ddc_dec/lib.rs
1//! ddc-dec — DEX → Java decompiler front-end over `jdc-core`.
2//!
3//! Pipeline per method: DEX code units → CFG → per-block register→IR
4//! lifting → `jdc-core` structuring/conversion → refinement passes →
5//! emission. Class-level rendering mirrors jcdc's `classdec` but reads
6//! DEX metadata (no signatures, no generics).
7
8pub mod cfg;
9pub mod cure_pass;
10pub mod classdec;
11pub mod fwdb;
12
13pub use classdec::{sanitize_fq, ClassOptions};
14pub mod ctx;
15pub mod lift;
16pub mod method;
17pub mod passes;
18pub mod platform;
19mod refscan;
20pub mod ssa_census;
21
22use jdc_core::FxHashMap as HashMap;
23
24use ddc_dex::annotations::{self, EncodedValue};
25
26/// A resolved static-field initializer.
27#[derive(Debug, Clone)]
28pub enum StaticValue {
29 Int(i64),
30 Float(f32),
31 Double(f64),
32 Str(String),
33 Type(String),
34 Boolean(bool),
35 Null,
36 /// (declaring class, field name, field descriptor) — enum constants /
37 /// const field refs. The desc participates in the rename-registry key
38 /// (owner, name, desc) so the render site can consult field_display.
39 Field(String, String, String),
40 Other,
41}
42use ddc_dex::{ClassDef, DexFile};
43use jdc_core::types::{
44 parse_field_descriptor, parse_method_descriptor, JavaType, MethodDescriptor,
45};
46
47// ---------------------------------------------------------------------------
48// Access flags (Dalvik numbering, JVM-compatible subset).
49// ---------------------------------------------------------------------------
50pub mod access {
51 pub const ACC_PUBLIC: u32 = 0x1;
52 pub const ACC_PRIVATE: u32 = 0x2;
53 pub const ACC_PROTECTED: u32 = 0x4;
54 pub const ACC_STATIC: u32 = 0x8;
55 pub const ACC_FINAL: u32 = 0x10;
56 pub const ACC_SYNCHRONIZED: u32 = 0x20;
57 pub const ACC_BRIDGE: u32 = 0x40;
58 pub const ACC_VARARGS: u32 = 0x80;
59 pub const ACC_NATIVE: u32 = 0x100;
60 pub const ACC_INTERFACE: u32 = 0x200;
61 pub const ACC_ABSTRACT: u32 = 0x400;
62 pub const ACC_STRICT: u32 = 0x800;
63 pub const ACC_SYNTHETIC: u32 = 0x1000;
64 pub const ACC_ANNOTATION: u32 = 0x2000;
65 pub const ACC_ENUM: u32 = 0x4000;
66 pub const ACC_CONSTRUCTOR: u32 = 0x1_0000;
67 pub const ACC_DECLARED_SYNCHRONIZED: u32 = 0x2_0000;
68}
69
70/// One field of a pooled class.
71#[derive(Debug, Clone)]
72pub struct PoolField {
73 /// Shared with the dex string table (field names repeat massively
74 /// under R8's single-letter namespace — one Arc per unique name).
75 pub name: std::sync::Arc<str>,
76 /// Field descriptor (`I`, `Ljava/lang/String;`, ...) — shared per
77 /// unique descriptor through the same table.
78 pub desc: std::sync::Arc<str>,
79 pub access: u32,
80 pub is_static: bool,
81}
82
83/// One method of a pooled class.
84#[derive(Debug, Clone)]
85pub struct PoolMethod {
86 /// Shared with the dex string table (the idx IS the dedup key).
87 pub name: std::sync::Arc<str>,
88 /// Method descriptor, shared per proto (a dex's proto table is the
89 /// descriptor dedup layer — 1.9M methods on lark share ~1 proto
90 /// table's worth of unique descriptors).
91 pub desc: std::sync::Arc<str>,
92 pub access: u32,
93 pub code_off: u32,
94 pub debug_info_off: u32,
95 /// Which DEX image the body lives in.
96 pub dex_idx: usize,
97}
98
99impl PoolMethod {
100 pub fn parsed_desc(&self) -> Option<MethodDescriptor> {
101 parse_method_descriptor(&self.desc)
102 }
103 pub fn is_static(&self) -> bool {
104 self.access & access::ACC_STATIC != 0
105 }
106 pub fn is_abstract_or_native(&self) -> bool {
107 self.access & (access::ACC_ABSTRACT | access::ACC_NATIVE) != 0
108 }
109}
110
111/// One pool entry: either already built, or name-registered with its
112/// (dex index, class_def index) locator for on-demand construction.
113enum ClassEntry {
114 Eager(Box<PoolClass>),
115 Lazy {
116 at: (usize, usize),
117 pc: std::sync::OnceLock<Box<PoolClass>>,
118 },
119 /// Name registered without an image: used by the two-tier single-class
120 /// path to keep global name-dependent rules (case/package/deshadow)
121 /// identical while only the target's dependency closure is parsed.
122 NameOnly,
123}
124
125/// A class materialized from one `class_def_item` (with ids resolved to
126/// names and the class data expanded).
127#[derive(Debug, Clone)]
128pub struct PoolClass {
129 pub name: String,
130 pub access: u32,
131 /// `None` for `java/lang/Object` roots.
132 pub super_name: Option<String>,
133 pub interfaces: Vec<String>,
134 pub source_file: Option<String>,
135 pub static_fields: Vec<PoolField>,
136 pub instance_fields: Vec<PoolField>,
137 pub direct_methods: Vec<PoolMethod>,
138 pub virtual_methods: Vec<PoolMethod>,
139 /// Static initial values aligned with the head of `static_fields`.
140 pub static_values: Vec<StaticValue>,
141 /// Nesting evidence with annotation ids resolved to names.
142 pub nesting: ResolvedNesting,
143 /// Which DEX image the class body lives in (provenance header).
144 pub dex_idx: usize,
145}
146
147/// Nesting evidence, names resolved at pooling time.
148#[derive(Debug, Clone, Default)]
149pub struct ResolvedNesting {
150 pub enclosing_class: Option<String>,
151 /// (class internal name, method name) of the enclosing method.
152 pub enclosing_method: Option<(String, String)>,
153 pub member_classes: Vec<String>,
154}
155
156impl PoolClass {
157 pub fn is_interface(&self) -> bool {
158 self.access & access::ACC_INTERFACE != 0
159 }
160
161 pub fn is_enum(&self) -> bool {
162 self.access & access::ACC_ENUM != 0
163 }
164
165 pub fn is_synthetic(&self) -> bool {
166 self.access & access::ACC_SYNTHETIC != 0
167 }
168
169 pub fn is_static_nested(&self) -> bool {
170 self.access & access::ACC_STATIC != 0
171 }
172
173 /// All methods in declaration order.
174 pub fn all_methods(&self) -> impl Iterator<Item = &PoolMethod> {
175 self.direct_methods
176 .iter()
177 .chain(self.virtual_methods.iter())
178 }
179
180 /// Find a method by name + descriptor.
181 pub fn find_method(&self, name: &str, desc: &str) -> Option<&PoolMethod> {
182 self.all_methods()
183 .find(|m| &*m.name == name && &*m.desc == desc)
184 }
185
186 /// Constructors `<init>` matching `arity` descriptor arguments.
187 pub fn ctors_by_arity(&self, arity: usize) -> Vec<&PoolMethod> {
188 self.all_methods()
189 .filter(|m| &*m.name == "<init>")
190 .filter(|m| {
191 m.parsed_desc()
192 .map(|d| d.args.len() == arity)
193 .unwrap_or(false)
194 })
195 .collect()
196 }
197
198 pub fn field_flags_of(&self, name: &str) -> Option<u32> {
199 self.static_fields
200 .iter()
201 .chain(self.instance_fields.iter())
202 .find(|f| f.name.as_ref() == name)
203 .map(|f| f.access)
204 }
205}
206
207/// Multi-DEX class pool: classes from all images, first definition wins.
208/// One facade-map entry: the public base plus, for PARTIALLY covered
209/// parts, the exact static members the base declares (None = the whole
210/// part rewrites; Some = per-member gate). Single map = one hash probe
211/// per static reference (two maps measured +2.4s on weibo).
212type FacadeTarget = (
213 String,
214 Option<
215 std::sync::Arc<(
216 jdc_core::FxHashSet<(String, String)>,
217 jdc_core::FxHashSet<String>,
218 )>,
219 >,
220);
221
222pub struct DexPool {
223 /// Images as Arcs: `dex()` hands out snapshots whose borrows stay
224 /// valid for the Arc's lifetime (decompile workers hold them), while
225 /// `release_images` empties the image data through exclusive access
226 /// when all of an image's classes are done (full-decompile driver
227 /// only — progressive callers keep images whole).
228 dexes: Vec<std::sync::Arc<DexFile>>,
229 /// Human label per image ("weibo!classes.dex") for the provenance
230 /// header; defaults to "dex N" until the CLI names the inputs.
231 pub dex_labels: Vec<String>,
232 /// Name → class entry. Eager pools materialize everything at add
233 /// time (full decompile, tests); lazy pools register names only and
234 /// materialize on first `get` (progressive `getclass` — one class's
235 /// PoolClass costs annotation reads, 145k of them cost ~0.2s).
236 classes: HashMap<String, ClassEntry>,
237 /// Class names in insertion order (stable for output).
238 pub order: Vec<String>,
239 /// Per-image remaining class count once `arm_retirement` fires.
240 retire_counts: std::sync::Mutex<Vec<u64>>,
241 retire_armed: std::sync::atomic::AtomicBool,
242 /// name → outer (computed once; `$` heuristic + dalvik annotations).
243 outers: std::sync::OnceLock<HashMap<String, Option<String>>>,
244 /// outer → direct children (computed once).
245 children: std::sync::OnceLock<HashMap<String, Vec<String>>>,
246 /// Per-image hot-reference interning (parallel to `dexes`).
247 ref_caches: Vec<DexRefCache>,
248 /// Set when the renderer materializes a NameOnly entry: the two-tier
249 /// single-class path then falls back to a full parse instead of
250 /// emitting degraded types.
251 name_only_missed: std::sync::atomic::AtomicBool,
252 /// Every NameOnly class the renderer asked for (drives on-demand image
253 /// loading in the two-tier path; the first also serves diagnostics).
254 name_only_miss: std::sync::Mutex<Vec<String>>,
255 /// Static-closure pools: facade analysis skips unparsed NameOnly classes
256 /// instead of recording misses (unrelated facades must not force images).
257 facade_skip_unparsed: std::sync::atomic::AtomicBool,
258 /// Synthetic-static accessor code snapshots keyed by (dex_idx,
259 /// code_off). `inline_accessors` reads accessor bodies ACROSS images
260 /// during worker runs; image retirement can release those bytes
261 /// mid-run, making the inline decision — and with it var numbering
262 /// downstream — dependent on worker completion interleaving
263 /// (observed: reqable enum-constant args flipping `var0`/`var1`
264 /// between identical runs; DDC_NORETIRE=1 stable). The full-decompile
265 /// driver fills this BEFORE `arm_retirement`; progressive/lazy pools
266 /// never retire and fall through to live reads.
267 accessor_code: std::sync::OnceLock<AccessorSnapshots>,
268 /// Cached package → direct-class simple names (import-collision gate).
269 pkg_simples:
270 std::sync::OnceLock<jdc_core::FxHashMap<String, std::sync::Arc<jdc_core::FxHashSet<String>>>>,
271 /// First path segments of all packages (`v2` for `v2/n`): local or
272 /// field names equal to one capture package-qualified renders
273 /// (`v2.n.a`) — deshadow_locals consults this.
274 root_segs: std::sync::OnceLock<jdc_core::FxHashSet<String>>,
275 /// Partially-covered facade parts (see kotlin_facade_partial_map).
276 /// Kotlin multi-file facade parts → public facade (see
277 /// `kotlin_facade_map`).
278 kotlin_facades: std::sync::OnceLock<jdc_core::FxHashMap<String, FacadeTarget>>,
279 /// Per class internal name: the set of `<init>` ARGUMENT descriptors
280 /// (the `(..)` inner text) REFERENCED through that class anywhere in
281 /// the pool. Dex resolves a `<init>` method-ref up the superclass
282 /// chain, so a subclass `super(..)` may target `C.<init>(sig)` that C
283 /// does NOT declare (an ancestor does) — Java has no inherited ctors,
284 /// so C needs a bridge. Computed once from the method-id tables
285 /// (parsed, survive image retirement). See `inherited_ctor_bridges`.
286 ctor_refs: std::sync::OnceLock<jdc_core::FxHashMap<String, jdc_core::FxHashSet<String>>>,
287}
288
289/// Raw code_item bytes of every synthetic-static accessor, keyed by
290/// (dex_idx, code_off) — see `snapshot_accessor_code`.
291type AccessorSnapshots = HashMap<(usize, u32), std::sync::Arc<[u8]>>;
292
293/// Lazily-filled, thread-shared reference interning for one image.
294///
295/// The lifter used to rebuild class names, field/method names and WHOLE
296/// method descriptors per INSTRUCTION (`method_ref` alone: N param
297/// Strings → `format!` descriptor → re-parse — 6-10 allocations per
298/// invoke). Every entry here is built at most once per (image, table
299/// index); lifts and the endless IR clones downstream are refcount
300/// bumps. Slots are `OnceLock` so worker threads fill them race-free
301/// without a lock on the hot path.
302pub(crate) struct DexRefCache {
303 proto_descs: Box<[std::sync::OnceLock<std::sync::Arc<jdc_core::types::MethodDescriptor>>]>,
304 type_names: Box<[std::sync::OnceLock<std::sync::Arc<str>>]>,
305 type_javas: Box<[std::sync::OnceLock<std::sync::Arc<jdc_core::types::JavaType>>]>,
306}
307
308impl DexRefCache {
309 fn new(protos: usize, types: usize) -> Self {
310 DexRefCache {
311 proto_descs: (0..protos).map(|_| std::sync::OnceLock::new()).collect(),
312 type_names: (0..types).map(|_| std::sync::OnceLock::new()).collect(),
313 type_javas: (0..types).map(|_| std::sync::OnceLock::new()).collect(),
314 }
315 }
316}
317
318impl DexPool {
319 pub fn new() -> Self {
320 DexPool {
321 dexes: Vec::new(),
322 retire_armed: std::sync::atomic::AtomicBool::new(false),
323 dex_labels: Vec::new(),
324 classes: HashMap::default(),
325 order: Vec::new(),
326 retire_counts: std::sync::Mutex::new(Vec::new()),
327 outers: std::sync::OnceLock::new(),
328 children: std::sync::OnceLock::new(),
329 ref_caches: Vec::new(),
330 name_only_missed: std::sync::atomic::AtomicBool::new(false),
331 name_only_miss: std::sync::Mutex::new(Vec::new()),
332 facade_skip_unparsed: std::sync::atomic::AtomicBool::new(false),
333 accessor_code: std::sync::OnceLock::new(),
334 pkg_simples: std::sync::OnceLock::new(),
335 root_segs: std::sync::OnceLock::new(),
336 kotlin_facades: std::sync::OnceLock::new(),
337 ctor_refs: std::sync::OnceLock::new(),
338 }
339 }
340
341 /// Referenced `<init>` argument-descriptor set for a class internal
342 /// name (see `ctor_refs`). Builds the whole index on first call by
343 /// scanning every image's method-id table for `<init>` refs — cheap
344 /// (flat table, integer name-idx compare against the interned
345 /// "<init>" string) and done once for the whole decompile.
346 pub fn ctor_ref_sigs(&self, class: &str) -> Option<&jdc_core::FxHashSet<String>> {
347 self.ctor_refs
348 .get_or_init(|| {
349 let mut map: jdc_core::FxHashMap<String, jdc_core::FxHashSet<String>> =
350 jdc_core::FxHashMap::default();
351 for dex in &self.dexes {
352 let n = dex.method_count();
353 for i in 0..n {
354 let m = dex.method(i as u32);
355 if dex.string(m.name_idx) != "<init>" {
356 continue;
357 }
358 let desc = dex.proto_desc(m.proto_idx);
359 let (Some(lo), Some(hi)) = (desc.find('('), desc.find(')')) else {
360 continue;
361 };
362 map.entry(dex.class_name(m.class_idx))
363 .or_default()
364 .insert(desc[lo + 1..hi].to_string());
365 }
366 }
367 map
368 })
369 .get(class)
370 }
371
372 /// Shared `MethodDescriptor` of a proto (one parse per image).
373 pub fn proto_desc_parsed(
374 &self,
375 di: usize,
376 proto_idx: u32,
377 ) -> std::sync::Arc<jdc_core::types::MethodDescriptor> {
378 let Some(dex) = self.dexes.get(di) else {
379 return std::sync::Arc::new(jdc_core::types::MethodDescriptor {
380 args: Vec::new(),
381 ret: jdc_core::types::JavaType::Void,
382 });
383 };
384 let build = || {
385 std::sync::Arc::new(parse_proto_desc(dex, proto_idx))
386 };
387 match self
388 .ref_caches
389 .get(di)
390 .and_then(|c| c.proto_descs.get(proto_idx as usize))
391 {
392 Some(slot) => slot.get_or_init(build).clone(),
393 None => build(),
394 }
395 }
396
397 /// Shared internal class name of a type id (one strip+alloc per image).
398 pub fn type_name_arc(&self, di: usize, type_idx: u32) -> std::sync::Arc<str> {
399 let Some(dex) = self.dexes.get(di) else {
400 return std::sync::Arc::from("");
401 };
402 let build = || std::sync::Arc::from(dex.class_name(type_idx).as_str());
403 match self
404 .ref_caches
405 .get(di)
406 .and_then(|c| c.type_names.get(type_idx as usize))
407 {
408 Some(slot) => slot.get_or_init(build).clone(),
409 None => build(),
410 }
411 }
412
413 /// Shared parsed type of a type id.
414 pub fn type_java(&self, di: usize, type_idx: u32) -> std::sync::Arc<jdc_core::types::JavaType> {
415 let Some(dex) = self.dexes.get(di) else {
416 return std::sync::Arc::new(jdc_core::types::JavaType::Object(
417 std::sync::Arc::from("java/lang/Object"),
418 ));
419 };
420 let build = || std::sync::Arc::new(desc_type(dex.type_name(type_idx)));
421 match self
422 .ref_caches
423 .get(di)
424 .and_then(|c| c.type_javas.get(type_idx as usize))
425 {
426 Some(slot) => slot.get_or_init(build).clone(),
427 None => build(),
428 }
429 }
430
431 /// name → outer map, computed lazily once.
432 fn outer_map(&self) -> &HashMap<String, Option<String>> {
433 self.outers.get_or_init(|| {
434 self.order
435 .iter()
436 .map(|n| (n.clone(), find_outer_name(self, n)))
437 .collect()
438 })
439 }
440
441 /// The outer class of `name`, from the cached map (borrowed).
442 pub fn outer_of(&self, name: &str) -> Option<&str> {
443 self.outer_map().get(name).and_then(|o| o.as_deref())
444 }
445
446 /// Direct nested children of `internal` (any `$` depth 1), cached.
447 /// (Perf: returns a borrowed slice — the owned-Vec clone ran once per
448 /// class on 98k-class runs.)
449 pub fn children_of(&self, internal: &str) -> &[String] {
450 if self.children.get().is_none() {
451 // Build FIRST, set ONCE: the previous shape set an empty map
452 // to claim the OnceLock and then silently failed to store the
453 // real index (`let _ = set(...)` on an initialized lock) —
454 // children_of returned [] forever, so member nested classes
455 // were neither inlined nor emitted as files (Guard$Report
456 // vanished whole; every corpus run since the borrow refactor
457 // dropped them, invisible to the syntax-only gate).
458 let mut idx: HashMap<String, Vec<String>> = HashMap::default();
459 for name in &self.order {
460 if let Some(outer) = self.outer_of(name).map(str::to_string) {
461 idx.entry(outer).or_default().push(name.clone());
462 }
463 }
464 // A racing thread may have set an identical map first — the
465 // index is a pure function of `order`, so either copy is right.
466 let _ = self.children.set(idx);
467 }
468 self.children
469 .get()
470 .and_then(|m| m.get(internal))
471 .map(|v| v.as_slice())
472 .unwrap_or(&[])
473 }
474
475 /// Adds one image; duplicate class names keep their first definition.
476 /// Returns the image's index — call `set_dex_label` with it to give the
477 /// provenance header a real origin instead of the default "dex N".
478 pub fn add_dex(&mut self, dex: DexFile) -> usize {
479 let dex_idx = self.dexes.len();
480 // Pre-size the name map: 240k-entry growth rehashed repeatedly
481 // (reserve_rehash showed up in corpus profiles).
482 self.classes.reserve(dex.class_defs.len());
483 // The annotation reader borrows the image; pool classes are built
484 // before ownership moves into `self.dexes` (no full-image copy).
485 for cd in &dex.class_defs {
486 let name = dex.class_name(cd.class_idx);
487 if self.classes.contains_key(&name) {
488 continue;
489 }
490 let pc = pool_class_of(&dex, dex.raw(), cd, dex_idx);
491 self.classes.insert(name.clone(), ClassEntry::Eager(Box::new(pc)));
492 self.order.push(name);
493 }
494 self.ref_caches
495 .push(DexRefCache::new(dex.proto_count(), dex.type_count()));
496 self.dexes.push(std::sync::Arc::new(dex));
497 self.retire_counts.lock().unwrap().push(0);
498 self.dex_labels.push(format!("dex {}", dex_idx));
499 dex_idx
500 }
501
502 /// Registers class NAMES only; the PoolClass (annotation reads
503 /// included) is materialized on first `get`. Duplicate class names
504 /// keep their first definition, matching `add_dex`.
505 /// Register class names without any image. Later `add_dex_lazy` for a
506 /// real image keeps the first registration (first-wins, matching the
507 /// eager/lazy builders); names never materialize.
508 pub fn add_names(&mut self, names: impl IntoIterator<Item = String>) {
509 for name in names {
510 if self.classes.contains_key(&name) {
511 continue;
512 }
513 self.classes.insert(name.clone(), ClassEntry::NameOnly);
514 self.order.push(name);
515 }
516 }
517
518 pub fn add_dex_lazy(&mut self, dex: DexFile) -> usize {
519 self.add_dex_lazy_arc(std::sync::Arc::new(dex))
520 }
521
522 /// Lazy registration from a shared image. The two-tier single-class
523 /// path rebuilds its pool after loading additional images; keeping the
524 /// parsed image in an Arc avoids re-parsing it on every round.
525 pub fn add_dex_lazy_arc(&mut self, dex: std::sync::Arc<DexFile>) -> usize {
526 let dex_idx = self.dexes.len();
527 self.classes.reserve(dex.class_defs.len());
528 for (ci, cd) in dex.class_defs.iter().enumerate() {
529 let name = dex.class_name(cd.class_idx);
530 if self.classes.contains_key(&name) {
531 continue;
532 }
533 self.classes.insert(
534 name.clone(),
535 ClassEntry::Lazy {
536 at: (dex_idx, ci),
537 pc: std::sync::OnceLock::new(),
538 },
539 );
540 self.order.push(name);
541 }
542 self.ref_caches
543 .push(DexRefCache::new(dex.proto_count(), dex.type_count()));
544 self.dexes.push(dex);
545 self.retire_counts.lock().unwrap().push(0);
546 self.dex_labels.push(format!("dex {}", dex_idx));
547 dex_idx
548 }
549
550 /// Label the image at `idx` (see `add_dex`).
