fix: P0 output-correctness & coverage-recovery batch (content, byte bounds, parser linking)

[jlucaso1]

Addresses the five P0 findings from the multi-dimension analysis: the two output-correctness bugs (the IR/reference misrepresenting the wire) and the three already-extractable-but-unlinked typing recoveries. Each is a self-contained commit; the committed generated/ artifacts are regenerated per change and every emitted value was cross-checked against the raw bundle.

Enabling fix (prerequisite)

• fix(scan): deterministic byteLengthSource — the content-length index recorded the first module seen to pin a (parentTag, tag, length), so provenance depended on bundle read order — a live violation of the byte-identical-output guarantee (regenerating flipped six values between WAWebDigestKeyJob/WAWebRetryRequestParser). Now tie-broken by lexicographically-smallest module name. This makes the committed artifacts reproducible from the cached bundles, which the coverage changes below rely on for clean regeneration. Guarded by a new order-independence unit test.

P0 · Output-correctness bugs

• fix(codegen): emit WapChildNode.content in IQ request builders — the emitter never read child.content, so every leaf carrying a byte payload built empty, making the 20 key-material builders (prekey upload, retry receipts, link-code pairing) structurally valid but unusable on the wire. Now a compile-time constant (const_bytes, e.g. the 00 nonce) emits .bytes(vec[0x00]), and a caller-supplied buffer threads a Vec<u8> spec field as .bytes(self.<tag>_content.clone()). Verified the emitted .bytes(...) matches the real wacore_binary::builder::NodeBuilder API. Generated reference Rust only (gitignored); the committed IR is unchanged.

• feat(iq): preserve ranged contentBytesRange bounds as byteMin/byteMax — a true range (min != max, e.g. a media buffer 1..1048576, a token 1..128) was silently dropped; only the fixed min == max case survived as byteLength. The IR carried intMin/intMax for integers but nothing for byte ranges. Added byteMin/byteMax to ParsedField (mirroring the int bounds) and threaded a byte_range through the smax classifier. Regenerated: 108 ranged byte fields across iq/srvreq now carry bounds — every one cross-checked against a real contentBytesRange site in the bundle. Purely additive.

P0 · Coverage recovery (already-extracted, unlinked typing)

• fix(scan): recover variable-named WADeprecatedWapParser — a parser whose name is hoisted into a variable (d = "mexNotificationParser"; new WADeprecatedWapParser(d, fn), as WAWebHandleMexNotification does) was dropped for lack of an inline string name, so the type="mex" notification carried no typed content. Now resolves the name from the module's var name = "literal" binding (requiring a unique binding). notif degraded 9→8, typedContent 18→19.

• feat(iq): link legacy responses from sibling parser modules at the send site — a job that hands a sibling module's parser to deprecatedSendIq(iq, o("Mod").parser) rather than constructing one inline lost its typed response. A new send-site index resolves the referenced parser (only when the target defines a single one, so an untracked export name is never guessed) and attaches it to any request left unknown, keyed by the sending module. Recovers WAWebQueryMediaConnsJob's w:m media-connection response; iq degradedResponses 19→18, typedResponses 140→141. The cross-module gdpr case (send site in a different module from the <iq> builder) is left for a future change.

• feat(notif): recover degraded content from srvreq read-shapes — several server-initiated notification types are parsed via the smax receive-RPC path, so their field tree lives in a WASmaxIn<X>...Request module (the srvreq domain) rather than an inline handler parser. Added a Phase 2b fallback matching the srvreq notification read-shape whose parser asserts that type, keeping only types with a single unambiguous shape (hosted/link_code_companion_reg have two shapes each and stay degraded). Recovers crsc_continuation, passkey_prologue_request, waffle; notif degraded 8→5, typedContent 19→22.

