rumdl_lib/utils/
regex_cache.rs

1//!
2//! Cached Regex Patterns and Fast Content Checks for Markdown Linting
3//!
4//! This module provides a centralized collection of pre-compiled, cached regex patterns
5//! for all major Markdown constructs (headings, lists, code blocks, links, images, etc.).
6//! It also includes fast-path utility functions for quickly checking if content
7//! potentially contains certain Markdown elements, allowing rules to skip expensive
8//! processing when unnecessary.
9//!
10//! # Performance
11//!
12//! All regexes are compiled once at startup using `lazy_static`, avoiding repeated
13//! compilation and improving performance across the linter. Use these shared patterns
14//! in rules instead of compiling new regexes.
15//!
16//! # Usage
17//!
18//! - Use the provided statics for common Markdown patterns.
19//! - Use the `regex_lazy!` macro for ad-hoc regexes that are not predefined.
20//! - Use the utility functions for fast content checks before running regexes.
21
22use fancy_regex::Regex as FancyRegex;
23use regex::Regex;
24use std::collections::HashMap;
25use std::sync::LazyLock;
26use std::sync::{Arc, Mutex};
27
28/// Global regex cache for dynamic patterns
29#[derive(Debug)]
30pub struct RegexCache {
31    cache: HashMap<String, Arc<Regex>>,
32    fancy_cache: HashMap<String, Arc<FancyRegex>>,
33    usage_stats: HashMap<String, u64>,
34}
35
36impl Default for RegexCache {
37    fn default() -> Self {
38        Self::new()
39    }
40}
41
42impl RegexCache {
43    pub fn new() -> Self {
44        Self {
45            cache: HashMap::new(),
46            fancy_cache: HashMap::new(),
47            usage_stats: HashMap::new(),
48        }
49    }
50
51    /// Get or compile a regex pattern
52    pub fn get_regex(&mut self, pattern: &str) -> Result<Arc<Regex>, regex::Error> {
53        if let Some(regex) = self.cache.get(pattern) {
54            *self.usage_stats.entry(pattern.to_string()).or_insert(0) += 1;
55            return Ok(regex.clone());
56        }
57
58        let regex = Arc::new(Regex::new(pattern)?);
59        self.cache.insert(pattern.to_string(), regex.clone());
60        *self.usage_stats.entry(pattern.to_string()).or_insert(0) += 1;
61        Ok(regex)
62    }
63
64    /// Get or compile a fancy regex pattern
65    pub fn get_fancy_regex(&mut self, pattern: &str) -> Result<Arc<FancyRegex>, Box<fancy_regex::Error>> {
66        if let Some(regex) = self.fancy_cache.get(pattern) {
67            *self.usage_stats.entry(pattern.to_string()).or_insert(0) += 1;
68            return Ok(regex.clone());
69        }
70
71        match FancyRegex::new(pattern) {
72            Ok(regex) => {
73                let arc_regex = Arc::new(regex);
74                self.fancy_cache.insert(pattern.to_string(), arc_regex.clone());
75                *self.usage_stats.entry(pattern.to_string()).or_insert(0) += 1;
76                Ok(arc_regex)
77            }
78            Err(e) => Err(Box::new(e)),
79        }
80    }
81
82    /// Get cache statistics
83    pub fn get_stats(&self) -> HashMap<String, u64> {
84        self.usage_stats.clone()
85    }
86
87    /// Clear cache (useful for testing)
88    pub fn clear(&mut self) {
89        self.cache.clear();
90        self.fancy_cache.clear();
91        self.usage_stats.clear();
92    }
93}
94
95/// Global regex cache instance
96static GLOBAL_REGEX_CACHE: LazyLock<Arc<Mutex<RegexCache>>> = LazyLock::new(|| Arc::new(Mutex::new(RegexCache::new())));
97
98/// Get a regex from the global cache
99pub fn get_cached_regex(pattern: &str) -> Result<Arc<Regex>, regex::Error> {
100    let mut cache = GLOBAL_REGEX_CACHE.lock().expect("Regex cache mutex poisoned");
101    cache.get_regex(pattern)
102}
103
104/// Get a fancy regex from the global cache
105pub fn get_cached_fancy_regex(pattern: &str) -> Result<Arc<FancyRegex>, Box<fancy_regex::Error>> {
106    let mut cache = GLOBAL_REGEX_CACHE.lock().expect("Regex cache mutex poisoned");
107    cache.get_fancy_regex(pattern)
108}
109
110/// Get cache usage statistics
111pub fn get_cache_stats() -> HashMap<String, u64> {
112    let cache = GLOBAL_REGEX_CACHE.lock().expect("Regex cache mutex poisoned");
113    cache.get_stats()
114}
115
116/// Macro for defining a lazily-initialized, cached regex pattern.
117///
118/// Use this for ad-hoc regexes that are not already defined in this module.
119///
120/// # Panics
121///
122/// This macro will panic at initialization if the regex pattern is invalid.
123/// This is intentional for compile-time constant patterns - we want to catch
124/// invalid patterns during development, not at runtime.