551 pub fn set_dex_label(&mut self, idx: usize, label: impl Into<String>) {
552 if let Some(slot) = self.dex_labels.get_mut(idx) {
553 *slot = label.into();
554 }
555 }
556
557 pub fn len(&self) -> usize {
558 self.classes.len()
559 }
560
561 pub fn is_empty(&self) -> bool {
562 self.classes.is_empty()
563 }
564
565 pub fn get(&self, internal: &str) -> Option<&PoolClass> {
566 if matches!(self.classes.get(internal), Some(ClassEntry::NameOnly)) {
567 self.name_only_missed
568 .store(true, std::sync::atomic::Ordering::Relaxed);
569 let mut missed = self.name_only_miss.lock().unwrap();
570 if !missed.iter().any(|n| n == internal) {
571 missed.push(internal.to_string());
572 }
573 return None;
574 }
575 self.get_inner(internal, None)
576 }
577
578 /// Skip unparsed NameOnly classes during Kotlin facade analysis.
579 pub fn set_facade_skip_unparsed(&self) {
580 self.facade_skip_unparsed
581 .store(true, std::sync::atomic::Ordering::Relaxed);
582 }
583
584 fn facade_get(&self, name: &str) -> Option<&PoolClass> {
585 if self
586 .facade_skip_unparsed
587 .load(std::sync::atomic::Ordering::Relaxed)
588 {
589 match self.classes.get(name)? {
590 ClassEntry::NameOnly => None,
591 _ => self.get_inner(name, None),
592 }
593 } else {
594 self.get(name)
595 }
596 }
597
598 /// Clear the NameOnly-materialization marker (call before rendering).
599 pub fn clear_name_only_miss(&self) {
600 self.name_only_missed
601 .store(false, std::sync::atomic::Ordering::Relaxed);
602 self.name_only_miss.lock().unwrap().clear();
603 }
604
605 /// Name of the first NameOnly class the renderer requested.
606 pub fn name_only_miss_name(&self) -> Option<String> {
607 self.name_only_miss.lock().unwrap().first().cloned()
608 }
609
610 /// Every NameOnly class the renderer requested since the last clear.
611 pub fn name_only_misses(&self) -> Vec<String> {
612 self.name_only_miss.lock().unwrap().clone()
613 }
614
615 /// True when rendering needed a class whose image was not parsed.
616 pub fn name_only_missed(&self) -> bool {
617 self.name_only_missed
618 .load(std::sync::atomic::Ordering::Relaxed)
619 }
620
621 /// `get` with optional pre-taken raw-image snapshots (one per
622 /// image). `materialize_all` hoists the `DexFile::raw()` lock out of
623 /// the per-class loop — the parallel materialize used to take the
624 /// same per-image lock once PER CLASS across all threads.
625 fn get_inner(&self, internal: &str, raws: Option<&[&[u8]]>) -> Option<&PoolClass> {
626 match self.classes.get(internal)? {
627 ClassEntry::NameOnly => None,
628 ClassEntry::Eager(pc) => Some(&**pc),
629 ClassEntry::Lazy { at, pc } => {
630 if let Some(built) = pc.get() {
631 return Some(&**built);
632 }
633 let (di, ci) = *at;
634 let dex = self.dexes.get(di)?;
635 let cd = dex.class_defs.get(ci)?;
636 let raw = match raws {
637 Some(r) => r.get(di).copied().unwrap_or(&[]),
638 None => dex.raw(),
639 };
640 let built = Box::new(pool_class_of(dex, raw, cd, di));
641 Some(&**pc.get_or_init(move || built))
642 }
643 }
644 }
645
646 /// Materialize every lazy entry (parallel across the given chunk
647 /// count). After this the pool is observationally identical to an
648 /// eagerly-built one — including the annotation-aware outer-map —
649 /// while the annotation reads ran on worker threads instead of the
650 /// serial pool build (~0.2s on weibo).
651 pub fn materialize_all(&self, threads: usize) {
652 let keys: Vec<String> = self
653 .classes
654 .iter()
655 .filter(|(_, e)| matches!(e, ClassEntry::Lazy { .. }))
656 .map(|(k, _)| k.clone())
657 .collect();
658 if keys.is_empty() {
659 return;
660 }
661 // One raw() snapshot per image for the whole sweep (see
662 // get_inner). Safe window: materialize_all completes before
663 // retirement is armed, so no image releases while these slices
664 // are alive.
665 let raws: Vec<&[u8]> = self.dexes.iter().map(|d| d.raw()).collect();
666 let threads = threads.max(1);
667 let next = std::sync::atomic::AtomicUsize::new(0);
668 std::thread::scope(|scope| {
669 for _ in 0..threads {
670 scope.spawn(|| loop {
671 let i = next.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
672 let Some(k) = keys.get(i) else { break };
673 let _ = self.get_inner(k, Some(&raws));
674 });
675 }
676 });
677 }
678
679 /// `get` without materializing: None for not-yet-built lazy entries.
680 fn get_if_materialized(&self, internal: &str) -> Option<&PoolClass> {
681 match self.classes.get(internal)? {
682 ClassEntry::NameOnly => None,
683 ClassEntry::Eager(pc) => Some(&**pc),
684 ClassEntry::Lazy { pc, .. } => pc.get().map(|b| &**b),
685 }
686 }
687
688 /// Name registered (no materialization).
689 pub fn has_name(&self, internal: &str) -> bool {
690 self.classes.contains_key(internal)
691 }
692
693 /// One image as an Arc snapshot (borrows of the snapshot live as
694 /// long as the caller holds the Arc).
695 pub fn dex(&self, idx: usize) -> Option<std::sync::Arc<DexFile>> {
696 self.dexes.get(idx).cloned()
697 }
698
699 pub fn dex_count(&self) -> usize {
700 self.dexes.len()
701 }
702
703 /// Raw image list (install-phase whole-program scans).
704 pub fn dexes(&self) -> &[std::sync::Arc<ddc_dex::DexFile>] {
705 &self.dexes
706 }
707
708 /// First path segments of every package in the pool (cached once).
709 pub fn root_pkg_segs(&self) -> &jdc_core::FxHashSet<String> {
710 self.root_segs.get_or_init(|| {
711 self.order
712 .iter()
713 .filter_map(|n| n.find('/').map(|i| n[..i].to_string()))
714 .collect()
715 })
716 }
717
718 /// Package → simple names of its direct classes, cached once per
719 /// pool (the import-collision gate consults it per class — the
720 /// uncached scan was O(classes²) and 10×'d weixin's wall time).
721 pub fn package_simples(
722 &self,
723 ) -> &jdc_core::FxHashMap<String, std::sync::Arc<jdc_core::FxHashSet<String>>> {
724 // Arc'd sets: the per-file import machinery BORROWS a package's
725 // set (an Arc clone) instead of deep-copying it — WhatsApp's
726 // ~10k-class package paid two full copies per class (~200M
727 // String allocations), the #2/#3 CPU frames on its profile.
728 self.pkg_simples.get_or_init(|| {
729 let mut m: jdc_core::FxHashMap<String, std::sync::Arc<jdc_core::FxHashSet<String>>> =
730 jdc_core::FxHashMap::default();
731 let mut raw: jdc_core::FxHashMap<String, jdc_core::FxHashSet<String>> =
732 jdc_core::FxHashMap::default();
733 for n in &self.order {
734 if let Some((pkg, simple)) = n.rsplit_once('/') {
735 raw.entry(pkg.to_string())
736 .or_default()
737 .insert(simple.to_string());
738 }
739 }
740 m.extend(raw.into_iter().map(|(k, v)| (k, std::sync::Arc::new(v))));
741 m
742 })
743 }
744
745 /// Kotlin multi-file class facades: the compiler splits a file facade
746 /// (`kotlin.text.StringsKt`) into package-private parts
747 /// (`StringsKt__StringsKt`, `StringsKt__StringsJVMKt`) and call sites
748 /// target the PARTS directly. Java cannot access a package-private
749 /// class cross-package ("StringsKt__StringsKt在kotlin.text中不是公共的"
750 /// — 1.4k lark root errors), but the public facade EXTENDS the parts,
751 /// so static members resolve through it: rewrite the reference's owner
752 /// class to the facade. part → base, gated on base existing, being
753 /// public, and transitively extending the part (static inheritance is
754 /// the resolution guarantee).
755 pub fn kotlin_facade_map(&self) -> &jdc_core::FxHashMap<String, FacadeTarget> {
756 self.kotlin_facades.get_or_init(|| {
757 let t_fb = std::time::Instant::now();
758 let mut m: jdc_core::FxHashMap<String, FacadeTarget> =
759 jdc_core::FxHashMap::default();
760 let mut npartial = 0usize;
761 for name in &self.order {
762 let Some(i) = name.find("__") else { continue };
763 let base = &name[..i];
764 if base.is_empty() {
765 continue;
766 }
767 let Some(basec) = self.facade_get(base) else {
768 continue;
769 };
770 if basec.access & crate::access::ACC_PUBLIC == 0 {
771 continue;
772 }
773 // base must transitively extend the part...
774 let mut cur = basec.super_name.clone();
775 let mut hops = 0u32;
776 let mut reaches = false;
777 while let Some(c) = cur {
778 if c == *name {
779 reaches = true;
780 break;
781 }
782 if hops >= 8 {
783 break;
784 }
785 hops += 1;
786 cur = self.facade_get(&c).and_then(|cc| cc.super_name.clone());
787 }
788 // ...OR cover the part's static members: the Kotlin
789 // compiler MERGES every part's statics into the public
790 // facade, but the facade's super chain only includes ONE
791 // part — lark's `CollectionsKt extends
792 // CollectionsKt___CollectionsKt` leaves
793 // `CollectionsKt__CollectionsJVMKt` uncovered (463
794 // 不是公共的 errors). Member coverage (name+desc for
795 // methods, name for fields) is the precise rewrite
796 // condition: every reference through the facade then
797 // resolves exactly as through the part.
798 if reaches {
799 m.insert(name.clone(), (base.to_string(), None));
800 continue;
801 }
802 // Member-coverage fallback: the Kotlin compiler MERGES
803 // the parts' statics into the public facade, but R8 can
804 // sever the super chain (lark: `CollectionsKt extends
805 // CollectionsKt___CollectionsKt extends
806 // kotlin/collections/s` — `__CollectionsJVMKt` is off
807 // the chain, 463 不是公共的 errors) and may drop facade
808 // members no direct facade call needed. Map at (name,
809 // desc) granularity then: covered members rewrite, the
810 // rest keep the part owner (a latent error either way —
811 // Java cannot access the package-private part).
812 let Some(part) = self.facade_get(name) else {
813 continue;
814 };
815 let base_methods: jdc_core::FxHashSet<(&str, &str)> = basec
816 .direct_methods
817 .iter()
818 .chain(basec.virtual_methods.iter())
819 .map(|mm| (&*mm.name, &*mm.desc))
820 .collect();
821 let base_fields: jdc_core::FxHashSet<&str> =
822 basec.static_fields.iter().map(|f| &*f.name).collect();
823 let mut methods: jdc_core::FxHashSet<(String, String)> =
824 jdc_core::FxHashSet::default();
825 let mut fields: jdc_core::FxHashSet<String> =
826 jdc_core::FxHashSet::default();
827 let mut all = true;
828 for mm in part
829 .direct_methods
830 .iter()
831 .chain(part.virtual_methods.iter())
832 .filter(|mm| mm.access & crate::access::ACC_STATIC != 0)
833 {
834 if base_methods.contains(&(&*mm.name, &*mm.desc)) {
835 methods.insert((mm.name.to_string(), mm.desc.to_string()));
836 } else {
837 all = false;
838 }
839 }
840 for pf in &part.static_fields {
841 if base_fields.contains(&*pf.name) {
842 fields.insert(pf.name.to_string());
843 } else {
844 all = false;
845 }
846 }
847 if all {
848 m.insert(name.clone(), (base.to_string(), None));
849 } else if !methods.is_empty() || !fields.is_empty() {
850 npartial += 1;
851 m.insert(
852 name.clone(),
853 (base.to_string(), Some(std::sync::Arc::new((methods, fields)))),
854 );
855 }
856 }
857 if std::env::var("DDC_STATS").is_ok() {
858 eprintln!(
859 "[renames] facade map: entries={} partial={} build={:?}",
860 m.len(),
861 npartial,
862 t_fb.elapsed()
863 );
864 }
865 m
866 })
867 }
868
869 pub fn class_names(&self) -> impl Iterator<Item = &str> {
870 self.order.iter().map(|s| s.as_str())
871 }
872
873 /// True when `sub` is assignable to `sup` (internal names), walking the
874 /// pool's hierarchy. Unknown classes are never assignable (conservative).
875 pub fn is_subtype(&self, sub: &str, sup: &str) -> bool {
876 if sub == sup {
877 return true;
878 }
879 // Arrays: covariance by element type when both are arrays.
880 if let (Some(sub_elem), Some(sup_elem)) = (array_elem(sub), array_elem(sup)) {
881 if sub_elem.starts_with('L')
882 && sup_elem.starts_with('L')
883 && sub_elem.ends_with(';')
884 && sup_elem.ends_with(';')
885 {
886 return self.is_subtype(
887 &sub_elem[1..sub_elem.len() - 1],
888 &sup_elem[1..sup_elem.len() - 1],
889 );
890 }
891 // Multi-dimensional: recurse on the element descriptors
892 // (`[[LSavedState;` <: `[[LObject;` peels one dimension at a
893 // time); the equality fallback below stays for primitive
894 // elements (`[I` vs `[J` are invariant).
895 if sub_elem.starts_with('[') && sup_elem.starts_with('[') {
896 return self.is_subtype(sub_elem, sup_elem);
897 }
898 return sub_elem == sup_elem;
899 }
900 // Everything reference-typed is Object-assignable (the pool walk
901 // used to fall off the chain end and answer false for
902 // String <: Object — the hand-written Object-formal special case
903 // in the ambiguity pin existed for exactly this hole).
904 if sup == "java/lang/Object" {
905 return !matches!(
906 sub,
907 "I" | "Z" | "B" | "S" | "C" | "J" | "F" | "D" | "V"
908 );
909 }
910 let mut cur = self.get(sub);
911 if cur.is_none() {
912 // `sub` is a framework (or phantom) type: the embedded API
913 // database decides (phantoms are absent there too → false).
914 return crate::fwdb::is_subtype(sub, sup);
915 }
916 let mut hops = 0;
917 while let Some(c) = cur {
918 hops += 1;
919 if hops > 64 {
920 return false;
921 }
922 for i in &c.interfaces {
923 if i == sup || self.is_subtype(i, sup) {
924 return true;
925 }
926 }
927 match &c.super_name {
928 Some(s) if s == sup => return true,
929 Some(s) if s != "java/lang/Object" => {
930 match self.get(s) {
931 Some(sc) => cur = Some(sc),
932 // Framework boundary: continue in the embedded DB.
933 None => return crate::fwdb::is_subtype(s, sup),
934 }
935 }
936 _ => return false,
937 }
938 }
939 false
940 }
941}
942
943impl Default for DexPool {
944 fn default() -> Self {
945 Self::new()
946 }
947}
948
949fn array_elem(desc: &str) -> Option<&str> {
950 desc.strip_prefix('[')
951}
952
953fn resolve_static_value(v: &EncodedValue, dex: &DexFile) -> StaticValue {
954 match v {
955 EncodedValue::Byte(x) => StaticValue::Int(*x as i64),
956 EncodedValue::Short(x) => StaticValue::Int(*x as i64),
957 EncodedValue::Char(x) => StaticValue::Int(*x as i64),
958 EncodedValue::Int(x) => StaticValue::Int(*x as i64),
959 EncodedValue::Long(x) => StaticValue::Int(*x),
960 EncodedValue::Float(x) => StaticValue::Float(*x),
961 EncodedValue::Double(x) => StaticValue::Double(*x),
962 EncodedValue::String(i) => StaticValue::Str(dex.string(*i).to_string()),
963 EncodedValue::Type(i) => StaticValue::Type(dex.class_name(*i)),
964 EncodedValue::Boolean(b) => StaticValue::Boolean(*b),
965 EncodedValue::Null => StaticValue::Null,
966 EncodedValue::Enum(i) | EncodedValue::Field(i) => {
967 let f = dex.field(*i);
968 StaticValue::Field(
969 dex.class_name(f.class_idx),
970 dex.string(f.name_idx).to_string(),
971 dex.type_name(f.type_idx).to_string(),
972 )
973 }
974 _ => StaticValue::Other,
975 }
976}
977
978fn pool_class_of(dex: &DexFile, raw: &[u8], cd: &ClassDef, dex_idx: usize) -> PoolClass {
979 let name = dex.class_name(cd.class_idx);
980 let super_name = if cd.superclass_idx == ddc_dex::NO_INDEX {
981 None
982 } else {
983 let s = dex.class_name(cd.superclass_idx);
984 if s.is_empty() || s == "java/lang/Object" {
985 None
986 } else {
987 Some(s)
988 }
989 };
990 let interfaces = dex
991 .interfaces_of(cd)
992 .into_iter()
993 .map(|t| dex.class_name(t))
994 .collect();
995 let source_file = if cd.source_file_idx == ddc_dex::NO_INDEX {
996 None
997 } else {
998 Some(dex.string(cd.source_file_idx).to_string())
999 };
1000
1001 let data = dex.class_data(cd);
1002 let mk_field = |ef: &ddc_dex::EncodedField, is_static: bool| {
1003 let f = dex.field(ef.field_idx);
1004 PoolField {
1005 name: dex.string_arc(f.name_idx),
1006 desc: dex.type_name_arc(f.type_idx),
1007 access: ef.access_flags,
1008 is_static,
1009 }
1010 };
1011 let static_fields: Vec<PoolField> = data
1012 .static_fields
1013 .iter()
1014 .map(|f| mk_field(f, true))
1015 .collect();
1016 let instance_fields: Vec<PoolField> = data
1017 .instance_fields
1018 .iter()
1019 .map(|f| mk_field(f, false))
1020 .collect();
1021 let mk_method = |em: &ddc_dex::EncodedMethod| {
1022 let m = dex.method(em.method_idx);
1023 let (code_off, debug_info_off) = match dex.debug_info_off_at(em.code_off) {
1024 Some(d) => (em.code_off, d),
1025 None => (0, 0),
1026 };
1027 PoolMethod {
1028 name: dex.string_arc(m.name_idx),
1029 desc: dex.proto_desc(m.proto_idx),
1030 access: em.access_flags,
1031 code_off,
1032 debug_info_off,
1033 dex_idx,
1034 }
1035 };
1036 let direct_methods: Vec<PoolMethod> = data.direct_methods.iter().map(&mk_method).collect();
1037 let virtual_methods: Vec<PoolMethod> = data.virtual_methods.iter().map(&mk_method).collect();
1038
1039 let static_values: Vec<StaticValue> = dex
1040 .static_values(cd.static_values_off)
1041 .into_iter()
1042 .map(|v| resolve_static_value(&v, dex))
1043 .collect();
1044 let anns = if cd.annotations_off != 0 {
1045 annotations::read_class_annotations(raw, cd.annotations_off)
1046 } else {
1047 Vec::new()
1048 };
1049 let raw = annotations::nesting_from(&anns, &|t: u32| dex.type_name(t).to_string());
1050 let nesting = ResolvedNesting {
1051 enclosing_class: raw.enclosing_class.map(|t| dex.class_name(t)),
1052 enclosing_method: raw.enclosing_method.map(|m| {
1053 let mid = dex.method(m);
1054 (
1055 dex.class_name(mid.class_idx),
1056 dex.string(mid.name_idx).to_string(),
1057 )
1058 }),
1059 member_classes: raw
1060 .member_classes
1061 .into_iter()
1062 .map(|t| dex.class_name(t))
1063 .collect(),
1064 };
1065
1066 PoolClass {
1067 name,
1068 access: cd.access_flags,
1069 super_name,
1070 interfaces,
1071 source_file,
1072 static_fields,
1073 instance_fields,
1074 direct_methods,
1075 virtual_methods,
1076 static_values,
1077 nesting,
1078 dex_idx,
1079 }
1080}
1081
1082/// JavaType for a field/method descriptor segment.
1083/// Parse a proto's descriptor into a MethodDescriptor straight from the
1084/// tables (no string round-trip).
1085fn parse_proto_desc(
1086 dex: &ddc_dex::DexFile,
1087 proto_idx: u32,
1088) -> jdc_core::types::MethodDescriptor {
1089 let proto = dex.proto(proto_idx);
1090 let args = dex
1091 .proto_params(proto_idx)
1092 .iter()
1093 .map(|&t| desc_type(dex.type_name(t)))
1094 .collect();
1095 jdc_core::types::MethodDescriptor {
1096 args,
1097 ret: desc_type(dex.type_name(proto.return_type_idx)),
1098 }
1099}
1100
1101pub fn desc_type(desc: &str) -> JavaType {
1102 parse_field_descriptor(desc).unwrap_or(JavaType::Object("java/lang/Object".into()))
1103}
1104
1105/// Nesting evidence for `internal`: dalvik annotations first, then the
1106/// `$`-name heuristic against the pool.
1107pub fn find_outer_name(pool: &DexPool, internal: &str) -> Option<String> {
1108 // Annotation refinement only for ALREADY-MATERIALIZED classes: eager
1109 // pools (full decompile) behave exactly as before; lazy pools
1110 // (progressive getclass) fall straight to the `$` chain without
1111 // materializing the world.
1112 if let Some(pc) = pool.get_if_materialized(internal) {
1113 if let Some(enc) = &pc.nesting.enclosing_class {
1114 // The annotation outer must still EXIST: R8 deletes whole
1115 // outer classes and leaves the nestlings as flat `$` files
1116 // (weixin matrix JiffiesMonitorFeature deleted,
1117 // $JiffiesSnapshot survives) — an unresolved annotation
1118 // outer made refs render dotted THROUGH a missing type
1119 // ("是在不可访问的类或接口中定义的"). Fall through to the
1120 // existence-checked `$` walk.
1121 if pool.has_name(enc) {
1122 return Some(enc.clone());
1123 }
1124 }
1125 }
1126 let mut rest = internal;
1127 while let Some(d) = rest.rfind('$') {
1128 // A `$` followed by `/` is a PACKAGE-path dollar (desugar's
1129 // `j$/util/...`), never a nesting boundary: without this the
1130 // whole j$ tree attached to a root-package class `j` as family
1131 // members (deepseek's 62k-line j.java absorbing 549 desugar
1132 // classes, double-emitted against their flat j$/ files).
1133 if rest.as_bytes().get(d + 1) == Some(&b'/') {
1134 rest = &rest[..d];
1135 continue;
1136 }
1137 let cand = &rest[..d];
1138 if pool.has_name(cand) {
1139 return Some(cand.to_string());
1140 }
1141 rest = cand;
1142 }
1143 None
1144}
1145
1146/// True when the `$` tail names a plain member class (not anonymous/local/
1147/// lambda), i.e. the class renders inside its outer's compilation unit.
1148fn clean_member_tail(rest: &str) -> bool {
1149 if rest.starts_with('-') {
1150 return false;
1151 }
1152 // A member tail can never contain `/` — that is a package-path
1153 // fragment from a package-dollar name (`j$/util/X` stripped against
1154 // a root class `j`); defensive backstop for the find_outer_name
1155 // guard.
