1use crate::{create_file, json_from_path, source_from_path};
8use oak_core::{
9 Language, Lexer, Source, TokenType,
10 errors::{OakDiagnostics, OakError},
11};
12use serde::{Deserialize, Serialize};
13
14use std::{
15 path::{Path, PathBuf},
16 sync::{Arc, Mutex},
17 thread,
18 time::{Duration, Instant},
19};
20use walkdir::WalkDir;
21
22pub struct LexerTester {
28 root: PathBuf,
29 extensions: Vec<String>,
30 timeout: Duration,
31}
32
33#[derive(Debug, PartialEq, Serialize, Deserialize)]
38pub struct LexerTestExpected {
39 pub success: bool,
41 pub count: usize,
43 pub tokens: Vec<TokenData>,
45 pub errors: Vec<String>,
47}
48
49#[derive(Debug, PartialEq, Serialize, Deserialize)]
54pub struct TokenData {
55 pub kind: String,
57 pub text: String,
59 pub start: usize,
61 pub end: usize,
63}
64
65impl LexerTester {
66 pub fn new<P: AsRef<Path>>(root: P) -> Self {
68 Self { root: root.as_ref().to_path_buf(), extensions: vec![], timeout: Duration::from_secs(10) }
69 }
70
71 pub fn with_extension(mut self, extension: impl ToString) -> Self {
73 self.extensions.push(extension.to_string());
74 self
75 }
76
77 pub fn with_timeout(mut self, time: Duration) -> Self {
79 self.timeout = time;
80 self
81 }
82
83 pub fn run_tests<L, Lex>(self, lexer: &Lex) -> Result<(), OakError>
85 where
86 L: Language + Send + Sync,
87 L::TokenType: Serialize + std::fmt::Debug + Send + Sync,
88 Lex: Lexer<L> + Send + Sync + Clone,
89 {
90 let test_files = self.find_test_files()?;
91 let force_regenerated = std::env::var("REGENERATE_TESTS").unwrap_or("0".to_string()) == "1";
92 let mut regenerated_any = false;
93
94 for file_path in test_files {
95 println!("Testing file: {}", file_path.display());
96 regenerated_any |= self.test_single_file::<L, Lex>(&file_path, lexer, force_regenerated)?
97 }
98
99 if regenerated_any && force_regenerated {
100 println!("Tests regenerated for: {}", self.root.display());
101 Ok(())
102 }
103 else {
104 Ok(())
105 }
106 }
107
108 fn find_test_files(&self) -> Result<Vec<PathBuf>, OakError> {
109 let mut files = Vec::new();
110
111 for entry in WalkDir::new(&self.root) {
112 let entry = entry.unwrap();
113 let path = entry.path();
114
115 if path.is_file() {
116 if let Some(ext) = path.extension() {
117 let ext_str = ext.to_str().unwrap_or("");
118 if self.extensions.iter().any(|e| e == ext_str) {
119 let file_name = path.file_name().and_then(|n| n.to_str()).unwrap_or("");
121 let is_output_file = file_name.ends_with(".parsed.json") || file_name.ends_with(".lexed.json") || file_name.ends_with(".built.json") || file_name.ends_with(".expected.json");
122
123 if !is_output_file {
124 files.push(path.to_path_buf())
125 }
126 }
127 }
128 }
129 }
130
131 Ok(files)
132 }
133
134 fn test_single_file<L, Lex>(&self, file_path: &Path, lexer: &Lex, force_regenerated: bool) -> Result<bool, OakError>
135 where
136 L: Language + Send + Sync,
137 L::TokenType: Serialize + std::fmt::Debug + Send + Sync,
138 Lex: Lexer<L> + Send + Sync + Clone,
139 {
140 let source = source_from_path(file_path)?;
141
142 let result = Arc::new(Mutex::new(None));
144 let result_clone = Arc::clone(&result);
145
146 let lexer_clone = lexer.clone();
148 let source_arc = Arc::new(source);
150 let source_clone = Arc::clone(&source_arc);
151
152 std::thread::scope(|s| {
154 let handle = s.spawn(move || {
155 let mut cache = oak_core::parser::ParseSession::<L>::default();
156 let output = lexer_clone.lex(&*source_clone, &[], &mut cache);
157 let mut result = result_clone.lock().unwrap();
158 *result = Some(output)
159 });
160
161 let start_time = Instant::now();
163 let timeout_occurred = loop {
164 if handle.is_finished() {
166 break false;
167 }
168
169 if start_time.elapsed() > self.timeout {
171 break true;
172 }
173
174 thread::sleep(Duration::from_millis(10));
176 };
177
178 if timeout_occurred {
180 return Err(OakError::custom_error(&format!("Lexer test timed out after {:?} for file: {}", self.timeout, file_path.display())));
181 }
182
183 Ok(())
184 })?;
185
186 let OakDiagnostics { result: tokens_result, mut diagnostics } = {
188 let result_guard = result.lock().unwrap();
189 match result_guard.as_ref() {
190 Some(output) => output.clone(),
191 None => return Err(OakError::custom_error("Failed to get lexer result")),
192 }
193 };
194
195 let mut success = true;
197 let tokens = match tokens_result {
198 Ok(tokens) => tokens,
199 Err(e) => {
200 success = false;
201 diagnostics.push(e);
202 oak_core::Tokens::default()
203 }
204 };
205
206 if !diagnostics.is_empty() {
207 success = false;
208 }
209
210 let tokens: Vec<TokenData> = tokens
211 .iter()
212 .filter(|token| !token.kind.is_ignored())
213 .map(|token| {
214 let len = source_arc.as_ref().length();
215 let start = token.span.start.min(len);
216 let end = token.span.end.min(len).max(start);
217 let text = source_arc.as_ref().get_text_in((start..end).into()).to_string();
218 TokenData { kind: format!("{:?}", token.kind), text, start: token.span.start, end: token.span.end }
219 })
220 .collect();
221
222 let errors: Vec<String> = diagnostics.iter().map(|e| e.to_string()).collect();
223 let test_result = LexerTestExpected { success, count: tokens.len(), tokens, errors };
224
225 let expected_file = file_path.with_extension(format!("{}.lexed.json", file_path.extension().unwrap_or_default().to_str().unwrap_or("")));
227
228 if !expected_file.exists() {
230 let legacy_file = file_path.with_extension("expected.json");
231 if legacy_file.exists() {
232 let _ = std::fs::rename(&legacy_file, &expected_file);
233 }
234 }
235
236 let mut regenerated = false;
237 if expected_file.exists() && !force_regenerated {
238 let expected_json = json_from_path(&expected_file)?;
239 let expected: LexerTestExpected = serde_json::from_value(expected_json).map_err(|e| OakError::custom_error(e.to_string()))?;
240 if test_result != expected {
241 return Err(OakError::test_failure(file_path.to_path_buf(), format!("{:#?}", expected), format!("{:#?}", test_result)));
242 }
243 }
244 else {
245 use std::io::Write;
246 let mut file = create_file(&expected_file)?;
247 let json_val = serde_json::to_string_pretty(&test_result).map_err(|e| OakError::custom_error(e.to_string()))?;
248 file.write_all(json_val.as_bytes()).map_err(|e| OakError::custom_error(e.to_string()))?;
249 regenerated = true;
250 }
251
252 Ok(regenerated)
253 }
254}