| 1 | //! Smoke-level fuzz tests for the lexer and parser. |
| 2 | //! |
| 3 | //! These don't use libfuzzer — they run deterministic random inputs |
| 4 | //! through the lexer and parser to verify no panics. For real fuzzing, |
| 5 | //! use `cargo fuzz run fuzz_lexer` / `cargo fuzz run fuzz_parser`. |
| 6 | |
| 7 | use armfortas::lexer::{self, SourceForm}; |
| 8 | use armfortas::parser::Parser; |
| 9 | |
| 10 | /// Feed N random-ish strings through the lexer. |
| 11 | fn fuzz_lexer_deterministic(seed: u64, count: usize) { |
| 12 | let mut state = seed; |
| 13 | for _ in 0..count { |
| 14 | // Simple xorshift64 PRNG. |
| 15 | state ^= state << 13; |
| 16 | state ^= state >> 7; |
| 17 | state ^= state << 17; |
| 18 | |
| 19 | let len = (state % 256) as usize; |
| 20 | let bytes: Vec<u8> = (0..len) |
| 21 | .map(|i| { |
| 22 | let mut s = state.wrapping_add(i as u64); |
| 23 | s ^= s << 13; |
| 24 | s ^= s >> 7; |
| 25 | s ^= s << 17; |
| 26 | (s & 0x7F) as u8 // ASCII range |
| 27 | }) |
| 28 | .collect(); |
| 29 | |
| 30 | if let Ok(src) = std::str::from_utf8(&bytes) { |
| 31 | let _ = lexer::tokenize(src, 0, SourceForm::FreeForm); |
| 32 | } |
| 33 | } |
| 34 | } |
| 35 | |
| 36 | /// Feed N random-ish strings through lexer → parser. |
| 37 | fn fuzz_parser_deterministic(seed: u64, count: usize) { |
| 38 | let mut state = seed; |
| 39 | for _ in 0..count { |
| 40 | state ^= state << 13; |
| 41 | state ^= state >> 7; |
| 42 | state ^= state << 17; |
| 43 | |
| 44 | let len = (state % 512) as usize; |
| 45 | let bytes: Vec<u8> = (0..len) |
| 46 | .map(|i| { |
| 47 | let mut s = state.wrapping_add(i as u64); |
| 48 | s ^= s << 13; |
| 49 | s ^= s >> 7; |
| 50 | s ^= s << 17; |
| 51 | (s & 0x7F) as u8 |
| 52 | }) |
| 53 | .collect(); |
| 54 | |
| 55 | if let Ok(src) = std::str::from_utf8(&bytes) { |
| 56 | if let Ok(tokens) = lexer::tokenize(src, 0, SourceForm::FreeForm) { |
| 57 | let mut parser = Parser::new(&tokens); |
| 58 | let _ = parser.parse_file(); |
| 59 | } |
| 60 | } |
| 61 | } |
| 62 | } |
| 63 | |
| 64 | /// Feed malformed but plausible Fortran through the parser. |
| 65 | /// Runs without threads — each input goes directly through the parser. |
| 66 | fn fuzz_parser_with_fortran_fragments(count: usize) { |
| 67 | let fragments = [ |
| 68 | "program p\nend program\n", "module m\nend module\n", |
| 69 | "integer :: x\n", "real, allocatable :: a(:,:)\n", |
| 70 | "do i = 1, 10\nend do\n", "if (x > 0) then\nend if\n", |
| 71 | "select case (x)\ncase (1)\nend select\n", |
| 72 | "type :: t\n integer :: f\nend type\n", |
| 73 | "interface\nend interface\n", |
| 74 | "goto 100\n", "100 continue\n", |
| 75 | "use iso_c_binding, only: c_int\n", |
| 76 | "", "!\n", "! comment\n", |
| 77 | "end\n", |
| 78 | ]; |
| 79 | |
| 80 | let mut state: u64 = 0xDEADBEEF; |
| 81 | for _ in 0..count { |
| 82 | state ^= state << 13; |
| 83 | state ^= state >> 7; |
| 84 | state ^= state << 17; |
| 85 | |
| 86 | // Pick 1-3 random fragments and concatenate. |
| 87 | let n_frags = 1 + (state % 3) as usize; |
| 88 | let mut src = String::new(); |
| 89 | for _ in 0..n_frags { |
| 90 | state ^= state << 13; |
| 91 | state ^= state >> 7; |
| 92 | state ^= state << 17; |
| 93 | let idx = (state as usize) % fragments.len(); |
| 94 | src.push_str(fragments[idx]); |
| 95 | } |
| 96 | |
| 97 | if let Ok(tokens) = lexer::tokenize(&src, 0, SourceForm::FreeForm) { |
| 98 | if tokens.len() > 100 { continue; } |
| 99 | let mut parser = Parser::new(&tokens); |
| 100 | let _ = parser.parse_file(); |
| 101 | } |
| 102 | } |
| 103 | } |
| 104 | |
| 105 | #[test] |
| 106 | fn lexer_no_panic_on_random_ascii() { |
| 107 | fuzz_lexer_deterministic(0x12345678, 10_000); |
| 108 | } |
| 109 | |
| 110 | #[test] |
| 111 | fn parser_no_panic_on_random_ascii() { |
| 112 | fuzz_parser_deterministic(0x87654321, 5_000); |
| 113 | } |
| 114 | |
| 115 | #[test] |
| 116 | fn parser_no_panic_on_fortran_fragments() { |
| 117 | fuzz_parser_with_fortran_fragments(5_000); |
| 118 | } |
| 119 |