1156 if rest.contains('/') {
1157 return false;
1158 }
1159 // d8's synthetic outline/lambda/backport classes (`X$$ExternalSynthetic…`)
1160 // are TOP-LEVEL synthetics, not nested members: `$$` is d8's marker, and
1161 // find_outer_name's `$`-chain walk mistakes them for children of `X`.
1162 // Emitting them nested strands every cross-package reference on a bare
1163 // simple name ("找不到符号 变量 ExternalSyntheticBackportWithForwarding0",
1164 // weibo ×467). Targeted at `ExternalSynthetic` only — Kotlin's
1165 // `$$serializer` / `$$delegate` stay nested members.
1166 if rest.starts_with("$ExternalSynthetic") {
1167 return false;
1168 }
1169 let tail = rest.rsplit('$').next().unwrap_or(rest);
1170 if tail.is_empty() || tail.starts_with(|c: char| c.is_ascii_digit()) {
1171 return false;
1172 }
1173 true
1174}
1175
1176/// Classes to emit as their own compilation units: top-level classes plus
1177/// anonymous/local/lambda-shaped ones (clean members render inline).
1178impl DexPool {
1179 /// Arm image retirement (full-decompile driver only): count one
1180 /// pending class per image by pool ownership.
1181 /// Snapshot every synthetic-static accessor body (the exact predicate
1182 /// `inline_accessors` uses) as RAW code_item bytes, so cross-image
1183 /// accessor reads during worker runs survive image retirement.
1184 /// Call AFTER materialization and BEFORE `arm_retirement`; idempotent
1185 /// (first fill wins). Raw memcpy, no decode — accessors are numerous
1186 /// (weixin ~100k) but only a fraction are ever inlined, so decoding
1187 /// stays lazy at the use site. The slice bound covers the fixed
1188 /// header + insns + try items with generous handler slack (accessor
1189 /// tries are vanishingly rare; the snapshot is a deterministic
1190 /// function of the image either way).
1191 pub fn snapshot_accessor_code(&self) {
1192 let mut snap: AccessorSnapshots = HashMap::default();
1193 let mut bytes_total = 0usize;
1194 for name in &self.order {
1195 let Some(pc) = self.get_if_materialized(name) else {
1196 continue;
1197 };
1198 for m in pc.direct_methods.iter().chain(pc.virtual_methods.iter()) {
1199 if m.code_off == 0
1200 || !m.is_static()
1201 || m.access & crate::access::ACC_SYNTHETIC == 0
1202 {
1203 continue;
1204 }
1205 if snap.contains_key(&(m.dex_idx, m.code_off)) {
1206 continue;
1207 }
1208 let Some(dex) = self.dex(m.dex_idx) else {
1209 continue;
1210 };
1211 let raw = dex.raw();
1212 let off = m.code_off as usize;
1213 if off + 16 > raw.len() {
1214 continue;
1215 }
1216 let u2 = |o: usize| u16::from_le_bytes([raw[o], raw[o + 1]]) as usize;
1217 let u4 = |o: usize| {
1218 u32::from_le_bytes([raw[o], raw[o + 1], raw[o + 2], raw[o + 3]]) as usize
1219 };
1220 let tries = u2(off + 6);
1221 let insns = u4(off + 12);
1222 let mut end = off + 16 + 2 * insns;
1223 if tries > 0 {
1224 // pad to 4-align + try_item[tries] + handler blobs
1225 // (uleb-coded; slack covers realistic accessor shapes).
1226 end = end.next_multiple_of(4) + 8 * tries + 256 + 64 * tries;
1227 }
1228 let end = end.min(raw.len());
1229 let slice: std::sync::Arc<[u8]> = raw[off..end].into();
1230 bytes_total += slice.len();
1231 snap.insert((m.dex_idx, m.code_off), slice);
1232 }
1233 }
1234 if std::env::var("DDC_WALL").is_ok() {
1235 eprintln!(
1236 "[wall] accessor snapshot: {} entries, {} KB",
1237 snap.len(),
1238 bytes_total / 1024
1239 );
1240 }
1241 let _ = self.accessor_code.set(snap);
1242 }
1243
1244 /// Snapshotted accessor code_item bytes, when the snapshot was taken.
1245 /// Parse with `CodeItem::parse(&bytes, 0)`.
1246 pub fn accessor_code(&self, dex_idx: usize, code_off: u32) -> Option<std::sync::Arc<[u8]>> {
1247 self.accessor_code.get()?.get(&(dex_idx, code_off)).cloned()
1248 }
1249
1250 pub fn arm_retirement(&self) {
1251 self.retire_armed
1252 .store(true, std::sync::atomic::Ordering::Release);
1253 let mut counts = self.retire_counts.lock().unwrap();
1254 for c in counts.iter_mut() {
1255 *c = 0;
1256 }
1257 for entry in self.classes.values() {
1258 let di = match entry {
1259 ClassEntry::Eager(pc) => pc.dex_idx,
1260 ClassEntry::Lazy { at, .. } => at.0,
1261 ClassEntry::NameOnly => continue,
1262 };
1263 if let Some(c) = counts.get_mut(di) {
1264 *c += 1;
1265 }
1266 }
1267 }
1268
1269 /// Report one class finished; Some(image) when it was the image's
1270 /// last class (driver batches these into release_images).
1271 pub fn report_class_done(&self, class_name: &str) -> Option<usize> {
1272 if !self.retire_armed.load(std::sync::atomic::Ordering::Acquire) {
1273 return None;
1274 }
1275 let entry = self.classes.get(class_name)?;
1276 let di = match entry {
1277 ClassEntry::Eager(pc) => pc.dex_idx,
1278 ClassEntry::Lazy { at, .. } => at.0,
1279 ClassEntry::NameOnly => return None,
1280 };
1281 let mut counts = self.retire_counts.lock().unwrap();
1282 match counts.get_mut(di) {
1283 Some(c) => {
1284 *c = c.saturating_sub(1);
1285 if *c == 0 {
1286 Some(di)
1287 } else {
1288 None
1289 }
1290 }
1291 None => None,
1292 }
1293 }
1294
1295 /// Release the images' inflated bytes. Safe: the driver calls this
1296 /// only when every class of each image is emitted or failed; new
1297 /// `dex()` snapshots still work (tables valid, code accessors empty).
1298 pub fn release_images(&self, indexes: &[usize]) {
1299 let mut counts = self.retire_counts.lock().unwrap();
1300 for &i in indexes {
1301 if let Some(dex) = self.dexes.get(i) {
1302 // Mark the shared image: code accessors go empty
1303 // immediately; the bytes drop when the last snapshot
1304 // drops (workers hold snapshots only mid-class).
1305 dex.mark_released();
1306 }
1307 if let Some(c) = counts.get_mut(i) {
1308 *c = u64::MAX; // released marker
1309 }
1310 }
1311 }
1312}
1313
1314pub fn top_level_classes(pool: &DexPool) -> Vec<String> {
1315 // Progressive (lazy) pools register 100k+ names but materialize only
1316 // the target family. `outer_of` would build the global outer map and
1317 // `get(outer)` would materialize every outer class — a single-class
1318 // query paid ~450MB and hundreds of ms for a boolean. The structural
1319 // decision needs neither: `find_outer_name` already resolves the
1320 // `$`-chain against registered names, and `has_name` proves existence
1321 // without materializing.
1322 let lazy = !pool_majority_materialized(pool);
1323 pool.class_names()
1324 .filter(|name| {
1325 let outer = if lazy {
1326 find_outer_name(pool, name)
1327 } else {
1328 pool.outer_of(name).map(str::to_string)
1329 };
1330 let Some(outer) = outer else {
1331 return true;
1332 };
1333 if !pool.has_name(&outer) {
1334 return true;
1335 }
1336 // The outer can come from an ANNOTATION (EnclosingClass)
1337 // with no naming relationship to this class — obfuscated
1338 // apps pair a 1-char name with a long enclosing descriptor
1339 // (weixin), which made the old blind slice panic. Only a
1340 // real `outer$tail` prefix yields a member tail; anything
1341 // else is a standalone unit.
1342 let rest = name
1343 .strip_prefix(outer.as_str())
1344 .and_then(|t| t.strip_prefix('$'))
1345 .unwrap_or("");
1346 !clean_member_tail(rest)
1347 })
1348 .map(|n| n.to_string())
1349 .collect()
1350}
1351
1352/// Sanitized form of one internal-name segment — the shared mapping the
1353/// declaration site (`print_class_name` → `sanitize_ref`), the CLASS
1354/// lookup and the CLI writer must all agree on. Exported so the writer
1355/// cannot drift from the declaration (see `sanitize_fq`).
1356pub fn sanitize_seg(seg: &str) -> String {
1357 classdec::sanitize_fq(seg)
1358}
1359
1360/// Sanitized internal PATH form of a class name: the exact on-disk
1361/// identity `source_path` writes and the declaration claims.
1362pub fn sanitize_internal(internal: &str) -> String {
1363 internal
1364 .split('/')
1365 .map(classdec::sanitize_fq)
1366 .collect::<Vec<_>>()
1367 .join("/")
1368}
1369
1370/// Deterministic case-collision renames over the FILE-emission set:
1371/// classes whose internal names differ only in letter case cannot share
1372/// one case-insensitive directory; the first (sorted) member of each
1373/// group keeps its name, the others gain `_2`, `_3`, … on the simple
1374/// segment. The map carries identity entries for unrenamed file-level
1375/// classes (they anchor nested prefix walks in apply_class_rename).
1376pub fn case_rename_map(pool: &DexPool) -> HashMap<String, String> {
1377 use std::collections::HashMap;
1378 let mut groups: HashMap<String, Vec<String>> = HashMap::default();
1379 for t in top_level_classes(pool) {
1380 groups.entry(t.to_lowercase()).or_default().push(t);
1381 }
1382 // No `folds` clone: the fold check consults `groups` itself (the old
1383 // full HashSet<String> of every lowercase key was a ~25-30MB pure
1384 // duplicate on the large corpora). Sort in place first, then an
1385 // IMMUTABLE walk keeps the keys queryable for the fold lookup.
1386 for v in groups.values_mut() {
1387 v.sort();
1388 }
1389 let mut map = HashMap::default();
1390 for members in groups.values() {
1391 for (i, m) in members.iter().enumerate() {
1392 if i == 0 {
1393 // Identity anchors are only needed for NESTED names: a
1394 // top-level name has no `$` prefix for apply_class_rename
1395 // to compose, so storing `m -> m` there is pure overhead
1396 // (118k entries on lark).
1397 if m.contains('$') {
1398 map.insert(m.clone(), m.clone());
1399 }
1400 continue;
1401 }
1402 // Suffix the simple segment; keep incrementing if the fold
1403 // of the suffixed name is already taken by a real class.
1404 let cut = m.rfind('/').map(|x| x + 1).unwrap_or(0);
1405 let mut n = i + 1;
1406 loop {
1407 let cand = format!("{}{}_{}", &m[..cut], &m[cut..], n);
1408 let fold = cand.to_lowercase();
1409 if !groups.contains_key(&fold) {
1410 map.insert(m.clone(), cand);
1411 break;
1412 }
1413 n += 1;
1414 }
1415 }
1416 }
1417 map
1418}
1419
1420/// Display-level renames for two nested-class shapes that render
1421/// uncompilable or inconsistent output:
1422///
1423/// 1. A member class whose simple tail equals its rendered parent's
1424/// tail (`x0$a$a` inside `x0$a`): javac rejects a member class with
1425/// the same simple name as its immediately enclosing class ("已在类
1426/// x0中定义了类 x0.a") — a shape Kotlin lambda families hit by the
1427/// thousand (lark alone: 22k). Bump the tail: `a` → `a2`, `a3`, …
1428/// 2. An orphaned intermediate (`x0$a$b` with `x0$a` absent from the
1429/// pool): the member DECLARES as `b` inside `x0` but references
1430/// printed `x0.a$b` — a type that exists nowhere. Flatten the
1431/// display name to `x0$b` so every site agrees.
1432///
1433/// Both are pure display renames keyed by internal name; declarations,
1434/// ctor names, file names and every type reference funnel through
1435/// apply_class_rename, so one map keeps all sites consistent. Children
1436/// Display-level renames for nested member classes that render
1437/// uncompilable or inconsistent output:
1438///
1439/// 1. A member whose simple tail equals ANY enclosing class's simple
1440/// name in its nesting chain. javac rejects more than the immediate
1441/// parent (`class a { static class a2 { static class a {} } }` —
1442/// member `a` two levels under `a` — is "已在类 X中定义了类 X.a");
1443/// Kotlin lambda families (`x0$a$a` inside `x0$a`) hit it by the
1444/// thousand (lark alone: 22k). Bump the tail until it clears every
1445/// ancestor name: `a` → `a3`.
1446/// 2. An orphaned intermediate (`x0$a$b` with `x0$a` absent): the
1447/// member DECLARES as `b` inside `x0` but references printed
1448/// `x0.a$b` — a type that exists nowhere. Flatten the display to
1449/// `x0$b` so every site agrees.
1450///
1451/// Candidates build on the parent's DISPLAY name (ancestors are
1452/// renamed first — ascending `$`-depth order), so declarations,
1453/// constructor names, file names and every type reference stay
1454/// consistent through apply_class_rename. All are pure display
1455/// renames keyed by internal name.
1456/// In-progress-map display composition, mirroring
1457/// `rename::apply_class_rename`'s ancestor walk (the registry is not
1458/// installed yet during map construction, so mint rules must compose
1459/// through renamed ancestors themselves). A nested name whose DIRECT
1460/// parent has no entry still inherits every renamed higher prefix:
1461/// grandparent `yN5` -> `yN5_3` must turn minted `yN5$a$a2` into
1462/// `yN5_3$a$a2` (xingye case-quads: the raw-prefix entry short-
1463/// circuited the render walk — 程序包ddcroot.yN5.a不存在 ×134-family).
1464fn display_of(name: &str, map: &HashMap<String, String>) -> String {
1465 // EXACT hit first — the walk truncates the full name before ever
1466 // consulting it, and an identity entry on an ancestor prefix
1467 // (case_rename_map seeds those for every fold-group first member)
1468 // would short-circuit to the raw name (lark aa2/d$c17 minted, then
1469 // aa2/d$c$a composed against the identity of aa2/d — the child
1470 // stranded on the dead `c` prefix, 找不到符号 类 a ×1,874).
1471 if let Some(d) = map.get(name) {
1472 return d.clone();
1473 }
1474 let mut s = name;
1475 while let Some(i) = s.rfind('$') {
1476 s = &s[..i];
1477 match map.get(s) {
1478 Some(d) if d.as_str() != s => return format!("{d}{}", &name[s.len()..]),
1479 Some(_) => return name.to_string(),
1480 None => continue,
1481 }
1482 }
1483 map.get(name).cloned().unwrap_or_else(|| name.to_string())
1484}
1485
1486fn nested_collision_renames(
1487 pool: &DexPool,
1488 map: &mut HashMap<String, String>,
1489 fam_segs: &SegMap,
1490) {
1491 // Ancestors before descendants: a parent's rename must be in the
1492 // map when its children compute their display chains.
1493 let mut names: Vec<&String> = pool.order.iter().collect();
1494 names.sort_by_key(|n| n.matches('$').count());
1495 let mut assigned: jdc_core::FxHashSet<String> = jdc_core::FxHashSet::default();
1496 // Top-level package segments. A nested member type DISPLAYED as `a2`
1497 // shadows package `a2` for every qualified `a2.x` reference in its
1498 // file (member types outrank package names in class scope), so a
1499 // renamed tail must not become a package shadow — round 47's failure
1500 // mode at nested level: candidate `a2` (for j2/g$a) broke the SAME
1501 // file's `extends a2.a` ("找不到符号 类 a" + a 9450-error Object
1502 // cascade as every dependent of the corrupt head failed too).
1503 let mut pkg_segments: jdc_core::FxHashSet<String> = jdc_core::FxHashSet::default();
1504 for n in &pool.order {
1505 if let Some((seg, _)) = n.split_once('/') {
1506 pkg_segments.insert(seg.to_string());
1507 }
1508 }
1509 // LOCAL-OK gate (round-59 Design B): a field-obscuring rename may
1510 // only fire when every class referencing the candidate lives inside
1511 // the root file's family, so the rename perturbs exactly ONE rendered
1512 // file. Ungated (Design A) the rule netted reqable −1653 / lark −5221
1513 // but weibo +36916: renames there flipped ambiguous type-vs-package
1514 // resolutions across thousands of pre-existing conflict families and
1515 // javac's error recovery turned "missing supertype, body suppressed"
1516 // files into full Object cascades. Lazy (progressive-browse) pools
1517 // skip the gate AND the rule — the multi-second image scan must not
1518 // stall single-class queries; the full-decompile pipeline always
1519 // materializes everything before installing renames.
1520 // Round-59 Design B stands: ungated field-clash renames were
1521 // retried AFTER the package-obscuring rules landed (weibo gate,
1522 // DDC_FC_UNGATED A/B) and still exploded to the 500k error cap —
1523 // fifth falsification of the ungated rename route.
1524 let local_ok: jdc_core::FxHashSet<String> = if pool_majority_materialized(pool) {
1525 refscan::local_ok(&pool.dexes, &field_clash_cands(pool, map))
1526 } else {
1527 jdc_core::FxHashSet::default()
1528 };
1529 let mut taken_sibling_displays: jdc_core::FxHashSet<String> =
1530 map.values().cloned().collect();
1531 // Stage-2 queue: nested tails equal to a top-level package segment
1532 // whose family showed NO descriptor-level trigger — the victim ref
1533 // may live in method bodies only (weixin e32/a2's `a.b.c()` bound
1534 // to its own nested `class a`, 位置: 类 a ×169).
1535 let mut body_cands: Vec<(String, String, String)> = Vec::new();
1536 for name in names {
1537 if !name.contains('$') || map.contains_key(name) {
1538 continue;
1539 }
1540 let Some(parent) = find_outer_name(pool, name) else {
1541 continue;
1542 };
1543 // Only a REAL `parent$rest` name renders as an inline member;
1544 // anonymous/local/lambda tails keep their flat own-file names.
1545 let Some(rest) = name
1546 .strip_prefix(parent.as_str())
1547 .and_then(|t| t.strip_prefix('$'))
1548 else {
1549 continue;
1550 };
1551 if !clean_member_tail(rest) {
1552 continue;
1553 }
1554 // Display simple names of every enclosing level, bottom-up; `cur`
1555 // ends at the ROOT (the top-level class whose file renders this
1556 // nested type — the scope the renamed tail lives in).
1557 let mut chain: Vec<String> = Vec::new();
1558 let mut cur = parent.clone();
1559 loop {
1560 let disp = map.get(&cur).cloned().unwrap_or_else(|| cur.clone());
1561 let simple = disp.rsplit('/').next().unwrap_or(&disp);
1562 chain.push(simple.rsplit('$').next().unwrap_or(simple).to_string());
1563 match find_outer_name(pool, &cur) {
1564 Some(o) if o != cur => cur = o,
1565 _ => break,
1566 }
1567 }
1568 let root: &str = &cur;
1569 let disp_parent = display_of(&parent, map);
1570 let tail = rest.rsplit('$').next().unwrap_or(rest);
1571 let orphan = rest.contains('$');
1572 // Field obscuring (JLS 6.4.2): a nested type whose simple name
1573 // equals an enclosing-class field is OBSCURED by it in a
1574 // `<Parent>.<tail>` qualified reference — `io/flutter/view/g$i`
1575 // (a `static enum i`) renders `g.i.t`, but g also declares an
1576 // instance field `i`, so javac resolves `g.i` to the field:
1577 // "无法从静态上下文中引用非静态 变量 i" (975 of 1140 reqable
1578 // static-context errors). The fix renames the NESTED TYPE (the
1579 // registry carries every render path: shorten, inner_simple —
1580 // whose bypass was the round-55 blowup, fixed in jdc-core —
1581 // classdec declarations), gated to LOCAL-OK candidates by the
1582 // refscan above. PoC: in-file rename of flutter g's `enum i`
1583 // took the class from 2452 errors to 1.
1584 let field_clash = local_ok.contains(name.as_str());
1585 // Package shadow (JLS 6.4.2 at nested level): a nested type
1586 // DISPLAYED with a simple name equal to the first segment of a
1587 // package its FAMILY references binds every `s.x` qualified ref
1588 // in the root file to the member type — the family's own outer
1589 // rename can EXPOSE this (e/b → e/b2 un-collided the nested `b`,
1590 // which then hijacked `b.b2` — 循环继承). The round-47 candidate
1591 // guard only kept RENAMED tails off package segments; pre-
1592 // existing shadows need the rename too, family-gated: fire only
1593 // when this family's own descriptors reference the segment.
1594 // Orphans INCLUDED: an `Outer$$x` orphan normalizes to the
1595 // nested display `x` (k=1 invisible rename), and a clean tail
1596 // renders BARE in the family file — same package shadow as a
1597 // true nested (weixin li/u0$$r → `class r` hijacked `r.a` for
1598 // root package r).
1599 let pkg_shadow = fam_segs
1600 .get(root)
1601 .is_some_and(|segs| segs.contains(tail));
1602 let clash = chain.iter().any(|c| c == tail) || field_clash || pkg_shadow;
1603 // Stage-2 queue: NO descriptor trigger, but the tail equals a
1604 // top-level package segment — the victim ref may be body-level
1605 // only. ORPHANS included: their pass-1 k=1 mint is the invisible
1606 // normalization (`a$$a` → display `a`), which leaves the package
1607 // shadow fully alive (weixin e32/a$$a hijacking `a.b.c()` ×169).
1608 if !clash && pkg_segments.contains(tail) {
1609 body_cands.push((name.clone(), root.to_string(), tail.to_string()));
1610 }
1611 if !orphan && !clash {
1612 continue;
1613 }
1614 // The enclosing class's field names (sanitized like the renderer)
1615 // — only needed once a rename actually fires, to keep the new
1616 // tail from being obscured again. Computed lazily so progressive
1617 // pools don't materialize parents for rules that don't rename.
1618 let mut enc_fields: jdc_core::FxHashSet<String> = jdc_core::FxHashSet::default();
1619 if field_clash || pkg_shadow {
1620 if let Some(pc) = pool.get(&parent) {
1621 for f in pc.static_fields.iter().chain(pc.instance_fields.iter()) {
1622 enc_fields.insert(crate::classdec::java_ident(&f.name).into_owned());
1623 }
1624 }
1625 }
1626 let mut k = 0u32;
1627 loop {
1628 k += 1;
1629 let cand_tail = if k == 1 {
1630 tail.to_string()
1631 } else {
1632 format!("{tail}{k}")
1633 };
1634 // The candidate must clear every ancestor display name (javac
1635 // checks the whole chain) AND every enclosing field name (else
1636 // the renamed type is just obscured again). Sibling nested
1637 // tails are covered by the pool.has_name check below.
1638 if chain.contains(&cand_tail) || enc_fields.contains(&cand_tail) {
1639 continue;
1640 }
1641 // Package shadow: the renamed tail becomes a member type of
1642 // the root file's class, outranking any package of the same
1643 // name for qualified refs in that file. Same-package top-level
1644 // classes likewise outrank-then-break in-file simple refs.
