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
99///
100/// If the mutex is poisoned (another thread panicked while holding the lock),
101/// this function recovers by clearing the cache and continuing. This ensures
102/// the library never panics due to mutex poisoning.
103pub fn get_cached_regex(pattern: &str) -> Result<Arc<Regex>, regex::Error> {
104    let mut cache = GLOBAL_REGEX_CACHE.lock().unwrap_or_else(|poisoned| {
105        // Recover from poisoned mutex by clearing the cache
106        let mut guard = poisoned.into_inner();
107        guard.clear();
108        guard
109    });
110    cache.get_regex(pattern)
111}
112
113/// Get a fancy regex from the global cache
114///
115/// If the mutex is poisoned (another thread panicked while holding the lock),
116/// this function recovers by clearing the cache and continuing. This ensures
117/// the library never panics due to mutex poisoning.
118pub fn get_cached_fancy_regex(pattern: &str) -> Result<Arc<FancyRegex>, Box<fancy_regex::Error>> {
119    let mut cache = GLOBAL_REGEX_CACHE.lock().unwrap_or_else(|poisoned| {
120        // Recover from poisoned mutex by clearing the cache
121        let mut guard = poisoned.into_inner();
122        guard.clear();
123        guard
124    });
125    cache.get_fancy_regex(pattern)
126}
127
128/// Get cache usage statistics
129///
130/// If the mutex is poisoned, returns an empty HashMap rather than panicking.
131pub fn get_cache_stats() -> HashMap<String, u64> {
132    match GLOBAL_REGEX_CACHE.lock() {
133        Ok(cache) => cache.get_stats(),
134        Err(_) => HashMap::new(),
135    }
136}
137
138/// Macro for defining a lazily-initialized, cached regex pattern.
139///
140/// Use this for ad-hoc regexes that are not already defined in this module.
141///
142/// # Panics
143///
144/// This macro will panic at initialization if the regex pattern is invalid.
145/// This is intentional for compile-time constant patterns - we want to catch
146/// invalid patterns during development, not at runtime.
147///
148/// # Example
149///
150/// ```
151/// use rumdl_lib::regex_lazy;
152/// let my_re = regex_lazy!(r"^foo.*bar$");
153/// assert!(my_re.is_match("foobar"));
154/// ```
155#[macro_export]
156macro_rules! regex_lazy {
157    ($pattern:expr) => {{
158        static REGEX: LazyLock<regex::Regex> = LazyLock::new(|| regex::Regex::new($pattern).unwrap());
159        &*REGEX
160    }};
161}
162
163/// Macro for getting regex from global cache.
164///
165/// # Panics
166///
167/// Panics if the regex pattern is invalid. This is acceptable for static patterns
168/// where we want to fail fast during development.
169#[macro_export]
170macro_rules! regex_cached {
171    ($pattern:expr) => {{ $crate::utils::regex_cache::get_cached_regex($pattern).expect("Failed to compile regex") }};
172}
173
174/// Macro for getting fancy regex from global cache.
175///
176/// # Panics
177///
178/// Panics if the regex pattern is invalid. This is acceptable for static patterns
179/// where we want to fail fast during development.
180#[macro_export]
181macro_rules! fancy_regex_cached {
182    ($pattern:expr) => {{ $crate::utils::regex_cache::get_cached_fancy_regex($pattern).expect("Failed to compile fancy regex") }};
183}
184
185// Also make the macro available directly from this module
186pub use crate::regex_lazy;
187
188// =============================================================================
189// URL REGEX PATTERNS - Centralized URL Detection
190// =============================================================================
191//
192// ## Pattern Hierarchy (use the most specific pattern for your needs):
193//
194// | Pattern              | Use Case                                    | Parens | Trailing Punct |
195// |----------------------|---------------------------------------------|--------|----------------|
196// | URL_STANDARD_REGEX   | MD034 bare URL detection with auto-fix      | Yes    | Captured*      |
197// | URL_WWW_REGEX        | www.domain URLs without protocol            | Yes    | Captured*      |
198// | URL_IPV6_REGEX       | IPv6 URLs like https://[::1]/path           | Yes    | Captured*      |
199// | URL_QUICK_CHECK_REGEX| Fast early-exit check (contains URL?)       | N/A    | N/A            |
200// | URL_SIMPLE_REGEX     | Content detection, line length exemption    | No     | Excluded       |
201//
202// *Trailing punctuation is captured by the regex; use trim_trailing_punctuation() to clean.
203//
204// ## Design Principles:
205// 1. Parentheses in paths are allowed for Wikipedia-style URLs (Issue #240)
206// 2. Host portion excludes / so path is captured separately
207// 3. Unbalanced trailing parens are handled by trim_trailing_punctuation()
208// 4. All patterns exclude angle brackets <> to avoid matching autolinks
209//
210// ## URL Structure: protocol://host[:port][/path][?query][#fragment]
211
212/// Pattern for standard HTTP(S)/FTP(S) URLs with full path support.
213///
214/// Use this for bare URL detection where you need the complete URL including
215/// Wikipedia-style parentheses in paths. Trailing punctuation like `,;.!?` may
216/// be captured and should be trimmed by the caller.
217///
218/// # Examples
219/// - `https://example.com/path_(with_parens)?query#fragment`
220/// - `https://en.wikipedia.org/wiki/Rust_(programming_language)`
221pub const URL_STANDARD_STR: &str = concat!(
222    r#"(?:https?|ftps?|ftp)://"#, // Protocol
223    r#"(?:"#,
224    r#"\[[0-9a-fA-F:%.\-a-zA-Z]+\]"#, // IPv6 host OR
225    r#"|"#,
226    r#"[^\s<>\[\]()\\'\"`/]+"#, // Standard host (no parens, no /)
227    r#")"#,
228    r#"(?::\d+)?"#,                 // Optional port
229    r#"(?:/[^\s<>\[\]\\'\"`]*)?"#,  // Optional path (allows parens)
230    r#"(?:\?[^\s<>\[\]\\'\"`]*)?"#, // Optional query (allows parens)
231    r#"(?:#[^\s<>\[\]\\'\"`]*)?"#,  // Optional fragment (allows parens)
232);
233
234/// Pattern for www URLs without protocol.
235///
236/// Matches URLs starting with `www.` that lack a protocol prefix.
237/// These should be converted to proper URLs or flagged as bare URLs.
238/// Supports port, path, query string, and fragment like URL_STANDARD_STR.
239///
240/// # Examples
241/// - `www.example.com`
242/// - `www.example.com:8080`
243/// - `www.example.com/path`
244/// - `www.example.com?query=value`
245/// - `www.example.com#section`
246pub const URL_WWW_STR: &str = concat!(
247    r#"www\.(?:[a-zA-Z0-9][-a-zA-Z0-9]*\.)+[a-zA-Z]{2,}"#, // www.domain.tld
248    r#"(?::\d+)?"#,                                        // Optional port
249    r#"(?:/[^\s<>\[\]\\'\"`]*)?"#,                         // Optional path (allows parens)
250    r#"(?:\?[^\s<>\[\]\\'\"`]*)?"#,                        // Optional query (allows parens)
251    r#"(?:#[^\s<>\[\]\\'\"`]*)?"#,                         // Optional fragment (allows parens)
252);
253
254/// Pattern for IPv6 URLs specifically.
255///
256/// Matches URLs with IPv6 addresses in brackets, including zone identifiers.
