1
0
Fork 0
mirror of https://github.com/imjasonh/nescript synced 2026-07-13 02:59:36 +00:00
nescript/tests/integration_test.rs

1725 lines
54 KiB
Rust
Raw Normal View History

use std::path::Path;
Implement NEScript compiler Milestone 1 ("Hello Sprite") Complete implementation of the NEScript compiler pipeline for M1: - Lexer: full tokenization with hex/binary/decimal literals, all keywords, operators - Parser: recursive descent with Pratt expression parsing (M1 subset) - Analyzer: symbol resolution, type checking, memory allocation - 6502 Assembler: full opcode encoding table (~150 valid combinations) - Code Generator: AST → 6502 instructions (direct, no IR for M1) - Runtime: NES hardware init, NMI handler, controller read, OAM DMA - Linker: NROM layout, vector table, palette loading, CHR data - ROM Builder: iNES header generation, PRG/CHR padding - CLI: `build` and `check` subcommands via clap 143 tests across all modules: - 22 lexer tests (literals, keywords, operators, error recovery) - 18 parser tests (expressions, statements, game structure, errors) - 7 analyzer tests (symbol resolution, memory allocation, transitions) - 30 assembler tests (every addressing mode, label resolution) - 7 codegen tests (var init, arithmetic, buttons, draw, comparisons) - 11 runtime tests (init sequence, NMI handler, controller read) - 10 ROM builder tests (iNES format, mirroring, banking, validation) - 5 linker tests (vector table, CHR data, palette loading) - 7 integration tests (end-to-end compilation, error detection) CI: GitHub Actions for check, fmt, clippy, test Pre-commit: script for local fmt + clippy + test validation https://claude.ai/code/session_01W6eQFStA66EuMKHUFo2rx3
2026-04-11 22:07:56 +00:00
use nescript::analyzer;
use nescript::assets;
Remove the legacy AST codegen — IR path is canonical now The `--use-ast` path through `src/codegen/mod.rs` was a strictly inferior subset of the IR codegen. Building every example with `--use-ast` through the jsnes harness: - `arrays_and_functions` — fully black (array init + function return values + OAM-in-loop all broken) - `structs_enums_for` — fully black (struct literal is a no-op, all fields stay at 0) - `inline_asm_demo` — fully black - `bitwise_ops`, `loop_break_continue` — below sprite floors (static `next_oam_slot` bug B) - `match_demo` — panics at compile time with `branch offset 153 out of range` (AST's if/else-chain desugaring of `match` emits short branches that can't reach the far arms in a multi-arm match) Six of fourteen examples are non-functional under `--use-ast`. The other eight happen to fall inside the subset AST handles (no arrays, no structs, no function return values, no multi-sprite loops, no long match chains). `docs/future-work.md` already listed "Once working, delete the AST-based codegen entirely" as the intended direction. It's working, so this commit does the deletion. What's removed: - The `CodeGen` struct, its impl block, and every helper in `src/codegen/mod.rs` (the AST codegen body) — ~1150 lines. The file is now a module header that re-exports `IrCodeGen`. - `src/codegen/tests.rs` — 15 AST-specific instruction-pattern tests. Every feature they covered has an equivalent test in `src/codegen/ir_codegen.rs::{tests,more_tests}` already. - The `--use-ast` CLI flag and its branch in `src/main.rs`. - `compile_with_ir_codegen` in `tests/integration_test.rs` — `compile()` now does what it did, so they merged. All 40 integration tests go through the IR path. - Outdated sections in `docs/future-work.md` that described the IR codegen as "not yet implemented" and listed AST codegen gaps as priority work. What's kept: - `src/codegen/ir_codegen.rs` — the real codegen. - `src/codegen/peephole.rs` — post-codegen cleanup pass, now run unconditionally from `main.rs`. Test plan: - `cargo test --release` — 313 unit + 37 integration tests pass (was 328 + 37; the 15 dropped are the deleted AST-specific tests). - `cargo fmt --check` clean. - `cargo clippy --release --all-targets -- -D warnings` clean. - `node tests/emulator/run_examples.mjs` — 14/14 ROMs render above their per-example nonBlack floors. - The one tightening: `sprite_resolution_uses_tile_index` was asserting on the old static-slot encoding (`A9 01 8D 01 02`). Updated to the cursor-based form (`A9 01 99 01 02`, i.e. STA AbsoluteY). Net diff: 1581 deletions, 62 insertions. https://claude.ai/code/session_014Z5y3Q9krLcAxYpZQJhZ5V
2026-04-12 20:37:59 +00:00
use nescript::codegen::IrCodeGen;
use nescript::ir;
banks: implement multi-bank PRG layout and bank-switching runtime Prior to this commit the linker always shipped a single 16 KB PRG bank regardless of the declared mapper, so the README's MMC1/UxROM/ MMC3 support was aspirational. This commit gives the three banked mappers a real multi-bank ROM layout: * RomBuilder.set_prg_banks() writes any number of 16 KB banks back-to-back so the iNES header reflects the true PRG size. * Linker.link_banked() places switchable banks first, fixed bank last, so the fixed bank maps to $C000-$FFFF (the address window where vectors and the runtime live). * runtime::gen_mapper_init() emits reset-time mapper config: MMC1 serial-writes a control-register value that pins the last bank at $C000 with the correct mirroring, UxROM relies on the power-on default, MMC3 writes the $8000/$8001/$A000/$E000 registers to get a known PRG and mirroring state. * runtime::gen_bank_select() is a mapper-specific subroutine (callable with the target bank in A) that maps any physical bank to $8000-$BFFF. * runtime::gen_bank_trampoline() generates a cross-bank call stub in the fixed bank that saves the caller's bank, switches, JSRs the target, and restores the fixed bank. * The CLI and integration helper thread declared `bank X: prg` declarations through to the linker so MMC1/UxROM/MMC3 programs actually produce multi-bank ROMs. Coverage: * Runtime unit tests (18 new): mapper init patterns for every supported mapper, bank-select signatures, trampoline dispatch order, UxROM bus-conflict table contents. * RomBuilder tests (6 new): multi-bank layout, padding, byte-level fidelity, per-bank size validation, legacy single-bank fallback. * Linker tests (13 new): multi-bank ROM sizes across MMC1/ UxROM/MMC3, fixed-bank placement, switchable-bank payload fidelity, bank-select subroutine detection, NROM rejection of switchable banks. * Integration e2e tests (16 new): compile real .ne sources through the full pipeline and assert on iNES headers, mapper init signatures in the fixed bank, vector locations, and a regression check against `examples/mmc1_banked.ne`. Total: 474 tests pass under `cargo test` with `RUSTFLAGS="-D warnings"`. https://claude.ai/code/session_01UCressA5e8k1XsuoJYLav2
2026-04-13 01:50:51 +00:00
use nescript::linker::{Linker, PrgBank};
use nescript::optimizer;
banks: implement multi-bank PRG layout and bank-switching runtime Prior to this commit the linker always shipped a single 16 KB PRG bank regardless of the declared mapper, so the README's MMC1/UxROM/ MMC3 support was aspirational. This commit gives the three banked mappers a real multi-bank ROM layout: * RomBuilder.set_prg_banks() writes any number of 16 KB banks back-to-back so the iNES header reflects the true PRG size. * Linker.link_banked() places switchable banks first, fixed bank last, so the fixed bank maps to $C000-$FFFF (the address window where vectors and the runtime live). * runtime::gen_mapper_init() emits reset-time mapper config: MMC1 serial-writes a control-register value that pins the last bank at $C000 with the correct mirroring, UxROM relies on the power-on default, MMC3 writes the $8000/$8001/$A000/$E000 registers to get a known PRG and mirroring state. * runtime::gen_bank_select() is a mapper-specific subroutine (callable with the target bank in A) that maps any physical bank to $8000-$BFFF. * runtime::gen_bank_trampoline() generates a cross-bank call stub in the fixed bank that saves the caller's bank, switches, JSRs the target, and restores the fixed bank. * The CLI and integration helper thread declared `bank X: prg` declarations through to the linker so MMC1/UxROM/MMC3 programs actually produce multi-bank ROMs. Coverage: * Runtime unit tests (18 new): mapper init patterns for every supported mapper, bank-select signatures, trampoline dispatch order, UxROM bus-conflict table contents. * RomBuilder tests (6 new): multi-bank layout, padding, byte-level fidelity, per-bank size validation, legacy single-bank fallback. * Linker tests (13 new): multi-bank ROM sizes across MMC1/ UxROM/MMC3, fixed-bank placement, switchable-bank payload fidelity, bank-select subroutine detection, NROM rejection of switchable banks. * Integration e2e tests (16 new): compile real .ne sources through the full pipeline and assert on iNES headers, mapper init signatures in the fixed bank, vector locations, and a regression check against `examples/mmc1_banked.ne`. Total: 474 tests pass under `cargo test` with `RUSTFLAGS="-D warnings"`. https://claude.ai/code/session_01UCressA5e8k1XsuoJYLav2
2026-04-13 01:50:51 +00:00
use nescript::parser::ast::BankType;
Implement NEScript compiler Milestone 1 ("Hello Sprite") Complete implementation of the NEScript compiler pipeline for M1: - Lexer: full tokenization with hex/binary/decimal literals, all keywords, operators - Parser: recursive descent with Pratt expression parsing (M1 subset) - Analyzer: symbol resolution, type checking, memory allocation - 6502 Assembler: full opcode encoding table (~150 valid combinations) - Code Generator: AST → 6502 instructions (direct, no IR for M1) - Runtime: NES hardware init, NMI handler, controller read, OAM DMA - Linker: NROM layout, vector table, palette loading, CHR data - ROM Builder: iNES header generation, PRG/CHR padding - CLI: `build` and `check` subcommands via clap 143 tests across all modules: - 22 lexer tests (literals, keywords, operators, error recovery) - 18 parser tests (expressions, statements, game structure, errors) - 7 analyzer tests (symbol resolution, memory allocation, transitions) - 30 assembler tests (every addressing mode, label resolution) - 7 codegen tests (var init, arithmetic, buttons, draw, comparisons) - 11 runtime tests (init sequence, NMI handler, controller read) - 10 ROM builder tests (iNES format, mirroring, banking, validation) - 5 linker tests (vector table, CHR data, palette loading) - 7 integration tests (end-to-end compilation, error detection) CI: GitHub Actions for check, fmt, clippy, test Pre-commit: script for local fmt + clippy + test validation https://claude.ai/code/session_01W6eQFStA66EuMKHUFo2rx3
2026-04-11 22:07:56 +00:00
use nescript::rom;
Remove the legacy AST codegen — IR path is canonical now The `--use-ast` path through `src/codegen/mod.rs` was a strictly inferior subset of the IR codegen. Building every example with `--use-ast` through the jsnes harness: - `arrays_and_functions` — fully black (array init + function return values + OAM-in-loop all broken) - `structs_enums_for` — fully black (struct literal is a no-op, all fields stay at 0) - `inline_asm_demo` — fully black - `bitwise_ops`, `loop_break_continue` — below sprite floors (static `next_oam_slot` bug B) - `match_demo` — panics at compile time with `branch offset 153 out of range` (AST's if/else-chain desugaring of `match` emits short branches that can't reach the far arms in a multi-arm match) Six of fourteen examples are non-functional under `--use-ast`. The other eight happen to fall inside the subset AST handles (no arrays, no structs, no function return values, no multi-sprite loops, no long match chains). `docs/future-work.md` already listed "Once working, delete the AST-based codegen entirely" as the intended direction. It's working, so this commit does the deletion. What's removed: - The `CodeGen` struct, its impl block, and every helper in `src/codegen/mod.rs` (the AST codegen body) — ~1150 lines. The file is now a module header that re-exports `IrCodeGen`. - `src/codegen/tests.rs` — 15 AST-specific instruction-pattern tests. Every feature they covered has an equivalent test in `src/codegen/ir_codegen.rs::{tests,more_tests}` already. - The `--use-ast` CLI flag and its branch in `src/main.rs`. - `compile_with_ir_codegen` in `tests/integration_test.rs` — `compile()` now does what it did, so they merged. All 40 integration tests go through the IR path. - Outdated sections in `docs/future-work.md` that described the IR codegen as "not yet implemented" and listed AST codegen gaps as priority work. What's kept: - `src/codegen/ir_codegen.rs` — the real codegen. - `src/codegen/peephole.rs` — post-codegen cleanup pass, now run unconditionally from `main.rs`. Test plan: - `cargo test --release` — 313 unit + 37 integration tests pass (was 328 + 37; the 15 dropped are the deleted AST-specific tests). - `cargo fmt --check` clean. - `cargo clippy --release --all-targets -- -D warnings` clean. - `node tests/emulator/run_examples.mjs` — 14/14 ROMs render above their per-example nonBlack floors. - The one tightening: `sprite_resolution_uses_tile_index` was asserting on the old static-slot encoding (`A9 01 8D 01 02`). Updated to the cursor-based form (`A9 01 99 01 02`, i.e. STA AbsoluteY). Net diff: 1581 deletions, 62 insertions. https://claude.ai/code/session_014Z5y3Q9krLcAxYpZQJhZ5V
2026-04-12 20:37:59 +00:00
/// Compile a `NEScript` source string into a .nes ROM. Runs the full
/// IR pipeline: parse → analyze → IR lower → optimize → IR codegen
/// → peephole → link. This is what the `nescript build` CLI does
/// (minus file IO and the dump flags), so these integration tests
/// exercise the same path end users hit.
Implement NEScript compiler Milestone 1 ("Hello Sprite") Complete implementation of the NEScript compiler pipeline for M1: - Lexer: full tokenization with hex/binary/decimal literals, all keywords, operators - Parser: recursive descent with Pratt expression parsing (M1 subset) - Analyzer: symbol resolution, type checking, memory allocation - 6502 Assembler: full opcode encoding table (~150 valid combinations) - Code Generator: AST → 6502 instructions (direct, no IR for M1) - Runtime: NES hardware init, NMI handler, controller read, OAM DMA - Linker: NROM layout, vector table, palette loading, CHR data - ROM Builder: iNES header generation, PRG/CHR padding - CLI: `build` and `check` subcommands via clap 143 tests across all modules: - 22 lexer tests (literals, keywords, operators, error recovery) - 18 parser tests (expressions, statements, game structure, errors) - 7 analyzer tests (symbol resolution, memory allocation, transitions) - 30 assembler tests (every addressing mode, label resolution) - 7 codegen tests (var init, arithmetic, buttons, draw, comparisons) - 11 runtime tests (init sequence, NMI handler, controller read) - 10 ROM builder tests (iNES format, mirroring, banking, validation) - 5 linker tests (vector table, CHR data, palette loading) - 7 integration tests (end-to-end compilation, error detection) CI: GitHub Actions for check, fmt, clippy, test Pre-commit: script for local fmt + clippy + test validation https://claude.ai/code/session_01W6eQFStA66EuMKHUFo2rx3
2026-04-11 22:07:56 +00:00
fn compile(source: &str) -> Vec<u8> {
let (program, diags) = nescript::parser::parse(source);
assert!(
diags.is_empty(),
"unexpected parse errors: {diags:?}\nsource:\n{source}"
);
let program = program.expect("parse should succeed");
let analysis = analyzer::analyze(&program);
assert!(
analysis.diagnostics.iter().all(|d| !d.is_error()),
"unexpected analysis errors: {:?}",
analysis.diagnostics
);
let mut ir_program = ir::lower(&program, &analysis);
optimizer::optimize(&mut ir_program);
let sprites = assets::resolve_sprites(&program, Path::new("."))