1645 // BOTH checks apply to field-clash renames only: for the
1646 // orphan rule the k=1 candidate is the natural display name
1647 // (`g0$$r` → `g0$r` renders identically — an invisible rename),
1648 // and rejecting it forced weixin's whole `g0$$X` Runnable
1649 // family to visible `X17` renames, breaking every lucky
1650 // `g0.r`-style resolution corpus-wide (+2911 `变量 r` — the
1651 // orphan rule is ungated by design, predating LOCAL-OK).
1652 if field_clash || pkg_shadow {
1653 if pkg_segments.contains(&cand_tail) {
1654 continue;
1655 }
1656 if pkg_shadow
1657 && fam_segs
1658 .get(root)
1659 .is_some_and(|segs| segs.contains(cand_tail.as_str()))
1660 {
1661 continue;
1662 }
1663 if let Some((root_pkg, _)) = root.rsplit_once('/') {
1664 let sib = format!("{root_pkg}/{cand_tail}");
1665 if pool.has_name(&sib) || taken_sibling_displays.contains(&sib)
1666 {
1667 // A top-level sibling exists — or an EARLIER rule
1668 // minted that display (obscuring_class_renames
1669 // moved ssosdk/b → ssosdk/b2 while this loop was
1670 // about to mint nested $b → b2: two `b2`s in one
1671 // scope, member type outranks the package sibling
1672 // and hijacked every bare `b2.b(..)` call).
1673 continue;
1674 }
1675 taken_sibling_displays.insert(sib);
1676 }
1677 }
1678 let cand = format!("{disp_parent}${cand_tail}");
1679 if pool.has_name(&cand) || assigned.contains(&cand) {
1680 continue;
1681 }
1682 assigned.insert(cand.clone());
1683 map.insert(name.clone(), cand);
1684 break;
1685 }
1686 }
1687 // Stage 2 (body-gated package shadow): one body-level image scan
1688 // over the queued families (family_ref_segments — the field-
1689 // deshadow mechanism; instruction-level, cross-package refs), then
1690 // mint for the families whose members actually reference the
1691 // segment. Mirrors obscuring_class_renames' two-stage design at
1692 // family level. Lazy pools skip (same gate as stage 1's scan).
1693 let mut body_renamed = 0usize;
1694 if !body_cands.is_empty() && pool_majority_materialized(pool) {
1695 let roots: jdc_core::FxHashSet<&str> =
1696 body_cands.iter().map(|(_, r, _)| r.as_str()).collect();
1697 let mut members: jdc_core::FxHashSet<String> = jdc_core::FxHashSet::default();
1698 for n in &pool.order {
1699 let r = match n.find('$') {
1700 Some(i) => &n[..i],
1701 None => n.as_str(),
1702 };
1703 if roots.contains(r) {
1704 members.insert(n.clone());
1705 }
1706 }
1707 let t0 = std::time::Instant::now();
1708 let (by_referrer, _) = refscan::family_ref_segments(&pool.dexes, &members);
1709 // Referrer-keyed → family-root-keyed: every member renders in
1710 // the root file's scope.
1711 let mut root_segs: jdc_core::FxHashMap<&str, jdc_core::FxHashSet<&str>> =
1712 jdc_core::FxHashMap::default();
1713 for (referrer, segs) in &by_referrer {
1714 let r = referrer.split('$').next().unwrap_or(referrer);
1715 root_segs.entry(r).or_default().extend(segs.iter().map(|s| s.as_str()));
1716 }
1717 if std::env::var("DDC_STATS").is_ok() {
1718 eprintln!(
1719 "[renames] nested body-shadow scan: cands={} families={} scan={:?}",
1720 body_cands.len(),
1721 root_segs.len(),
1722 t0.elapsed()
1723 );
1724 }
1725 for (name, root, tail) in &body_cands {
1726 if !root_segs
1727 .get(root.as_str())
1728 .is_some_and(|s| s.contains(tail.as_str()))
1729 {
1730 continue;
1731 }
1732 if let Some(d) = map.get(name) {
1733 // Pass-1 orphan mints are invisible (display tail ==
1734 // tail) — those get re-minted visibly here. Anything
1735 // already visibly renamed stays as the earlier rule set it.
1736 let cur_tail = d.rsplit(['/', '$']).next().unwrap_or("");
1737 if cur_tail != tail.as_str() {
1738 continue;
1739 }
1740 }
1741 // Fresh display context: stage-1 renames may have moved the
1742 // parent/chain since this candidate was queued.
1743 let Some(parent) = find_outer_name(pool, name) else {
1744 continue;
1745 };
1746 let Some(rest) = name
1747 .strip_prefix(parent.as_str())
1748 .and_then(|t| t.strip_prefix('$'))
1749 else {
1750 continue;
1751 };
1752 if !clean_member_tail(rest) {
1753 continue;
1754 }
1755 let mut chain: Vec<String> = Vec::new();
1756 let mut cur = parent.clone();
1757 loop {
1758 let disp = map.get(&cur).cloned().unwrap_or_else(|| cur.clone());
1759 let simple = disp.rsplit('/').next().unwrap_or(&disp);
1760 chain.push(simple.rsplit('$').next().unwrap_or(simple).to_string());
1761 match find_outer_name(pool, &cur) {
1762 Some(o) if o != cur => cur = o,
1763 _ => break,
1764 }
1765 }
1766 let disp_parent = display_of(&parent, map);
1767 let mut enc_fields: jdc_core::FxHashSet<String> =
1768 jdc_core::FxHashSet::default();
1769 if let Some(pc) = pool.get(&parent) {
1770 for f in pc.static_fields.iter().chain(pc.instance_fields.iter()) {
1771 enc_fields.insert(crate::classdec::java_ident(&f.name).into_owned());
1772 }
1773 }
1774 let mut k = 0u32;
1775 loop {
1776 k += 1;
1777 let cand_tail = if k == 1 {
1778 tail.clone()
1779 } else {
1780 format!("{tail}{k}")
1781 };
1782 if chain.contains(&cand_tail) || enc_fields.contains(&cand_tail) {
1783 continue;
1784 }
1785 // The renamed tail stays a member type of the root
1786 // file — it must not shadow ANOTHER package segment the
1787 // family references (descriptor OR body level) nor a
1788 // same-package top-level sibling.
1789 if pkg_segments.contains(&cand_tail) {
1790 continue;
1791 }
1792 if fam_segs
1793 .get(root.as_str())
1794 .is_some_and(|segs| segs.contains(cand_tail.as_str()))
1795 {
1796 continue;
1797 }
1798 if root_segs
1799 .get(root.as_str())
1800 .is_some_and(|segs| segs.contains(cand_tail.as_str()))
1801 {
1802 continue;
1803 }
1804 if let Some((root_pkg, _)) = root.rsplit_once('/') {
1805 let sib = format!("{root_pkg}/{cand_tail}");
1806 if pool.has_name(&sib) || taken_sibling_displays.contains(&sib) {
1807 continue;
1808 }
1809 taken_sibling_displays.insert(sib);
1810 }
1811 let cand = format!("{disp_parent}${cand_tail}");
1812 if pool.has_name(&cand) || assigned.contains(&cand) {
1813 continue;
1814 }
1815 // Repair pass-1 mints whose display chain ran through
1816 // THIS name's old display: stage 2 renames ancestors
1817 // after descendants were minted (weibo mobileads k$a →
1818 // a5 left pass-1's k$a$b$a → "k$a$b$a2" pointing through
1819 // the dead `a` — 找不到符号 类 a ×48). Ancestor-first
1820 // within each pass, but stage 2 as a whole runs last.
1821 let old_disp = format!("{disp_parent}${tail}");
1822 if old_disp != cand {
1823 for (mk, mv) in map.iter_mut() {
1824 if mk == name {
1825 continue;
1826 }
1827 if let Some(rest) = mv.strip_prefix(old_disp.as_str()) {
1828 if rest.starts_with('$') {
1829 let repaired = format!("{cand}{rest}");
1830 assigned.insert(repaired.clone());
1831 *mv = repaired;
1832 }
1833 }
1834 }
1835 }
1836 assigned.insert(cand.clone());
1837 map.insert(name.clone(), cand);
1838 body_renamed += 1;
1839 break;
1840 }
1841 }
1842 }
1843 if body_renamed > 0 && std::env::var("DDC_STATS").is_ok() {
1844 eprintln!("[renames] nested body-shadow renames: {body_renamed}");
1845 }
1846}
1847
1848/// Field-obscuring rename candidates: nested member types whose simple
1849/// name equals a field of the enclosing class, mapped to the top-level
1850/// class whose file renders them (the LOCAL-OK family root). Mirrors the
1851/// skip conditions of `nested_collision_renames` (same-name entries must
1852/// agree, or the gate proves locality for a rename that never fires — or
1853/// worse, skips one that does).
1854fn field_clash_cands(pool: &DexPool, map: &HashMap<String, String>) -> HashMap<String, String> {
1855 let mut out: HashMap<String, String> = HashMap::default();
1856 for name in &pool.order {
1857 if !name.contains('$') || map.contains_key(name) {
1858 continue;
1859 }
1860 let Some(parent) = find_outer_name(pool, name) else {
1861 continue;
1862 };
1863 let Some(rest) = name
1864 .strip_prefix(parent.as_str())
1865 .and_then(|t| t.strip_prefix('$'))
1866 else {
1867 continue;
1868 };
1869 if !clean_member_tail(rest) {
1870 continue;
1871 }
1872 let tail = rest.rsplit('$').next().unwrap_or(rest);
1873 let Some(pc) = pool.get(&parent) else {
1874 continue;
1875 };
1876 let clash = pc
1877 .static_fields
1878 .iter()
1879 .chain(pc.instance_fields.iter())
1880 .any(|f| crate::classdec::java_ident(&f.name).as_ref() == tail);
1881 if !clash {
1882 continue;
1883 }
1884 let mut root = parent.clone();
1885 loop {
1886 match find_outer_name(pool, &root) {
1887 Some(o) if o != root => root = o,
1888 _ => break,
1889 }
1890 }
1891 out.insert(name.clone(), root);
1892 }
1893 out
1894}
1895
1896/// Whether the pool is (majority-)materialized: full-decompile pipelines
1897/// call `materialize_all` before installing renames, progressive browse
1898/// pools materialize a handful of classes on demand. The refscan gate
1899/// must not run (or stall startup) in the progressive case.
1900fn pool_majority_materialized(pool: &DexPool) -> bool {
1901 let (mut mat, mut lazy) = (0usize, 0usize);
1902 for e in pool.classes.values() {
1903 match e {
1904 ClassEntry::Eager(_) => mat += 1,
1905 ClassEntry::Lazy { pc, .. } => {
1906 if pc.get().is_some() {
1907 mat += 1;
1908 } else {
1909 lazy += 1;
1910 }
1911 }
1912 ClassEntry::NameOnly => lazy += 1,
1913 }
1914 }
1915 lazy == 0 || mat >= lazy
1916}
1917
1918/// Compute and install the registry (call before worker threads spawn).
1919/// CURRENT resident size (MB) — ru_maxrss is monotone (historical peak),
1920/// so purge effects are invisible there; live-vs-retained attribution
1921/// needs the instantaneous figure (macOS mach task_info).
1922#[cfg(target_vendor = "apple")]
1923fn current_rss_mb() -> f64 {
1924 #[repr(C)]
1925 struct TaskBasicInfo {
1926 virtual_size: u64,
1927 resident_size: u64,
1928 resident_size_max: u64,
1929 // time_value_t = {i32 secs, i32 usec} = 8B each.
1930 total_user_time: u64,
1931 total_sys_time: u64,
1932 policy: i32,
1933 suspend_count: i32,
1934 }
1935 extern "C" {
1936 fn mach_task_self() -> u32;
1937 fn task_info(
1938 target_port: u32,
1939 flavor: i32,
1940 task_info_out: *mut TaskBasicInfo,
1941 task_info_out_cnt: *mut u32,
1942 ) -> i32;
1943 }
1944 let mut info = TaskBasicInfo {
1945 virtual_size: 0,
1946 resident_size: 0,
1947 resident_size_max: 0,
1948 total_user_time: 0,
1949 total_sys_time: 0,
1950 policy: 0,
1951 suspend_count: 0,
1952 };
1953 // count is in natural_t (4B) units; a smaller-than-kernel version
1954 // still fills the prefix fields we read.
1955 let mut cnt = (std::mem::size_of::<TaskBasicInfo>() / 4) as u32;
1956 let kr = unsafe { task_info(mach_task_self(), 20, &mut info, &mut cnt) };
1957 if kr == 0 {
1958 info.resident_size as f64 / (1024.0 * 1024.0)
1959 } else {
1960 0.0
1961 }
1962}
1963#[cfg(not(target_vendor = "apple"))]
1964fn current_rss_mb() -> f64 {
1965 0.0
1966}
1967
1968/// ru_maxrss peak so far (MB). macOS reports bytes, Linux KB; Windows
1969/// has no getrusage (the old not(apple) fallback referenced it and
1970/// failed the MSVC link — LNK2019 — in the v0.1.18 CI).
1971#[cfg(unix)]
1972fn peak_rss_mb() -> f64 {
1973 #[repr(C)]
1974 struct Rusage {
1975 pad: [i64; 4],
1976 maxrss: i64,
1977 // Oversized on purpose: the kernel writes sizeof(struct rusage)
1978 // (~144B on macOS) regardless of what we read back.
1979 tail: [i64; 16],
1980 }
1981 extern "C" {
1982 fn getrusage(who: i32, r: *mut Rusage) -> i32;
1983 }
1984 let mut r = Rusage {
1985 pad: [0; 4],
1986 maxrss: 0,
1987 tail: [0; 16],
1988 };
1989 unsafe { getrusage(0, &mut r) };
1990 #[cfg(target_vendor = "apple")]
1991 let scale = 1024.0 * 1024.0;
1992 #[cfg(not(target_vendor = "apple"))]
1993 let scale = 1024.0;
1994 r.maxrss as f64 / scale
1995}
1996#[cfg(not(unix))]
1997fn peak_rss_mb() -> f64 {
1998 0.0
1999}
2000
2001/// DDC_STATS phase attribution of the RSS high-water mark: peak AND
2002/// instantaneous live (ru_maxrss is monotone, so purge effects are
2003/// invisible there).
2004pub(crate) fn stats_rss(phase: &str) {
2005 if std::env::var("DDC_STATS").is_ok() {
2006 eprintln!(
2007 "[rss] {phase}: peak {:.0} MB, live {:.0} MB",
2008 peak_rss_mb(),
2009 current_rss_mb()
2010 );
2011 }
2012}
2013
2014pub fn install_case_renames(pool: &DexPool) {
2015 let dbg = std::env::var("DDC_DBG_REN").is_ok();
2016 let mut tick = std::time::Instant::now();
2017 let mut lap = |tag: &str| {
2018 if dbg {
2019 eprintln!("[ren] {:?} {}", tick.elapsed(), tag);
2020 tick = std::time::Instant::now();
2021 }
2022 };
2023 let mut map = case_rename_map(pool);
2024 lap("case_rename_map");
2025 crate::stats_rss(" case_rename_map");
2026 // Cross-package reference first segments (descriptor level). Lazy
2027 // (progressive-browse) pools skip the scan — materializing every
2028 // class must not stall single-class queries; the obscuring rules
2029 // then simply never fire (browse mode has no whole-program javac).
2030 let (pkg_segs, fam_segs) = if pool_majority_materialized(pool) {
2031 ref_segments(pool)
2032 } else {
2033 (
2034 jdc_core::FxHashMap::default(),
2035 jdc_core::FxHashMap::default(),
2036 )
2037 };
2038 // Package-leaf shadows FIRST: the nested-collision renames compute
2039 // display chains off the (renamed) parent display names — running
2040 // them before the parent rename left both rules minting the same
2041 // display (`a2` twice in weibo's AIDL families).
2042 crate::stats_rss(" ref_segments");
2043 lap("ref_segments");
2044 pkg_leaf_shadow_renames(pool, &mut map);
2045 lap("pkg_leaf_shadow");
2046 class_pkg_collision_renames(pool, &mut map);
2047 lap("class_pkg_collision");
2048 obscuring_class_renames(pool, &mut map, &pkg_segs);
2049 lap("obscuring");
2050 nested_collision_renames(pool, &mut map, &fam_segs);
2051 lap("nested_collision");
2052 // LAST: lossy-sanitize collisions key on the display form every rule
2053 // above has already settled (`l.᩻ܶ` → `l/__.java`, 22,636 classes onto
2054 // 3 paths on bin.mt.plus — the silent-overwrite data loss).
2055 crate::stats_rss(" shadow+collision+obscuring+nested rules");
2056 lap("rules");
2057 lossy_sanitize_renames(pool, &mut map);
2058 // Default-package relocation keys on the settled displays too (it
2059 // PREFIXES them with the synthetic package) — runs after every
2060 // simple-name rule has minted.
2061 root_pkg_relocation(pool, &mut map);
2062 fw_shadow_relocation(pool, &mut map);
2063 crate::stats_rss(" lossy+root-pkg rules");
2064 lap("lossy+root");
2065 jdc_core::rename::set_class_renames(map);
2066 crate::stats_rss("class-renames installed");
2067 lap("set_class_renames");
2068 // Field-rename registry build and the access-widening image scan are
2069 // independent: both only READ the pool, they install into disjoint
2070 // statics (FIELD_RENAMES vs WIDEN), the WIDEN accessors are
2071 // is_some_and-guarded with no reader before render time, and lazy
2072 // materialization is OnceLock-safe. Run them concurrently — the 1.6s
2073 // QQ widening scan fully overlaps the field-rename build.
2074 if pool_majority_materialized(pool) {
2075 std::thread::scope(|s| {
2076 let h = s.spawn(|| crate::classdec::install_access_widening(pool));
2077 jdc_core::rename::set_field_renames(combined_field_renames(pool));
2078 let _ = h.join();
2079 });
2080 crate::stats_rss("field-renames + widening installed");
2081 lap("field+widen");
2082 } else {
2083 jdc_core::rename::set_field_renames(combined_field_renames(pool));
2084 lap("field_renames(lazy)");
2085 }
2086}
2087
2088/// A top-level class `P/s` whose simple name equals the FIRST PACKAGE
2089/// SEGMENT of types that other classes in P reference cross-package.
2090/// Such a reference renders as the FQN `s.rest...`, and javac binds the
2091/// first segment to the nearest type in scope — the same-package class
2092/// `P/s` (JLS 6.4.2): `n91.f.a(x)` inside pc5/bd looks up member `f` OF
2093/// CLASS pc5.n91 (weixin's n0/n91/uc6/vc6 protobuf cluster: ~12k root
2094/// files, ~90k cascade lines; weibo's `a.g.b.a.b` chains root the same
2095/// way — class a beside subpackage a/ inside package a/g/b). No Java
2096/// syntax escapes an obscured first segment (imports can bypass it, but
2097/// single-type imports are blocked exactly where obfuscated leaf names
2098/// collide with same-package siblings), so the class moves: the r66
2099/// rename mechanism at a new trigger. Gated on DESCRIPTOR-level refs of
2100/// the candidate's own package (supers/fields/protos) — body-only refs
2101/// are missed by design; the import layer covers unblocked leaves there.
2102/// Descriptor-level cross-package reference first segments, aggregated
2103/// per PACKAGE (top-level obscurers: every file in P shares the scope)
2104/// and per FAMILY root (nested obscurers: a nested simple name is in
2105/// scope only inside its own root file). Raw `L..;` scans — no JavaType
2106/// parsing. Body-level refs are missed by design (the import layer
2107/// catches unblocked leaves there); descriptors cover supers, field
2108/// types and method protos — the header positions whose breakage starts
2109/// the error-type cascades.
2110/// Package (or family) → first segments of the cross-package types its
2111/// classes reference at descriptor level. Obscuring-rename trigger sets.
2112type SegMap<'a> = jdc_core::FxHashMap<&'a str, jdc_core::FxHashSet<&'a str>>;
2113
2114fn ref_segments(pool: &DexPool) -> (SegMap<'_>, SegMap<'_>) {
2115 fn pkg_of(n: &str) -> &str {
2116 match n.rsplit_once('/') {
2117 Some((p, _)) => p,
2118 None => "",
2119 }
2120 }
2121 fn note<'a>(
2122 pkg_segs: &mut jdc_core::FxHashMap<&'a str, jdc_core::FxHashSet<&'a str>>,
2123 fam_segs: &mut jdc_core::FxHashMap<&'a str, jdc_core::FxHashSet<&'a str>>,
2124 pkg: &'a str,
2125 fam: &'a str,
2126 t: &'a str,
2127 ) {
2128 if !t.contains('/') {
2129 return; // root-package target: unreferenceable from named pkgs
2130 }
2131 let _ = pkg; // same-package refs COUNT: ddc renders headers/field
2132 // types fully qualified (print_class_name), so `a.g.b.a.b`
2133 // inside package a/g/b is obscured by class a/g/b/a just
2134 // like a cross-package FQN (weibo's whole `a`-tree family).
2135 let seg = t.split('/').next().unwrap_or("");
2136 if !seg.is_empty() {
2137 pkg_segs.entry(pkg).or_default().insert(seg);
2138 fam_segs.entry(fam).or_default().insert(seg);
2139 }
2140 }
2141 fn scan<'a>(
2142 pkg_segs: &mut jdc_core::FxHashMap<&'a str, jdc_core::FxHashSet<&'a str>>,
2143 fam_segs: &mut jdc_core::FxHashMap<&'a str, jdc_core::FxHashSet<&'a str>>,
2144 pkg: &'a str,
2145 fam: &'a str,
2146 desc: &'a str,
2147 ) {
2148 let b = desc.as_bytes();
2149 let mut i = 0usize;
2150 while i < b.len() {
2151 if b[i] == b'L' {
2152 if let Some(end) = desc[i + 1..].find(';') {
2153 note(pkg_segs, fam_segs, pkg, fam, &desc[i + 1..i + 1 + end]);
2154 i += end + 2;
2155 continue;
2156 }
2157 }
2158 i += 1;
2159 }
2160 }
2161 let mut pkg_segs: jdc_core::FxHashMap<&str, jdc_core::FxHashSet<&str>> =
2162 jdc_core::FxHashMap::default();
2163 let mut fam_segs: jdc_core::FxHashMap<&str, jdc_core::FxHashSet<&str>> =
2164 jdc_core::FxHashMap::default();
2165 for name in &pool.order {
2166 let Some(c) = pool.get(name) else { continue };
2167 let pkg = pkg_of(name);
2168 let fam = match name.find('$') {
2169 Some(i) => &name[..i],
2170 None => name.as_str(),
2171 };
2172 if let Some(sup) = &c.super_name {
2173 note(&mut pkg_segs, &mut fam_segs, pkg, fam, sup);
2174 }
2175 for sup in c.interfaces.iter() {
2176 note(&mut pkg_segs, &mut fam_segs, pkg, fam, sup);
2177 }
2178 for f in c.static_fields.iter().chain(c.instance_fields.iter()) {
2179 scan(&mut pkg_segs, &mut fam_segs, pkg, fam, &f.desc);
2180 }
2181 for m in c.direct_methods.iter().chain(c.virtual_methods.iter()) {
2182 scan(&mut pkg_segs, &mut fam_segs, pkg, fam, &m.desc);
2183 }
2184 }
2185 (pkg_segs, fam_segs)
2186}
2187
2188/// A top-level class `P/s` whose simple name equals the FIRST PACKAGE
2189/// SEGMENT of types that classes in P reference cross-package. Such a
2190/// reference renders as the FQN `s.rest...`, and javac binds the first
2191/// segment to the nearest type in scope — the same-package class `P/s`
2192/// (JLS 6.4.2, no package fallback — verified): `n91.f.a(x)` inside
2193/// pc5/bd looks up member `f` OF CLASS pc5.n91 (weixin's n0/n91/uc6/vc6
2194/// protobuf cluster: ~12k root files, ~90k cascade lines; weibo's
2195/// `a.g.b.a.b` chains root the same way). No Java syntax escapes an
2196/// obscured first segment, so the class moves: the r66 rename mechanism
2197/// at a new trigger, gated on DESCRIPTOR-level refs of the candidate's
2198/// package (body-only refs go to the import layer). Nested obscurers
2199/// (a nested simple name shadowing a package inside its family file)
2200/// are the companion trigger in nested_collision_renames.