125///
126/// # Example
127///
128/// ```
129/// use rumdl_lib::regex_lazy;
130/// let my_re = regex_lazy!(r"^foo.*bar$");
131/// assert!(my_re.is_match("foobar"));
132/// ```
133#[macro_export]
134macro_rules! regex_lazy {
135    ($pattern:expr) => {{
136        static REGEX: LazyLock<regex::Regex> = LazyLock::new(|| regex::Regex::new($pattern).unwrap());
137        &*REGEX
138    }};
139}
140
141/// Macro for getting regex from global cache.
142///
143/// # Panics
144///
145/// Panics if the regex pattern is invalid. This is acceptable for static patterns
146/// where we want to fail fast during development.
147#[macro_export]
148macro_rules! regex_cached {
149    ($pattern:expr) => {{ $crate::utils::regex_cache::get_cached_regex($pattern).expect("Failed to compile regex") }};
150}
151
152/// Macro for getting fancy regex from global cache.
153///
154/// # Panics
155///
156/// Panics if the regex pattern is invalid. This is acceptable for static patterns
157/// where we want to fail fast during development.
158#[macro_export]
159macro_rules! fancy_regex_cached {
160    ($pattern:expr) => {{ $crate::utils::regex_cache::get_cached_fancy_regex($pattern).expect("Failed to compile fancy regex") }};
161}
162
163// Also make the macro available directly from this module
164pub use crate::regex_lazy;
165
166// URL patterns
167pub static URL_REGEX: LazyLock<Regex> =
168    LazyLock::new(|| Regex::new(r#"(?:https?|ftp)://[^\s<>\[\]()'"]+[^\s<>\[\]()"'.,]"#).unwrap());
169pub static BARE_URL_REGEX: LazyLock<Regex> =
170    LazyLock::new(|| Regex::new(r"(?:https?|ftp)://[^\s<>]+[^\s<>.]").unwrap());
171pub static URL_PATTERN: LazyLock<Regex> =
172    LazyLock::new(|| Regex::new(r"((?:https?|ftp)://[^\s\)<>]+[^\s\)<>.,])").unwrap());
173
174// Heading patterns
175pub static ATX_HEADING_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s*)(#{1,6})(\s+|$)").unwrap());
176pub static CLOSED_ATX_HEADING_REGEX: LazyLock<Regex> =
177    LazyLock::new(|| Regex::new(r"^(\s*)(#{1,6})(\s+)(.*)(\s+)(#+)(\s*)$").unwrap());
178pub static SETEXT_HEADING_REGEX: LazyLock<Regex> =
179    LazyLock::new(|| Regex::new(r"^(\s*)[^\s]+.*\n(\s*)(=+|-+)\s*$").unwrap());
180pub static TRAILING_PUNCTUATION_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"[.,:);!?]$").unwrap());
181
182// ATX heading patterns for MD051 and other rules
183pub static ATX_HEADING_WITH_CAPTURE: LazyLock<Regex> =
184    LazyLock::new(|| Regex::new(r"^(#{1,6})\s+(.+?)(?:\s+#*\s*)?$").unwrap());
185pub static SETEXT_HEADING_WITH_CAPTURE: LazyLock<FancyRegex> =
186    LazyLock::new(|| FancyRegex::new(r"^([^\n]+)\n([=\-])\2+\s*$").unwrap());
187
188// List patterns
189pub static UNORDERED_LIST_MARKER_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s*)([*+-])(\s+)").unwrap());
190pub static ORDERED_LIST_MARKER_REGEX: LazyLock<Regex> =
191    LazyLock::new(|| Regex::new(r"^(\s*)(\d+)([.)])(\s+)").unwrap());
192pub static LIST_MARKER_ANY_REGEX: LazyLock<Regex> =
193    LazyLock::new(|| Regex::new(r"^(\s*)(?:([*+-])|(\d+)[.)])(\s+)").unwrap());
194
195// Code block patterns
196pub static FENCED_CODE_BLOCK_START_REGEX: LazyLock<Regex> =
197    LazyLock::new(|| Regex::new(r"^(\s*)(```|~~~)(.*)$").unwrap());
198pub static FENCED_CODE_BLOCK_END_REGEX: LazyLock<Regex> =
199    LazyLock::new(|| Regex::new(r"^(\s*)(```|~~~)(\s*)$").unwrap());
200pub static INDENTED_CODE_BLOCK_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s{4,})(.*)$").unwrap());
201pub static CODE_FENCE_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(`{3,}|~{3,})").unwrap());
202
203// Emphasis patterns
204pub static EMPHASIS_REGEX: LazyLock<FancyRegex> =
205    LazyLock::new(|| FancyRegex::new(r"(\s|^)(\*{1,2}|_{1,2})(?=\S)(.+?)(?<=\S)(\2)(\s|$)").unwrap());
206pub static SPACE_IN_EMPHASIS_REGEX: LazyLock<FancyRegex> =
207    LazyLock::new(|| FancyRegex::new(r"(\*|_)(\s+)(.+?)(\s+)(\1)").unwrap());