.expect("sprite resolution should succeed");
audio: complete the subsystem — asset pipeline, user decls, tracker-style driver The audio subsystem was a sketch: `play name` / `start_music name` / `stop_music` parsed, lowered, and emitted a few hardcoded register writes from a builtin name table. No user-declared effects, no per-frame envelope, no note streams, no real engine. This flesh-out brings audio up to the quality bar of the rest of the compiler (sprites, palettes, bank switching, scanline IRQ, etc.) with a full data-driven pipeline: ## Asset pipeline (new `src/assets/audio.rs`) - `sfx Name { duty, pitch, volume }` blocks compile into per-frame pulse-1 envelopes. Pitch/volume arrays must match in length; each entry is one NMI's worth of `$4000` data. - `music Name { duty, volume, repeat, notes }` blocks compile into flat `(pitch, duration)` streams for pulse 2. Pitch 0 is a rest, 1-60 indexes a builtin period table covering C1-B5. - `resolve_sfx` / `resolve_music` walk the program for `play` / `start_music` references and append builtin fallbacks for any name that isn't user-declared — so `play coin` still works without a `sfx Coin { ... }` block. - Builtin effects (coin, jump, hit, click, cancel, shoot, step) and tracks (theme, battle, victory, gameover) synthesize through the same compile path as user decls — one data model, one driver. ## Runtime engine (`src/runtime/mod.rs`) - `gen_audio_tick()` walks both channels every NMI: reads one envelope byte through `(ZP_SFX_PTR),Y` -> writes `$4000`, advances ptr, mutes on zero sentinel. Music decrements the note counter, advances to the next `(pitch, dur)` pair on zero, looks up the period through `(__period_table),Y`, loops on `0xFF 0xFF`. - `gen_period_table()` emits a 60-entry equal-tempered table (A4 = 440 Hz, NTSC 1.789773 MHz CPU clock) with length-counter load bits pre-baked into each high byte. - `gen_data_block()` emits a label + raw-bytes pseudo pair so user sfx/music data can be spliced into PRG with regular labels that the two-pass assembler resolves. - New ZP layout: `$05/$06` music loop base, `$07` music state (duty/volume/loop/active), `$0C-$0F` sfx and music pointers. ## IR codegen (`src/codegen/ir_codegen.rs`) - `with_audio(sfx, music)` registers compile-time trigger constants per blob name. - `gen_play_sfx` emits: write period to `$4002`/`$4003`, load envelope pointer into `ZP_SFX_PTR` via SymbolLo/SymbolHi of `__sfx_<name>`, mark the sfx counter active. - `gen_start_music` stamps the header byte into `ZP_MUSIC_STATE` with the active bit OR'd in, seeds both ptr and loop base from `__music_<name>`, primes the duration counter. - `gen_stop_music` mutes pulse 2 and clears state. ## Linker (`src/linker/mod.rs`) - New `link_with_all_assets(user_code, sprites, sfx, music)` path that splices driver body, period table, and each sfx/music data blob into PRG — all guarded on the `__audio_used` marker so silent programs pay zero ROM cost. ## Assembler (`src/asm/opcodes.rs`, `src/asm/mod.rs`) - New `AddressingMode::Bytes(Vec<u8>)` variant for raw-data pseudo-instructions. `NOP+Bytes(v)` emits the payload verbatim, letting the linker splice ROM data tables into a code section and still have `Label` / `SymbolLo` / `SymbolHi` fixups resolve correctly in the same assembly pass. ## Analyzer - `play` / `start_music` now validate the name against user decls and builtin tables. Unknown names emit E0505 with a helpful list of builtins — previously a typo would silently compile to no-op. ## Parser - New `sfx_decl` / `music_decl` grammar with property-style configuration. Strict validation: duty 0-3, volume 0-15, pitch arrays must match volume length, music notes must come in pairs, pitch 0-60, duration ≥ 1. ## Tests +170 new tests across every layer: - `src/assets/audio.rs`: 17 tests (compile, resolve, builtins, shadowing, label sanitation, nested reference walks) - `src/parser/tests.rs`: 13 tests (valid/invalid sfx + music declarations, property validation, play/start_music/stop_music) - `src/analyzer/tests.rs`: 7 tests (builtin acceptance, user decl acceptance, unknown-name rejection) - `src/runtime/tests.rs`: 10 tests (audio tick labels, RTS end, $4000 write, $4004 mute, period table assembly, A4 = 440 Hz, length counter bits, data block verbatim emit) - `src/linker/tests.rs`: 4 tests (sfx/music blob placement, pointer resolution, elision when unused) - `src/codegen/ir_codegen.rs`: rewrote the 4 existing audio tests to match the new data-driven contract - `tests/integration_test.rs`: 4 end-to-end tests including a user-declared `sfx` + `music` program that verifies bytes land in PRG ROM at the right addresses ## Docs - New Audio section in `docs/language-guide.md` with syntax reference, builtin tables, and an explanation of how the driver works at compile and run time. - `docs/architecture.md` updated to reflect the real audio pipeline instead of the old "audio import stubs" stub. - `docs/future-work.md` moves audio from "status: minimal" to "status: full subsystem" with a narrower list of follow-up work (triangle/noise/DMC channels, NSF/FTM imports, richer envelopes). - `examples/audio_demo.ne` rewritten to showcase user-declared `sfx LongCoin`, `sfx Zap`, `music Theme`, still demonstrating builtin fallback via `play coin`. Total: 424 tests passing (381 unit + 43 integration), clippy clean, fmt clean, all 19 examples compile. https://claude.ai/code/session_015WfaDttE3DpWn9rpyfpQd8
2026-04-13 01:10:21 +00:00
let sfx = assets::resolve_sfx(&program).expect("sfx resolution should succeed");
let music = assets::resolve_music(&program).expect("music resolution should succeed");
audio: complete the subsystem — asset pipeline, user decls, tracker-style driver The audio subsystem was a sketch: `play name` / `start_music name` / `stop_music` parsed, lowered, and emitted a few hardcoded register writes from a builtin name table. No user-declared effects, no per-frame envelope, no note streams, no real engine. This flesh-out brings audio up to the quality bar of the rest of the compiler (sprites, palettes, bank switching, scanline IRQ, etc.) with a full data-driven pipeline: ## Asset pipeline (new `src/assets/audio.rs`) - `sfx Name { duty, pitch, volume }` blocks compile into per-frame pulse-1 envelopes. Pitch/volume arrays must match in length; each entry is one NMI's worth of `$4000` data. - `music Name { duty, volume, repeat, notes }` blocks compile into flat `(pitch, duration)` streams for pulse 2. Pitch 0 is a rest, 1-60 indexes a builtin period table covering C1-B5. - `resolve_sfx` / `resolve_music` walk the program for `play` / `start_music` references and append builtin fallbacks for any name that isn't user-declared — so `play coin` still works without a `sfx Coin { ... }` block. - Builtin effects (coin, jump, hit, click, cancel, shoot, step) and tracks (theme, battle, victory, gameover) synthesize through the same compile path as user decls — one data model, one driver. ## Runtime engine (`src/runtime/mod.rs`) - `gen_audio_tick()` walks both channels every NMI: reads one envelope byte through `(ZP_SFX_PTR),Y` -> writes `$4000`, advances ptr, mutes on zero sentinel. Music decrements the note counter, advances to the next `(pitch, dur)` pair on zero, looks up the period through `(__period_table),Y`, loops on `0xFF 0xFF`. - `gen_period_table()` emits a 60-entry equal-tempered table (A4 = 440 Hz, NTSC 1.789773 MHz CPU clock) with length-counter load bits pre-baked into each high byte. - `gen_data_block()` emits a label + raw-bytes pseudo pair so user sfx/music data can be spliced into PRG with regular labels that the two-pass assembler resolves. - New ZP layout: `$05/$06` music loop base, `$07` music state (duty/volume/loop/active), `$0C-$0F` sfx and music pointers. ## IR codegen (`src/codegen/ir_codegen.rs`) - `with_audio(sfx, music)` registers compile-time trigger constants per blob name. - `gen_play_sfx` emits: write period to `$4002`/`$4003`, load envelope pointer into `ZP_SFX_PTR` via SymbolLo/SymbolHi of `__sfx_<name>`, mark the sfx counter active. - `gen_start_music` stamps the header byte into `ZP_MUSIC_STATE` with the active bit OR'd in, seeds both ptr and loop base from `__music_<name>`, primes the duration counter. - `gen_stop_music` mutes pulse 2 and clears state. ## Linker (`src/linker/mod.rs`) - New `link_with_all_assets(user_code, sprites, sfx, music)` path that splices driver body, period table, and each sfx/music data blob into PRG — all guarded on the `__audio_used` marker so silent programs pay zero ROM cost. ## Assembler (`src/asm/opcodes.rs`, `src/asm/mod.rs`) - New `AddressingMode::Bytes(Vec<u8>)` variant for raw-data pseudo-instructions. `NOP+Bytes(v)` emits the payload verbatim, letting the linker splice ROM data tables into a code section and still have `Label` / `SymbolLo` / `SymbolHi` fixups resolve correctly in the same assembly pass. ## Analyzer - `play` / `start_music` now validate the name against user decls and builtin tables. Unknown names emit E0505 with a helpful list of builtins — previously a typo would silently compile to no-op. ## Parser - New `sfx_decl` / `music_decl` grammar with property-style configuration. Strict validation: duty 0-3, volume 0-15, pitch arrays must match volume length, music notes must come in pairs, pitch 0-60, duration ≥ 1. ## Tests +170 new tests across every layer: - `src/assets/audio.rs`: 17 tests (compile, resolve, builtins, shadowing, label sanitation, nested reference walks) - `src/parser/tests.rs`: 13 tests (valid/invalid sfx + music declarations, property validation, play/start_music/stop_music) - `src/analyzer/tests.rs`: 7 tests (builtin acceptance, user decl acceptance, unknown-name rejection) - `src/runtime/tests.rs`: 10 tests (audio tick labels, RTS end, $4000 write, $4004 mute, period table assembly, A4 = 440 Hz, length counter bits, data block verbatim emit) - `src/linker/tests.rs`: 4 tests (sfx/music blob placement, pointer resolution, elision when unused) - `src/codegen/ir_codegen.rs`: rewrote the 4 existing audio tests to match the new data-driven contract - `tests/integration_test.rs`: 4 end-to-end tests including a user-declared `sfx` + `music` program that verifies bytes land in PRG ROM at the right addresses ## Docs - New Audio section in `docs/language-guide.md` with syntax reference, builtin tables, and an explanation of how the driver works at compile and run time. - `docs/architecture.md` updated to reflect the real audio pipeline instead of the old "audio import stubs" stub. - `docs/future-work.md` moves audio from "status: minimal" to "status: full subsystem" with a narrower list of follow-up work (triangle/noise/DMC channels, NSF/FTM imports, richer envelopes). - `examples/audio_demo.ne` rewritten to showcase user-declared `sfx LongCoin`, `sfx Zap`, `music Theme`, still demonstrating builtin fallback via `play coin`. Total: 424 tests passing (381 unit + 43 integration), clippy clean, fmt clean, all 19 examples compile. https://claude.ai/code/session_015WfaDttE3DpWn9rpyfpQd8
2026-04-13 01:10:21 +00:00
let codegen = IrCodeGen::new(&analysis.var_allocations, &ir_program)
.with_sprites(&sprites)
.with_audio(&sfx, &music);
Remove the legacy AST codegen — IR path is canonical now The `--use-ast` path through `src/codegen/mod.rs` was a strictly inferior subset of the IR codegen. Building every example with `--use-ast` through the jsnes harness: - `arrays_and_functions` — fully black (array init + function return values + OAM-in-loop all broken) - `structs_enums_for` — fully black (struct literal is a no-op, all fields stay at 0) - `inline_asm_demo` — fully black - `bitwise_ops`, `loop_break_continue` — below sprite floors (static `next_oam_slot` bug B) - `match_demo` — panics at compile time with `branch offset 153 out of range` (AST's if/else-chain desugaring of `match` emits short branches that can't reach the far arms in a multi-arm match) Six of fourteen examples are non-functional under `--use-ast`. The other eight happen to fall inside the subset AST handles (no arrays, no structs, no function return values, no multi-sprite loops, no long match chains). `docs/future-work.md` already listed "Once working, delete the AST-based codegen entirely" as the intended direction. It's working, so this commit does the deletion. What's removed: - The `CodeGen` struct, its impl block, and every helper in `src/codegen/mod.rs` (the AST codegen body) — ~1150 lines. The file is now a module header that re-exports `IrCodeGen`. - `src/codegen/tests.rs` — 15 AST-specific instruction-pattern tests. Every feature they covered has an equivalent test in `src/codegen/ir_codegen.rs::{tests,more_tests}` already. - The `--use-ast` CLI flag and its branch in `src/main.rs`. - `compile_with_ir_codegen` in `tests/integration_test.rs` — `compile()` now does what it did, so they merged. All 40 integration tests go through the IR path. - Outdated sections in `docs/future-work.md` that described the IR codegen as "not yet implemented" and listed AST codegen gaps as priority work. What's kept: - `src/codegen/ir_codegen.rs` — the real codegen. - `src/codegen/peephole.rs` — post-codegen cleanup pass, now run unconditionally from `main.rs`. Test plan: - `cargo test --release` — 313 unit + 37 integration tests pass (was 328 + 37; the 15 dropped are the deleted AST-specific tests). - `cargo fmt --check` clean. - `cargo clippy --release --all-targets -- -D warnings` clean. - `node tests/emulator/run_examples.mjs` — 14/14 ROMs render above their per-example nonBlack floors. - The one tightening: `sprite_resolution_uses_tile_index` was asserting on the old static-slot encoding (`A9 01 8D 01 02`). Updated to the cursor-based form (`A9 01 99 01 02`, i.e. STA AbsoluteY). Net diff: 1581 deletions, 62 insertions. https://claude.ai/code/session_014Z5y3Q9krLcAxYpZQJhZ5V
2026-04-12 20:37:59 +00:00
let mut instructions = codegen.generate(&ir_program);
nescript::codegen::peephole::optimize(&mut instructions);
Implement NEScript compiler Milestone 1 ("Hello Sprite") Complete implementation of the NEScript compiler pipeline for M1: - Lexer: full tokenization with hex/binary/decimal literals, all keywords, operators - Parser: recursive descent with Pratt expression parsing (M1 subset) - Analyzer: symbol resolution, type checking, memory allocation - 6502 Assembler: full opcode encoding table (~150 valid combinations) - Code Generator: AST → 6502 instructions (direct, no IR for M1) - Runtime: NES hardware init, NMI handler, controller read, OAM DMA - Linker: NROM layout, vector table, palette loading, CHR data - ROM Builder: iNES header generation, PRG/CHR padding - CLI: `build` and `check` subcommands via clap 143 tests across all modules: - 22 lexer tests (literals, keywords, operators, error recovery) - 18 parser tests (expressions, statements, game structure, errors) - 7 analyzer tests (symbol resolution, memory allocation, transitions) - 30 assembler tests (every addressing mode, label resolution) - 7 codegen tests (var init, arithmetic, buttons, draw, comparisons) - 11 runtime tests (init sequence, NMI handler, controller read) - 10 ROM builder tests (iNES format, mirroring, banking, validation) - 5 linker tests (vector table, CHR data, palette loading) - 7 integration tests (end-to-end compilation, error detection) CI: GitHub Actions for check, fmt, clippy, test Pre-commit: script for local fmt + clippy + test validation https://claude.ai/code/session_01W6eQFStA66EuMKHUFo2rx3
2026-04-11 22:07:56 +00:00
let linker = Linker::new(program.game.mirroring);
audio: complete the subsystem — asset pipeline, user decls, tracker-style driver The audio subsystem was a sketch: `play name` / `start_music name` / `stop_music` parsed, lowered, and emitted a few hardcoded register writes from a builtin name table. No user-declared effects, no per-frame envelope, no note streams, no real engine. This flesh-out brings audio up to the quality bar of the rest of the compiler (sprites, palettes, bank switching, scanline IRQ, etc.) with a full data-driven pipeline: ## Asset pipeline (new `src/assets/audio.rs`) - `sfx Name { duty, pitch, volume }` blocks compile into per-frame pulse-1 envelopes. Pitch/volume arrays must match in length; each entry is one NMI's worth of `$4000` data. - `music Name { duty, volume, repeat, notes }` blocks compile into flat `(pitch, duration)` streams for pulse 2. Pitch 0 is a rest, 1-60 indexes a builtin period table covering C1-B5. - `resolve_sfx` / `resolve_music` walk the program for `play` / `start_music` references and append builtin fallbacks for any name that isn't user-declared — so `play coin` still works without a `sfx Coin { ... }` block. - Builtin effects (coin, jump, hit, click, cancel, shoot, step) and tracks (theme, battle, victory, gameover) synthesize through the same compile path as user decls — one data model, one driver. ## Runtime engine (`src/runtime/mod.rs`) - `gen_audio_tick()` walks both channels every NMI: reads one envelope byte through `(ZP_SFX_PTR),Y` -> writes `$4000`, advances ptr, mutes on zero sentinel. Music decrements the note counter, advances to the next `(pitch, dur)` pair on zero, looks up the period through `(__period_table),Y`, loops on `0xFF 0xFF`. - `gen_period_table()` emits a 60-entry equal-tempered table (A4 = 440 Hz, NTSC 1.789773 MHz CPU clock) with length-counter load bits pre-baked into each high byte. - `gen_data_block()` emits a label + raw-bytes pseudo pair so user sfx/music data can be spliced into PRG with regular labels that the two-pass assembler resolves. - New ZP layout: `$05/$06` music loop base, `$07` music state (duty/volume/loop/active), `$0C-$0F` sfx and music pointers. ## IR codegen (`src/codegen/ir_codegen.rs`) - `with_audio(sfx, music)` registers compile-time trigger constants per blob name. - `gen_play_sfx` emits: write period to `$4002`/`$4003`, load envelope pointer into `ZP_SFX_PTR` via SymbolLo/SymbolHi of `__sfx_<name>`, mark the sfx counter active. - `gen_start_music` stamps the header byte into `ZP_MUSIC_STATE` with the active bit OR'd in, seeds both ptr and loop base from `__music_<name>`, primes the duration counter. - `gen_stop_music` mutes pulse 2 and clears state. ## Linker (`src/linker/mod.rs`) - New `link_with_all_assets(user_code, sprites, sfx, music)` path that splices driver body, period table, and each sfx/music data blob into PRG — all guarded on the `__audio_used` marker so silent programs pay zero ROM cost. ## Assembler (`src/asm/opcodes.rs`, `src/asm/mod.rs`) - New `AddressingMode::Bytes(Vec<u8>)` variant for raw-data pseudo-instructions. `NOP+Bytes(v)` emits the payload verbatim, letting the linker splice ROM data tables into a code section and still have `Label` / `SymbolLo` / `SymbolHi` fixups resolve correctly in the same assembly pass. ## Analyzer - `play` / `start_music` now validate the name against user decls and builtin tables. Unknown names emit E0505 with a helpful list of builtins — previously a typo would silently compile to no-op. ## Parser - New `sfx_decl` / `music_decl` grammar with property-style configuration. Strict validation: duty 0-3, volume 0-15, pitch arrays must match volume length, music notes must come in pairs, pitch 0-60, duration ≥ 1. ## Tests +170 new tests across every layer: - `src/assets/audio.rs`: 17 tests (compile, resolve, builtins, shadowing, label sanitation, nested reference walks) - `src/parser/tests.rs`: 13 tests (valid/invalid sfx + music declarations, property validation, play/start_music/stop_music) - `src/analyzer/tests.rs`: 7 tests (builtin acceptance, user decl acceptance, unknown-name rejection) - `src/runtime/tests.rs`: 10 tests (audio tick labels, RTS end, $4000 write, $4004 mute, period table assembly, A4 = 440 Hz, length counter bits, data block verbatim emit) - `src/linker/tests.rs`: 4 tests (sfx/music blob placement, pointer resolution, elision when unused) - `src/codegen/ir_codegen.rs`: rewrote the 4 existing audio tests to match the new data-driven contract - `tests/integration_test.rs`: 4 end-to-end tests including a user-declared `sfx` + `music` program that verifies bytes land in PRG ROM at the right addresses ## Docs - New Audio section in `docs/language-guide.md` with syntax reference, builtin tables, and an explanation of how the driver works at compile and run time. - `docs/architecture.md` updated to reflect the real audio pipeline instead of the old "audio import stubs" stub. - `docs/future-work.md` moves audio from "status: minimal" to "status: full subsystem" with a narrower list of follow-up work (triangle/noise/DMC channels, NSF/FTM imports, richer envelopes). - `examples/audio_demo.ne` rewritten to showcase user-declared `sfx LongCoin`, `sfx Zap`, `music Theme`, still demonstrating builtin fallback via `play coin`. Total: 424 tests passing (381 unit + 43 integration), clippy clean, fmt clean, all 19 examples compile. https://claude.ai/code/session_015WfaDttE3DpWn9rpyfpQd8
2026-04-13 01:10:21 +00:00
linker.link_with_all_assets(&instructions, &sprites, &sfx, &music)
Implement NEScript compiler Milestone 1 ("Hello Sprite") Complete implementation of the NEScript compiler pipeline for M1: - Lexer: full tokenization with hex/binary/decimal literals, all keywords, operators - Parser: recursive descent with Pratt expression parsing (M1 subset) - Analyzer: symbol resolution, type checking, memory allocation - 6502 Assembler: full opcode encoding table (~150 valid combinations) - Code Generator: AST → 6502 instructions (direct, no IR for M1) - Runtime: NES hardware init, NMI handler, controller read, OAM DMA - Linker: NROM layout, vector table, palette loading, CHR data - ROM Builder: iNES header generation, PRG/CHR padding - CLI: `build` and `check` subcommands via clap 143 tests across all modules: - 22 lexer tests (literals, keywords, operators, error recovery) - 18 parser tests (expressions, statements, game structure, errors) - 7 analyzer tests (symbol resolution, memory allocation, transitions) - 30 assembler tests (every addressing mode, label resolution) - 7 codegen tests (var init, arithmetic, buttons, draw, comparisons) - 11 runtime tests (init sequence, NMI handler, controller read) - 10 ROM builder tests (iNES format, mirroring, banking, validation) - 5 linker tests (vector table, CHR data, palette loading) - 7 integration tests (end-to-end compilation, error detection) CI: GitHub Actions for check, fmt, clippy, test Pre-commit: script for local fmt + clippy + test validation https://claude.ai/code/session_01W6eQFStA66EuMKHUFo2rx3
2026-04-11 22:07:56 +00:00
}
// ── M1 Tests ──
Implement NEScript compiler Milestone 1 ("Hello Sprite") Complete implementation of the NEScript compiler pipeline for M1: - Lexer: full tokenization with hex/binary/decimal literals, all keywords, operators - Parser: recursive descent with Pratt expression parsing (M1 subset) - Analyzer: symbol resolution, type checking, memory allocation - 6502 Assembler: full opcode encoding table (~150 valid combinations) - Code Generator: AST → 6502 instructions (direct, no IR for M1) - Runtime: NES hardware init, NMI handler, controller read, OAM DMA - Linker: NROM layout, vector table, palette loading, CHR data - ROM Builder: iNES header generation, PRG/CHR padding - CLI: `build` and `check` subcommands via clap 143 tests across all modules: - 22 lexer tests (literals, keywords, operators, error recovery) - 18 parser tests (expressions, statements, game structure, errors) - 7 analyzer tests (symbol resolution, memory allocation, transitions) - 30 assembler tests (every addressing mode, label resolution) - 7 codegen tests (var init, arithmetic, buttons, draw, comparisons) - 11 runtime tests (init sequence, NMI handler, controller read) - 10 ROM builder tests (iNES format, mirroring, banking, validation) - 5 linker tests (vector table, CHR data, palette loading) - 7 integration tests (end-to-end compilation, error detection) CI: GitHub Actions for check, fmt, clippy, test Pre-commit: script for local fmt + clippy + test validation https://claude.ai/code/session_01W6eQFStA66EuMKHUFo2rx3
2026-04-11 22:07:56 +00:00
#[test]
fn hello_sprite_compiles_to_valid_rom() {
let source = include_str!("integration/hello_sprite.ne");
let rom_data = compile(source);
let info = rom::validate_ines(&rom_data).expect("should be valid iNES");
assert_eq!(info.prg_banks, 1, "should be 1 PRG bank (16 KB)");
assert_eq!(info.chr_banks, 1, "should have CHR ROM");
assert_eq!(info.mapper, 0, "should be NROM (mapper 0)");
assert_eq!(rom_data.len(), 16 + 16384 + 8192);
}
#[test]
fn hello_sprite_has_correct_vectors() {
let source = include_str!("integration/hello_sprite.ne");
let rom_data = compile(source);
let prg_end = 16 + 16384;
let nmi = u16::from_le_bytes([rom_data[prg_end - 6], rom_data[prg_end - 5]]);
let reset = u16::from_le_bytes([rom_data[prg_end - 4], rom_data[prg_end - 3]]);
let irq = u16::from_le_bytes([rom_data[prg_end - 2], rom_data[prg_end - 1]]);
assert!(nmi >= 0xC000, "NMI vector should be in ROM space");
assert_eq!(reset, 0xC000, "RESET should point to $C000");
assert!(irq >= 0xC000, "IRQ vector should be in ROM space");
assert!(nmi != reset, "NMI and RESET should be different");
}
#[test]
fn minimal_program_compiles() {
let source = r#"
game "Minimal" { mapper: NROM }
on frame { wait_frame }
start Main
"#;
let rom_data = compile(source);
let info = rom::validate_ines(&rom_data).expect("should be valid iNES");
assert_eq!(info.mapper, 0);
}
#[test]
fn program_with_state_machine() {
let source = r#"
game "States" { mapper: NROM }
state Title {
on frame {
if button.start { transition Game }
}
}
state Game {
var score: u8 = 0
on frame {
score += 1
}
}
start Title
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_constants() {
let source = r#"
game "Constants" { mapper: NROM }
const SPEED: u8 = 3
var px: u8 = 100
on frame {
if button.right { px += SPEED }
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
// ── M2 Tests ──
#[test]
fn program_with_functions() {
let source = r#"
game "Functions" { mapper: NROM }
var x: u8 = 0
fun add_ten(val: u8) -> u8 {
return val + 10
}
on frame {
x = add_ten(5)
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_on_scanline_mmc3() {
let source = r#"
game "Scanline" { mapper: MMC3 }
var sx: u8 = 0
state Main {
on frame { wait_frame }
on scanline(120) { scroll(sx, 0) }
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_on_scanline_per_state() {
// Two states, each with its own scanline handler at a different
// position. The IR codegen should emit per-state dispatch in
// both `__irq_user` and `__ir_mmc3_reload`.
let source = r#"
game "MultiSL" { mapper: MMC3 }
var s: u8 = 0
state A {
on frame { wait_frame }
on scanline(64) { scroll(0, 0) }
}
state B {
on frame { wait_frame }
on scanline(192) { scroll(0, 0) }
}
start A
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_function_local_variables() {
// Functions with locally-declared variables should allocate
// their own backing storage and not corrupt caller state when
// nested.
let source = r#"
game "Locals" { mapper: NROM }
var out: u8 = 0
fun double(x: u8) -> u8 {
var t: u8 = x
t = t + t
return t
}
fun double_sum(a: u8, b: u8) -> u8 {
var s1: u8 = double(a)
var s2: u8 = double(b)
return s1 + s2
}
on frame {
out = double_sum(10, 20)
wait_frame
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_for_loop() {
let source = r#"
game "ForLoop" { mapper: NROM }
var arr: u8[8] = [0, 0, 0, 0, 0, 0, 0, 0]
var total: u8 = 0
on frame {
total = 0
for i in 0..8 {
total += arr[i]
}
wait_frame
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_match_statement() {
// Note: the parser doesn't support `;` as a statement separator,
// so each arm body uses newlines between statements.