257/// Examples: `https://[::1]/path`, `https://[fe80::1%eth0]:8080/`
258pub const URL_IPV6_STR: &str = concat!(
259    r#"(?:https?|ftps?|ftp)://"#,
260    r#"\[[0-9a-fA-F:%.\-a-zA-Z]+\]"#, // IPv6 host in brackets
261    r#"(?::\d+)?"#,                   // Optional port
262    r#"(?:/[^\s<>\[\]\\'\"`]*)?"#,    // Optional path
263    r#"(?:\?[^\s<>\[\]\\'\"`]*)?"#,   // Optional query
264    r#"(?:#[^\s<>\[\]\\'\"`]*)?"#,    // Optional fragment
265);
266
267/// Pattern for XMPP URIs per GFM extended autolinks specification.
268///
269/// XMPP URIs use the format `xmpp:user@domain/resource` (without `://`).
270/// Reference: <https://github.github.com/gfm/#autolinks-extension->
271///
272/// # Examples
273/// - `xmpp:foo@bar.baz`
274/// - `xmpp:foo@bar.baz/txt`
275pub const XMPP_URI_STR: &str = r#"xmpp:[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}(?:/[^\s<>\[\]\\'\"`]*)?"#;
276
277/// Quick check pattern for early exits.
278///
279/// Use this for fast pre-filtering before running more expensive patterns.
280/// Matches if the text likely contains a URL or email address.
281/// Includes `xmpp:` for GFM extended autolinks.
282pub const URL_QUICK_CHECK_STR: &str = r#"(?:https?|ftps?|ftp|xmpp)://|xmpp:|@|www\."#;
283
284/// Simple URL pattern for content detection.
285///
286/// Less strict pattern that excludes trailing sentence punctuation (.,).
287/// Use for line length exemption checks or content characteristic detection
288/// where you just need to know if a URL exists, not extract it precisely.
289pub const URL_SIMPLE_STR: &str = r#"(?:https?|ftps?|ftp)://[^\s<>]+[^\s<>.,]"#;
290
291// Pre-compiled static patterns for performance
292
293/// Standard URL regex - primary pattern for bare URL detection (MD034).
294/// See [`URL_STANDARD_STR`] for documentation.
295pub static URL_STANDARD_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(URL_STANDARD_STR).unwrap());
296
297/// WWW URL regex - for URLs starting with www. without protocol.
298/// See [`URL_WWW_STR`] for documentation.
299pub static URL_WWW_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(URL_WWW_STR).unwrap());
300
301/// IPv6 URL regex - for URLs with IPv6 addresses.
302/// See [`URL_IPV6_STR`] for documentation.
303pub static URL_IPV6_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(URL_IPV6_STR).unwrap());
304
305/// Quick check regex - fast early-exit test.
306/// See [`URL_QUICK_CHECK_STR`] for documentation.
307pub static URL_QUICK_CHECK_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(URL_QUICK_CHECK_STR).unwrap());
308
309/// Simple URL regex - for content detection and line length exemption.
310/// See [`URL_SIMPLE_STR`] for documentation.
311pub static URL_SIMPLE_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(URL_SIMPLE_STR).unwrap());
312
313/// Alias for `URL_SIMPLE_REGEX`. Used by MD013 for line length exemption.
314pub static URL_PATTERN: LazyLock<Regex> = LazyLock::new(|| URL_SIMPLE_REGEX.clone());
315
316/// XMPP URI regex - for GFM extended autolinks.
317/// See [`XMPP_URI_STR`] for documentation.
318pub static XMPP_URI_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(XMPP_URI_STR).unwrap());
319
320// Heading patterns
321pub static ATX_HEADING_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s*)(#{1,6})(\s+|$)").unwrap());
322pub static CLOSED_ATX_HEADING_REGEX: LazyLock<Regex> =
323    LazyLock::new(|| Regex::new(r"^(\s*)(#{1,6})(\s+)(.*)(\s+)(#+)(\s*)$").unwrap());
324pub static SETEXT_HEADING_REGEX: LazyLock<Regex> =
325    LazyLock::new(|| Regex::new(r"^(\s*)[^\s]+.*\n(\s*)(=+|-+)\s*$").unwrap());
326pub static TRAILING_PUNCTUATION_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"[.,:);!?]$").unwrap());
327
328// ATX heading patterns for MD051 and other rules
329pub static ATX_HEADING_WITH_CAPTURE: LazyLock<Regex> =
330    LazyLock::new(|| Regex::new(r"^(#{1,6})\s+(.+?)(?:\s+#*\s*)?$").unwrap());
331pub static SETEXT_HEADING_WITH_CAPTURE: LazyLock<FancyRegex> =
332    LazyLock::new(|| FancyRegex::new(r"^([^\n]+)\n([=\-])\2+\s*$").unwrap());
333
334// List patterns
335pub static UNORDERED_LIST_MARKER_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s*)([*+-])(\s+)").unwrap());
336pub static ORDERED_LIST_MARKER_REGEX: LazyLock<Regex> =
337    LazyLock::new(|| Regex::new(r"^(\s*)(\d+)([.)])(\s+)").unwrap());
338pub static LIST_MARKER_ANY_REGEX: LazyLock<Regex> =
339    LazyLock::new(|| Regex::new(r"^(\s*)(?:([*+-])|(\d+)[.)])(\s+)").unwrap());
340
341// Code block patterns
342pub static FENCED_CODE_BLOCK_START_REGEX: LazyLock<Regex> =
343    LazyLock::new(|| Regex::new(r"^(\s*)(```|~~~)(.*)$").unwrap());
344pub static FENCED_CODE_BLOCK_END_REGEX: LazyLock<Regex> =
345    LazyLock::new(|| Regex::new(r"^(\s*)(```|~~~)(\s*)$").unwrap());
346pub static INDENTED_CODE_BLOCK_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s{4,})(.*)$").unwrap());
347pub static CODE_FENCE_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(`{3,}|~{3,})").unwrap());
348
349// Emphasis patterns
350pub static EMPHASIS_REGEX: LazyLock<FancyRegex> =
351    LazyLock::new(|| FancyRegex::new(r"(\s|^)(\*{1,2}|_{1,2})(?=\S)(.+?)(?<=\S)(\2)(\s|$)").unwrap());
352pub static SPACE_IN_EMPHASIS_REGEX: LazyLock<FancyRegex> =
353    LazyLock::new(|| FancyRegex::new(r"(\*|_)(\s+)(.+?)(\s+)(\1)").unwrap());
354
355// MD037 specific emphasis patterns - improved to avoid false positives
356// Only match emphasis with spaces that are actually complete emphasis blocks
357// Use word boundaries and negative lookbehind/lookahead to avoid matching across emphasis boundaries
358pub static ASTERISK_EMPHASIS: LazyLock<Regex> =
359    LazyLock::new(|| Regex::new(r"(?:^|[^*])\*(\s+[^*]+\s*|\s*[^*]+\s+)\*(?:[^*]|$)").unwrap());
360pub static UNDERSCORE_EMPHASIS: LazyLock<Regex> =
361    LazyLock::new(|| Regex::new(r"(?:^|[^_])_(\s+[^_]+\s*|\s*[^_]+\s+)_(?:[^_]|$)").unwrap());
362pub static DOUBLE_UNDERSCORE_EMPHASIS: LazyLock<Regex> =
363    LazyLock::new(|| Regex::new(r"(?:^|[^_])__(\s+[^_]+\s*|\s*[^_]+\s+)__(?:[^_]|$)").unwrap());
364pub static DOUBLE_ASTERISK_EMPHASIS: LazyLock<FancyRegex> =
365    LazyLock::new(|| FancyRegex::new(r"\*\*\s+([^*]+?)\s+\*\*").unwrap());
366pub static DOUBLE_ASTERISK_SPACE_START: LazyLock<FancyRegex> =