2201fn obscuring_class_renames(
2202 pool: &DexPool,
2203 map: &mut HashMap<String, String>,
2204 pkg_segs: &SegMap,
2205) {
2206 fn pkg_of(n: &str) -> &str {
2207 match n.rsplit_once('/') {
2208 Some((p, _)) => p,
2209 None => "",
2210 }
2211 }
2212 // Root package first segments (segments some class lives under).
2213 let mut root_segs: jdc_core::FxHashSet<&str> = jdc_core::FxHashSet::default();
2214 for n in &pool.order {
2215 if let Some(i) = n.find('/') {
2216 root_segs.insert(&n[..i]);
2217 }
2218 }
2219 let mut cands: Vec<&String> = Vec::new();
2220 for name in &pool.order {
2221 let simple = name.rsplit('/').next().unwrap_or(name);
2222 let pkg = pkg_of(name);
2223 if simple.is_empty() || simple.contains('$') || pkg.is_empty() {
2224 continue;
2225 }
2226 if !root_segs.contains(simple) {
2227 continue;
2228 }
2229 if map.get(name).is_some_and(|v| v != name) {
2230 continue; // an earlier rule moved it
2231 }
2232 // NB: NO "class lives under the s tree" exclusion. weibo's
2233 // a/a/b/c/l/a shadowed the FQN refs `a.a.a.a2` that its own
2234 // package files render for cross-subpackage targets (the
2235 // starts_with("a/") skip left the WHOLE a-tree unrenamed —
2236 // 找不到符号 类 a chains). The two-stage refsegs gates below
2237 // require an actual cross-package FQN ref for the trigger.
2238 cands.push(name);
2239 }
2240 let ncands = cands.len();
2241 if std::env::var("DDC_STATS").is_ok() {
2242 eprintln!(
2243 "[renames] obscuring candidates={ncands} pkgs-with-segs={}",
2244 pkg_segs.len()
2245 );
2246 }
2247 if cands.is_empty() {
2248 return;
2249 }
2250 // Fast collision oracle: sorted names + prefix range scan (the r66
2251 // candidate loop's `order.iter().any(starts_with)` is O(n) per
2252 // candidate — this rule can mint tens of thousands of renames).
2253 let mut sorted: Vec<&String> = pool.order.iter().collect();
2254 sorted.sort_unstable();
2255 let prefix_free = |cand: &str| -> bool {
2256 // No class named cand, nothing nested under cand$..., no package cand/...
2257 let idx = sorted.partition_point(|n| n.as_str() < cand);
2258 !sorted[idx..].first().is_some_and(|n| n.starts_with(cand))
2259 };
2260 let pkg_simples = pool.package_simples();
2261 let mut taken_displays: jdc_core::FxHashSet<String> = map.values().cloned().collect();
2262 let mut renamed = 0usize;
2263 // Stage 1: descriptor-level hits rename now; the rest queue for the
2264 // BODY-level image scan (weixin pc5's `invoke-static Ln91/f;.a` —
2265 // 7.8k+6.6k error lines the descriptor gate cannot see).
2266 let mut stage2: Vec<&String> = Vec::new();
2267 for name in cands {
2268 let pkg = pkg_of(name);
2269 let simple = name.rsplit('/').next().unwrap_or(name);
2270 if !pkg_segs.get(pkg).is_some_and(|segs| segs.contains(simple)) {
2271 stage2.push(name);
2272 continue;
2273 }
2274 let siblings = pkg_simples.get(pkg);
2275 let mut k = 1u32;
2276 loop {
2277 k += 1;
2278 let new_simple = format!("{simple}{k}");
2279 let cand = if pkg.is_empty() {
2280 new_simple.clone()
2281 } else {
2282 format!("{pkg}/{new_simple}")
2283 };
2284 // Exact/prefix pool clash, another rule's display target, the
2285 // package's own referenced segments (the new name must not
2286 // obscure ANOTHER package the same files reference), or a
2287 // case-insensitive clash with a same-package sibling
2288 // (case-insensitive filesystems fold the files).
2289 let clash = !prefix_free(&cand)
2290 || taken_displays.contains(&cand)
2291 || pkg_segs
2292 .get(pkg)
2293 .is_some_and(|segs| segs.contains(new_simple.as_str()))
2294 || siblings.is_some_and(|s| {
2295 s.iter()
2296 .any(|p| p.eq_ignore_ascii_case(&new_simple) && *p != simple)
2297 });
2298 if !clash {
2299 taken_displays.insert(cand.clone());
2300 map.insert(name.clone(), cand);
2301 renamed += 1;
2302 break;
2303 }
2304 }
2305 }
2306 let mut body_renamed = 0usize;
2307 if !stage2.is_empty() && pool_majority_materialized(pool) {
2308 let stage2_pkgs: jdc_core::FxHashSet<String> =
2309 stage2.iter().map(|n| pkg_of(n).to_string()).collect();
2310 let body_segs = crate::refscan::body_ref_segments(&pool.dexes, &stage2_pkgs);
2311 for name in stage2 {
2312 let pkg = pkg_of(name);
2313 let simple = name.rsplit('/').next().unwrap_or(name);
2314 if !body_segs
2315 .get(pkg)
2316 .is_some_and(|segs| segs.contains(simple))
2317 {
2318 continue;
2319 }
2320 let siblings = pkg_simples.get(pkg);
2321 let mut k = 1u32;
2322 loop {
2323 k += 1;
2324 let new_simple = format!("{simple}{k}");
2325 let cand = format!("{pkg}/{new_simple}");
2326 let clash = !prefix_free(&cand)
2327 || taken_displays.contains(&cand)
2328 || pkg_segs
2329 .get(pkg)
2330 .is_some_and(|segs| segs.contains(new_simple.as_str()))
2331 || body_segs
2332 .get(pkg)
2333 .is_some_and(|segs| segs.contains(new_simple.as_str()))
2334 || siblings.is_some_and(|s| {
2335 s.iter()
2336 .any(|p| p.eq_ignore_ascii_case(&new_simple) && *p != simple)
2337 });
2338 if !clash {
2339 taken_displays.insert(cand.clone());
2340 map.insert(name.clone(), cand);
2341 renamed += 1;
2342 body_renamed += 1;
2343 break;
2344 }
2345 }
2346 }
2347 }
2348 if std::env::var("DDC_STATS").is_ok() {
2349 eprintln!(
2350 "[renames] obscuring-class renames: {renamed} (body-gated {body_renamed}; pkgs-with-segs={}, cands={ncands})",
2351 pkg_segs.len()
2352 );
2353 }
2354}
2355
2356/// Class-level counterpart of the non-ASCII rule already applied to
2357/// MEMBERS by `member_collision_renames`, kept as the SAFETY NET behind
2358/// the injective `sanitize_fq`.
2359///
2360/// `sanitize_fq` escapes non-identifier characters as `_u<hex>`, so two
2361/// distinct classes normally reach two distinct file names. It cannot be
2362/// injective against *lookalikes*, because ident-safe ASCII must pass
2363/// through unchanged: a literal class named `_u1a7b` collides with the
2364/// class whose single character is U+1A7B, and the pre-existing keyword
2365/// repair makes `_do` collide with `do`. Without this pass those pairs
2366/// repeat the original disaster — the writer opens with `create_new`,
2367/// takes EEXIST, unlinks and rewrites, so one class is destroyed with
2368/// exit code 0 (on bin.mt.plus the lossy predecessor of this rule hid
2369/// 22,633 of 30,768 classes exactly this way).
2370///
2371/// Groups are keyed by the SANITIZED path — the exact string both the
2372/// declaration site (`sanitize_ref`) and the writer (`source_path`)
2373/// produce. The first (sorted) member of each group keeps its name; the
2374/// rest become `<sanitized>_2`, `<sanitized>_3`, … Pure ASCII, so a
2375/// minted name is a fixed point of the sanitizer: it cannot silently
2376/// join a second collision group, and the declaration, every reference
2377/// and the file name agree by construction through `apply_class_rename`.
2378///
2379/// Costs nothing on a healthy pool: one pass over the emission set, and
2380/// it returns before touching the registry when every key is unique.
2381/// Runs LAST in `install_case_renames`: it keys on the display form the
2382/// earlier rules have already settled.
2383fn lossy_sanitize_renames(pool: &DexPool, map: &mut HashMap<String, String>) {
2384 let names = top_level_classes(pool);
2385 // Current display of an emission-set name. `case_rename_map` has
2386 // already inserted an entry for every one of these (identity or
2387 // renamed) and an exact registry hit short-circuits
2388 // `apply_class_rename`, so `map` already holds the final display —
2389 // no `$`-prefix walk is needed here.
2390 fn key_of(map: &HashMap<String, String>, n: &str) -> String {
2391 let disp = map.get(n).map(|s| s.as_str()).unwrap_or(n);
2392 sanitize_internal(disp)
2393 }
2394 let mut counts: HashMap<String, u32> = HashMap::default();
2395 for n in &names {
2396 *counts.entry(key_of(map, n)).or_insert(0) += 1;
2397 }
2398 if counts.values().all(|c| *c == 1) {
2399 return;
2400 }
2401 // Every sanitized path in the POOL is off-limits for a minted name,
2402 // not just the emission set: an inlined nested class still owns the
2403 // display it renders under inside its outer's file.
2404 let mut taken: jdc_core::FxHashSet<String> = pool
2405 .order
2406 .iter()
2407 .map(|n| sanitize_internal(map.get(n).map(|s| s.as_str()).unwrap_or(n)))
2408 .collect();
2409 let mut sorted: Vec<&String> = names.iter().collect();
2410 sorted.sort();
2411 let mut seen: jdc_core::FxHashSet<String> = jdc_core::FxHashSet::default();
2412 let mut renamed = 0usize;
2413 for n in sorted {
2414 let key = key_of(map, n);
2415 if counts.get(&key).copied().unwrap_or(0) < 2 {
2416 continue;
2417 }
2418 if seen.insert(key.clone()) {
2419 continue; // first (sorted) member of the group keeps its name
2420 }
2421 let (pkg, simple) = match key.rsplit_once('/') {
2422 Some((p, s)) => (p, s),
2423 None => ("", key.as_str()),
2424 };
2425 let mut i = 1u32;
2426 loop {
2427 i += 1;
2428 let cand = if pkg.is_empty() {
2429 format!("{simple}_{i}")
2430 } else {
2431 format!("{pkg}/{simple}_{i}")
2432 };
2433 if !taken.insert(cand.clone()) {
2434 continue; // already owned by a real class or another mint
2435 }
2436 // The minted display IS the sanitized name, so
2437 // sanitize_internal(cand) == cand and the writer's file name
2438 // matches the declaration it writes into it.
2439 map.insert(n.clone(), cand);
2440 renamed += 1;
2441 break;
2442 }
2443 }
2444 if std::env::var("DDC_STATS").is_ok() {
2445 eprintln!("[renames] lossy-sanitize collisions renamed={renamed}");
2446 }
2447}
2448
2449/// A class named `P/s` while OTHER classes live under `P/s/` — the dex
2450/// namespace is flat, so obfuscators freely mint a class whose simple
2451/// name equals a subpackage of its own package (`ptr.a` +
2452/// `ptr.a.a`, weibo rsplay/ptr families). That namespace is
2453/// unrepresentable in Java: the declaration itself is
2454/// "类 a与带有相同名称的程序包冲突" and every qualified reference
2455/// through the segment becomes unresolvable (the 53.8k weibo
2456/// missing-class errors root here, with Object/getClass cascades on
2457/// top). Rename the class — the registry carries declarations and
2458/// references — with a candidate that cannot recreate the shape (not a
2459/// class, not a package, not a nested-class prefix, not another rule's
2460/// target).
2461fn class_pkg_collision_renames(pool: &DexPool, map: &mut HashMap<String, String>) {
2462 // Every package that exists in the pool — ALL ancestor prefixes of
2463 // every class path, not just direct parents: Java sees package
2464 // `z` as existing when any class lives at `z/a/b` (the weibo
2465 // `com.sina.weibo.z` interface collides with a subpackage-only
2466 // `z/` that has no directly-resident class).
2467 // Borrowed slices: the owned set allocated a fresh String per
2468 // ancestor prefix of every class (~600k allocations on weibo).
2469 let mut pkgs: jdc_core::FxHashSet<&str> = jdc_core::FxHashSet::default();
2470 for n in &pool.order {
2471 let mut rest = n.as_str();
2472 while let Some(i) = rest.rfind('/') {
2473 rest = &rest[..i];
2474 pkgs.insert(rest);
2475 }
2476 }
2477 // Exact-class and minted-display sets make the clash checks O(1):
2478 // the old `order.iter().any(starts_with(&cand))` scanned every name
2479 // per candidate per k-step (weibo: the second quadratic of the
2480 // single-class getclass path, ~1s) — and its prefix match also
2481 // counted SIBLING classes (`a/b2` matching `a/b23`) that map to
2482 // different files and never collide.
2483 let taken: jdc_core::FxHashSet<&str> = pool.order.iter().map(|s| s.as_str()).collect();
2484 let mut used: jdc_core::FxHashSet<String> = map.values().cloned().collect();
2485 for name in &pool.order {
2486 let Some((pkg, simple)) = name.rsplit_once('/') else {
2487 continue;
2488 };
2489 if simple.is_empty() || pkg.is_empty() || simple.contains('$') {
2490 // Nested classes render dotted (`Outer.a`) — no package
2491 // effect from their `$` tail.
2492 continue;
2493 }
2494 if !pkgs.contains(name.as_str()) || map.get(name).is_some_and(|v| v != name) {
2495 continue; // no subpackage under it, or an earlier rule moved it
2496 }
2497 let mut k = 1u32;
2498 loop {
2499 k += 1;
2500 let cand = format!("{pkg}/{simple}{k}");
2501 let clash = pkgs.contains(cand.as_str())
2502 || taken.contains(cand.as_str())
2503 || used.contains(&cand);
2504 if !clash {
2505 used.insert(cand.clone());
2506 map.insert(name.clone(), cand);
2507 break;
2508 }
2509 }
2510 }
2511}
2512
2513/// Obfuscators can name a class after its own package leaf (`package k;`
2514/// `class k extends EditText`). Every SAME-PACKAGE reference to `k.Anything`
2515/// then resolves the first segment to the CLASS, not the package —
2516/// javac looks for `Anything` as a member of class k ("找不到符号 class p2,
2517/// location: class k"). Renaming the class (display-level, the registry
2518/// carries declarations and references) makes the package the only
2519/// thing the leaf name resolves to.
2520fn pkg_leaf_shadow_renames(pool: &DexPool, map: &mut HashMap<String, String>) {
2521 let taken: jdc_core::FxHashSet<&str> = pool.order.iter().map(|s| s.as_str()).collect();
2522 let mut used: jdc_core::FxHashSet<String> = map.values().cloned().collect();
2523 // Classes per package in ONE pass — the per-candidate
2524 // `order.iter().filter(starts_with).count()` was a quadratic scan
2525 // on obfuscated corpora (weibo: 2124 leaf-shadow candidates × 98k
2526 // names ≈ 2×10⁸ memcmp — the dominant single-class getclass cost,
2527 // and paid again by every full run).
2528 let mut under_pkg: jdc_core::FxHashMap<&str, usize> = jdc_core::FxHashMap::default();
2529 for n in &pool.order {
2530 if let Some(i) = n.rfind('/') {
2531 *under_pkg.entry(&n[..i]).or_default() += 1;
2532 }
2533 }
2534 for name in &pool.order {
2535 let Some((pkg, simple)) = name.rsplit_once('/') else {
2536 continue;
2537 };
2538 if simple.is_empty() || pkg.is_empty() {
2539 continue;
2540 }
2541 let leaf = pkg.rsplit('/').next().unwrap_or(pkg);
2542 // class name == package leaf, and the package really exists
2543 // (more classes than just this one live under it).
2544 if simple != leaf {
2545 continue;
2546 }
2547 let siblings = under_pkg.get(pkg).copied().unwrap_or(0).saturating_sub(1);
2548 if siblings == 0 || map.get(name).is_some_and(|v| v != name) {
2549 continue; // already renamed by an earlier rule; identity anchors are fine to overwrite
2550 }
2551 let mut k = 1u32;
2552 loop {
2553 k += 1;
2554 let cand = format!("{pkg}/{simple}{k}");
2555 if !taken.contains(cand.as_str()) && !used.contains(&cand) {
2556 used.insert(cand.clone());
2557 map.insert(name.clone(), cand);
2558 break;
2559 }
2560 }
2561 }
2562}
2563
2564/// Member-level collision renames (fields AND methods of one class).
2565/// Two sources collapse onto one display name: obfuscators renaming a
2566/// synthetic outer reference (`this$0`) onto a real field's name, and
2567/// the identifier sanitizer mapping non-ASCII to `_` (whole
2568/// `ERROR_中文` families render one `ERROR________`).
2569/// Both are legal in bytecode (members resolve by index) and both are
2570/// javac "already defined" errors in source. Policy per colliding group
2571/// (fields keep the first non-synthetic name; a synthetic outer
2572/// reference restores `this$0`; everything else gets `2`/`3` suffixes;
2573/// method keys include erased parameter types):
2574/// Keys stay ORIGINAL (name, descriptor) so references resolve exactly;
2575/// only the rendered identifier changes.
2576/// Descriptor type string → internal class name (arrays/primitives →
2577/// None).
2578/// Can a method returning descriptor `m_ret` override an ancestor
2579/// declaration returning `r`? Equal, or a covariant subtype. ARRAY
2580/// returns must reach is_subtype in DESCRIPTOR form (it peels `[` for
2581/// element covariance): the old `strip L…;`-only comparison returned
2582/// false for `SavedState[] <: Object[]`, so Parcelable.Creator's real
2583/// `newArray(I)[LSavedState;` was judged an IMPOSTOR and renamed
2584/// (newArray3 — m_taken already held newArray + the bridge's newArray2),
2585/// breaking the interface override: "X$1不是抽象的, 并且未覆盖…
2586/// newArray(int)" (weibo ×504, battery ~789, rimet/news/uuyc ~655).
2587/// Object returns keep the internal-name form is_subtype's pool/fwdb
2588/// walk expects; an array m_ret against an object r resolves through
2589/// is_subtype's `sup == Object` / fwdb path (arrays ARE Objects).
2590fn ret_satisfies(pool: &DexPool, m_ret: &str, r: &str) -> bool {
2591 if m_ret == r {
2592 return true;
2593 }
2594 if m_ret.starts_with('[') {
2595 let sup: std::borrow::Cow<str> = if r.starts_with('[') {
2596 std::borrow::Cow::Borrowed(r)
2597 } else {
2598 match r.strip_prefix('L').and_then(|x| x.strip_suffix(';')) {
2599 Some(rc) => std::borrow::Cow::Borrowed(rc),
2600 None => return false, // primitive sup: only equality (above)
2601 }
2602 };
2603 return pool.is_subtype(m_ret, &sup);
2604 }
2605 if r.starts_with('[') {
2606 return false; // nothing but an array satisfies an array return
2607 }
2608 match (
2609 r.strip_prefix('L').and_then(|x| x.strip_suffix(';')),
2610 m_ret.strip_prefix('L').and_then(|x| x.strip_suffix(';')),
2611 ) {
2612 (Some(rc), Some(mc)) => pool.is_subtype(mc, rc),
2613 _ => false,
2614 }
2615}
2616
2617/// Raw name → (argsig, ret) JVM-descriptor slices of one class's own
2618/// virtual methods — the key space the collision queries use.
2619type AncMethods = jdc_core::FxHashMap<
2620 std::sync::Arc<str>,
2621 Vec<(std::sync::Arc<str>, std::sync::Arc<str>)>,
2622>;
2623
2624/// One provenance bucket of an analyzed class's ancestor chain (cached
2625/// layers, no merge — see `layer_rets`).
2626type AncLayers = Vec<std::sync::Arc<AncEntry>>;
2627
2628/// Parsed direct virtual methods of one ancestor class + its chain
2629/// edges, cached per class name for the member-collision walk. The
2630/// framework chains (View/Activity/Fragment…) and pool hub supers repeat
2631/// across thousands of analyzed classes — parsing each ancestor's
2632/// descriptors ONCE per run (instead of per analyzed subclass) removes
2633/// the dominant main-loop cost.
2634struct AncEntry {
2635 methods: AncMethods,
2636 /// Provenance bucket: FRAMEWORK ancestors (fwdb classes AND
2637 /// desugared `java.*`/`javax.*` pool copies — javac compiles against
2638 /// the android.jar originals, the covariant source truth) vs POOL
2639 /// ancestors (rendered RAW — the erased bridge is what satisfies the
2640 /// raw interface). The keeper preference flips on this axis (weibo
2641 /// Function0.invoke bridge must keep the name; WhatsApp SortedMap
2642 /// .keySet real override must).
2643 is_fw: bool,
2644 /// super + direct interfaces (chain continuation), Object-stripped.
2645 supers: Vec<std::sync::Arc<str>>,
2646}
2647
2648fn parse_anc_entry(pool: &DexPool, cn: &str) -> std::sync::Arc<AncEntry> {
2649 let mut methods: AncMethods = jdc_core::FxHashMap::default();
2650 let mut supers: Vec<std::sync::Arc<str>> = Vec::new();
2651 let is_fw;
2652 if let Some(ac) = pool.get_if_materialized(cn) {
2653 // Desugared `java.*`/`javax.*` copies in the dex are
2654 // pool-materialized, but javac compiles against the android.jar
2655 // ORIGINALS (covariant generics truth) — their declarations
2656 // belong to the fw bucket (WhatsApp bundles desugared
2657 // java.util.SortedMap; the pool bucket kept the
2658 // flattened-erasure bridge and renamed the real SortedSet
2659 // keySet).
2660 is_fw = cn.starts_with("java/") || cn.starts_with("javax/");
2661 for m in ac.all_methods() {
2662 if m.is_static() || &*m.name == "<init>" || &*m.name == "<clinit>"
2663 {
2664 continue;
2665 }
2666 let d: &str = &m.desc;
2667 let lo = d.find('(').map(|i| i + 1).unwrap_or(0);
2668 let hi = d.find(')').unwrap_or(d.len());
2669 methods
2670 .entry(m.name.clone())
2671 .or_default()
2672 .push((
2673 std::sync::Arc::from(&d[lo..hi]),
2674 std::sync::Arc::from(&d[hi + 1..]),
2675 ));
2676 }
2677 if let Some(s) = &ac.super_name {
2678 if s != "java/lang/Object" {
2679 supers.push(std::sync::Arc::from(s.as_str()));
2680 }
2681 }
2682 supers.extend(ac.interfaces.iter().map(|i| std::sync::Arc::from(i.as_str())));
2683 } else {
2684 // FRAMEWORK ancestors bind names too: a group member
2685 // implementing a framework declaration must keep the original
2686 // name (renaming WhatsApp's Map.keySet-style overrides broke
2687 // the interface chain — cannot-find:变量 ×18.6k). fwdb
2688 // descriptors are erased JVM form — directly comparable with
2689 // dex descs. A class that is NEITHER pool nor fwdb (phantom)
2690 // parses to an empty entry — a chain dead-end, same as before.