Net effect

Diagnostic	Before	After
iq degradedResponses	19	18
iq typedResponses	140	141
notif degraded	9	5
notif typedContent	18	22

Verification

Full CI-equivalent suite green locally: fmt --check, clippy --workspace --all-targets and --all-features (both -D warnings), cargo test --workspace, schema self-tests, wa-ir/wa-fetch WASM checks, whatspec --no-default-features build, C-free tree check, and Python schema validation (11 domains). New unit tests cover each change. Determinism re-verified: regenerating generated/ from the cached bundles is byte-identical to the committed output. generated/ changes are purely additive except the six intended deterministic byteLengthSource values and the one recovered iq response replacing its unknown placeholder.

🤖 Generated with Claude Code

https://claude.ai/code/session_013111jePsSca7epxMugRn2A

---

Generated by Claude Code

---

Summary by cubic

Fixes P0 wire correctness by emitting byte content in IQ builders and preserving byte range limits, and recovers missing typed responses/notifications. Makes byteLengthSource deterministic for reproducible builds. Net impact: iq typedResponses +1 (140→141), iq degradedResponses −1 (19→18); notif typedContent +4 (18→22), notif degraded −4 (9→5).

• Bug Fixes

  • IQ builders now emit leaf node content: supports constant bytes (e.g. 00) and caller-supplied buffers via .bytes(...).
  • byteLengthSource provenance is order-independent (ties broken by smallest module name) to ensure byte-identical regeneration.
  • Ranged contentBytesRange is preserved as byteMin/byteMax in IR and schemas (additive; no breaking changes).
  • Robustness: safe const_bytes decoding; no fallthrough to builder fields on decode failure; only rename the terminal _node in content var names; attach a send-site parser only when a module has a single unknown stanza; match only the member-form o("WADeprecatedSendIq").deprecatedSendIq(...) send sites with require("WADeprecatedSendIq") provenance (bare calls are intentionally not matched). No output changes.

• Coverage

  • Recovers variable-named WADeprecatedWapParser instances, fixing the type="mex" notification envelope.
  • Links legacy responses referenced at the send site (e.g. media-connection response in WAWebQueryMediaConnsJob).
  • Restores degraded notification content by matching unique srvreq read-shapes (crsc_continuation, passkey_prologue_request, waffle).

^{Written for commit 63aa321. Summary will update on new commits.}

Summary by CodeRabbit

• New Features

  • Improved recovery of message and notification parsing so more items now show structured content instead of “unknown” or degraded data.
  • Added support for byte-field ranges, including minimum and maximum length constraints, in parsed results.
  • Byte payloads are now generated more reliably for leaf nodes, including both fixed and user-provided values.

• Bug Fixes

  • Fixed ambiguous parser and notification cases to avoid guessing when multiple matches exist.
  • Made byte-length source selection consistent regardless of bundle or module ordering.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, add credits to your account and enable them for code reviews in your settings.

[coderabbitai]

📝 Walkthrough

Walkthrough

This PR adds byte-payload emission for request node builders, introduces byte-range constraints (byte_min/byte_max) on ParsedField, adds Phase 2b srvreq-based notification content recovery, makes content-length source selection deterministic, and adds send-site parser resolution wired into IQ scanning.

Changes

Request Node Byte Payload Emission

Layer / File(s)	Summary
Emit const/dynamic byte content crates/wa-codegen/src/emit.rs	emit_child_builder calls new emit_node_content and decode_hex helpers to emit literal .bytes(vec![...]) for constant hex payloads or thread caller-supplied bytes through a generated spec field, with tests for both paths.

Byte-Range Constraint Modeling

Layer / File(s)	Summary
ParsedField schema and computation crates/wa-ir/src/iq.rs, crates/wa-scan/src/response_smax.rs	ParsedField gains optional byte_min/byte_max; Binding::Field gains byte_range; content_byte_length now returns both fixed length and range bounds from contentBytesRange.
Materialization and tests crates/wa-scan/src/response_smax.rs	Field collection threads byte_range into byte_min/byte_max; tests cover fixed, ranged, plain, and optional contentBytesRange cases.

