashMap<String, std::sync::Arc<jdc_core::FxHashSet<String>>>>,
257 /// First path segments of all packages (`v2` for `v2/n`): local or
258 /// field names equal to one capture package-qualified renders
259 /// (`v2.n.a`) — deshadow_locals consults this.
260 root_segs: std::sync::OnceLock<jdc_core::FxHashSet<String>>,
261 /// Partially-covered facade parts (see kotlin_facade_partial_map).
262 /// Kotlin multi-file facade parts → public facade (see
263 /// `kotlin_facade_map`).
264 kotlin_facades: std::sync::OnceLock<jdc_core::FxHashMap<String, FacadeTarget>>,
265 /// Per class internal name: the set of `<init>` ARGUMENT descriptors
266 /// (the `(..)` inner text) REFERENCED through that class anywhere in
267 /// the pool. Dex resolves a `<init>` method-ref up the superclass
268 /// chain, so a subclass `super(..)` may target `C.<init>(sig)` that C
269 /// does NOT declare (an ancestor does) — Java has no inherited ctors,
270 /// so C needs a bridge. Computed once from the method-id tables
271 /// (parsed, survive image retirement). See `inherited_ctor_bridges`.
272 ctor_refs: std::sync::OnceLock<jdc_core::FxHashMap<String, jdc_core::FxHashSet<String>>>,
273}
274
275/// Raw code_item bytes of every synthetic-static accessor, keyed by
276/// (dex_idx, code_off) — see `snapshot_accessor_code`.
277type AccessorSnapshots = HashMap<(usize, u32), std::sync::Arc<[u8]>>;
278
279/// Lazily-filled, thread-shared reference interning for one image.
280///
281/// The lifter used to rebuild class names, field/method names and WHOLE
282/// method descriptors per INSTRUCTION (`method_ref` alone: N param
283/// Strings → `format!` descriptor → re-parse — 6-10 allocations per
284/// invoke). Every entry here is built at most once per (image, table
285/// index); lifts and the endless IR clones downstream are refcount
286/// bumps. Slots are `OnceLock` so worker threads fill them race-free
287/// without a lock on the hot path.
288pub(crate) struct DexRefCache {
289 proto_descs: Box<[std::sync::OnceLock<std::sync::Arc<jdc_core::types::MethodDescriptor>>]>,
290 type_names: Box<[std::sync::OnceLock<std::sync::Arc<str>>]>,
291 type_javas: Box<[std::sync::OnceLock<std::sync::Arc<jdc_core::types::JavaType>>]>,
292}
293
294impl DexRefCache {
295 fn new(protos: usize, types: usize) -> Self {
296 DexRefCache {
297 proto_descs: (0..protos).map(|_| std::sync::OnceLock::new()).collect(),
298 type_names: (0..types).map(|_| std::sync::OnceLock::new()).collect(),
299 type_javas: (0..types).map(|_| std::sync::OnceLock::new()).collect(),
300 }
301 }
302}
303
304impl DexPool {
305 pub fn new() -> Self {
306 DexPool {
307 dexes: Vec::new(),
308 retire_armed: std::sync::atomic::AtomicBool::new(false),
309 dex_labels: Vec::new(),
310 classes: HashMap::default(),
311 order: Vec::new(),
312 retire_counts: std::sync::Mutex::new(Vec::new()),
313 outers: std::sync::OnceLock::new(),
314 children: std::sync::OnceLock::new(),
315 ref_caches: Vec::new(),
316 accessor_code: std::sync::OnceLock::new(),
317 pkg_simples: std::sync::OnceLock::new(),
318 root_segs: std::sync::OnceLock::new(),
319 kotlin_facades: std::sync::OnceLock::new(),
320 ctor_refs: std::sync::OnceLock::new(),
321 }
322 }
323
324 /// Referenced `<init>` argument-descriptor set for a class internal
325 /// name (see `ctor_refs`). Builds the whole index on first call by
326 /// scanning every image's method-id table for `<init>` refs — cheap
327 /// (flat table, integer name-idx compare against the interned
328 /// "<init>" string) and done once for the whole decompile.
329 pub fn ctor_ref_sigs(&self, class: &str) -> Option<&jdc_core::FxHashSet<String>> {
330 self.ctor_refs
331 .get_or_init(|| {
332 let mut map: jdc_core::FxHashMap<String, jdc_core::FxHashSet<String>> =
333 jdc_core::FxHashMap::default();
334 for dex in &self.dexes {
335 let n = dex.method_count();
336 for i in 0..n {
337 let m = dex.method(i as u32);
338 if dex.string(m.name_idx) != "<init>" {
339 continue;
340 }
341 let desc = dex.proto_desc(m.proto_idx);
342 let (Some(lo), Some(hi)) = (desc.find('('), desc.find(')')) else {
343 continue;
344 };
345 map.entry(dex.class_name(m.class_idx))
346 .or_default()
347 .insert(desc[lo + 1..hi].to_string());
348 }
349 }
350 map
351 })
352 .get(class)
353 }
354
355 /// Shared `MethodDescriptor` of a proto (one parse per image).
356 pub fn proto_desc_parsed(
357 &self,
358 di: usize,
359 proto_idx: u32,
360 ) -> std::sync::Arc<jdc_core::types::MethodDescriptor> {
361 let Some(dex) = self.dexes.get(di) else {
362 return std::sync::Arc::new(jdc_core::types::MethodDescriptor {
363 args: Vec::new(),
364 ret: jdc_core::types::JavaType::Void,
365 });
366 };
367 let build = || {
368 std::sync::Arc::new(parse_proto_desc(dex, proto_idx))
369 };
370 match self
371 .ref_caches
372 .get(di)
373 .and_then(|c| c.proto_descs.get(proto_idx as usize))
374 {
375 Some(slot) => slot.get_or_init(build).clone(),
376 None => build(),
377 }
378 }
379
380 /// Shared internal class name of a type id (one strip+alloc per image).
381 pub fn type_name_arc(&self, di: usize, type_idx: u32) -> std::sync::Arc<str> {
382 let Some(dex) = self.dexes.get(di) else {
383 return std::sync::Arc::from("");
384 };
385 let build = || std::sync::Arc::from(dex.class_name(type_idx).as_str());
386 match self
387 .ref_caches
388 .get(di)
389 .and_then(|c| c.type_names.get(type_idx as usize))
390 {
391 Some(slot) => slot.get_or_init(build).clone(),
392 None => build(),
393 }
394 }
395
396 /// Shared parsed type of a type id.
397 pub fn type_java(&self, di: usize, type_idx: u32) -> std::sync::Arc<jdc_core::types::JavaType> {
398 let Some(dex) = self.dexes.get(di) else {
399 return std::sync::Arc::new(jdc_core::types::JavaType::Object(
400 std::sync::Arc::from("java/lang/Object"),
401 ));
402 };
403 let build = || std::sync::Arc::new(desc_type(dex.type_name(type_idx)));
404 match self
405 .ref_caches
406 .get(di)
407 .and_then(|c| c.type_javas.get(type_idx as usize))
408 {
409 Some(slot) => slot.get_or_init(build).clone(),
410 None => build(),
411 }
412 }
413
414 /// name → outer map, computed lazily once.
415 fn outer_map(&self) -> &HashMap<String, Option<String>> {
416 self.outers.get_or_init(|| {
417 self.order
418 .iter()
419 .map(|n| (n.clone(), find_outer_name(self, n)))
420 .collect()
421 })
422 }
423
424 /// The outer class of `name`, from the cached map (borrowed).
425 pub fn outer_of(&self, name: &str) -> Option<&str> {
426 self.outer_map().get(name).and_then(|o| o.as_deref())
427 }
428
429 /// Direct nested children of `internal` (any `$` depth 1), cached.
430 /// (Perf: returns a borrowed slice — the owned-Vec clone ran once per
431 /// class on 98k-class runs.)
432 pub fn children_of(&self, internal: &str) -> &[String] {
433 if self.children.get().is_none() {
434 // Build FIRST, set ONCE: the previous shape set an empty map
435 // to claim the OnceLock and then silently failed to store the
436 // real index (`let _ = set(...)` on an initialized lock) —
437 // children_of returned [] forever, so member nested classes
438 // were neither inlined nor emitted as files (Guard$Report
439 // vanished whole; every corpus run since the borrow refactor
440 // dropped them, invisible to the syntax-only gate).
441 let mut idx: HashMap<String, Vec<String>> = HashMap::default();
442 for name in &self.order {
443 if let Some(outer) = self.outer_of(name).map(str::to_string) {
444 idx.entry(outer).or_default().push(name.clone());
445 }
446 }
447 // A racing thread may have set an identical map first — the
448 // index is a pure function of `order`, so either copy is right.
449 let _ = self.children.set(idx);
450 }
451 self.children
452 .get()
453 .and_then(|m| m.get(internal))
454 .map(|v| v.as_slice())
455 .unwrap_or(&[])
456 }
457
458 /// Adds one image; duplicate class names keep their first definition.
459 /// Returns the image's index — call `set_dex_label` with it to give the
460 /// provenance header a real origin instead of the default "dex N".
461 pub fn add_dex(&mut self, dex: DexFile) -> usize {
462 let dex_idx = self.dexes.len();
463 // Pre-size the name map: 240k-entry growth rehashed repeatedly
464 // (reserve_rehash showed up in corpus profiles).
465 self.classes.reserve(dex.class_defs.len());
466 // The annotation reader borrows the image; pool classes are built
467 // before ownership moves into `self.dexes` (no full-image copy).
468 for cd in &dex.class_defs {
469 let name = dex.class_name(cd.class_idx);
470 if self.classes.contains_key(&name) {
471 continue;
472 }
473 let pc = pool_class_of(&dex, dex.raw(), cd, dex_idx);
474 self.classes.insert(name.clone(), ClassEntry::Eager(pc));
475 self.order.push(name);
476 }
477 self.ref_caches
478 .push(DexRefCache::new(dex.proto_count(), dex.type_count()));
479 self.dexes.push(std::sync::Arc::new(dex));
480 self.retire_counts.lock().unwrap().push(0);
481 self.dex_labels.push(format!("dex {}", dex_idx));
482 dex_idx
483 }
484
485 /// Registers class NAMES only; the PoolClass (annotation reads
486 /// included) is materialized on first `get`. Duplicate class names
487 /// keep their first definition, matching `add_dex`.
488 pub fn add_dex_lazy(&mut self, dex: DexFile) -> usize {
489 let dex_idx = self.dexes.len();
490 self.classes.reserve(dex.class_defs.len());
491 for (ci, cd) in dex.class_defs.iter().enumerate() {
492 let name = dex.class_name(cd.class_idx);
493 if self.classes.contains_key(&name) {
494 continue;
495 }
496 self.classes.insert(
497 name.clone(),
498 ClassEntry::Lazy {
499 at: (dex_idx, ci),
500 pc: std::sync::OnceLock::new(),
501 },
502 );
503 self.order.push(name);
504 }
505 self.ref_caches
506 .push(DexRefCache::new(dex.proto_count(), dex.type_count()));
507 self.dexes.push(std::sync::Arc::new(dex));
508 self.retire_counts.lock().unwrap().push(0);
509 self.dex_labels.push(format!("dex {}", dex_idx));
510 dex_idx
511 }
512
513 /// Label the image at `idx` (see `add_dex`).
514 pub fn set_dex_label(&mut self, idx: usize, label: impl Into<String>) {
515 if let Some(slot) = self.dex_labels.get_mut(idx) {
516 *slot = label.into();
517 }
518 }
519
520 pub fn len(&self) -> usize {
521 self.classes.len()
522 }
523
524 pub fn is_empty(&self) -> bool {
525 self.classes.is_empty()
526 }
527
528 pub fn get(&self, internal: &str) -> Option<&PoolClass> {
529 self.get_inner(internal, None)
530 }
531
532 /// `get` with optional pre-taken raw-image snapshots (one per
533 /// image). `materialize_all` hoists the `DexFile::raw()` lock out of
534 /// the per-class loop — the parallel materialize used to take the
535 /// same per-image lock once PER CLASS across all threads.
536 fn get_inner(&self, internal: &str, raws: Option<&[&[u8]]>) -> Option<&PoolClass> {
537 match self.classes.get(internal)? {
538 ClassEntry::Eager(pc) => Some(pc),
539 ClassEntry::Lazy { at, pc } => {
540 if let Some(built) = pc.get() {
541 return Some(built);
542 }
543 let (di, ci) = *at;
544 let dex = self.dexes.get(di)?;
545 let cd = dex.class_defs.get(ci)?;
546 let raw = match raws {
547 Some(r) => r.get(di).copied().unwrap_or(&[]),
548 None => dex.raw(),
549 };
550 let built = pool_class_of(dex, raw, cd, di);
551 Some(pc.get_or_init(move || built))
552 }
553 }
554 }
555
556 /// Materialize every lazy entry (parallel across the given chunk
557 /// count). After this the pool is observationally identical to an
558 /// eagerly-built one — including the annotation-aware outer-map —
559 /// while the annotation reads ran on worker threads instead of the
560 /// serial pool build (~0.2s on weibo).
561 pub fn materialize_all(&self, threads: usize) {
562 let keys: Vec<String> = self
563 .classes
564 .iter()
565 .filter(|(_, e)| matches!(e, ClassEntry::Lazy { .. }))
566 .map(|(k, _)| k.clone())
567 .collect();
568 if keys.is_empty() {
569 return;
570 }
571 // One raw() snapshot per image for the whole sweep (see
572 // get_inner). Safe window: materialize_all completes before
573 // retirement is armed, so no image releases while these slices
574 // are alive.
575 let raws: Vec<&[u8]> = self.dexes.iter().map(|d| d.raw()).collect();
576 let threads = threads.max(1);
577 let next = std::sync::atomic::AtomicUsize::new(0);
578 std::thread::scope(|scope| {
579 for _ in 0..threads {
580 scope.spawn(|| loop {
581 let i = next.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
582 let Some(k) = keys.get(i) else { break };
583 let _ = self.get_inner(k, Some(&raws));
584 });
585 }
586 });
587 }
588
589 /// `get` without materializing: None for not-yet-built lazy entries.
590 fn get_if_materialized(&self, internal: &str) -> Option<&PoolClass> {
591 match self.classes.get(internal)? {
592 ClassEntry::Eager(pc) => Some(pc),
593 ClassEntry::Lazy { pc, .. } => pc.get(),
594 }
595 }
596
597 /// Name registered (no materialization).
598 pub fn has_name(&self, internal: &str) -> bool {
599 self.classes.contains_key(internal)
600 }
601
602 /// One image as an Arc snapshot (borrows of the snapshot live as
603 /// long as the caller holds the Arc).
604 pub fn dex(&self, idx: usize) -> Option<std::sync::Arc<DexFile>> {
605 self.dexes.get(idx).cloned()
606 }
607
608 pub fn dex_count(&self) -> usize {
609 self.dexes.len()
610 }
611
612 /// Raw image list (install-phase whole-program scans).
613 pub fn dexes(&self) -> &[std::sync::Arc<ddc_dex::DexFile>] {
614 &self.dexes
615 }
616
617 /// First path segments of every package in the pool (cached once).
618 pub fn root_pkg_segs(&self) -> &jdc_core::FxHashSet<String> {
619 self.root_segs.get_or_init(|| {
620 self.order
621 .iter()
622 .filter_map(|n| n.find('/').map(|i| n[..i].to_string()))
623 .collect()
624 })
625 }
626
627 /// Package → simple names of its direct classes, cached once per
628 /// pool (the import-collision gate consults it per class — the
629 /// uncached scan was O(classes²) and 10×'d weixin's wall time).
630 pub fn package_simples(
631 &self,
632 ) -> &jdc_core::FxHashMap<String, std::sync::Arc<jdc_core::FxHashSet<String>>> {
633 // Arc'd sets: the per-file import machinery BORROWS a package's
634 // set (an Arc clone) instead of deep-copying it — WhatsApp's
635 // ~10k-class package paid two full copies per class (~200M
636 // String allocations), the #2/#3 CPU frames on its profile.
637 self.pkg_simples.get_or_init(|| {
638 let mut m: jdc_core::FxHashMap<String, std::sync::Arc<jdc_core::FxHashSet<String>>> =
639 jdc_core::FxHashMap::default();
640 let mut raw: jdc_core::FxHashMap<String, jdc_core::FxHashSet<String>> =
641 jdc_core::FxHashMap::default();
642 for n in &self.order {
643 if let Some((pkg, simple)) = n.rsplit_once('/') {
644 raw.entry(pkg.to_string())
645 .or_default()
646 .insert(simple.to_string());
647 }
648 }
649 m.extend(raw.into_iter().map(|(k, v)| (k, std::sync::Arc::new(v))));
650 m
651 })
652 }
653
654 /// Kotlin multi-file class facades: the compiler splits a file facade
655 /// (`kotlin.text.StringsKt`) into package-private parts
656 /// (`StringsKt__StringsKt`, `StringsKt__StringsJVMKt`) and call sites
657 /// target the PARTS directly. Java cannot access a package-private
658 /// class cross-package ("StringsKt__StringsKt在kotlin.text中不是公共的"
659 /// — 1.4k lark root errors), but the public facade EXTENDS the parts,
660 /// so static members resolve through it: rewrite the reference's owner
661 /// class to the facade. part → base, gated on base existing, being
662 /// public, and transitively extending the part (static inheritance is
663 /// the resolution guarantee).
664 pub fn kotlin_facade_map(&self) -> &jdc_core::FxHashMap<String, FacadeTarget> {
665 self.kotlin_facades.get_or_init(|| {
666 let t_fb = std::time::Instant::now();
667 let mut m: jdc_core::FxHashMap<String, FacadeTarget> =
668 jdc_core::FxHashMap::default();
669 let mut npartial = 0usize;
670 for name in &self.order {
671 let Some(i) = name.find("__") else { continue };
672 let base = &name[..i];
673 if base.is_empty() {
674 continue;
675 }
676 let Some(basec) = self.get(base) else {
677 continue;
678 };
679 if basec.access & crate::access::ACC_PUBLIC == 0 {
680 continue;
681 }
682 // base must transitively extend the part...
683 let mut cur = basec.super_name.clone();
684 let mut hops = 0u32;
685 let mut reaches = false;
686 while let Some(c) = cur {
687 if c == *name {
688 reaches = true;
689 break;
690 }
691 if hops >= 8 {
692 break;
693 }
694 hops += 1;
695 cur = self.get(&c).and_then(|cc| cc.super_name.clone());
696 }
697 // ...OR cover the part's static members: the Kotlin
698 // compiler MERGES every part's statics into the public
699 // facade, but the facade's super chain only includes ONE
700 // part — lark's `CollectionsKt extends
701 // CollectionsKt___CollectionsKt` leaves
702 // `CollectionsKt__CollectionsJVMKt` uncovered (463
703 // 不是公共的 errors). Member coverage (name+desc for
704 // methods, name for fields) is the precise rewrite
705 // condition: every reference through the facade then
706 // resolves exactly as through the part.
707 if reaches {
708 m.insert(name.clone(), (base.to_string(), None));
709 continue;
710 }
711 // Member-coverage fallback: the Kotlin compiler MERGES
712 // the parts' statics into the public facade, but R8 can
713 // sever the super chain (lark: `CollectionsKt extends
714 // CollectionsKt___CollectionsKt extends
715 // kotlin/collections/s` — `__CollectionsJVMKt` is off
716 // the chain, 463 不是公共的 errors) and may drop facade
717 // members no direct facade call needed. Map at (name,
718 // desc) granularity then: covered members rewrite, the
719 // rest keep the part owner (a latent error either way —
720 // Java cannot access the package-private part).
721 let Some(part) = self.get(name) else {
722 continue;
723 };
724 let base_methods: jdc_core::FxHashSet<(&str, &str)> = basec
725 .direct_methods
726 .iter()
727 .chain(basec.virtual_methods.iter())
728 .map(|mm| (&*mm.name, &*mm.desc))
729 .collect();
730 let base_fields: jdc_core::FxHashSet<&str> =
731 basec.static_fields.iter().map(|f| &*f.name).collect();
732 let mut methods: jdc_core::FxHashSet<(String, String)> =
733 jdc_core::FxHashSet::default();
734 let mut fields: jdc_core::FxHashSet<String> =
735 jdc_core::FxHashSet::default();
736 let mut all = true;
737 for mm in part
738 .direct_methods
739 .iter()
740 .chain(part.virtual_methods.iter())
741 .filter(|mm| mm.access & crate::access::ACC_STATIC != 0)
742 {
743 if base_methods.contains(&(&*mm.name, &*mm.desc)) {
744 methods.insert((mm.name.to_string(), mm.desc.to_string()));
745 } else {
746 all = false;
747 }
748 }
749 for pf in &part.static_fields {
750 if base_fields.contains(&*pf.name) {
751 fields.insert(pf.name.to_string());
752 } else {
753 all = false;
754 }
755 }
756 if all {
757 m.insert(name.clone(), (base.to_string(), None));
758 } else if !methods.is_empty() || !fields.is_empty() {
759 npartial += 1;
760 m.insert(
761 name.clone(),
762 (base.to_string(), Some(std::sync::Arc::new((methods, fields)))),
763 );
764 }
765 }
766 if std::env::var("DDC_STATS").is_ok() {
767 eprintln!(
768 "[renames] facade map: entries={} partial={} build={:?}",
769 m.len(),
770 npartial,
771 t_fb.elapsed()
772 );
773 }
774 m
775 })
776 }
777
778 pub fn class_names(&self) -> impl Iterator<Item = &str> {
779 self.order.iter().map(|s| s.as_str())
780 }
781
782 /// True when `sub` is assignable to `sup` (internal names), walking the
783 /// pool's hierarchy. Unknown classes are never assignable (conservative).