367    LazyLock::new(|| FancyRegex::new(r"\*\*\s+([^*]+?)\*\*").unwrap());
368pub static DOUBLE_ASTERISK_SPACE_END: LazyLock<FancyRegex> =
369    LazyLock::new(|| FancyRegex::new(r"\*\*([^*]+?)\s+\*\*").unwrap());
370
371// Code block patterns
372pub static FENCED_CODE_BLOCK_START: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s*)```(?:[^`\r\n]*)$").unwrap());
373pub static FENCED_CODE_BLOCK_END: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s*)```\s*$").unwrap());
374pub static ALTERNATE_FENCED_CODE_BLOCK_START: LazyLock<Regex> =
375    LazyLock::new(|| Regex::new(r"^(\s*)~~~(?:[^~\r\n]*)$").unwrap());
376pub static ALTERNATE_FENCED_CODE_BLOCK_END: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s*)~~~\s*$").unwrap());
377pub static INDENTED_CODE_BLOCK_PATTERN: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s{4,})").unwrap());
378
379// HTML patterns
380pub static HTML_TAG_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"<([a-zA-Z][^>]*)>").unwrap());
381pub static HTML_SELF_CLOSING_TAG_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"<([a-zA-Z][^>]*/)>").unwrap());
382pub static HTML_TAG_FINDER: LazyLock<Regex> = LazyLock::new(|| Regex::new("(?i)</?[a-zA-Z][^>]*>").unwrap());
383pub static HTML_OPENING_TAG_FINDER: LazyLock<Regex> = LazyLock::new(|| Regex::new("(?i)<[a-zA-Z][^>]*>").unwrap());
384pub static HTML_TAG_QUICK_CHECK: LazyLock<Regex> = LazyLock::new(|| Regex::new("(?i)</?[a-zA-Z]").unwrap());
385
386// Link patterns for MD051 and other rules
387pub static LINK_REFERENCE_DEFINITION_REGEX: LazyLock<Regex> =
388    LazyLock::new(|| Regex::new(r"^\s*\[([^\]]+)\]:\s+(.+)$").unwrap());
389pub static INLINE_LINK_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\[([^\]]+)\]\(([^)]+)\)").unwrap());
390pub static LINK_TEXT_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\[([^\]]*)\]").unwrap());
391pub static LINK_REGEX: LazyLock<FancyRegex> =
392    LazyLock::new(|| FancyRegex::new(r"(?<!\\)\[([^\]]*)\]\(([^)#]*)#([^)]+)\)").unwrap());
393pub static EXTERNAL_URL_REGEX: LazyLock<FancyRegex> =
394    LazyLock::new(|| FancyRegex::new(r"^(https?://|ftp://|www\.|[^/]+\.[a-z]{2,})").unwrap());
395
396// Image patterns
397pub static IMAGE_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"!\[([^\]]*)\]\(([^)]+)\)").unwrap());
398
399// Whitespace patterns
400pub static TRAILING_WHITESPACE_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\s+$").unwrap());
401pub static MULTIPLE_BLANK_LINES_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\n{3,}").unwrap());
402
403// Front matter patterns
404pub static FRONT_MATTER_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^---\n.*?\n---\n").unwrap());
405
406// MD051 specific patterns
407pub static INLINE_CODE_REGEX: LazyLock<FancyRegex> = LazyLock::new(|| FancyRegex::new(r"`[^`]+`").unwrap());
408pub static ITALIC_ASTERISK_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\*([^*]+?)\*").unwrap());
409pub static ITALIC_UNDERSCORE_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"_([^_]+?)_").unwrap());
410pub static LINK_TEXT_FULL_REGEX: LazyLock<FancyRegex> =
411    LazyLock::new(|| FancyRegex::new(r"\[([^\]]*)\]\([^)]*\)").unwrap());
412pub static STRIKETHROUGH_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"~~(.+?)~~").unwrap());
413pub static MULTIPLE_HYPHENS: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"-{2,}").unwrap());
414pub static TOC_SECTION_START: LazyLock<Regex> =
415    LazyLock::new(|| Regex::new(r"^#+\s*(?:Table of Contents|Contents|TOC)\s*$").unwrap());
416
417// Blockquote patterns
418pub static BLOCKQUOTE_PREFIX_RE: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s*>+\s*)").unwrap());
419
420/// Check if a line is blank in the context of blockquotes.
421///
422/// A line is considered "blank" if:
423/// - It's empty or contains only whitespace
424/// - It's a blockquote continuation line with no content (e.g., ">", ">>", "> ")
425///
426/// This is essential for rules like MD058 (blanks-around-tables), MD065 (blanks-around-horizontal-rules),
427/// and any other rule that needs to detect blank lines that might be inside blockquotes.
428///
429/// # Examples
430/// ```
431/// use rumdl_lib::utils::regex_cache::is_blank_in_blockquote_context;
432///
433/// assert!(is_blank_in_blockquote_context(""));           // Empty line
434/// assert!(is_blank_in_blockquote_context("   "));        // Whitespace only
435/// assert!(is_blank_in_blockquote_context(">"));          // Blockquote continuation
436/// assert!(is_blank_in_blockquote_context("> "));         // Blockquote with trailing space
437/// assert!(is_blank_in_blockquote_context(">>"));         // Nested blockquote
438/// assert!(is_blank_in_blockquote_context("> > "));       // Spaced nested blockquote
439/// assert!(!is_blank_in_blockquote_context("> text"));    // Blockquote with content
440/// assert!(!is_blank_in_blockquote_context("text"));      // Regular text
441/// ```
442pub fn is_blank_in_blockquote_context(line: &str) -> bool {
443    if line.trim().is_empty() {
444        return true;
445    }
446    // Check if line is a blockquote prefix with no content after it
447    // Handle spaced nested blockquotes like "> > " by recursively checking remainder
448    if let Some(m) = BLOCKQUOTE_PREFIX_RE.find(line) {
449        let remainder = &line[m.end()..];
450        // The remainder should be empty/whitespace OR another blockquote prefix (for spaced nesting)
451        is_blank_in_blockquote_context(remainder)
452    } else {
453        false
454    }
455}
456
457// MD013 specific patterns
458pub static IMAGE_REF_PATTERN: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^!\[.*?\]\[.*?\]$").unwrap());
459pub static LINK_REF_PATTERN: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^\[.*?\]:\s*https?://\S+$").unwrap());
460/// Greedy URL pattern for finding URLs in text for length calculation.
461///
462/// Pattern `https?://\S+` matches until whitespace, which may include trailing
463/// punctuation. This is intentional for MD013 line length calculation where
464/// we replace URLs with fixed-length placeholders.
465///
466/// For precise URL extraction, use `URL_STANDARD_REGEX` instead.