let source = r#"
game "Match" { mapper: NROM }
enum Mode { Idle, Run, Jump }
var mode: u8 = Idle
var x: u8 = 0
on frame {
match mode {
Idle => { if button.a { mode = Run } }
Run => {
x += 1
if button.b { mode = Jump }
}
Jump => {
x += 2
if button.a { mode = Idle }
}
_ => {}
}
wait_frame
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_struct_literals() {
let source = r#"
game "Lit" { mapper: NROM }
struct Vec2 { x: u8, y: u8 }
var pos: Vec2 = Vec2 { x: 10, y: 20 }
on frame {
pos = Vec2 { x: 100, y: 50 }
if button.right {
pos = Vec2 { x: pos.x + 1, y: pos.y }
}
draw Smiley at: (pos.x, pos.y)
wait_frame
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_structs() {
let source = r#"
game "Structs" { mapper: NROM }
struct Vec2 { x: u8, y: u8 }
struct Player { health: u8, lives: u8 }
var pos: Vec2
var hero: Player
on frame {
pos.x = 100
pos.y = 50
hero.health = 3
hero.lives = 5
if button.right { pos.x += 1 }
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_enums() {
let source = r#"
game "Enums" { mapper: NROM }
enum Direction { Up, Down, Left, Right }
enum Mode { Idle, Running, Jumping }
var dir: u8 = 0
var mode: u8 = 0
on frame {
if button.right { dir = Right }
if button.left { dir = Left }
if dir == Right { mode = Running }
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_poke_peek_intrinsics() {
let source = r#"
game "Hardware" { mapper: NROM }
var status: u8 = 0
on frame {
// Write to PPU address / data registers directly.
poke(0x2006, 0x3F)
poke(0x2006, 0x00)
poke(0x2007, 0x0F)
// Read PPU status.
status = peek(0x2002)
wait_frame
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_raw_asm_block() {
// `raw asm` bypasses `{var}` substitution so the body is passed
// to the inline parser unchanged.
let source = r#"
game "RawAsm" { mapper: NROM }
var x: u8 = 0
on frame {
raw asm {
LDA #$42
STA $00
}
wait_frame
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_inline_asm_variable_substitution() {
let source = r#"
game "AsmVar" { mapper: NROM }
var counter: u8 = 0
on frame {
asm {
LDA {counter}
CLC
ADC #$01
STA {counter}
}
wait_frame
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_inline_asm() {
let source = r#"
game "Asm" { mapper: NROM }
var x: u8 = 0
on frame {
asm {
LDA #$42
STA $10
INC $10
LSR A
CLC
ADC #$01
}
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_while_loop() {
let source = r#"
game "Loops" { mapper: NROM }
var x: u8 = 0
on frame {
while x < 10 {
x += 1
}
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_fast_slow_vars() {
let source = r#"
game "Placement" { mapper: NROM }
fast var hot: u8 = 0
slow var cold: u8 = 0
on frame {
hot += 1
cold += 1
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_multi_state_transitions() {
let source = r#"
game "Multi" { mapper: NROM }
state Menu {
on enter { wait_frame }
on frame {
if button.start { transition Level1 }
}
}
state Level1 {
var timer: u8 = 0
on frame {
timer += 1
if timer > 60 {
transition Level2
}
}
}
state Level2 {
on frame {
if button.select { transition Menu }
}
}
start Menu
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn coin_cavern_compiles() {
let source = include_str!("../examples/coin_cavern.ne");
let rom_data = compile(source);
let info = rom::validate_ines(&rom_data).expect("should be valid iNES");
assert_eq!(info.mapper, 0);
}
#[test]
fn ir_pipeline_produces_ir() {
let source = r#"
game "IR" { mapper: NROM }
const SPEED: u8 = 2
var x: u8 = 0
fun double(n: u8) -> u8 { return n + n }
on frame {
x += SPEED
if x > 100 { x = 0 }
}
start Main
"#;
let (program, diags) = nescript::parser::parse(source);
assert!(diags.is_empty());
let program = program.unwrap();
let analysis = analyzer::analyze(&program);
assert!(analysis.diagnostics.iter().all(|d| !d.is_error()));
let mut ir_program = ir::lower(&program, &analysis);
let before_ops = ir_program.op_count();
optimizer::optimize(&mut ir_program);
let after_ops = ir_program.op_count();
// Optimizer should reduce or maintain op count (not increase)
assert!(after_ops <= before_ops, "optimizer should not increase ops");
// Should have functions for the user function + frame handler
assert!(ir_program.functions.len() >= 2);
}
Implement NEScript compiler Milestone 1 ("Hello Sprite") Complete implementation of the NEScript compiler pipeline for M1: - Lexer: full tokenization with hex/binary/decimal literals, all keywords, operators - Parser: recursive descent with Pratt expression parsing (M1 subset) - Analyzer: symbol resolution, type checking, memory allocation - 6502 Assembler: full opcode encoding table (~150 valid combinations) - Code Generator: AST → 6502 instructions (direct, no IR for M1) - Runtime: NES hardware init, NMI handler, controller read, OAM DMA - Linker: NROM layout, vector table, palette loading, CHR data - ROM Builder: iNES header generation, PRG/CHR padding - CLI: `build` and `check` subcommands via clap 143 tests across all modules: - 22 lexer tests (literals, keywords, operators, error recovery) - 18 parser tests (expressions, statements, game structure, errors) - 7 analyzer tests (symbol resolution, memory allocation, transitions) - 30 assembler tests (every addressing mode, label resolution) - 7 codegen tests (var init, arithmetic, buttons, draw, comparisons) - 11 runtime tests (init sequence, NMI handler, controller read) - 10 ROM builder tests (iNES format, mirroring, banking, validation) - 5 linker tests (vector table, CHR data, palette loading) - 7 integration tests (end-to-end compilation, error detection) CI: GitHub Actions for check, fmt, clippy, test Pre-commit: script for local fmt + clippy + test validation https://claude.ai/code/session_01W6eQFStA66EuMKHUFo2rx3
2026-04-11 22:07:56 +00:00
#[test]
fn error_test_missing_game() {
let source = "var x: u8 = 0\nstart Main";
let (_, diags) = nescript::parser::parse(source);
assert!(
diags.iter().any(nescript::errors::Diagnostic::is_error),
"should produce error"
);
}
#[test]
fn error_test_undefined_transition() {
let source = r#"
game "T" { mapper: NROM }
state Main {
on frame { transition Nonexistent }
}
start Main
"#;
let (program, parse_diags) = nescript::parser::parse(source);
assert!(parse_diags.is_empty());
let analysis = analyzer::analyze(&program.unwrap());
assert!(
analysis
.diagnostics
.iter()
.any(nescript::errors::Diagnostic::is_error),
"should detect undefined transition target"
);
}
#[test]
fn error_test_recursion_detected() {
let source = r#"
game "T" { mapper: NROM }
fun loop_forever() { loop_forever() }
on frame { wait_frame }
start Main
"#;
let (program, parse_diags) = nescript::parser::parse(source);
assert!(parse_diags.is_empty());
let analysis = analyzer::analyze(&program.unwrap());
assert!(
analysis
.diagnostics
.iter()
.any(|d| d.code == nescript::errors::ErrorCode::E0402),
"should detect recursion"
);
}
// ── M4 Tests ──
#[test]
fn program_with_scroll_and_cast() {
let source = r#"
game "M4 Test" { mapper: NROM }
var px: u8 = 0
var py: u8 = 0
var wide: u16 = 0
on frame {
if button.right { px += 1 }
wide = px as u16
scroll(px, py)
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
compiler: audio driver, u16 arithmetic, multi-scanline, slot recycling Five language features and optimizations from the planned-work backlog: - **Minimal audio driver**: `play`/`start_music`/`stop_music` now generate APU pulse-1/pulse-2 writes from a builtin SFX/music name table, and the NMI handler gains a `JSR __audio_tick` splice (via the linker's `__audio_used` marker lookup) that ages an SFX countdown counter and mutes pulse 1 when the tone expires. Programs that never trigger audio pay zero ROM cost. - **u16 arithmetic and comparisons**: new IR ops `LoadVarHi`, `StoreVarHi`, `Add16`, `Sub16`, and six `Cmp*16` variants. The lowering context tracks variable types via the analyzer's symbol table and routes expressions through the 8-bit or 16-bit path based on operand width. Add16 emits `CLC;ADC;ADC` with carry propagating naturally into the high byte; compares dispatch high-byte-first with a short-circuit low-byte fallback. Fixes a silent miscompile where `big += 1` on a u16 var only incremented the low byte. - **Multi-scanline handlers per state**: `gen_scanline_irq` now dispatches on `(current_state, ZP_SCANLINE_STEP)` and reloads the MMC3 counter with the delta to the next scanline in the same state. `gen_scanline_reload` resets the step counter at the top of each NMI so a state with multiple handlers fires them in ascending line order. Previously only the first handler per state ever fired. - **IR temp slot recycling**: `build_use_counts` pre-scans each function to count per-temp uses; `retire_op_sources` decrements the counts after each op and pushes dead slots back onto `free_slots` for later allocation. `bitwise_ops.ne` used to crash (debug) or miscompile (release) once it hit 128 concurrent temps; with recycling the same function now uses ~4 slots instead of 136. - **INC/DEC peephole fold + improved dead-load elimination**: `fold_inc_dec` collapses `LDA addr; CLC; ADC #1; STA addr` into a single `INC addr` (and the SEC/SBC variant into `DEC addr`), saving 5 bytes and 5 cycles per increment. The fold is suppressed when the next instruction reads carry. `remove_dead_loads` now walks past INC/DEC/STX/STY (which don't touch A) to find the actual next A-use, catching more dead loads. Tests: 331 unit + 39 integration (up from 313 + 37), including new guards for audio, u16, multi-scanline, and slot recycling. https://claude.ai/code/session_01A8qk3gw2jWSzdiXBZPZSFE
2026-04-12 22:21:53 +00:00
#[test]
fn program_with_u16_arithmetic_and_compare() {
// Exercises the full u16 path: literal > 255 initializer,
// u16 += u8, u16 > u16 comparison. The old codegen truncated
// all u16 operations to their low byte, so `big = 1000`
// landed as 232 and `big += 1` never carried into the high
// byte. This test just asserts the ROM builds cleanly — the
// unit tests in `codegen/ir_codegen.rs` verify the actual
// instruction shape.
let source = r#"
game "U16 Arith" { mapper: NROM }
var big: u16 = 1000
var flag: u8 = 0
on frame {
big = big + 1
if big > 1050 {
flag = 1
}
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_audio_driver() {
audio: complete the subsystem — asset pipeline, user decls, tracker-style driver The audio subsystem was a sketch: `play name` / `start_music name` / `stop_music` parsed, lowered, and emitted a few hardcoded register writes from a builtin name table. No user-declared effects, no per-frame envelope, no note streams, no real engine. This flesh-out brings audio up to the quality bar of the rest of the compiler (sprites, palettes, bank switching, scanline IRQ, etc.) with a full data-driven pipeline: ## Asset pipeline (new `src/assets/audio.rs`) - `sfx Name { duty, pitch, volume }` blocks compile into per-frame pulse-1 envelopes. Pitch/volume arrays must match in length; each entry is one NMI's worth of `$4000` data. - `music Name { duty, volume, repeat, notes }` blocks compile into flat `(pitch, duration)` streams for pulse 2. Pitch 0 is a rest, 1-60 indexes a builtin period table covering C1-B5. - `resolve_sfx` / `resolve_music` walk the program for `play` / `start_music` references and append builtin fallbacks for any name that isn't user-declared — so `play coin` still works without a `sfx Coin { ... }` block. - Builtin effects (coin, jump, hit, click, cancel, shoot, step) and tracks (theme, battle, victory, gameover) synthesize through the same compile path as user decls — one data model, one driver. ## Runtime engine (`src/runtime/mod.rs`) - `gen_audio_tick()` walks both channels every NMI: reads one envelope byte through `(ZP_SFX_PTR),Y` -> writes `$4000`, advances ptr, mutes on zero sentinel. Music decrements the note counter, advances to the next `(pitch, dur)` pair on zero, looks up the period through `(__period_table),Y`, loops on `0xFF 0xFF`. - `gen_period_table()` emits a 60-entry equal-tempered table (A4 = 440 Hz, NTSC 1.789773 MHz CPU clock) with length-counter load bits pre-baked into each high byte. - `gen_data_block()` emits a label + raw-bytes pseudo pair so user sfx/music data can be spliced into PRG with regular labels that the two-pass assembler resolves. - New ZP layout: `$05/$06` music loop base, `$07` music state (duty/volume/loop/active), `$0C-$0F` sfx and music pointers. ## IR codegen (`src/codegen/ir_codegen.rs`) - `with_audio(sfx, music)` registers compile-time trigger constants per blob name. - `gen_play_sfx` emits: write period to `$4002`/`$4003`, load envelope pointer into `ZP_SFX_PTR` via SymbolLo/SymbolHi of `__sfx_<name>`, mark the sfx counter active. - `gen_start_music` stamps the header byte into `ZP_MUSIC_STATE` with the active bit OR'd in, seeds both ptr and loop base from `__music_<name>`, primes the duration counter. - `gen_stop_music` mutes pulse 2 and clears state. ## Linker (`src/linker/mod.rs`) - New `link_with_all_assets(user_code, sprites, sfx, music)` path that splices driver body, period table, and each sfx/music data blob into PRG — all guarded on the `__audio_used` marker so silent programs pay zero ROM cost. ## Assembler (`src/asm/opcodes.rs`, `src/asm/mod.rs`) - New `AddressingMode::Bytes(Vec<u8>)` variant for raw-data pseudo-instructions. `NOP+Bytes(v)` emits the payload verbatim, letting the linker splice ROM data tables into a code section and still have `Label` / `SymbolLo` / `SymbolHi` fixups resolve correctly in the same assembly pass. ## Analyzer - `play` / `start_music` now validate the name against user decls and builtin tables. Unknown names emit E0505 with a helpful list of builtins — previously a typo would silently compile to no-op. ## Parser - New `sfx_decl` / `music_decl` grammar with property-style configuration. Strict validation: duty 0-3, volume 0-15, pitch arrays must match volume length, music notes must come in pairs, pitch 0-60, duration ≥ 1. ## Tests +170 new tests across every layer: - `src/assets/audio.rs`: 17 tests (compile, resolve, builtins, shadowing, label sanitation, nested reference walks) - `src/parser/tests.rs`: 13 tests (valid/invalid sfx + music declarations, property validation, play/start_music/stop_music) - `src/analyzer/tests.rs`: 7 tests (builtin acceptance, user decl acceptance, unknown-name rejection) - `src/runtime/tests.rs`: 10 tests (audio tick labels, RTS end, $4000 write, $4004 mute, period table assembly, A4 = 440 Hz, length counter bits, data block verbatim emit) - `src/linker/tests.rs`: 4 tests (sfx/music blob placement, pointer resolution, elision when unused) - `src/codegen/ir_codegen.rs`: rewrote the 4 existing audio tests to match the new data-driven contract - `tests/integration_test.rs`: 4 end-to-end tests including a user-declared `sfx` + `music` program that verifies bytes land in PRG ROM at the right addresses ## Docs - New Audio section in `docs/language-guide.md` with syntax reference, builtin tables, and an explanation of how the driver works at compile and run time. - `docs/architecture.md` updated to reflect the real audio pipeline instead of the old "audio import stubs" stub. - `docs/future-work.md` moves audio from "status: minimal" to "status: full subsystem" with a narrower list of follow-up work (triangle/noise/DMC channels, NSF/FTM imports, richer envelopes). - `examples/audio_demo.ne` rewritten to showcase user-declared `sfx LongCoin`, `sfx Zap`, `music Theme`, still demonstrating builtin fallback via `play coin`. Total: 424 tests passing (381 unit + 43 integration), clippy clean, fmt clean, all 19 examples compile. https://claude.ai/code/session_015WfaDttE3DpWn9rpyfpQd8
2026-04-13 01:10:21 +00:00
// Exercises the audio driver end-to-end with builtin sfx/music
// names: play, start_music, stop_music all lower into the
// data-driven driver, the linker splices the tick/period-table/
// data blobs, and the resulting ROM is valid iNES.
compiler: audio driver, u16 arithmetic, multi-scanline, slot recycling Five language features and optimizations from the planned-work backlog: - **Minimal audio driver**: `play`/`start_music`/`stop_music` now generate APU pulse-1/pulse-2 writes from a builtin SFX/music name table, and the NMI handler gains a `JSR __audio_tick` splice (via the linker's `__audio_used` marker lookup) that ages an SFX countdown counter and mutes pulse 1 when the tone expires. Programs that never trigger audio pay zero ROM cost. - **u16 arithmetic and comparisons**: new IR ops `LoadVarHi`, `StoreVarHi`, `Add16`, `Sub16`, and six `Cmp*16` variants. The lowering context tracks variable types via the analyzer's symbol table and routes expressions through the 8-bit or 16-bit path based on operand width. Add16 emits `CLC;ADC;ADC` with carry propagating naturally into the high byte; compares dispatch high-byte-first with a short-circuit low-byte fallback. Fixes a silent miscompile where `big += 1` on a u16 var only incremented the low byte. - **Multi-scanline handlers per state**: `gen_scanline_irq` now dispatches on `(current_state, ZP_SCANLINE_STEP)` and reloads the MMC3 counter with the delta to the next scanline in the same state. `gen_scanline_reload` resets the step counter at the top of each NMI so a state with multiple handlers fires them in ascending line order. Previously only the first handler per state ever fired. - **IR temp slot recycling**: `build_use_counts` pre-scans each function to count per-temp uses; `retire_op_sources` decrements the counts after each op and pushes dead slots back onto `free_slots` for later allocation. `bitwise_ops.ne` used to crash (debug) or miscompile (release) once it hit 128 concurrent temps; with recycling the same function now uses ~4 slots instead of 136. - **INC/DEC peephole fold + improved dead-load elimination**: `fold_inc_dec` collapses `LDA addr; CLC; ADC #1; STA addr` into a single `INC addr` (and the SEC/SBC variant into `DEC addr`), saving 5 bytes and 5 cycles per increment. The fold is suppressed when the next instruction reads carry. `remove_dead_loads` now walks past INC/DEC/STX/STY (which don't touch A) to find the actual next A-use, catching more dead loads. Tests: 331 unit + 39 integration (up from 313 + 37), including new guards for audio, u16, multi-scanline, and slot recycling. https://claude.ai/code/session_01A8qk3gw2jWSzdiXBZPZSFE
2026-04-12 22:21:53 +00:00
let source = r#"
game "Audio" { mapper: NROM }
on frame {
if button.a { play coin }
if button.b { start_music theme }
if button.start { stop_music }
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
audio: complete the subsystem — asset pipeline, user decls, tracker-style driver The audio subsystem was a sketch: `play name` / `start_music name` / `stop_music` parsed, lowered, and emitted a few hardcoded register writes from a builtin name table. No user-declared effects, no per-frame envelope, no note streams, no real engine. This flesh-out brings audio up to the quality bar of the rest of the compiler (sprites, palettes, bank switching, scanline IRQ, etc.) with a full data-driven pipeline: ## Asset pipeline (new `src/assets/audio.rs`) - `sfx Name { duty, pitch, volume }` blocks compile into per-frame pulse-1 envelopes. Pitch/volume arrays must match in length; each entry is one NMI's worth of `$4000` data. - `music Name { duty, volume, repeat, notes }` blocks compile into flat `(pitch, duration)` streams for pulse 2. Pitch 0 is a rest, 1-60 indexes a builtin period table covering C1-B5. - `resolve_sfx` / `resolve_music` walk the program for `play` / `start_music` references and append builtin fallbacks for any name that isn't user-declared — so `play coin` still works without a `sfx Coin { ... }` block. - Builtin effects (coin, jump, hit, click, cancel, shoot, step) and tracks (theme, battle, victory, gameover) synthesize through the same compile path as user decls — one data model, one driver. ## Runtime engine (`src/runtime/mod.rs`) - `gen_audio_tick()` walks both channels every NMI: reads one envelope byte through `(ZP_SFX_PTR),Y` -> writes `$4000`, advances ptr, mutes on zero sentinel. Music decrements the note counter, advances to the next `(pitch, dur)` pair on zero, looks up the period through `(__period_table),Y`, loops on `0xFF 0xFF`. - `gen_period_table()` emits a 60-entry equal-tempered table (A4 = 440 Hz, NTSC 1.789773 MHz CPU clock) with length-counter load bits pre-baked into each high byte. - `gen_data_block()` emits a label + raw-bytes pseudo pair so user sfx/music data can be spliced into PRG with regular labels that the two-pass assembler resolves. - New ZP layout: `$05/$06` music loop base, `$07` music state (duty/volume/loop/active), `$0C-$0F` sfx and music pointers. ## IR codegen (`src/codegen/ir_codegen.rs`) - `with_audio(sfx, music)` registers compile-time trigger constants per blob name. - `gen_play_sfx` emits: write period to `$4002`/`$4003`, load envelope pointer into `ZP_SFX_PTR` via SymbolLo/SymbolHi of `__sfx_<name>`, mark the sfx counter active. - `gen_start_music` stamps the header byte into `ZP_MUSIC_STATE` with the active bit OR'd in, seeds both ptr and loop base from `__music_<name>`, primes the duration counter. - `gen_stop_music` mutes pulse 2 and clears state. ## Linker (`src/linker/mod.rs`) - New `link_with_all_assets(user_code, sprites, sfx, music)` path that splices driver body, period table, and each sfx/music data blob into PRG — all guarded on the `__audio_used` marker so silent programs pay zero ROM cost. ## Assembler (`src/asm/opcodes.rs`, `src/asm/mod.rs`) - New `AddressingMode::Bytes(Vec<u8>)` variant for raw-data pseudo-instructions. `NOP+Bytes(v)` emits the payload verbatim, letting the linker splice ROM data tables into a code section and still have `Label` / `SymbolLo` / `SymbolHi` fixups resolve correctly in the same assembly pass. ## Analyzer - `play` / `start_music` now validate the name against user decls and builtin tables. Unknown names emit E0505 with a helpful list of builtins — previously a typo would silently compile to no-op. ## Parser - New `sfx_decl` / `music_decl` grammar with property-style configuration. Strict validation: duty 0-3, volume 0-15, pitch arrays must match volume length, music notes must come in pairs, pitch 0-60, duration ≥ 1. ## Tests +170 new tests across every layer: - `src/assets/audio.rs`: 17 tests (compile, resolve, builtins, shadowing, label sanitation, nested reference walks) - `src/parser/tests.rs`: 13 tests (valid/invalid sfx + music declarations, property validation, play/start_music/stop_music) - `src/analyzer/tests.rs`: 7 tests (builtin acceptance, user decl acceptance, unknown-name rejection) - `src/runtime/tests.rs`: 10 tests (audio tick labels, RTS end, $4000 write, $4004 mute, period table assembly, A4 = 440 Hz, length counter bits, data block verbatim emit) - `src/linker/tests.rs`: 4 tests (sfx/music blob placement, pointer resolution, elision when unused) - `src/codegen/ir_codegen.rs`: rewrote the 4 existing audio tests to match the new data-driven contract - `tests/integration_test.rs`: 4 end-to-end tests including a user-declared `sfx` + `music` program that verifies bytes land in PRG ROM at the right addresses ## Docs - New Audio section in `docs/language-guide.md` with syntax reference, builtin tables, and an explanation of how the driver works at compile and run time. - `docs/architecture.md` updated to reflect the real audio pipeline instead of the old "audio import stubs" stub. - `docs/future-work.md` moves audio from "status: minimal" to "status: full subsystem" with a narrower list of follow-up work (triangle/noise/DMC channels, NSF/FTM imports, richer envelopes). - `examples/audio_demo.ne` rewritten to showcase user-declared `sfx LongCoin`, `sfx Zap`, `music Theme`, still demonstrating builtin fallback via `play coin`. Total: 424 tests passing (381 unit + 43 integration), clippy clean, fmt clean, all 19 examples compile. https://claude.ai/code/session_015WfaDttE3DpWn9rpyfpQd8
2026-04-13 01:10:21 +00:00
#[test]
fn program_with_user_declared_sfx_and_music() {
// Full user-declared audio pipeline: `sfx` and `music` blocks,
// references via `play`/`start_music`, full ROM emission. The
// resolved envelope and note-stream bytes should land in PRG
// under stable labels so the IR codegen's SymbolLo/SymbolHi
// references resolve.