784 pub fn is_subtype(&self, sub: &str, sup: &str) -> bool {
785 if sub == sup {
786 return true;
787 }
788 // Arrays: covariance by element type when both are arrays.
789 if let (Some(sub_elem), Some(sup_elem)) = (array_elem(sub), array_elem(sup)) {
790 if sub_elem.starts_with('L')
791 && sup_elem.starts_with('L')
792 && sub_elem.ends_with(';')
793 && sup_elem.ends_with(';')
794 {
795 return self.is_subtype(
796 &sub_elem[1..sub_elem.len() - 1],
797 &sup_elem[1..sup_elem.len() - 1],
798 );
799 }
800 // Multi-dimensional: recurse on the element descriptors
801 // (`[[LSavedState;` <: `[[LObject;` peels one dimension at a
802 // time); the equality fallback below stays for primitive
803 // elements (`[I` vs `[J` are invariant).
804 if sub_elem.starts_with('[') && sup_elem.starts_with('[') {
805 return self.is_subtype(sub_elem, sup_elem);
806 }
807 return sub_elem == sup_elem;
808 }
809 // Everything reference-typed is Object-assignable (the pool walk
810 // used to fall off the chain end and answer false for
811 // String <: Object — the hand-written Object-formal special case
812 // in the ambiguity pin existed for exactly this hole).
813 if sup == "java/lang/Object" {
814 return !matches!(
815 sub,
816 "I" | "Z" | "B" | "S" | "C" | "J" | "F" | "D" | "V"
817 );
818 }
819 let mut cur = self.get(sub);
820 if cur.is_none() {
821 // `sub` is a framework (or phantom) type: the embedded API
822 // database decides (phantoms are absent there too → false).
823 return crate::fwdb::is_subtype(sub, sup);
824 }
825 let mut hops = 0;
826 while let Some(c) = cur {
827 hops += 1;
828 if hops > 64 {
829 return false;
830 }
831 for i in &c.interfaces {
832 if i == sup || self.is_subtype(i, sup) {
833 return true;
834 }
835 }
836 match &c.super_name {
837 Some(s) if s == sup => return true,
838 Some(s) if s != "java/lang/Object" => {
839 match self.get(s) {
840 Some(sc) => cur = Some(sc),
841 // Framework boundary: continue in the embedded DB.
842 None => return crate::fwdb::is_subtype(s, sup),
843 }
844 }
845 _ => return false,
846 }
847 }
848 false
849 }
850}
851
852impl Default for DexPool {
853 fn default() -> Self {
854 Self::new()
855 }
856}
857
858fn array_elem(desc: &str) -> Option<&str> {
859 desc.strip_prefix('[')
860}
861
862fn resolve_static_value(v: &EncodedValue, dex: &DexFile) -> StaticValue {
863 match v {
864 EncodedValue::Byte(x) => StaticValue::Int(*x as i64),
865 EncodedValue::Short(x) => StaticValue::Int(*x as i64),
866 EncodedValue::Char(x) => StaticValue::Int(*x as i64),
867 EncodedValue::Int(x) => StaticValue::Int(*x as i64),
868 EncodedValue::Long(x) => StaticValue::Int(*x),
869 EncodedValue::Float(x) => StaticValue::Float(*x),
870 EncodedValue::Double(x) => StaticValue::Double(*x),
871 EncodedValue::String(i) => StaticValue::Str(dex.string(*i).to_string()),
872 EncodedValue::Type(i) => StaticValue::Type(dex.class_name(*i)),
873 EncodedValue::Boolean(b) => StaticValue::Boolean(*b),
874 EncodedValue::Null => StaticValue::Null,
875 EncodedValue::Enum(i) | EncodedValue::Field(i) => {
876 let f = dex.field(*i);
877 StaticValue::Field(
878 dex.class_name(f.class_idx),
879 dex.string(f.name_idx).to_string(),
880 dex.type_name(f.type_idx).to_string(),
881 )
882 }
883 _ => StaticValue::Other,
884 }
885}
886
887fn pool_class_of(dex: &DexFile, raw: &[u8], cd: &ClassDef, dex_idx: usize) -> PoolClass {
888 let name = dex.class_name(cd.class_idx);
889 let super_name = if cd.superclass_idx == ddc_dex::NO_INDEX {
890 None
891 } else {
892 let s = dex.class_name(cd.superclass_idx);
893 if s.is_empty() || s == "java/lang/Object" {
894 None
895 } else {
896 Some(s)
897 }
898 };
899 let interfaces = dex
900 .interfaces_of(cd)
901 .into_iter()
902 .map(|t| dex.class_name(t))
903 .collect();
904 let source_file = if cd.source_file_idx == ddc_dex::NO_INDEX {
905 None
906 } else {
907 Some(dex.string(cd.source_file_idx).to_string())
908 };
909
910 let data = dex.class_data(cd);
911 let mk_field = |ef: &ddc_dex::EncodedField, is_static: bool| {
912 let f = dex.field(ef.field_idx);
913 PoolField {
914 name: dex.string_arc(f.name_idx),
915 desc: dex.type_name_arc(f.type_idx),
916 access: ef.access_flags,
917 is_static,
918 }
919 };
920 let static_fields: Vec<PoolField> = data
921 .static_fields
922 .iter()
923 .map(|f| mk_field(f, true))
924 .collect();
925 let instance_fields: Vec<PoolField> = data
926 .instance_fields
927 .iter()
928 .map(|f| mk_field(f, false))
929 .collect();
930 let mk_method = |em: &ddc_dex::EncodedMethod| {
931 let m = dex.method(em.method_idx);
932 let (code_off, debug_info_off) = match dex.debug_info_off_at(em.code_off) {
933 Some(d) => (em.code_off, d),
934 None => (0, 0),
935 };
936 PoolMethod {
937 name: dex.string_arc(m.name_idx),
938 desc: dex.proto_desc(m.proto_idx),
939 access: em.access_flags,
940 code_off,
941 debug_info_off,
942 dex_idx,
943 }
944 };
945 let direct_methods: Vec<PoolMethod> = data.direct_methods.iter().map(&mk_method).collect();
946 let virtual_methods: Vec<PoolMethod> = data.virtual_methods.iter().map(&mk_method).collect();
947
948 let static_values: Vec<StaticValue> = dex
949 .static_values(cd.static_values_off)
950 .into_iter()
951 .map(|v| resolve_static_value(&v, dex))
952 .collect();
953 let anns = if cd.annotations_off != 0 {
954 annotations::read_class_annotations(raw, cd.annotations_off)
955 } else {
956 Vec::new()
957 };
958 let raw = annotations::nesting_from(&anns, &|t: u32| dex.type_name(t).to_string());
959 let nesting = ResolvedNesting {
960 enclosing_class: raw.enclosing_class.map(|t| dex.class_name(t)),
961 enclosing_method: raw.enclosing_method.map(|m| {
962 let mid = dex.method(m);
963 (
964 dex.class_name(mid.class_idx),
965 dex.string(mid.name_idx).to_string(),
966 )
967 }),
968 member_classes: raw
969 .member_classes
970 .into_iter()
971 .map(|t| dex.class_name(t))
972 .collect(),
973 };
974
975 PoolClass {
976 name,
977 access: cd.access_flags,
978 super_name,
979 interfaces,
980 source_file,
981 static_fields,
982 instance_fields,
983 direct_methods,
984 virtual_methods,
985 static_values,
986 nesting,
987 dex_idx,
988 }
989}
990
991/// JavaType for a field/method descriptor segment.
992/// Parse a proto's descriptor into a MethodDescriptor straight from the
993/// tables (no string round-trip).
994fn parse_proto_desc(
995 dex: &ddc_dex::DexFile,
996 proto_idx: u32,
997) -> jdc_core::types::MethodDescriptor {
998 let proto = dex.proto(proto_idx);
999 let args = dex
1000 .proto_params(proto_idx)
1001 .iter()
1002 .map(|&t| desc_type(dex.type_name(t)))
1003 .collect();
1004 jdc_core::types::MethodDescriptor {
1005 args,
1006 ret: desc_type(dex.type_name(proto.return_type_idx)),
1007 }
1008}
1009
1010pub fn desc_type(desc: &str) -> JavaType {
1011 parse_field_descriptor(desc).unwrap_or(JavaType::Object("java/lang/Object".into()))
1012}
1013
1014/// Nesting evidence for `internal`: dalvik annotations first, then the
1015/// `$`-name heuristic against the pool.
1016pub fn find_outer_name(pool: &DexPool, internal: &str) -> Option<String> {
1017 // Annotation refinement only for ALREADY-MATERIALIZED classes: eager
1018 // pools (full decompile) behave exactly as before; lazy pools
1019 // (progressive getclass) fall straight to the `$` chain without
1020 // materializing the world.
1021 if let Some(pc) = pool.get_if_materialized(internal) {
1022 if let Some(enc) = &pc.nesting.enclosing_class {
1023 // The annotation outer must still EXIST: R8 deletes whole
1024 // outer classes and leaves the nestlings as flat `$` files
1025 // (weixin matrix JiffiesMonitorFeature deleted,
1026 // $JiffiesSnapshot survives) — an unresolved annotation
1027 // outer made refs render dotted THROUGH a missing type
1028 // ("是在不可访问的类或接口中定义的"). Fall through to the
1029 // existence-checked `$` walk.
1030 if pool.has_name(enc) {
1031 return Some(enc.clone());
1032 }
1033 }
1034 }
1035 let mut rest = internal;
1036 while let Some(d) = rest.rfind('$') {
1037 // A `$` followed by `/` is a PACKAGE-path dollar (desugar's
1038 // `j$/util/...`), never a nesting boundary: without this the
1039 // whole j$ tree attached to a root-package class `j` as family
1040 // members (deepseek's 62k-line j.java absorbing 549 desugar
1041 // classes, double-emitted against their flat j$/ files).
1042 if rest.as_bytes().get(d + 1) == Some(&b'/') {
1043 rest = &rest[..d];
1044 continue;
1045 }
1046 let cand = &rest[..d];
1047 if pool.has_name(cand) {
1048 return Some(cand.to_string());
1049 }
1050 rest = cand;
1051 }
1052 None
1053}
1054
1055/// True when the `$` tail names a plain member class (not anonymous/local/
1056/// lambda), i.e. the class renders inside its outer's compilation unit.
1057fn clean_member_tail(rest: &str) -> bool {
1058 if rest.starts_with('-') {
1059 return false;
1060 }
1061 // A member tail can never contain `/` — that is a package-path
1062 // fragment from a package-dollar name (`j$/util/X` stripped against
1063 // a root class `j`); defensive backstop for the find_outer_name
1064 // guard.
1065 if rest.contains('/') {
1066 return false;
1067 }
1068 // d8's synthetic outline/lambda/backport classes (`X$$ExternalSynthetic…`)
1069 // are TOP-LEVEL synthetics, not nested members: `$$` is d8's marker, and
1070 // find_outer_name's `$`-chain walk mistakes them for children of `X`.
1071 // Emitting them nested strands every cross-package reference on a bare
1072 // simple name ("找不到符号 变量 ExternalSyntheticBackportWithForwarding0",
1073 // weibo ×467). Targeted at `ExternalSynthetic` only — Kotlin's
1074 // `$$serializer` / `$$delegate` stay nested members.
1075 if rest.starts_with("$ExternalSynthetic") {
1076 return false;
1077 }
1078 let tail = rest.rsplit('$').next().unwrap_or(rest);
1079 if tail.is_empty() || tail.starts_with(|c: char| c.is_ascii_digit()) {
1080 return false;
1081 }
1082 true
1083}
1084
1085/// Classes to emit as their own compilation units: top-level classes plus
1086/// anonymous/local/lambda-shaped ones (clean members render inline).
1087impl DexPool {
1088 /// Arm image retirement (full-decompile driver only): count one
1089 /// pending class per image by pool ownership.
1090 /// Snapshot every synthetic-static accessor body (the exact predicate
1091 /// `inline_accessors` uses) as RAW code_item bytes, so cross-image
1092 /// accessor reads during worker runs survive image retirement.
1093 /// Call AFTER materialization and BEFORE `arm_retirement`; idempotent
1094 /// (first fill wins). Raw memcpy, no decode — accessors are numerous
1095 /// (weixin ~100k) but only a fraction are ever inlined, so decoding
1096 /// stays lazy at the use site. The slice bound covers the fixed
1097 /// header + insns + try items with generous handler slack (accessor
1098 /// tries are vanishingly rare; the snapshot is a deterministic
1099 /// function of the image either way).
1100 pub fn snapshot_accessor_code(&self) {
1101 let mut snap: AccessorSnapshots = HashMap::default();
1102 let mut bytes_total = 0usize;
1103 for name in &self.order {
1104 let Some(pc) = self.get_if_materialized(name) else {
1105 continue;
1106 };
1107 for m in pc.direct_methods.iter().chain(pc.virtual_methods.iter()) {
1108 if m.code_off == 0
1109 || !m.is_static()
1110 || m.access & crate::access::ACC_SYNTHETIC == 0
1111 {
1112 continue;
1113 }
1114 if snap.contains_key(&(m.dex_idx, m.code_off)) {
1115 continue;
1116 }
1117 let Some(dex) = self.dex(m.dex_idx) else {
1118 continue;
1119 };
1120 let raw = dex.raw();
1121 let off = m.code_off as usize;
1122 if off + 16 > raw.len() {
1123 continue;
1124 }
1125 let u2 = |o: usize| u16::from_le_bytes([raw[o], raw[o + 1]]) as usize;
1126 let u4 = |o: usize| {
1127 u32::from_le_bytes([raw[o], raw[o + 1], raw[o + 2], raw[o + 3]]) as usize
1128 };
1129 let tries = u2(off + 6);
1130 let insns = u4(off + 12);
1131 let mut end = off + 16 + 2 * insns;
1132 if tries > 0 {
1133 // pad to 4-align + try_item[tries] + handler blobs
1134 // (uleb-coded; slack covers realistic accessor shapes).
1135 end = end.next_multiple_of(4) + 8 * tries + 256 + 64 * tries;
1136 }
1137 let end = end.min(raw.len());
1138 let slice: std::sync::Arc<[u8]> = raw[off..end].into();
1139 bytes_total += slice.len();
1140 snap.insert((m.dex_idx, m.code_off), slice);
1141 }
1142 }
1143 if std::env::var("DDC_WALL").is_ok() {
1144 eprintln!(
1145 "[wall] accessor snapshot: {} entries, {} KB",
1146 snap.len(),
1147 bytes_total / 1024
1148 );
1149 }
1150 let _ = self.accessor_code.set(snap);
1151 }
1152
1153 /// Snapshotted accessor code_item bytes, when the snapshot was taken.
1154 /// Parse with `CodeItem::parse(&bytes, 0)`.
1155 pub fn accessor_code(&self, dex_idx: usize, code_off: u32) -> Option<std::sync::Arc<[u8]>> {
1156 self.accessor_code.get()?.get(&(dex_idx, code_off)).cloned()
1157 }
1158
1159 pub fn arm_retirement(&self) {
1160 self.retire_armed
1161 .store(true, std::sync::atomic::Ordering::Release);
1162 let mut counts = self.retire_counts.lock().unwrap();
1163 for c in counts.iter_mut() {
1164 *c = 0;
1165 }
1166 for entry in self.classes.values() {
1167 let di = match entry {
1168 ClassEntry::Eager(pc) => pc.dex_idx,
1169 ClassEntry::Lazy { at, .. } => at.0,
1170 };
1171 if let Some(c) = counts.get_mut(di) {
1172 *c += 1;
1173 }
1174 }
1175 }
1176
1177 /// Report one class finished; Some(image) when it was the image's
1178 /// last class (driver batches these into release_images).
1179 pub fn report_class_done(&self, class_name: &str) -> Option<usize> {
1180 if !self.retire_armed.load(std::sync::atomic::Ordering::Acquire) {
1181 return None;
1182 }
1183 let entry = self.classes.get(class_name)?;
1184 let di = match entry {
1185 ClassEntry::Eager(pc) => pc.dex_idx,
1186 ClassEntry::Lazy { at, .. } => at.0,
1187 };
1188 let mut counts = self.retire_counts.lock().unwrap();
1189 match counts.get_mut(di) {
1190 Some(c) => {
1191 *c = c.saturating_sub(1);
1192 if *c == 0 {
1193 Some(di)
1194 } else {
1195 None
1196 }
1197 }
1198 None => None,
1199 }
1200 }
1201
1202 /// Release the images' inflated bytes. Safe: the driver calls this
1203 /// only when every class of each image is emitted or failed; new
1204 /// `dex()` snapshots still work (tables valid, code accessors empty).
1205 pub fn release_images(&self, indexes: &[usize]) {
1206 let mut counts = self.retire_counts.lock().unwrap();
1207 for &i in indexes {
1208 if let Some(dex) = self.dexes.get(i) {
1209 // Mark the shared image: code accessors go empty
1210 // immediately; the bytes drop when the last snapshot
1211 // drops (workers hold snapshots only mid-class).
1212 dex.mark_released();
1213 }
1214 if let Some(c) = counts.get_mut(i) {
1215 *c = u64::MAX; // released marker
1216 }
1217 }
1218 }
1219}
1220
1221pub fn top_level_classes(pool: &DexPool) -> Vec<String> {
1222 pool.class_names()
1223 .filter(|name| match pool.outer_of(name) {
1224 None => true,
1225 Some(outer) => {
1226 if pool.get(outer).is_none() {
1227 return true;
1228 }
1229 // The outer can come from an ANNOTATION (EnclosingClass)
1230 // with no naming relationship to this class — obfuscated
1231 // apps pair a 1-char name with a long enclosing descriptor
1232 // (weixin), which made the old blind slice panic. Only a
1233 // real `outer$tail` prefix yields a member tail; anything
1234 // else is a standalone unit.
1235 let rest = name
1236 .strip_prefix(outer)
1237 .and_then(|t| t.strip_prefix('$'))
1238 .unwrap_or("");
1239 !clean_member_tail(rest)
1240 }
1241 })
1242 .map(|n| n.to_string())
1243 .collect()
1244}
1245
1246/// Sanitized form of one internal-name segment — the shared mapping the
1247/// declaration site (`print_class_name` → `sanitize_ref`), the CLASS
1248/// lookup and the CLI writer must all agree on. Exported so the writer
1249/// cannot drift from the declaration (see `sanitize_fq`).
1250pub fn sanitize_seg(seg: &str) -> String {
1251 classdec::sanitize_fq(seg)
1252}
1253
1254/// Sanitized internal PATH form of a class name: the exact on-disk
1255/// identity `source_path` writes and the declaration claims.
1256pub fn sanitize_internal(internal: &str) -> String {
1257 internal
1258 .split('/')
1259 .map(classdec::sanitize_fq)
1260 .collect::<Vec<_>>()
1261 .join("/")
1262}
1263
1264/// Deterministic case-collision renames over the FILE-emission set:
1265/// classes whose internal names differ only in letter case cannot share
1266/// one case-insensitive directory; the first (sorted) member of each
1267/// group keeps its name, the others gain `_2`, `_3`, … on the simple
1268/// segment. The map carries identity entries for unrenamed file-level
1269/// classes (they anchor nested prefix walks in apply_class_rename).
1270pub fn case_rename_map(pool: &DexPool) -> HashMap<String, String> {
1271 use std::collections::HashMap;
1272 let mut groups: HashMap<String, Vec<String>> = HashMap::default();
1273 for t in top_level_classes(pool) {
1274 groups.entry(t.to_lowercase()).or_default().push(t);
1275 }
1276 // No `folds` clone: the fold check consults `groups` itself (the old
1277 // full HashSet<String> of every lowercase key was a ~25-30MB pure
1278 // duplicate on the large corpora). Sort in place first, then an
1279 // IMMUTABLE walk keeps the keys queryable for the fold lookup.
1280 for v in groups.values_mut() {
1281 v.sort();
1282 }
1283 let mut map = HashMap::default();
1284 for members in groups.values() {
1285 for (i, m) in members.iter().enumerate() {
1286 if i == 0 {
1287 map.insert(m.clone(), m.clone());
1288 continue;
1289 }
1290 // Suffix the simple segment; keep incrementing if the fold
1291 // of the suffixed name is already taken by a real class.
1292 let cut = m.rfind('/').map(|x| x + 1).unwrap_or(0);
1293 let mut n = i + 1;
1294 loop {
1295 let cand = format!("{}{}_{}", &m[..cut], &m[cut..], n);
1296 let fold = cand.to_lowercase();
1297 if !groups.contains_key(&fold) {
1298 map.insert(m.clone(), cand);
1299 break;
1300 }
1301 n += 1;
1302 }
1303 }
1304 }
1305 map
1306}
1307
1308/// Display-level renames for two nested-class shapes that render
1309/// uncompilable or inconsistent output:
1310///
1311/// 1. A member class whose simple tail equals its rendered parent's
1312/// tail (`x0$a$a` inside `x0$a`): javac rejects a member class with
1313/// the same simple name as its immediately enclosing class ("已在类
1314/// x0中定义了类 x0.a") — a shape Kotlin lambda families hit by the
1315/// thousand (lark alone: 22k). Bump the tail: `a` → `a2`, `a3`, …
1316/// 2. An orphaned intermediate (`x0$a$b` with `x0$a` absent from the
1317/// pool): the member DECLARES as `b` inside `x0` but references
1318/// printed `x0.a$b` — a type that exists nowhere. Flatten the
1319/// display name to `x0$b` so every site agrees.