467pub static URL_IN_TEXT: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"https?://\S+").unwrap());
468pub static SENTENCE_END: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"[.!?]\s+[A-Z]").unwrap());
469pub static ABBREVIATION: LazyLock<Regex> = LazyLock::new(|| {
470    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()
471});
472pub static DECIMAL_NUMBER: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\d+\.\s*\d+").unwrap());
473pub static LIST_ITEM: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^\s*\d+\.\s+").unwrap());
474pub static REFERENCE_LINK: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"\[([^\]]*)\]\[([^\]]*)\]").unwrap());
475
476// Email pattern
477pub static EMAIL_PATTERN: LazyLock<Regex> =
478    LazyLock::new(|| Regex::new(r"[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}").unwrap());
479
480// Third lazy_static block for link and image patterns used by MD052 and text_reflow
481// Reference link patterns (shared by MD052 and text_reflow)
482// Pattern to match reference links: [text][reference] or [text][]
483pub static REF_LINK_REGEX: LazyLock<FancyRegex> =
484    LazyLock::new(|| FancyRegex::new(r"(?<!\\)\[((?:[^\[\]\\]|\\.|\[[^\]]*\])*)\]\[([^\]]*)\]").unwrap());
485
486// Pattern for shortcut reference links: [reference]
487// Must not be preceded by ] or ) (to avoid matching second part of [text][ref])
488// Must not be followed by [ or ( (to avoid matching first part of [text][ref] or [text](url))
489// The capturing group handles nested brackets to support cases like [`Union[T, None]`]
490pub static SHORTCUT_REF_REGEX: LazyLock<FancyRegex> =
491    LazyLock::new(|| FancyRegex::new(r"(?<![\\)\]])\[((?:[^\[\]\\]|\\.|\[[^\]]*\])*)\](?!\s*[\[\(])").unwrap());
492
493// Inline link with fancy regex for better escaping handling (used by text_reflow)
494pub static INLINE_LINK_FANCY_REGEX: LazyLock<FancyRegex> =
495    LazyLock::new(|| FancyRegex::new(r"(?<!\\)\[([^\]]+)\]\(([^)]+)\)").unwrap());
496
497// Inline image with fancy regex (used by MD052 and text_reflow)
498pub static INLINE_IMAGE_FANCY_REGEX: LazyLock<FancyRegex> =
499    LazyLock::new(|| FancyRegex::new(r"!\[([^\]]*)\]\(([^)]+)\)").unwrap());
500
501// Linked images (clickable badges) - all 4 variants
502// Must be detected before inline_image and inline_link to treat as atomic units
503//
504// Limitation: Alt text containing brackets like [![[v1.0]](img)](link) is not supported.
505// The [^\]]* pattern cannot match nested brackets. This is rare in practice.
506//
507// Pattern 1: Inline image in inline link - [![alt](img-url)](link-url)
508pub static LINKED_IMAGE_INLINE_INLINE: LazyLock<FancyRegex> =
509    LazyLock::new(|| FancyRegex::new(r"\[!\[([^\]]*)\]\(([^)]+)\)\]\(([^)]+)\)").unwrap());
510
511// Pattern 2: Reference image in inline link - [![alt][img-ref]](link-url)
512pub static LINKED_IMAGE_REF_INLINE: LazyLock<FancyRegex> =
513    LazyLock::new(|| FancyRegex::new(r"\[!\[([^\]]*)\]\[([^\]]*)\]\]\(([^)]+)\)").unwrap());
514
515// Pattern 3: Inline image in reference link - [![alt](img-url)][link-ref]
516pub static LINKED_IMAGE_INLINE_REF: LazyLock<FancyRegex> =
517    LazyLock::new(|| FancyRegex::new(r"\[!\[([^\]]*)\]\(([^)]+)\)\]\[([^\]]*)\]").unwrap());
518
519// Pattern 4: Reference image in reference link - [![alt][img-ref]][link-ref]
520pub static LINKED_IMAGE_REF_REF: LazyLock<FancyRegex> =
521    LazyLock::new(|| FancyRegex::new(r"\[!\[([^\]]*)\]\[([^\]]*)\]\]\[([^\]]*)\]").unwrap());
522
523// Reference image: ![alt][ref] or ![alt][]
524pub static REF_IMAGE_REGEX: LazyLock<FancyRegex> =
525    LazyLock::new(|| FancyRegex::new(r"!\[((?:[^\[\]\\]|\\.|\[[^\]]*\])*)\]\[([^\]]*)\]").unwrap());
526
527// Footnote reference: [^note]
528pub static FOOTNOTE_REF_REGEX: LazyLock<FancyRegex> = LazyLock::new(|| FancyRegex::new(r"\[\^([^\]]+)\]").unwrap());
529
530// Strikethrough with fancy regex: ~~text~~
531pub static STRIKETHROUGH_FANCY_REGEX: LazyLock<FancyRegex> = LazyLock::new(|| FancyRegex::new(r"~~([^~]+)~~").unwrap());
532
533// Wiki-style links: [[wiki]] or [[wiki|display text]]
534pub static WIKI_LINK_REGEX: LazyLock<FancyRegex> = LazyLock::new(|| FancyRegex::new(r"\[\[([^\]]+)\]\]").unwrap());
535
536// Math formulas: $inline$ or $$display$$
537pub static INLINE_MATH_REGEX: LazyLock<FancyRegex> =
538    LazyLock::new(|| FancyRegex::new(r"(?<!\$)\$(?!\$)([^\$]+)\$(?!\$)").unwrap());
539pub static DISPLAY_MATH_REGEX: LazyLock<FancyRegex> = LazyLock::new(|| FancyRegex::new(r"\$\$([^\$]+)\$\$").unwrap());
540
541// Emoji shortcodes: :emoji:
542pub static EMOJI_SHORTCODE_REGEX: LazyLock<FancyRegex> =
543    LazyLock::new(|| FancyRegex::new(r":([a-zA-Z0-9_+-]+):").unwrap());
544
545// HTML tags (opening, closing, self-closing)
546pub static HTML_TAG_PATTERN: LazyLock<FancyRegex> =
547    LazyLock::new(|| FancyRegex::new(r"</?[a-zA-Z][^>]*>|<[a-zA-Z][^>]*/\s*>").unwrap());
548
549// HTML entities: &nbsp; &mdash; etc
550pub static HTML_ENTITY_REGEX: LazyLock<FancyRegex> =
551    LazyLock::new(|| FancyRegex::new(r"&[a-zA-Z][a-zA-Z0-9]*;|&#\d+;|&#x[0-9a-fA-F]+;").unwrap());
552
553// Hugo/Go template shortcodes: {{< figure ... >}} and {{% shortcode %}}
554// Matches both delimiters: {{< ... >}} (shortcode) and {{% ... %}} (template)
555// Handles multi-line content with embedded quotes and newlines
556pub static HUGO_SHORTCODE_REGEX: LazyLock<FancyRegex> =
557    LazyLock::new(|| FancyRegex::new(r"\{\{[<%][\s\S]*?[%>]\}\}").unwrap());
558
559// HTML comment pattern
560pub static HTML_COMMENT_PATTERN: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"<!--[\s\S]*?-->").unwrap());
561
562// HTML heading pattern (matches <h1> through <h6> tags)
563pub static HTML_HEADING_PATTERN: LazyLock<FancyRegex> =
564    LazyLock::new(|| FancyRegex::new(r"^\s*<h([1-6])(?:\s[^>]*)?>.*</h\1>\s*$").unwrap());
565
566// Heading quick check pattern
567pub static HEADING_CHECK: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"(?m)^(?:\s*)#").unwrap());
568
569// Horizontal rule patterns
570pub static HR_DASH: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^\-{3,}\s*$").unwrap());
571pub static HR_ASTERISK: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^\*{3,}\s*$").unwrap());
572pub static HR_UNDERSCORE: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^_{3,}\s*$").unwrap());
573pub static HR_SPACED_DASH: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\-\s+){2,}\-\s*$").unwrap());
574pub static HR_SPACED_ASTERISK: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\*\s+){2,}\*\s*$").unwrap());