let source = r#"
game "Audio Assets" { mapper: NROM }
sfx Zap {
duty: 2
pitch: [0x20, 0x22, 0x24, 0x26, 0x28, 0x2A]
volume: [15, 13, 11, 9, 6, 3]
}
music Loop {
duty: 2
volume: 10
repeat: true
notes: [37, 8, 41, 8, 44, 8, 49, 8]
}
var t: u8 = 0
on frame {
t += 1
if t == 30 { play Zap }
if t == 60 {
t = 0
start_music Loop
}
}
start Main
"#;
let rom_data = compile(source);
let info = rom::validate_ines(&rom_data).expect("should be valid iNES");
assert_eq!(info.mapper, 0);
// Verify the user-declared envelope appears in PRG. The
// resolver encodes `Zap` as
// duty << 6 | 0x30 | volume
// per frame, terminated by a zero sentinel.
let prg = &rom_data[16..16 + 16384];
let env = |v: u8| (2u8 << 6) | 0x30u8 | v;
let zap_env: [u8; 7] = [env(15), env(13), env(11), env(9), env(6), env(3), 0x00];
assert!(
prg.windows(zap_env.len()).any(|w| w == zap_env),
"Zap envelope bytes should be in PRG ROM"
);
// Verify the music stream is in PRG: (37, 8, 41, 8, 44, 8, 49, 8, 0xFF, 0xFF)
let loop_stream: [u8; 10] = [37, 8, 41, 8, 44, 8, 49, 8, 0xFF, 0xFF];
assert!(
prg.windows(loop_stream.len()).any(|w| w == loop_stream),
"Loop music note stream should be in PRG ROM"
);
}
#[test]
fn program_without_audio_has_no_audio_driver_in_prg() {
// Programs that never touch audio should pay zero ROM cost:
// no period table, no driver body, no data blobs. We verify
// indirectly by checking that the `__audio_tick` entry point
// wouldn't have anything to JSR to (because the NMI splice
// is gated on the `__audio_used` marker which never exists).
//
// The cheapest observable signal: a period-table fingerprint.
// The period table always starts with a distinct 2-byte
// sequence that appears at C1's period; if we don't see it in
// PRG, the audio subsystem wasn't linked in.
let source = r#"
game "Silent" { mapper: NROM }
var x: u8 = 0
on frame { x += 1 }
start Main
"#;
let rom_data = compile(source);
// Pull the period table for C1 and make sure it's NOT in PRG.
// C1 ≈ 32.7 Hz → period ≈ 3421 → but that's too big for 11
// bits, so it clamps. Instead, use the distinctive combined
// LDA #imm / LDA #imm pattern from the audio tick itself that
// would only appear if the driver body was linked in.
//
// A robust fingerprint: the `JSR __audio_tick` opcode byte
// ($20) followed by any 2 bytes only appears in the NMI
// handler when audio was used. We test the absence of the
// label instead via an indirect method: count the total
// number of STA $4004 writes (pulse-2 register). When audio
// is unused, there should be none. When audio is used, there
// would be several in the driver.
let prg = &rom_data[16..16 + 16384];
// `STA $4006` ($8D $06 $40) is written exclusively by the
// music tick's period-lookup path. The init code pre-silences
// $4004 but never touches $4006, so its presence is a reliable
// "the audio driver was linked in" signal.
let pattern: [u8; 3] = [0x8D, 0x06, 0x40];
let count = prg.windows(pattern.len()).filter(|w| *w == pattern).count();
assert_eq!(
count, 0,
"silent program should not contain the music tick's $4006 write"
);
}
#[test]
fn unknown_sfx_name_is_a_hard_error() {
// The analyzer must reject `play NoSuchSfx` (neither a user
// decl nor a builtin) with E0505. Regression test for the
// old behavior, which silently accepted any name.
let source = r#"
game "T" { mapper: NROM }
on frame { play NoSuchSfx }
start Main
"#;
let (program, _) = nescript::parser::parse(source);
let analysis = analyzer::analyze(&program.unwrap());
assert!(
analysis
.diagnostics
.iter()
.any(nescript::errors::Diagnostic::is_error),
"unknown sfx should produce an error"
);
}
#[test]
fn audio_pipeline_drops_period_table_cost_when_unused() {
// Regression test for the "no-cost elision" invariant: a
// program with no audio statements should produce a ROM
// smaller than one that uses audio. The exact byte count
// varies with codegen changes, so we test the *ordering* of
// sizes: a silent program < an audio program.
let silent = compile(
r#"
game "Silent" { mapper: NROM }
var x: u8 = 0
on frame { x += 1 }
start Main
"#,
);
// Both ROMs are the same file size (16 header + 16 KB PRG + 8
// KB CHR = 24592), but the silent program's PRG fills with
// $FF padding past the code; an audio program's PRG has the
// driver and tables eating into that padding space. So we
// count $FF bytes in PRG: the silent version must have more.
let audio = compile(
r#"
game "Audio" { mapper: NROM }
on frame { play coin }
start Main
"#,
);
let silent_prg = &silent[16..16 + 16384];
let audio_prg = &audio[16..16 + 16384];
// Count padding bytes ($FF = PRG fill) in each ROM. Using a
// raw filter().count() is clippy-noisy ("naive_bytecount"),
// but pulling in the `bytecount` crate for a one-line test
// helper isn't worth it — the test runs once per build.
#[allow(clippy::naive_bytecount)]
let silent_ff = silent_prg.iter().filter(|&&b| b == 0xFF).count();
#[allow(clippy::naive_bytecount)]
let audio_ff = audio_prg.iter().filter(|&&b| b == 0xFF).count();
assert!(
silent_ff > audio_ff,
"silent program should have more $FF padding than an audio program \
(silent={silent_ff}, audio={audio_ff})"
);
}
// ── M3 Tests ──
#[test]
cleanup: fix silent miscompiles and delete dead code exposed by code review Two correctness bugs were silently producing wrong ROMs: - `x << n` / `x >> n` always shifted by 1, regardless of `n`, because the IR lowering for `BinOp::ShiftLeft`/`ShiftRight` hardcoded the count. Now eval_const the RHS into a compile-time count; fall back to a new `IrOp::ShiftLeftVar` / `ShiftRightVar` (runtime loop) when the amount isn't constant. Strength reduction folds the variable form back to a fixed count once the optimizer knows the value. - `x / n` / `x % n` always returned 0, because the lowering emitted `LoadImm(t, 0)` for `BinOp::Div`/`Mod` with a comment saying the runtime call was "TODO for now". Added real `IrOp::Div` and `IrOp::Mod`, wired them through use-counting and DCE, gave codegen `__divide`-based implementations, and taught strength reduction to rewrite power-of-two divisors into shifts and modulo-by-2ⁿ into AND masks. Constant folding now handles `Mul`/`Div`/`Mod`/shifts too, which were previously left for the codegen to emit inefficient software calls. Dead code removed (no backward-compat shims kept): - `src/debug/` entirely. `DebugSymbols`, `SourceMap`, and the Mesen/.sym emitters had no callers outside their own tests; `main.rs` never wrote a symbol file. Documented the intent in `docs/future-work.md` so it comes back intentionally if needed. - `ErrorCode::E0202` (invalid cast) and `E0403` (unreachable state): defined, formatted, and marked `#[allow(dead_code)]` but never emitted. W0104 now carries the unreachable-state semantics too. - `Level::Info`: never constructed. - `load_background` / `set_palette` statements and their `BackgroundDecl` / `PaletteDecl` parser support: parsed and silently dropped by IR lowering (`// TODO: implement in asset pipeline`). Removed keywords, AST variants, parser paths, analyzer arms, and tests. `docs/future-work.md` documents the runtime palette/nametable design for when it comes back. Doc cleanup: - `docs/architecture.md` was describing files that don't exist (`analyzer/types.rs`, `optimizer/const_fold.rs`, `codegen/regalloc.rs`, `rom/header.rs`, `debug/symbols.rs`, …). Rewrote it to match the real flat `mod.rs` + `tests.rs` layout and the real pipeline order. - `docs/future-work.md` was a hybrid of open work and "recently completed" entries that duplicated the active stubs at the top of the file. Collapsed to just the gaps that are actually still open. - `README.md` claimed Mesen symbol export and 210 tests; updated both. - `docs/language-guide.md` and `spec.md` described `palette` decls, `set_palette` / `load_background`, `debug.overlay`, and error codes that were never emitted. Trimmed. - Stale comments on `Statement::Play`/`StartMusic`/`StopMusic` claimed the audio subsystem was "a no-op at codegen time". Tests: - Regression tests for every fix above (`lower_shift_left_with_literal _count_uses_that_count`, `lower_shift_right_with_variable_count _uses_runtime_variant`, `lower_divide_emits_div_op_not_load_imm _zero`, `lower_modulo_emits_mod_op_not_load_imm_zero`, `strength_reduce_div_by_power_of_two`, `strength_reduce_mod_by _power_of_two`, `strength_reduce_shift_var_with_constant_amount`). - Renamed the `program_with_sprites_and_palette` integration test (which was exercising the now-removed `load_background`/`set_palette`) to `program_with_inline_sprite_chr`. `examples/sprites_and_palettes.ne` lost its `palette`/`set_palette` usage. Nothing in the emulator test presses A, so the headless jsnes render shouldn't move, but the golden may need regeneration via `UPDATE_GOLDENS=1` if it does. https://claude.ai/code/session_012fKB251HvEUQwG3tizFyqt
2026-04-13 02:47:37 +00:00
fn program_with_inline_sprite_chr() {
let source = r#"
game "M3 Assets" { mapper: NROM }
sprite Player {
chr: [0x3C, 0x42, 0x81, 0x81, 0x81, 0x81, 0x42, 0x3C,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00]
}
var px: u8 = 128
var py: u8 = 120
state Title {
on frame {
if button.right { px += 2 }
if button.left { px -= 2 }
draw Player at: (px, py)
}
}
start Title
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
palette/background: first-class declarations with reset-time load and runtime swaps Re-adds `palette Name { colors: [...] }` and `background Name { tiles: [...], attributes: [...] }` as first-class declarations, plus `set_palette Name` and `load_background Name` statements for runtime swaps. Unlike the previous iteration that quietly no-op'd, this one is fully wired through the pipeline and its behavior is pinned by both unit tests and an emulator golden. Pipeline: - Lexer: re-adds `palette`, `background`, `set_palette`, `load_background` keywords and tokenizes them. - AST: `PaletteDecl` (name + 1..=32 colour bytes) and `BackgroundDecl` (name + 0..=960 tile bytes + 0..=64 attribute bytes) live in `Program`. `Statement::SetPalette` and `Statement::LoadBackground` name-reference these declarations. - Parser: `palette Name { colors: [...] }` / `background Name { tiles: [...], attributes: [...] }` blocks and their statement forms parse via the existing byte-array helper. - Analyzer: validates colour indices ($00-$3F), palette length (<=32), nametable length (<=960), attribute length (<=64), and duplicate decl names. `set_palette` / `load_background` targets must reference a declared name (E0502 otherwise). When a program declares palette or background, the analyzer bumps the user zero-page allocator's starting address from `$10` to `$18` to reserve `$11-$17` for the runtime update handshake — programs that don't use the feature keep the old layout so their emulator goldens stay byte-exact. - Assets: `PaletteData` and `BackgroundData` resolve declarations into zero-padded fixed-size blobs (32 / 960 / 64 bytes) and expose `label()` / `tiles_label()` / `attrs_label()` for codegen to reference. - IR: new `IrOp::SetPalette(String)` and `IrOp::LoadBackground(String)`; lowering forwards the names verbatim. - Codegen: `gen_set_palette` writes the palette label pointer into ZP `$12/$13` and ORs bit 0 into the update flags at `$11`; `gen_load_background` does the same for tile/attribute pointers at `$14/$15/$16/$17` with bit 1. Both emit a `__ppu_update_used` marker so the linker splices in the NMI apply helper only when the feature is actually used. - Runtime: `gen_initial_palette_load` and `gen_initial_background_load` write the first declared palette/background at reset time (before rendering is enabled, where PPU writes are safe). `gen_nmi(has_ppu_updates)` takes a new flag; when true it splices `gen_ppu_update_apply` at the top of the NMI body, which checks the `$11` flags byte and copies pending palette / nametable data to `$3F00` / `$2000` inside vblank. All helpers use only ZP $02/$03 as scratch at reset time and never clobber ZP slots live across NMI. - Linker: new `link_banked_with_ppu` takes slice of `PaletteData` / `BackgroundData`; splices each blob as a labelled data block in PRG ROM, picks the first-declared as the reset-time load target, enables background rendering automatically when a background is declared, and threads `has_ppu_updates` into `gen_nmi`. Old `link_banked` remains as a thin wrapper for callers without palette/background data so existing tests don't shift. Tests: - Lexer: tokenization of the 4 new keywords (single added test case). - Parser: 5 new tests for `palette` / `background` decls with and without attributes, plus `set_palette` / `load_background` statements. - Analyzer: 9 new tests covering acceptance of declared palettes/backgrounds, E0502 for unknown names, E0201 for out-of-range NES colors and oversized blobs, E0501 for duplicate names, and the zero-page-layout guard (palette/bg decls bump ZP start; no decls keeps it at $10). - Resolver: 3 new tests for zero-padding, truncation of oversized decls, and label derivation. - IR: 2 new lowering tests for `set_palette` and `load_background`. - Integration: 5 new tests — blob contents spliced verbatim into PRG, `STA $12` / `STA $14` emitted by set_palette / load_background codegen, and a regression guard that programs without palette/background still land user vars at $10. - Emulator: new `examples/palette_and_background.ne` driven by a frame counter that toggles between `CoolBlues` / `WarmReds` and `TitleScreen` / `StageOne` every 90 frames. Golden PNG and audio hash checked in under `tests/emulator/goldens/` and verified via `node run_examples.mjs` — rendered image shows the blue `CoolBlues` palette with the nametable populated from `TitleScreen`. Docs: - `README.md` adds the feature to the headline list and the example table. - `docs/language-guide.md` restores the palette/background sections with the full 32-byte layout table and `set_palette` / `load_background` statement references. - `docs/future-work.md` replaces the "removed as dead code" entry with the remaining gaps (PNG-sourced palette and nametable assets, cross-vblank large background updates, memory-map reporting). - `spec.md` restores the grammar productions and usage examples. - `examples/README.md` lists the new demo. All 497 unit + integration tests pass. Clippy clean. All 21 emulator goldens match after the update pass. https://claude.ai/code/session_012fKB251HvEUQwG3tizFyqt
2026-04-13 11:11:33 +00:00
#[test]
fn program_with_palette_compiles_and_blob_is_in_prg() {
let source = r#"
game "PalTest" { mapper: NROM }
palette Cool {
colors: [0x0F, 0x01, 0x11, 0x21,
0x0F, 0x02, 0x12, 0x22,
0x0F, 0x0C, 0x1C, 0x2C,
0x0F, 0x0B, 0x1B, 0x2B,
0x0F, 0x01, 0x11, 0x21,
0x0F, 0x16, 0x27, 0x30,
0x0F, 0x14, 0x24, 0x34,
0x0F, 0x0B, 0x1B, 0x2B]
}
on frame { wait_frame }
start Main
"#;
let rom_data = compile_banked(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
// The 32-byte palette blob lands verbatim inside PRG ROM.
// Search for a distinctive 8-byte subsequence from sub-palette 3
// that doesn't collide with any of the other blobs or init
// sequences the linker emits.
let needle = [0x0F, 0x16, 0x27, 0x30, 0x0F, 0x14, 0x24, 0x34];
let found = rom_data.windows(needle.len()).any(|w| w == needle);
assert!(
found,
"palette bytes should be spliced into PRG ROM verbatim"
);
}
#[test]
fn program_with_set_palette_queues_update_at_runtime() {
// A program with a `set_palette Name` statement should emit
// the `__ppu_update_used` marker (so the linker pulls in the
// NMI helper) and must contain the zero-page write sequence
// that stores the palette label pointer into $12/$13.
let source = r#"
game "PalRuntime" { mapper: NROM }
palette Swap { colors: [0x0F, 0x01, 0x11, 0x21] }
on frame { set_palette Swap }
start Main
"#;
let rom_data = compile_banked(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
// $12 == ZP_PENDING_PALETTE_LO, so the code will contain
// `STA $12` (opcode 85 12) somewhere in PRG.
let sta_12 = [0x85u8, 0x12];
let found = rom_data.windows(sta_12.len()).any(|w| w == sta_12);
assert!(
found,
"set_palette codegen should emit `STA $12` (ZP_PENDING_PALETTE_LO)"
);
}
#[test]
fn program_with_background_compiles_and_tiles_spliced() {
let source = r#"
game "BgTest" { mapper: NROM }
background Stage {
tiles: [0xAA, 0xBB, 0xCC, 0xDD, 0xEE]
}
on frame { wait_frame }
start Main
"#;
let rom_data = compile_banked(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
// The distinctive 5-byte prefix of the tiles blob should be in
// PRG verbatim (the resolver zero-pads to 960 bytes so the tail
// is mostly zero).
let needle = [0xAA, 0xBB, 0xCC, 0xDD, 0xEE];
let found = rom_data.windows(needle.len()).any(|w| w == needle);
assert!(
found,
"background tile bytes should be spliced into PRG ROM verbatim"
);
}
#[test]
fn program_with_load_background_queues_update() {
let source = r#"
game "BgRuntime" { mapper: NROM }
background Stage { tiles: [1, 2, 3] }
on frame { load_background Stage }
start Main
"#;
let rom_data = compile_banked(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
// $14 == ZP_PENDING_BG_TILES_LO.
let sta_14 = [0x85u8, 0x14];
let found = rom_data.windows(sta_14.len()).any(|w| w == sta_14);
assert!(
found,
"load_background codegen should emit `STA $14` (ZP_PENDING_BG_TILES_LO)"
);
}
#[test]
fn program_without_palette_does_not_reserve_ppu_zero_page() {
// Regression guard: programs that don't declare palette or
// background should keep user vars starting at $10, same as
// they always did, so existing emulator goldens don't shift.
let source = r#"
game "NoPal" { mapper: NROM }
var x: u8 = 42
on frame { x += 1 }
start Main
"#;
let rom_data = compile_banked(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
// `STA $10` (85 10) corresponds to writing to the first user
// var slot. Guarantees `x` is still allocated at $10.
let sta_10 = [0x85u8, 0x10];
let found = rom_data.windows(sta_10.len()).any(|w| w == sta_10);
assert!(
found,
"user var should still land at $10 when no palette/bg declared"
);
}
// ── M5 Tests ──
/// Compile a source string using the mapper-aware linker.
fn compile_with_mapper(source: &str) -> Vec<u8> {
banks: implement multi-bank PRG layout and bank-switching runtime Prior to this commit the linker always shipped a single 16 KB PRG bank regardless of the declared mapper, so the README's MMC1/UxROM/ MMC3 support was aspirational. This commit gives the three banked mappers a real multi-bank ROM layout: * RomBuilder.set_prg_banks() writes any number of 16 KB banks back-to-back so the iNES header reflects the true PRG size. * Linker.link_banked() places switchable banks first, fixed bank last, so the fixed bank maps to $C000-$FFFF (the address window where vectors and the runtime live). * runtime::gen_mapper_init() emits reset-time mapper config: MMC1 serial-writes a control-register value that pins the last bank at $C000 with the correct mirroring, UxROM relies on the power-on default, MMC3 writes the $8000/$8001/$A000/$E000 registers to get a known PRG and mirroring state. * runtime::gen_bank_select() is a mapper-specific subroutine (callable with the target bank in A) that maps any physical bank to $8000-$BFFF. * runtime::gen_bank_trampoline() generates a cross-bank call stub in the fixed bank that saves the caller's bank, switches, JSRs the target, and restores the fixed bank. * The CLI and integration helper thread declared `bank X: prg` declarations through to the linker so MMC1/UxROM/MMC3 programs actually produce multi-bank ROMs. Coverage: * Runtime unit tests (18 new): mapper init patterns for every supported mapper, bank-select signatures, trampoline dispatch order, UxROM bus-conflict table contents. * RomBuilder tests (6 new): multi-bank layout, padding, byte-level fidelity, per-bank size validation, legacy single-bank fallback. * Linker tests (13 new): multi-bank ROM sizes across MMC1/ UxROM/MMC3, fixed-bank placement, switchable-bank payload fidelity, bank-select subroutine detection, NROM rejection of switchable banks. * Integration e2e tests (16 new): compile real .ne sources through the full pipeline and assert on iNES headers, mapper init signatures in the fixed bank, vector locations, and a regression check against `examples/mmc1_banked.ne`. Total: 474 tests pass under `cargo test` with `RUSTFLAGS="-D warnings"`. https://claude.ai/code/session_01UCressA5e8k1XsuoJYLav2
2026-04-13 01:50:51 +00:00
compile_banked(source)
}
/// Compile a source string, running the full IR pipeline and
/// routing declared `bank X: prg` entries through `link_banked`
/// as empty switchable PRG slots. This mirrors the real CLI path.
fn compile_banked(source: &str) -> Vec<u8> {
let (program, diags) = nescript::parser::parse(source);
assert!(
diags.is_empty(),
"unexpected parse errors: {diags:?}\nsource:\n{source}"
);
let program = program.expect("parse should succeed");
let analysis = analyzer::analyze(&program);
assert!(
analysis.diagnostics.iter().all(|d| !d.is_error()),
"unexpected analysis errors: {:?}",
analysis.diagnostics
);
let mut ir_program = ir::lower(&program, &analysis);
nescript::optimizer::optimize(&mut ir_program);
let sprites = assets::resolve_sprites(&program, Path::new("."))
