rumdl_lib/
lib.rs

1pub mod config;
2pub mod exit_codes;
3pub mod filtered_lines;
4pub mod fix_coordinator;
5pub mod inline_config;
6pub mod lint_context;
7pub mod markdownlint_config;
8pub mod profiling;
9pub mod rule;
10#[cfg(feature = "native")]
11pub mod vscode;
12pub mod workspace_index;
13#[macro_use]
14pub mod rule_config;
15#[macro_use]
16pub mod rule_config_serde;
17pub mod rules;
18pub mod types;
19pub mod utils;
20
21// Native-only modules (require tokio, tower-lsp, etc.)
22#[cfg(feature = "native")]
23pub mod lsp;
24#[cfg(feature = "native")]
25pub mod output;
26#[cfg(feature = "native")]
27pub mod parallel;
28#[cfg(feature = "native")]
29pub mod performance;
30
31// WASM module
32#[cfg(all(target_arch = "wasm32", feature = "wasm"))]
33pub mod wasm;
34
35pub use rules::heading_utils::{Heading, HeadingStyle};
36pub use rules::*;
37
38pub use crate::lint_context::{LineInfo, LintContext, ListItemInfo};
39use crate::rule::{LintResult, Rule, RuleCategory};
40#[cfg(not(target_arch = "wasm32"))]
41use std::time::Instant;
42
43/// Content characteristics for efficient rule filtering
44#[derive(Debug, Default)]
45struct ContentCharacteristics {
46    has_headings: bool,    // # or setext headings
47    has_lists: bool,       // *, -, +, 1. etc
48    has_links: bool,       // [text](url) or [text][ref]
49    has_code: bool,        // ``` or ~~~ or indented code
50    has_emphasis: bool,    // * or _ for emphasis
51    has_html: bool,        // < > tags
52    has_tables: bool,      // | pipes
53    has_blockquotes: bool, // > markers
54    has_images: bool,      // ![alt](url)
55}
56
57impl ContentCharacteristics {
58    fn analyze(content: &str) -> Self {
59        let mut chars = Self { ..Default::default() };
60
61        // Quick single-pass analysis
62        let mut has_atx_heading = false;
63        let mut has_setext_heading = false;
64
65        for line in content.lines() {
66            let trimmed = line.trim();
67
68            // Headings: ATX (#) or Setext (underlines)
69            if !has_atx_heading && trimmed.starts_with('#') {
70                has_atx_heading = true;
71            }
72            if !has_setext_heading && (trimmed.chars().all(|c| c == '=' || c == '-') && trimmed.len() > 1) {
73                has_setext_heading = true;
74            }
75
76            // Quick character-based detection (more efficient than regex)
77            if !chars.has_lists && (line.contains("* ") || line.contains("- ") || line.contains("+ ")) {
78                chars.has_lists = true;
79            }
80            if !chars.has_lists && line.chars().next().is_some_and(|c| c.is_ascii_digit()) && line.contains(". ") {
81                chars.has_lists = true;
82            }
83            if !chars.has_links
84                && (line.contains('[')
85                    || line.contains("http://")
86                    || line.contains("https://")
87                    || line.contains("ftp://"))
88            {
89                chars.has_links = true;
90            }
91            if !chars.has_images && line.contains("![") {
92                chars.has_images = true;
93            }
94            if !chars.has_code && (line.contains('`') || line.contains("~~~")) {
95                chars.has_code = true;
96            }
97            if !chars.has_emphasis && (line.contains('*') || line.contains('_')) {
98                chars.has_emphasis = true;
99            }
100            if !chars.has_html && line.contains('<') {
101                chars.has_html = true;
102            }
103            if !chars.has_tables && line.contains('|') {
104                chars.has_tables = true;
105            }
106            if !chars.has_blockquotes && line.starts_with('>') {
107                chars.has_blockquotes = true;
108            }
109        }
110
111        chars.has_headings = has_atx_heading || has_setext_heading;
112        chars
113    }
114
115    /// Check if a rule should be skipped based on content characteristics
116    fn should_skip_rule(&self, rule: &dyn Rule) -> bool {
117        match rule.category() {
118            RuleCategory::Heading => !self.has_headings,
119            RuleCategory::List => !self.has_lists,
120            RuleCategory::Link => !self.has_links && !self.has_images,
121            RuleCategory::Image => !self.has_images,
122            RuleCategory::CodeBlock => !self.has_code,
123            RuleCategory::Html => !self.has_html,
124            RuleCategory::Emphasis => !self.has_emphasis,
125            RuleCategory::Blockquote => !self.has_blockquotes,
126            RuleCategory::Table => !self.has_tables,
127            // Always check these categories as they apply to all content
128            RuleCategory::Whitespace | RuleCategory::FrontMatter | RuleCategory::Other => false,
129        }
130    }
131}
132
133/// Compute content hash for incremental indexing change detection
134///
135/// Uses blake3 for native builds (fast, cryptographic-strength hash)
136/// Falls back to std::hash for WASM builds
137#[cfg(feature = "native")]
138fn compute_content_hash(content: &str) -> String {
139    blake3::hash(content.as_bytes()).to_hex().to_string()
140}
141
142/// Compute content hash for WASM builds using std::hash
143#[cfg(not(feature = "native"))]
144fn compute_content_hash(content: &str) -> String {
145    use std::hash::{DefaultHasher, Hash, Hasher};
146    let mut hasher = DefaultHasher::new();
147    content.hash(&mut hasher);
148    format!("{:016x}", hasher.finish())
149}
150
151/// Lint a file against the given rules with intelligent rule filtering
152/// Assumes the provided `rules` vector contains the final,
153/// configured, and filtered set of rules to be executed.