208
209// MD037 specific emphasis patterns - improved to avoid false positives
210// Only match emphasis with spaces that are actually complete emphasis blocks
211// Use word boundaries and negative lookbehind/lookahead to avoid matching across emphasis boundaries
212pub static ASTERISK_EMPHASIS: LazyLock<Regex> =
213    LazyLock::new(|| Regex::new(r"(?:^|[^*])\*(\s+[^*]+\s*|\s*[^*]+\s+)\*(?:[^*]|$)").unwrap());
214pub static UNDERSCORE_EMPHASIS: LazyLock<Regex> =
215    LazyLock::new(|| Regex::new(r"(?:^|[^_])_(\s+[^_]+\s*|\s*[^_]+\s+)_(?:[^_]|$)").unwrap());
216pub static DOUBLE_UNDERSCORE_EMPHASIS: LazyLock<Regex> =
217    LazyLock::new(|| Regex::new(r"(?:^|[^_])__(\s+[^_]+\s*|\s*[^_]+\s+)__(?:[^_]|$)").unwrap());
218pub static DOUBLE_ASTERISK_EMPHASIS: LazyLock<FancyRegex> =
219    LazyLock::new(|| FancyRegex::new(r"\*\*\s+([^*]+?)\s+\*\*").unwrap());
220pub static DOUBLE_ASTERISK_SPACE_START: LazyLock<FancyRegex> =
221    LazyLock::new(|| FancyRegex::new(r"\*\*\s+([^*]+?)\*\*").unwrap());
222pub static DOUBLE_ASTERISK_SPACE_END: LazyLock<FancyRegex> =
223    LazyLock::new(|| FancyRegex::new(r"\*\*([^*]+?)\s+\*\*").unwrap());
224
225// Code block patterns
226pub static FENCED_CODE_BLOCK_START: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s*)```(?:[^`\r\n]*)$").unwrap());
227pub static FENCED_CODE_BLOCK_END: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s*)```\s*$").unwrap());
228pub static ALTERNATE_FENCED_CODE_BLOCK_START: LazyLock<Regex> =
229    LazyLock::new(|| Regex::new(r"^(\s*)~~~(?:[^~\r\n]*)$").unwrap());
230pub static ALTERNATE_FENCED_CODE_BLOCK_END: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s*)~~~\s*$").unwrap());
231pub static INDENTED_CODE_BLOCK_PATTERN: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s{4,})").unwrap());
232
233// HTML patterns
234pub static HTML_TAG_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"<([a-zA-Z][^>]*)>").unwrap());
235pub static HTML_SELF_CLOSING_TAG_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"<([a-zA-Z][^>]*/)>").unwrap());
236pub static HTML_TAG_FINDER: LazyLock<Regex> = LazyLock::new(|| Regex::new("(?i)</?[a-zA-Z][^>]*>").unwrap());
237pub static HTML_OPENING_TAG_FINDER: LazyLock<Regex> = LazyLock::new(|| Regex::new("(?i)<[a-zA-Z][^>]*>").unwrap());
238pub static HTML_TAG_QUICK_CHECK: LazyLock<Regex> = LazyLock::new(|| Regex::new("(?i)</?[a-zA-Z]").unwrap());
239
240// Link patterns for MD051 and other rules
241pub static LINK_REFERENCE_DEFINITION_REGEX: LazyLock<Regex> =
242    LazyLock::new(|| Regex::new(r"^\s*\[([^\]]+)\]:\s+(.+)$").unwrap());
243pub static INLINE_LINK_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\[([^\]]+)\]\(([^)]+)\)").unwrap());
244pub static LINK_TEXT_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\[([^\]]*)\]").unwrap());
245pub static LINK_REGEX: LazyLock<FancyRegex> =
246    LazyLock::new(|| FancyRegex::new(r"(?<!\\)\[([^\]]*)\]\(([^)#]*)#([^)]+)\)").unwrap());
247pub static EXTERNAL_URL_REGEX: LazyLock<FancyRegex> =
248    LazyLock::new(|| FancyRegex::new(r"^(https?://|ftp://|www\.|[^/]+\.[a-z]{2,})").unwrap());
249
250// Image patterns
251pub static IMAGE_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"!\[([^\]]*)\]\(([^)]+)\)").unwrap());
252
253// Whitespace patterns
254pub static TRAILING_WHITESPACE_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\s+$").unwrap());
255pub static MULTIPLE_BLANK_LINES_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\n{3,}").unwrap());
256
257// Front matter patterns
258pub static FRONT_MATTER_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^---\n.*?\n---\n").unwrap());
259
260// MD051 specific patterns
261pub static INLINE_CODE_REGEX: LazyLock<FancyRegex> = LazyLock::new(|| FancyRegex::new(r"`[^`]+`").unwrap());
262pub static BOLD_ASTERISK_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\*\*(.+?)\*\*").unwrap());