.expect("sprite resolution should succeed");
audio: complete the subsystem — asset pipeline, user decls, tracker-style driver The audio subsystem was a sketch: `play name` / `start_music name` / `stop_music` parsed, lowered, and emitted a few hardcoded register writes from a builtin name table. No user-declared effects, no per-frame envelope, no note streams, no real engine. This flesh-out brings audio up to the quality bar of the rest of the compiler (sprites, palettes, bank switching, scanline IRQ, etc.) with a full data-driven pipeline: ## Asset pipeline (new `src/assets/audio.rs`) - `sfx Name { duty, pitch, volume }` blocks compile into per-frame pulse-1 envelopes. Pitch/volume arrays must match in length; each entry is one NMI's worth of `$4000` data. - `music Name { duty, volume, repeat, notes }` blocks compile into flat `(pitch, duration)` streams for pulse 2. Pitch 0 is a rest, 1-60 indexes a builtin period table covering C1-B5. - `resolve_sfx` / `resolve_music` walk the program for `play` / `start_music` references and append builtin fallbacks for any name that isn't user-declared — so `play coin` still works without a `sfx Coin { ... }` block. - Builtin effects (coin, jump, hit, click, cancel, shoot, step) and tracks (theme, battle, victory, gameover) synthesize through the same compile path as user decls — one data model, one driver. ## Runtime engine (`src/runtime/mod.rs`) - `gen_audio_tick()` walks both channels every NMI: reads one envelope byte through `(ZP_SFX_PTR),Y` -> writes `$4000`, advances ptr, mutes on zero sentinel. Music decrements the note counter, advances to the next `(pitch, dur)` pair on zero, looks up the period through `(__period_table),Y`, loops on `0xFF 0xFF`. - `gen_period_table()` emits a 60-entry equal-tempered table (A4 = 440 Hz, NTSC 1.789773 MHz CPU clock) with length-counter load bits pre-baked into each high byte. - `gen_data_block()` emits a label + raw-bytes pseudo pair so user sfx/music data can be spliced into PRG with regular labels that the two-pass assembler resolves. - New ZP layout: `$05/$06` music loop base, `$07` music state (duty/volume/loop/active), `$0C-$0F` sfx and music pointers. ## IR codegen (`src/codegen/ir_codegen.rs`) - `with_audio(sfx, music)` registers compile-time trigger constants per blob name. - `gen_play_sfx` emits: write period to `$4002`/`$4003`, load envelope pointer into `ZP_SFX_PTR` via SymbolLo/SymbolHi of `__sfx_<name>`, mark the sfx counter active. - `gen_start_music` stamps the header byte into `ZP_MUSIC_STATE` with the active bit OR'd in, seeds both ptr and loop base from `__music_<name>`, primes the duration counter. - `gen_stop_music` mutes pulse 2 and clears state. ## Linker (`src/linker/mod.rs`) - New `link_with_all_assets(user_code, sprites, sfx, music)` path that splices driver body, period table, and each sfx/music data blob into PRG — all guarded on the `__audio_used` marker so silent programs pay zero ROM cost. ## Assembler (`src/asm/opcodes.rs`, `src/asm/mod.rs`) - New `AddressingMode::Bytes(Vec<u8>)` variant for raw-data pseudo-instructions. `NOP+Bytes(v)` emits the payload verbatim, letting the linker splice ROM data tables into a code section and still have `Label` / `SymbolLo` / `SymbolHi` fixups resolve correctly in the same assembly pass. ## Analyzer - `play` / `start_music` now validate the name against user decls and builtin tables. Unknown names emit E0505 with a helpful list of builtins — previously a typo would silently compile to no-op. ## Parser - New `sfx_decl` / `music_decl` grammar with property-style configuration. Strict validation: duty 0-3, volume 0-15, pitch arrays must match volume length, music notes must come in pairs, pitch 0-60, duration ≥ 1. ## Tests +170 new tests across every layer: - `src/assets/audio.rs`: 17 tests (compile, resolve, builtins, shadowing, label sanitation, nested reference walks) - `src/parser/tests.rs`: 13 tests (valid/invalid sfx + music declarations, property validation, play/start_music/stop_music) - `src/analyzer/tests.rs`: 7 tests (builtin acceptance, user decl acceptance, unknown-name rejection) - `src/runtime/tests.rs`: 10 tests (audio tick labels, RTS end, $4000 write, $4004 mute, period table assembly, A4 = 440 Hz, length counter bits, data block verbatim emit) - `src/linker/tests.rs`: 4 tests (sfx/music blob placement, pointer resolution, elision when unused) - `src/codegen/ir_codegen.rs`: rewrote the 4 existing audio tests to match the new data-driven contract - `tests/integration_test.rs`: 4 end-to-end tests including a user-declared `sfx` + `music` program that verifies bytes land in PRG ROM at the right addresses ## Docs - New Audio section in `docs/language-guide.md` with syntax reference, builtin tables, and an explanation of how the driver works at compile and run time. - `docs/architecture.md` updated to reflect the real audio pipeline instead of the old "audio import stubs" stub. - `docs/future-work.md` moves audio from "status: minimal" to "status: full subsystem" with a narrower list of follow-up work (triangle/noise/DMC channels, NSF/FTM imports, richer envelopes). - `examples/audio_demo.ne` rewritten to showcase user-declared `sfx LongCoin`, `sfx Zap`, `music Theme`, still demonstrating builtin fallback via `play coin`. Total: 424 tests passing (381 unit + 43 integration), clippy clean, fmt clean, all 19 examples compile. https://claude.ai/code/session_015WfaDttE3DpWn9rpyfpQd8
2026-04-13 01:10:21 +00:00
let sfx = assets::resolve_sfx(&program).expect("sfx resolution should succeed");
let music = assets::resolve_music(&program).expect("music resolution should succeed");
palette/background: first-class declarations with reset-time load and runtime swaps Re-adds `palette Name { colors: [...] }` and `background Name { tiles: [...], attributes: [...] }` as first-class declarations, plus `set_palette Name` and `load_background Name` statements for runtime swaps. Unlike the previous iteration that quietly no-op'd, this one is fully wired through the pipeline and its behavior is pinned by both unit tests and an emulator golden. Pipeline: - Lexer: re-adds `palette`, `background`, `set_palette`, `load_background` keywords and tokenizes them. - AST: `PaletteDecl` (name + 1..=32 colour bytes) and `BackgroundDecl` (name + 0..=960 tile bytes + 0..=64 attribute bytes) live in `Program`. `Statement::SetPalette` and `Statement::LoadBackground` name-reference these declarations. - Parser: `palette Name { colors: [...] }` / `background Name { tiles: [...], attributes: [...] }` blocks and their statement forms parse via the existing byte-array helper. - Analyzer: validates colour indices ($00-$3F), palette length (<=32), nametable length (<=960), attribute length (<=64), and duplicate decl names. `set_palette` / `load_background` targets must reference a declared name (E0502 otherwise). When a program declares palette or background, the analyzer bumps the user zero-page allocator's starting address from `$10` to `$18` to reserve `$11-$17` for the runtime update handshake — programs that don't use the feature keep the old layout so their emulator goldens stay byte-exact. - Assets: `PaletteData` and `BackgroundData` resolve declarations into zero-padded fixed-size blobs (32 / 960 / 64 bytes) and expose `label()` / `tiles_label()` / `attrs_label()` for codegen to reference. - IR: new `IrOp::SetPalette(String)` and `IrOp::LoadBackground(String)`; lowering forwards the names verbatim. - Codegen: `gen_set_palette` writes the palette label pointer into ZP `$12/$13` and ORs bit 0 into the update flags at `$11`; `gen_load_background` does the same for tile/attribute pointers at `$14/$15/$16/$17` with bit 1. Both emit a `__ppu_update_used` marker so the linker splices in the NMI apply helper only when the feature is actually used. - Runtime: `gen_initial_palette_load` and `gen_initial_background_load` write the first declared palette/background at reset time (before rendering is enabled, where PPU writes are safe). `gen_nmi(has_ppu_updates)` takes a new flag; when true it splices `gen_ppu_update_apply` at the top of the NMI body, which checks the `$11` flags byte and copies pending palette / nametable data to `$3F00` / `$2000` inside vblank. All helpers use only ZP $02/$03 as scratch at reset time and never clobber ZP slots live across NMI. - Linker: new `link_banked_with_ppu` takes slice of `PaletteData` / `BackgroundData`; splices each blob as a labelled data block in PRG ROM, picks the first-declared as the reset-time load target, enables background rendering automatically when a background is declared, and threads `has_ppu_updates` into `gen_nmi`. Old `link_banked` remains as a thin wrapper for callers without palette/background data so existing tests don't shift. Tests: - Lexer: tokenization of the 4 new keywords (single added test case). - Parser: 5 new tests for `palette` / `background` decls with and without attributes, plus `set_palette` / `load_background` statements. - Analyzer: 9 new tests covering acceptance of declared palettes/backgrounds, E0502 for unknown names, E0201 for out-of-range NES colors and oversized blobs, E0501 for duplicate names, and the zero-page-layout guard (palette/bg decls bump ZP start; no decls keeps it at $10). - Resolver: 3 new tests for zero-padding, truncation of oversized decls, and label derivation. - IR: 2 new lowering tests for `set_palette` and `load_background`. - Integration: 5 new tests — blob contents spliced verbatim into PRG, `STA $12` / `STA $14` emitted by set_palette / load_background codegen, and a regression guard that programs without palette/background still land user vars at $10. - Emulator: new `examples/palette_and_background.ne` driven by a frame counter that toggles between `CoolBlues` / `WarmReds` and `TitleScreen` / `StageOne` every 90 frames. Golden PNG and audio hash checked in under `tests/emulator/goldens/` and verified via `node run_examples.mjs` — rendered image shows the blue `CoolBlues` palette with the nametable populated from `TitleScreen`. Docs: - `README.md` adds the feature to the headline list and the example table. - `docs/language-guide.md` restores the palette/background sections with the full 32-byte layout table and `set_palette` / `load_background` statement references. - `docs/future-work.md` replaces the "removed as dead code" entry with the remaining gaps (PNG-sourced palette and nametable assets, cross-vblank large background updates, memory-map reporting). - `spec.md` restores the grammar productions and usage examples. - `examples/README.md` lists the new demo. All 497 unit + integration tests pass. Clippy clean. All 21 emulator goldens match after the update pass. https://claude.ai/code/session_012fKB251HvEUQwG3tizFyqt
2026-04-13 11:11:33 +00:00
let palettes = assets::resolve_palettes(&program);
let backgrounds = assets::resolve_backgrounds(&program);
audio: complete the subsystem — asset pipeline, user decls, tracker-style driver The audio subsystem was a sketch: `play name` / `start_music name` / `stop_music` parsed, lowered, and emitted a few hardcoded register writes from a builtin name table. No user-declared effects, no per-frame envelope, no note streams, no real engine. This flesh-out brings audio up to the quality bar of the rest of the compiler (sprites, palettes, bank switching, scanline IRQ, etc.) with a full data-driven pipeline: ## Asset pipeline (new `src/assets/audio.rs`) - `sfx Name { duty, pitch, volume }` blocks compile into per-frame pulse-1 envelopes. Pitch/volume arrays must match in length; each entry is one NMI's worth of `$4000` data. - `music Name { duty, volume, repeat, notes }` blocks compile into flat `(pitch, duration)` streams for pulse 2. Pitch 0 is a rest, 1-60 indexes a builtin period table covering C1-B5. - `resolve_sfx` / `resolve_music` walk the program for `play` / `start_music` references and append builtin fallbacks for any name that isn't user-declared — so `play coin` still works without a `sfx Coin { ... }` block. - Builtin effects (coin, jump, hit, click, cancel, shoot, step) and tracks (theme, battle, victory, gameover) synthesize through the same compile path as user decls — one data model, one driver. ## Runtime engine (`src/runtime/mod.rs`) - `gen_audio_tick()` walks both channels every NMI: reads one envelope byte through `(ZP_SFX_PTR),Y` -> writes `$4000`, advances ptr, mutes on zero sentinel. Music decrements the note counter, advances to the next `(pitch, dur)` pair on zero, looks up the period through `(__period_table),Y`, loops on `0xFF 0xFF`. - `gen_period_table()` emits a 60-entry equal-tempered table (A4 = 440 Hz, NTSC 1.789773 MHz CPU clock) with length-counter load bits pre-baked into each high byte. - `gen_data_block()` emits a label + raw-bytes pseudo pair so user sfx/music data can be spliced into PRG with regular labels that the two-pass assembler resolves. - New ZP layout: `$05/$06` music loop base, `$07` music state (duty/volume/loop/active), `$0C-$0F` sfx and music pointers. ## IR codegen (`src/codegen/ir_codegen.rs`) - `with_audio(sfx, music)` registers compile-time trigger constants per blob name. - `gen_play_sfx` emits: write period to `$4002`/`$4003`, load envelope pointer into `ZP_SFX_PTR` via SymbolLo/SymbolHi of `__sfx_<name>`, mark the sfx counter active. - `gen_start_music` stamps the header byte into `ZP_MUSIC_STATE` with the active bit OR'd in, seeds both ptr and loop base from `__music_<name>`, primes the duration counter. - `gen_stop_music` mutes pulse 2 and clears state. ## Linker (`src/linker/mod.rs`) - New `link_with_all_assets(user_code, sprites, sfx, music)` path that splices driver body, period table, and each sfx/music data blob into PRG — all guarded on the `__audio_used` marker so silent programs pay zero ROM cost. ## Assembler (`src/asm/opcodes.rs`, `src/asm/mod.rs`) - New `AddressingMode::Bytes(Vec<u8>)` variant for raw-data pseudo-instructions. `NOP+Bytes(v)` emits the payload verbatim, letting the linker splice ROM data tables into a code section and still have `Label` / `SymbolLo` / `SymbolHi` fixups resolve correctly in the same assembly pass. ## Analyzer - `play` / `start_music` now validate the name against user decls and builtin tables. Unknown names emit E0505 with a helpful list of builtins — previously a typo would silently compile to no-op. ## Parser - New `sfx_decl` / `music_decl` grammar with property-style configuration. Strict validation: duty 0-3, volume 0-15, pitch arrays must match volume length, music notes must come in pairs, pitch 0-60, duration ≥ 1. ## Tests +170 new tests across every layer: - `src/assets/audio.rs`: 17 tests (compile, resolve, builtins, shadowing, label sanitation, nested reference walks) - `src/parser/tests.rs`: 13 tests (valid/invalid sfx + music declarations, property validation, play/start_music/stop_music) - `src/analyzer/tests.rs`: 7 tests (builtin acceptance, user decl acceptance, unknown-name rejection) - `src/runtime/tests.rs`: 10 tests (audio tick labels, RTS end, $4000 write, $4004 mute, period table assembly, A4 = 440 Hz, length counter bits, data block verbatim emit) - `src/linker/tests.rs`: 4 tests (sfx/music blob placement, pointer resolution, elision when unused) - `src/codegen/ir_codegen.rs`: rewrote the 4 existing audio tests to match the new data-driven contract - `tests/integration_test.rs`: 4 end-to-end tests including a user-declared `sfx` + `music` program that verifies bytes land in PRG ROM at the right addresses ## Docs - New Audio section in `docs/language-guide.md` with syntax reference, builtin tables, and an explanation of how the driver works at compile and run time. - `docs/architecture.md` updated to reflect the real audio pipeline instead of the old "audio import stubs" stub. - `docs/future-work.md` moves audio from "status: minimal" to "status: full subsystem" with a narrower list of follow-up work (triangle/noise/DMC channels, NSF/FTM imports, richer envelopes). - `examples/audio_demo.ne` rewritten to showcase user-declared `sfx LongCoin`, `sfx Zap`, `music Theme`, still demonstrating builtin fallback via `play coin`. Total: 424 tests passing (381 unit + 43 integration), clippy clean, fmt clean, all 19 examples compile. https://claude.ai/code/session_015WfaDttE3DpWn9rpyfpQd8
2026-04-13 01:10:21 +00:00
let codegen = IrCodeGen::new(&analysis.var_allocations, &ir_program)
.with_sprites(&sprites)
.with_audio(&sfx, &music);
Remove the legacy AST codegen — IR path is canonical now The `--use-ast` path through `src/codegen/mod.rs` was a strictly inferior subset of the IR codegen. Building every example with `--use-ast` through the jsnes harness: - `arrays_and_functions` — fully black (array init + function return values + OAM-in-loop all broken) - `structs_enums_for` — fully black (struct literal is a no-op, all fields stay at 0) - `inline_asm_demo` — fully black - `bitwise_ops`, `loop_break_continue` — below sprite floors (static `next_oam_slot` bug B) - `match_demo` — panics at compile time with `branch offset 153 out of range` (AST's if/else-chain desugaring of `match` emits short branches that can't reach the far arms in a multi-arm match) Six of fourteen examples are non-functional under `--use-ast`. The other eight happen to fall inside the subset AST handles (no arrays, no structs, no function return values, no multi-sprite loops, no long match chains). `docs/future-work.md` already listed "Once working, delete the AST-based codegen entirely" as the intended direction. It's working, so this commit does the deletion. What's removed: - The `CodeGen` struct, its impl block, and every helper in `src/codegen/mod.rs` (the AST codegen body) — ~1150 lines. The file is now a module header that re-exports `IrCodeGen`. - `src/codegen/tests.rs` — 15 AST-specific instruction-pattern tests. Every feature they covered has an equivalent test in `src/codegen/ir_codegen.rs::{tests,more_tests}` already. - The `--use-ast` CLI flag and its branch in `src/main.rs`. - `compile_with_ir_codegen` in `tests/integration_test.rs` — `compile()` now does what it did, so they merged. All 40 integration tests go through the IR path. - Outdated sections in `docs/future-work.md` that described the IR codegen as "not yet implemented" and listed AST codegen gaps as priority work. What's kept: - `src/codegen/ir_codegen.rs` — the real codegen. - `src/codegen/peephole.rs` — post-codegen cleanup pass, now run unconditionally from `main.rs`. Test plan: - `cargo test --release` — 313 unit + 37 integration tests pass (was 328 + 37; the 15 dropped are the deleted AST-specific tests). - `cargo fmt --check` clean. - `cargo clippy --release --all-targets -- -D warnings` clean. - `node tests/emulator/run_examples.mjs` — 14/14 ROMs render above their per-example nonBlack floors. - The one tightening: `sprite_resolution_uses_tile_index` was asserting on the old static-slot encoding (`A9 01 8D 01 02`). Updated to the cursor-based form (`A9 01 99 01 02`, i.e. STA AbsoluteY). Net diff: 1581 deletions, 62 insertions. https://claude.ai/code/session_014Z5y3Q9krLcAxYpZQJhZ5V
2026-04-12 20:37:59 +00:00
let mut instructions = codegen.generate(&ir_program);
nescript::codegen::peephole::optimize(&mut instructions);
let linker = Linker::with_mapper(program.game.mirroring, program.game.mapper);
banks: implement multi-bank PRG layout and bank-switching runtime Prior to this commit the linker always shipped a single 16 KB PRG bank regardless of the declared mapper, so the README's MMC1/UxROM/ MMC3 support was aspirational. This commit gives the three banked mappers a real multi-bank ROM layout: * RomBuilder.set_prg_banks() writes any number of 16 KB banks back-to-back so the iNES header reflects the true PRG size. * Linker.link_banked() places switchable banks first, fixed bank last, so the fixed bank maps to $C000-$FFFF (the address window where vectors and the runtime live). * runtime::gen_mapper_init() emits reset-time mapper config: MMC1 serial-writes a control-register value that pins the last bank at $C000 with the correct mirroring, UxROM relies on the power-on default, MMC3 writes the $8000/$8001/$A000/$E000 registers to get a known PRG and mirroring state. * runtime::gen_bank_select() is a mapper-specific subroutine (callable with the target bank in A) that maps any physical bank to $8000-$BFFF. * runtime::gen_bank_trampoline() generates a cross-bank call stub in the fixed bank that saves the caller's bank, switches, JSRs the target, and restores the fixed bank. * The CLI and integration helper thread declared `bank X: prg` declarations through to the linker so MMC1/UxROM/MMC3 programs actually produce multi-bank ROMs. Coverage: * Runtime unit tests (18 new): mapper init patterns for every supported mapper, bank-select signatures, trampoline dispatch order, UxROM bus-conflict table contents. * RomBuilder tests (6 new): multi-bank layout, padding, byte-level fidelity, per-bank size validation, legacy single-bank fallback. * Linker tests (13 new): multi-bank ROM sizes across MMC1/ UxROM/MMC3, fixed-bank placement, switchable-bank payload fidelity, bank-select subroutine detection, NROM rejection of switchable banks. * Integration e2e tests (16 new): compile real .ne sources through the full pipeline and assert on iNES headers, mapper init signatures in the fixed bank, vector locations, and a regression check against `examples/mmc1_banked.ne`. Total: 474 tests pass under `cargo test` with `RUSTFLAGS="-D warnings"`. https://claude.ai/code/session_01UCressA5e8k1XsuoJYLav2
2026-04-13 01:50:51 +00:00
let switchable_banks: Vec<PrgBank> = program
.banks
.iter()
.filter(|b| b.bank_type == BankType::Prg)
.map(|b| PrgBank::empty(&b.name))
.collect();