2691 is_fw = true;
2692 crate::fwdb::for_each_method(cn, |d, f| {
2693 if f & crate::fwdb::MF_STATIC != 0 {
2694 return;
2695 }
2696 let Some(open) = d.find('(') else {
2697 return;
2698 };
2699 let lo = open + 1;
2700 let hi = d.find(')').unwrap_or(0);
2701 // NOTE: hi == lo is the zero-arg case, not malformed; and
2702 // the name slice ends at `open` (a `d[..lo]` cut kept the
2703 // `(` — every fw key missed its gkey lookup).
2704 if hi == 0 || hi + 1 >= d.len() {
2705 return;
2706 }
2707 methods
2708 .entry(std::sync::Arc::from(&d[..open]))
2709 .or_default()
2710 .push((
2711 std::sync::Arc::from(&d[lo..hi]),
2712 std::sync::Arc::from(&d[hi + 1..]),
2713 ));
2714 });
2715 if let Some(sup) = crate::fwdb::super_of(cn) {
2716 if sup != "java/lang/Object" {
2717 supers.push(std::sync::Arc::from(sup));
2718 }
2719 }
2720 crate::fwdb::for_each_interface(cn, |i| supers.push(std::sync::Arc::from(i)));
2721 }
2722 std::sync::Arc::new(AncEntry {
2723 methods,
2724 is_fw,
2725 supers,
2726 })
2727}
2728
2729/// Rets declared for `(name, argsig)` by any layer of a chain — lazy,
2730/// no merge (every consumer is an "any ancestor declares this" query,
2731/// so a per-layer scan replaces the old per-class eager map build).
2732/// Alloc-free: `Arc<str>: Borrow<str>` keys the hash lookup by `&str`.
2733fn layer_rets<'a>(
2734 layers: &'a [std::sync::Arc<AncEntry>],
2735 name: &'a str,
2736 argsig: &'a str,
2737) -> impl Iterator<Item = &'a str> {
2738 layers
2739 .iter()
2740 .filter_map(move |e| e.methods.get(name))
2741 .flatten()
2742 .filter(move |(a, _)| &**a == argsig)
2743 .map(|(_, r)| &**r)
2744}
2745
2746fn member_collision_renames(
2747 pool: &DexPool,
2748) -> HashMap<std::sync::Arc<str>, Vec<jdc_core::rename::FieldRename>> {
2749 let mut out: HashMap<std::sync::Arc<str>, Vec<jdc_core::rename::FieldRename>> =
2750 HashMap::default();
2751 // Direct nested tails per base class (`C$a` → a for base C), for
2752 // the companion-field rule below.
2753 let mut child_tails: jdc_core::FxHashMap<&str, Vec<&str>> =
2754 jdc_core::FxHashMap::default();
2755 // Per-package RENAMED class display heads: a MINTED field display
2756 // equal to one of these shadows same-package qualified type refs
2757 // (`a2.changeQuickRedirect` binds to the field when the class a2's
2758 // display is a2 — the inherited rule minting field a→a2 ONTO the
2759 // pkg-renamed class display created exactly this: weibo +194
2760 // cannot-find). Keyed on the POST-rename display (what refs render),
2761 // covering each class's own display too.
2762 let root_segs = pool.root_pkg_segs();
2763 // Lazy single-class pools build the global tables from MATERIALIZED
2764 // classes only (the target + its on-demand family): unmateralized
2765 // siblings matter to whole-program javac, which a one-class stdout
2766 // dump does not have — the same reduced-fidelity contract the
2767 // obscuring rules already document for lazy pools. The walk over
2768 // every registered name was the dominant field-rename cost of
2769 // getclass (~0.5s on weibo for one class).
2770 let lazy = !pool_majority_materialized(pool);
2771 let mut pkg_displays: jdc_core::FxHashMap<&str, jdc_core::FxHashSet<String>> =
2772 jdc_core::FxHashMap::default();
2773 for n in &pool.order {
2774 if lazy && pool.get_if_materialized(n).is_none() {
2775 continue;
2776 }
2777 let renamed = crate::apply_class_rename(n);
2778 let (pkg, simple) = match renamed.rfind('/') {
2779 Some(i) => (renamed[..i].to_string(), renamed[i + 1..].to_string()),
2780 None => (String::new(), renamed.to_string()),
2781 };
2782 let head = simple.split('$').next().unwrap_or(&simple).to_string();
2783 if !head.is_empty() {
2784 let pkg_key = n.rfind('/').map(|i| &n[..i]).unwrap_or("");
2785 let _ = pkg;
2786 pkg_displays.entry(pkg_key).or_default().insert(head);
2787 }
2788 }
2789 // Reverse super map — the obscuring rename must register under every
2790 // transitive subclass (dex refs of an inherited field may name any
2791 // hierarchy class as owner).
2792 let mut subs: jdc_core::FxHashMap<&str, Vec<&str>> = jdc_core::FxHashMap::default();
2793 for n in &pool.order {
2794 if lazy && pool.get_if_materialized(n).is_none() {
2795 continue;
2796 }
2797 // EVERY `$` boundary: each prefix is an owner and the segment
2798 // after it one of its direct member-type tails. The old
2799 // first-segment-only map left NESTED owners (ta8/t$a, androidx
2800 // i0$e) with no member displays — their fields kept or minted
2801 // names colliding with their own member types, and in qualified
2802 // chains a VARIABLE OBSCURES a member type (JLS 6.4.2):
2803 // `ta8.t.a.a100.a` (Kotlin-object singleton sget, nova
2804 // static-ctx ×1,281) and `i0.e.b.ADDING` (lark ×145) bound to
2805 // the field instead of the nested class.
2806 for (i, _) in n.match_indices('$') {
2807 let owner = &n[..i];
2808 let rest = &n[i + 1..];
2809 let tail = match rest.find('$') {
2810 Some(j) => &rest[..j],
2811 None => rest,
2812 };
2813 if owner.is_empty() || tail.is_empty() {
2814 continue;
2815 }
2816 let v = child_tails.entry(owner).or_default();
2817 if !v.contains(&tail) {
2818 v.push(tail);
2819 }
2820 }
2821 if let Some(pc) = pool.get_if_materialized(n) {
2822 if let Some(sup) = &pc.super_name {
2823 if sup != "java/lang/Object" {
2824 subs.entry(sup.as_str()).or_default().push(n.as_str());
2825 }
2826 }
2827 }
2828 }
2829 // Per-ancestor parse cache (function-local: lives exactly as long
2830 // as this rename install, so materialization state is consistent by
2831 // construction; freed on return — no cross-run staleness, no RSS
2832 // tail).
2833 let mut anc_cache: jdc_core::FxHashMap<
2834 std::sync::Arc<str>,
2835 std::sync::Arc<AncEntry>,
2836 > = jdc_core::FxHashMap::default();
2837 for name in &pool.order {
2838 let Some(pc) = pool.get_if_materialized(name) else {
2839 continue;
2840 };
2841 let outer_ref_desc = find_outer_name(pool, name)
2842 .filter(|o| o != name)
2843 .map(|o| format!("L{o};"));
2844 // ---- fields: one shared namespace (static + instance) ----
2845 let fields: Vec<&PoolField> = pc
2846 .static_fields
2847 .iter()
2848 .chain(pc.instance_fields.iter())
2849 .collect();
2850 let fgroups: HashMap<String, Vec<&PoolField>> = HashMap::default();
2851 let mut fgroups = fgroups;
2852 for f in &fields {
2853 let disp = crate::classdec::java_ident(&f.name).into_owned();
2854 fgroups.entry(disp).or_default().push(f);
2855 }
2856 let mut f_taken: jdc_core::FxHashSet<String> = fields
2857 .iter()
2858 .map(|f| crate::classdec::java_ident(&f.name).into_owned())
2859 .collect();
2860 // Displays already registered under THIS class by ancestor
2861 // subs-propagation (an ancestor processed earlier in pool.order
2862 // pushed its renames down): the own mints must avoid them or
2863 // two fields in one class render the same display
2864 // ("已在类中定义了变量 a2", weibo bx$b: subs-pushed bv→a2 beside
2865 // its own String→a2).
2866 if let Some(v) = out.get(&std::sync::Arc::from(name.as_str())) {
2867 for fr in v {
2868 f_taken.insert(fr.display.to_string());
2869 }
2870 }
2871 for group in fgroups.values() {
2872 if group.len() < 2 {
2873 continue;
2874 }
2875 let keeper_pos = group
2876 .iter()
2877 .position(|f| f.access & crate::access::ACC_SYNTHETIC == 0)
2878 .unwrap_or(0);
2879 for (gi, f) in group.iter().enumerate() {
2880 if gi == keeper_pos {
2881 continue;
2882 }
2883 let base = crate::classdec::java_ident(&f.name).into_owned();
2884 let display: String = match &outer_ref_desc {
2885 Some(ord)
2886 if f.access & crate::access::ACC_SYNTHETIC != 0
2887 && !f.is_static
2888 && *ord == *f.desc =>
2889 {
2890 let mut k = 0u32;
2891 loop {
2892 let cand = if k == 0 {
2893 "this$0".to_string()
2894 } else {
2895 format!("this$0{k}")
2896 };
2897 if f_taken.insert(cand.clone()) {
2898 break cand;
2899 }
2900 k += 1;
2901 }
2902 }
2903 _ => suffix_unique(&base, &mut f_taken),
2904 };
2905 // Register with subclass propagation: dex refs of an
2906 // inherited field may name ANY hierarchy class as owner
2907 // (R8 writes the referrer), and without the propagation
2908 // the declaring class's rename does not reach the
2909 // subclass-keyed lookups — the ref renders the RAW name
2910 // against a renamed declaration (bx$b read its inherited
2911 // `a:Lcom/xiaomi/push/bv;` as `this.a` while bx$a
2912 // declared it `a2`). Mirrors the method-side hierarchy
2913 // consistency rule.
2914 register_field_rename_with_subs(
2915 pool, &subs, &mut out, name, &f.name, &f.desc, &display,
2916 );
2917 }
2918 }
2919 // ---- fields obscuring nested types (JLS 6.4.2) ----
2920 // `C.b.a(..)` with BOTH a field `b` and a member class `C$b`
2921 // binds to the FIELD — the qualified type reading is obscured
2922 // ("无法从静态上下文中引用非静态 变量 b" + the paired cannot-find
2923 // cascade; lark LKEvent/MotionLayout family, static-ctx ×1,154).
2924 // The CHILD-side rule (nested_collision_renames) already renames
2925 // colliding children when its local_ok gate passes (`a$a` renders
2926 // as `a2`); this field-side rule covers the gate-REJECTED residual
2927 // (child keeps display `b`, program-wide refs block its rename).
2928 // A field's refs all funnel through field_display (lift call
2929 // sites, declarations, const inits), so the field side needs no
2930 // locality gate. TWO invariants the first attempt violated (lark
2931 // +8,129): fire on the child's FINAL DISPLAY last segment, not
2932 // the raw tail (a child the child-side rule already renamed does
2933 // NOT obscure), and mint avoiding every child display (`a`→`a2`
2934 // landed on the existing `interface a2`, minting a fresh
2935 // obscuring pair).
2936 {
2937 let tails: &[&str] = child_tails
2938 .get(name.as_str())
2939 .map(|v| v.as_slice())
2940 .unwrap_or(&[]);
2941 let mut member_displays: jdc_core::FxHashSet<String> =
2942 jdc_core::FxHashSet::default();
2943 let mut obscuring: jdc_core::FxHashSet<String> =
2944 jdc_core::FxHashSet::default();
2945 for t in tails {
2946 let full = format!("{name}${t}");
2947 let renamed = crate::apply_class_rename(&full);
2948 let last = renamed
2949 .rsplit('/')
2950 .next()
2951 .unwrap_or(&renamed)
2952 .rsplit('$')
2953 .next()
2954 .unwrap_or(t)
2955 .to_string();
2956 member_displays.insert(last.clone());
2957 // Only an INLINE-rendered member is reachable by the bare
2958 // simple name (own-file children are referenced by their
2959 // flat `$` form, which fields cannot obscure).
2960 let member_rendered = clean_member_tail(t)
2961 && crate::find_outer_name(pool, &full).as_deref()
2962 == Some(name.as_str());
2963 if member_rendered {
2964 obscuring.insert(last);
2965 }
2966 }
2967 // The class's OWN simple display name joins the obscuring
2968 // set: a field named like its class captures every
2969 // qualified self-static ref from nested scopes
2970 // (`a2.changeQuickRedirect` bound to the String FIELD a2 —
2971 // weibo robust refs ×317; bare refs survive via the
2972 // variable namespace, the qualified form does not).
2973 let self_renamed = crate::apply_class_rename(name);
2974 let self_simple = self_renamed
2975 .rsplit('/')
2976 .next()
2977 .unwrap_or(&self_renamed)
2978 .rsplit('$')
2979 .next()
2980 .unwrap_or("")
2981 .to_string();
2982 if !self_simple.is_empty() {
2983 obscuring.insert(crate::classdec::java_ident(&self_simple).into_owned());
2984 }
2985 let self_pkg = name.rfind('/').map(|i| &name[..i]).unwrap_or("");
2986 let pkg_names = pkg_displays.get(self_pkg);
2987 for f in &fields {
2988 let fdisp = crate::classdec::java_ident(&f.name).into_owned();
2989 if fdisp.starts_with("this$") || !obscuring.contains(fdisp.as_str()) {
2990 continue;
2991 }
2992 // Already renamed by the same-class duplicate-group rule?
2993 // Its display no longer equals the obscuring simple name,
2994 // so this rule's target collision does not exist — and
2995 // minting a SECOND entry for the same (name, desc) with
2996 // subs-propagation put a CONFLICTING display under every
2997 // subclass: refs naming the subclass resolved to one name
2998 // while the declaring class's declaration rendered the
2999 // first entry's (weibo bx$b read its inherited
3000 // `a:Lcom/xiaomi/push/bv;` as `this.a8` while bx$a
3001 // declared it `a2` — the R8 same-name-field family).
3002 let already = out
3003 .get(&std::sync::Arc::from(name.as_str()))
3004 .is_some_and(|v| {
3005 v.iter().any(|fr| *fr.name == *f.name && *fr.desc == *f.desc)
3006 });
3007 if already {
3008 continue;
3009 }
3010 let mut k = 1u32;
3011 let display = loop {
3012 k += 1;
3013 let cand = format!("{fdisp}{k}");
3014 // root_segs: a minted display equal to a PACKAGE
3015 // first segment captures every qualified ref to
3016 // that package inside the family's file (field e2
3017 // ate `e2.c.b(..)` — lark pq/e "变量 c" ×47). The
3018 // deshadow builder cannot repair this after the
3019 // fact: it scans RAW names, and the collision is
3020 // born here in display space.
3021 if !f_taken.contains(&cand)
3022 && !member_displays.contains(&cand)
3023 && !root_segs.contains(&cand)
3024 && !pkg_names.is_some_and(|p| p.contains(cand.as_str()))
3025 {
3026 f_taken.insert(cand.clone());
3027 member_displays.insert(cand.clone());
3028 break cand;
3029 }
3030 };
3031 register_field_rename_with_subs(
3032 pool, &subs, &mut out, name, &f.name, &f.desc, &display,
3033 );
3034 if std::env::var("DDC_OBSCURE_STATS").is_ok() {
3035 eprintln!("[obscure] {name} field {} -> {}", f.name, display);
3036 }
3037 }
3038 }
3039 // ---- companion-holder fields vs nested types (JLS 6.4.2 in
3040 // expression context): Kotlin compiles `object`/companion
3041 // holders as a static field whose name EQUALS the nested class
3042 // it instances (`static final j$a a`), and static members of the
3043 // nested are invoked as `j.a.c(..)` — but `j.a` in an EXPRESSION
3044 // resolves to the FIELD (variables outrank member types), so the
3045 // call becomes a member lookup on the instance type... which is
3046 // the same class, yet javac reports 找不到符号 for statics
3047 // reached through the shadowed path, and the whole file
3048 // error-types (weibo feed/business/j: 654 errors + closure
3049 // cascade; the nested-side rename for this is the LOCAL-OK-gated
3050 // rule that explodes ungated — the FIELD side is safe because
3051 // dex field refs always carry the declaring class, so this
3052 // registry keys every consumer). Gate: the nested class has
3053 // static members (companion shape) and the field's descriptor
3054 // is exactly the nested type.
3055 if let Some(tails) = child_tails.get(name.as_str()) {
3056 for tail in tails {
3057 let nested = format!("{name}${tail}");
3058 let Some(nc) = pool.get_if_materialized(&nested) else {
3059 continue;
3060 };
3061 // Only CLEAN tails render dotted (`j.a`); flat-rendered
3062 // tails (`j$1`) carry the `$` and never clash.
3063 if !clean_member_tail(tail) {
3064 continue;
3065 }
3066 let has_statics = nc
3067 .static_fields
3068 .iter()
3069 .chain(nc.instance_fields.iter())
3070 .any(|f| f.is_static)
3071 || nc.all_methods().any(|m| m.access & crate::access::ACC_STATIC != 0);
3072 if !has_statics {
3073 continue;
3074 }
3075 // STRICT companion shape: the field's descriptor IS the
3076 // nested type (`static final j$a a`). The relaxed form
3077 // (any same-named field beside a statics-bearing nested)
3078 // minted thousands of renames on weibo and exploded the
3079 // battery to the 500k cap — the field funnel does not
3080 // hold at that scale (改名面 vs 引用覆盖面, again).
3081 let want_desc = format!("L{nested};");
3082 for f in pc.static_fields.iter().chain(pc.instance_fields.iter()) {
3083 if f.desc.as_ref() != want_desc.as_str() {
3084 continue;
3085 }
3086 let disp = crate::classdec::java_ident(&f.name).into_owned();
3087 if disp != crate::classdec::java_ident(tail) {
3088 continue;
3089 }
3090 // Already renamed by the duplicate-group rule above?
3091 let already = out
3092 .get(&std::sync::Arc::from(name.as_str()))
3093 .is_some_and(|v| {
3094 v.iter().any(|fr| *fr.name == *f.name && *fr.desc == *f.desc)
3095 });
3096 if already {
3097 continue;
3098 }
3099 let display = suffix_unique(&disp, &mut f_taken);
3100 f_taken.insert(display.clone());
3101 out.entry(std::sync::Arc::from(name.as_str()))
3102 .or_default()
3103 .push(jdc_core::rename::FieldRename {
3104 name: f.name.clone(),
3105 desc: f.desc.clone(),
3106 display: std::sync::Arc::from(display.as_str()),
3107 });
3108 }
3109 }
3110 }
3111 // ---- methods: display key = sanitized name + erased params ----
3112 let methods: Vec<&PoolMethod> = pc.all_methods().collect();
3113 // Ancestor virtual-method returns per (name, argsig) — one
3114 // chain walk per non-interface class, shared by the
3115 // covariant-keeper selection and the impostor-override
3116 // detection below. Layers stay PER-ANCESTOR (cached Arcs, no
3117 // merge): every consumer is an "any ancestor declares
3118 // (name, argsig) → ret" query, so a per-layer scan replaces
3119 // the old per-class eager map build. Provenance splits the
3120 // chain into fw / pool buckets (see AncEntry.is_fw).
3121 let mut anc_chain: Option<(AncLayers, AncLayers)> = None;
3122 if pc.access & crate::access::ACC_INTERFACE == 0 {
3123 let mut fw_layers: AncLayers = Vec::new();
3124 let mut pool_layers: AncLayers = Vec::new();
3125 let mut stack: Vec<std::sync::Arc<str>> = Vec::new();
3126 if let Some(sup) = &pc.super_name {
3127 if sup != "java/lang/Object" {
3128 stack.push(std::sync::Arc::from(sup.as_str()));
3129 }
3130 }
3131 stack.extend(
3132 pc.interfaces
3133 .iter()
3134 .map(|i| std::sync::Arc::from(i.as_str())),
3135 );
3136 let mut seen_cls: jdc_core::FxHashSet<std::sync::Arc<str>> =
3137 jdc_core::FxHashSet::default();
3138 while let Some(cn) = stack.pop() {
3139 if !seen_cls.insert(cn.clone()) {
3140 continue;
3141 }
3142 let entry = match anc_cache.get(cn.as_ref()) {
3143 Some(e) => e.clone(),
3144 None => {
3145 let e = parse_anc_entry(pool, &cn);
3146 anc_cache.insert(cn.clone(), e.clone());
3147 e
3148 }
3149 };
3150 if !entry.methods.is_empty() {
3151 if entry.is_fw {
3152 fw_layers.push(entry.clone());
3153 } else {
3154 pool_layers.push(entry.clone());
3155 }
3156 }
3157 stack.extend(entry.supers.iter().cloned());
3158 }
3159 anc_chain = Some((fw_layers, pool_layers));
3160 }
3161 let mut m_taken: jdc_core::FxHashSet<String> = methods
3162 .iter()
3163 .map(|m| crate::classdec::java_ident(&m.name).into_owned())
3164 .collect();
3165 let mut mgroups: HashMap<(String, String), Vec<&PoolMethod>> = HashMap::default();
3166 for m in &methods {
3167 let d: &str = &m.desc;
3168 let lo = d.find('(').map(|i| i + 1).unwrap_or(0);
3169 let hi = d.find(')').unwrap_or(d.len());
3170 let key = (
3171 crate::classdec::java_ident(&m.name).into_owned(),
3172 d[lo..hi].to_string(),
3173 );
3174 mgroups.entry(key).or_default().push(m);
3175 }
3176 for ((base, _), group) in &mgroups {
3177 if group.len() < 2 {
3178 continue;
3179 }
3180 // R8 return-type specialization: a non-isolated class carries
3181 // `M(args)void` (the real logic) beside the interface bridge
3182 // `M(args)R` (unbox → call void → return Unit). Same erasure,
3183 // so Java source can't hold both, and `void` never satisfies
3184 // the interface's `R` — the claim map keeps the void side and
3185 // drops the bridge, yielding "X不是抽象的, 并且未覆盖…invoke"
3186 // (Kotlin suspend lambdas, weixin/lark ×1.1k). This is NOT the
3187 // covariant-override shape below (void has no subtype
3188 // relation): keep the BRIDGE under the original name so it
3189 // satisfies the interface, and rename the void specialization
3190 // so both render and the bridge's call to it resolves to the
3191 // renamed method.
3192 // `)Ljava/lang/Void;` counts as void-like: a Kotlin `Nothing`
3193 // function (an unsupported-op that always throws) compiles to
3194 // the Void CLASS return, not the `V` primitive, but it IS the
3195 // real-logic side of the same specialization pair. Missing it
3196 // sent the group down the covariant path, which kept the Void
3197 // method and renamed the bridge, then the impostor pass renamed
3198 // the Void method too — leaving NO `addAll` to satisfy Set
3199 // (weibo compose SnapshotMapEntrySet "未覆盖Set中的addAll").