263pub static BOLD_UNDERSCORE_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"__(.+?)__").unwrap());
264pub static ITALIC_ASTERISK_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\*([^*]+?)\*").unwrap());
265pub static ITALIC_UNDERSCORE_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"_([^_]+?)_").unwrap());
266pub static LINK_TEXT_FULL_REGEX: LazyLock<FancyRegex> =
267    LazyLock::new(|| FancyRegex::new(r"\[([^\]]*)\]\([^)]*\)").unwrap());
268pub static STRIKETHROUGH_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"~~(.+?)~~").unwrap());
269pub static MULTIPLE_HYPHENS: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"-{2,}").unwrap());
270pub static TOC_SECTION_START: LazyLock<Regex> =
271    LazyLock::new(|| Regex::new(r"^#+\s*(?:Table of Contents|Contents|TOC)\s*$").unwrap());
272
273// Blockquote patterns
274pub static BLOCKQUOTE_PREFIX_RE: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s*>+\s*)").unwrap());
275
276// MD013 specific patterns
277pub static IMAGE_REF_PATTERN: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^!\[.*?\]\[.*?\]$").unwrap());
278pub static LINK_REF_PATTERN: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^\[.*?\]:\s*https?://\S+$").unwrap());
279pub static URL_IN_TEXT: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"https?://\S+").unwrap());
280pub static SENTENCE_END: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"[.!?]\s+[A-Z]").unwrap());
281pub static ABBREVIATION: LazyLock<Regex> = LazyLock::new(|| {
282    Regex::new(r"\b(?:Mr|Mrs|Ms|Dr|Prof|Sr|Jr|vs|etc|i\.e|e\.g|Inc|Corp|Ltd|Co|St|Ave|Blvd|Rd|Ph\.D|M\.D|B\.A|M\.A|Ph\.D|U\.S|U\.K|U\.N|N\.Y|L\.A|D\.C)\.\s+[A-Z]").unwrap()
283});
284pub static DECIMAL_NUMBER: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\d+\.\s*\d+").unwrap());
285pub static LIST_ITEM: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^\s*\d+\.\s+").unwrap());
286pub static REFERENCE_LINK: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\[([^\]]*)\]\[([^\]]*)\]").unwrap());
287
288// Email pattern
289pub static EMAIL_PATTERN: LazyLock<Regex> =
290    LazyLock::new(|| Regex::new(r"[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}").unwrap());
291
292// Third lazy_static block for link and image patterns used by MD052 and text_reflow
293// Reference link patterns (shared by MD052 and text_reflow)
294// Pattern to match reference links: [text][reference] or [text][]
295pub static REF_LINK_REGEX: LazyLock<FancyRegex> =
296    LazyLock::new(|| FancyRegex::new(r"(?<!\\)\[((?:[^\[\]\\]|\\.|\[[^\]]*\])*)\]\[([^\]]*)\]").unwrap());
297
298// Pattern for shortcut reference links: [reference]
299// Must not be preceded by ] or ) (to avoid matching second part of [text][ref])
300// Must not be followed by [ or ( (to avoid matching first part of [text][ref] or [text](url))
301// The capturing group handles nested brackets to support cases like [`Union[T, None]`]
302pub static SHORTCUT_REF_REGEX: LazyLock<FancyRegex> =
303    LazyLock::new(|| FancyRegex::new(r"(?<![\\)\]])\[((?:[^\[\]\\]|\\.|\[[^\]]*\])*)\](?!\s*[\[\(])").unwrap());
304
305// Inline link with fancy regex for better escaping handling (used by text_reflow)
306pub static INLINE_LINK_FANCY_REGEX: LazyLock<FancyRegex> =
307    LazyLock::new(|| FancyRegex::new(r"(?<!\\)\[([^\]]+)\]\(([^)]+)\)").unwrap());
308
309// Inline image with fancy regex (used by MD052 and text_reflow)
310pub static INLINE_IMAGE_FANCY_REGEX: LazyLock<FancyRegex> =
311    LazyLock::new(|| FancyRegex::new(r"!\[([^\]]*)\]\(([^)]+)\)").unwrap());
312
313// Linked images (clickable badges) - all 4 variants
314// Must be detected before inline_image and inline_link to treat as atomic units
315//
316// Limitation: Alt text containing brackets like [![[v1.0]](img)](link) is not supported.
317// The [^\]]* pattern cannot match nested brackets. This is rare in practice.
