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