154pub fn lint(
155    content: &str,
156    rules: &[Box<dyn Rule>],
157    verbose: bool,
158    flavor: crate::config::MarkdownFlavor,
159) -> LintResult {
160    // Use lint_and_index but discard the FileIndex for backward compatibility
161    let (result, _file_index) = lint_and_index(content, rules, verbose, flavor, None);
162    result
163}
164
165/// Build FileIndex only (no linting) for cross-file analysis on cache hits
166///
167/// This is a lightweight function that only builds the FileIndex without running
168/// any rules. Used when we have a cache hit but still need the FileIndex for
169/// cross-file validation.
170///
171/// This avoids the overhead of re-running all rules when only the index data is needed.
172pub fn build_file_index_only(
173    content: &str,
174    rules: &[Box<dyn Rule>],
175    flavor: crate::config::MarkdownFlavor,
176) -> crate::workspace_index::FileIndex {
177    // Compute content hash for change detection
178    let content_hash = compute_content_hash(content);
179    let mut file_index = crate::workspace_index::FileIndex::with_hash(content_hash);
180
181    // Early return for empty content
182    if content.is_empty() {
183        return file_index;
184    }
185
186    // Parse LintContext once with the provided flavor
187    let lint_ctx = crate::lint_context::LintContext::new(content, flavor, None);
188
189    // Only call contribute_to_index for cross-file rules (no rule checking!)
190    for rule in rules {
191        if rule.cross_file_scope() == crate::rule::CrossFileScope::Workspace {
192            rule.contribute_to_index(&lint_ctx, &mut file_index);
193        }
194    }
195
196    file_index
197}
198
199/// Lint a file and contribute to workspace index for cross-file analysis
200///
201/// This variant performs linting and optionally populates a `FileIndex` with data
202/// needed for cross-file validation. The FileIndex is populated during linting,
203/// avoiding duplicate parsing.
204///
205/// Returns: (warnings, FileIndex) - the FileIndex contains headings/links for cross-file rules
206pub fn lint_and_index(
207    content: &str,
208    rules: &[Box<dyn Rule>],
209    _verbose: bool,
210    flavor: crate::config::MarkdownFlavor,
211    source_file: Option<std::path::PathBuf>,
212) -> (LintResult, crate::workspace_index::FileIndex) {
213    let mut warnings = Vec::new();
214    // Compute content hash for change detection
215    let content_hash = compute_content_hash(content);
216    let mut file_index = crate::workspace_index::FileIndex::with_hash(content_hash);
217
218    #[cfg(not(target_arch = "wasm32"))]
219    let _overall_start = Instant::now();
220
221    // Early return for empty content
222    if content.is_empty() {
223        return (Ok(warnings), file_index);
224    }
225
226    // Parse inline configuration comments once
227    let inline_config = crate::inline_config::InlineConfig::from_content(content);
228
229    // Export inline config data to FileIndex for cross-file rule filtering
230    let (file_disabled, line_disabled) = inline_config.export_for_file_index();
231    file_index.file_disabled_rules = file_disabled;
232    file_index.line_disabled_rules = line_disabled;
233
234    // Analyze content characteristics for rule filtering
235    let characteristics = ContentCharacteristics::analyze(content);
236
237    // Filter rules based on content characteristics
238    let applicable_rules: Vec<_> = rules
239        .iter()
240        .filter(|rule| !characteristics.should_skip_rule(rule.as_ref()))
241        .collect();
242
243    // Calculate skipped rules count before consuming applicable_rules
244    let _total_rules = rules.len();
245    let _applicable_count = applicable_rules.len();
246
247    // Parse LintContext once with the provided flavor
248    let lint_ctx = crate::lint_context::LintContext::new(content, flavor, source_file);
249
250    #[cfg(not(target_arch = "wasm32"))]