1320///
1321/// Both are pure display renames keyed by internal name; declarations,
1322/// ctor names, file names and every type reference funnel through
1323/// apply_class_rename, so one map keeps all sites consistent. Children
1324/// Display-level renames for nested member classes that render
1325/// uncompilable or inconsistent output:
1326///
1327/// 1. A member whose simple tail equals ANY enclosing class's simple
1328/// name in its nesting chain. javac rejects more than the immediate
1329/// parent (`class a { static class a2 { static class a {} } }` —
1330/// member `a` two levels under `a` — is "已在类 X中定义了类 X.a");
1331/// Kotlin lambda families (`x0$a$a` inside `x0$a`) hit it by the
1332/// thousand (lark alone: 22k). Bump the tail until it clears every
1333/// ancestor name: `a` → `a3`.
1334/// 2. An orphaned intermediate (`x0$a$b` with `x0$a` absent): the
1335/// member DECLARES as `b` inside `x0` but references printed
1336/// `x0.a$b` — a type that exists nowhere. Flatten the display to
1337/// `x0$b` so every site agrees.
1338///
1339/// Candidates build on the parent's DISPLAY name (ancestors are
1340/// renamed first — ascending `$`-depth order), so declarations,
1341/// constructor names, file names and every type reference stay
1342/// consistent through apply_class_rename. All are pure display
1343/// renames keyed by internal name.
1344/// In-progress-map display composition, mirroring
1345/// `rename::apply_class_rename`'s ancestor walk (the registry is not
1346/// installed yet during map construction, so mint rules must compose
1347/// through renamed ancestors themselves). A nested name whose DIRECT
1348/// parent has no entry still inherits every renamed higher prefix:
1349/// grandparent `yN5` -> `yN5_3` must turn minted `yN5$a$a2` into
1350/// `yN5_3$a$a2` (xingye case-quads: the raw-prefix entry short-
1351/// circuited the render walk — 程序包ddcroot.yN5.a不存在 ×134-family).
1352fn display_of(name: &str, map: &HashMap<String, String>) -> String {
1353 // EXACT hit first — the walk truncates the full name before ever
1354 // consulting it, and an identity entry on an ancestor prefix
1355 // (case_rename_map seeds those for every fold-group first member)
1356 // would short-circuit to the raw name (lark aa2/d$c17 minted, then
1357 // aa2/d$c$a composed against the identity of aa2/d — the child
1358 // stranded on the dead `c` prefix, 找不到符号 类 a ×1,874).
1359 if let Some(d) = map.get(name) {
1360 return d.clone();
1361 }
1362 let mut s = name;
1363 while let Some(i) = s.rfind('$') {
1364 s = &s[..i];
1365 match map.get(s) {
1366 Some(d) if d.as_str() != s => return format!("{d}{}", &name[s.len()..]),
1367 Some(_) => return name.to_string(),
1368 None => continue,
1369 }
1370 }
1371 map.get(name).cloned().unwrap_or_else(|| name.to_string())
1372}
1373
1374fn nested_collision_renames(
1375 pool: &DexPool,
1376 map: &mut HashMap<String, String>,
1377 fam_segs: &SegMap,
1378) {
1379 // Ancestors before descendants: a parent's rename must be in the
1380 // map when its children compute their display chains.
1381 let mut names: Vec<&String> = pool.order.iter().collect();
1382 names.sort_by_key(|n| n.matches('$').count());
1383 let mut assigned: jdc_core::FxHashSet<String> = jdc_core::FxHashSet::default();
1384 // Top-level package segments. A nested member type DISPLAYED as `a2`
1385 // shadows package `a2` for every qualified `a2.x` reference in its
1386 // file (member types outrank package names in class scope), so a
1387 // renamed tail must not become a package shadow — round 47's failure
1388 // mode at nested level: candidate `a2` (for j2/g$a) broke the SAME
1389 // file's `extends a2.a` ("找不到符号 类 a" + a 9450-error Object
1390 // cascade as every dependent of the corrupt head failed too).
1391 let mut pkg_segments: jdc_core::FxHashSet<String> = jdc_core::FxHashSet::default();
1392 for n in &pool.order {
1393 if let Some((seg, _)) = n.split_once('/') {
1394 pkg_segments.insert(seg.to_string());
1395 }
1396 }
1397 // LOCAL-OK gate (round-59 Design B): a field-obscuring rename may
1398 // only fire when every class referencing the candidate lives inside
1399 // the root file's family, so the rename perturbs exactly ONE rendered
1400 // file. Ungated (Design A) the rule netted reqable −1653 / lark −5221
1401 // but weibo +36916: renames there flipped ambiguous type-vs-package
1402 // resolutions across thousands of pre-existing conflict families and
1403 // javac's error recovery turned "missing supertype, body suppressed"
1404 // files into full Object cascades. Lazy (progressive-browse) pools
1405 // skip the gate AND the rule — the multi-second image scan must not
1406 // stall single-class queries; the full-decompile pipeline always
1407 // materializes everything before installing renames.
1408 // Round-59 Design B stands: ungated field-clash renames were
1409 // retried AFTER the package-obscuring rules landed (weibo gate,
1410 // DDC_FC_UNGATED A/B) and still exploded to the 500k error cap —
1411 // fifth falsification of the ungated rename route.
1412 let local_ok: jdc_core::FxHashSet<String> = if pool_majority_materialized(pool) {
1413 refscan::local_ok(&pool.dexes, &field_clash_cands(pool, map))
1414 } else {
1415 jdc_core::FxHashSet::default()
1416 };
1417 let mut taken_sibling_displays: jdc_core::FxHashSet<String> =
1418 map.values().cloned().collect();
1419 // Stage-2 queue: nested tails equal to a top-level package segment
1420 // whose family showed NO descriptor-level trigger — the victim ref
1421 // may live in method bodies only (weixin e32/a2's `a.b.c()` bound
1422 // to its own nested `class a`, 位置: 类 a ×169).
1423 let mut body_cands: Vec<(String, String, String)> = Vec::new();
1424 for name in names {
1425 if !name.contains('$') || map.contains_key(name) {
1426 continue;
1427 }
1428 let Some(parent) = find_outer_name(pool, name) else {
1429 continue;
1430 };
1431 // Only a REAL `parent$rest` name renders as an inline member;
1432 // anonymous/local/lambda tails keep their flat own-file names.
1433 let Some(rest) = name
1434 .strip_prefix(parent.as_str())
1435 .and_then(|t| t.strip_prefix('$'))
1436 else {
1437 continue;
1438 };
1439 if !clean_member_tail(rest) {
1440 continue;
1441 }
1442 // Display simple names of every enclosing level, bottom-up; `cur`
1443 // ends at the ROOT (the top-level class whose file renders this
1444 // nested type — the scope the renamed tail lives in).
1445 let mut chain: Vec<String> = Vec::new();
1446 let mut cur = parent.clone();
1447 loop {
1448 let disp = map.get(&cur).cloned().unwrap_or_else(|| cur.clone());
1449 let simple = disp.rsplit('/').next().unwrap_or(&disp);
1450 chain.push(simple.rsplit('$').next().unwrap_or(simple).to_string());
1451 match find_outer_name(pool, &cur) {
1452 Some(o) if o != cur => cur = o,
1453 _ => break,
1454 }
1455 }
1456 let root: &str = &cur;
1457 let disp_parent = display_of(&parent, map);
1458 let tail = rest.rsplit('$').next().unwrap_or(rest);
1459 let orphan = rest.contains('$');
1460 // Field obscuring (JLS 6.4.2): a nested type whose simple name
1461 // equals an enclosing-class field is OBSCURED by it in a
1462 // `<Parent>.<tail>` qualified reference — `io/flutter/view/g$i`
1463 // (a `static enum i`) renders `g.i.t`, but g also declares an
1464 // instance field `i`, so javac resolves `g.i` to the field:
1465 // "无法从静态上下文中引用非静态 变量 i" (975 of 1140 reqable
1466 // static-context errors). The fix renames the NESTED TYPE (the
1467 // registry carries every render path: shorten, inner_simple —
1468 // whose bypass was the round-55 blowup, fixed in jdc-core —
1469 // classdec declarations), gated to LOCAL-OK candidates by the
1470 // refscan above. PoC: in-file rename of flutter g's `enum i`
1471 // took the class from 2452 errors to 1.
1472 let field_clash = local_ok.contains(name.as_str());
1473 // Package shadow (JLS 6.4.2 at nested level): a nested type
1474 // DISPLAYED with a simple name equal to the first segment of a
1475 // package its FAMILY references binds every `s.x` qualified ref
1476 // in the root file to the member type — the family's own outer
1477 // rename can EXPOSE this (e/b → e/b2 un-collided the nested `b`,
1478 // which then hijacked `b.b2` — 循环继承). The round-47 candidate
1479 // guard only kept RENAMED tails off package segments; pre-
1480 // existing shadows need the rename too, family-gated: fire only
1481 // when this family's own descriptors reference the segment.
1482 // Orphans INCLUDED: an `Outer$$x` orphan normalizes to the
1483 // nested display `x` (k=1 invisible rename), and a clean tail
1484 // renders BARE in the family file — same package shadow as a
1485 // true nested (weixin li/u0$$r → `class r` hijacked `r.a` for
1486 // root package r).
1487 let pkg_shadow = fam_segs
1488 .get(root)
1489 .is_some_and(|segs| segs.contains(tail));
1490 let clash = chain.iter().any(|c| c == tail) || field_clash || pkg_shadow;
1491 // Stage-2 queue: NO descriptor trigger, but the tail equals a
1492 // top-level package segment — the victim ref may be body-level
1493 // only. ORPHANS included: their pass-1 k=1 mint is the invisible
1494 // normalization (`a$$a` → display `a`), which leaves the package
1495 // shadow fully alive (weixin e32/a$$a hijacking `a.b.c()` ×169).
1496 if !clash && pkg_segments.contains(tail) {
1497 body_cands.push((name.clone(), root.to_string(), tail.to_string()));
1498 }
1499 if !orphan && !clash {
1500 continue;
1501 }
1502 // The enclosing class's field names (sanitized like the renderer)
1503 // — only needed once a rename actually fires, to keep the new
1504 // tail from being obscured again. Computed lazily so progressive
1505 // pools don't materialize parents for rules that don't rename.
1506 let mut enc_fields: jdc_core::FxHashSet<String> = jdc_core::FxHashSet::default();
1507 if field_clash || pkg_shadow {
1508 if let Some(pc) = pool.get(&parent) {
1509 for f in pc.static_fields.iter().chain(pc.instance_fields.iter()) {
1510 enc_fields.insert(crate::classdec::java_ident(&f.name).into_owned());
1511 }
1512 }
1513 }
1514 let mut k = 0u32;
1515 loop {
1516 k += 1;
1517 let cand_tail = if k == 1 {
1518 tail.to_string()
1519 } else {
1520 format!("{tail}{k}")
1521 };
1522 // The candidate must clear every ancestor display name (javac
1523 // checks the whole chain) AND every enclosing field name (else
1524 // the renamed type is just obscured again). Sibling nested
1525 // tails are covered by the pool.has_name check below.
1526 if chain.contains(&cand_tail) || enc_fields.contains(&cand_tail) {
1527 continue;
1528 }
1529 // Package shadow: the renamed tail becomes a member type of
1530 // the root file's class, outranking any package of the same
1531 // name for qualified refs in that file. Same-package top-level
1532 // classes likewise outrank-then-break in-file simple refs.
1533 // BOTH checks apply to field-clash renames only: for the
1534 // orphan rule the k=1 candidate is the natural display name
1535 // (`g0$$r` → `g0$r` renders identically — an invisible rename),
1536 // and rejecting it forced weixin's whole `g0$$X` Runnable
1537 // family to visible `X17` renames, breaking every lucky
1538 // `g0.r`-style resolution corpus-wide (+2911 `变量 r` — the
1539 // orphan rule is ungated by design, predating LOCAL-OK).
1540 if field_clash || pkg_shadow {
1541 if pkg_segments.contains(&cand_tail) {
1542 continue;
1543 }
1544 if pkg_shadow
1545 && fam_segs
1546 .get(root)
1547 .is_some_and(|segs| segs.contains(cand_tail.as_str()))
1548 {
1549 continue;
1550 }
1551 if let Some((root_pkg, _)) = root.rsplit_once('/') {
1552 let sib = format!("{root_pkg}/{cand_tail}");
1553 if pool.has_name(&sib) || taken_sibling_displays.contains(&sib)
1554 {
1555 // A top-level sibling exists — or an EARLIER rule
1556 // minted that display (obscuring_class_renames
1557 // moved ssosdk/b → ssosdk/b2 while this loop was
1558 // about to mint nested $b → b2: two `b2`s in one
1559 // scope, member type outranks the package sibling
1560 // and hijacked every bare `b2.b(..)` call).
1561 continue;
1562 }
1563 taken_sibling_displays.insert(sib);
1564 }
1565 }
1566 let cand = format!("{disp_parent}${cand_tail}");
1567 if pool.has_name(&cand) || assigned.contains(&cand) {
1568 continue;
1569 }
1570 assigned.insert(cand.clone());
1571 map.insert(name.clone(), cand);
1572 break;
1573 }
1574 }
1575 // Stage 2 (body-gated package shadow): one body-level image scan
1576 // over the queued families (family_ref_segments — the field-
1577 // deshadow mechanism; instruction-level, cross-package refs), then
1578 // mint for the families whose members actually reference the
1579 // segment. Mirrors obscuring_class_renames' two-stage design at
1580 // family level. Lazy pools skip (same gate as stage 1's scan).
1581 let mut body_renamed = 0usize;
1582 if !body_cands.is_empty() && pool_majority_materialized(pool) {
1583 let roots: jdc_core::FxHashSet<&str> =
1584 body_cands.iter().map(|(_, r, _)| r.as_str()).collect();
1585 let mut members: jdc_core::FxHashSet<String> = jdc_core::FxHashSet::default();
1586 for n in &pool.order {
1587 let r = match n.find('$') {
1588 Some(i) => &n[..i],
1589 None => n.as_str(),
1590 };
1591 if roots.contains(r) {
1592 members.insert(n.clone());
1593 }
1594 }
1595 let t0 = std::time::Instant::now();
1596 let (by_referrer, _) = refscan::family_ref_segments(&pool.dexes, &members);
1597 // Referrer-keyed → family-root-keyed: every member renders in
1598 // the root file's scope.
1599 let mut root_segs: jdc_core::FxHashMap<&str, jdc_core::FxHashSet<&str>> =
1600 jdc_core::FxHashMap::default();
1601 for (referrer, segs) in &by_referrer {
1602 let r = referrer.split('$').next().unwrap_or(referrer);
1603 root_segs.entry(r).or_default().extend(segs.iter().map(|s| s.as_str()));
1604 }
1605 if std::env::var("DDC_STATS").is_ok() {
1606 eprintln!(
1607 "[renames] nested body-shadow scan: cands={} families={} scan={:?}",
1608 body_cands.len(),
1609 root_segs.len(),
1610 t0.elapsed()
1611 );
1612 }
1613 for (name, root, tail) in &body_cands {
1614 if !root_segs
1615 .get(root.as_str())
1616 .is_some_and(|s| s.contains(tail.as_str()))
1617 {
1618 continue;
1619 }
1620 if let Some(d) = map.get(name) {
1621 // Pass-1 orphan mints are invisible (display tail ==
1622 // tail) — those get re-minted visibly here. Anything
1623 // already visibly renamed stays as the earlier rule set it.
1624 let cur_tail = d.rsplit(['/', '$']).next().unwrap_or("");
1625 if cur_tail != tail.as_str() {
1626 continue;
1627 }
1628 }
1629 // Fresh display context: stage-1 renames may have moved the
1630 // parent/chain since this candidate was queued.
1631 let Some(parent) = find_outer_name(pool, name) else {
1632 continue;
1633 };
1634 let Some(rest) = name
1635 .strip_prefix(parent.as_str())
1636 .and_then(|t| t.strip_prefix('$'))
1637 else {
1638 continue;
1639 };
1640 if !clean_member_tail(rest) {
1641 continue;
1642 }
1643 let mut chain: Vec<String> = Vec::new();
1644 let mut cur = parent.clone();
1645 loop {
1646 let disp = map.get(&cur).cloned().unwrap_or_else(|| cur.clone());
1647 let simple = disp.rsplit('/').next().unwrap_or(&disp);
1648 chain.push(simple.rsplit('$').next().unwrap_or(simple).to_string());
1649 match find_outer_name(pool, &cur) {
1650 Some(o) if o != cur => cur = o,
1651 _ => break,
1652 }
1653 }
1654 let disp_parent = display_of(&parent, map);
1655 let mut enc_fields: jdc_core::FxHashSet<String> =
1656 jdc_core::FxHashSet::default();
1657 if let Some(pc) = pool.get(&parent) {
1658 for f in pc.static_fields.iter().chain(pc.instance_fields.iter()) {
1659 enc_fields.insert(crate::classdec::java_ident(&f.name).into_owned());
1660 }
1661 }
1662 let mut k = 0u32;
1663 loop {
1664 k += 1;
1665 let cand_tail = if k == 1 {
1666 tail.clone()
1667 } else {
1668 format!("{tail}{k}")
1669 };
1670 if chain.contains(&cand_tail) || enc_fields.contains(&cand_tail) {
1671 continue;
1672 }
1673 // The renamed tail stays a member type of the root
1674 // file — it must not shadow ANOTHER package segment the
1675 // family references (descriptor OR body level) nor a
1676 // same-package top-level sibling.
1677 if pkg_segments.contains(&cand_tail) {
1678 continue;
1679 }
1680 if fam_segs
1681 .get(root.as_str())
1682 .is_some_and(|segs| segs.contains(cand_tail.as_str()))
1683 {
1684 continue;
1685 }
1686 if root_segs
1687 .get(root.as_str())
1688 .is_some_and(|segs| segs.contains(cand_tail.as_str()))
1689 {
1690 continue;
1691 }
1692 if let Some((root_pkg, _)) = root.rsplit_once('/') {
1693 let sib = format!("{root_pkg}/{cand_tail}");
1694 if pool.has_name(&sib) || taken_sibling_displays.contains(&sib) {
1695 continue;
1696 }
1697 taken_sibling_displays.insert(sib);
1698 }
1699 let cand = format!("{disp_parent}${cand_tail}");
1700 if pool.has_name(&cand) || assigned.contains(&cand) {
1701 continue;
1702 }
1703 // Repair pass-1 mints whose display chain ran through
1704 // THIS name's old display: stage 2 renames ancestors
1705 // after descendants were minted (weibo mobileads k$a →
1706 // a5 left pass-1's k$a$b$a → "k$a$b$a2" pointing through
1707 // the dead `a` — 找不到符号 类 a ×48). Ancestor-first
1708 // within each pass, but stage 2 as a whole runs last.
1709 let old_disp = format!("{disp_parent}${tail}");
1710 if old_disp != cand {
1711 for (mk, mv) in map.iter_mut() {
1712 if mk == name {
1713 continue;
1714 }
1715 if let Some(rest) = mv.strip_prefix(old_disp.as_str()) {
1716 if rest.starts_with('$') {
1717 let repaired = format!("{cand}{rest}");
1718 assigned.insert(repaired.clone());
1719 *mv = repaired;
1720 }
1721 }
1722 }
1723 }
1724 assigned.insert(cand.clone());
1725 map.insert(name.clone(), cand);
1726 body_renamed += 1;
1727 break;
1728 }
1729 }
1730 }
1731 if body_renamed > 0 && std::env::var("DDC_STATS").is_ok() {
1732 eprintln!("[renames] nested body-shadow renames: {body_renamed}");
1733 }
1734}
1735
1736/// Field-obscuring rename candidates: nested member types whose simple
1737/// name equals a field of the enclosing class, mapped to the top-level
1738/// class whose file renders them (the LOCAL-OK family root). Mirrors the
1739/// skip conditions of `nested_collision_renames` (same-name entries must
1740/// agree, or the gate proves locality for a rename that never fires — or
1741/// worse, skips one that does).
1742fn field_clash_cands(pool: &DexPool, map: &HashMap<String, String>) -> HashMap<String, String> {
1743 let mut out: HashMap<String, String> = HashMap::default();
1744 for name in &pool.order {
1745 if !name.contains('$') || map.contains_key(name) {
1746 continue;
1747 }
1748 let Some(parent) = find_outer_name(pool, name) else {
1749 continue;
1750 };
1751 let Some(rest) = name
1752 .strip_prefix(parent.as_str())
1753 .and_then(|t| t.strip_prefix('$'))
1754 else {
1755 continue;
1756 };
1757 if !clean_member_tail(rest) {
1758 continue;
1759 }
1760 let tail = rest.rsplit('$').next().unwrap_or(rest);
1761 let Some(pc) = pool.get(&parent) else {
1762 continue;
1763 };
1764 let clash = pc
1765 .static_fields
1766 .iter()
1767 .chain(pc.instance_fields.iter())
1768 .any(|f| crate::classdec::java_ident(&f.name).as_ref() == tail);
1769 if !clash {
1770 continue;
1771 }
1772 let mut root = parent.clone();
1773 loop {
1774 match find_outer_name(pool, &root) {
1775 Some(o) if o != root => root = o,
1776 _ => break,
1777 }
1778 }
1779 out.insert(name.clone(), root);
1780 }
1781 out
1782}
1783
1784/// Whether the pool is (majority-)materialized: full-decompile pipelines
1785/// call `materialize_all` before installing renames, progressive browse
1786/// pools materialize a handful of classes on demand. The refscan gate
1787/// must not run (or stall startup) in the progressive case.