Srvreq-Based Notification Content Recovery

Layer / File(s)	Summary
Phase 2b recovery logic crates/wa-notif/src/lib.rs	New srvreq_notification_shapes matches degraded notifications against unambiguous server-request shapes by asserted type, filling content when resolved.
Fixtures and tests crates/wa-notif/src/tests.rs	Adds SRVREQ_LINK_BUNDLE fixture and tests validating content recovery and continued degradation for ambiguous types.

Deterministic Content-Length Provenance

Layer / File(s)	Summary
Order-independent source selection crates/wa-scan/src/content_length_index.rs	build_pass now keeps the lexicographically-smallest module name as byteLengthSource, verified by a new order-independence test.

Send-Site Parser Resolution and IQ Scan Integration

Layer / File(s)	Summary
send_parser_index pass crates/wa-scan/src/send_parser_index.rs	New build_pass, find_send_parser_ref, and SendFinder resolve unambiguous typed parsers for modules calling deprecatedSendIq(iq, o("Mod").parser), with tests.
IQ scan wiring crates/wa-scan/src/lib.rs	scan_iq_with_diagnostics builds send_parsers and replaces "unknown" response parsers with resolved entries.
Variable-named parser recovery crates/wa-scan/src/response.rs	Adds resolve_parser_name/resolve_string_binding/StringBindingFinder to recover hoisted variable-bound legacy parser names, refusing ambiguous bindings, with tests.

Estimated code review effort: 3 (Moderate) | ~30 minutes

Sequence Diagram(s)

sequenceDiagram
  participant ScanPipeline as scan_iq_with_diagnostics
  participant SendParserIndex as send_parser_index::build_pass
  participant ResponseResolver as resolve_parser_name
  participant Stanzas as IQ Stanzas

  ScanPipeline->>SendParserIndex: build send-site parser map
  SendParserIndex->>ResponseResolver: resolve legacy parser name (literal or variable)
  ResponseResolver-->>SendParserIndex: resolved parser name or None
  SendParserIndex-->>ScanPipeline: module -> ParsedResponse map
  ScanPipeline->>Stanzas: replace "unknown" response/parser_name with resolved entry

Loading

Suggested reviewers: greptile-apps, cubic-dev-ai

Poem

A rabbit hops through bytes and range,
Nibbling parsers, watching them change,
Waffle notifications no longer stray,
Deterministic modules light the way,
Thump-thump, ship it — hooray! 🐇✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately summarizes the main fixes: content emission, byte-bound handling, and parser-linking recovery.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[greptile-apps]

Greptile Summary

This PR fixes five wire-correctness and coverage-recovery bugs: leaf node byte payloads were never emitted in IQ builders, ranged byte bounds were silently dropped, the byteLengthSource provenance depended on bundle read order, variable-named WADeprecatedWapParser constructors were lost, and several notification types whose parsers live in srvreq modules were left degraded.

• emit_node_content (codegen): leaf nodes now emit .bytes(vec![…]) for compile-time constants and thread a Vec<u8> spec field for caller-supplied buffers; uses rfind(\"_node\") to rename only the terminal segment so tag names containing _node are not corrupted.
• byteLengthSource determinism (scan): tie-broken by lexicographically smallest module name instead of first-seen, making regenerated artifacts byte-identical across input orderings.
• byteMin/byteMax in ParsedField (IR + smax): contentBytesRange with min != max is now threaded into the IR as range bounds rather than silently dropped; purely additive with skip_serializing_if.
• Variable-named parser recovery (scan): WADeprecatedWapParser(d, fn) where d is a var binding is now resolved; an ambiguous name (bound to multiple distinct strings anywhere in the module) is rejected rather than guessed.
• Send-site parser index (scan) and srvreq Phase 2b fallback (notif): recover typed responses and notification content that previously stayed unknown/degraded.

Confidence Score: 5/5

All changes are additive or isolated fixes with no breaking schema changes; each fix is independently unit-tested and the regenerated artifacts were verified byte-identical to the committed output.