251    let profile_rules = std::env::var("RUMDL_PROFILE_RULES").is_ok();
252    #[cfg(target_arch = "wasm32")]
253    let profile_rules = false;
254
255    for rule in &applicable_rules {
256        #[cfg(not(target_arch = "wasm32"))]
257        let _rule_start = Instant::now();
258
259        // Run single-file check
260        let result = rule.check(&lint_ctx);
261
262        match result {
263            Ok(rule_warnings) => {
264                // Filter out warnings for rules disabled via inline comments
265                let filtered_warnings: Vec<_> = rule_warnings
266                    .into_iter()
267                    .filter(|warning| {
268                        // Use the warning's rule_name if available, otherwise use the rule's name
269                        let rule_name_to_check = warning.rule_name.as_deref().unwrap_or(rule.name());
270
271                        // Extract the base rule name for sub-rules like "MD029-style" -> "MD029"
272                        let base_rule_name = if let Some(dash_pos) = rule_name_to_check.find('-') {
273                            &rule_name_to_check[..dash_pos]
274                        } else {
275                            rule_name_to_check
276                        };
277
278                        !inline_config.is_rule_disabled(
279                            base_rule_name,
280                            warning.line, // Already 1-indexed
281                        )
282                    })
283                    .collect();
284                warnings.extend(filtered_warnings);
285            }
286            Err(e) => {
287                log::error!("Error checking rule {}: {}", rule.name(), e);
288                return (Err(e), file_index);
289            }
290        }
291
292        #[cfg(not(target_arch = "wasm32"))]
293        {
294            let rule_duration = _rule_start.elapsed();
295            if profile_rules {
296                eprintln!("[RULE] {:6} {:?}", rule.name(), rule_duration);
297            }
298
299            #[cfg(not(test))]
300            if _verbose && rule_duration.as_millis() > 500 {
301                log::debug!("Rule {} took {:?}", rule.name(), rule_duration);
302            }
303        }
304    }
305
306    // Contribute to index for cross-file rules (done after all rules checked)
307    // NOTE: We iterate over ALL rules (not just applicable_rules) because cross-file
308    // rules need to extract data from every file in the workspace, regardless of whether
309    // that file has content that would trigger the rule. For example, MD051 needs to
310    // index headings from files that have no links (like target.md) so that links
311    // FROM other files TO those headings can be validated.
312    for rule in rules {
313        if rule.cross_file_scope() == crate::rule::CrossFileScope::Workspace {
314            rule.contribute_to_index(&lint_ctx, &mut file_index);
315        }
316    }
317
318    #[cfg(not(test))]
319    if _verbose {
320        let skipped_rules = _total_rules - _applicable_count;
321        if skipped_rules > 0 {
322            log::debug!("Skipped {skipped_rules} of {_total_rules} rules based on content analysis");
323        }
324    }
325
326    (Ok(warnings), file_index)
327}
328
329/// Run cross-file checks for rules that need workspace-wide validation
330///
331/// This should be called after all files have been linted and the WorkspaceIndex
332/// has been built from the accumulated FileIndex data.
333///
334/// Note: This takes the FileIndex instead of content to avoid re-parsing each file.
335/// The FileIndex was already populated during contribute_to_index in the linting phase.
336///
337/// Rules can use workspace_index methods for cross-file validation:
338/// - `get_file(path)` - to look up headings in target files (for MD051)
339///
340/// Returns additional warnings from cross-file validation.