1788fn pool_majority_materialized(pool: &DexPool) -> bool {
1789 let (mut mat, mut lazy) = (0usize, 0usize);
1790 for e in pool.classes.values() {
1791 match e {
1792 ClassEntry::Eager(_) => mat += 1,
1793 ClassEntry::Lazy { pc, .. } => {
1794 if pc.get().is_some() {
1795 mat += 1;
1796 } else {
1797 lazy += 1;
1798 }
1799 }
1800 }
1801 }
1802 lazy == 0 || mat >= lazy
1803}
1804
1805/// Compute and install the registry (call before worker threads spawn).
1806/// CURRENT resident size (MB) — ru_maxrss is monotone (historical peak),
1807/// so purge effects are invisible there; live-vs-retained attribution
1808/// needs the instantaneous figure (macOS mach task_info).
1809#[cfg(target_vendor = "apple")]
1810fn current_rss_mb() -> f64 {
1811 #[repr(C)]
1812 struct TaskBasicInfo {
1813 virtual_size: u64,
1814 resident_size: u64,
1815 resident_size_max: u64,
1816 // time_value_t = {i32 secs, i32 usec} = 8B each.
1817 total_user_time: u64,
1818 total_sys_time: u64,
1819 policy: i32,
1820 suspend_count: i32,
1821 }
1822 extern "C" {
1823 fn mach_task_self() -> u32;
1824 fn task_info(
1825 target_port: u32,
1826 flavor: i32,
1827 task_info_out: *mut TaskBasicInfo,
1828 task_info_out_cnt: *mut u32,
1829 ) -> i32;
1830 }
1831 let mut info = TaskBasicInfo {
1832 virtual_size: 0,
1833 resident_size: 0,
1834 resident_size_max: 0,
1835 total_user_time: 0,
1836 total_sys_time: 0,
1837 policy: 0,
1838 suspend_count: 0,
1839 };
1840 // count is in natural_t (4B) units; a smaller-than-kernel version
1841 // still fills the prefix fields we read.
1842 let mut cnt = (std::mem::size_of::<TaskBasicInfo>() / 4) as u32;
1843 let kr = unsafe { task_info(mach_task_self(), 20, &mut info, &mut cnt) };
1844 if kr == 0 {
1845 info.resident_size as f64 / (1024.0 * 1024.0)
1846 } else {
1847 0.0
1848 }
1849}
1850#[cfg(not(target_vendor = "apple"))]
1851fn current_rss_mb() -> f64 {
1852 0.0
1853}
1854
1855/// ru_maxrss peak so far (MB). macOS reports bytes, Linux KB; Windows
1856/// has no getrusage (the old not(apple) fallback referenced it and
1857/// failed the MSVC link — LNK2019 — in the v0.1.18 CI).
1858#[cfg(unix)]
1859fn peak_rss_mb() -> f64 {
1860 #[repr(C)]
1861 struct Rusage {
1862 pad: [i64; 4],
1863 maxrss: i64,
1864 // Oversized on purpose: the kernel writes sizeof(struct rusage)
1865 // (~144B on macOS) regardless of what we read back.
1866 tail: [i64; 16],
1867 }
1868 extern "C" {
1869 fn getrusage(who: i32, r: *mut Rusage) -> i32;
1870 }
1871 let mut r = Rusage {
1872 pad: [0; 4],
1873 maxrss: 0,
1874 tail: [0; 16],
1875 };
1876 unsafe { getrusage(0, &mut r) };
1877 #[cfg(target_vendor = "apple")]
1878 let scale = 1024.0 * 1024.0;
1879 #[cfg(not(target_vendor = "apple"))]
1880 let scale = 1024.0;
1881 r.maxrss as f64 / scale
1882}
1883#[cfg(not(unix))]
1884fn peak_rss_mb() -> f64 {
1885 0.0
1886}
1887
1888/// DDC_STATS phase attribution of the RSS high-water mark: peak AND
1889/// instantaneous live (ru_maxrss is monotone, so purge effects are
1890/// invisible there).
1891pub(crate) fn stats_rss(phase: &str) {
1892 if std::env::var("DDC_STATS").is_ok() {
1893 eprintln!(
1894 "[rss] {phase}: peak {:.0} MB, live {:.0} MB",
1895 peak_rss_mb(),
1896 current_rss_mb()
1897 );
1898 }
1899}
1900
1901pub fn install_case_renames(pool: &DexPool) {
1902 let dbg = std::env::var("DDC_DBG_REN").is_ok();
1903 let mut tick = std::time::Instant::now();
1904 let mut lap = |tag: &str| {
1905 if dbg {
1906 eprintln!("[ren] {:?} {}", tick.elapsed(), tag);
1907 tick = std::time::Instant::now();
1908 }
1909 };
1910 let mut map = case_rename_map(pool);
1911 lap("case_rename_map");
1912 crate::stats_rss(" case_rename_map");
1913 // Cross-package reference first segments (descriptor level). Lazy
1914 // (progressive-browse) pools skip the scan — materializing every
1915 // class must not stall single-class queries; the obscuring rules
1916 // then simply never fire (browse mode has no whole-program javac).
1917 let (pkg_segs, fam_segs) = if pool_majority_materialized(pool) {
1918 ref_segments(pool)
1919 } else {
1920 (
1921 jdc_core::FxHashMap::default(),
1922 jdc_core::FxHashMap::default(),
1923 )
1924 };
1925 // Package-leaf shadows FIRST: the nested-collision renames compute
1926 // display chains off the (renamed) parent display names — running
1927 // them before the parent rename left both rules minting the same
1928 // display (`a2` twice in weibo's AIDL families).
1929 crate::stats_rss(" ref_segments");
1930 lap("ref_segments");
1931 pkg_leaf_shadow_renames(pool, &mut map);
1932 lap("pkg_leaf_shadow");
1933 class_pkg_collision_renames(pool, &mut map);
1934 lap("class_pkg_collision");
1935 obscuring_class_renames(pool, &mut map, &pkg_segs);
1936 lap("obscuring");
1937 nested_collision_renames(pool, &mut map, &fam_segs);
1938 lap("nested_collision");
1939 // LAST: lossy-sanitize collisions key on the display form every rule
1940 // above has already settled (`l.᩻ܶ` → `l/__.java`, 22,636 classes onto
1941 // 3 paths on bin.mt.plus — the silent-overwrite data loss).
1942 crate::stats_rss(" shadow+collision+obscuring+nested rules");
1943 lap("rules");
1944 lossy_sanitize_renames(pool, &mut map);
1945 // Default-package relocation keys on the settled displays too (it
1946 // PREFIXES them with the synthetic package) — runs after every
1947 // simple-name rule has minted.
1948 root_pkg_relocation(pool, &mut map);
1949 fw_shadow_relocation(pool, &mut map);
1950 crate::stats_rss(" lossy+root-pkg rules");
1951 lap("lossy+root");
1952 jdc_core::rename::set_class_renames(map);
1953 crate::stats_rss("class-renames installed");
1954 lap("set_class_renames");
1955 // Field-rename registry build and the access-widening image scan are
1956 // independent: both only READ the pool, they install into disjoint
1957 // statics (FIELD_RENAMES vs WIDEN), the WIDEN accessors are
1958 // is_some_and-guarded with no reader before render time, and lazy
1959 // materialization is OnceLock-safe. Run them concurrently — the 1.6s
1960 // QQ widening scan fully overlaps the field-rename build.
1961 if pool_majority_materialized(pool) {
1962 std::thread::scope(|s| {
1963 let h = s.spawn(|| crate::classdec::install_access_widening(pool));
1964 jdc_core::rename::set_field_renames(combined_field_renames(pool));
1965 let _ = h.join();
1966 });
1967 crate::stats_rss("field-renames + widening installed");
1968 lap("field+widen");
1969 } else {
1970 jdc_core::rename::set_field_renames(combined_field_renames(pool));
1971 lap("field_renames(lazy)");
1972 }
1973}
1974
1975/// A top-level class `P/s` whose simple name equals the FIRST PACKAGE
1976/// SEGMENT of types that other classes in P reference cross-package.
1977/// Such a reference renders as the FQN `s.rest...`, and javac binds the
1978/// first segment to the nearest type in scope — the same-package class
1979/// `P/s` (JLS 6.4.2): `n91.f.a(x)` inside pc5/bd looks up member `f` OF
1980/// CLASS pc5.n91 (weixin's n0/n91/uc6/vc6 protobuf cluster: ~12k root
1981/// files, ~90k cascade lines; weibo's `a.g.b.a.b` chains root the same
1982/// way — class a beside subpackage a/ inside package a/g/b). No Java
1983/// syntax escapes an obscured first segment (imports can bypass it, but
1984/// single-type imports are blocked exactly where obfuscated leaf names
1985/// collide with same-package siblings), so the class moves: the r66
1986/// rename mechanism at a new trigger. Gated on DESCRIPTOR-level refs of
1987/// the candidate's own package (supers/fields/protos) — body-only refs
1988/// are missed by design; the import layer covers unblocked leaves there.
1989/// Descriptor-level cross-package reference first segments, aggregated
1990/// per PACKAGE (top-level obscurers: every file in P shares the scope)
1991/// and per FAMILY root (nested obscurers: a nested simple name is in
1992/// scope only inside its own root file). Raw `L..;` scans — no JavaType
1993/// parsing. Body-level refs are missed by design (the import layer
1994/// catches unblocked leaves there); descriptors cover supers, field
1995/// types and method protos — the header positions whose breakage starts
1996/// the error-type cascades.
1997/// Package (or family) → first segments of the cross-package types its
1998/// classes reference at descriptor level. Obscuring-rename trigger sets.
1999type SegMap<'a> = jdc_core::FxHashMap<&'a str, jdc_core::FxHashSet<&'a str>>;
2000
2001fn ref_segments(pool: &DexPool) -> (SegMap<'_>, SegMap<'_>) {
2002 fn pkg_of(n: &str) -> &str {
2003 match n.rsplit_once('/') {
2004 Some((p, _)) => p,
2005 None => "",
2006 }
2007 }
2008 fn note<'a>(
2009 pkg_segs: &mut jdc_core::FxHashMap<&'a str, jdc_core::FxHashSet<&'a str>>,
2010 fam_segs: &mut jdc_core::FxHashMap<&'a str, jdc_core::FxHashSet<&'a str>>,
2011 pkg: &'a str,
2012 fam: &'a str,
2013 t: &'a str,
2014 ) {
2015 if !t.contains('/') {
2016 return; // root-package target: unreferenceable from named pkgs
2017 }
2018 let _ = pkg; // same-package refs COUNT: ddc renders headers/field
2019 // types fully qualified (print_class_name), so `a.g.b.a.b`
2020 // inside package a/g/b is obscured by class a/g/b/a just
2021 // like a cross-package FQN (weibo's whole `a`-tree family).
2022 let seg = t.split('/').next().unwrap_or("");
2023 if !seg.is_empty() {
2024 pkg_segs.entry(pkg).or_default().insert(seg);
2025 fam_segs.entry(fam).or_default().insert(seg);
2026 }
2027 }
2028 fn scan<'a>(
2029 pkg_segs: &mut jdc_core::FxHashMap<&'a str, jdc_core::FxHashSet<&'a str>>,
2030 fam_segs: &mut jdc_core::FxHashMap<&'a str, jdc_core::FxHashSet<&'a str>>,
2031 pkg: &'a str,
2032 fam: &'a str,
2033 desc: &'a str,
2034 ) {
2035 let b = desc.as_bytes();
2036 let mut i = 0usize;
2037 while i < b.len() {
2038 if b[i] == b'L' {
2039 if let Some(end) = desc[i + 1..].find(';') {
2040 note(pkg_segs, fam_segs, pkg, fam, &desc[i + 1..i + 1 + end]);
2041 i += end + 2;
2042 continue;
2043 }
2044 }
2045 i += 1;
2046 }
2047 }
2048 let mut pkg_segs: jdc_core::FxHashMap<&str, jdc_core::FxHashSet<&str>> =
2049 jdc_core::FxHashMap::default();
2050 let mut fam_segs: jdc_core::FxHashMap<&str, jdc_core::FxHashSet<&str>> =
2051 jdc_core::FxHashMap::default();
2052 for name in &pool.order {
2053 let Some(c) = pool.get(name) else { continue };
2054 let pkg = pkg_of(name);
2055 let fam = match name.find('$') {
2056 Some(i) => &name[..i],
2057 None => name.as_str(),
2058 };
2059 if let Some(sup) = &c.super_name {
2060 note(&mut pkg_segs, &mut fam_segs, pkg, fam, sup);
2061 }
2062 for sup in c.interfaces.iter() {
2063 note(&mut pkg_segs, &mut fam_segs, pkg, fam, sup);
2064 }
2065 for f in c.static_fields.iter().chain(c.instance_fields.iter()) {
2066 scan(&mut pkg_segs, &mut fam_segs, pkg, fam, &f.desc);
2067 }
2068 for m in c.direct_methods.iter().chain(c.virtual_methods.iter()) {
2069 scan(&mut pkg_segs, &mut fam_segs, pkg, fam, &m.desc);
2070 }
2071 }
2072 (pkg_segs, fam_segs)
2073}
2074
2075/// A top-level class `P/s` whose simple name equals the FIRST PACKAGE
2076/// SEGMENT of types that classes in P reference cross-package. Such a
2077/// reference renders as the FQN `s.rest...`, and javac binds the first
2078/// segment to the nearest type in scope — the same-package class `P/s`
2079/// (JLS 6.4.2, no package fallback — verified): `n91.f.a(x)` inside
2080/// pc5/bd looks up member `f` OF CLASS pc5.n91 (weixin's n0/n91/uc6/vc6
2081/// protobuf cluster: ~12k root files, ~90k cascade lines; weibo's
2082/// `a.g.b.a.b` chains root the same way). No Java syntax escapes an
2083/// obscured first segment, so the class moves: the r66 rename mechanism
2084/// at a new trigger, gated on DESCRIPTOR-level refs of the candidate's
2085/// package (body-only refs go to the import layer). Nested obscurers
2086/// (a nested simple name shadowing a package inside its family file)
2087/// are the companion trigger in nested_collision_renames.
2088fn obscuring_class_renames(
2089 pool: &DexPool,
2090 map: &mut HashMap<String, String>,
2091 pkg_segs: &SegMap,
2092) {
2093 fn pkg_of(n: &str) -> &str {
2094 match n.rsplit_once('/') {
2095 Some((p, _)) => p,
2096 None => "",
2097 }
2098 }
2099 // Root package first segments (segments some class lives under).
2100 let mut root_segs: jdc_core::FxHashSet<&str> = jdc_core::FxHashSet::default();
2101 for n in &pool.order {
2102 if let Some(i) = n.find('/') {
2103 root_segs.insert(&n[..i]);
2104 }
2105 }
2106 let mut cands: Vec<&String> = Vec::new();
2107 for name in &pool.order {
2108 let simple = name.rsplit('/').next().unwrap_or(name);
2109 let pkg = pkg_of(name);
2110 if simple.is_empty() || simple.contains('$') || pkg.is_empty() {
2111 continue;
2112 }
2113 if !root_segs.contains(simple) {
2114 continue;
2115 }
2116 if map.get(name).is_some_and(|v| v != name) {
2117 continue; // an earlier rule moved it
2118 }
2119 // NB: NO "class lives under the s tree" exclusion. weibo's
2120 // a/a/b/c/l/a shadowed the FQN refs `a.a.a.a2` that its own
2121 // package files render for cross-subpackage targets (the
2122 // starts_with("a/") skip left the WHOLE a-tree unrenamed —
2123 // 找不到符号 类 a chains). The two-stage refsegs gates below
2124 // require an actual cross-package FQN ref for the trigger.
2125 cands.push(name);
2126 }
2127 let ncands = cands.len();
2128 if std::env::var("DDC_STATS").is_ok() {
2129 eprintln!(
2130 "[renames] obscuring candidates={ncands} pkgs-with-segs={}",
2131 pkg_segs.len()
2132 );
2133 }
2134 if cands.is_empty() {
2135 return;
2136 }
2137 // Fast collision oracle: sorted names + prefix range scan (the r66
2138 // candidate loop's `order.iter().any(starts_with)` is O(n) per
2139 // candidate — this rule can mint tens of thousands of renames).
2140 let mut sorted: Vec<&String> = pool.order.iter().collect();
2141 sorted.sort_unstable();
2142 let prefix_free = |cand: &str| -> bool {
2143 // No class named cand, nothing nested under cand$..., no package cand/...
2144 let idx = sorted.partition_point(|n| n.as_str() < cand);
2145 !sorted[idx..].first().is_some_and(|n| n.starts_with(cand))
2146 };
2147 let pkg_simples = pool.package_simples();
2148 let mut taken_displays: jdc_core::FxHashSet<String> = map.values().cloned().collect();
2149 let mut renamed = 0usize;
2150 // Stage 1: descriptor-level hits rename now; the rest queue for the
2151 // BODY-level image scan (weixin pc5's `invoke-static Ln91/f;.a` —
2152 // 7.8k+6.6k error lines the descriptor gate cannot see).
2153 let mut stage2: Vec<&String> = Vec::new();
2154 for name in cands {
2155 let pkg = pkg_of(name);
2156 let simple = name.rsplit('/').next().unwrap_or(name);
2157 if !pkg_segs.get(pkg).is_some_and(|segs| segs.contains(simple)) {
2158 stage2.push(name);
2159 continue;
2160 }
2161 let siblings = pkg_simples.get(pkg);
2162 let mut k = 1u32;
2163 loop {
2164 k += 1;
2165 let new_simple = format!("{simple}{k}");
2166 let cand = if pkg.is_empty() {
2167 new_simple.clone()
2168 } else {
2169 format!("{pkg}/{new_simple}")
2170 };
2171 // Exact/prefix pool clash, another rule's display target, the
2172 // package's own referenced segments (the new name must not
2173 // obscure ANOTHER package the same files reference), or a
2174 // case-insensitive clash with a same-package sibling
2175 // (case-insensitive filesystems fold the files).
2176 let clash = !prefix_free(&cand)
2177 || taken_displays.contains(&cand)
2178 || pkg_segs
2179 .get(pkg)
2180 .is_some_and(|segs| segs.contains(new_simple.as_str()))
2181 || siblings.is_some_and(|s| {
2182 s.iter()
2183 .any(|p| p.eq_ignore_ascii_case(&new_simple) && *p != simple)
2184 });
2185 if !clash {
2186 taken_displays.insert(cand.clone());
2187 map.insert(name.clone(), cand);
2188 renamed += 1;
2189 break;
2190 }
2191 }
2192 }
2193 let mut body_renamed = 0usize;
2194 if !stage2.is_empty() && pool_majority_materialized(pool) {
2195 let stage2_pkgs: jdc_core::FxHashSet<String> =
2196 stage2.iter().map(|n| pkg_of(n).to_string()).collect();
2197 let body_segs = crate::refscan::body_ref_segments(&pool.dexes, &stage2_pkgs);
2198 for name in stage2 {
2199 let pkg = pkg_of(name);
2200 let simple = name.rsplit('/').next().unwrap_or(name);
2201 if !body_segs
2202 .get(pkg)
2203 .is_some_and(|segs| segs.contains(simple))
2204 {
2205 continue;
2206 }
2207 let siblings = pkg_simples.get(pkg);
2208 let mut k = 1u32;
2209 loop {
2210 k += 1;
2211 let new_simple = format!("{simple}{k}");
2212 let cand = format!("{pkg}/{new_simple}");
2213 let clash = !prefix_free(&cand)
2214 || taken_displays.contains(&cand)
2215 || pkg_segs
2216 .get(pkg)
2217 .is_some_and(|segs| segs.contains(new_simple.as_str()))
2218 || body_segs
2219 .get(pkg)
2220 .is_some_and(|segs| segs.contains(new_simple.as_str()))
2221 || siblings.is_some_and(|s| {
2222 s.iter()
2223 .any(|p| p.eq_ignore_ascii_case(&new_simple) && *p != simple)
2224 });
2225 if !clash {
2226 taken_displays.insert(cand.clone());
2227 map.insert(name.clone(), cand);
2228 renamed += 1;
2229 body_renamed += 1;
2230 break;
2231 }
2232 }
2233 }
2234 }
2235 if std::env::var("DDC_STATS").is_ok() {
2236 eprintln!(
2237 "[renames] obscuring-class renames: {renamed} (body-gated {body_renamed}; pkgs-with-segs={}, cands={ncands})",
2238 pkg_segs.len()
2239 );
2240 }
2241}
2242
2243/// Class-level counterpart of the non-ASCII rule already applied to
2244/// MEMBERS by `member_collision_renames`, kept as the SAFETY NET behind
2245/// the injective `sanitize_fq`.
2246///
2247/// `sanitize_fq` escapes non-identifier characters as `_u<hex>`, so two
2248/// distinct classes normally reach two distinct file names. It cannot be
2249/// injective against *lookalikes*, because ident-safe ASCII must pass
2250/// through unchanged: a literal class named `_u1a7b` collides with the
2251/// class whose single character is U+1A7B, and the pre-existing keyword
2252/// repair makes `_do` collide with `do`. Without this pass those pairs
2253/// repeat the original disaster — the writer opens with `create_new`,
2254/// takes EEXIST, unlinks and rewrites, so one class is destroyed with
2255/// exit code 0 (on bin.mt.plus the lossy predecessor of this rule hid
2256/// 22,633 of 30,768 classes exactly this way).