palette/background: first-class declarations with reset-time load and runtime swaps Re-adds `palette Name { colors: [...] }` and `background Name { tiles: [...], attributes: [...] }` as first-class declarations, plus `set_palette Name` and `load_background Name` statements for runtime swaps. Unlike the previous iteration that quietly no-op'd, this one is fully wired through the pipeline and its behavior is pinned by both unit tests and an emulator golden. Pipeline: - Lexer: re-adds `palette`, `background`, `set_palette`, `load_background` keywords and tokenizes them. - AST: `PaletteDecl` (name + 1..=32 colour bytes) and `BackgroundDecl` (name + 0..=960 tile bytes + 0..=64 attribute bytes) live in `Program`. `Statement::SetPalette` and `Statement::LoadBackground` name-reference these declarations. - Parser: `palette Name { colors: [...] }` / `background Name { tiles: [...], attributes: [...] }` blocks and their statement forms parse via the existing byte-array helper. - Analyzer: validates colour indices ($00-$3F), palette length (<=32), nametable length (<=960), attribute length (<=64), and duplicate decl names. `set_palette` / `load_background` targets must reference a declared name (E0502 otherwise). When a program declares palette or background, the analyzer bumps the user zero-page allocator's starting address from `$10` to `$18` to reserve `$11-$17` for the runtime update handshake — programs that don't use the feature keep the old layout so their emulator goldens stay byte-exact. - Assets: `PaletteData` and `BackgroundData` resolve declarations into zero-padded fixed-size blobs (32 / 960 / 64 bytes) and expose `label()` / `tiles_label()` / `attrs_label()` for codegen to reference. - IR: new `IrOp::SetPalette(String)` and `IrOp::LoadBackground(String)`; lowering forwards the names verbatim. - Codegen: `gen_set_palette` writes the palette label pointer into ZP `$12/$13` and ORs bit 0 into the update flags at `$11`; `gen_load_background` does the same for tile/attribute pointers at `$14/$15/$16/$17` with bit 1. Both emit a `__ppu_update_used` marker so the linker splices in the NMI apply helper only when the feature is actually used. - Runtime: `gen_initial_palette_load` and `gen_initial_background_load` write the first declared palette/background at reset time (before rendering is enabled, where PPU writes are safe). `gen_nmi(has_ppu_updates)` takes a new flag; when true it splices `gen_ppu_update_apply` at the top of the NMI body, which checks the `$11` flags byte and copies pending palette / nametable data to `$3F00` / `$2000` inside vblank. All helpers use only ZP $02/$03 as scratch at reset time and never clobber ZP slots live across NMI. - Linker: new `link_banked_with_ppu` takes slice of `PaletteData` / `BackgroundData`; splices each blob as a labelled data block in PRG ROM, picks the first-declared as the reset-time load target, enables background rendering automatically when a background is declared, and threads `has_ppu_updates` into `gen_nmi`. Old `link_banked` remains as a thin wrapper for callers without palette/background data so existing tests don't shift. Tests: - Lexer: tokenization of the 4 new keywords (single added test case). - Parser: 5 new tests for `palette` / `background` decls with and without attributes, plus `set_palette` / `load_background` statements. - Analyzer: 9 new tests covering acceptance of declared palettes/backgrounds, E0502 for unknown names, E0201 for out-of-range NES colors and oversized blobs, E0501 for duplicate names, and the zero-page-layout guard (palette/bg decls bump ZP start; no decls keeps it at $10). - Resolver: 3 new tests for zero-padding, truncation of oversized decls, and label derivation. - IR: 2 new lowering tests for `set_palette` and `load_background`. - Integration: 5 new tests — blob contents spliced verbatim into PRG, `STA $12` / `STA $14` emitted by set_palette / load_background codegen, and a regression guard that programs without palette/background still land user vars at $10. - Emulator: new `examples/palette_and_background.ne` driven by a frame counter that toggles between `CoolBlues` / `WarmReds` and `TitleScreen` / `StageOne` every 90 frames. Golden PNG and audio hash checked in under `tests/emulator/goldens/` and verified via `node run_examples.mjs` — rendered image shows the blue `CoolBlues` palette with the nametable populated from `TitleScreen`. Docs: - `README.md` adds the feature to the headline list and the example table. - `docs/language-guide.md` restores the palette/background sections with the full 32-byte layout table and `set_palette` / `load_background` statement references. - `docs/future-work.md` replaces the "removed as dead code" entry with the remaining gaps (PNG-sourced palette and nametable assets, cross-vblank large background updates, memory-map reporting). - `spec.md` restores the grammar productions and usage examples. - `examples/README.md` lists the new demo. All 497 unit + integration tests pass. Clippy clean. All 21 emulator goldens match after the update pass. https://claude.ai/code/session_012fKB251HvEUQwG3tizFyqt
2026-04-13 11:11:33 +00:00
linker.link_banked_with_ppu(
&instructions,
&sprites,
&sfx,
&music,
&palettes,
&backgrounds,
&switchable_banks,
)
}
#[test]
fn sprite_resolution_uses_tile_index() {
// The Player sprite has 16 unique bytes of CHR data. Because tile index 0
// is reserved for the built-in smiley, the compiler should place Player
// at tile index 1 and `draw Player` should store that tile index in OAM.
//
// We check this in two ways:
// 1. The CHR ROM contains Player's bytes at tile 1 (offset 16).
// 2. The PRG ROM contains the immediate-load sequence `A9 01 8D 01 02`
// (LDA #$01 ; STA $0201) — writing tile index 1 into OAM byte 1.
let source = r#"
game "SpriteTile" { mapper: NROM }
sprite Player {
chr: [0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17,
0x18, 0x19, 0x1A, 0x1B, 0x1C, 0x1D, 0x1E, 0x1F]
}
var px: u8 = 128
var py: u8 = 120
state Title {
on frame {
draw Player at: (px, py)
}
}
start Title
"#;
let rom_data = compile(source);
// CHR ROM begins right after PRG ROM (16 header + 16384 PRG).
let chr_start = 16 + 16384;
// Tile 1 lives at CHR offset 16 (16 bytes per tile).
let tile1 = &rom_data[chr_start + 16..chr_start + 32];
assert_eq!(
tile1,
&[
0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, 0x18, 0x19, 0x1A, 0x1B, 0x1C, 0x1D,
0x1E, 0x1F
],
"Player sprite CHR bytes should be placed at tile index 1",
);
// The default smiley tile at index 0 should still be non-zero (untouched).
let tile0 = &rom_data[chr_start..chr_start + 16];
assert_ne!(
tile0, &[0u8; 16],
"tile 0 should still contain the default smiley",
);
Remove the legacy AST codegen — IR path is canonical now The `--use-ast` path through `src/codegen/mod.rs` was a strictly inferior subset of the IR codegen. Building every example with `--use-ast` through the jsnes harness: - `arrays_and_functions` — fully black (array init + function return values + OAM-in-loop all broken) - `structs_enums_for` — fully black (struct literal is a no-op, all fields stay at 0) - `inline_asm_demo` — fully black - `bitwise_ops`, `loop_break_continue` — below sprite floors (static `next_oam_slot` bug B) - `match_demo` — panics at compile time with `branch offset 153 out of range` (AST's if/else-chain desugaring of `match` emits short branches that can't reach the far arms in a multi-arm match) Six of fourteen examples are non-functional under `--use-ast`. The other eight happen to fall inside the subset AST handles (no arrays, no structs, no function return values, no multi-sprite loops, no long match chains). `docs/future-work.md` already listed "Once working, delete the AST-based codegen entirely" as the intended direction. It's working, so this commit does the deletion. What's removed: - The `CodeGen` struct, its impl block, and every helper in `src/codegen/mod.rs` (the AST codegen body) — ~1150 lines. The file is now a module header that re-exports `IrCodeGen`. - `src/codegen/tests.rs` — 15 AST-specific instruction-pattern tests. Every feature they covered has an equivalent test in `src/codegen/ir_codegen.rs::{tests,more_tests}` already. - The `--use-ast` CLI flag and its branch in `src/main.rs`. - `compile_with_ir_codegen` in `tests/integration_test.rs` — `compile()` now does what it did, so they merged. All 40 integration tests go through the IR path. - Outdated sections in `docs/future-work.md` that described the IR codegen as "not yet implemented" and listed AST codegen gaps as priority work. What's kept: - `src/codegen/ir_codegen.rs` — the real codegen. - `src/codegen/peephole.rs` — post-codegen cleanup pass, now run unconditionally from `main.rs`. Test plan: - `cargo test --release` — 313 unit + 37 integration tests pass (was 328 + 37; the 15 dropped are the deleted AST-specific tests). - `cargo fmt --check` clean. - `cargo clippy --release --all-targets -- -D warnings` clean. - `node tests/emulator/run_examples.mjs` — 14/14 ROMs render above their per-example nonBlack floors. - The one tightening: `sprite_resolution_uses_tile_index` was asserting on the old static-slot encoding (`A9 01 8D 01 02`). Updated to the cursor-based form (`A9 01 99 01 02`, i.e. STA AbsoluteY). Net diff: 1581 deletions, 62 insertions. https://claude.ai/code/session_014Z5y3Q9krLcAxYpZQJhZ5V
2026-04-12 20:37:59 +00:00
// In PRG ROM, look for `LDA #$01 ; STA $0201,Y` which writes
// the Player's tile index (1) into the tile-index byte of the
// current OAM slot (the slot is computed at runtime via the
// OAM cursor in Y). The STA AbsoluteY opcode is $99.
let prg = &rom_data[16..16 + 16384];
Remove the legacy AST codegen — IR path is canonical now The `--use-ast` path through `src/codegen/mod.rs` was a strictly inferior subset of the IR codegen. Building every example with `--use-ast` through the jsnes harness: - `arrays_and_functions` — fully black (array init + function return values + OAM-in-loop all broken) - `structs_enums_for` — fully black (struct literal is a no-op, all fields stay at 0) - `inline_asm_demo` — fully black - `bitwise_ops`, `loop_break_continue` — below sprite floors (static `next_oam_slot` bug B) - `match_demo` — panics at compile time with `branch offset 153 out of range` (AST's if/else-chain desugaring of `match` emits short branches that can't reach the far arms in a multi-arm match) Six of fourteen examples are non-functional under `--use-ast`. The other eight happen to fall inside the subset AST handles (no arrays, no structs, no function return values, no multi-sprite loops, no long match chains). `docs/future-work.md` already listed "Once working, delete the AST-based codegen entirely" as the intended direction. It's working, so this commit does the deletion. What's removed: - The `CodeGen` struct, its impl block, and every helper in `src/codegen/mod.rs` (the AST codegen body) — ~1150 lines. The file is now a module header that re-exports `IrCodeGen`. - `src/codegen/tests.rs` — 15 AST-specific instruction-pattern tests. Every feature they covered has an equivalent test in `src/codegen/ir_codegen.rs::{tests,more_tests}` already. - The `--use-ast` CLI flag and its branch in `src/main.rs`. - `compile_with_ir_codegen` in `tests/integration_test.rs` — `compile()` now does what it did, so they merged. All 40 integration tests go through the IR path. - Outdated sections in `docs/future-work.md` that described the IR codegen as "not yet implemented" and listed AST codegen gaps as priority work. What's kept: - `src/codegen/ir_codegen.rs` — the real codegen. - `src/codegen/peephole.rs` — post-codegen cleanup pass, now run unconditionally from `main.rs`. Test plan: - `cargo test --release` — 313 unit + 37 integration tests pass (was 328 + 37; the 15 dropped are the deleted AST-specific tests). - `cargo fmt --check` clean. - `cargo clippy --release --all-targets -- -D warnings` clean. - `node tests/emulator/run_examples.mjs` — 14/14 ROMs render above their per-example nonBlack floors. - The one tightening: `sprite_resolution_uses_tile_index` was asserting on the old static-slot encoding (`A9 01 8D 01 02`). Updated to the cursor-based form (`A9 01 99 01 02`, i.e. STA AbsoluteY). Net diff: 1581 deletions, 62 insertions. https://claude.ai/code/session_014Z5y3Q9krLcAxYpZQJhZ5V
2026-04-12 20:37:59 +00:00
let pattern = [0xA9u8, 0x01, 0x99, 0x01, 0x02];
assert!(
prg.windows(pattern.len()).any(|w| w == pattern),
Remove the legacy AST codegen — IR path is canonical now The `--use-ast` path through `src/codegen/mod.rs` was a strictly inferior subset of the IR codegen. Building every example with `--use-ast` through the jsnes harness: - `arrays_and_functions` — fully black (array init + function return values + OAM-in-loop all broken) - `structs_enums_for` — fully black (struct literal is a no-op, all fields stay at 0) - `inline_asm_demo` — fully black - `bitwise_ops`, `loop_break_continue` — below sprite floors (static `next_oam_slot` bug B) - `match_demo` — panics at compile time with `branch offset 153 out of range` (AST's if/else-chain desugaring of `match` emits short branches that can't reach the far arms in a multi-arm match) Six of fourteen examples are non-functional under `--use-ast`. The other eight happen to fall inside the subset AST handles (no arrays, no structs, no function return values, no multi-sprite loops, no long match chains). `docs/future-work.md` already listed "Once working, delete the AST-based codegen entirely" as the intended direction. It's working, so this commit does the deletion. What's removed: - The `CodeGen` struct, its impl block, and every helper in `src/codegen/mod.rs` (the AST codegen body) — ~1150 lines. The file is now a module header that re-exports `IrCodeGen`. - `src/codegen/tests.rs` — 15 AST-specific instruction-pattern tests. Every feature they covered has an equivalent test in `src/codegen/ir_codegen.rs::{tests,more_tests}` already. - The `--use-ast` CLI flag and its branch in `src/main.rs`. - `compile_with_ir_codegen` in `tests/integration_test.rs` — `compile()` now does what it did, so they merged. All 40 integration tests go through the IR path. - Outdated sections in `docs/future-work.md` that described the IR codegen as "not yet implemented" and listed AST codegen gaps as priority work. What's kept: - `src/codegen/ir_codegen.rs` — the real codegen. - `src/codegen/peephole.rs` — post-codegen cleanup pass, now run unconditionally from `main.rs`. Test plan: - `cargo test --release` — 313 unit + 37 integration tests pass (was 328 + 37; the 15 dropped are the deleted AST-specific tests). - `cargo fmt --check` clean. - `cargo clippy --release --all-targets -- -D warnings` clean. - `node tests/emulator/run_examples.mjs` — 14/14 ROMs render above their per-example nonBlack floors. - The one tightening: `sprite_resolution_uses_tile_index` was asserting on the old static-slot encoding (`A9 01 8D 01 02`). Updated to the cursor-based form (`A9 01 99 01 02`, i.e. STA AbsoluteY). Net diff: 1581 deletions, 62 insertions. https://claude.ai/code/session_014Z5y3Q9krLcAxYpZQJhZ5V
2026-04-12 20:37:59 +00:00
"PRG ROM should contain LDA #$01 ; STA $0201,Y for draw Player",
);
}
Implement codegen for state dispatch, functions, arrays, math, scroll State machine dispatch: - Assign each state a numeric index, store in ZP $03 - Main loop dispatch table: CMP + BNE + JMP trampoline pattern (avoids branch range limits for large programs) - on_enter/on_exit handlers generated as JSR targets - Transition statement writes state index + JSR enter/exit handlers Function calls: - Function bodies emitted as labeled subroutines with RTS - Statement::Call generates parameter passing via ZP + JSR - Statement::Return generates RTS (with value in A if present) - Parameter slots at ZP $04-$07 Break/continue: - Loop stack tracks continue/break label pairs - Break generates JMP to break_label - Continue generates JMP to continue_label - While and Loop push/pop the stack Array indexing: - LValue::ArrayIndex generates TAX + STA absolute,X - Expr::ArrayIndex generates TAX + LDA absolute,X / ZP,X - Compound array assignments (+=, -=, &=, |=, ^=) load-modify-store Scroll: - scroll(x, y) writes to PPU $2005 twice (X then Y) Math: - Multiply generates JSR __multiply (shift-and-add routine) - Divide generates JSR __divide (restoring division) - Modulo loads remainder from $03 after divide - ShiftLeft generates ASL A, ShiftRight generates LSR A - Math routines wired into linker Error validations: - E0203 for assignment to const variables - Break/continue outside loop detection (in_loop tracking) 233 tests (8 new codegen + 2 analyzer + 2 integration), all passing. https://claude.ai/code/session_01W6eQFStA66EuMKHUFo2rx3
2026-04-12 02:04:49 +00:00
#[test]
fn program_with_arrays_and_math() {
let source = r#"
game "ArrayMath" { mapper: NROM }
var arr: u8[4] = [10, 20, 30, 40]
var idx: u8 = 0
var result: u8 = 0
on frame {
result = arr[idx] * 2
idx += 1
}
start Main
"#;
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn program_with_mmc1() {
let source = r#"
game "MMC1 Game" { mapper: MMC1 }
var px: u8 = 128
on frame {
if button.right { px += 2 }
}
start Main
"#;
let rom_data = compile_with_mapper(source);
let info = rom::validate_ines(&rom_data).expect("should be valid iNES");
assert_eq!(info.mapper, 1, "should be MMC1 (mapper 1)");
}
Implement IR-based code generator (--use-ir) New src/codegen/ir_codegen.rs walks IrProgram and emits 6502 instructions. This enables optimizer passes to actually affect the output ROM. Design: - Each IR temp gets a zero-page slot at $80 + temp_index - Functions reset the temp counter at entry (temps don't outlive functions) - Globals map by name to their analyzer-assigned zero-page addresses - Operands are loaded into A, computed, stored back to the dest slot Handles all IrOp variants: - LoadImm, LoadVar, StoreVar (basic loads/stores) - Add/Sub (CLC+ADC / SEC+SBC) - Mul (JSR __multiply runtime routine) - And/Or/Xor (zero-page operands) - ShiftLeft/ShiftRight (repeated ASL/LSR) - Negate/Complement (EOR #$FF + optional two's complement) - CmpEq/Ne/Lt/Gt/LtEq/GtEq (CMP + conditional branch + 0/1) - ArrayLoad/ArrayStore (TAX + ZeroPageX/AbsoluteX) - Call (ZP param passing + JSR) - DrawSprite (OAM slot 0 write, uses sprite_tiles map) - ReadInput (LDA $01, P1 input) - WaitFrame (poll frame flag at $00) All terminators: - Jump (JMP to block label) - Branch (LDA temp + BNE true / JMP false) - Return (optional value in A + RTS) - Unreachable (BRK) IR lowering fixes: - ReadInput now has a destination IrTemp (was a side-effect-only op) - ButtonRead uses the proper input temp instead of uninitialized register - Logical AND/OR use new emit_move helper (OR with zero) instead of bogus raw VarId for path merging CLI: - New --use-ir flag on `build` subcommand to opt in to IR codegen - Default remains AST codegen (for now); IR codegen is experimental All 7 examples compile through the IR pipeline and produce valid iNES ROMs. Tests: 266 total (7 new ir_codegen unit + 2 new integration). https://claude.ai/code/session_01W6eQFStA66EuMKHUFo2rx3
2026-04-12 10:23:43 +00:00
// ── IR Codegen Tests ──
Remove the legacy AST codegen — IR path is canonical now The `--use-ast` path through `src/codegen/mod.rs` was a strictly inferior subset of the IR codegen. Building every example with `--use-ast` through the jsnes harness: - `arrays_and_functions` — fully black (array init + function return values + OAM-in-loop all broken) - `structs_enums_for` — fully black (struct literal is a no-op, all fields stay at 0) - `inline_asm_demo` — fully black - `bitwise_ops`, `loop_break_continue` — below sprite floors (static `next_oam_slot` bug B) - `match_demo` — panics at compile time with `branch offset 153 out of range` (AST's if/else-chain desugaring of `match` emits short branches that can't reach the far arms in a multi-arm match) Six of fourteen examples are non-functional under `--use-ast`. The other eight happen to fall inside the subset AST handles (no arrays, no structs, no function return values, no multi-sprite loops, no long match chains). `docs/future-work.md` already listed "Once working, delete the AST-based codegen entirely" as the intended direction. It's working, so this commit does the deletion. What's removed: - The `CodeGen` struct, its impl block, and every helper in `src/codegen/mod.rs` (the AST codegen body) — ~1150 lines. The file is now a module header that re-exports `IrCodeGen`. - `src/codegen/tests.rs` — 15 AST-specific instruction-pattern tests. Every feature they covered has an equivalent test in `src/codegen/ir_codegen.rs::{tests,more_tests}` already. - The `--use-ast` CLI flag and its branch in `src/main.rs`. - `compile_with_ir_codegen` in `tests/integration_test.rs` — `compile()` now does what it did, so they merged. All 40 integration tests go through the IR path. - Outdated sections in `docs/future-work.md` that described the IR codegen as "not yet implemented" and listed AST codegen gaps as priority work. What's kept: - `src/codegen/ir_codegen.rs` — the real codegen. - `src/codegen/peephole.rs` — post-codegen cleanup pass, now run unconditionally from `main.rs`. Test plan: - `cargo test --release` — 313 unit + 37 integration tests pass (was 328 + 37; the 15 dropped are the deleted AST-specific tests). - `cargo fmt --check` clean. - `cargo clippy --release --all-targets -- -D warnings` clean. - `node tests/emulator/run_examples.mjs` — 14/14 ROMs render above their per-example nonBlack floors. - The one tightening: `sprite_resolution_uses_tile_index` was asserting on the old static-slot encoding (`A9 01 8D 01 02`). Updated to the cursor-based form (`A9 01 99 01 02`, i.e. STA AbsoluteY). Net diff: 1581 deletions, 62 insertions. https://claude.ai/code/session_014Z5y3Q9krLcAxYpZQJhZ5V
2026-04-12 20:37:59 +00:00
//
// These tests exercise specific end-to-end IR codegen behavior.