341pub fn run_cross_file_checks(
342    file_path: &std::path::Path,
343    file_index: &crate::workspace_index::FileIndex,
344    rules: &[Box<dyn Rule>],
345    workspace_index: &crate::workspace_index::WorkspaceIndex,
346) -> LintResult {
347    use crate::rule::CrossFileScope;
348
349    let mut warnings = Vec::new();
350
351    // Only check rules that need cross-file analysis
352    for rule in rules {
353        if rule.cross_file_scope() != CrossFileScope::Workspace {
354            continue;
355        }
356
357        match rule.cross_file_check(file_path, file_index, workspace_index) {
358            Ok(rule_warnings) => {
359                // Filter cross-file warnings based on inline config stored in file_index
360                let filtered: Vec<_> = rule_warnings
361                    .into_iter()
362                    .filter(|w| !file_index.is_rule_disabled_at_line(rule.name(), w.line))
363                    .collect();
364                warnings.extend(filtered);
365            }
366            Err(e) => {
367                log::error!("Error in cross-file check for rule {}: {}", rule.name(), e);
368                return Err(e);
369            }
370        }
371    }
372
373    Ok(warnings)
374}
375
376/// Get the profiling report
377pub fn get_profiling_report() -> String {
378    profiling::get_report()
379}
380
381/// Reset the profiling data
382pub fn reset_profiling() {
383    profiling::reset()
384}
385
386/// Get regex cache statistics for performance monitoring
387pub fn get_regex_cache_stats() -> std::collections::HashMap<String, u64> {
388    crate::utils::regex_cache::get_cache_stats()
389}
390
391#[cfg(test)]
392mod tests {
393    use super::*;
394    use crate::rule::Rule;
395    use crate::rules::{MD001HeadingIncrement, MD009TrailingSpaces};
396
397    #[test]
398    fn test_content_characteristics_analyze() {
399        // Test empty content
400        let chars = ContentCharacteristics::analyze("");
401        assert!(!chars.has_headings);
402        assert!(!chars.has_lists);
403        assert!(!chars.has_links);
404        assert!(!chars.has_code);
405        assert!(!chars.has_emphasis);
406        assert!(!chars.has_html);
407        assert!(!chars.has_tables);
408        assert!(!chars.has_blockquotes);
409        assert!(!chars.has_images);
410
411        // Test content with headings
412        let chars = ContentCharacteristics::analyze("# Heading");
413        assert!(chars.has_headings);
414
415        // Test setext headings
416        let chars = ContentCharacteristics::analyze("Heading\n=======");
417        assert!(chars.has_headings);
418
419        // Test lists
420        let chars = ContentCharacteristics::analyze("* Item\n- Item 2\n+ Item 3");
421        assert!(chars.has_lists);
422
423        // Test ordered lists
424        let chars = ContentCharacteristics::analyze("1. First\n2. Second");
425        assert!(chars.has_lists);
426
427        // Test links
428        let chars = ContentCharacteristics::analyze("[link](url)");
429        assert!(chars.has_links);
430
431        // Test URLs
432        let chars = ContentCharacteristics::analyze("Visit https://example.com");
433        assert!(chars.has_links);
434
435        // Test images
436        let chars = ContentCharacteristics::analyze("![alt text](image.png)");
437        assert!(chars.has_images);
438
439        // Test code
440        let chars = ContentCharacteristics::analyze("`inline code`");
441        assert!(chars.has_code);
442
443        let chars = ContentCharacteristics::analyze("~~~\ncode block\n~~~");
444        assert!(chars.has_code);
445
446        // Test emphasis
447        let chars = ContentCharacteristics::analyze("*emphasis* and _more_");
448        assert!(chars.has_emphasis);
449
450        // Test HTML
451        let chars = ContentCharacteristics::analyze("<div>HTML content</div>");
452        assert!(chars.has_html);
453
454        // Test tables
455        let chars = ContentCharacteristics::analyze("| Header | Header |\n|--------|--------|");
456        assert!(chars.has_tables);
457
458        // Test blockquotes
459        let chars = ContentCharacteristics::analyze("> Quote");
460        assert!(chars.has_blockquotes);
461
462        // Test mixed content
463        let content = "# Heading\n* List item\n[link](url)\n`code`\n*emphasis*\n<p>html</p>\n| table |\n> quote\n![image](img.png)";
464        let chars = ContentCharacteristics::analyze(content);
465        assert!(chars.has_headings);
466        assert!(chars.has_lists);
467        assert!(chars.has_links);
468        assert!(chars.has_code);
469        assert!(chars.has_emphasis);
470        assert!(chars.has_html);
471        assert!(chars.has_tables);
472        assert!(chars.has_blockquotes);
473        assert!(chars.has_images);
474    }
475
476    #[test]