575pub static HR_SPACED_UNDERSCORE: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(_\s+){2,}_\s*$").unwrap());
576
577/// Utility functions for quick content checks
578/// Check if content contains any headings (quick check before regex)
579pub fn has_heading_markers(content: &str) -> bool {
580    content.contains('#')
581}
582
583/// Check if content contains any lists (quick check before regex)
584pub fn has_list_markers(content: &str) -> bool {
585    content.contains('*')
586        || content.contains('-')
587        || content.contains('+')
588        || (content.contains('.') && content.contains(|c: char| c.is_ascii_digit()))
589}
590
591/// Check if content contains any code blocks (quick check before regex)
592pub fn has_code_block_markers(content: &str) -> bool {
593    content.contains("```") || content.contains("~~~") || content.contains("\n    ")
594    // Indented code block potential
595}
596
597/// Check if content contains any emphasis markers (quick check before regex)
598pub fn has_emphasis_markers(content: &str) -> bool {
599    content.contains('*') || content.contains('_')
600}
601
602/// Check if content contains any HTML tags (quick check before regex)
603pub fn has_html_tags(content: &str) -> bool {
604    content.contains('<') && (content.contains('>') || content.contains("/>"))
605}
606
607/// Check if content contains any links (quick check before regex)
608pub fn has_link_markers(content: &str) -> bool {
609    (content.contains('[') && content.contains(']'))
610        || content.contains("http://")
611        || content.contains("https://")
612        || content.contains("ftp://")
613}
614
615/// Check if content contains any images (quick check before regex)
616pub fn has_image_markers(content: &str) -> bool {
617    content.contains("![")
618}
619
620/// Optimize URL detection by implementing a character-by-character scanner
621/// that's much faster than regex for cases where we know there's no URL
622pub fn contains_url(content: &str) -> bool {
623    // Fast check - if these substrings aren't present, there's no URL
624    if !content.contains("://") {
625        return false;
626    }
627
628    let chars: Vec<char> = content.chars().collect();
629    let mut i = 0;
630
631    while i < chars.len() {
632        // Look for the start of a URL protocol
633        if i + 2 < chars.len()
634            && ((chars[i] == 'h' && chars[i + 1] == 't' && chars[i + 2] == 't')
635                || (chars[i] == 'f' && chars[i + 1] == 't' && chars[i + 2] == 'p'))
636        {
637            // Scan forward to find "://"
638            let mut j = i;
639            while j + 2 < chars.len() {
640                if chars[j] == ':' && chars[j + 1] == '/' && chars[j + 2] == '/' {
641                    return true;
642                }
643                j += 1;
644
645                // Don't scan too far ahead for the protocol
646                if j > i + 10 {
647                    break;
648                }
649            }
650        }
651        i += 1;
652    }
653
654    false
655}
656
657/// Escapes a string to be used in a regex pattern
658pub fn escape_regex(s: &str) -> String {
659    let mut result = String::with_capacity(s.len() * 2);
660
661    for c in s.chars() {
662        // Use matches! for O(1) lookup instead of array.contains() which is O(n)
663        if matches!(
664            c,
665            '.' | '+' | '*' | '?' | '^' | '$' | '(' | ')' | '[' | ']' | '{' | '}' | '|' | '\\'
666        ) {
667            result.push('\\');
668        }
669        result.push(c);
670    }
671
672    result
673}
674
675#[cfg(test)]
676mod tests {
677    use super::*;
678
679    #[test]
680    fn test_regex_cache_new() {
681        let cache = RegexCache::new();
682        assert!(cache.cache.is_empty());
683        assert!(cache.fancy_cache.is_empty());
684        assert!(cache.usage_stats.is_empty());
685    }
686
687    #[test]
688    fn test_regex_cache_default() {
689        let cache = RegexCache::default();
690        assert!(cache.cache.is_empty());
691        assert!(cache.fancy_cache.is_empty());
692        assert!(cache.usage_stats.is_empty());
693    }
694
695    #[test]
696    fn test_get_regex_compilation() {
697        let mut cache = RegexCache::new();
698
699        // First call compiles and caches
700        let regex1 = cache.get_regex(r"\d+").unwrap();
701        assert_eq!(cache.cache.len(), 1);
702        assert_eq!(cache.usage_stats.get(r"\d+"), Some(&1));
703
704        // Second call returns cached version
705        let regex2 = cache.get_regex(r"\d+").unwrap();
706        assert_eq!(cache.cache.len(), 1);
707        assert_eq!(cache.usage_stats.get(r"\d+"), Some(&2));
708
709        // Both should be the same Arc
710        assert!(Arc::ptr_eq(&regex1, &regex2));
711    }
712
713    #[test]
714    fn test_get_regex_invalid_pattern() {
715        let mut cache = RegexCache::new();
716        let result = cache.get_regex(r"[unterminated");
717        assert!(result.is_err());
718        assert!(cache.cache.is_empty());
719    }
720
721    #[test]
722    fn test_get_fancy_regex_compilation() {
723        let mut cache = RegexCache::new();
724
725        // First call compiles and caches
726        let regex1 = cache.get_fancy_regex(r"(?<=foo)bar").unwrap();
727        assert_eq!(cache.fancy_cache.len(), 1);
728        assert_eq!(cache.usage_stats.get(r"(?<=foo)bar"), Some(&1));
729
730        // Second call returns cached version
731        let regex2 = cache.get_fancy_regex(r"(?<=foo)bar").unwrap();
732        assert_eq!(cache.fancy_cache.len(), 1);
733        assert_eq!(cache.usage_stats.get(r"(?<=foo)bar"), Some(&2));
734
735        // Both should be the same Arc
736        assert!(Arc::ptr_eq(&regex1, &regex2));
737    }
738
739    #[test]
740    fn test_get_fancy_regex_invalid_pattern() {
741        let mut cache = RegexCache::new();
742        let result = cache.get_fancy_regex(r"(?<=invalid");
743        assert!(result.is_err());
744        assert!(cache.fancy_cache.is_empty());
745    }
746
747    #[test]
748    fn test_get_stats() {
749        let mut cache = RegexCache::new();
750
751        // Use some patterns
752        let _ = cache.get_regex(r"\d+").unwrap();
753        let _ = cache.get_regex(r"\d+").unwrap();
754        let _ = cache.get_regex(r"\w+").unwrap();
755        let _ = cache.get_fancy_regex(r"(?<=foo)bar").unwrap();
756
757        let stats = cache.get_stats();
758        assert_eq!(stats.get(r"\d+"), Some(&2));