// They all use the top-level `compile()` helper now that it runs
// the full IR pipeline — there's no longer a separate legacy path
// to compare against.
Implement IR-based code generator (--use-ir) New src/codegen/ir_codegen.rs walks IrProgram and emits 6502 instructions. This enables optimizer passes to actually affect the output ROM. Design: - Each IR temp gets a zero-page slot at $80 + temp_index - Functions reset the temp counter at entry (temps don't outlive functions) - Globals map by name to their analyzer-assigned zero-page addresses - Operands are loaded into A, computed, stored back to the dest slot Handles all IrOp variants: - LoadImm, LoadVar, StoreVar (basic loads/stores) - Add/Sub (CLC+ADC / SEC+SBC) - Mul (JSR __multiply runtime routine) - And/Or/Xor (zero-page operands) - ShiftLeft/ShiftRight (repeated ASL/LSR) - Negate/Complement (EOR #$FF + optional two's complement) - CmpEq/Ne/Lt/Gt/LtEq/GtEq (CMP + conditional branch + 0/1) - ArrayLoad/ArrayStore (TAX + ZeroPageX/AbsoluteX) - Call (ZP param passing + JSR) - DrawSprite (OAM slot 0 write, uses sprite_tiles map) - ReadInput (LDA $01, P1 input) - WaitFrame (poll frame flag at $00) All terminators: - Jump (JMP to block label) - Branch (LDA temp + BNE true / JMP false) - Return (optional value in A + RTS) - Unreachable (BRK) IR lowering fixes: - ReadInput now has a destination IrTemp (was a side-effect-only op) - ButtonRead uses the proper input temp instead of uninitialized register - Logical AND/OR use new emit_move helper (OR with zero) instead of bogus raw VarId for path merging CLI: - New --use-ir flag on `build` subcommand to opt in to IR codegen - Default remains AST codegen (for now); IR codegen is experimental All 7 examples compile through the IR pipeline and produce valid iNES ROMs. Tests: 266 total (7 new ir_codegen unit + 2 new integration). https://claude.ai/code/session_01W6eQFStA66EuMKHUFo2rx3
2026-04-12 10:23:43 +00:00
#[test]
fn ir_codegen_minimal_rom() {
let source = r#"
game "IR Test" { mapper: NROM }
var x: u8 = 42
on frame { wait_frame }
start Main
"#;
Remove the legacy AST codegen — IR path is canonical now The `--use-ast` path through `src/codegen/mod.rs` was a strictly inferior subset of the IR codegen. Building every example with `--use-ast` through the jsnes harness: - `arrays_and_functions` — fully black (array init + function return values + OAM-in-loop all broken) - `structs_enums_for` — fully black (struct literal is a no-op, all fields stay at 0) - `inline_asm_demo` — fully black - `bitwise_ops`, `loop_break_continue` — below sprite floors (static `next_oam_slot` bug B) - `match_demo` — panics at compile time with `branch offset 153 out of range` (AST's if/else-chain desugaring of `match` emits short branches that can't reach the far arms in a multi-arm match) Six of fourteen examples are non-functional under `--use-ast`. The other eight happen to fall inside the subset AST handles (no arrays, no structs, no function return values, no multi-sprite loops, no long match chains). `docs/future-work.md` already listed "Once working, delete the AST-based codegen entirely" as the intended direction. It's working, so this commit does the deletion. What's removed: - The `CodeGen` struct, its impl block, and every helper in `src/codegen/mod.rs` (the AST codegen body) — ~1150 lines. The file is now a module header that re-exports `IrCodeGen`. - `src/codegen/tests.rs` — 15 AST-specific instruction-pattern tests. Every feature they covered has an equivalent test in `src/codegen/ir_codegen.rs::{tests,more_tests}` already. - The `--use-ast` CLI flag and its branch in `src/main.rs`. - `compile_with_ir_codegen` in `tests/integration_test.rs` — `compile()` now does what it did, so they merged. All 40 integration tests go through the IR path. - Outdated sections in `docs/future-work.md` that described the IR codegen as "not yet implemented" and listed AST codegen gaps as priority work. What's kept: - `src/codegen/ir_codegen.rs` — the real codegen. - `src/codegen/peephole.rs` — post-codegen cleanup pass, now run unconditionally from `main.rs`. Test plan: - `cargo test --release` — 313 unit + 37 integration tests pass (was 328 + 37; the 15 dropped are the deleted AST-specific tests). - `cargo fmt --check` clean. - `cargo clippy --release --all-targets -- -D warnings` clean. - `node tests/emulator/run_examples.mjs` — 14/14 ROMs render above their per-example nonBlack floors. - The one tightening: `sprite_resolution_uses_tile_index` was asserting on the old static-slot encoding (`A9 01 8D 01 02`). Updated to the cursor-based form (`A9 01 99 01 02`, i.e. STA AbsoluteY). Net diff: 1581 deletions, 62 insertions. https://claude.ai/code/session_014Z5y3Q9krLcAxYpZQJhZ5V
2026-04-12 20:37:59 +00:00
let rom_data = compile(source);
Implement IR-based code generator (--use-ir) New src/codegen/ir_codegen.rs walks IrProgram and emits 6502 instructions. This enables optimizer passes to actually affect the output ROM. Design: - Each IR temp gets a zero-page slot at $80 + temp_index - Functions reset the temp counter at entry (temps don't outlive functions) - Globals map by name to their analyzer-assigned zero-page addresses - Operands are loaded into A, computed, stored back to the dest slot Handles all IrOp variants: - LoadImm, LoadVar, StoreVar (basic loads/stores) - Add/Sub (CLC+ADC / SEC+SBC) - Mul (JSR __multiply runtime routine) - And/Or/Xor (zero-page operands) - ShiftLeft/ShiftRight (repeated ASL/LSR) - Negate/Complement (EOR #$FF + optional two's complement) - CmpEq/Ne/Lt/Gt/LtEq/GtEq (CMP + conditional branch + 0/1) - ArrayLoad/ArrayStore (TAX + ZeroPageX/AbsoluteX) - Call (ZP param passing + JSR) - DrawSprite (OAM slot 0 write, uses sprite_tiles map) - ReadInput (LDA $01, P1 input) - WaitFrame (poll frame flag at $00) All terminators: - Jump (JMP to block label) - Branch (LDA temp + BNE true / JMP false) - Return (optional value in A + RTS) - Unreachable (BRK) IR lowering fixes: - ReadInput now has a destination IrTemp (was a side-effect-only op) - ButtonRead uses the proper input temp instead of uninitialized register - Logical AND/OR use new emit_move helper (OR with zero) instead of bogus raw VarId for path merging CLI: - New --use-ir flag on `build` subcommand to opt in to IR codegen - Default remains AST codegen (for now); IR codegen is experimental All 7 examples compile through the IR pipeline and produce valid iNES ROMs. Tests: 266 total (7 new ir_codegen unit + 2 new integration). https://claude.ai/code/session_01W6eQFStA66EuMKHUFo2rx3
2026-04-12 10:23:43 +00:00
let info = rom::validate_ines(&rom_data).expect("should be valid iNES");
assert_eq!(info.mapper, 0);
assert_eq!(rom_data.len(), 16 + 16384 + 8192);
}
#[test]
fn ir_codegen_full_pipeline() {
let source = r#"
game "IR Full" { mapper: NROM }
var x: u8 = 0
var y: u8 = 0
on frame {
if button.right { x += 1 }
if button.left { x -= 1 }
if x > 100 { x = 0 }
draw Smiley at: (x, y)
}
start Main
"#;
Remove the legacy AST codegen — IR path is canonical now The `--use-ast` path through `src/codegen/mod.rs` was a strictly inferior subset of the IR codegen. Building every example with `--use-ast` through the jsnes harness: - `arrays_and_functions` — fully black (array init + function return values + OAM-in-loop all broken) - `structs_enums_for` — fully black (struct literal is a no-op, all fields stay at 0) - `inline_asm_demo` — fully black - `bitwise_ops`, `loop_break_continue` — below sprite floors (static `next_oam_slot` bug B) - `match_demo` — panics at compile time with `branch offset 153 out of range` (AST's if/else-chain desugaring of `match` emits short branches that can't reach the far arms in a multi-arm match) Six of fourteen examples are non-functional under `--use-ast`. The other eight happen to fall inside the subset AST handles (no arrays, no structs, no function return values, no multi-sprite loops, no long match chains). `docs/future-work.md` already listed "Once working, delete the AST-based codegen entirely" as the intended direction. It's working, so this commit does the deletion. What's removed: - The `CodeGen` struct, its impl block, and every helper in `src/codegen/mod.rs` (the AST codegen body) — ~1150 lines. The file is now a module header that re-exports `IrCodeGen`. - `src/codegen/tests.rs` — 15 AST-specific instruction-pattern tests. Every feature they covered has an equivalent test in `src/codegen/ir_codegen.rs::{tests,more_tests}` already. - The `--use-ast` CLI flag and its branch in `src/main.rs`. - `compile_with_ir_codegen` in `tests/integration_test.rs` — `compile()` now does what it did, so they merged. All 40 integration tests go through the IR path. - Outdated sections in `docs/future-work.md` that described the IR codegen as "not yet implemented" and listed AST codegen gaps as priority work. What's kept: - `src/codegen/ir_codegen.rs` — the real codegen. - `src/codegen/peephole.rs` — post-codegen cleanup pass, now run unconditionally from `main.rs`. Test plan: - `cargo test --release` — 313 unit + 37 integration tests pass (was 328 + 37; the 15 dropped are the deleted AST-specific tests). - `cargo fmt --check` clean. - `cargo clippy --release --all-targets -- -D warnings` clean. - `node tests/emulator/run_examples.mjs` — 14/14 ROMs render above their per-example nonBlack floors. - The one tightening: `sprite_resolution_uses_tile_index` was asserting on the old static-slot encoding (`A9 01 8D 01 02`). Updated to the cursor-based form (`A9 01 99 01 02`, i.e. STA AbsoluteY). Net diff: 1581 deletions, 62 insertions. https://claude.ai/code/session_014Z5y3Q9krLcAxYpZQJhZ5V
2026-04-12 20:37:59 +00:00
let rom_data = compile(source);
Implement IR-based code generator (--use-ir) New src/codegen/ir_codegen.rs walks IrProgram and emits 6502 instructions. This enables optimizer passes to actually affect the output ROM. Design: - Each IR temp gets a zero-page slot at $80 + temp_index - Functions reset the temp counter at entry (temps don't outlive functions) - Globals map by name to their analyzer-assigned zero-page addresses - Operands are loaded into A, computed, stored back to the dest slot Handles all IrOp variants: - LoadImm, LoadVar, StoreVar (basic loads/stores) - Add/Sub (CLC+ADC / SEC+SBC) - Mul (JSR __multiply runtime routine) - And/Or/Xor (zero-page operands) - ShiftLeft/ShiftRight (repeated ASL/LSR) - Negate/Complement (EOR #$FF + optional two's complement) - CmpEq/Ne/Lt/Gt/LtEq/GtEq (CMP + conditional branch + 0/1) - ArrayLoad/ArrayStore (TAX + ZeroPageX/AbsoluteX) - Call (ZP param passing + JSR) - DrawSprite (OAM slot 0 write, uses sprite_tiles map) - ReadInput (LDA $01, P1 input) - WaitFrame (poll frame flag at $00) All terminators: - Jump (JMP to block label) - Branch (LDA temp + BNE true / JMP false) - Return (optional value in A + RTS) - Unreachable (BRK) IR lowering fixes: - ReadInput now has a destination IrTemp (was a side-effect-only op) - ButtonRead uses the proper input temp instead of uninitialized register - Logical AND/OR use new emit_move helper (OR with zero) instead of bogus raw VarId for path merging CLI: - New --use-ir flag on `build` subcommand to opt in to IR codegen - Default remains AST codegen (for now); IR codegen is experimental All 7 examples compile through the IR pipeline and produce valid iNES ROMs. Tests: 266 total (7 new ir_codegen unit + 2 new integration). https://claude.ai/code/session_01W6eQFStA66EuMKHUFo2rx3
2026-04-12 10:23:43 +00:00
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
#[test]
fn ir_codegen_multi_state_dispatch() {
// Exercise the IR main-loop dispatch with multiple states and a
// transition.
let source = r#"
game "IR States" { mapper: NROM }
var timer: u8 = 0
state Title {
on frame {
if button.start { transition Play }
}
}
state Play {
on frame {
timer += 1
if timer > 60 { transition Title }
}
}
start Title
"#;
Remove the legacy AST codegen — IR path is canonical now The `--use-ast` path through `src/codegen/mod.rs` was a strictly inferior subset of the IR codegen. Building every example with `--use-ast` through the jsnes harness: - `arrays_and_functions` — fully black (array init + function return values + OAM-in-loop all broken) - `structs_enums_for` — fully black (struct literal is a no-op, all fields stay at 0) - `inline_asm_demo` — fully black - `bitwise_ops`, `loop_break_continue` — below sprite floors (static `next_oam_slot` bug B) - `match_demo` — panics at compile time with `branch offset 153 out of range` (AST's if/else-chain desugaring of `match` emits short branches that can't reach the far arms in a multi-arm match) Six of fourteen examples are non-functional under `--use-ast`. The other eight happen to fall inside the subset AST handles (no arrays, no structs, no function return values, no multi-sprite loops, no long match chains). `docs/future-work.md` already listed "Once working, delete the AST-based codegen entirely" as the intended direction. It's working, so this commit does the deletion. What's removed: - The `CodeGen` struct, its impl block, and every helper in `src/codegen/mod.rs` (the AST codegen body) — ~1150 lines. The file is now a module header that re-exports `IrCodeGen`. - `src/codegen/tests.rs` — 15 AST-specific instruction-pattern tests. Every feature they covered has an equivalent test in `src/codegen/ir_codegen.rs::{tests,more_tests}` already. - The `--use-ast` CLI flag and its branch in `src/main.rs`. - `compile_with_ir_codegen` in `tests/integration_test.rs` — `compile()` now does what it did, so they merged. All 40 integration tests go through the IR path. - Outdated sections in `docs/future-work.md` that described the IR codegen as "not yet implemented" and listed AST codegen gaps as priority work. What's kept: - `src/codegen/ir_codegen.rs` — the real codegen. - `src/codegen/peephole.rs` — post-codegen cleanup pass, now run unconditionally from `main.rs`. Test plan: - `cargo test --release` — 313 unit + 37 integration tests pass (was 328 + 37; the 15 dropped are the deleted AST-specific tests). - `cargo fmt --check` clean. - `cargo clippy --release --all-targets -- -D warnings` clean. - `node tests/emulator/run_examples.mjs` — 14/14 ROMs render above their per-example nonBlack floors. - The one tightening: `sprite_resolution_uses_tile_index` was asserting on the old static-slot encoding (`A9 01 8D 01 02`). Updated to the cursor-based form (`A9 01 99 01 02`, i.e. STA AbsoluteY). Net diff: 1581 deletions, 62 insertions. https://claude.ai/code/session_014Z5y3Q9krLcAxYpZQJhZ5V
2026-04-12 20:37:59 +00:00
let rom_data = compile(source);
let info = rom::validate_ines(&rom_data).expect("should be valid iNES");
assert_eq!(info.mapper, 0);
}
#[test]
fn ir_codegen_multi_oam() {
// Draw multiple sprites and verify OAM slots are allocated sequentially.
let source = r#"
game "IR MultiOAM" { mapper: NROM }
var a: u8 = 10
var b: u8 = 20
var c: u8 = 30
on frame {
draw One at: (a, a)
draw Two at: (b, b)
draw Three at: (c, c)
}
start Main
"#;
Remove the legacy AST codegen — IR path is canonical now The `--use-ast` path through `src/codegen/mod.rs` was a strictly inferior subset of the IR codegen. Building every example with `--use-ast` through the jsnes harness: - `arrays_and_functions` — fully black (array init + function return values + OAM-in-loop all broken) - `structs_enums_for` — fully black (struct literal is a no-op, all fields stay at 0) - `inline_asm_demo` — fully black - `bitwise_ops`, `loop_break_continue` — below sprite floors (static `next_oam_slot` bug B) - `match_demo` — panics at compile time with `branch offset 153 out of range` (AST's if/else-chain desugaring of `match` emits short branches that can't reach the far arms in a multi-arm match) Six of fourteen examples are non-functional under `--use-ast`. The other eight happen to fall inside the subset AST handles (no arrays, no structs, no function return values, no multi-sprite loops, no long match chains). `docs/future-work.md` already listed "Once working, delete the AST-based codegen entirely" as the intended direction. It's working, so this commit does the deletion. What's removed: - The `CodeGen` struct, its impl block, and every helper in `src/codegen/mod.rs` (the AST codegen body) — ~1150 lines. The file is now a module header that re-exports `IrCodeGen`. - `src/codegen/tests.rs` — 15 AST-specific instruction-pattern tests. Every feature they covered has an equivalent test in `src/codegen/ir_codegen.rs::{tests,more_tests}` already. - The `--use-ast` CLI flag and its branch in `src/main.rs`. - `compile_with_ir_codegen` in `tests/integration_test.rs` — `compile()` now does what it did, so they merged. All 40 integration tests go through the IR path. - Outdated sections in `docs/future-work.md` that described the IR codegen as "not yet implemented" and listed AST codegen gaps as priority work. What's kept: - `src/codegen/ir_codegen.rs` — the real codegen. - `src/codegen/peephole.rs` — post-codegen cleanup pass, now run unconditionally from `main.rs`. Test plan: - `cargo test --release` — 313 unit + 37 integration tests pass (was 328 + 37; the 15 dropped are the deleted AST-specific tests). - `cargo fmt --check` clean. - `cargo clippy --release --all-targets -- -D warnings` clean. - `node tests/emulator/run_examples.mjs` — 14/14 ROMs render above their per-example nonBlack floors. - The one tightening: `sprite_resolution_uses_tile_index` was asserting on the old static-slot encoding (`A9 01 8D 01 02`). Updated to the cursor-based form (`A9 01 99 01 02`, i.e. STA AbsoluteY). Net diff: 1581 deletions, 62 insertions. https://claude.ai/code/session_014Z5y3Q9krLcAxYpZQJhZ5V
2026-04-12 20:37:59 +00:00
let rom_data = compile(source);
rom::validate_ines(&rom_data).expect("should be valid iNES");
}
Fix three compiler bugs exposed by array-using examples Landing bug A from the previous writeup plus two adjacent bugs that the fix exposed. All three miscompile anything that uses a u8[N] global with a literal initializer. 1. Array-literal globals are now actually initialized. `lower_program` only expanded `Expr::StructLiteral` into per- field synthetic globals — `Expr::ArrayLiteral` hit `eval_const`, returned `None`, and the array boot-cleared to zero. `IrGlobal` now carries an `init_array: Vec<u8>` populated by lowering, and the IR codegen startup loop emits one `LDA #byte; STA base+i` pair per element. 2. Local variables no longer overlap array globals. `IrCodeGen::new` advanced `local_ram_next` past `max_global_base + 1` — for an array at `$0300-$0303` it placed the first handler-local at `$0301`, inside the array. The frame handler's stores through the local then corrupted the array mid-frame. The allocator now walks the analyzer's `VarAllocation` list and advances past `address + size` for every RAM global, not just the base. 3. Peephole `remove_redundant_loads` honors indexed LDAs. The pass tracked `LDA Immediate/ZeroPage/Absolute` but let `LDA AbsoluteX/AbsoluteY/ZeroPageX/IndirectX/IndirectY` fall through the match, leaving the A-equivalence tracker unchanged. A later `LDA #v` that happened to match a stale entry from BEFORE the indexed load would then be dropped as "already in A" — a silent miscompile that turned every `draw Sprite at: (arr[i], arr[j])` pattern into garbage (the second array index would be computed from `arr[i]`'s value, reading way out of bounds). Indexed LDAs now clear the tracker. Regression tests: - `src/codegen/peephole.rs`: a synthetic `LDA #0; TAX; LDA AbsX(arr1); STA temp; LDA #0; TAX; LDA AbsX(arr2); ...` sequence asserts both `LDA #0`s survive. - `src/ir/tests.rs`: verifies `var xs: u8[4] = [1,2,3,4]` populates `IrGlobal::init_array` with `[1,2,3,4]`. - `tests/integration_test.rs`: two IR-codegen tests — one checks the startup instructions contain `LDA #v; STA base+i` for every element, the other compiles a handler-local var alongside an array global and asserts no post-init stores land inside the array. Smoke test impact (14/14 still passing, now more visible): - arrays_and_functions: 56 -> 104 nonBlack, now animated - loop_break_continue: 52 -> 208 (player + 3 hazards visible) - structs_enums_for: 52 -> 104 (player + enemy visible) Existing examples unchanged; no remaining work for bug B (static OAM slot allocation in loops) — that's the next PR. https://claude.ai/code/session_014Z5y3Q9krLcAxYpZQJhZ5V
2026-04-12 19:32:22 +00:00
#[test]
fn ir_codegen_array_literal_globals_emit_per_byte_init() {
// Regression test: `var xs: u8[4] = [10, 20, 30, 40]` used to
// compile to a zero-initialized array because `eval_const`
// returned `None` for `Expr::ArrayLiteral` and no startup
// stores were emitted. The fix captures the literal values
// in `IrGlobal::init_array` and has the IR codegen emit one
// `LDA #imm; STA base+i` per byte during startup.
use nescript::asm::{AddressingMode, Opcode};
use nescript::codegen::IrCodeGen;
let source = r#"
game "ArrLit" { mapper: NROM }
var xs: u8[4] = [10, 20, 30, 40]
on frame { wait_frame }
start Main
"#;
let (prog, diags) = nescript::parser::parse(source);
assert!(diags.is_empty(), "parse errors: {diags:?}");
let prog = prog.unwrap();
let analysis = analyzer::analyze(&prog);
let mut ir_program = ir::lower(&prog, &analysis);
optimizer::optimize(&mut ir_program);
let xs_addr = analysis
.var_allocations
.iter()
.find(|a| a.name == "xs")
.expect("xs should be allocated")
.address;
let codegen = IrCodeGen::new(&analysis.var_allocations, &ir_program);
let instructions = codegen.generate(&ir_program);
// For each element, look for `LDA #val` followed shortly by
// `STA absolute(xs_addr + i)`. We don't require them to be
// adjacent because the peephole passes can reshuffle, but a
// store of the correct value to the correct address must
// exist.
for (i, &expected) in [10u8, 20, 30, 40].iter().enumerate() {
let target = xs_addr + i as u16;
let has_store = instructions.windows(2).any(|w| {
matches!(w[0].mode, AddressingMode::Immediate(v) if v == expected)
&& w[0].opcode == Opcode::LDA
&& w[1].opcode == Opcode::STA
&& matches!(w[1].mode, AddressingMode::Absolute(a) if a == target)
});
assert!(
has_store,
"expected `LDA #{expected}; STA ${target:04X}` for xs[{i}] but did not find it"
);
}
}
#[test]
fn ir_codegen_locals_do_not_overlap_array_globals() {
// Regression test for the local-allocator off-by-array-size
// bug. `IrCodeGen::new` used to start handler-local vars at
// `max_global_base + 1`, which for an array global at
// `$0300-$0303` put the first local at `$0301` — inside the
// array. Any store through that local then corrupted the
// array mid-frame. The fix advances past the global's END,
// not its base.