477    fn test_content_characteristics_should_skip_rule() {
478        let chars = ContentCharacteristics {
479            has_headings: true,
480            has_lists: false,
481            has_links: true,
482            has_code: false,
483            has_emphasis: true,
484            has_html: false,
485            has_tables: true,
486            has_blockquotes: false,
487            has_images: false,
488        };
489
490        // Create test rules for different categories
491        let heading_rule = MD001HeadingIncrement;
492        assert!(!chars.should_skip_rule(&heading_rule));
493
494        let trailing_spaces_rule = MD009TrailingSpaces::new(2, false);
495        assert!(!chars.should_skip_rule(&trailing_spaces_rule)); // Whitespace rules always run
496
497        // Test skipping based on content
498        let chars_no_headings = ContentCharacteristics {
499            has_headings: false,
500            ..Default::default()
501        };
502        assert!(chars_no_headings.should_skip_rule(&heading_rule));
503    }
504
505    #[test]
506    fn test_lint_empty_content() {
507        let rules: Vec<Box<dyn Rule>> = vec![Box::new(MD001HeadingIncrement)];
508
509        let result = lint("", &rules, false, crate::config::MarkdownFlavor::Standard);
510        assert!(result.is_ok());
511        assert!(result.unwrap().is_empty());
512    }
513
514    #[test]
515    fn test_lint_with_violations() {
516        let content = "## Level 2\n#### Level 4"; // Skips level 3
517        let rules: Vec<Box<dyn Rule>> = vec![Box::new(MD001HeadingIncrement)];
518
519        let result = lint(content, &rules, false, crate::config::MarkdownFlavor::Standard);
520        assert!(result.is_ok());
521        let warnings = result.unwrap();
522        assert!(!warnings.is_empty());
523        // Check the rule field of LintWarning struct
524        assert_eq!(warnings[0].rule_name.as_deref(), Some("MD001"));
525    }
526
527    #[test]
528    fn test_lint_with_inline_disable() {
529        let content = "<!-- rumdl-disable MD001 -->\n## Level 2\n#### Level 4";
530        let rules: Vec<Box<dyn Rule>> = vec![Box::new(MD001HeadingIncrement)];
531
532        let result = lint(content, &rules, false, crate::config::MarkdownFlavor::Standard);
533        assert!(result.is_ok());
534        let warnings = result.unwrap();
535        assert!(warnings.is_empty()); // Should be disabled by inline comment
536    }
537
538    #[test]
539    fn test_lint_rule_filtering() {
540        // Content with no lists
541        let content = "# Heading\nJust text";
542        let rules: Vec<Box<dyn Rule>> = vec![
543            Box::new(MD001HeadingIncrement),
544            // A list-related rule would be skipped
545        ];
546
547        let result = lint(content, &rules, false, crate::config::MarkdownFlavor::Standard);
548        assert!(result.is_ok());
549    }
550
551    #[test]
552    fn test_get_profiling_report() {
553        // Just test that it returns a string without panicking
554        let report = get_profiling_report();
555        assert!(!report.is_empty());
556        assert!(report.contains("Profiling"));
557    }
558
559    #[test]
560    fn test_reset_profiling() {
561        // Test that reset_profiling doesn't panic
562        reset_profiling();
563
564        // After reset, report should indicate no measurements or profiling disabled
565        let report = get_profiling_report();
566        assert!(report.contains("disabled") || report.contains("no measurements"));
567    }
568
569    #[test]
570    fn test_get_regex_cache_stats() {
571        let stats = get_regex_cache_stats();
572        // Stats should be a valid HashMap (might be empty)
573        assert!(stats.is_empty() || !stats.is_empty());
574
575        // If not empty, all values should be positive
576        for count in stats.values() {
577            assert!(*count > 0);
578        }
579    }
580
581    #[test]
582    fn test_content_characteristics_edge_cases() {
583        // Test setext heading edge case
584        let chars = ContentCharacteristics::analyze("-"); // Single dash, not a heading
585        assert!(!chars.has_headings);
586
587        let chars = ContentCharacteristics::analyze("--"); // Two dashes, valid setext
588        assert!(chars.has_headings);
589
590        // Test list detection edge cases
591        let chars = ContentCharacteristics::analyze("*emphasis*"); // Not a list
592        assert!(!chars.has_lists);
593
594        let chars = ContentCharacteristics::analyze("1.Item"); // No space after period
595        assert!(!chars.has_lists);
596
597        // Test blockquote must be at start of line
598        let chars = ContentCharacteristics::analyze("text > not a quote");
599        assert!(!chars.has_blockquotes);
600    }
601}
rumdl_lib/lib.rs

rumdl_lib/
lib.rs