759        assert_eq!(stats.get(r"\w+"), Some(&1));
760        assert_eq!(stats.get(r"(?<=foo)bar"), Some(&1));
761    }
762
763    #[test]
764    fn test_clear_cache() {
765        let mut cache = RegexCache::new();
766
767        // Add some patterns
768        let _ = cache.get_regex(r"\d+").unwrap();
769        let _ = cache.get_fancy_regex(r"(?<=foo)bar").unwrap();
770
771        assert!(!cache.cache.is_empty());
772        assert!(!cache.fancy_cache.is_empty());
773        assert!(!cache.usage_stats.is_empty());
774
775        // Clear cache
776        cache.clear();
777
778        assert!(cache.cache.is_empty());
779        assert!(cache.fancy_cache.is_empty());
780        assert!(cache.usage_stats.is_empty());
781    }
782
783    #[test]
784    fn test_global_cache_functions() {
785        // Test get_cached_regex
786        let regex1 = get_cached_regex(r"\d{3}").unwrap();
787        let regex2 = get_cached_regex(r"\d{3}").unwrap();
788        assert!(Arc::ptr_eq(&regex1, &regex2));
789
790        // Test get_cached_fancy_regex
791        let fancy1 = get_cached_fancy_regex(r"(?<=test)ing").unwrap();
792        let fancy2 = get_cached_fancy_regex(r"(?<=test)ing").unwrap();
793        assert!(Arc::ptr_eq(&fancy1, &fancy2));
794
795        // Test stats
796        let stats = get_cache_stats();
797        assert!(stats.contains_key(r"\d{3}"));
798        assert!(stats.contains_key(r"(?<=test)ing"));
799    }
800
801    #[test]
802    fn test_regex_lazy_macro() {
803        let re = regex_lazy!(r"^test.*end$");
804        assert!(re.is_match("test something end"));
805        assert!(!re.is_match("test something"));
806
807        // The macro creates a new static for each invocation location,
808        // so we can't test pointer equality across different invocations
809        // But we can test that the regex works correctly
810        let re2 = regex_lazy!(r"^start.*finish$");
811        assert!(re2.is_match("start and finish"));
812        assert!(!re2.is_match("start without end"));
813    }
814
815    #[test]
816    fn test_has_heading_markers() {
817        assert!(has_heading_markers("# Heading"));
818        assert!(has_heading_markers("Text with # symbol"));
819        assert!(!has_heading_markers("Text without heading marker"));
820    }
821
822    #[test]
823    fn test_has_list_markers() {
824        assert!(has_list_markers("* Item"));
825        assert!(has_list_markers("- Item"));
826        assert!(has_list_markers("+ Item"));
827        assert!(has_list_markers("1. Item"));
828        assert!(!has_list_markers("Text without list markers"));
829    }
830
831    #[test]
832    fn test_has_code_block_markers() {
833        assert!(has_code_block_markers("```code```"));
834        assert!(has_code_block_markers("~~~code~~~"));
835        assert!(has_code_block_markers("Text\n    indented code"));
836        assert!(!has_code_block_markers("Text without code blocks"));
837    }
838
839    #[test]
840    fn test_has_emphasis_markers() {
841        assert!(has_emphasis_markers("*emphasis*"));
842        assert!(has_emphasis_markers("_emphasis_"));
843        assert!(has_emphasis_markers("**bold**"));
844        assert!(has_emphasis_markers("__bold__"));
845        assert!(!has_emphasis_markers("no emphasis"));
846    }
847
848    #[test]
849    fn test_has_html_tags() {
850        assert!(has_html_tags("<div>content</div>"));
851        assert!(has_html_tags("<br/>"));
852        assert!(has_html_tags("<img src='test.jpg'>"));
853        assert!(!has_html_tags("no html tags"));
854        assert!(!has_html_tags("less than < but no tag"));
855    }
856
857    #[test]
858    fn test_has_link_markers() {
859        assert!(has_link_markers("[text](url)"));
860        assert!(has_link_markers("[reference][1]"));
861        assert!(has_link_markers("http://example.com"));
862        assert!(has_link_markers("https://example.com"));
863        assert!(has_link_markers("ftp://example.com"));
864        assert!(!has_link_markers("no links here"));
865    }
866
867    #[test]
868    fn test_has_image_markers() {
869        assert!(has_image_markers("![alt text](image.png)"));
870        assert!(has_image_markers("![](image.png)"));
871        assert!(!has_image_markers("[link](url)"));
872        assert!(!has_image_markers("no images"));
873    }
874
875    #[test]
876    fn test_contains_url() {
877        assert!(contains_url("http://example.com"));
878        assert!(contains_url("Text with https://example.com link"));
879        assert!(contains_url("ftp://example.com"));
880        assert!(!contains_url("Text without URL"));
881        assert!(!contains_url("http not followed by ://"));
882
883        // Edge cases
884        assert!(!contains_url("http"));
885        assert!(!contains_url("https"));
886        assert!(!contains_url("://"));
887        assert!(contains_url("Visit http://site.com now"));
888        assert!(contains_url("See https://secure.site.com/path"));
889    }
890
891    #[test]
892    fn test_contains_url_performance() {
893        // Test early exit for strings without "://"
894        let long_text = "a".repeat(10000);
895        assert!(!contains_url(&long_text));
896
897        // Test with URL at the end
898        let text_with_url = format!("{long_text}https://example.com");
899        assert!(contains_url(&text_with_url));
900    }
901
902    #[test]
903    fn test_escape_regex() {
904        assert_eq!(escape_regex("a.b"), "a\\.b");
905        assert_eq!(escape_regex("a+b*c"), "a\\+b\\*c");
906        assert_eq!(escape_regex("(test)"), "\\(test\\)");
907        assert_eq!(escape_regex("[a-z]"), "\\[a-z\\]");
908        assert_eq!(escape_regex("normal text"), "normal text");
909
910        // Test all special characters
911        assert_eq!(escape_regex(".$^{[(|)*+?\\"), "\\.\\$\\^\\{\\[\\(\\|\\)\\*\\+\\?\\\\");
912
913        // Test empty string
914        assert_eq!(escape_regex(""), "");
915
916        // Test mixed content
917        assert_eq!(escape_regex("test.com/path?query=1"), "test\\.com/path\\?query=1");
918    }
919
920    #[test]
921    fn test_static_regex_patterns() {
922        // Test URL patterns
923        assert!(URL_SIMPLE_REGEX.is_match("https://example.com"));
924        assert!(URL_SIMPLE_REGEX.is_match("http://test.org/path"));
925        assert!(URL_SIMPLE_REGEX.is_match("ftp://files.com"));
926        assert!(!URL_SIMPLE_REGEX.is_match("not a url"));
927
928        // Test heading patterns