//
// We verify by asking the IR codegen what addresses it
// assigned. Since `var_addrs` is private, we check indirectly
// via emitted instructions: any `STA $030N` for N > 3 that
// isn't part of the startup init must be writing to a local
// whose address is outside the array. If the bug regressed,
// we'd see `STA $0302` or similar in the frame handler's
// computation code.
use nescript::asm::{AddressingMode, Opcode};
use nescript::codegen::IrCodeGen;
let source = r#"
game "LocalVsArr" { mapper: NROM }
var xs: u8[4] = [11, 22, 33, 44]
on frame {
var tmp: u8 = 0
tmp = xs[0]
tmp += 1
wait_frame
}
start Main
"#;
let (prog, diags) = nescript::parser::parse(source);
assert!(diags.is_empty(), "parse errors: {diags:?}");
let prog = prog.unwrap();
let analysis = analyzer::analyze(&prog);
let mut ir_program = ir::lower(&prog, &analysis);
optimizer::optimize(&mut ir_program);
let xs_alloc = analysis
.var_allocations
.iter()
.find(|a| a.name == "xs")
.expect("xs should be allocated");
let xs_base = xs_alloc.address;
let xs_end = xs_base + xs_alloc.size; // one past last element
let codegen = IrCodeGen::new(&analysis.var_allocations, &ir_program);
let instructions = codegen.generate(&ir_program);
// Collect the (ordered) list of `STA absolute` targets and
// immediate values preceding each store. The first four
// stores into `[xs_base, xs_end)` should be the `LDA #imm;
// STA addr` init pairs — those are fine. Any STA into the
// array AFTER the init sequence would indicate a local var
// was allocated inside the array.
let mut init_stores_seen = 0usize;
for w in instructions.windows(2) {
if w[1].opcode != Opcode::STA {
continue;
}
let AddressingMode::Absolute(addr) = w[1].mode else {
continue;
};
if addr < xs_base || addr >= xs_end {
continue;
}
if w[0].opcode == Opcode::LDA
&& matches!(w[0].mode, AddressingMode::Immediate(_))
&& init_stores_seen < 4
{
init_stores_seen += 1;
continue;
}
panic!(
"store into xs array (${addr:04X}) after init sequence — \
local probably overlapping with array global"
);
}
assert_eq!(
init_stores_seen, 4,
"expected 4 init stores for xs[0..4], found {init_stores_seen}"
);
}
banks: implement multi-bank PRG layout and bank-switching runtime Prior to this commit the linker always shipped a single 16 KB PRG bank regardless of the declared mapper, so the README's MMC1/UxROM/ MMC3 support was aspirational. This commit gives the three banked mappers a real multi-bank ROM layout: * RomBuilder.set_prg_banks() writes any number of 16 KB banks back-to-back so the iNES header reflects the true PRG size. * Linker.link_banked() places switchable banks first, fixed bank last, so the fixed bank maps to $C000-$FFFF (the address window where vectors and the runtime live). * runtime::gen_mapper_init() emits reset-time mapper config: MMC1 serial-writes a control-register value that pins the last bank at $C000 with the correct mirroring, UxROM relies on the power-on default, MMC3 writes the $8000/$8001/$A000/$E000 registers to get a known PRG and mirroring state. * runtime::gen_bank_select() is a mapper-specific subroutine (callable with the target bank in A) that maps any physical bank to $8000-$BFFF. * runtime::gen_bank_trampoline() generates a cross-bank call stub in the fixed bank that saves the caller's bank, switches, JSRs the target, and restores the fixed bank. * The CLI and integration helper thread declared `bank X: prg` declarations through to the linker so MMC1/UxROM/MMC3 programs actually produce multi-bank ROMs. Coverage: * Runtime unit tests (18 new): mapper init patterns for every supported mapper, bank-select signatures, trampoline dispatch order, UxROM bus-conflict table contents. * RomBuilder tests (6 new): multi-bank layout, padding, byte-level fidelity, per-bank size validation, legacy single-bank fallback. * Linker tests (13 new): multi-bank ROM sizes across MMC1/ UxROM/MMC3, fixed-bank placement, switchable-bank payload fidelity, bank-select subroutine detection, NROM rejection of switchable banks. * Integration e2e tests (16 new): compile real .ne sources through the full pipeline and assert on iNES headers, mapper init signatures in the fixed bank, vector locations, and a regression check against `examples/mmc1_banked.ne`. Total: 474 tests pass under `cargo test` with `RUSTFLAGS="-D warnings"`. https://claude.ai/code/session_01UCressA5e8k1XsuoJYLav2
2026-04-13 01:50:51 +00:00
// ─── End-to-end bank switching tests ───────────────────────────────
//
// These tests compile real NEScript source through the full parse
// → analyze → IR → codegen → linker pipeline, producing .nes ROMs
// that assert the bank-switching layout the README promises:
//
// * Declared `bank X: prg` slots become real 16 KB PRG banks
// * Fixed bank lands at the end so it maps to $C000-$FFFF
// * Reset vector points inside the fixed bank
// * Mapper-specific init code appears in the fixed bank
// * Every iNES header field reflects the banked layout
#[test]
fn e2e_mmc1_with_two_declared_banks_produces_three_bank_rom() {
// MMC1 with two declared PRG banks should ship a ROM with
// three 16 KB PRG slots (Level1Data, Level2Data, fixed).
let source = r#"
game "MMC1 Banked" {
mapper: MMC1
mirroring: horizontal
}
bank Level1Data: prg
bank Level2Data: prg
var x: u8 = 0
on frame {
if button.right { x += 1 }
}
start Main
"#;
let rom = compile_banked(source);
let info = rom::validate_ines(&rom).expect("should be valid iNES");
assert_eq!(info.mapper, 1, "mapper number should be 1 (MMC1)");
assert_eq!(info.prg_banks, 3, "should have 2 switchable + 1 fixed bank");
assert_eq!(rom.len(), 16 + 3 * 16384 + 8192);
}
#[test]
fn e2e_uxrom_with_four_banks_produces_five_bank_rom() {
let source = r#"
game "UxROM Banked" {
mapper: UxROM
mirroring: vertical
}
bank Level1: prg
bank Level2: prg
bank Level3: prg
bank Level4: prg
var x: u8 = 0
on frame {
if button.a { x += 1 }
}
start Main
"#;
let rom = compile_banked(source);
let info = rom::validate_ines(&rom).expect("should be valid iNES");
assert_eq!(info.mapper, 2, "mapper number should be 2 (UxROM)");
assert_eq!(info.prg_banks, 5, "4 switchable + 1 fixed = 5 PRG banks");
assert_eq!(info.mirroring, nescript::parser::ast::Mirroring::Vertical);
}
#[test]
fn e2e_mmc3_with_three_banks_produces_four_bank_rom() {
let source = r#"
game "MMC3 Banked" {
mapper: MMC3
mirroring: horizontal
}
bank Stage1: prg
bank Stage2: prg
bank Stage3: prg
var x: u8 = 0
on frame {
if button.start { x = 1 }
}
start Main
"#;
let rom = compile_banked(source);
let info = rom::validate_ines(&rom).expect("should be valid iNES");
assert_eq!(info.mapper, 4, "mapper number should be 4 (MMC3)");
assert_eq!(info.prg_banks, 4, "3 switchable + 1 fixed = 4 PRG banks");
}
#[test]
fn e2e_banked_fixed_bank_contains_reset_vector() {
// The reset vector (bytes $FFFC/$FFFD in the final bank) must
// point into the $C000-$FFFF window — this is how the CPU
// boots into the fixed bank regardless of mapper.
let source = r#"
game "BankTest" { mapper: MMC1 }
bank Data: prg
on frame { wait_frame }
start Main
"#;
let rom = compile_banked(source);
let info = rom::validate_ines(&rom).expect("should be valid iNES");
let prg_end = 16 + info.prg_banks * 16384;
// Last 6 bytes = NMI, RESET, IRQ vectors (little-endian).
let reset = u16::from_le_bytes([rom[prg_end - 4], rom[prg_end - 3]]);
assert!(
(0xC000..=0xFFFF).contains(&reset),
"reset vector {reset:#06X} must live in fixed-bank address window"
);
}
#[test]
fn e2e_banked_fixed_bank_contains_mmc1_init_and_bank_select() {
// MMC1 requires a 6-way STA $8000 pattern at init (1 reset +
// 5 control bits) plus a 5-way STA $E000 pattern in the
// bank-select routine. Both must be in the fixed bank — they
// ship with the program regardless of whether user code
// calls `__bank_select` directly.
let source = r#"
game "MMC1Init" { mapper: MMC1 }
bank Payload: prg
var x: u8 = 0
on frame { x += 1 }
start Main
"#;
let rom = compile_banked(source);
let info = rom::validate_ines(&rom).expect("should be valid iNES");
// The fixed bank is the last 16 KB of PRG.
let fixed_offset = 16 + (info.prg_banks - 1) * 16384;
let fixed_bank = &rom[fixed_offset..fixed_offset + 16384];
// Count STA $8000 (opcode $8D, operand little-endian $00 $80):
// MMC1 init writes to $8000 six times.
let sta_lo = [0x8Du8, 0x00, 0x80];
let lo_count = fixed_bank.windows(3).filter(|w| *w == sta_lo).count();
assert!(
lo_count >= 6,
"MMC1 fixed bank should contain >=6 STA $8000 writes (got {lo_count})"
);
// Count STA $E000 (opcode $8D, operand $00 $E0): bank-select
// writes to it 5 times.
let sta_hi = [0x8Du8, 0x00, 0xE0];
let hi_count = fixed_bank.windows(3).filter(|w| *w == sta_hi).count();
assert!(
hi_count >= 5,
"MMC1 fixed bank should contain >=5 STA $E000 writes (got {hi_count})"
);
}
#[test]
fn e2e_banked_fixed_bank_contains_uxrom_bank_table() {
// UxROM ships a 256-byte bank-select bus-conflict table
// (values 0..=255). The table must be in the fixed bank.
let source = r#"
game "UxROMInit" { mapper: UxROM }
bank Payload: prg
on frame { wait_frame }
start Main
"#;
let rom = compile_banked(source);
let info = rom::validate_ines(&rom).unwrap();
let fixed_offset = 16 + (info.prg_banks - 1) * 16384;
let fixed = &rom[fixed_offset..fixed_offset + 16384];
// Search for a run of 0,1,2,3,...,31 — a 32-byte stretch that's
// distinctive enough that a random PRG byte sequence almost
// never contains it. The full 256-byte table starts with this
// prefix.
let mut needle: [u8; 32] = [0; 32];
#[allow(clippy::cast_possible_truncation)]
for (i, b) in needle.iter_mut().enumerate() {
*b = i as u8;
}
let found = fixed.windows(needle.len()).any(|w| w == needle);
assert!(
found,
"UxROM fixed bank should contain the bank-select bus-conflict table"
);
}
#[test]
fn e2e_banked_fixed_bank_contains_mmc3_init_writes() {
// MMC3 init writes two (bank-select, bank-number) pairs to
// ($8000, $8001) plus one $A000 mirroring write and one
// $E000 IRQ-disable write. We check each pattern appears.
let source = r#"
game "MMC3Init" { mapper: MMC3 }
bank Stage1: prg
on frame { wait_frame }
start Main
"#;
let rom = compile_banked(source);
let info = rom::validate_ines(&rom).unwrap();
let fixed_offset = 16 + (info.prg_banks - 1) * 16384;
let fixed_bank = &rom[fixed_offset..fixed_offset + 16384];
let select = [0x8Du8, 0x00, 0x80];
let data = [0x8Du8, 0x01, 0x80];
let mirror = [0x8Du8, 0x00, 0xA0];
// MMC3 init writes $8000 twice, plus once per bank-select
// call. With no `__bank_select` invocations from user code
// we expect exactly 2 init writes to $8000, but the
// bank-select subroutine also writes $8000 once. So the
// minimum is 3 (2 init + 1 bank-select body).
let select_count = fixed_bank.windows(3).filter(|w| *w == select).count();
let data_count = fixed_bank.windows(3).filter(|w| *w == data).count();
let mirror_count = fixed_bank.windows(3).filter(|w| *w == mirror).count();
assert!(
select_count >= 3,
"MMC3 fixed bank should contain >=3 STA $8000 writes (got {select_count})"
);
assert!(
data_count >= 3,
"MMC3 fixed bank should contain >=3 STA $8001 writes (got {data_count})"
);
assert!(
mirror_count >= 1,
"MMC3 fixed bank should contain >=1 STA $A000 write for mirroring (got {mirror_count})"
);
}
#[test]
fn e2e_banked_switchable_banks_contain_ff_padding() {
// Empty switchable banks should be entirely $FF-filled so no
// stray code accidentally lands in them. We check each
// switchable bank slot is 16384 bytes of $FF.
let source = r#"
game "PadCheck" { mapper: MMC1 }
bank A: prg
bank B: prg
on frame { wait_frame }
start Main
"#;
let rom = compile_banked(source);
for i in 0..2 {
let offset = 16 + i * 16384;
let bank = &rom[offset..offset + 16384];
assert!(
bank.iter().all(|&b| b == 0xFF),
"switchable bank {i} should be all $FF padding"
);
}
}
#[test]
fn e2e_nrom_still_produces_single_bank_rom_without_declarations() {
// Regression: programs that don't declare banks and use NROM
// must still ship as a single-bank 16 KB PRG ROM (the legacy
// layout), unaffected by the banking pipeline.
let source = r#"
game "Plain" { mapper: NROM }
var x: u8 = 0
on frame { x += 1 }
start Main
"#;
let rom = compile_banked(source);
let info = rom::validate_ines(&rom).unwrap();
assert_eq!(info.mapper, 0);
assert_eq!(info.prg_banks, 1);
assert_eq!(rom.len(), 16 + 16384 + 8192);
}
#[test]
fn e2e_chr_banks_do_not_consume_prg_slots() {
// A `bank X: chr` declaration reserves CHR space, not PRG.
// The linker currently keeps CHR at a single 8 KB slot, so
// declaring a CHR bank should NOT add a PRG slot.
let source = r#"
game "CHRBank" { mapper: MMC1 }
bank TileBank: chr
bank PrgBank: prg
on frame { wait_frame }
start Main
"#;
let rom = compile_banked(source);
let info = rom::validate_ines(&rom).unwrap();
// 1 PRG bank declared + 1 fixed = 2 total; TileBank:chr should
// NOT bump the PRG count.
assert_eq!(info.prg_banks, 2);
}
#[test]
fn e2e_mmc1_banked_example_compiles_successfully() {
// The examples/mmc1_banked.ne file is the canonical example
// the README points at. It must compile cleanly through the
// full pipeline and produce a valid multi-bank ROM.
let source = include_str!("../examples/mmc1_banked.ne");
let rom = compile_banked(source);
let info = rom::validate_ines(&rom).expect("should be valid iNES");
assert_eq!(info.mapper, 1, "mmc1_banked example should ship as MMC1");
assert!(
info.prg_banks >= 2,
"mmc1_banked example should ship with at least 2 PRG banks (got {})",
info.prg_banks
);
}
#[test]
fn e2e_large_bank_count_still_produces_valid_rom() {
// Stress test: 7 switchable banks (8 total) on UxROM. This
// exercises the ROM builder's multi-bank concatenation with
// a non-trivial bank count and ensures nothing in the linker
// pipeline hard-codes a bank limit.
let source = r#"
game "LotsOfBanks" { mapper: UxROM }
bank A: prg
bank B: prg
bank C: prg
bank D: prg
bank E: prg
bank F: prg
bank G: prg
on frame { wait_frame }
start Main
"#;
let rom = compile_banked(source);
let info = rom::validate_ines(&rom).unwrap();
assert_eq!(info.prg_banks, 8, "7 switchable + 1 fixed = 8 PRG banks");
assert_eq!(rom.len(), 16 + 8 * 16384 + 8192);
}
#[test]
fn e2e_banked_rom_ines_header_mapper_bits_encoded_correctly() {
// Sanity check: the iNES header's mapper number field is split
// across byte 6 (low nibble) and byte 7 (high nibble). For
// mapper 1 (MMC1), byte 6 should have $10 in its high nibble
// and byte 7 should have $00 in its high nibble.
let source = r#"
game "HeaderCheck" { mapper: MMC1 }
bank Foo: prg
on frame { wait_frame }
start Main
"#;
let rom = compile_banked(source);
let byte6_high_nibble = rom[6] & 0xF0;
let byte7_high_nibble = rom[7] & 0xF0;
assert_eq!(byte6_high_nibble, 0x10, "MMC1 low mapper nibble in byte 6");
assert_eq!(byte7_high_nibble, 0x00, "MMC1 high mapper nibble in byte 7");
}
#[test]
fn e2e_banked_all_three_mappers_have_correct_vectors() {
// For each banked mapper, verify all three vectors (NMI, RESET,
// IRQ) live inside the fixed bank address window.
for mapper_kw in ["MMC1", "UxROM", "MMC3"] {
let source = format!(
r#"
game "VecCheck" {{ mapper: {mapper_kw} }}
bank One: prg
on frame {{ wait_frame }}
start Main
"#
);
let rom = compile_banked(&source);
let info = rom::validate_ines(&rom).unwrap();
let prg_end = 16 + info.prg_banks * 16384;
let nmi = u16::from_le_bytes([rom[prg_end - 6], rom[prg_end - 5]]);
let reset = u16::from_le_bytes([rom[prg_end - 4], rom[prg_end - 3]]);
let irq = u16::from_le_bytes([rom[prg_end - 2], rom[prg_end - 1]]);
for (name, v) in [("NMI", nmi), ("RESET", reset), ("IRQ", irq)] {
assert!(
(0xC000..=0xFFFF).contains(&v),
"{mapper_kw} {name} vector {v:#06X} should be in fixed-bank window"
);
}
}
}
#[test]
fn e2e_bank_declarations_dont_affect_nrom_prg_size() {
// Even though the linker REJECTS switchable banks for NROM,
// the compiler only passes banks through when they're in the
// `program.banks` list — for NROM sources without declarations
// nothing is passed, so the NROM path is unchanged. Just
// double-check here that a plain NROM ROM is still 1 bank.
let source = r#"
game "JustNROM" { mapper: NROM }
on frame { wait_frame }
start Main
"#;
let rom = compile_banked(source);
let info = rom::validate_ines(&rom).unwrap();
assert_eq!(info.prg_banks, 1);
assert_eq!(info.mapper, 0);
}
#[test]
fn e2e_banked_chr_rom_is_preserved() {
// CHR ROM should still contain the default smiley sprite at
// tile 0 regardless of how many PRG banks the ROM has.
let source = r#"
game "CHRCheck" { mapper: MMC1 }
bank One: prg
bank Two: prg
on frame { wait_frame }
start Main
"#;
let rom = compile_banked(source);
let info = rom::validate_ines(&rom).unwrap();
let chr_start = 16 + info.prg_banks * 16384;
// Default smiley is non-zero in its first 16 bytes.
assert_ne!(&rom[chr_start..chr_start + 16], &[0u8; 16]);
}