2257///
2258/// Groups are keyed by the SANITIZED path — the exact string both the
2259/// declaration site (`sanitize_ref`) and the writer (`source_path`)
2260/// produce. The first (sorted) member of each group keeps its name; the
2261/// rest become `<sanitized>_2`, `<sanitized>_3`, … Pure ASCII, so a
2262/// minted name is a fixed point of the sanitizer: it cannot silently
2263/// join a second collision group, and the declaration, every reference
2264/// and the file name agree by construction through `apply_class_rename`.
2265///
2266/// Costs nothing on a healthy pool: one pass over the emission set, and
2267/// it returns before touching the registry when every key is unique.
2268/// Runs LAST in `install_case_renames`: it keys on the display form the
2269/// earlier rules have already settled.
2270fn lossy_sanitize_renames(pool: &DexPool, map: &mut HashMap<String, String>) {
2271 let names = top_level_classes(pool);
2272 // Current display of an emission-set name. `case_rename_map` has
2273 // already inserted an entry for every one of these (identity or
2274 // renamed) and an exact registry hit short-circuits
2275 // `apply_class_rename`, so `map` already holds the final display —
2276 // no `$`-prefix walk is needed here.
2277 fn key_of(map: &HashMap<String, String>, n: &str) -> String {
2278 let disp = map.get(n).map(|s| s.as_str()).unwrap_or(n);
2279 sanitize_internal(disp)
2280 }
2281 let mut counts: HashMap<String, u32> = HashMap::default();
2282 for n in &names {
2283 *counts.entry(key_of(map, n)).or_insert(0) += 1;
2284 }
2285 if counts.values().all(|c| *c == 1) {
2286 return;
2287 }
2288 // Every sanitized path in the POOL is off-limits for a minted name,
2289 // not just the emission set: an inlined nested class still owns the
2290 // display it renders under inside its outer's file.
2291 let mut taken: jdc_core::FxHashSet<String> = pool
2292 .order
2293 .iter()
2294 .map(|n| sanitize_internal(map.get(n).map(|s| s.as_str()).unwrap_or(n)))
2295 .collect();
2296 let mut sorted: Vec<&String> = names.iter().collect();
2297 sorted.sort();
2298 let mut seen: jdc_core::FxHashSet<String> = jdc_core::FxHashSet::default();
2299 let mut renamed = 0usize;
2300 for n in sorted {
2301 let key = key_of(map, n);
2302 if counts.get(&key).copied().unwrap_or(0) < 2 {
2303 continue;
2304 }
2305 if seen.insert(key.clone()) {
2306 continue; // first (sorted) member of the group keeps its name
2307 }
2308 let (pkg, simple) = match key.rsplit_once('/') {
2309 Some((p, s)) => (p, s),
2310 None => ("", key.as_str()),
2311 };
2312 let mut i = 1u32;
2313 loop {
2314 i += 1;
2315 let cand = if pkg.is_empty() {
2316 format!("{simple}_{i}")
2317 } else {
2318 format!("{pkg}/{simple}_{i}")
2319 };
2320 if !taken.insert(cand.clone()) {
2321 continue; // already owned by a real class or another mint
2322 }
2323 // The minted display IS the sanitized name, so
2324 // sanitize_internal(cand) == cand and the writer's file name
2325 // matches the declaration it writes into it.
2326 map.insert(n.clone(), cand);
2327 renamed += 1;
2328 break;
2329 }
2330 }
2331 if std::env::var("DDC_STATS").is_ok() {
2332 eprintln!("[renames] lossy-sanitize collisions renamed={renamed}");
2333 }
2334}
2335
2336/// A class named `P/s` while OTHER classes live under `P/s/` — the dex
2337/// namespace is flat, so obfuscators freely mint a class whose simple
2338/// name equals a subpackage of its own package (`ptr.a` +
2339/// `ptr.a.a`, weibo rsplay/ptr families). That namespace is
2340/// unrepresentable in Java: the declaration itself is
2341/// "类 a与带有相同名称的程序包冲突" and every qualified reference
2342/// through the segment becomes unresolvable (the 53.8k weibo
2343/// missing-class errors root here, with Object/getClass cascades on
2344/// top). Rename the class — the registry carries declarations and
2345/// references — with a candidate that cannot recreate the shape (not a
2346/// class, not a package, not a nested-class prefix, not another rule's
2347/// target).
2348fn class_pkg_collision_renames(pool: &DexPool, map: &mut HashMap<String, String>) {
2349 // Every package that exists in the pool — ALL ancestor prefixes of
2350 // every class path, not just direct parents: Java sees package
2351 // `z` as existing when any class lives at `z/a/b` (the weibo
2352 // `com.sina.weibo.z` interface collides with a subpackage-only
2353 // `z/` that has no directly-resident class).
2354 // Borrowed slices: the owned set allocated a fresh String per
2355 // ancestor prefix of every class (~600k allocations on weibo).
2356 let mut pkgs: jdc_core::FxHashSet<&str> = jdc_core::FxHashSet::default();
2357 for n in &pool.order {
2358 let mut rest = n.as_str();
2359 while let Some(i) = rest.rfind('/') {
2360 rest = &rest[..i];
2361 pkgs.insert(rest);
2362 }
2363 }
2364 // Exact-class and minted-display sets make the clash checks O(1):
2365 // the old `order.iter().any(starts_with(&cand))` scanned every name
2366 // per candidate per k-step (weibo: the second quadratic of the
2367 // single-class getclass path, ~1s) — and its prefix match also
2368 // counted SIBLING classes (`a/b2` matching `a/b23`) that map to
2369 // different files and never collide.
2370 let taken: jdc_core::FxHashSet<&str> = pool.order.iter().map(|s| s.as_str()).collect();
2371 let mut used: jdc_core::FxHashSet<String> = map.values().cloned().collect();
2372 for name in &pool.order {
2373 let Some((pkg, simple)) = name.rsplit_once('/') else {
2374 continue;
2375 };
2376 if simple.is_empty() || pkg.is_empty() || simple.contains('$') {
2377 // Nested classes render dotted (`Outer.a`) — no package
2378 // effect from their `$` tail.
2379 continue;
2380 }
2381 if !pkgs.contains(name.as_str()) || map.get(name).is_some_and(|v| v != name) {
2382 continue; // no subpackage under it, or an earlier rule moved it
2383 }
2384 let mut k = 1u32;
2385 loop {
2386 k += 1;
2387 let cand = format!("{pkg}/{simple}{k}");
2388 let clash = pkgs.contains(cand.as_str())
2389 || taken.contains(cand.as_str())
2390 || used.contains(&cand);
2391 if !clash {
2392 used.insert(cand.clone());
2393 map.insert(name.clone(), cand);
2394 break;
2395 }
2396 }
2397 }
2398}
2399
2400/// Obfuscators can name a class after its own package leaf (`package k;`
2401/// `class k extends EditText`). Every SAME-PACKAGE reference to `k.Anything`
2402/// then resolves the first segment to the CLASS, not the package —
2403/// javac looks for `Anything` as a member of class k ("找不到符号 class p2,
2404/// location: class k"). Renaming the class (display-level, the registry
2405/// carries declarations and references) makes the package the only
2406/// thing the leaf name resolves to.
2407fn pkg_leaf_shadow_renames(pool: &DexPool, map: &mut HashMap<String, String>) {
2408 let taken: jdc_core::FxHashSet<&str> = pool.order.iter().map(|s| s.as_str()).collect();
2409 let mut used: jdc_core::FxHashSet<String> = map.values().cloned().collect();
2410 // Classes per package in ONE pass — the per-candidate
2411 // `order.iter().filter(starts_with).count()` was a quadratic scan
2412 // on obfuscated corpora (weibo: 2124 leaf-shadow candidates × 98k
2413 // names ≈ 2×10⁸ memcmp — the dominant single-class getclass cost,
2414 // and paid again by every full run).
2415 let mut under_pkg: jdc_core::FxHashMap<&str, usize> = jdc_core::FxHashMap::default();
2416 for n in &pool.order {
2417 if let Some(i) = n.rfind('/') {
2418 *under_pkg.entry(&n[..i]).or_default() += 1;
2419 }
2420 }
2421 for name in &pool.order {
2422 let Some((pkg, simple)) = name.rsplit_once('/') else {
2423 continue;
2424 };
2425 if simple.is_empty() || pkg.is_empty() {
2426 continue;
2427 }
2428 let leaf = pkg.rsplit('/').next().unwrap_or(pkg);
2429 // class name == package leaf, and the package really exists
2430 // (more classes than just this one live under it).
2431 if simple != leaf {
2432 continue;
2433 }
2434 let siblings = under_pkg.get(pkg).copied().unwrap_or(0).saturating_sub(1);
2435 if siblings == 0 || map.get(name).is_some_and(|v| v != name) {
2436 continue; // already renamed by an earlier rule; identity anchors are fine to overwrite
2437 }
2438 let mut k = 1u32;
2439 loop {
2440 k += 1;
2441 let cand = format!("{pkg}/{simple}{k}");
2442 if !taken.contains(cand.as_str()) && !used.contains(&cand) {
2443 used.insert(cand.clone());
2444 map.insert(name.clone(), cand);
2445 break;
2446 }
2447 }
2448 }
2449}
2450
2451/// Member-level collision renames (fields AND methods of one class).
2452/// Two sources collapse onto one display name: obfuscators renaming a
2453/// synthetic outer reference (`this$0`) onto a real field's name, and
2454/// the identifier sanitizer mapping non-ASCII to `_` (whole
2455/// `ERROR_中文` families render one `ERROR________`).
2456/// Both are legal in bytecode (members resolve by index) and both are
2457/// javac "already defined" errors in source. Policy per colliding group
2458/// (fields keep the first non-synthetic name; a synthetic outer
2459/// reference restores `this$0`; everything else gets `2`/`3` suffixes;
2460/// method keys include erased parameter types):
2461/// Keys stay ORIGINAL (name, descriptor) so references resolve exactly;
2462/// only the rendered identifier changes.
2463/// Descriptor type string → internal class name (arrays/primitives →
2464/// None).
2465/// Can a method returning descriptor `m_ret` override an ancestor
2466/// declaration returning `r`? Equal, or a covariant subtype. ARRAY
2467/// returns must reach is_subtype in DESCRIPTOR form (it peels `[` for
2468/// element covariance): the old `strip L…;`-only comparison returned
2469/// false for `SavedState[] <: Object[]`, so Parcelable.Creator's real
2470/// `newArray(I)[LSavedState;` was judged an IMPOSTOR and renamed
2471/// (newArray3 — m_taken already held newArray + the bridge's newArray2),
2472/// breaking the interface override: "X$1不是抽象的, 并且未覆盖…
2473/// newArray(int)" (weibo ×504, battery ~789, rimet/news/uuyc ~655).
2474/// Object returns keep the internal-name form is_subtype's pool/fwdb
2475/// walk expects; an array m_ret against an object r resolves through
2476/// is_subtype's `sup == Object` / fwdb path (arrays ARE Objects).
2477fn ret_satisfies(pool: &DexPool, m_ret: &str, r: &str) -> bool {
2478 if m_ret == r {
2479 return true;
2480 }
2481 if m_ret.starts_with('[') {
2482 let sup: std::borrow::Cow<str> = if r.starts_with('[') {
2483 std::borrow::Cow::Borrowed(r)
2484 } else {
2485 match r.strip_prefix('L').and_then(|x| x.strip_suffix(';')) {
2486 Some(rc) => std::borrow::Cow::Borrowed(rc),
2487 None => return false, // primitive sup: only equality (above)
2488 }
2489 };
2490 return pool.is_subtype(m_ret, &sup);
2491 }
2492 if r.starts_with('[') {
2493 return false; // nothing but an array satisfies an array return
2494 }
2495 match (
2496 r.strip_prefix('L').and_then(|x| x.strip_suffix(';')),
2497 m_ret.strip_prefix('L').and_then(|x| x.strip_suffix(';')),
2498 ) {
2499 (Some(rc), Some(mc)) => pool.is_subtype(mc, rc),
2500 _ => false,
2501 }
2502}
2503
2504/// Raw name → (argsig, ret) JVM-descriptor slices of one class's own
2505/// virtual methods — the key space the collision queries use.
2506type AncMethods = jdc_core::FxHashMap<
2507 std::sync::Arc<str>,
2508 Vec<(std::sync::Arc<str>, std::sync::Arc<str>)>,
2509>;
2510
2511/// One provenance bucket of an analyzed class's ancestor chain (cached
2512/// layers, no merge — see `layer_rets`).
2513type AncLayers = Vec<std::sync::Arc<AncEntry>>;
2514
2515/// Parsed direct virtual methods of one ancestor class + its chain
2516/// edges, cached per class name for the member-collision walk. The
2517/// framework chains (View/Activity/Fragment…) and pool hub supers repeat
2518/// across thousands of analyzed classes — parsing each ancestor's
2519/// descriptors ONCE per run (instead of per analyzed subclass) removes
2520/// the dominant main-loop cost.
2521struct AncEntry {
2522 methods: AncMethods,
2523 /// Provenance bucket: FRAMEWORK ancestors (fwdb classes AND
2524 /// desugared `java.*`/`javax.*` pool copies — javac compiles against
2525 /// the android.jar originals, the covariant source truth) vs POOL
2526 /// ancestors (rendered RAW — the erased bridge is what satisfies the
2527 /// raw interface). The keeper preference flips on this axis (weibo
2528 /// Function0.invoke bridge must keep the name; WhatsApp SortedMap
2529 /// .keySet real override must).
2530 is_fw: bool,
2531 /// super + direct interfaces (chain continuation), Object-stripped.
2532 supers: Vec<std::sync::Arc<str>>,
2533}
2534
2535fn parse_anc_entry(pool: &DexPool, cn: &str) -> std::sync::Arc<AncEntry> {
2536 let mut methods: AncMethods = jdc_core::FxHashMap::default();
2537 let mut supers: Vec<std::sync::Arc<str>> = Vec::new();
2538 let is_fw;
2539 if let Some(ac) = pool.get_if_materialized(cn) {
2540 // Desugared `java.*`/`javax.*` copies in the dex are
2541 // pool-materialized, but javac compiles against the android.jar
2542 // ORIGINALS (covariant generics truth) — their declarations
2543 // belong to the fw bucket (WhatsApp bundles desugared
2544 // java.util.SortedMap; the pool bucket kept the
2545 // flattened-erasure bridge and renamed the real SortedSet
2546 // keySet).
2547 is_fw = cn.starts_with("java/") || cn.starts_with("javax/");
2548 for m in ac.all_methods() {
2549 if m.is_static() || &*m.name == "<init>" || &*m.name == "<clinit>"
2550 {
2551 continue;
2552 }
2553 let d: &str = &m.desc;
2554 let lo = d.find('(').map(|i| i + 1).unwrap_or(0);
2555 let hi = d.find(')').unwrap_or(d.len());
2556 methods
2557 .entry(m.name.clone())
2558 .or_default()
2559 .push((
2560 std::sync::Arc::from(&d[lo..hi]),
2561 std::sync::Arc::from(&d[hi + 1..]),
2562 ));
2563 }
2564 if let Some(s) = &ac.super_name {
2565 if s != "java/lang/Object" {
2566 supers.push(std::sync::Arc::from(s.as_str()));
2567 }
2568 }
2569 supers.extend(ac.interfaces.iter().map(|i| std::sync::Arc::from(i.as_str())));
2570 } else {
2571 // FRAMEWORK ancestors bind names too: a group member
2572 // implementing a framework declaration must keep the original
2573 // name (renaming WhatsApp's Map.keySet-style overrides broke
2574 // the interface chain — cannot-find:变量 ×18.6k). fwdb
2575 // descriptors are erased JVM form — directly comparable with
2576 // dex descs. A class that is NEITHER pool nor fwdb (phantom)
2577 // parses to an empty entry — a chain dead-end, same as before.
2578 is_fw = true;
2579 crate::fwdb::for_each_method(cn, |d, f| {
2580 if f & crate::fwdb::MF_STATIC != 0 {
2581 return;
2582 }
2583 let Some(open) = d.find('(') else {
2584 return;
2585 };
2586 let lo = open + 1;
2587 let hi = d.find(')').unwrap_or(0);
2588 // NOTE: hi == lo is the zero-arg case, not malformed; and
2589 // the name slice ends at `open` (a `d[..lo]` cut kept the
2590 // `(` — every fw key missed its gkey lookup).
2591 if hi == 0 || hi + 1 >= d.len() {
2592 return;
2593 }
2594 methods
2595 .entry(std::sync::Arc::from(&d[..open]))
2596 .or_default()
2597 .push((
2598 std::sync::Arc::from(&d[lo..hi]),
2599 std::sync::Arc::from(&d[hi + 1..]),
2600 ));
2601 });
2602 if let Some(sup) = crate::fwdb::super_of(cn) {
2603 if sup != "java/lang/Object" {
2604 supers.push(std::sync::Arc::from(sup));
2605 }
2606 }
2607 crate::fwdb::for_each_interface(cn, |i| supers.push(std::sync::Arc::from(i)));
2608 }
2609 std::sync::Arc::new(AncEntry {
2610 methods,
2611 is_fw,
2612 supers,
2613 })
2614}
2615
2616/// Rets declared for `(name, argsig)` by any layer of a chain — lazy,
2617/// no merge (every consumer is an "any ancestor declares this" query,
2618/// so a per-layer scan replaces the old per-class eager map build).
2619/// Alloc-free: `Arc<str>: Borrow<str>` keys the hash lookup by `&str`.
2620fn layer_rets<'a>(
2621 layers: &'a [std::sync::Arc<AncEntry>],
2622 name: &'a str,
2623 argsig: &'a str,
2624) -> impl Iterator<Item = &'a str> {
2625 layers
2626 .iter()
2627 .filter_map(move |e| e.methods.get(name))
2628 .flatten()
2629 .filter(move |(a, _)| &**a == argsig)
2630 .map(|(_, r)| &**r)
2631}
2632
2633fn member_collision_renames(
2634 pool: &DexPool,
2635) -> HashMap<std::sync::Arc<str>, Vec<jdc_core::rename::FieldRename>> {
2636 let mut out: HashMap<std::sync::Arc<str>, Vec<jdc_core::rename::FieldRename>> =
2637 HashMap::default();
2638 // Direct nested tails per base class (`C$a` → a for base C), for
2639 // the companion-field rule below.
2640 let mut child_tails: jdc_core::FxHashMap<&str, Vec<&str>> =
2641 jdc_core::FxHashMap::default();
2642 // Per-package RENAMED class display heads: a MINTED field display
2643 // equal to one of these shadows same-package qualified type refs
2644 // (`a2.changeQuickRedirect` binds to the field when the class a2's
2645 // display is a2 — the inherited rule minting field a→a2 ONTO the
2646 // pkg-renamed class display created exactly this: weibo +194
2647 // cannot-find). Keyed on the POST-rename display (what refs render),
2648 // covering each class's own display too.
2649 let root_segs = pool.root_pkg_segs();
2650 // Lazy single-class pools build the global tables from MATERIALIZED
2651 // classes only (the target + its on-demand family): unmateralized
2652 // siblings matter to whole-program javac, which a one-class stdout
2653 // dump does not have — the same reduced-fidelity contract the
2654 // obscuring rules already document for lazy pools. The walk over
2655 // every registered name was the dominant field-rename cost of
2656 // getclass (~0.5s on weibo for one class).
2657 let lazy = !pool_majority_materialized(pool);
2658 let mut pkg_displays: jdc_core::FxHashMap<&str, jdc_core::FxHashSet<String>> =
2659 jdc_core::FxHashMap::default();
2660 for n in &pool.order {
2661 if lazy && pool.get_if_materialized(n).is_none() {
2662 continue;
2663 }
2664 let renamed = crate::apply_class_rename(n);
2665 let (pkg, simple) = match renamed.rfind('/') {
2666 Some(i) => (renamed[..i].to_string(), renamed[i + 1..].to_string()),
2667 None => (String::new(), renamed.to_string()),
2668 };
2669 let head = simple.split('$').next().unwrap_or(&simple).to_string();
2670 if !head.is_empty() {
2671 let pkg_key = n.rfind('/').map(|i| &n[..i]).unwrap_or("");
2672 let _ = pkg;
2673 pkg_displays.entry(pkg_key).or_default().insert(head);
2674 }
2675 }
2676 // Reverse super map — the obscuring rename must register under every
2677 // transitive subclass (dex refs of an inherited field may name any
2678 // hierarchy class as owner).
2679 let mut subs: jdc_core::FxHashMap<&str, Vec<&str>> = jdc_core::FxHashMap::default();
2680 for n in &pool.order {
2681 if lazy && pool.get_if_materialized(n).is_none() {
2682 continue;
2683 }
2684 // EVERY `$` boundary: each prefix is an owner and the segment
2685 // after it one of its direct member-type tails. The old
2686 // first-segment-only map left NESTED owners (ta8/t$a, androidx
2687 // i0$e) with no member displays — their fields kept or minted
2688 // names colliding with their own member types, and in qualified
2689 // chains a VARIABLE OBSCURES a member type (JLS 6.4.2):
2690 // `ta8.t.a.a100.a` (Kotlin-object singleton sget, nova
2691 // static-ctx ×1,281) and `i0.e.b.ADDING` (lark ×145) bound to
2692 // the field instead of the nested class.