3200 let ret_void = |m: &PoolMethod| {
3201 m.desc.ends_with(")V") || m.desc.ends_with(")Ljava/lang/Void;")
3202 };
3203 let isolated = pc
3204 .super_name
3205 .as_ref()
3206 .is_none_or(|s| s == "java/lang/Object")
3207 && pc.interfaces.is_empty();
3208 let has_void = group.iter().any(|m| ret_void(m) && &*m.name != "<init>");
3209 let has_nonvoid = group.iter().any(|m| !ret_void(m));
3210 if !isolated && has_void && has_nonvoid {
3211 let keeper_desc: Option<&str> = group
3212 .iter()
3213 .copied()
3214 .find(|m| {
3215 m.access & crate::access::ACC_BRIDGE != 0
3216 && !ret_void(m)
3217 && &*m.name != "<init>"
3218 })
3219 .or_else(|| {
3220 group
3221 .iter()
3222 .copied()
3223 .find(|m| !ret_void(m) && &*m.name != "<init>")
3224 })
3225 .map(|m| &*m.desc);
3226 let mut seen_orig: jdc_core::FxHashSet<(&str, &str)> =
3227 jdc_core::FxHashSet::default();
3228 for m in group.iter().copied() {
3229 if keeper_desc == Some(&*m.desc) {
3230 seen_orig.insert((&*m.name, &*m.desc));
3231 continue; // bridge keeps the interface name
3232 }
3233 if !seen_orig.insert((&*m.name, &*m.desc)) {
3234 continue; // exact duplicate: emitter drops it
3235 }
3236 let display = suffix_unique(base, &mut m_taken);
3237 out.entry(std::sync::Arc::from(name.as_str()))
3238 .or_default()
3239 .push(jdc_core::rename::FieldRename {
3240 name: m.name.clone(),
3241 desc: m.desc.clone(),
3242 display: std::sync::Arc::from(display.as_str()),
3243 });
3244 }
3245 continue;
3246 }
3247 // A COVARIANT override pair (same ORIGINAL name, different
3248 // descriptors — the compiler's bridge shape: the interface's
3249 // `deserialize(e)` plus the narrower implementation). Renaming
3250 // either side breaks @Override ("is not abstract and does not
3251 // override abstract method deserialize(e)", ~2.2k on reqable);
3252 // the emitter's claim logic already renders only the
3253 // non-bridge member of the pair. An ISOLATED class (extends
3254 // Object, implements nothing — d8's ExternalSyntheticApiModel
3255 // Outline shells) overrides nothing: its same-(name,params)
3256 // pairs differing only by RETURN type are dex-legal distinct
3257 // methods, and dropping one strands its callers on the
3258 // survivor (`HalfKt$$…0.m()Class` vs `.m()V` — "void无法
3259 // 转换为Class", weibo ×85+).
3260 let covariant = !isolated
3261 && group
3262 .iter()
3263 .any(|m| group.iter().any(|o| o.name == m.name && o.desc != m.desc));
3264 if covariant {
3265 // Return-type-only clash in a class WITH a hierarchy:
3266 // Java cannot declare both, and the claim logic would
3267 // keep an arbitrary one — when that is NOT the member
3268 // an ancestor declares, every concrete subclass fails
3269 // "X不是抽象的, 并且未覆盖…" (weibo ×590: viewmodel.a
3270 // kept getUsageInfo()e while interface d declares
3271 // getUsageInfo()g; the kept side also strands the
3272 // dropped side's callers). Keep the ancestor-declared
3273 // member under the original name and rename the rest —
3274 // the registry carries every call site, and the
3275 // propagation below shares the display down the family.
3276 // 0 or 2+ ancestor-matched members = the shape Java
3277 // cannot express at all (two interfaces demanding
3278 // different returns): keep the old claim behavior.
3279 let keepers: Vec<&&PoolMethod> = group
3280 .iter()
3281 .filter(|m| {
3282 let d: &str = &m.desc;
3283 let lo = d.find('(').map(|i| i + 1).unwrap_or(0);
3284 let hi = d.find(')').unwrap_or(d.len());
3285 if let Some((fwl, pll)) = &anc_chain {
3286 layer_rets(fwl, &m.name, &d[lo..hi])
3287 .chain(layer_rets(pll, &m.name, &d[lo..hi]))
3288 .any(|r| r == &d[hi + 1..])
3289 } else {
3290 false
3291 }
3292 })
3293 .collect();
3294 if keepers.len() >= 2 {
3295 // The shape Java cannot express at all (2+
3296 // ancestors demanding different returns): no
3297 // rename target — keep the claim behavior.
3298 continue;
3299 }
3300 // fwdb flattens JDK covariant overrides to the erased
3301 // declaring return (SortedMap.keySet records Set, the
3302 // source return is SortedSet) — the sole "keeper" can
3303 // be the synthetic BRIDGE while the real override is
3304 // the non-bridge sibling whose return is a subtype of
3305 // the recorded one. Keeping the bridge renamed the
3306 // real method and broke the interface chain (WhatsApp
3307 // X/Jl2 SortedMap, cannot-find:变量 ×18.6k). Prefer the
3308 // non-bridge in that shape: the claim map dropped the
3309 // bridge before this machinery existed anyway.
3310 // Keeper selection by ancestor PROVENANCE:
3311 // - POOL ancestor matches (raw-rendered): the member
3312 // whose return EXACTLY matches a pool erasure that
3313 // is NOT purely framework-sourced keeps the name —
3314 // the erased bridge is what satisfies the raw
3315 // interface (weibo Function0.invoke / GlideRequest
3316 // .load; renaming the real one instead cost +375).
3317 // - FRAMEWORK-only match: javac reads the compiled
3318 // covariant truth (SortedMap.keySet→SortedSet) while
3319 // fwdb records the flattened erasure (→Set) that the
3320 // BRIDGE matches — prefer the non-bridge sibling
3321 // whose return is a subtype (WhatsApp X/Jl2).
3322 // - no ancestor: first non-bridge (the old claim
3323 // survivor), else dex order.
3324 fn ret_desc(m: &PoolMethod) -> &str {
3325 let d: &str = &m.desc;
3326 &d[d.find(')').map(|i| i + 1).unwrap_or(d.len())..]
3327 }
3328 let gkey = {
3329 let d: &str = &group[0].desc;
3330 let lo = d.find('(').map(|i| i + 1).unwrap_or(0);
3331 let hi = d.find(')').unwrap_or(d.len());
3332 (
3333 crate::classdec::java_ident(&group[0].name).into_owned(),
3334 d[lo..hi].to_string(),
3335 )
3336 };
3337 // fw match: a NON-BRIDGE member whose return is
3338 // assignable to a framework ret (covariant source
3339 // truth — it satisfies the raw pool ancestor too,
3340 // overrides being covariant-legal in Java).
3341 let fw_i = anc_chain.as_ref().and_then(|(fwl, _)| {
3342 let fws: Vec<&str> =
3343 layer_rets(fwl, &gkey.0, &gkey.1).collect();
3344 if fws.is_empty() {
3345 return None;
3346 }
3347 group.iter().position(|m| {
3348 m.access & crate::access::ACC_BRIDGE == 0
3349 && fws.iter().any(|fr| {
3350 // ret_satisfies: descriptor-level, so
3351 // ARRAY covariant returns match too (the
3352 // old ret_internal strip returned None
3353 // for `[LX;` and fell back to descriptor
3354 // equality — Creator newArray's real
3355 // `[LSavedState;` never matched fw's
3356 // `[LObject;`).
3357 ret_satisfies(pool, ret_desc(m), fr)
3358 })
3359 })
3360 });
3361 let pool_i = anc_chain.as_ref().and_then(|(_, pll)| {
3362 let rets: Vec<&str> =
3363 layer_rets(pll, &gkey.0, &gkey.1).collect();
3364 if rets.is_empty() {
3365 return None;
3366 }
3367 group
3368 .iter()
3369 .position(|m| rets.iter().any(|r| *r == ret_desc(m)))
3370 });
3371 let keeper_idx = match (fw_i, pool_i) {
3372 (Some(fi), _) => fi,
3373 (None, Some(pi)) => pi,
3374 (None, None) => group
3375 .iter()
3376 .position(|m| m.access & crate::access::ACC_BRIDGE == 0)
3377 .unwrap_or(0),
3378 };
3379 // keepers==1: the ancestor-declared member keeps the
3380 // original name. keepers==0: NOTHING in the POOL binds
3381 // the name — keep the member the old claim map would
3382 // have kept (first NON-BRIDGE, else dex order) and
3383 // rename the rest through the registry: letting the
3384 // claim map silently drop a side stranded its callers
3385 // on the survivor's incompatible return (weibo xiaomi
3386 // push gk/dq/hj/hf/gt: `boolean b()Z` dropped beside
3387 // `String b()` — every `boolean v = this.b()` bound to
3388 // the String twin, "String无法转换为boolean" ×300+ in
3389 // the push package alone). NON-bridge preference is
3390 // load-bearing: a framework-ancestor covariant pair
3391 // (emoji2 n extends SpannableStringBuilder, Editable-
3392 // bridge beside the real override) satisfies the
3393 // ancestor through the REAL method — bridge-preference
3394 // renamed it and broke the override chain (reqable +8
3395 // "无法覆盖…返回类型不兼容").
3396 let keep_desc: &str = &group[keeper_idx].desc;
3397 let mut seen_orig: jdc_core::FxHashSet<(&str, &str)> =
3398 jdc_core::FxHashSet::default();
3399 for m in group.iter().copied() {
3400 if m.desc.as_ref() == keep_desc {
3401 seen_orig.insert((&*m.name, &*m.desc));
3402 continue;
3403 }
3404 if !seen_orig.insert((&*m.name, &*m.desc)) {
3405 continue;
3406 }
3407 let display = suffix_unique(base, &mut m_taken);
3408 out.entry(std::sync::Arc::from(name.as_str()))
3409 .or_default()
3410 .push(jdc_core::rename::FieldRename {
3411 name: m.name.clone(),
3412 desc: m.desc.clone(),
3413 display: std::sync::Arc::from(display.as_str()),
3414 });
3415 }
3416 continue;
3417 }
3418 // Real re-declarations (same original name AND descriptor —
3419 // R8 duplicates) are already deduped by the emitter's claim
3420 // logic; only SANITIZER collapses reach here and every one of
3421 // them is a distinct method.
3422 let mut base = base.clone();
3423 let mut seen_orig: jdc_core::FxHashSet<(&str, &str)> =
3424 jdc_core::FxHashSet::default();
3425 for m in group {
3426 if !seen_orig.insert((&m.name, &m.desc)) {
3427 continue; // exact duplicate: emitter drops it
3428 }
3429 let display = suffix_unique(&base, &mut m_taken);
3430 out.entry(std::sync::Arc::from(name.as_str()))
3431 .or_default()
3432 .push(jdc_core::rename::FieldRename {
3433 name: m.name.clone(),
3434 desc: m.desc.clone(),
3435 display: std::sync::Arc::from(display.as_str()),
3436 });
3437 // The base stays the sanitized ORIGINAL so the first
3438 // occurrence keeps the plain name.
3439 let _ = &mut base;
3440 }
3441 }
3442 // ---- impostor overrides (R8 generic specialization) ----
3443 // A class method whose (name, argsig) matches an ancestor
3444 // declaration but whose return type satisfies NONE of the
3445 // ancestor returns is not an override Java can express: the
3446 // pair shares an erasure, so the class cannot declare both, and
3447 // ART never resolved it either (interface dispatch matches
3448 // name+descriptor — the missing proto throws AbstractMethodError
3449 // at runtime; weibo IStreamPresenter.getView()IStreamView vs the
3450 // impl's getView()IPageView with no subtype edge, ×258 across
3451 // getView/getWidget/getListView families). Rename the impostor
3452 // (the registry carries every caller and the propagation below
3453 // carries subclasses) and let synth_missing_interface_stubs add
3454 // the ART-faithful stub — rename-aware provision makes the
3455 // requirement missing again.
3456 if let Some((fwl, pll)) = &anc_chain {
3457 for m in &methods {
3458 if m.is_static()
3459 || &*m.name == "<init>"
3460 || &*m.name == "<clinit>"
3461 {
3462 continue;
3463 }
3464 let d: &str = &m.desc;
3465 let lo = d.find('(').map(|i| i + 1).unwrap_or(0);
3466 let hi = d.find(')').unwrap_or(d.len());
3467 let m_ret = &d[hi + 1..];
3468 // Lazy per-layer scan: `found` replaces the old
3469 // `map.get(key)` presence gate (no ancestor declares
3470 // (name, argsig) ⇒ nothing to violate ⇒ skip).
3471 let mut found = false;
3472 let mut satisfied = false;
3473 'lay: for e in fwl.iter().chain(pll.iter()) {
3474 let Some(rets) = e.methods.get(m.name.as_ref()) else {
3475 continue;
3476 };
3477 for (a, r) in rets {
3478 if **a != d[lo..hi] {
3479 continue;
3480 }
3481 found = true;
3482 // ret_satisfies handles ARRAY covariance in
3483 // descriptor form — the old L…;-strip match
3484 // answered false for `SavedState[] <: Object[]`
3485 // and renamed the REAL Parcelable.Creator
3486 // newArray as an impostor (newArray3), breaking
3487 // the override ("不是抽象的…未覆盖 newArray(int)",
3488 // weibo ×504 / battery ~789).
3489 if ret_satisfies(pool, m_ret, r) {
3490 satisfied = true;
3491 break 'lay;
3492 }
3493 }
3494 }
3495 if !found || satisfied {
3496 continue;
3497 }
3498 // Already renamed by a clash-group rule above?
3499 let already = out
3500 .get(&std::sync::Arc::from(name.as_str()))
3501 .is_some_and(|v| {
3502 v.iter().any(|fr| *fr.name == *m.name && *fr.desc == *m.desc)
3503 });
3504 if already {
3505 continue;
3506 }
3507 let base = crate::classdec::java_ident(&m.name).into_owned();
3508 let display = suffix_unique(&base, &mut m_taken);
3509 out.entry(std::sync::Arc::from(name.as_str()))
3510 .or_default()
3511 .push(jdc_core::rename::FieldRename {
3512 name: m.name.clone(),
3513 desc: m.desc.clone(),
3514 display: std::sync::Arc::from(display.as_str()),
3515 });
3516 }
3517 }
3518 }
3519 // ---- method-rename propagation down the hierarchy ----
3520 // A renamed method must keep ONE display across its whole override
3521 // family: the parent's abstract declaration and every subclass
3522 // implementation/caller render through the registry, and a missed
3523 // link re-breaks the override ("未覆盖" at each concrete subclass)
3524 // or the call site (cannot-find). Register each method entry on the
3525 // full transitive subclass + implementer closure — a subclass that
3526 // does NOT declare the method still serves as a dex ref owner for
3527 // the inherited method, so it needs the entry too (same reasoning
3528 // as register_field_rename_with_subs for fields). Ancestor-forced
3529 // displays win; a colliding local entry re-mints off the original
3530 // method name and re-propagates.
3531 {
3532 let mut down: jdc_core::FxHashMap<&str, Vec<&str>> =
3533 jdc_core::FxHashMap::default();
3534 for (sup, children) in subs.iter() {
3535 down.entry(*sup).or_default().extend(children.iter().copied());
3536 }
3537 for n in &pool.order {
3538 if let Some(pc) = pool.get_if_materialized(n) {
3539 for i in pc.interfaces.iter() {
3540 down.entry(i.as_str()).or_default().push(n.as_str());
3541 }
3542 }
3543 }
3544 type PropEntry = (
3545 std::sync::Arc<str>,
3546 std::sync::Arc<str>,
3547 std::sync::Arc<str>,
3548 std::sync::Arc<str>,
3549 );
3550 let mut queue: std::collections::VecDeque<PropEntry> =
3551 std::collections::VecDeque::new();
3552 {
3553 // Sorted seed: `out` is a std HashMap (RandomState) — an
3554 // unsorted seed would make conflicting ancestor forces
3555 // resolve in per-process order (display names flipping
3556 // run-to-run).
3557 let mut seed: Vec<PropEntry> = Vec::new();
3558 for (cls, entries) in out.iter() {
3559 for fr in entries {
3560 if fr.desc.contains('(') {
3561 seed.push((
3562 cls.clone(),
3563 fr.name.clone(),
3564 fr.desc.clone(),
3565 fr.display.clone(),
3566 ));
3567 }
3568 }
3569 }
3570 seed.sort();
3571 for t in seed {
3572 queue.push_back(t);
3573 }
3574 }
3575 let mut guard = 0usize;
3576 while let Some((cls, n, d, disp)) = queue.pop_front() {
3577 guard += 1;
3578 if guard > 2_000_000 {
3579 break;
3580 }
3581 let Some(children) = down.get(cls.as_ref()).cloned() else {
3582 continue;
3583 };
3584 for child in children {
3585 let mut changed = false;
3586 let mut reminted: Vec<PropEntry> = Vec::new();
3587 {
3588 let child_key: std::sync::Arc<str> = std::sync::Arc::from(child);
3589 let entries = out.entry(child_key.clone()).or_default();
3590 // A raw method of the child literally named `disp`
3591 // with a different signature blocks the force —
3592 // renaming that raw method instead is the over-reach
3593 // class (500k blowup history); skip this child.
3594 let raw_blocked = pool
3595 .get_if_materialized(child)
3596 .is_some_and(|pc| {
3597 pc.all_methods().any(|m| {
3598 crate::classdec::java_ident(&m.name) == disp.as_ref()
3599 && m.desc != d
3600 })
3601 });
3602 if raw_blocked {
3603 continue;
3604 }
3605 match entries
3606 .iter_mut()
3607 .find(|fr| *fr.name == *n && *fr.desc == *d)
3608 {
3609 Some(e) if e.display == disp => {}
3610 Some(e) => {
3611 e.display = disp.clone();
3612 changed = true;
3613 }
3614 None => {
3615 entries.push(jdc_core::rename::FieldRename {
3616 name: n.clone(),
3617 desc: d.clone(),
3618 display: disp.clone(),
3619 });
3620 changed = true;
3621 }
3622 }
3623 if changed {
3624 // Display collisions inside the child: the
3625 // ancestor-forced entry keeps `disp`; others
3626 // re-mint from their original method name.
3627 let mut taken: jdc_core::FxHashSet<String> =
3628 jdc_core::FxHashSet::default();
3629 if let Some(pc) = pool.get_if_materialized(child) {
3630 for m in pc.all_methods() {
3631 taken.insert(
3632 crate::classdec::java_ident(&m.name).into_owned(),
3633 );
3634 }
3635 }
3636 for e in entries.iter() {
3637 taken.insert(e.display.to_string());
3638 }
3639 let mut seen: jdc_core::FxHashSet<String> =
3640 jdc_core::FxHashSet::default();
3641 seen.insert(disp.to_string());
3642 for e in entries.iter_mut() {
3643 // FIELD entries never re-mint here: this is
3644 // the METHOD-name propagation, and a field
3645 // display colliding with a forced method
3646 // display is resolved at the field's own mint
3647 // (f_taken seeding above), not by renaming the
3648 // field out from under its subs-registered
3649 // references (bx$b's inherited bv→a2 was
3650 // reminted to a7 while every ref kept a2).
3651 if !e.desc.contains('(') {
3652 continue;
3653 }
3654 if *e.name == *n && *e.desc == *d {
3655 continue;
3656 }
3657 if seen.insert(e.display.to_string()) {
3658 continue;
3659 }
3660 let base =
3661 crate::classdec::java_ident(&e.name).into_owned();
3662 let nd: std::sync::Arc<str> = std::sync::Arc::from(
3663 suffix_unique(&base, &mut taken).as_str(),
3664 );
3665 e.display = nd.clone();
3666 reminted.push((
3667 child_key.clone(),
3668 e.name.clone(),
3669 e.desc.clone(),
3670 nd,
3671 ));
3672 }
3673 }
3674 }
3675 if changed {
3676 let child_key: std::sync::Arc<str> = std::sync::Arc::from(child);
3677 queue.push_back((child_key.clone(), n.clone(), d.clone(), disp.clone()));
3678 for r in reminted {
3679 queue.push_back(r);
3680 }
3681 }
3682 }
3683 }
3684 }
3685 inherited_obscuring_renames(pool, &mut out, &pkg_displays);
3686 out
3687}
3688
3689/// Inherited-field variant of the obscuring rule: class C renders a
3690/// member nested type `d` while INHERITING a field `d` from a pool
3691/// superclass — `C.d.b(..)` binds to the inherited variable (fields
3692/// obscure types, JLS 6.4.2) and fails ("无法从静态上下文中引用非静态
3693/// 变量 d" + "无法取消引用int" pairs, LynxBaseScrollViewDragging ×84 —
3694/// latent behind the broken-superclass hub until the same-class rule
3695/// healed it). The rename lands on the DECLARING class (one declaration
3696/// to re-render) and is registered under every transitive subclass as
3697/// well, because dex field refs may name any subclass as owner and the
3698/// lift-time consult keys on the ref's owner.
3699/// Register a field rename under the declaring class AND every transitive
3700/// subclass: dex field refs for an inherited access may name ANY class in
3701/// the hierarchy as owner (`iput Lh$a;->a` inside the constant-subclass
3702/// ctor of `h`), and the lift-time consult keys on the ref's owner — a
3703/// declaring-class-only entry leaves those refs stale (org/a/c/h ×3).
3704/// Subclasses that REDECLARE (name, desc) hide the field — their own
3705/// field is distinct and must keep its name (and their own obscuring,
3706/// if any, is handled when the loop reaches them).
3707fn register_field_rename_with_subs(
3708 pool: &DexPool,
3709 subs: &jdc_core::FxHashMap<&str, Vec<&str>>,
3710 out: &mut HashMap<std::sync::Arc<str>, Vec<jdc_core::rename::FieldRename>>,
3711 owner: &str,
3712 fname: &str,
3713 fdesc: &str,
3714 display: &str,
3715) {
3716 let mut queue: Vec<&str> = vec![owner];
3717 let mut seen: jdc_core::FxHashSet<&str> = jdc_core::FxHashSet::default();
3718 seen.insert(owner);
3719 let mut first = true;
3720 while let Some(c) = queue.pop() {
3721 let redeclares = if first {
3722 first = false;
3723 false // the declaring class itself: always register
3724 } else {
3725 pool.get_if_materialized(c)
3726 .map(|pc| {
3727 pc.static_fields
3728 .iter()
3729 .chain(pc.instance_fields.iter())
3730 .any(|f| f.name.as_ref() == fname && f.desc.as_ref() == fdesc)
3731 })
3732 .unwrap_or(false)
3733 };
3734 if redeclares {
3735 continue; // hidden below this point: its subtree refs bind to its own field
3736 }
3737 out.entry(std::sync::Arc::from(c))
3738 .or_default()
3739 .push(jdc_core::rename::FieldRename {
3740 name: std::sync::Arc::from(fname),
3741 desc: std::sync::Arc::from(fdesc),
3742 display: std::sync::Arc::from(display),
3743 });
3744 if let Some(children) = subs.get(c) {
3745 for ch in children {
3746 if seen.insert(*ch) {
3747 queue.push(*ch);
3748 }
3749 }
3750 }
3751 }
3752}
3753
3754fn inherited_obscuring_renames(
3755 pool: &DexPool,
3756 out: &mut HashMap<std::sync::Arc<str>, Vec<jdc_core::rename::FieldRename>>,
3757 pkg_displays: &jdc_core::FxHashMap<&str, jdc_core::FxHashSet<String>>,
3758) {
3759 let root_segs = pool.root_pkg_segs();
3760 // Reverse super map over materialized classes.