318//
319// Pattern 1: Inline image in inline link - [![alt](img-url)](link-url)
320pub static LINKED_IMAGE_INLINE_INLINE: LazyLock<FancyRegex> =
321    LazyLock::new(|| FancyRegex::new(r"\[!\[([^\]]*)\]\(([^)]+)\)\]\(([^)]+)\)").unwrap());
322
323// Pattern 2: Reference image in inline link - [![alt][img-ref]](link-url)
324pub static LINKED_IMAGE_REF_INLINE: LazyLock<FancyRegex> =
325    LazyLock::new(|| FancyRegex::new(r"\[!\[([^\]]*)\]\[([^\]]*)\]\]\(([^)]+)\)").unwrap());
326
327// Pattern 3: Inline image in reference link - [![alt](img-url)][link-ref]
328pub static LINKED_IMAGE_INLINE_REF: LazyLock<FancyRegex> =
329    LazyLock::new(|| FancyRegex::new(r"\[!\[([^\]]*)\]\(([^)]+)\)\]\[([^\]]*)\]").unwrap());
330
331// Pattern 4: Reference image in reference link - [![alt][img-ref]][link-ref]
332pub static LINKED_IMAGE_REF_REF: LazyLock<FancyRegex> =
333    LazyLock::new(|| FancyRegex::new(r"\[!\[([^\]]*)\]\[([^\]]*)\]\]\[([^\]]*)\]").unwrap());
334
335// Reference image: ![alt][ref] or ![alt][]
336pub static REF_IMAGE_REGEX: LazyLock<FancyRegex> =
337    LazyLock::new(|| FancyRegex::new(r"!\[((?:[^\[\]\\]|\\.|\[[^\]]*\])*)\]\[([^\]]*)\]").unwrap());
338
339// Footnote reference: [^note]
340pub static FOOTNOTE_REF_REGEX: LazyLock<FancyRegex> = LazyLock::new(|| FancyRegex::new(r"\[\^([^\]]+)\]").unwrap());
341
342// Strikethrough with fancy regex: ~~text~~
343pub static STRIKETHROUGH_FANCY_REGEX: LazyLock<FancyRegex> = LazyLock::new(|| FancyRegex::new(r"~~([^~]+)~~").unwrap());
344
345// Wiki-style links: [[wiki]] or [[wiki|display text]]
346pub static WIKI_LINK_REGEX: LazyLock<FancyRegex> = LazyLock::new(|| FancyRegex::new(r"\[\[([^\]]+)\]\]").unwrap());
347
348// Math formulas: $inline$ or $$display$$
349pub static INLINE_MATH_REGEX: LazyLock<FancyRegex> =
350    LazyLock::new(|| FancyRegex::new(r"(?<!\$)\$(?!\$)([^\$]+)\$(?!\$)").unwrap());
351pub static DISPLAY_MATH_REGEX: LazyLock<FancyRegex> = LazyLock::new(|| FancyRegex::new(r"\$\$([^\$]+)\$\$").unwrap());
352
353// Emoji shortcodes: :emoji:
354pub static EMOJI_SHORTCODE_REGEX: LazyLock<FancyRegex> =
355    LazyLock::new(|| FancyRegex::new(r":([a-zA-Z0-9_+-]+):").unwrap());
356
357// HTML tags (opening, closing, self-closing)
358pub static HTML_TAG_PATTERN: LazyLock<FancyRegex> =
359    LazyLock::new(|| FancyRegex::new(r"</?[a-zA-Z][^>]*>|<[a-zA-Z][^>]*/\s*>").unwrap());
360
361// HTML entities: &nbsp; &mdash; etc
362pub static HTML_ENTITY_REGEX: LazyLock<FancyRegex> =
363    LazyLock::new(|| FancyRegex::new(r"&[a-zA-Z][a-zA-Z0-9]*;|&#\d+;|&#x[0-9a-fA-F]+;").unwrap());
364
365// Fourth lazy_static block for additional patterns
366// HTML comment patterns
367pub static HTML_COMMENT_START: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"<!--").unwrap());
368pub static HTML_COMMENT_END: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"-->").unwrap());
369pub static HTML_COMMENT_PATTERN: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"<!--[\s\S]*?-->").unwrap());
370
371// HTML heading pattern (matches <h1> through <h6> tags)
372pub static HTML_HEADING_PATTERN: LazyLock<FancyRegex> =
373    LazyLock::new(|| FancyRegex::new(r"^\s*<h([1-6])(?:\s[^>]*)?>.*</h\1>\s*$").unwrap());
374
375// Heading quick check pattern
376pub static HEADING_CHECK: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"(?m)^(?:\s*)#").unwrap());
377
378// Horizontal rule patterns
379pub static HR_DASH: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^\-{3,}\s*$").unwrap());
380pub static HR_ASTERISK: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^\*{3,}\s*$").unwrap());
381pub static HR_UNDERSCORE: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^_{3,}\s*$").unwrap());
382pub static HR_SPACED_DASH: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\-\s+){2,}\-\s*$").unwrap());
383pub static HR_SPACED_ASTERISK: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\*\s+){2,}\*\s*$").unwrap());
384pub static HR_SPACED_UNDERSCORE: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(_\s+){2,}_\s*$").unwrap());
385
386/// Utility functions for quick content checks
387/// Check if content contains any headings (quick check before regex)
388pub fn has_heading_markers(content: &str) -> bool {
389    content.contains('#')
390}
391
392/// Check if content contains any lists (quick check before regex)
393pub fn has_list_markers(content: &str) -> bool {
394    content.contains('*')
395        || content.contains('-')
396        || content.contains('+')
397        || (content.contains('.') && content.contains(|c: char| c.is_ascii_digit()))
398}
399
400/// Check if content contains any code blocks (quick check before regex)
401pub fn has_code_block_markers(content: &str) -> bool {
402    content.contains("```") || content.contains("~~~") || content.contains("\n    ")
403    // Indented code block potential
404}
405
406/// Check if content contains any emphasis markers (quick check before regex)
