oak_testing/
lexing.rs

1//! Lexer testing utilities for the Oak ecosystem.
2//!
3//! This module provides comprehensive testing infrastructure for lexers,
4//! including file-based testing, expected output comparison, and
5//! test result serialization.
6
7use 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
22/// A lexer testing utility that can run tests against multiple files.
23///
24/// The `LexerTester` provides functionality to test lexers against a directory
25/// of files with specific extensions, comparing actual output against expected
26/// results stored in JSON files.
27pub struct LexerTester {
28    root: PathBuf,
29    extensions: Vec<String>,
30    timeout: Duration,
31}
32
33/// Expected lexer test results for comparison.
34///
35/// This struct represents the expected output of a lexer test, including
36/// success status, token count, token data, and any expected errors.
37#[derive(Debug, PartialEq, Serialize, Deserialize)]
38pub struct LexerTestExpected {
39    pub success: bool,
40    pub count: usize,
41    pub tokens: Vec<TokenData>,
42    pub errors: Vec<String>,
43}
44
45/// Individual token data for lexer testing.
46///
47/// Represents a single token with its kind, text content, and position
48/// information used for testing lexer output.
49#[derive(Debug, PartialEq, Serialize, Deserialize)]
50pub struct TokenData {
51    pub kind: String,
52    pub text: String,
53    pub start: usize,
54    pub end: usize,
55}
56
57impl LexerTester {
58    /// Creates a new lexer tester with the specified root directory.
59    pub fn new<P: AsRef<Path>>(root: P) -> Self {
60        Self { root: root.as_ref().to_path_buf(), extensions: vec![], timeout: Duration::from_secs(10) }
61    }
62
63    /// Adds a file extension to test against.
64    pub fn with_extension(mut self, extension: impl ToString) -> Self {
65        self.extensions.push(extension.to_string());
66        self
67    }
68    /// Sets the timeout duration for each test.
69    pub fn with_timeout(mut self, time: Duration) -> Self {
70        self.timeout = time;
71        self
72    }
73
74    /// Run tests for the given lexer against all files in the root directory with the specified extensions.
75    pub fn run_tests<L, Lex>(self, lexer: &Lex) -> Result<(), OakError>
76    where
77        L: Language + Send + Sync + 'static,
78        L::TokenType: Serialize + std::fmt::Debug + Send + Sync,
79        Lex: Lexer<L> + Send + Sync + 'static + Clone,
80    {
81        let test_files = self.find_test_files()?;
82        let force_regenerated = std::env::var("REGENERATE_TESTS").unwrap_or("0".to_string()) == "1";
83        let mut regenerated_any = false;
84
85        for file_path in test_files {
86            println!("Testing file: {}", file_path.display());
87            regenerated_any |= self.test_single_file::<L, Lex>(&file_path, lexer, force_regenerated)?;
88        }
89
90        if regenerated_any && force_regenerated { Err(OakError::test_regenerated(self.root)) } else { Ok(()) }
91    }
92
93    fn find_test_files(&self) -> Result<Vec<PathBuf>, OakError> {
94        let mut files = Vec::new();
95
96        for entry in WalkDir::new(&self.root) {
97            let entry = entry.unwrap();
98            let path = entry.path();
99
100            if path.is_file() {
101                if let Some(ext) = path.extension() {
102                    let ext_str = ext.to_str().unwrap_or("");
103                    if self.extensions.iter().any(|e| e == ext_str) {
104                        // 忽略由 Tester 自身生成的输出文件，防止递归包含
105                        let file_name = path.file_name().and_then(|n| n.to_str()).unwrap_or("");
106                        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");
107
108                        if !is_output_file {
109                            files.push(path.to_path_buf());
110                        }
111                    }
112                }
113            }
114        }
115
116        Ok(files)
117    }
118
119    fn test_single_file<L, Lex>(&self, file_path: &Path, lexer: &Lex, force_regenerated: bool) -> Result<bool, OakError>
120    where
121        L: Language + Send + Sync + 'static,
122        L::TokenType: Serialize + std::fmt::Debug + Send + Sync,
123        Lex: Lexer<L> + Send + Sync + 'static + Clone,
124    {
125        let source = source_from_path(file_path)?;
126
127        // Use Arc and Mutex to share results between threads
128        let result = Arc::new(Mutex::new(None));
129        let result_clone = Arc::clone(&result);
130
131        // Clone lexer for use in thread