929        assert!(ATX_HEADING_REGEX.is_match("# Heading"));
930        assert!(ATX_HEADING_REGEX.is_match("  ## Indented"));
931        assert!(ATX_HEADING_REGEX.is_match("### "));
932        assert!(!ATX_HEADING_REGEX.is_match("Not a heading"));
933
934        // Test list patterns
935        assert!(UNORDERED_LIST_MARKER_REGEX.is_match("* Item"));
936        assert!(UNORDERED_LIST_MARKER_REGEX.is_match("- Item"));
937        assert!(UNORDERED_LIST_MARKER_REGEX.is_match("+ Item"));
938        assert!(ORDERED_LIST_MARKER_REGEX.is_match("1. Item"));
939        assert!(ORDERED_LIST_MARKER_REGEX.is_match("99. Item"));
940
941        // Test code block patterns
942        assert!(FENCED_CODE_BLOCK_START_REGEX.is_match("```"));
943        assert!(FENCED_CODE_BLOCK_START_REGEX.is_match("```rust"));
944        assert!(FENCED_CODE_BLOCK_START_REGEX.is_match("~~~"));
945        assert!(FENCED_CODE_BLOCK_END_REGEX.is_match("```"));
946        assert!(FENCED_CODE_BLOCK_END_REGEX.is_match("~~~"));
947
948        // Test emphasis patterns
949        assert!(ITALIC_ASTERISK_REGEX.is_match("*italic*"));
950        assert!(ITALIC_UNDERSCORE_REGEX.is_match("_italic_"));
951
952        // Test HTML patterns
953        assert!(HTML_TAG_REGEX.is_match("<div>"));
954        assert!(HTML_TAG_REGEX.is_match("<span class='test'>"));
955        assert!(HTML_SELF_CLOSING_TAG_REGEX.is_match("<br/>"));
956        assert!(HTML_SELF_CLOSING_TAG_REGEX.is_match("<img src='test'/>"));
957
958        // Test whitespace patterns
959        assert!(TRAILING_WHITESPACE_REGEX.is_match("line with spaces   "));
960        assert!(TRAILING_WHITESPACE_REGEX.is_match("tabs\t\t"));
961        assert!(MULTIPLE_BLANK_LINES_REGEX.is_match("\n\n\n"));
962        assert!(MULTIPLE_BLANK_LINES_REGEX.is_match("\n\n\n\n"));
963
964        // Test blockquote pattern
965        assert!(BLOCKQUOTE_PREFIX_RE.is_match("> Quote"));
966        assert!(BLOCKQUOTE_PREFIX_RE.is_match("  > Indented quote"));
967        assert!(BLOCKQUOTE_PREFIX_RE.is_match(">> Nested"));
968    }
969
970    #[test]
971    fn test_thread_safety() {
972        use std::thread;
973
974        let handles: Vec<_> = (0..10)
975            .map(|i| {
976                thread::spawn(move || {
977                    let pattern = format!(r"\d{{{i}}}");
978                    let regex = get_cached_regex(&pattern).unwrap();
979                    assert!(regex.is_match(&"1".repeat(i)));
980                })
981            })
982            .collect();
983
984        for handle in handles {
985            handle.join().unwrap();
986        }
987    }
988
989    // ==========================================================================
990    // Comprehensive URL Regex Tests
991    // ==========================================================================
992
993    #[test]
994    fn test_url_standard_basic() {
995        // Basic HTTP/HTTPS URLs
996        assert!(URL_STANDARD_REGEX.is_match("https://example.com"));
997        assert!(URL_STANDARD_REGEX.is_match("http://example.com"));
998        assert!(URL_STANDARD_REGEX.is_match("https://example.com/"));
999        assert!(URL_STANDARD_REGEX.is_match("https://example.com/path"));
1000        assert!(URL_STANDARD_REGEX.is_match("ftp://files.example.com"));
1001        assert!(URL_STANDARD_REGEX.is_match("ftps://secure.example.com"));
1002
1003        // Should not match non-URLs
1004        assert!(!URL_STANDARD_REGEX.is_match("not a url"));
1005        assert!(!URL_STANDARD_REGEX.is_match("example.com"));
1006        assert!(!URL_STANDARD_REGEX.is_match("www.example.com"));
1007    }
1008
1009    #[test]
1010    fn test_url_standard_with_path() {
1011        assert!(URL_STANDARD_REGEX.is_match("https://example.com/path/to/page"));
1012        assert!(URL_STANDARD_REGEX.is_match("https://example.com/path/to/page.html"));
1013        assert!(URL_STANDARD_REGEX.is_match("https://example.com/path/to/page/"));
1014    }
1015
1016    #[test]
1017    fn test_url_standard_with_query() {
1018        assert!(URL_STANDARD_REGEX.is_match("https://example.com?query=value"));
1019        assert!(URL_STANDARD_REGEX.is_match("https://example.com/path?query=value"));
1020        assert!(URL_STANDARD_REGEX.is_match("https://example.com/path?a=1&b=2"));
1021    }
1022
1023    #[test]
1024    fn test_url_standard_with_fragment() {
1025        assert!(URL_STANDARD_REGEX.is_match("https://example.com#section"));
1026        assert!(URL_STANDARD_REGEX.is_match("https://example.com/path#section"));
1027        assert!(URL_STANDARD_REGEX.is_match("https://example.com/path?query=value#section"));
1028    }
1029
1030    #[test]
1031    fn test_url_standard_with_port() {
1032        assert!(URL_STANDARD_REGEX.is_match("https://example.com:8080"));
1033        assert!(URL_STANDARD_REGEX.is_match("https://example.com:443/path"));
1034        assert!(URL_STANDARD_REGEX.is_match("http://localhost:3000"));
1035        assert!(URL_STANDARD_REGEX.is_match("https://192.168.1.1:8080/path"));
1036    }
1037
1038    #[test]
1039    fn test_url_standard_wikipedia_style_parentheses() {
1040        // Wikipedia-style URLs with parentheses in path (Issue #240)
1041        let url = "https://en.wikipedia.org/wiki/Rust_(programming_language)";
1042        assert!(URL_STANDARD_REGEX.is_match(url));
1043
1044        // Verify the full URL is captured
1045        let cap = URL_STANDARD_REGEX.find(url).unwrap();
1046        assert_eq!(cap.as_str(), url);
1047
1048        // Multiple parentheses pairs
1049        let url2 = "https://example.com/path_(foo)_(bar)";
1050        let cap2 = URL_STANDARD_REGEX.find(url2).unwrap();
1051        assert_eq!(cap2.as_str(), url2);
1052    }
1053
1054    #[test]
1055    fn test_url_standard_ipv6() {
1056        // IPv6 addresses in URLs
1057        assert!(URL_STANDARD_REGEX.is_match("https://[::1]/path"));
1058        assert!(URL_STANDARD_REGEX.is_match("https://[2001:db8::1]:8080/path"));
1059        assert!(URL_STANDARD_REGEX.is_match("http://[fe80::1%eth0]/"));
1060    }
1061
1062    #[test]
1063    fn test_url_www_basic() {
1064        // www URLs without protocol
1065        assert!(URL_WWW_REGEX.is_match("www.example.com"));
1066        assert!(URL_WWW_REGEX.is_match("www.example.co.uk"));
1067        assert!(URL_WWW_REGEX.is_match("www.sub.example.com"));
1068
1069        // Should not match plain domains without www
1070        assert!(!URL_WWW_REGEX.is_match("example.com"));
1071
1072        // Note: https://www.example.com DOES match because it contains "www."