407pub fn has_emphasis_markers(content: &str) -> bool {
408    content.contains('*') || content.contains('_')
409}
410
411/// Check if content contains any HTML tags (quick check before regex)
412pub fn has_html_tags(content: &str) -> bool {
413    content.contains('<') && (content.contains('>') || content.contains("/>"))
414}
415
416/// Check if content contains any links (quick check before regex)
417pub fn has_link_markers(content: &str) -> bool {
418    (content.contains('[') && content.contains(']'))
419        || content.contains("http://")
420        || content.contains("https://")
421        || content.contains("ftp://")
422}
423
424/// Check if content contains any images (quick check before regex)
425pub fn has_image_markers(content: &str) -> bool {
426    content.contains("![")
427}
428
429/// Optimize URL detection by implementing a character-by-character scanner
430/// that's much faster than regex for cases where we know there's no URL
431pub fn contains_url(content: &str) -> bool {
432    // Fast check - if these substrings aren't present, there's no URL
433    if !content.contains("://") {
434        return false;
435    }
436
437    let chars: Vec<char> = content.chars().collect();
438    let mut i = 0;
439
440    while i < chars.len() {
441        // Look for the start of a URL protocol
442        if i + 2 < chars.len()
443            && ((chars[i] == 'h' && chars[i + 1] == 't' && chars[i + 2] == 't')
444                || (chars[i] == 'f' && chars[i + 1] == 't' && chars[i + 2] == 'p'))
445        {
446            // Scan forward to find "://"
447            let mut j = i;
448            while j + 2 < chars.len() {
449                if chars[j] == ':' && chars[j + 1] == '/' && chars[j + 2] == '/' {
450                    return true;
451                }
452                j += 1;
453
454                // Don't scan too far ahead for the protocol
455                if j > i + 10 {
456                    break;
457                }
458            }
459        }
460        i += 1;
461    }
462
463    false
464}
465
466/// Escapes a string to be used in a regex pattern
467pub fn escape_regex(s: &str) -> String {
468    let special_chars = ['.', '+', '*', '?', '^', '$', '(', ')', '[', ']', '{', '}', '|', '\\'];
469    let mut result = String::with_capacity(s.len() * 2);
470
471    for c in s.chars() {
472        if special_chars.contains(&c) {
473            result.push('\\');
474        }
475        result.push(c);
476    }
477
478    result
479}
480
481#[cfg(test)]
482mod tests {
483    use super::*;
484
485    #[test]
486    fn test_regex_cache_new() {
487        let cache = RegexCache::new();
488        assert!(cache.cache.is_empty());
489        assert!(cache.fancy_cache.is_empty());
490        assert!(cache.usage_stats.is_empty());
491    }
492
493    #[test]
494    fn test_regex_cache_default() {
495        let cache = RegexCache::default();
496        assert!(cache.cache.is_empty());
497        assert!(cache.fancy_cache.is_empty());
498        assert!(cache.usage_stats.is_empty());
499    }
500
501    #[test]
502    fn test_get_regex_compilation() {
503        let mut cache = RegexCache::new();
504
505        // First call compiles and caches
506        let regex1 = cache.get_regex(r"\d+").unwrap();
507        assert_eq!(cache.cache.len(), 1);
508        assert_eq!(cache.usage_stats.get(r"\d+"), Some(&1));
509
510        // Second call returns cached version
511        let regex2 = cache.get_regex(r"\d+").unwrap();
512        assert_eq!(cache.cache.len(), 1);
513        assert_eq!(cache.usage_stats.get(r"\d+"), Some(&2));
514
515        // Both should be the same Arc
516        assert!(Arc::ptr_eq(&regex1, &regex2));
517    }
518
519    #[test]
520    fn test_get_regex_invalid_pattern() {
521        let mut cache = RegexCache::new();
522        let result = cache.get_regex(r"[unterminated");
523        assert!(result.is_err());
524        assert!(cache.cache.is_empty());
525    }
526
527    #[test]
528    fn test_get_fancy_regex_compilation() {
529        let mut cache = RegexCache::new();
530
531        // First call compiles and caches
532        let regex1 = cache.get_fancy_regex(r"(?<=foo)bar").unwrap();
533        assert_eq!(cache.fancy_cache.len(), 1);
534        assert_eq!(cache.usage_stats.get(r"(?<=foo)bar"), Some(&1));
535
536        // Second call returns cached version
537        let regex2 = cache.get_fancy_regex(r"(?<=foo)bar").unwrap();
538        assert_eq!(cache.fancy_cache.len(), 1);
539        assert_eq!(cache.usage_stats.get(r"(?<=foo)bar"), Some(&2));
540
541        // Both should be the same Arc
542        assert!(Arc::ptr_eq(&regex1, &regex2));
543    }
544
545    #[test]
546    fn test_get_fancy_regex_invalid_pattern() {
547        let mut cache = RegexCache::new();
548        let result = cache.get_fancy_regex(r"(?<=invalid");