3761 let mut subs: jdc_core::FxHashMap<&str, Vec<&str>> = jdc_core::FxHashMap::default();
3762 for n in &pool.order {
3763 let Some(pc) = pool.get_if_materialized(n) else {
3764 continue;
3765 };
3766 if let Some(sup) = &pc.super_name {
3767 if sup != "java/lang/Object" {
3768 subs.entry(sup.as_str()).or_default().push(n.as_str());
3769 }
3770 }
3771 }
3772 let mut renamed: jdc_core::FxHashSet<(String, String, String)> =
3773 jdc_core::FxHashSet::default();
3774 for n in &pool.order {
3775 let Some(pc) = pool.get_if_materialized(n) else {
3776 continue;
3777 };
3778 if pc.is_interface() {
3779 continue;
3780 }
3781 // Member-rendered child displays of THIS class.
3782 let mut child_displays: Vec<String> = Vec::new();
3783 for child in pool.children_of(n.as_str()) {
3784 let Some(rest) = child
3785 .strip_prefix(n.as_str())
3786 .and_then(|t| t.strip_prefix('$'))
3787 else {
3788 continue;
3789 };
3790 if !clean_member_tail(rest) {
3791 continue;
3792 }
3793 if crate::find_outer_name(pool, child).as_deref() != Some(n.as_str()) {
3794 continue;
3795 }
3796 let renamed_disp = crate::apply_class_rename(child);
3797 let last = renamed_disp
3798 .rsplit('/')
3799 .next()
3800 .unwrap_or(&renamed_disp)
3801 .rsplit('$')
3802 .next()
3803 .unwrap_or(rest)
3804 .to_string();
3805 child_displays.push(last);
3806 }
3807 if child_displays.is_empty() {
3808 continue;
3809 }
3810 // Walk the super chain for an inherited field whose display
3811 // collides; own fields are the same-class rule's job.
3812 let own: jdc_core::FxHashSet<String> = pc
3813 .static_fields
3814 .iter()
3815 .chain(pc.instance_fields.iter())
3816 .map(|f| crate::classdec::java_ident(&f.name).into_owned())
3817 .collect();
3818 let mut sup = pc.super_name.clone();
3819 while let Some(sname) = sup {
3820 if sname == "java/lang/Object" {
3821 break;
3822 }
3823 let Some(sc) = pool.get_if_materialized(&sname) else {
3824 break;
3825 };
3826 for f in sc.static_fields.iter().chain(sc.instance_fields.iter()) {
3827 let fdisp = crate::classdec::java_ident(&f.name).into_owned();
3828 if fdisp.starts_with("this$") || !child_displays.contains(&fdisp) {
3829 continue;
3830 }
3831 if !renamed.insert((sname.clone(), f.name.to_string(), f.desc.to_string())) {
3832 continue;
3833 }
3834 // Mint in the DECLARING class's namespace: avoid its
3835 // fields + its own member-rendered child displays.
3836 let mut avoid: jdc_core::FxHashSet<String> = sc
3837 .static_fields
3838 .iter()
3839 .chain(sc.instance_fields.iter())
3840 .map(|g| crate::classdec::java_ident(&g.name).into_owned())
3841 .collect();
3842 // The declaring class's PACKAGE class displays (post-
3843 // rename): minting onto one shadows its qualified refs
3844 // from inside the class (the a→a2-onto-display-a2 bug).
3845 if let Some(names) = pkg_displays
3846 .get(sname.rfind('/').map(|i| &sname[..i]).unwrap_or(""))
3847 {
3848 avoid.extend(names.iter().cloned());
3849 }
3850 for child in pool.children_of(sname.as_str()) {
3851 let Some(rest) = child
3852 .strip_prefix(sname.as_str())
3853 .and_then(|t| t.strip_prefix('$'))
3854 else {
3855 continue;
3856 };
3857 if !clean_member_tail(rest) {
3858 continue;
3859 }
3860 let rd = crate::apply_class_rename(child);
3861 avoid.insert(
3862 rd.rsplit('/')
3863 .next()
3864 .unwrap_or(&rd)
3865 .rsplit('$')
3866 .next()
3867 .unwrap_or(rest)
3868 .to_string(),
3869 );
3870 }
3871 let mut k = 1u32;
3872 let display = loop {
3873 k += 1;
3874 let cand = format!("{fdisp}{k}");
3875 // root_segs: see the own-field mint site — a
3876 // package first segment must never be minted.
3877 if !avoid.contains(&cand) && !root_segs.contains(&cand) {
3878 break cand;
3879 }
3880 };
3881 register_field_rename_with_subs(
3882 pool, &subs, out, &sname, &f.name, &f.desc, &display,
3883 );
3884 if std::env::var("DDC_OBSCURE_STATS").is_ok() {
3885 eprintln!("[obscure-inh] {sname} field {} -> {display} (child of {n})", f.name);
3886 }
3887 }
3888 sup = sc.super_name.clone();
3889 }
3890 let _ = &own;
3891 }
3892}
3893
3894fn suffix_unique(base: &str, taken: &mut jdc_core::FxHashSet<String>) -> String {
3895 let mut k = 1u32;
3896 loop {
3897 k += 1;
3898 let cand = format!("{base}{k}");
3899 if taken.insert(cand.clone()) {
3900 return cand;
3901 }
3902 }
3903}
3904
3905/// member_collision_renames + field_deshadow_renames, collision-registry
3906/// entries winning per (owner, name, desc).
3907fn combined_field_renames(
3908 pool: &DexPool,
3909) -> HashMap<std::sync::Arc<str>, Vec<jdc_core::rename::FieldRename>> {
3910 let t = std::time::Instant::now();
3911 let mut base = member_collision_renames(pool);
3912 if std::env::var("DDC_DBG_REN").is_ok() {
3913 eprintln!("[ren] {:?} member_collision_renames", t.elapsed());
3914 }
3915 let t = std::time::Instant::now();
3916 crate::stats_rss(" member_collision_renames");
3917 let extra = field_deshadow_renames(pool);
3918 if std::env::var("DDC_DBG_REN").is_ok() {
3919 eprintln!("[ren] {:?} field_deshadow_renames", t.elapsed());
3920 }
3921 let mut merged = 0usize;
3922 for (owner, v) in extra {
3923 let e = base.entry(owner).or_default();
3924 for fr in v {
3925 if !e.iter().any(|x| x.name == fr.name && x.desc == fr.desc) {
3926 e.push(fr);
3927 merged += 1;
3928 }
3929 }
3930 }
3931 if std::env::var("DDC_STATS").is_ok() {
3932 eprintln!("[renames] field-deshadow renames: {merged}");
3933 }
3934 crate::stats_rss(" field_deshadow merged");
3935 base
3936}
3937
3938/// Field deshadow (JLS 6.4.2, expression position): a FIELD whose display
3939/// name equals a root package first segment captures package-qualified
3940/// renders inside its family's file — androidx z2's field `j` binds the
3941/// `j` of `j.a.a` (找不到符号 变量 a). Rename through the field registry
3942/// (declaration and every reference funnel through field_display).
3943/// Gate: only when the family actually references that segment
3944/// (family_ref_segments body scan) or the segment is the family's OWN
3945/// package root (same-package refs fall back to `root.x` FQN renders) —
3946/// ungated field renames are the six-times-falsified blast radius.
3947fn field_deshadow_renames(
3948 pool: &DexPool,
3949) -> HashMap<std::sync::Arc<str>, Vec<jdc_core::rename::FieldRename>> {
3950 let mut out: HashMap<std::sync::Arc<str>, Vec<jdc_core::rename::FieldRename>> =
3951 HashMap::default();
3952 let root_segs = pool.root_pkg_segs();
3953 if root_segs.is_empty() {
3954 return out;
3955 }
3956 // Candidate fields per family (display, original name, desc).
3957 let mut cand: HashMap<
3958 String,
3959 Vec<(String, String, String, u32)>,
3960 > = HashMap::default();
3961 for name in &pool.order {
3962 let Some(pc) = pool.get_if_materialized(name) else {
3963 continue;
3964 };
3965 for f in pc.static_fields.iter().chain(pc.instance_fields.iter()) {
3966 let disp = crate::classdec::java_ident(&f.name).into_owned();
3967 if root_segs.contains(&disp) {
3968 cand
3969 .entry(name.clone())
3970 .or_default()
3971 .push((disp, f.name.to_string(), f.desc.to_string(), f.access));
3972 }
3973 }
3974 }
3975 if cand.is_empty() {
3976 return out;
3977 }
3978 let cand_set: jdc_core::FxHashSet<String> = cand.keys().cloned().collect();
3979 let t0 = std::time::Instant::now();
3980 let (fam_segs, fam_nest) =
3981 crate::refscan::family_ref_segments(&pool.dexes, &cand_set);
3982 if std::env::var("DDC_STATS").is_ok() {
3983 eprintln!(
3984 "[renames] field-deshadow scan: cands={} scan={:?}",
3985 cand_set.len(),
3986 t0.elapsed()
3987 );
3988 }
3989 for (fam, fields) in cand {
3990 let own_root = fam.find('/').map(|i| fam[..i].to_string());
3991 let scanned = fam_segs.get(&fam);
3992 // NO nested-tail collision trigger here: a private field named
3993 // like a nested class does capture `Fam.b.member` in expression
3994 // position (lark LKEvent static-ctx family, ~1k lines), but
3995 // renaming it is corpus-toxic even gated private+body-
3996 // referenced — lark +32k (43,643 renames), weixin/weibo −200.
3997 // Eighth falsification of loose rename gates: the resolution
3998 // perturbation surface of mass field renames dwarfs the target
3999 // family. The fam_nest scan stays wired for a future
4000 // render-side fix (the capture is only unfixable INSIDE the
4001 // family file; foreign files could import the nested type).
4002 let _ = &fam_nest;
4003 let hits: Vec<(String, String, String)> = fields
4004 .into_iter()
4005 .filter(|(disp, _, _, _access)| {
4006 scanned.is_some_and(|s| s.contains(disp))
4007 || own_root.as_deref() == Some(disp.as_str())
4008 })
4009 .map(|(d, n, de, _)| (d, n, de))
4010 .collect();
4011 if hits.is_empty() {
4012 continue;
4013 }
4014 // Names already taken in the family's member scope: field and
4015 // method displays plus direct nested-class tails (member types
4016 // share the lexical scope of the rendered file). Built only for
4017 // hit families — the per-family setup must stay off the 175k-
4018 // candidate loop (an O(pool.order) nested scan there cost 60s on
4019 // weixin).
4020 let mut taken: jdc_core::FxHashSet<String> = jdc_core::FxHashSet::default();
4021 if let Some(pc) = pool.get_if_materialized(&fam) {
4022 for f in pc.static_fields.iter().chain(pc.instance_fields.iter()) {
4023 taken.insert(crate::classdec::java_ident(&f.name).into_owned());
4024 }
4025 for m in pc.direct_methods.iter().chain(pc.virtual_methods.iter()) {
4026 taken.insert(crate::classdec::java_ident(&m.name).into_owned());
4027 }
4028 }
4029 let mut nested_tails: jdc_core::FxHashSet<String> =
4030 jdc_core::FxHashSet::default();
4031 for child in pool.children_of(&fam) {
4032 if let Some(tail) = child.rsplit('$').next() {
4033 if !tail.is_empty() {
4034 taken.insert(tail.to_string());
4035 nested_tails.insert(tail.to_string());
4036 }
4037 }
4038 }
4039 for (disp, name, desc) in hits {
4040 let mut k = 0u32;
4041 let new_disp = loop {
4042 k += 1;
4043 let c = if k == 1 {
4044 format!("{disp}x")
4045 } else {
4046 format!("{disp}x{k}")
4047 };
4048 if !taken.contains(&c) && !root_segs.contains(&c) {
4049 break c;
4050 }
4051 };
4052 taken.insert(new_disp.clone());
4053 out.entry(std::sync::Arc::from(fam.as_str()))
4054 .or_default()
4055 .push(jdc_core::rename::FieldRename {
4056 name: std::sync::Arc::from(name.as_str()),
4057 desc: std::sync::Arc::from(desc.as_str()),
4058 display: std::sync::Arc::from(new_disp.as_str()),
4059 });
4060 }
4061 }
4062 out
4063}
4064pub use jdc_core::rename::apply_class_rename;
4065
4066// ---------------------------------------------------------------------------
4067// Default-package relocation.
4068// ---------------------------------------------------------------------------
4069
4070/// The synthetic package the default-package cohort currently relocates
4071/// to (Mutex, not OnceLock: `set_class_renames` replaces the registry per
4072/// pool, so a second pool in one process must be able to re-key this).
4073static ROOT_PKG: std::sync::Mutex<Option<String>> = std::sync::Mutex::new(None);
4074
4075/// The synthetic package display root-package classes were moved under
4076/// (`None` when no relocation ran). Consumers keying per-package caches
4077/// by the ORIGINAL package map this display back to `""`.
4078pub fn root_pkg_display() -> Option<String> {
4079 ROOT_PKG
4080 .lock()
4081 .unwrap_or_else(|e| e.into_inner())
4082 .clone()
4083}
4084
4085/// Default-package relocation: javac cannot resolve ANY reference from a
4086/// named package to an unnamed-package class (JLS 7.4.2 — `private it0 f;`
4087/// inside com.amap.X is 找不到符号 whatever the syntax; xingye ~198k +
4088/// Amap ~81k errors ≈ 73% of the whole-directory sweep total, and
4089/// kotlin.jvm.Function0 even EXTENDS a root class on xingye). No source
4090/// syntax escapes it — but the rename registry escapes the NAME: move the
4091/// whole root cohort under a synthetic package (`ddcroot/a.java`,
4092/// `package ddcroot;`) and every cross-package ref renders the resolvable
4093/// FQN `ddcroot.a`. Cohort-uniform: in-cohort refs keep resolving (one
4094/// new shared package), the case/lossy dedupes inside the cohort still
4095/// hold (a directory move preserves relative collisions), and the
4096/// synthetic segment is collision-bumped against every pool package and
4097/// class so it can never shadow or be shadowed. Runs LAST (prefixes the
4098/// settled displays). Fidelity note: declared package no longer matches
4099/// the binary name — the same trade every rename rule here makes, on a
4100/// cohort whose cross-package render is otherwise dead code.
4101/// Pool classes that shadow a FRAMEWORK class (d8 desugar stubs at the
4102/// framework FQN). The FILE identity relocates to the synthetic package,
4103/// but every REFERENCE renders the raw framework name — the bytecode's
4104/// FQN means the platform class (the stub is dead code the boot
4105/// classpath outranks).
4106pub(crate) static FW_SHADOW: std::sync::OnceLock<
4107 std::sync::Arc<jdc_core::FxHashSet<String>>,
4108> = std::sync::OnceLock::new();
4109
4110/// True when `internal` is a framework-shadow stub (fast empty-set
4111/// check when no corpus stubs exist).
4112pub fn is_fw_shadow(internal: &str) -> bool {
4113 let Some(s) = FW_SHADOW.get() else { return false };
4114 // Exact hit, or any $-chain PREFIX hits: a nested FRAMEWORK member
4115 // of a stub outer (`ImageDecoder$Source` with the stub
4116 // `ImageDecoder` in the pool) renders through the $-walk of the
4117 // stub's map entry — it must keep the raw framework name too.
4118 if s.contains(internal) {
4119 return true;
4120 }
4121 let mut idx = internal.len();
4122 while let Some(i) = internal[..idx].rfind('$') {
4123 if s.contains(&internal[..i]) {
4124 return true;
4125 }
4126 idx = i;
4127 }
4128 false
4129}
4130
4131/// d8's library/API desugaring embeds STUB classes under the framework's
4132/// own names (`android.graphics.Insets`, `android.app.Person` — minimal
4133/// shapes the runtime never loads, the boot classpath wins). Emitted at
4134/// their FQN they SHADOW the real framework class for every OTHER file
4135/// at compile time: javac resolves `android.graphics.Insets` to the
4136/// field-less stub and `insets.top` dies (reqable r2: the whole Insets
4137/// family). Relocate every pool class whose head resolves in the
4138/// embedded framework DB to a synthetic package; the rename registry
4139/// carries every reference.
4140fn fw_shadow_relocation(pool: &DexPool, map: &mut HashMap<String, String>) {
4141 let root_segs = pool.root_pkg_segs();
4142 // ANDROID-namespace only: a framework-shadow stub in another
4143 // namespace is usually a REAL bundled implementation (weibo ships
4144 // its own full org.json — relocating it broke every org.json user,
4145 // +86). d8's API-desugar stubs live under android/.
4146 let shadow_set: std::sync::Arc<jdc_core::FxHashSet<String>> =
4147 std::sync::Arc::new(
4148 pool.order
4149 .iter()
4150 .filter(|n| n.starts_with("android/"))
4151 .filter(|n| {
4152 let head = n.split('$').next().unwrap_or(n);
4153 crate::fwdb::find(head).is_some()
4154 })
4155 .map(|n| n.to_string())
4156 .collect(),
4157 );
4158 let _ = FW_SHADOW.set(shadow_set.clone());
4159 jdc_core::rename::set_fw_shadow(
4160 shadow_set.iter().cloned().collect(),
4161 );
4162 let mut pkg = "ddcfwstub".to_string();
4163 while pool.has_name(&pkg) || root_segs.contains(&pkg) {
4164 pkg.push('_');
4165 }
4166 let mut moved = 0usize;
4167 for name in &pool.order {
4168 // The relocation decision keys on the $-chain HEAD (a nested
4169 // stub rides its outer's identity); nested-of-pool-outer shapes
4170 // stay entry-less so apply_class_rename's $-walk composes them
4171 // through the relocated outer (the VasUIToken exact-entry
4172 // lesson from root_pkg_relocation). Android-namespace only —
4173 // see the set build above.
4174 if !name.starts_with("android/") {
4175 continue;
4176 }
4177 let head = name.split('$').next().unwrap_or(name);
4178 if crate::fwdb::find(head).is_none() {
4179 continue;
4180 }
4181 if head != name && pool.has_name(head) {
4182 continue;
4183 }
4184 let disp = map.get(name).cloned().unwrap_or_else(|| name.clone());
4185 if disp != *name {
4186 // An earlier rule renamed this class (obscuring/collision
4187 // family) — its display already diverges from the FQN and
4188 // the relocation prefix applies to the settled display.
4189 map.insert(name.clone(), format!("{pkg}/{disp}"));
4190 } else {
4191 map.insert(name.clone(), format!("{pkg}/{name}"));
4192 }
4193 moved += 1;
4194 }
4195 if std::env::var("DDC_STATS").is_ok() {
4196 eprintln!("[renames] fw-shadow relocation: {moved} classes -> {pkg}");
4197 }
4198}
4199
4200fn root_pkg_relocation(pool: &DexPool, map: &mut HashMap<String, String>) {
4201 // Gate on NAMES only (identical verdict in lazy browse and full
4202 // decompile — the display must not depend on materialization):
4203 // relocation matters only when BOTH cohorts exist — the JLS-dead
4204 // direction is named-package → root-package, so an all-root app
4205 // (hello fixtures, pure-demo dexes) keeps the faithful layout.
4206 let has_root = pool.order.iter().any(|n| !n.contains('/'));
4207 let has_named = pool.order.iter().any(|n| n.contains('/'));
4208 if !has_root || !has_named {
4209 *ROOT_PKG.lock().unwrap_or_else(|e| e.into_inner()) = None;
4210 return;
4211 }
4212 let root_segs = pool.root_pkg_segs();
4213 let mut pkg = "ddcroot".to_string();
4214 while pool.has_name(&pkg) || root_segs.contains(&pkg) {
4215 pkg.push('_');
4216 }
4217 // A root class whose simple name IS a top-level package segment
4218 // (`com`, `org`, ...) cannot simply MOVE into the synthetic
4219 // package: a same-package type shadows the package name in every
4220 // qualified ref from every cohort file (JLS 6.4.1 — rimet's root
4221 // `com`/`org` hijacked 235k `com.alibaba...` refs as 类 alibaba
4222 // 位置: 类 com). Those get a bumped first segment (`com` →
4223 // ddcroot/com2) — a display rename the registry carries through
4224 // every ref path, like the obscuring renames.
4225 const FW_SEGS: [&str; 10] = [
4226 "java", "javax", "android", "org", "junit", "dalvik", "libcore", "androidx", "kotlin",
4227 "kotlinx",
4228 ];
4229 let mut taken: jdc_core::FxHashSet<String> = jdc_core::FxHashSet::default();
4230 for name in &pool.order {
4231 if name.contains('/') {
4232 continue;
4233 }
4234 let disp = map.get(name).cloned().unwrap_or_else(|| name.clone());
4235 taken.insert(format!("{pkg}/{disp}"));
4236 }
4237 let mut moved = 0usize;
4238 let mut bumped = 0usize;
4239 for name in &pool.order {
4240 if name.contains('/') {
4241 continue;
4242 }
4243 let disp = map.get(name).cloned().unwrap_or_else(|| name.clone());
4244 if disp.contains('/') {
4245 continue; // defensive: some rule already packaged it
4246 }
4247 if disp == *name && name.contains('$') && find_outer_name(pool, name).is_some() {
4248 // An UNRENAMED nested name with a pool outer must stay
4249 // entry-less so apply_class_rename's $-walk composes it
4250 // through the outer's relocated+renamed display — an exact
4251 // entry short-circuits the walk and freezes the outer's
4252 // pre-rename simple (xingye's case-collision quads:
4253 // `ue5$a` exact-mapped to ddcroot/ue5$a instead of
4254 // ddcroot/ue5_4$a — 程序包ddcroot.ue5不存在 ×61k).
4255 // The case-collision machinery pre-inserts an IDENTITY
4256 // entry for every emission-set name, which short-circuits
4257 // the same walk and froze `$`-tails in the ROOT package
4258 // while their outer moved to ddcroot — family split: QQ's
4259 // VasUIToken$1..$1xx husks stayed at tree root, and every
4260 // `VasUIToken$N.INSTANCE` ref from ddcroot/VasUIToken.java
4261 // died (找不到符号 变量 VasUIToken$N ×202 in that one file).
4262 // Remove the pin; the walk composes through the outer.
4263 map.remove(name);
4264 continue;
4265 }
4266 let head = disp.split('$').next().unwrap_or(&disp);
4267 let collide = root_segs.contains(head) || FW_SEGS.contains(&head);
4268 if !collide {
4269 map.insert(name.clone(), format!("{pkg}/{disp}"));
4270 moved += 1;
4271 continue;
4272 }
4273 let rest = &disp[head.len()..];
4274 let mut k = 1u32;
4275 loop {
4276 k += 1;
4277 let nh = format!("{head}{k}");
4278 if root_segs.contains(&nh) || FW_SEGS.contains(&nh.as_str()) {
4279 continue;
4280 }
4281 let cand = format!("{pkg}/{nh}{rest}");
4282 if taken.contains(&cand) {
4283 continue;
4284 }
4285 taken.insert(cand.clone());
4286 map.insert(name.clone(), cand);
4287 moved += 1;
4288 bumped += 1;
4289 break;
4290 }
4291 }
4292 *ROOT_PKG.lock().unwrap_or_else(|e| e.into_inner()) = Some(pkg.clone());
4293 if std::env::var("DDC_STATS").is_ok() {
4294 eprintln!("[renames] root-pkg relocation: {moved} classes -> {pkg}/ ({bumped} segment-bumped)");
4295 }
4296}