2693 for (i, _) in n.match_indices('$') {
2694 let owner = &n[..i];
2695 let rest = &n[i + 1..];
2696 let tail = match rest.find('$') {
2697 Some(j) => &rest[..j],
2698 None => rest,
2699 };
2700 if owner.is_empty() || tail.is_empty() {
2701 continue;
2702 }
2703 let v = child_tails.entry(owner).or_default();
2704 if !v.contains(&tail) {
2705 v.push(tail);
2706 }
2707 }
2708 if let Some(pc) = pool.get_if_materialized(n) {
2709 if let Some(sup) = &pc.super_name {
2710 if sup != "java/lang/Object" {
2711 subs.entry(sup.as_str()).or_default().push(n.as_str());
2712 }
2713 }
2714 }
2715 }
2716 // Per-ancestor parse cache (function-local: lives exactly as long
2717 // as this rename install, so materialization state is consistent by
2718 // construction; freed on return — no cross-run staleness, no RSS
2719 // tail).
2720 let mut anc_cache: jdc_core::FxHashMap<
2721 std::sync::Arc<str>,
2722 std::sync::Arc<AncEntry>,
2723 > = jdc_core::FxHashMap::default();
2724 for name in &pool.order {
2725 let Some(pc) = pool.get_if_materialized(name) else {
2726 continue;
2727 };
2728 let outer_ref_desc = find_outer_name(pool, name)
2729 .filter(|o| o != name)
2730 .map(|o| format!("L{o};"));
2731 // ---- fields: one shared namespace (static + instance) ----
2732 let fields: Vec<&PoolField> = pc
2733 .static_fields
2734 .iter()
2735 .chain(pc.instance_fields.iter())
2736 .collect();
2737 let fgroups: HashMap<String, Vec<&PoolField>> = HashMap::default();
2738 let mut fgroups = fgroups;
2739 for f in &fields {
2740 let disp = crate::classdec::java_ident(&f.name).into_owned();
2741 fgroups.entry(disp).or_default().push(f);
2742 }
2743 let mut f_taken: jdc_core::FxHashSet<String> = fields
2744 .iter()
2745 .map(|f| crate::classdec::java_ident(&f.name).into_owned())
2746 .collect();
2747 // Displays already registered under THIS class by ancestor
2748 // subs-propagation (an ancestor processed earlier in pool.order
2749 // pushed its renames down): the own mints must avoid them or
2750 // two fields in one class render the same display
2751 // ("已在类中定义了变量 a2", weibo bx$b: subs-pushed bv→a2 beside
2752 // its own String→a2).
2753 if let Some(v) = out.get(&std::sync::Arc::from(name.as_str())) {
2754 for fr in v {
2755 f_taken.insert(fr.display.to_string());
2756 }
2757 }
2758 for group in fgroups.values() {
2759 if group.len() < 2 {
2760 continue;
2761 }
2762 let keeper_pos = group
2763 .iter()
2764 .position(|f| f.access & crate::access::ACC_SYNTHETIC == 0)
2765 .unwrap_or(0);
2766 for (gi, f) in group.iter().enumerate() {
2767 if gi == keeper_pos {
2768 continue;
2769 }
2770 let base = crate::classdec::java_ident(&f.name).into_owned();
2771 let display: String = match &outer_ref_desc {
2772 Some(ord)
2773 if f.access & crate::access::ACC_SYNTHETIC != 0
2774 && !f.is_static
2775 && *ord == *f.desc =>
2776 {
2777 let mut k = 0u32;
2778 loop {
2779 let cand = if k == 0 {
2780 "this$0".to_string()
2781 } else {
2782 format!("this$0{k}")
2783 };
2784 if f_taken.insert(cand.clone()) {
2785 break cand;
2786 }
2787 k += 1;
2788 }
2789 }
2790 _ => suffix_unique(&base, &mut f_taken),
2791 };
2792 // Register with subclass propagation: dex refs of an
2793 // inherited field may name ANY hierarchy class as owner
2794 // (R8 writes the referrer), and without the propagation
2795 // the declaring class's rename does not reach the
2796 // subclass-keyed lookups — the ref renders the RAW name
2797 // against a renamed declaration (bx$b read its inherited
2798 // `a:Lcom/xiaomi/push/bv;` as `this.a` while bx$a
2799 // declared it `a2`). Mirrors the method-side hierarchy
2800 // consistency rule.
2801 register_field_rename_with_subs(
2802 pool, &subs, &mut out, name, &f.name, &f.desc, &display,
2803 );
2804 }
2805 }
2806 // ---- fields obscuring nested types (JLS 6.4.2) ----
2807 // `C.b.a(..)` with BOTH a field `b` and a member class `C$b`
2808 // binds to the FIELD — the qualified type reading is obscured
2809 // ("无法从静态上下文中引用非静态 变量 b" + the paired cannot-find
2810 // cascade; lark LKEvent/MotionLayout family, static-ctx ×1,154).
2811 // The CHILD-side rule (nested_collision_renames) already renames
2812 // colliding children when its local_ok gate passes (`a$a` renders
2813 // as `a2`); this field-side rule covers the gate-REJECTED residual
2814 // (child keeps display `b`, program-wide refs block its rename).
2815 // A field's refs all funnel through field_display (lift call
2816 // sites, declarations, const inits), so the field side needs no
2817 // locality gate. TWO invariants the first attempt violated (lark
2818 // +8,129): fire on the child's FINAL DISPLAY last segment, not
2819 // the raw tail (a child the child-side rule already renamed does
2820 // NOT obscure), and mint avoiding every child display (`a`→`a2`
2821 // landed on the existing `interface a2`, minting a fresh
2822 // obscuring pair).
2823 {
2824 let tails: &[&str] = child_tails
2825 .get(name.as_str())
2826 .map(|v| v.as_slice())
2827 .unwrap_or(&[]);
2828 let mut member_displays: jdc_core::FxHashSet<String> =
2829 jdc_core::FxHashSet::default();
2830 let mut obscuring: jdc_core::FxHashSet<String> =
2831 jdc_core::FxHashSet::default();
2832 for t in tails {
2833 let full = format!("{name}${t}");
2834 let renamed = crate::apply_class_rename(&full);
2835 let last = renamed
2836 .rsplit('/')
2837 .next()
2838 .unwrap_or(&renamed)
2839 .rsplit('$')
2840 .next()
2841 .unwrap_or(t)
2842 .to_string();
2843 member_displays.insert(last.clone());
2844 // Only an INLINE-rendered member is reachable by the bare
2845 // simple name (own-file children are referenced by their
2846 // flat `$` form, which fields cannot obscure).
2847 let member_rendered = clean_member_tail(t)
2848 && crate::find_outer_name(pool, &full).as_deref()
2849 == Some(name.as_str());
2850 if member_rendered {
2851 obscuring.insert(last);
2852 }
2853 }
2854 // The class's OWN simple display name joins the obscuring
2855 // set: a field named like its class captures every
2856 // qualified self-static ref from nested scopes
2857 // (`a2.changeQuickRedirect` bound to the String FIELD a2 —
2858 // weibo robust refs ×317; bare refs survive via the
2859 // variable namespace, the qualified form does not).
2860 let self_renamed = crate::apply_class_rename(name);
2861 let self_simple = self_renamed
2862 .rsplit('/')
2863 .next()
2864 .unwrap_or(&self_renamed)
2865 .rsplit('$')
2866 .next()
2867 .unwrap_or("")
2868 .to_string();
2869 if !self_simple.is_empty() {
2870 obscuring.insert(crate::classdec::java_ident(&self_simple).into_owned());
2871 }
2872 let self_pkg = name.rfind('/').map(|i| &name[..i]).unwrap_or("");
2873 let pkg_names = pkg_displays.get(self_pkg);
2874 for f in &fields {
2875 let fdisp = crate::classdec::java_ident(&f.name).into_owned();
2876 if fdisp.starts_with("this$") || !obscuring.contains(fdisp.as_str()) {
2877 continue;
2878 }
2879 // Already renamed by the same-class duplicate-group rule?
2880 // Its display no longer equals the obscuring simple name,
2881 // so this rule's target collision does not exist — and
2882 // minting a SECOND entry for the same (name, desc) with
2883 // subs-propagation put a CONFLICTING display under every
2884 // subclass: refs naming the subclass resolved to one name
2885 // while the declaring class's declaration rendered the
2886 // first entry's (weibo bx$b read its inherited
2887 // `a:Lcom/xiaomi/push/bv;` as `this.a8` while bx$a
2888 // declared it `a2` — the R8 same-name-field family).
2889 let already = out
2890 .get(&std::sync::Arc::from(name.as_str()))
2891 .is_some_and(|v| {
2892 v.iter().any(|fr| *fr.name == *f.name && *fr.desc == *f.desc)
2893 });
2894 if already {
2895 continue;
2896 }
2897 let mut k = 1u32;
2898 let display = loop {
2899 k += 1;
2900 let cand = format!("{fdisp}{k}");
2901 // root_segs: a minted display equal to a PACKAGE
2902 // first segment captures every qualified ref to
2903 // that package inside the family's file (field e2
2904 // ate `e2.c.b(..)` — lark pq/e "变量 c" ×47). The
2905 // deshadow builder cannot repair this after the
2906 // fact: it scans RAW names, and the collision is
2907 // born here in display space.
2908 if !f_taken.contains(&cand)
2909 && !member_displays.contains(&cand)
2910 && !root_segs.contains(&cand)
2911 && !pkg_names.is_some_and(|p| p.contains(cand.as_str()))
2912 {
2913 f_taken.insert(cand.clone());
2914 member_displays.insert(cand.clone());
2915 break cand;
2916 }
2917 };
2918 register_field_rename_with_subs(
2919 pool, &subs, &mut out, name, &f.name, &f.desc, &display,
2920 );
2921 if std::env::var("DDC_OBSCURE_STATS").is_ok() {
2922 eprintln!("[obscure] {name} field {} -> {}", f.name, display);
2923 }
2924 }
2925 }
2926 // ---- companion-holder fields vs nested types (JLS 6.4.2 in
2927 // expression context): Kotlin compiles `object`/companion
2928 // holders as a static field whose name EQUALS the nested class
2929 // it instances (`static final j$a a`), and static members of the
2930 // nested are invoked as `j.a.c(..)` — but `j.a` in an EXPRESSION
2931 // resolves to the FIELD (variables outrank member types), so the
2932 // call becomes a member lookup on the instance type... which is
2933 // the same class, yet javac reports 找不到符号 for statics
2934 // reached through the shadowed path, and the whole file
2935 // error-types (weibo feed/business/j: 654 errors + closure
2936 // cascade; the nested-side rename for this is the LOCAL-OK-gated
2937 // rule that explodes ungated — the FIELD side is safe because
2938 // dex field refs always carry the declaring class, so this
2939 // registry keys every consumer). Gate: the nested class has
2940 // static members (companion shape) and the field's descriptor
2941 // is exactly the nested type.
2942 if let Some(tails) = child_tails.get(name.as_str()) {
2943 for tail in tails {
2944 let nested = format!("{name}${tail}");
2945 let Some(nc) = pool.get_if_materialized(&nested) else {
2946 continue;
2947 };
2948 // Only CLEAN tails render dotted (`j.a`); flat-rendered
2949 // tails (`j$1`) carry the `$` and never clash.
2950 if !clean_member_tail(tail) {
2951 continue;
2952 }
2953 let has_statics = nc
2954 .static_fields
2955 .iter()
2956 .chain(nc.instance_fields.iter())
2957 .any(|f| f.is_static)
2958 || nc.all_methods().any(|m| m.access & crate::access::ACC_STATIC != 0);
2959 if !has_statics {
2960 continue;
2961 }
2962 // STRICT companion shape: the field's descriptor IS the
2963 // nested type (`static final j$a a`). The relaxed form
2964 // (any same-named field beside a statics-bearing nested)
2965 // minted thousands of renames on weibo and exploded the
2966 // battery to the 500k cap — the field funnel does not
2967 // hold at that scale (改名面 vs 引用覆盖面, again).
2968 let want_desc = format!("L{nested};");
2969 for f in pc.static_fields.iter().chain(pc.instance_fields.iter()) {
2970 if f.desc.as_ref() != want_desc.as_str() {
2971 continue;
2972 }
2973 let disp = crate::classdec::java_ident(&f.name).into_owned();
2974 if disp != crate::classdec::java_ident(tail) {
2975 continue;
2976 }
2977 // Already renamed by the duplicate-group rule above?
2978 let already = out
2979 .get(&std::sync::Arc::from(name.as_str()))
2980 .is_some_and(|v| {
2981 v.iter().any(|fr| *fr.name == *f.name && *fr.desc == *f.desc)
2982 });
2983 if already {
2984 continue;
2985 }
2986 let display = suffix_unique(&disp, &mut f_taken);
2987 f_taken.insert(display.clone());
2988 out.entry(std::sync::Arc::from(name.as_str()))
2989 .or_default()
2990 .push(jdc_core::rename::FieldRename {
2991 name: f.name.clone(),
2992 desc: f.desc.clone(),
2993 display: std::sync::Arc::from(display.as_str()),
2994 });
2995 }
2996 }
2997 }
2998 // ---- methods: display key = sanitized name + erased params ----
2999 let methods: Vec<&PoolMethod> = pc.all_methods().collect();
3000 // Ancestor virtual-method returns per (name, argsig) — one
3001 // chain walk per non-interface class, shared by the
3002 // covariant-keeper selection and the impostor-override
3003 // detection below. Layers stay PER-ANCESTOR (cached Arcs, no
3004 // merge): every consumer is an "any ancestor declares
3005 // (name, argsig) → ret" query, so a per-layer scan replaces
3006 // the old per-class eager map build. Provenance splits the
3007 // chain into fw / pool buckets (see AncEntry.is_fw).
3008 let mut anc_chain: Option<(AncLayers, AncLayers)> = None;
3009 if pc.access & crate::access::ACC_INTERFACE == 0 {
3010 let mut fw_layers: AncLayers = Vec::new();
3011 let mut pool_layers: AncLayers = Vec::new();
3012 let mut stack: Vec<std::sync::Arc<str>> = Vec::new();
3013 if let Some(sup) = &pc.super_name {
3014 if sup != "java/lang/Object" {
3015 stack.push(std::sync::Arc::from(sup.as_str()));
3016 }
3017 }
3018 stack.extend(
3019 pc.interfaces
3020 .iter()
3021 .map(|i| std::sync::Arc::from(i.as_str())),
3022 );
3023 let mut seen_cls: jdc_core::FxHashSet<std::sync::Arc<str>> =
3024 jdc_core::FxHashSet::default();
3025 while let Some(cn) = stack.pop() {
3026 if !seen_cls.insert(cn.clone()) {
3027 continue;
3028 }
3029 let entry = match anc_cache.get(cn.as_ref()) {
3030 Some(e) => e.clone(),
3031 None => {
3032 let e = parse_anc_entry(pool, &cn);
3033 anc_cache.insert(cn.clone(), e.clone());
3034 e
3035 }
3036 };
3037 if !entry.methods.is_empty() {
3038 if entry.is_fw {
3039 fw_layers.push(entry.clone());
3040 } else {
3041 pool_layers.push(entry.clone());
3042 }
3043 }
3044 stack.extend(entry.supers.iter().cloned());
3045 }
3046 anc_chain = Some((fw_layers, pool_layers));
3047 }
3048 let mut m_taken: jdc_core::FxHashSet<String> = methods
3049 .iter()
3050 .map(|m| crate::classdec::java_ident(&m.name).into_owned())
3051 .collect();
3052 let mut mgroups: HashMap<(String, String), Vec<&PoolMethod>> = HashMap::default();
3053 for m in &methods {
3054 let d: &str = &m.desc;
3055 let lo = d.find('(').map(|i| i + 1).unwrap_or(0);
3056 let hi = d.find(')').unwrap_or(d.len());
3057 let key = (
3058 crate::classdec::java_ident(&m.name).into_owned(),
3059 d[lo..hi].to_string(),
3060 );
3061 mgroups.entry(key).or_default().push(m);
3062 }
3063 for ((base, _), group) in &mgroups {
3064 if group.len() < 2 {
3065 continue;
3066 }
3067 // R8 return-type specialization: a non-isolated class carries
3068 // `M(args)void` (the real logic) beside the interface bridge
3069 // `M(args)R` (unbox → call void → return Unit). Same erasure,
3070 // so Java source can't hold both, and `void` never satisfies
3071 // the interface's `R` — the claim map keeps the void side and
3072 // drops the bridge, yielding "X不是抽象的, 并且未覆盖…invoke"
3073 // (Kotlin suspend lambdas, weixin/lark ×1.1k). This is NOT the
3074 // covariant-override shape below (void has no subtype
3075 // relation): keep the BRIDGE under the original name so it
3076 // satisfies the interface, and rename the void specialization
3077 // so both render and the bridge's call to it resolves to the
3078 // renamed method.
3079 // `)Ljava/lang/Void;` counts as void-like: a Kotlin `Nothing`
3080 // function (an unsupported-op that always throws) compiles to
3081 // the Void CLASS return, not the `V` primitive, but it IS the
3082 // real-logic side of the same specialization pair. Missing it
3083 // sent the group down the covariant path, which kept the Void
3084 // method and renamed the bridge, then the impostor pass renamed
3085 // the Void method too — leaving NO `addAll` to satisfy Set
3086 // (weibo compose SnapshotMapEntrySet "未覆盖Set中的addAll").
3087 let ret_void = |m: &PoolMethod| {
3088 m.desc.ends_with(")V") || m.desc.ends_with(")Ljava/lang/Void;")
3089 };
3090 let isolated = pc
3091 .super_name
3092 .as_ref()
3093 .is_none_or(|s| s == "java/lang/Object")
3094 && pc.interfaces.is_empty();
3095 let has_void = group.iter().any(|m| ret_void(m) && &*m.name != "<init>");
3096 let has_nonvoid = group.iter().any(|m| !ret_void(m));
3097 if !isolated && has_void && has_nonvoid {
3098 let keeper_desc: Option<&str> = group
3099 .iter()
3100 .copied()
3101 .find(|m| {
3102 m.access & crate::access::ACC_BRIDGE != 0
3103 && !ret_void(m)
3104 && &*m.name != "<init>"
3105 })
3106 .or_else(|| {
3107 group
3108 .iter()
3109 .copied()
3110 .find(|m| !ret_void(m) && &*m.name != "<init>")
3111 })
3112 .map(|m| &*m.desc);
3113 let mut seen_orig: jdc_core::FxHashSet<(&str, &str)> =
3114 jdc_core::FxHashSet::default();
3115 for m in group.iter().copied() {
3116 if keeper_desc == Some(&*m.desc) {
3117 seen_orig.insert((&*m.name, &*m.desc));
3118 continue; // bridge keeps the interface name
3119 }
3120 if !seen_orig.insert((&*m.name, &*m.desc)) {
3121 continue; // exact duplicate: emitter drops it
3122 }
3123 let display = suffix_unique(base, &mut m_taken);
3124 out.entry(std::sync::Arc::from(name.as_str()))
3125 .or_default()
3126 .push(jdc_core::rename::FieldRename {
3127 name: m.name.clone(),
3128 desc: m.desc.clone(),
3129 display: std::sync::Arc::from(display.as_str()),
3130 });
3131 }
3132 continue;
3133 }
3134 // A COVARIANT override pair (same ORIGINAL name, different
3135 // descriptors — the compiler's bridge shape: the interface's
3136 // `deserialize(e)` plus the narrower implementation). Renaming
3137 // either side breaks @Override ("is not abstract and does not
3138 // override abstract method deserialize(e)", ~2.2k on reqable);
3139 // the emitter's claim logic already renders only the
3140 // non-bridge member of the pair. An ISOLATED class (extends
3141 // Object, implements nothing — d8's ExternalSyntheticApiModel
3142 // Outline shells) overrides nothing: its same-(name,params)
3143 // pairs differing only by RETURN type are dex-legal distinct
3144 // methods, and dropping one strands its callers on the
3145 // survivor (`HalfKt$$…0.m()Class` vs `.m()V` — "void无法
3146 // 转换为Class", weibo ×85+).
3147 let covariant = !isolated
3148 && group
3149 .iter()
3150 .any(|m| group.iter().any(|o| o.name == m.name && o.desc != m.desc));
3151 if covariant {
3152 // Return-type-only clash in a class WITH a hierarchy:
3153 // Java cannot declare both, and the claim logic would
3154 // keep an arbitrary one — when that is NOT the member
3155 // an ancestor declares, every concrete subclass fails
3156 // "X不是抽象的, 并且未覆盖…" (weibo ×590: viewmodel.a
3157 // kept getUsageInfo()e while interface d declares
3158 // getUsageInfo()g; the kept side also strands the
3159 // dropped side's callers). Keep the ancestor-declared
3160 // member under the original name and rename the rest —
3161 // the registry carries every call site, and the
3162 // propagation below shares the display down the family.
3163 // 0 or 2+ ancestor-matched members = the shape Java
3164 // cannot express at all (two interfaces demanding
3165 // different returns): keep the old claim behavior.
3166 let keepers: Vec<&&PoolMethod> = group
3167 .iter()
3168 .filter(|m| {
3169 let d: &str = &m.desc;
3170 let lo = d.find('(').map(|i| i + 1).unwrap_or(0);
3171 let hi = d.find(')').unwrap_or(d.len());
3172 if let Some((fwl, pll)) = &anc_chain {
3173 layer_rets(fwl, &m.name, &d[lo..hi])
3174 .chain(layer_rets(pll, &m.name, &d[lo..hi]))
3175 .any(|r| r == &d[hi + 1..])