132        let lexer_clone = lexer.clone();
133        // Wrap source in Arc for sharing between threads
134        let source_arc = Arc::new(source);
135        let source_clone = Arc::clone(&source_arc);
136
137        // Create a new thread to perform lexical analysis
138        let handle = thread::spawn(move || {
139            let mut cache = oak_core::parser::ParseSession::<L>::default();
140            let output = lexer_clone.lex(&*source_clone, &[], &mut cache);
141            let mut result = result_clone.lock().unwrap();
142            *result = Some(output);
143        });
144
145        // Wait for thread completion or timeout
146        let start_time = Instant::now();
147        let timeout_occurred = loop {
148            // Check if thread has finished
149            if handle.is_finished() {
150                break false;
151            }
152
153            // Check for timeout
154            if start_time.elapsed() > self.timeout {
155                break true;
156            }
157
158            // Sleep briefly to avoid busy waiting
159            thread::sleep(Duration::from_millis(10));
160        };
161
162        // Return error if timed out
163        if timeout_occurred {
164            return Err(OakError::custom_error(&format!("Lexer test timed out after {:?} for file: {}", self.timeout, file_path.display())));
165        }
166
167        // Get lexical analysis result
168        let OakDiagnostics { result: tokens_result, mut diagnostics } = {
169            let result_guard = result.lock().unwrap();
170            match result_guard.as_ref() {
171                Some(output) => output.clone(),
172                None => return Err(OakError::custom_error("Failed to get lexer result")),
173            }
174        };
175
176        // Construct test result
177        let mut success = true;
178        let tokens = match tokens_result {
179            Ok(tokens) => tokens,
180            Err(e) => {
181                success = false;
182                diagnostics.push(e);
183                triomphe::Arc::from_iter(Vec::new())
184            }
185        };
186
187        if !diagnostics.is_empty() {
188            success = false;
189        }
190
191        let tokens: Vec<TokenData> = tokens
192            .iter()
193            .filter(|token| !token.kind.is_ignored())
194            .map(|token| {
195                let len = source_arc.as_ref().length();
196                let start = token.span.start.min(len);
197                let end = token.span.end.min(len).max(start);
198                let text = source_arc.as_ref().get_text_in((start..end).into()).to_string();
199                TokenData { kind: format!("{:?}", token.kind), text, start: token.span.start, end: token.span.end }
200            })
201            .take(100)
202            .collect();
203
204        let errors: Vec<String> = diagnostics.iter().map(|e| e.to_string()).collect();
205        let test_result = LexerTestExpected { success, count: tokens.len(), tokens, errors };
206
207        // Process expected result file
208        let expected_file = file_path.with_extension(format!("{}.lexed.json", file_path.extension().unwrap_or_default().to_str().unwrap_or("")));
209
210        // Migration: If the new naming convention file doesn't exist, but the old one does, rename it
211        if !expected_file.exists() {
212            let legacy_file = file_path.with_extension("expected.json");
213            if legacy_file.exists() {
214                let _ = std::fs::rename(&legacy_file, &expected_file);
215            }
216        }
217
218        let mut regenerated = false;
219        if expected_file.exists() && !force_regenerated {
220            let expected_json = json_from_path(&expected_file)?;
221            let expected: LexerTestExpected = serde_json::from_value(expected_json).map_err(|e| OakError::custom_error(e.to_string()))?;
222            if test_result != expected {
223                return Err(OakError::test_failure(file_path.to_path_buf(), format!("{:#?}", expected), format!("{:#?}", test_result)));
224            }
225        }
226        else {
227            use std::io::Write;
228            let mut file = create_file(&expected_file)?;
229            let json_val = serde_json::to_string_pretty(&test_result).map_err(|e| OakError::custom_error(e.to_string()))?;
230            file.write_all(json_val.as_bytes()).map_err(|e| OakError::custom_error(e.to_string()))?;
231            regenerated = true;
232        }
233
234        Ok(regenerated)
235    }
236}
oak_testing/lexing.rs

oak_testing/
lexing.rs