549        assert!(result.is_err());
550        assert!(cache.fancy_cache.is_empty());
551    }
552
553    #[test]
554    fn test_get_stats() {
555        let mut cache = RegexCache::new();
556
557        // Use some patterns
558        let _ = cache.get_regex(r"\d+").unwrap();
559        let _ = cache.get_regex(r"\d+").unwrap();
560        let _ = cache.get_regex(r"\w+").unwrap();
561        let _ = cache.get_fancy_regex(r"(?<=foo)bar").unwrap();
562
563        let stats = cache.get_stats();
564        assert_eq!(stats.get(r"\d+"), Some(&2));
565        assert_eq!(stats.get(r"\w+"), Some(&1));
566        assert_eq!(stats.get(r"(?<=foo)bar"), Some(&1));
567    }
568
569    #[test]
570    fn test_clear_cache() {
571        let mut cache = RegexCache::new();
572
573        // Add some patterns
574        let _ = cache.get_regex(r"\d+").unwrap();
575        let _ = cache.get_fancy_regex(r"(?<=foo)bar").unwrap();
576
577        assert!(!cache.cache.is_empty());
578        assert!(!cache.fancy_cache.is_empty());
579        assert!(!cache.usage_stats.is_empty());
580
581        // Clear cache
582        cache.clear();
583
584        assert!(cache.cache.is_empty());
585        assert!(cache.fancy_cache.is_empty());
586        assert!(cache.usage_stats.is_empty());
587    }
588
589    #[test]
590    fn test_global_cache_functions() {
591        // Test get_cached_regex
592        let regex1 = get_cached_regex(r"\d{3}").unwrap();
593        let regex2 = get_cached_regex(r"\d{3}").unwrap();
594        assert!(Arc::ptr_eq(&regex1, &regex2));
595
596        // Test get_cached_fancy_regex
597        let fancy1 = get_cached_fancy_regex(r"(?<=test)ing").unwrap();
598        let fancy2 = get_cached_fancy_regex(r"(?<=test)ing").unwrap();
599        assert!(Arc::ptr_eq(&fancy1, &fancy2));
600
601        // Test stats
602        let stats = get_cache_stats();
603        assert!(stats.contains_key(r"\d{3}"));
604        assert!(stats.contains_key(r"(?<=test)ing"));
605    }
606
607    #[test]
608    fn test_regex_lazy_macro() {
609        let re = regex_lazy!(r"^test.*end$");
610        assert!(re.is_match("test something end"));
611        assert!(!re.is_match("test something"));
612
613        // The macro creates a new static for each invocation location,
614        // so we can't test pointer equality across different invocations
615        // But we can test that the regex works correctly
616        let re2 = regex_lazy!(r"^start.*finish$");
617        assert!(re2.is_match("start and finish"));
618        assert!(!re2.is_match("start without end"));
619    }
620
621    #[test]
622    fn test_has_heading_markers() {
623        assert!(has_heading_markers("# Heading"));
624        assert!(has_heading_markers("Text with # symbol"));
625        assert!(!has_heading_markers("Text without heading marker"));
626    }
627
628    #[test]
629    fn test_has_list_markers() {
630        assert!(has_list_markers("* Item"));
631        assert!(has_list_markers("- Item"));
632        assert!(has_list_markers("+ Item"));
633        assert!(has_list_markers("1. Item"));
634        assert!(!has_list_markers("Text without list markers"));
635    }
636
637    #[test]
638    fn test_has_code_block_markers() {
639        assert!(has_code_block_markers("```code```"));
640        assert!(has_code_block_markers("~~~code~~~"));
641        assert!(has_code_block_markers("Text\n    indented code"));
642        assert!(!has_code_block_markers("Text without code blocks"));
643    }
644
645    #[test]
646    fn test_has_emphasis_markers() {
647        assert!(has_emphasis_markers("*emphasis*"));
648        assert!(has_emphasis_markers("_emphasis_"));
649        assert!(has_emphasis_markers("**bold**"));
650        assert!(has_emphasis_markers("__bold__"));
651        assert!(!has_emphasis_markers("no emphasis"));
652    }
653
654    #[test]
655    fn test_has_html_tags() {
656        assert!(has_html_tags("<div>content</div>"));
657        assert!(has_html_tags("<br/>"));
658        assert!(has_html_tags("<img src='test.jpg'>"));
659        assert!(!has_html_tags("no html tags"));
660        assert!(!has_html_tags("less than < but no tag"));
661    }
662
663    #[test]
664    fn test_has_link_markers() {
665        assert!(has_link_markers("[text](url)"));
666        assert!(has_link_markers("[reference][1]"));
667        assert!(has_link_markers("http://example.com"));
668        assert!(has_link_markers("https://example.com"));
669        assert!(has_link_markers("ftp://example.com"));
670        assert!(!has_link_markers("no links here"));
671    }
672
673    #[test]
674    fn test_has_image_markers() {
675        assert!(has_image_markers("![alt text](image.png)"));
676        assert!(has_image_markers("![](image.png)"));
677        assert!(!has_image_markers("[link](url)"));
678        assert!(!has_image_markers("no images"));
679    }
680
681    #[test]