3176 } else {
3177 false
3178 }
3179 })
3180 .collect();
3181 if keepers.len() >= 2 {
3182 // The shape Java cannot express at all (2+
3183 // ancestors demanding different returns): no
3184 // rename target — keep the claim behavior.
3185 continue;
3186 }
3187 // fwdb flattens JDK covariant overrides to the erased
3188 // declaring return (SortedMap.keySet records Set, the
3189 // source return is SortedSet) — the sole "keeper" can
3190 // be the synthetic BRIDGE while the real override is
3191 // the non-bridge sibling whose return is a subtype of
3192 // the recorded one. Keeping the bridge renamed the
3193 // real method and broke the interface chain (WhatsApp
3194 // X/Jl2 SortedMap, cannot-find:变量 ×18.6k). Prefer the
3195 // non-bridge in that shape: the claim map dropped the
3196 // bridge before this machinery existed anyway.
3197 // Keeper selection by ancestor PROVENANCE:
3198 // - POOL ancestor matches (raw-rendered): the member
3199 // whose return EXACTLY matches a pool erasure that
3200 // is NOT purely framework-sourced keeps the name —
3201 // the erased bridge is what satisfies the raw
3202 // interface (weibo Function0.invoke / GlideRequest
3203 // .load; renaming the real one instead cost +375).
3204 // - FRAMEWORK-only match: javac reads the compiled
3205 // covariant truth (SortedMap.keySet→SortedSet) while
3206 // fwdb records the flattened erasure (→Set) that the
3207 // BRIDGE matches — prefer the non-bridge sibling
3208 // whose return is a subtype (WhatsApp X/Jl2).
3209 // - no ancestor: first non-bridge (the old claim
3210 // survivor), else dex order.
3211 fn ret_desc(m: &PoolMethod) -> &str {
3212 let d: &str = &m.desc;
3213 &d[d.find(')').map(|i| i + 1).unwrap_or(d.len())..]
3214 }
3215 let gkey = {
3216 let d: &str = &group[0].desc;
3217 let lo = d.find('(').map(|i| i + 1).unwrap_or(0);
3218 let hi = d.find(')').unwrap_or(d.len());
3219 (
3220 crate::classdec::java_ident(&group[0].name).into_owned(),
3221 d[lo..hi].to_string(),
3222 )
3223 };
3224 // fw match: a NON-BRIDGE member whose return is
3225 // assignable to a framework ret (covariant source
3226 // truth — it satisfies the raw pool ancestor too,
3227 // overrides being covariant-legal in Java).
3228 let fw_i = anc_chain.as_ref().and_then(|(fwl, _)| {
3229 let fws: Vec<&str> =
3230 layer_rets(fwl, &gkey.0, &gkey.1).collect();
3231 if fws.is_empty() {
3232 return None;
3233 }
3234 group.iter().position(|m| {
3235 m.access & crate::access::ACC_BRIDGE == 0
3236 && fws.iter().any(|fr| {
3237 // ret_satisfies: descriptor-level, so
3238 // ARRAY covariant returns match too (the
3239 // old ret_internal strip returned None
3240 // for `[LX;` and fell back to descriptor
3241 // equality — Creator newArray's real
3242 // `[LSavedState;` never matched fw's
3243 // `[LObject;`).
3244 ret_satisfies(pool, ret_desc(m), fr)
3245 })
3246 })
3247 });
3248 let pool_i = anc_chain.as_ref().and_then(|(_, pll)| {
3249 let rets: Vec<&str> =
3250 layer_rets(pll, &gkey.0, &gkey.1).collect();
3251 if rets.is_empty() {
3252 return None;
3253 }
3254 group
3255 .iter()
3256 .position(|m| rets.iter().any(|r| *r == ret_desc(m)))
3257 });
3258 let keeper_idx = match (fw_i, pool_i) {
3259 (Some(fi), _) => fi,
3260 (None, Some(pi)) => pi,
3261 (None, None) => group
3262 .iter()
3263 .position(|m| m.access & crate::access::ACC_BRIDGE == 0)
3264 .unwrap_or(0),
3265 };
3266 // keepers==1: the ancestor-declared member keeps the
3267 // original name. keepers==0: NOTHING in the POOL binds
3268 // the name — keep the member the old claim map would
3269 // have kept (first NON-BRIDGE, else dex order) and
3270 // rename the rest through the registry: letting the
3271 // claim map silently drop a side stranded its callers
3272 // on the survivor's incompatible return (weibo xiaomi
3273 // push gk/dq/hj/hf/gt: `boolean b()Z` dropped beside
3274 // `String b()` — every `boolean v = this.b()` bound to
3275 // the String twin, "String无法转换为boolean" ×300+ in
3276 // the push package alone). NON-bridge preference is
3277 // load-bearing: a framework-ancestor covariant pair
3278 // (emoji2 n extends SpannableStringBuilder, Editable-
3279 // bridge beside the real override) satisfies the
3280 // ancestor through the REAL method — bridge-preference
3281 // renamed it and broke the override chain (reqable +8
3282 // "无法覆盖…返回类型不兼容").
3283 let keep_desc: &str = &group[keeper_idx].desc;
3284 let mut seen_orig: jdc_core::FxHashSet<(&str, &str)> =
3285 jdc_core::FxHashSet::default();
3286 for m in group.iter().copied() {
3287 if m.desc.as_ref() == keep_desc {
3288 seen_orig.insert((&*m.name, &*m.desc));
3289 continue;
3290 }
3291 if !seen_orig.insert((&*m.name, &*m.desc)) {
3292 continue;
3293 }
3294 let display = suffix_unique(base, &mut m_taken);
3295 out.entry(std::sync::Arc::from(name.as_str()))
3296 .or_default()
3297 .push(jdc_core::rename::FieldRename {
3298 name: m.name.clone(),
3299 desc: m.desc.clone(),
3300 display: std::sync::Arc::from(display.as_str()),
3301 });
3302 }
3303 continue;
3304 }
3305 // Real re-declarations (same original name AND descriptor —
3306 // R8 duplicates) are already deduped by the emitter's claim
3307 // logic; only SANITIZER collapses reach here and every one of
3308 // them is a distinct method.
3309 let mut base = base.clone();
3310 let mut seen_orig: jdc_core::FxHashSet<(&str, &str)> =
3311 jdc_core::FxHashSet::default();
3312 for m in group {
3313 if !seen_orig.insert((&m.name, &m.desc)) {
3314 continue; // exact duplicate: emitter drops it
3315 }
3316 let display = suffix_unique(&base, &mut m_taken);
3317 out.entry(std::sync::Arc::from(name.as_str()))
3318 .or_default()
3319 .push(jdc_core::rename::FieldRename {
3320 name: m.name.clone(),
3321 desc: m.desc.clone(),
3322 display: std::sync::Arc::from(display.as_str()),
3323 });
3324 // The base stays the sanitized ORIGINAL so the first
3325 // occurrence keeps the plain name.
3326 let _ = &mut base;
3327 }
3328 }
3329 // ---- impostor overrides (R8 generic specialization) ----
3330 // A class method whose (name, argsig) matches an ancestor
3331 // declaration but whose return type satisfies NONE of the
3332 // ancestor returns is not an override Java can express: the
3333 // pair shares an erasure, so the class cannot declare both, and
3334 // ART never resolved it either (interface dispatch matches
3335 // name+descriptor — the missing proto throws AbstractMethodError
3336 // at runtime; weibo IStreamPresenter.getView()IStreamView vs the
3337 // impl's getView()IPageView with no subtype edge, ×258 across
3338 // getView/getWidget/getListView families). Rename the impostor
3339 // (the registry carries every caller and the propagation below
3340 // carries subclasses) and let synth_missing_interface_stubs add
3341 // the ART-faithful stub — rename-aware provision makes the
3342 // requirement missing again.
3343 if let Some((fwl, pll)) = &anc_chain {
3344 for m in &methods {
3345 if m.is_static()
3346 || &*m.name == "<init>"
3347 || &*m.name == "<clinit>"
3348 {
3349 continue;
3350 }
3351 let d: &str = &m.desc;
3352 let lo = d.find('(').map(|i| i + 1).unwrap_or(0);
3353 let hi = d.find(')').unwrap_or(d.len());
3354 let m_ret = &d[hi + 1..];
3355 // Lazy per-layer scan: `found` replaces the old
3356 // `map.get(key)` presence gate (no ancestor declares
3357 // (name, argsig) ⇒ nothing to violate ⇒ skip).
3358 let mut found = false;
3359 let mut satisfied = false;
3360 'lay: for e in fwl.iter().chain(pll.iter()) {
3361 let Some(rets) = e.methods.get(m.name.as_ref()) else {
3362 continue;
3363 };
3364 for (a, r) in rets {
3365 if **a != d[lo..hi] {
3366 continue;
3367 }
3368 found = true;
3369 // ret_satisfies handles ARRAY covariance in
3370 // descriptor form — the old L…;-strip match
3371 // answered false for `SavedState[] <: Object[]`
3372 // and renamed the REAL Parcelable.Creator
3373 // newArray as an impostor (newArray3), breaking
3374 // the override ("不是抽象的…未覆盖 newArray(int)",
3375 // weibo ×504 / battery ~789).
3376 if ret_satisfies(pool, m_ret, r) {
3377 satisfied = true;
3378 break 'lay;
3379 }
3380 }
3381 }
3382 if !found || satisfied {
3383 continue;
3384 }
3385 // Already renamed by a clash-group rule above?
3386 let already = out
3387 .get(&std::sync::Arc::from(name.as_str()))
3388 .is_some_and(|v| {
3389 v.iter().any(|fr| *fr.name == *m.name && *fr.desc == *m.desc)
3390 });
3391 if already {
3392 continue;
3393 }
3394 let base = crate::classdec::java_ident(&m.name).into_owned();
3395 let display = suffix_unique(&base, &mut m_taken);
3396 out.entry(std::sync::Arc::from(name.as_str()))
3397 .or_default()
3398 .push(jdc_core::rename::FieldRename {
3399 name: m.name.clone(),
3400 desc: m.desc.clone(),
3401 display: std::sync::Arc::from(display.as_str()),
3402 });
3403 }
3404 }
3405 }
3406 // ---- method-rename propagation down the hierarchy ----
3407 // A renamed method must keep ONE display across its whole override
3408 // family: the parent's abstract declaration and every subclass
3409 // implementation/caller render through the registry, and a missed
3410 // link re-breaks the override ("未覆盖" at each concrete subclass)
3411 // or the call site (cannot-find). Register each method entry on the
3412 // full transitive subclass + implementer closure — a subclass that
3413 // does NOT declare the method still serves as a dex ref owner for
3414 // the inherited method, so it needs the entry too (same reasoning
3415 // as register_field_rename_with_subs for fields). Ancestor-forced
3416 // displays win; a colliding local entry re-mints off the original
3417 // method name and re-propagates.
3418 {
3419 let mut down: jdc_core::FxHashMap<&str, Vec<&str>> =
3420 jdc_core::FxHashMap::default();
3421 for (sup, children) in subs.iter() {
3422 down.entry(*sup).or_default().extend(children.iter().copied());
3423 }
3424 for n in &pool.order {
3425 if let Some(pc) = pool.get_if_materialized(n) {
3426 for i in pc.interfaces.iter() {
3427 down.entry(i.as_str()).or_default().push(n.as_str());
3428 }
3429 }
3430 }
3431 type PropEntry = (
3432 std::sync::Arc<str>,
3433 std::sync::Arc<str>,
3434 std::sync::Arc<str>,
3435 std::sync::Arc<str>,
3436 );
3437 let mut queue: std::collections::VecDeque<PropEntry> =
3438 std::collections::VecDeque::new();
3439 {
3440 // Sorted seed: `out` is a std HashMap (RandomState) — an
3441 // unsorted seed would make conflicting ancestor forces
3442 // resolve in per-process order (display names flipping
3443 // run-to-run).
3444 let mut seed: Vec<PropEntry> = Vec::new();
3445 for (cls, entries) in out.iter() {
3446 for fr in entries {
3447 if fr.desc.contains('(') {
3448 seed.push((
3449 cls.clone(),
3450 fr.name.clone(),
3451 fr.desc.clone(),
3452 fr.display.clone(),
3453 ));
3454 }
3455 }
3456 }
3457 seed.sort();
3458 for t in seed {
3459 queue.push_back(t);
3460 }
3461 }
3462 let mut guard = 0usize;
3463 while let Some((cls, n, d, disp)) = queue.pop_front() {
3464 guard += 1;
3465 if guard > 2_000_000 {
3466 break;
3467 }
3468 let Some(children) = down.get(cls.as_ref()).cloned() else {
3469 continue;
3470 };
3471 for child in children {
3472 let mut changed = false;
3473 let mut reminted: Vec<PropEntry> = Vec::new();
3474 {
3475 let child_key: std::sync::Arc<str> = std::sync::Arc::from(child);
3476 let entries = out.entry(child_key.clone()).or_default();
3477 // A raw method of the child literally named `disp`
3478 // with a different signature blocks the force —
3479 // renaming that raw method instead is the over-reach
3480 // class (500k blowup history); skip this child.
3481 let raw_blocked = pool
3482 .get_if_materialized(child)
3483 .is_some_and(|pc| {
3484 pc.all_methods().any(|m| {
3485 crate::classdec::java_ident(&m.name) == disp.as_ref()
3486 && m.desc != d
3487 })
3488 });
3489 if raw_blocked {
3490 continue;
3491 }
3492 match entries
3493 .iter_mut()
3494 .find(|fr| *fr.name == *n && *fr.desc == *d)
3495 {
3496 Some(e) if e.display == disp => {}
3497 Some(e) => {
3498 e.display = disp.clone();
3499 changed = true;
3500 }
3501 None => {
3502 entries.push(jdc_core::rename::FieldRename {
3503 name: n.clone(),
3504 desc: d.clone(),
3505 display: disp.clone(),
3506 });
3507 changed = true;
3508 }
3509 }
3510 if changed {
3511 // Display collisions inside the child: the
3512 // ancestor-forced entry keeps `disp`; others
3513 // re-mint from their original method name.
3514 let mut taken: jdc_core::FxHashSet<String> =
3515 jdc_core::FxHashSet::default();
3516 if let Some(pc) = pool.get_if_materialized(child) {
3517 for m in pc.all_methods() {
3518 taken.insert(
3519 crate::classdec::java_ident(&m.name).into_owned(),
3520 );
3521 }
3522 }
3523 for e in entries.iter() {
3524 taken.insert(e.display.to_string());
3525 }
3526 let mut seen: jdc_core::FxHashSet<String> =
3527 jdc_core::FxHashSet::default();
3528 seen.insert(disp.to_string());
3529 for e in entries.iter_mut() {
3530 // FIELD entries never re-mint here: this is
3531 // the METHOD-name propagation, and a field
3532 // display colliding with a forced method
3533 // display is resolved at the field's own mint
3534 // (f_taken seeding above), not by renaming the
3535 // field out from under its subs-registered
3536 // references (bx$b's inherited bv→a2 was
3537 // reminted to a7 while every ref kept a2).
3538 if !e.desc.contains('(') {
3539 continue;
3540 }
3541 if *e.name == *n && *e.desc == *d {
3542 continue;
3543 }
3544 if seen.insert(e.display.to_string()) {
3545 continue;
3546 }
3547 let base =
3548 crate::classdec::java_ident(&e.name).into_owned();
3549 let nd: std::sync::Arc<str> = std::sync::Arc::from(
3550 suffix_unique(&base, &mut taken).as_str(),
3551 );
3552 e.display = nd.clone();
3553 reminted.push((
3554 child_key.clone(),
3555 e.name.clone(),
3556 e.desc.clone(),
3557 nd,
3558 ));
3559 }
3560 }
3561 }
3562 if changed {
3563 let child_key: std::sync::Arc<str> = std::sync::Arc::from(child);
3564 queue.push_back((child_key.clone(), n.clone(), d.clone(), disp.clone()));
3565 for r in reminted {
3566 queue.push_back(r);
3567 }
3568 }
3569 }
3570 }
3571 }
3572 inherited_obscuring_renames(pool, &mut out, &pkg_displays);
3573 out
3574}
3575
3576/// Inherited-field variant of the obscuring rule: class C renders a
3577/// member nested type `d` while INHERITING a field `d` from a pool
3578/// superclass — `C.d.b(..)` binds to the inherited variable (fields
3579/// obscure types, JLS 6.4.2) and fails ("无法从静态上下文中引用非静态
3580/// 变量 d" + "无法取消引用int" pairs, LynxBaseScrollViewDragging ×84 —
3581/// latent behind the broken-superclass hub until the same-class rule
3582/// healed it). The rename lands on the DECLARING class (one declaration
3583/// to re-render) and is registered under every transitive subclass as
3584/// well, because dex field refs may name any subclass as owner and the
3585/// lift-time consult keys on the ref's owner.
3586/// Register a field rename under the declaring class AND every transitive
3587/// subclass: dex field refs for an inherited access may name ANY class in
3588/// the hierarchy as owner (`iput Lh$a;->a` inside the constant-subclass
3589/// ctor of `h`), and the lift-time consult keys on the ref's owner — a
3590/// declaring-class-only entry leaves those refs stale (org/a/c/h ×3).
3591/// Subclasses that REDECLARE (name, desc) hide the field — their own
3592/// field is distinct and must keep its name (and their own obscuring,
3593/// if any, is handled when the loop reaches them).
3594fn register_field_rename_with_subs(
3595 pool: &DexPool,
3596 subs: &jdc_core::FxHashMap<&str, Vec<&str>>,
3597 out: &mut HashMap<std::sync::Arc<str>, Vec<jdc_core::rename::FieldRename>>,
3598 owner: &str,
3599 fname: &str,
3600 fdesc: &str,
3601 display: &str,
3602) {
3603 let mut queue: Vec<&str> = vec![owner];
3604 let mut seen: jdc_core::FxHashSet<&str> = jdc_core::FxHashSet::default();
3605 seen.insert(owner);
3606 let mut first = true;
3607 while let Some(c) = queue.pop() {
3608 let redeclares = if first {
3609 first = false;
3610 false // the declaring class itself: always register
3611 } else {
3612 pool.get_if_materialized(c)
3613 .map(|pc| {
3614 pc.static_fields
3615 .iter()
3616 .chain(pc.instance_fields.iter())
3617 .any(|f| f.name.as_ref() == fname && f.desc.as_ref() == fdesc)
3618 })
3619 .unwrap_or(false)
3620 };
3621 if redeclares {
3622 continue; // hidden below this point: its subtree refs bind to its own field
3623 }
3624 out.entry(std::sync::Arc::from(c))
3625 .or_default()
3626 .push(jdc_core::rename::FieldRename {
3627 name: std::sync::Arc::from(fname),
3628 desc: std::sync::Arc::from(fdesc),
3629 display: std::sync::Arc::from(display),
3630 });
3631 if let Some(children) = subs.get(c) {
3632 for ch in children {
3633 if seen.insert(*ch) {
3634 queue.push(*ch);
3635 }
3636 }
3637 }
3638 }
3639}
3640
3641fn inherited_obscuring_renames(
3642 pool: &DexPool,
3643 out: &mut HashMap<std::sync::Arc<str>, Vec<jdc_core::rename::FieldRename>>,
3644 pkg_displays: &jdc_core::FxHashMap<&str, jdc_core::FxHashSet<String>>,
3645) {
3646 let root_segs = pool.root_pkg_segs();
3647 // Reverse super map over materialized classes.
3648 let mut subs: jdc_core::FxHashMap<&str, Vec<&str>> = jdc_core::FxHashMap::default();
3649 for n in &pool.order {
3650 let Some(pc) = pool.get_if_materialized(n) else {
3651 continue;
3652 };
3653 if let Some(sup) = &pc.super_name {
3654 if sup != "java/lang/Object" {
3655 subs.entry(sup.as_str()).or_default().push(n.as_str());
3656 }
3657 }
3658 }
3659 let mut renamed: jdc_core::FxHashSet<(String, String, String)> =
3660 jdc_core::FxHashSet::default();
3661 for n in &pool.order {
3662 let Some(pc) = pool.get_if_materialized(n) else {
3663 continue;
3664 };
3665 if pc.is_interface() {
3666 continue;
3667 }
3668 // Member-rendered child displays of THIS class.
3669 let mut child_displays: Vec<String> = Vec::new();
3670 for child in pool.children_of(n.as_str()) {
3671 let Some(rest) = child
3672 .strip_prefix(n.as_str())
3673 .and_then(|t| t.strip_prefix('$'))
3674 else {
3675 continue;
3676 };
3677 if !clean_member_tail(rest) {
3678 continue;
3679 }
3680 if crate::find_outer_name(pool, child).as_deref() != Some(n.as_str()) {
3681 continue;
3682 }
3683 let renamed_disp = crate::apply_class_rename(child);
3684 let last = renamed_disp
3685 .rsplit('/')
3686 .next()
3687 .unwrap_or(&renamed_disp)
3688 .rsplit('$')
3689 .next()
3690 .unwrap_or(rest)
3691 .to_string();
3692 child_displays.push(last);
3693 }
3694 if child_displays.is_empty() {
3695 continue;
3696 }
3697 // Walk the super chain for an inherited field whose display
3698 // collides; own fields are the same-class rule's job.