1073        // The URL_WWW_REGEX is designed to find www. URLs that lack a protocol
1074        // Use URL_STANDARD_REGEX for full URLs with protocols
1075        assert!(URL_WWW_REGEX.is_match("https://www.example.com"));
1076    }
1077
1078    #[test]
1079    fn test_url_www_with_path() {
1080        assert!(URL_WWW_REGEX.is_match("www.example.com/path"));
1081        assert!(URL_WWW_REGEX.is_match("www.example.com/path/to/page"));
1082        assert!(URL_WWW_REGEX.is_match("www.example.com/path_(with_parens)"));
1083    }
1084
1085    #[test]
1086    fn test_url_ipv6_basic() {
1087        // IPv6 specific patterns
1088        assert!(URL_IPV6_REGEX.is_match("https://[::1]/"));
1089        assert!(URL_IPV6_REGEX.is_match("http://[2001:db8::1]/path"));
1090        assert!(URL_IPV6_REGEX.is_match("https://[fe80::1]:8080/path"));
1091        assert!(URL_IPV6_REGEX.is_match("ftp://[::ffff:192.168.1.1]/file"));
1092    }
1093
1094    #[test]
1095    fn test_url_ipv6_with_zone_id() {
1096        // IPv6 with zone identifiers
1097        assert!(URL_IPV6_REGEX.is_match("https://[fe80::1%eth0]/path"));
1098        assert!(URL_IPV6_REGEX.is_match("http://[fe80::1%25eth0]:8080/"));
1099    }
1100
1101    #[test]
1102    fn test_url_simple_detection() {
1103        // Simple pattern for content characteristic detection
1104        assert!(URL_SIMPLE_REGEX.is_match("https://example.com"));
1105        assert!(URL_SIMPLE_REGEX.is_match("http://test.org/path"));
1106        assert!(URL_SIMPLE_REGEX.is_match("ftp://files.com/file.zip"));
1107        assert!(!URL_SIMPLE_REGEX.is_match("not a url"));
1108    }
1109
1110    #[test]
1111    fn test_url_quick_check() {
1112        // Quick check pattern for early exits
1113        assert!(URL_QUICK_CHECK_REGEX.is_match("https://example.com"));
1114        assert!(URL_QUICK_CHECK_REGEX.is_match("http://example.com"));
1115        assert!(URL_QUICK_CHECK_REGEX.is_match("ftp://files.com"));
1116        assert!(URL_QUICK_CHECK_REGEX.is_match("www.example.com"));
1117        assert!(URL_QUICK_CHECK_REGEX.is_match("user@example.com"));
1118        assert!(!URL_QUICK_CHECK_REGEX.is_match("just plain text"));
1119    }
1120
1121    #[test]
1122    fn test_url_edge_cases() {
1123        // URLs with special characters that should be excluded
1124        let url = "https://example.com/path";
1125        assert!(URL_STANDARD_REGEX.is_match(url));
1126
1127        // URL followed by punctuation - the regex captures trailing punctuation
1128        // because trimming is done by `trim_trailing_punctuation()` in the rule
1129        let text = "Check https://example.com, it's great!";
1130        let cap = URL_STANDARD_REGEX.find(text).unwrap();
1131        // The comma IS captured by the regex - rule-level trimming handles this
1132        assert!(cap.as_str().ends_with(','));
1133
1134        // URL in angle brackets should still be found
1135        let text2 = "See <https://example.com> for more";
1136        assert!(URL_STANDARD_REGEX.is_match(text2));
1137
1138        // URL ending at angle bracket should stop at >
1139        let cap2 = URL_STANDARD_REGEX.find(text2).unwrap();
1140        assert!(!cap2.as_str().contains('>'));
1141    }
1142
1143    #[test]
1144    fn test_url_with_complex_paths() {
1145        // Complex real-world URLs
1146        let urls = [
1147            "https://github.com/owner/repo/blob/main/src/file.rs#L123",
1148            "https://docs.example.com/api/v2/endpoint?format=json&page=1",
1149            "https://cdn.example.com/assets/images/logo.png?v=2023",
1150            "https://search.example.com/results?q=test+query&filter=all",
1151        ];
1152
1153        for url in urls {
1154            assert!(URL_STANDARD_REGEX.is_match(url), "Should match: {url}");
1155        }
1156    }
1157
1158    #[test]
1159    fn test_url_pattern_strings_are_valid() {
1160        // Verify patterns compile into valid regexes by accessing them
1161        assert!(URL_STANDARD_REGEX.is_match("https://example.com"));
1162        assert!(URL_WWW_REGEX.is_match("www.example.com"));
1163        assert!(URL_IPV6_REGEX.is_match("https://[::1]/"));
1164        assert!(URL_QUICK_CHECK_REGEX.is_match("https://example.com"));
1165        assert!(URL_SIMPLE_REGEX.is_match("https://example.com"));
1166    }
1167
1168    // =========================================================================
1169    // Tests for is_blank_in_blockquote_context
1170    // This is a shared utility used by MD058, MD065, and other rules that need
1171    // to detect blank lines inside blockquotes (Issue #305)
1172    // =========================================================================
1173
1174    #[test]
1175    fn test_is_blank_in_blockquote_context_regular_blanks() {
1176        // Regular blank lines
1177        assert!(is_blank_in_blockquote_context(""));
1178        assert!(is_blank_in_blockquote_context("   "));
1179        assert!(is_blank_in_blockquote_context("\t"));
1180        assert!(is_blank_in_blockquote_context("  \t  "));
1181    }
1182
1183    #[test]
1184    fn test_is_blank_in_blockquote_context_blockquote_blanks() {
1185        // Blockquote continuation lines with no content (should be treated as blank)
1186        assert!(is_blank_in_blockquote_context(">"));
1187        assert!(is_blank_in_blockquote_context("> "));
1188        assert!(is_blank_in_blockquote_context(">  "));
1189        assert!(is_blank_in_blockquote_context(">>"));
1190        assert!(is_blank_in_blockquote_context(">> "));
1191        assert!(is_blank_in_blockquote_context(">>>"));
1192        assert!(is_blank_in_blockquote_context(">>> "));
1193    }
1194
1195    #[test]
1196    fn test_is_blank_in_blockquote_context_spaced_nested() {
1197        // Spaced nested blockquotes ("> > " style)
1198        assert!(is_blank_in_blockquote_context("> > "));
1199        assert!(is_blank_in_blockquote_context("> > > "));
1200        assert!(is_blank_in_blockquote_context(">  >  "));
1201    }
1202
1203    #[test]
1204    fn test_is_blank_in_blockquote_context_with_leading_space() {
1205        // Blockquote with leading whitespace
1206        assert!(is_blank_in_blockquote_context("  >"));
1207        assert!(is_blank_in_blockquote_context("  > "));
1208        assert!(is_blank_in_blockquote_context("  >>"));
1209    }
1210
1211    #[test]
1212    fn test_is_blank_in_blockquote_context_not_blank() {
1213        // Lines with actual content (should NOT be treated as blank)
1214        assert!(!is_blank_in_blockquote_context("text"));
1215        assert!(!is_blank_in_blockquote_context("> text"));
1216        assert!(!is_blank_in_blockquote_context(">> text"));
1217        assert!(!is_blank_in_blockquote_context("> | table |"));
1218        assert!(!is_blank_in_blockquote_context("| table |"));
1219        assert!(!is_blank_in_blockquote_context("> # Heading"));
1220        assert!(!is_blank_in_blockquote_context(">text")); // No space after > but has text
1221    }
1222
1223    #[test]
1224    fn test_is_blank_in_blockquote_context_edge_cases() {
1225        // Edge cases
1226        assert!(!is_blank_in_blockquote_context(">a")); // Content immediately after >
1227        assert!(!is_blank_in_blockquote_context("> a")); // Single char content
1228        assert!(is_blank_in_blockquote_context(">   ")); // Multiple spaces after >
1229        assert!(!is_blank_in_blockquote_context(">  text")); // Multiple spaces before content
1230    }
1231}
rumdl_lib/utils/regex_cache.rs

rumdl_lib/utils/
regex_cache.rs