Every correction is narrowly scoped: the byte-content emitter degrades safely on malformed input, the variable-binding resolver rejects ambiguous names rather than guessing, the send-site index guards against multiple unknowns per module (addressing the prior reviewer concern), and the ranged byte fields are optional in both the IR and the serde schema. No existing field or generated artifact is removed or changed in a breaking way.

No files require special attention. The new send_parser_index.rs is the most novel piece, but its four unit tests plus the outer guard in lib.rs cover the key invariants.

Important Files Changed

Filename	Overview
crates/wa-codegen/src/emit.rs	Adds emit_node_content and decode_hex; uses rfind("_node") (addressing prior reviewer concern); four new unit tests cover constant bytes, dynamic bytes, terminal-segment renaming, and malformed hex degradation.
crates/wa-ir/src/iq.rs	Adds byte_min/byte_max optional fields to ParsedField with serde(default, skip_serializing_if), mirroring the existing int_min/int_max pattern; purely additive, no breaking schema changes.
crates/wa-scan/src/response_smax.rs	Refactors content_byte_length to return (Option<u32>, Option<(u32,u32)>) and threads byte_range through all Binding::Field construction sites; all non-range arms correctly set byte_range: None; updated test extends coverage to range and optional cases.
crates/wa-scan/src/response.rs	Adds resolve_parser_name and StringBindingFinder; the module-wide (non-scope-aware) binding search is intentionally conservative — ambiguous names are rejected rather than guessed, with three unit tests covering recovery, ambiguous, and reuse cases.
crates/wa-scan/src/send_parser_index.rs	New module: resolves o("Mod").parser references at deprecatedSendIq call sites; provenance-checks the callee against WADeprecatedSendIq; only attaches when the target module has exactly one parser; four unit tests cover the happy path, bare-identifier exclusion, missing reference, and ambiguous target.
crates/wa-scan/src/lib.rs	Wires the send-parser index into the IQ scan; explicitly counts unknown stanzas per module and only attaches when exactly one unknown exists per module, addressing the prior reviewer concern.
crates/wa-notif/src/lib.rs	Adds Phase 2b fallback recovering degraded notification content from srvreq read-shapes; short-circuits when no degraded notifications exist; only attaches when a single unambiguous shape asserts the matching type.
crates/wa-scan/src/content_length_index.rs	Fixes tie-breaking to use and_modify + lexicographic minimum instead of first-wins; new unit test asserts that swapping bundle order produces identical provenance output.

_{Reviews (5): Last reviewed commit: "fix(scan): match only the member-form de..." | Re-trigger Greptile}

[coderabbitai]

Actionable comments posted: 4

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@crates/wa-codegen/src/emit.rs`:
- Around line 774-786: The decode_hex helper in emit.rs can panic on non-ASCII
input because it slices with byte indexes before validating UTF-8 boundaries.
Update decode_hex to avoid direct string slicing in the loop and instead iterate
over bytes or validated hex pairs so malformed input always returns None. Keep
the behavior aligned with the doc comment by making malformed const_bytes in
WapContent degrade gracefully rather than triggering a panic during code
generation.
- Around line 726-772: The emit_node_content path incorrectly falls through when
decode_hex fails, causing malformed const_bytes to be treated as caller-supplied
Bytes content and adding an unwanted Vec<u8> field. Update emit_node_content to
keep the const_bytes case isolated: if content.const_bytes is present but
decode_hex returns nothing, return early with no emitted content instead of
entering the WapContentKind::Bytes branch. Make sure the Bytes branch only runs
for true non-const byte content, and use the existing emit_node_content and
WapContentKind::Bytes logic to locate the fix.

In `@crates/wa-notif/src/lib.rs`:
- Around line 126-141: The notification backfill logic in the
`dispatch.notifications` loop uses `&& let`, which is not compatible with the
workspace’s stated minimum Rust version. Rewrite the conditional in the `Phase
2b` block inside `crates/wa-notif/src/lib.rs` using MSRV-safe syntax, or
alternatively bump the crate/workspace `rust-version` if that is the intended
support policy. Keep the behavior of `srvreq_notification_shapes` and the
`n.content` assignment unchanged.