3699 let own: jdc_core::FxHashSet<String> = pc
3700 .static_fields
3701 .iter()
3702 .chain(pc.instance_fields.iter())
3703 .map(|f| crate::classdec::java_ident(&f.name).into_owned())
3704 .collect();
3705 let mut sup = pc.super_name.clone();
3706 while let Some(sname) = sup {
3707 if sname == "java/lang/Object" {
3708 break;
3709 }
3710 let Some(sc) = pool.get_if_materialized(&sname) else {
3711 break;
3712 };
3713 for f in sc.static_fields.iter().chain(sc.instance_fields.iter()) {
3714 let fdisp = crate::classdec::java_ident(&f.name).into_owned();
3715 if fdisp.starts_with("this$") || !child_displays.contains(&fdisp) {
3716 continue;
3717 }
3718 if !renamed.insert((sname.clone(), f.name.to_string(), f.desc.to_string())) {
3719 continue;
3720 }
3721 // Mint in the DECLARING class's namespace: avoid its
3722 // fields + its own member-rendered child displays.
3723 let mut avoid: jdc_core::FxHashSet<String> = sc
3724 .static_fields
3725 .iter()
3726 .chain(sc.instance_fields.iter())
3727 .map(|g| crate::classdec::java_ident(&g.name).into_owned())
3728 .collect();
3729 // The declaring class's PACKAGE class displays (post-
3730 // rename): minting onto one shadows its qualified refs
3731 // from inside the class (the a→a2-onto-display-a2 bug).
3732 if let Some(names) = pkg_displays
3733 .get(sname.rfind('/').map(|i| &sname[..i]).unwrap_or(""))
3734 {
3735 avoid.extend(names.iter().cloned());
3736 }
3737 for child in pool.children_of(sname.as_str()) {
3738 let Some(rest) = child
3739 .strip_prefix(sname.as_str())
3740 .and_then(|t| t.strip_prefix('$'))
3741 else {
3742 continue;
3743 };
3744 if !clean_member_tail(rest) {
3745 continue;
3746 }
3747 let rd = crate::apply_class_rename(child);
3748 avoid.insert(
3749 rd.rsplit('/')
3750 .next()
3751 .unwrap_or(&rd)
3752 .rsplit('$')
3753 .next()
3754 .unwrap_or(rest)
3755 .to_string(),
3756 );
3757 }
3758 let mut k = 1u32;
3759 let display = loop {
3760 k += 1;
3761 let cand = format!("{fdisp}{k}");
3762 // root_segs: see the own-field mint site — a
3763 // package first segment must never be minted.
3764 if !avoid.contains(&cand) && !root_segs.contains(&cand) {
3765 break cand;
3766 }
3767 };
3768 register_field_rename_with_subs(
3769 pool, &subs, out, &sname, &f.name, &f.desc, &display,
3770 );
3771 if std::env::var("DDC_OBSCURE_STATS").is_ok() {
3772 eprintln!("[obscure-inh] {sname} field {} -> {display} (child of {n})", f.name);
3773 }
3774 }
3775 sup = sc.super_name.clone();
3776 }
3777 let _ = &own;
3778 }
3779}
3780
3781fn suffix_unique(base: &str, taken: &mut jdc_core::FxHashSet<String>) -> String {
3782 let mut k = 1u32;
3783 loop {
3784 k += 1;
3785 let cand = format!("{base}{k}");
3786 if taken.insert(cand.clone()) {
3787 return cand;
3788 }
3789 }
3790}
3791
3792/// member_collision_renames + field_deshadow_renames, collision-registry
3793/// entries winning per (owner, name, desc).
3794fn combined_field_renames(
3795 pool: &DexPool,
3796) -> HashMap<std::sync::Arc<str>, Vec<jdc_core::rename::FieldRename>> {
3797 let t = std::time::Instant::now();
3798 let mut base = member_collision_renames(pool);
3799 if std::env::var("DDC_DBG_REN").is_ok() {
3800 eprintln!("[ren] {:?} member_collision_renames", t.elapsed());
3801 }
3802 let t = std::time::Instant::now();
3803 crate::stats_rss(" member_collision_renames");
3804 let extra = field_deshadow_renames(pool);
3805 if std::env::var("DDC_DBG_REN").is_ok() {
3806 eprintln!("[ren] {:?} field_deshadow_renames", t.elapsed());
3807 }
3808 let mut merged = 0usize;
3809 for (owner, v) in extra {
3810 let e = base.entry(owner).or_default();
3811 for fr in v {
3812 if !e.iter().any(|x| x.name == fr.name && x.desc == fr.desc) {
3813 e.push(fr);
3814 merged += 1;
3815 }
3816 }
3817 }
3818 if std::env::var("DDC_STATS").is_ok() {
3819 eprintln!("[renames] field-deshadow renames: {merged}");
3820 }
3821 crate::stats_rss(" field_deshadow merged");
3822 base
3823}
3824
3825/// Field deshadow (JLS 6.4.2, expression position): a FIELD whose display
3826/// name equals a root package first segment captures package-qualified
3827/// renders inside its family's file — androidx z2's field `j` binds the
3828/// `j` of `j.a.a` (找不到符号 变量 a). Rename through the field registry
3829/// (declaration and every reference funnel through field_display).
3830/// Gate: only when the family actually references that segment
3831/// (family_ref_segments body scan) or the segment is the family's OWN
3832/// package root (same-package refs fall back to `root.x` FQN renders) —
3833/// ungated field renames are the six-times-falsified blast radius.
3834fn field_deshadow_renames(
3835 pool: &DexPool,
3836) -> HashMap<std::sync::Arc<str>, Vec<jdc_core::rename::FieldRename>> {
3837 let mut out: HashMap<std::sync::Arc<str>, Vec<jdc_core::rename::FieldRename>> =
3838 HashMap::default();
3839 let root_segs = pool.root_pkg_segs();
3840 if root_segs.is_empty() {
3841 return out;
3842 }
3843 // Candidate fields per family (display, original name, desc).
3844 let mut cand: HashMap<
3845 String,
3846 Vec<(String, String, String, u32)>,
3847 > = HashMap::default();
3848 for name in &pool.order {
3849 let Some(pc) = pool.get_if_materialized(name) else {
3850 continue;
3851 };
3852 for f in pc.static_fields.iter().chain(pc.instance_fields.iter()) {
3853 let disp = crate::classdec::java_ident(&f.name).into_owned();
3854 if root_segs.contains(&disp) {
3855 cand
3856 .entry(name.clone())
3857 .or_default()
3858 .push((disp, f.name.to_string(), f.desc.to_string(), f.access));
3859 }
3860 }
3861 }
3862 if cand.is_empty() {
3863 return out;
3864 }
3865 let cand_set: jdc_core::FxHashSet<String> = cand.keys().cloned().collect();
3866 let t0 = std::time::Instant::now();
3867 let (fam_segs, fam_nest) =
3868 crate::refscan::family_ref_segments(&pool.dexes, &cand_set);
3869 if std::env::var("DDC_STATS").is_ok() {
3870 eprintln!(
3871 "[renames] field-deshadow scan: cands={} scan={:?}",
3872 cand_set.len(),
3873 t0.elapsed()
3874 );
3875 }
3876 for (fam, fields) in cand {
3877 let own_root = fam.find('/').map(|i| fam[..i].to_string());
3878 let scanned = fam_segs.get(&fam);
3879 // NO nested-tail collision trigger here: a private field named
3880 // like a nested class does capture `Fam.b.member` in expression
3881 // position (lark LKEvent static-ctx family, ~1k lines), but
3882 // renaming it is corpus-toxic even gated private+body-
3883 // referenced — lark +32k (43,643 renames), weixin/weibo −200.
3884 // Eighth falsification of loose rename gates: the resolution
3885 // perturbation surface of mass field renames dwarfs the target
3886 // family. The fam_nest scan stays wired for a future
3887 // render-side fix (the capture is only unfixable INSIDE the
3888 // family file; foreign files could import the nested type).
3889 let _ = &fam_nest;
3890 let hits: Vec<(String, String, String)> = fields
3891 .into_iter()
3892 .filter(|(disp, _, _, _access)| {
3893 scanned.is_some_and(|s| s.contains(disp))
3894 || own_root.as_deref() == Some(disp.as_str())
3895 })
3896 .map(|(d, n, de, _)| (d, n, de))
3897 .collect();
3898 if hits.is_empty() {
3899 continue;
3900 }
3901 // Names already taken in the family's member scope: field and
3902 // method displays plus direct nested-class tails (member types
3903 // share the lexical scope of the rendered file). Built only for
3904 // hit families — the per-family setup must stay off the 175k-
3905 // candidate loop (an O(pool.order) nested scan there cost 60s on
3906 // weixin).
3907 let mut taken: jdc_core::FxHashSet<String> = jdc_core::FxHashSet::default();
3908 if let Some(pc) = pool.get_if_materialized(&fam) {
3909 for f in pc.static_fields.iter().chain(pc.instance_fields.iter()) {
3910 taken.insert(crate::classdec::java_ident(&f.name).into_owned());
3911 }
3912 for m in pc.direct_methods.iter().chain(pc.virtual_methods.iter()) {
3913 taken.insert(crate::classdec::java_ident(&m.name).into_owned());
3914 }
3915 }
3916 let mut nested_tails: jdc_core::FxHashSet<String> =
3917 jdc_core::FxHashSet::default();
3918 for child in pool.children_of(&fam) {
3919 if let Some(tail) = child.rsplit('$').next() {
3920 if !tail.is_empty() {
3921 taken.insert(tail.to_string());
3922 nested_tails.insert(tail.to_string());
3923 }
3924 }
3925 }
3926 for (disp, name, desc) in hits {
3927 let mut k = 0u32;
3928 let new_disp = loop {
3929 k += 1;
3930 let c = if k == 1 {
3931 format!("{disp}x")
3932 } else {
3933 format!("{disp}x{k}")
3934 };
3935 if !taken.contains(&c) && !root_segs.contains(&c) {
3936 break c;
3937 }
3938 };
3939 taken.insert(new_disp.clone());
3940 out.entry(std::sync::Arc::from(fam.as_str()))
3941 .or_default()
3942 .push(jdc_core::rename::FieldRename {
3943 name: std::sync::Arc::from(name.as_str()),
3944 desc: std::sync::Arc::from(desc.as_str()),
3945 display: std::sync::Arc::from(new_disp.as_str()),
3946 });
3947 }
3948 }
3949 out
3950}
3951pub use jdc_core::rename::apply_class_rename;
3952
3953// ---------------------------------------------------------------------------
3954// Default-package relocation.
3955// ---------------------------------------------------------------------------
3956
3957/// The synthetic package the default-package cohort currently relocates
3958/// to (Mutex, not OnceLock: `set_class_renames` replaces the registry per
3959/// pool, so a second pool in one process must be able to re-key this).
3960static ROOT_PKG: std::sync::Mutex<Option<String>> = std::sync::Mutex::new(None);
3961
3962/// The synthetic package display root-package classes were moved under
3963/// (`None` when no relocation ran). Consumers keying per-package caches
3964/// by the ORIGINAL package map this display back to `""`.
3965pub fn root_pkg_display() -> Option<String> {
3966 ROOT_PKG
3967 .lock()
3968 .unwrap_or_else(|e| e.into_inner())
3969 .clone()
3970}
3971
3972/// Default-package relocation: javac cannot resolve ANY reference from a
3973/// named package to an unnamed-package class (JLS 7.4.2 — `private it0 f;`
3974/// inside com.amap.X is 找不到符号 whatever the syntax; xingye ~198k +
3975/// Amap ~81k errors ≈ 73% of the whole-directory sweep total, and
3976/// kotlin.jvm.Function0 even EXTENDS a root class on xingye). No source
3977/// syntax escapes it — but the rename registry escapes the NAME: move the
3978/// whole root cohort under a synthetic package (`ddcroot/a.java`,
3979/// `package ddcroot;`) and every cross-package ref renders the resolvable
3980/// FQN `ddcroot.a`. Cohort-uniform: in-cohort refs keep resolving (one
3981/// new shared package), the case/lossy dedupes inside the cohort still
3982/// hold (a directory move preserves relative collisions), and the
3983/// synthetic segment is collision-bumped against every pool package and
3984/// class so it can never shadow or be shadowed. Runs LAST (prefixes the
3985/// settled displays). Fidelity note: declared package no longer matches
3986/// the binary name — the same trade every rename rule here makes, on a
3987/// cohort whose cross-package render is otherwise dead code.
3988/// Pool classes that shadow a FRAMEWORK class (d8 desugar stubs at the
3989/// framework FQN). The FILE identity relocates to the synthetic package,
3990/// but every REFERENCE renders the raw framework name — the bytecode's
3991/// FQN means the platform class (the stub is dead code the boot
3992/// classpath outranks).
3993pub(crate) static FW_SHADOW: std::sync::OnceLock<
3994 std::sync::Arc<jdc_core::FxHashSet<String>>,
3995> = std::sync::OnceLock::new();
3996
3997/// True when `internal` is a framework-shadow stub (fast empty-set
3998/// check when no corpus stubs exist).
3999pub fn is_fw_shadow(internal: &str) -> bool {
4000 let Some(s) = FW_SHADOW.get() else { return false };
4001 // Exact hit, or any $-chain PREFIX hits: a nested FRAMEWORK member
4002 // of a stub outer (`ImageDecoder$Source` with the stub
4003 // `ImageDecoder` in the pool) renders through the $-walk of the
4004 // stub's map entry — it must keep the raw framework name too.
4005 if s.contains(internal) {
4006 return true;
4007 }
4008 let mut idx = internal.len();
4009 while let Some(i) = internal[..idx].rfind('$') {
4010 if s.contains(&internal[..i]) {
4011 return true;
4012 }
4013 idx = i;
4014 }
4015 false
4016}
4017
4018/// d8's library/API desugaring embeds STUB classes under the framework's
4019/// own names (`android.graphics.Insets`, `android.app.Person` — minimal
4020/// shapes the runtime never loads, the boot classpath wins). Emitted at
4021/// their FQN they SHADOW the real framework class for every OTHER file
4022/// at compile time: javac resolves `android.graphics.Insets` to the
4023/// field-less stub and `insets.top` dies (reqable r2: the whole Insets
4024/// family). Relocate every pool class whose head resolves in the
4025/// embedded framework DB to a synthetic package; the rename registry
4026/// carries every reference.
4027fn fw_shadow_relocation(pool: &DexPool, map: &mut HashMap<String, String>) {
4028 let root_segs = pool.root_pkg_segs();
4029 // ANDROID-namespace only: a framework-shadow stub in another
4030 // namespace is usually a REAL bundled implementation (weibo ships
4031 // its own full org.json — relocating it broke every org.json user,
4032 // +86). d8's API-desugar stubs live under android/.
4033 let shadow_set: std::sync::Arc<jdc_core::FxHashSet<String>> =
4034 std::sync::Arc::new(
4035 pool.order
4036 .iter()
4037 .filter(|n| n.starts_with("android/"))
4038 .filter(|n| {
4039 let head = n.split('$').next().unwrap_or(n);
4040 crate::fwdb::find(head).is_some()
4041 })
4042 .map(|n| n.to_string())
4043 .collect(),
4044 );
4045 let _ = FW_SHADOW.set(shadow_set.clone());
4046 jdc_core::rename::set_fw_shadow(
4047 shadow_set.iter().cloned().collect(),
4048 );
4049 let mut pkg = "ddcfwstub".to_string();
4050 while pool.has_name(&pkg) || root_segs.contains(&pkg) {
4051 pkg.push('_');
4052 }
4053 let mut moved = 0usize;
4054 for name in &pool.order {
4055 // The relocation decision keys on the $-chain HEAD (a nested
4056 // stub rides its outer's identity); nested-of-pool-outer shapes
4057 // stay entry-less so apply_class_rename's $-walk composes them
4058 // through the relocated outer (the VasUIToken exact-entry
4059 // lesson from root_pkg_relocation). Android-namespace only —
4060 // see the set build above.
4061 if !name.starts_with("android/") {
4062 continue;
4063 }
4064 let head = name.split('$').next().unwrap_or(name);
4065 if crate::fwdb::find(head).is_none() {
4066 continue;
4067 }
4068 if head != name && pool.has_name(head) {
4069 continue;
4070 }
4071 let disp = map.get(name).cloned().unwrap_or_else(|| name.clone());
4072 if disp != *name {
4073 // An earlier rule renamed this class (obscuring/collision
4074 // family) — its display already diverges from the FQN and
4075 // the relocation prefix applies to the settled display.
4076 map.insert(name.clone(), format!("{pkg}/{disp}"));
4077 } else {
4078 map.insert(name.clone(), format!("{pkg}/{name}"));
4079 }
4080 moved += 1;
4081 }
4082 if std::env::var("DDC_STATS").is_ok() {
4083 eprintln!("[renames] fw-shadow relocation: {moved} classes -> {pkg}");
4084 }
4085}
4086
4087fn root_pkg_relocation(pool: &DexPool, map: &mut HashMap<String, String>) {
4088 // Gate on NAMES only (identical verdict in lazy browse and full
4089 // decompile — the display must not depend on materialization):
4090 // relocation matters only when BOTH cohorts exist — the JLS-dead
4091 // direction is named-package → root-package, so an all-root app
4092 // (hello fixtures, pure-demo dexes) keeps the faithful layout.
4093 let has_root = pool.order.iter().any(|n| !n.contains('/'));
4094 let has_named = pool.order.iter().any(|n| n.contains('/'));
4095 if !has_root || !has_named {
4096 *ROOT_PKG.lock().unwrap_or_else(|e| e.into_inner()) = None;
4097 return;
4098 }
4099 let root_segs = pool.root_pkg_segs();
4100 let mut pkg = "ddcroot".to_string();
4101 while pool.has_name(&pkg) || root_segs.contains(&pkg) {
4102 pkg.push('_');
4103 }
4104 // A root class whose simple name IS a top-level package segment
4105 // (`com`, `org`, ...) cannot simply MOVE into the synthetic
4106 // package: a same-package type shadows the package name in every
4107 // qualified ref from every cohort file (JLS 6.4.1 — rimet's root
4108 // `com`/`org` hijacked 235k `com.alibaba...` refs as 类 alibaba
4109 // 位置: 类 com). Those get a bumped first segment (`com` →
4110 // ddcroot/com2) — a display rename the registry carries through
4111 // every ref path, like the obscuring renames.
4112 const FW_SEGS: [&str; 10] = [
4113 "java", "javax", "android", "org", "junit", "dalvik", "libcore", "androidx", "kotlin",
4114 "kotlinx",
4115 ];
4116 let mut taken: jdc_core::FxHashSet<String> = jdc_core::FxHashSet::default();
4117 for name in &pool.order {
4118 if name.contains('/') {
4119 continue;
4120 }
4121 let disp = map.get(name).cloned().unwrap_or_else(|| name.clone());
4122 taken.insert(format!("{pkg}/{disp}"));
4123 }
4124 let mut moved = 0usize;
4125 let mut bumped = 0usize;
4126 for name in &pool.order {
4127 if name.contains('/') {
4128 continue;
4129 }
4130 let disp = map.get(name).cloned().unwrap_or_else(|| name.clone());
4131 if disp.contains('/') {
4132 continue; // defensive: some rule already packaged it
4133 }
4134 if disp == *name && name.contains('$') && find_outer_name(pool, name).is_some() {
4135 // An UNRENAMED nested name with a pool outer must stay
4136 // entry-less so apply_class_rename's $-walk composes it
4137 // through the outer's relocated+renamed display — an exact
4138 // entry short-circuits the walk and freezes the outer's
4139 // pre-rename simple (xingye's case-collision quads:
4140 // `ue5$a` exact-mapped to ddcroot/ue5$a instead of
4141 // ddcroot/ue5_4$a — 程序包ddcroot.ue5不存在 ×61k).
4142 // The case-collision machinery pre-inserts an IDENTITY
4143 // entry for every emission-set name, which short-circuits
4144 // the same walk and froze `$`-tails in the ROOT package
4145 // while their outer moved to ddcroot — family split: QQ's
4146 // VasUIToken$1..$1xx husks stayed at tree root, and every
4147 // `VasUIToken$N.INSTANCE` ref from ddcroot/VasUIToken.java
4148 // died (找不到符号 变量 VasUIToken$N ×202 in that one file).
4149 // Remove the pin; the walk composes through the outer.
4150 map.remove(name);
4151 continue;
4152 }
4153 let head = disp.split('$').next().unwrap_or(&disp);
4154 let collide = root_segs.contains(head) || FW_SEGS.contains(&head);
4155 if !collide {
4156 map.insert(name.clone(), format!("{pkg}/{disp}"));
4157 moved += 1;
4158 continue;
4159 }
4160 let rest = &disp[head.len()..];
4161 let mut k = 1u32;
4162 loop {
4163 k += 1;
4164 let nh = format!("{head}{k}");
4165 if root_segs.contains(&nh) || FW_SEGS.contains(&nh.as_str()) {
4166 continue;
4167 }
4168 let cand = format!("{pkg}/{nh}{rest}");
4169 if taken.contains(&cand) {
4170 continue;
4171 }
4172 taken.insert(cand.clone());
4173 map.insert(name.clone(), cand);
4174 moved += 1;
4175 bumped += 1;
4176 break;
4177 }
4178 }
4179 *ROOT_PKG.lock().unwrap_or_else(|e| e.into_inner()) = Some(pkg.clone());
4180 if std::env::var("DDC_STATS").is_ok() {
4181 eprintln!("[renames] root-pkg relocation: {moved} classes -> {pkg}/ ({bumped} segment-bumped)");
4182 }
4183}