682    fn test_contains_url() {
683        assert!(contains_url("http://example.com"));
684        assert!(contains_url("Text with https://example.com link"));
685        assert!(contains_url("ftp://example.com"));
686        assert!(!contains_url("Text without URL"));
687        assert!(!contains_url("http not followed by ://"));
688
689        // Edge cases
690        assert!(!contains_url("http"));
691        assert!(!contains_url("https"));
692        assert!(!contains_url("://"));
693        assert!(contains_url("Visit http://site.com now"));
694        assert!(contains_url("See https://secure.site.com/path"));
695    }
696
697    #[test]
698    fn test_contains_url_performance() {
699        // Test early exit for strings without "://"
700        let long_text = "a".repeat(10000);
701        assert!(!contains_url(&long_text));
702
703        // Test with URL at the end
704        let text_with_url = format!("{long_text}https://example.com");
705        assert!(contains_url(&text_with_url));
706    }
707
708    #[test]
709    fn test_escape_regex() {
710        assert_eq!(escape_regex("a.b"), "a\\.b");
711        assert_eq!(escape_regex("a+b*c"), "a\\+b\\*c");
712        assert_eq!(escape_regex("(test)"), "\\(test\\)");
713        assert_eq!(escape_regex("[a-z]"), "\\[a-z\\]");
714        assert_eq!(escape_regex("normal text"), "normal text");
715
716        // Test all special characters
717        assert_eq!(escape_regex(".$^{[(|)*+?\\"), "\\.\\$\\^\\{\\[\\(\\|\\)\\*\\+\\?\\\\");
718
719        // Test empty string
720        assert_eq!(escape_regex(""), "");
721
722        // Test mixed content
723        assert_eq!(escape_regex("test.com/path?query=1"), "test\\.com/path\\?query=1");
724    }
725
726    #[test]
727    fn test_static_regex_patterns() {
728        // Test URL patterns
729        assert!(URL_REGEX.is_match("https://example.com"));
730        assert!(URL_REGEX.is_match("http://test.org/path"));
731        assert!(URL_REGEX.is_match("ftp://files.com"));
732        assert!(!URL_REGEX.is_match("not a url"));
733
734        // Test heading patterns
735        assert!(ATX_HEADING_REGEX.is_match("# Heading"));
736        assert!(ATX_HEADING_REGEX.is_match("  ## Indented"));
737        assert!(ATX_HEADING_REGEX.is_match("### "));
738        assert!(!ATX_HEADING_REGEX.is_match("Not a heading"));
739
740        // Test list patterns
741        assert!(UNORDERED_LIST_MARKER_REGEX.is_match("* Item"));
742        assert!(UNORDERED_LIST_MARKER_REGEX.is_match("- Item"));
743        assert!(UNORDERED_LIST_MARKER_REGEX.is_match("+ Item"));
744        assert!(ORDERED_LIST_MARKER_REGEX.is_match("1. Item"));
745        assert!(ORDERED_LIST_MARKER_REGEX.is_match("99. Item"));
746
747        // Test code block patterns
748        assert!(FENCED_CODE_BLOCK_START_REGEX.is_match("```"));
749        assert!(FENCED_CODE_BLOCK_START_REGEX.is_match("```rust"));
750        assert!(FENCED_CODE_BLOCK_START_REGEX.is_match("~~~"));
751        assert!(FENCED_CODE_BLOCK_END_REGEX.is_match("```"));
752        assert!(FENCED_CODE_BLOCK_END_REGEX.is_match("~~~"));
753
754        // Test emphasis patterns
755        assert!(BOLD_ASTERISK_REGEX.is_match("**bold**"));
756        assert!(BOLD_UNDERSCORE_REGEX.is_match("__bold__"));
757        assert!(ITALIC_ASTERISK_REGEX.is_match("*italic*"));
758        assert!(ITALIC_UNDERSCORE_REGEX.is_match("_italic_"));
759
760        // Test HTML patterns
761        assert!(HTML_TAG_REGEX.is_match("<div>"));
762        assert!(HTML_TAG_REGEX.is_match("<span class='test'>"));
763        assert!(HTML_SELF_CLOSING_TAG_REGEX.is_match("<br/>"));
764        assert!(HTML_SELF_CLOSING_TAG_REGEX.is_match("<img src='test'/>"));
765
766        // Test whitespace patterns
767        assert!(TRAILING_WHITESPACE_REGEX.is_match("line with spaces   "));
768        assert!(TRAILING_WHITESPACE_REGEX.is_match("tabs\t\t"));
769        assert!(MULTIPLE_BLANK_LINES_REGEX.is_match("\n\n\n"));
770        assert!(MULTIPLE_BLANK_LINES_REGEX.is_match("\n\n\n\n"));
771
772        // Test blockquote pattern
773        assert!(BLOCKQUOTE_PREFIX_RE.is_match("> Quote"));
774        assert!(BLOCKQUOTE_PREFIX_RE.is_match("  > Indented quote"));
775        assert!(BLOCKQUOTE_PREFIX_RE.is_match(">> Nested"));
776    }
777
778    #[test]
779    fn test_thread_safety() {
780        use std::thread;
781
782        let handles: Vec<_> = (0..10)
783            .map(|i| {
784                thread::spawn(move || {
785                    let pattern = format!(r"\d{{{i}}}");
786                    let regex = get_cached_regex(&pattern).unwrap();
787                    assert!(regex.is_match(&"1".repeat(i)));
788                })
789            })
790            .collect();
791
792        for handle in handles {
793            handle.join().unwrap();
794        }
795    }
796}
rumdl_lib/utils/regex_cache.rs

rumdl_lib/utils/
regex_cache.rs