In `@crates/wa-scan/src/response.rs`:
- Around line 64-119: The binding lookup in resolve_string_binding is too broad
because it scans the entire module, so a reused name elsewhere can mislabel the
parser or make it ambiguous. Restrict lookup to the enclosing scope of
resolve_parser_name / the new WADeprecatedWapParser(...) call, or use semantic
scope resolution to bind the identifier to the correct declaration. Update
StringBindingFinder accordingly so it only considers the parser’s actual scope,
and add a regression test where an unrelated inner function reuses the same
variable name.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 2c5df8ea-0752-4f22-bcab-b250c4edf097

📥 Commits

Reviewing files that changed from the base of the PR and between f0cbe0f and a3a6125.

⛔ Files ignored due to path filters (9)

• generated/iq/index.json is excluded by !**/generated/**
• generated/manifest.json is excluded by !**/generated/**
• generated/notif/index.json is excluded by !**/generated/**
• generated/schema/incoming.schema.json is excluded by !**/generated/**
• generated/schema/iq.schema.json is excluded by !**/generated/**
• generated/schema/notif.schema.json is excluded by !**/generated/**
• generated/schema/srvreq.schema.json is excluded by !**/generated/**
• generated/schema/stanza.schema.json is excluded by !**/generated/**
• generated/srvreq/index.json is excluded by !**/generated/**

📒 Files selected for processing (9)

• crates/wa-codegen/src/emit.rs
• crates/wa-ir/src/iq.rs
• crates/wa-notif/src/lib.rs
• crates/wa-notif/src/tests.rs
• crates/wa-scan/src/content_length_index.rs
• crates/wa-scan/src/lib.rs
• crates/wa-scan/src/response.rs
• crates/wa-scan/src/response_smax.rs
• crates/wa-scan/src/send_parser_index.rs

[cubic-dev-ai]

1 issue found and verified against the latest diff

Confidence score: 3/5

• In crates/wa-scan/src/send_parser_index.rs, callee_method(call) only handling member calls means bare deprecatedSendIq(iq, o("Mod").parser) sites are skipped, so legacy response types stay unknown; merging as-is risks incomplete parser indexing and weaker downstream type safety/coverage for those send paths—add identifier-callee matching before merging.

<sub>Reply with feedback, questions, or to request a fix.

Re-trigger cubic</sub>

[cubic-dev-ai]

0 issues found across 3 files (changes from recent commits).

<sub>Requires human review: Auto-approval blocked by 1 unresolved issue from previous reviews.

Re-trigger cubic</sub>

Dismissed because a newer commit was pushed; Greptile will re-review the current head.

[cubic-dev-ai]

1 issue found across 1 file (changes from recent commits).

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="crates/wa-scan/src/send_parser_index.rs">

<violation number="1" location="crates/wa-scan/src/send_parser_index.rs:96">
P2: Modules with a local or shadowed helper named `deprecatedSendIq` can now get an unrelated typed response attached, because the bare-identifier path matches by name without checking the binding came from `WADeprecatedSendIq`. Consider tracking the local alias assignment/import before accepting bare calls, or keep the previous member-only match when the binding is unknown.</violation>
</file>

<sub>Tip: Review your code locally with the cubic CLI to iterate faster.

Re-trigger cubic</sub>

Dismissed because a newer commit was pushed; Greptile will re-review the current head.

[cubic-dev-ai]

1 issue found across 2 files (changes from recent commits).

<sub>Tip: Review your code locally with the cubic CLI to iterate faster.

Re-trigger cubic</sub>

Dismissed because a newer commit was pushed; Greptile will re-review the current head.

[cubic-dev-ai]

0 issues found across 1 file (changes from recent commits).

<sub>Requires human review: Core logic changes to IR schema, codegen, parser linking, and notification recovery; high risk despite being P0 fixes.

Re-trigger cubic</sub>