rumdl_lib/rules/
md026_no_trailing_punctuation.rs

1/// Rule MD026: No trailing punctuation in headings
2///
3/// See [docs/md026.md](../../docs/md026.md) for full documentation, configuration, and examples.
4use crate::rule::{Fix, LintError, LintResult, LintWarning, Rule, Severity};
5use crate::utils::range_utils::calculate_match_range;
6use lazy_static::lazy_static;
7use regex::Regex;
8use std::collections::HashMap;
9use std::ops::Range;
10use std::sync::RwLock;
11
12mod md026_config;
13use md026_config::{DEFAULT_PUNCTUATION, MD026Config};
14
15lazy_static! {
16    // Optimized single regex for all ATX heading types (normal, closed, indented 1-3 spaces)
17    static ref ATX_HEADING_UNIFIED: Regex = Regex::new(r"^( {0,3})(#{1,6})(\s+)(.+?)(\s+#{1,6})?$").unwrap();
18
19    // Fast check patterns for early returns - match defaults
20    static ref QUICK_PUNCTUATION_CHECK: Regex = Regex::new(&format!(r"[{}]", regex::escape(DEFAULT_PUNCTUATION))).unwrap();
21
22    // Regex cache for punctuation patterns
23    static ref PUNCTUATION_REGEX_CACHE: RwLock<HashMap<String, Regex>> = RwLock::new(HashMap::new());
24}
25
26/// Rule MD026: Trailing punctuation in heading
27#[derive(Clone, Default)]
28pub struct MD026NoTrailingPunctuation {
29    config: MD026Config,
30}
31
32impl MD026NoTrailingPunctuation {
33    pub fn new(punctuation: Option<String>) -> Self {
34        Self {
35            config: MD026Config {
36                punctuation: punctuation.unwrap_or_else(|| DEFAULT_PUNCTUATION.to_string()),
37            },
38        }
39    }
40
41    pub fn from_config_struct(config: MD026Config) -> Self {
42        Self { config }
43    }
44
45    #[inline]
46    fn get_punctuation_regex(&self) -> Result<Regex, regex::Error> {
47        // Check cache first
48        {
49            let cache = PUNCTUATION_REGEX_CACHE.read().unwrap();
50            if let Some(cached_regex) = cache.get(&self.config.punctuation) {
51                return Ok(cached_regex.clone());
52            }
53        }
54
55        // Compile and cache the regex
56        let pattern = format!(r"([{}]+)$", regex::escape(&self.config.punctuation));
57        let regex = Regex::new(&pattern)?;
58
59        {
60            let mut cache = PUNCTUATION_REGEX_CACHE.write().unwrap();
61            cache.insert(self.config.punctuation.clone(), regex.clone());
62        }
63
64        Ok(regex)
65    }
66
67    #[inline]
68    fn has_trailing_punctuation(&self, text: &str, re: &Regex) -> bool {
69        let trimmed = text.trim();
70        re.is_match(trimmed)
71    }
72
73    #[inline]
74    fn get_line_byte_range(&self, content: &str, line_num: usize) -> Range<usize> {
75        let mut start_pos = 0;
76
77        for (idx, line) in content.lines().enumerate() {
78            if idx + 1 == line_num {
79                return Range {
80                    start: start_pos,
81                    end: start_pos + line.len(),
82                };
83            }
84            // +1 for the newline character
85            start_pos += line.len() + 1;
86        }
87
88        Range {
89            start: content.len(),
90            end: content.len(),
91        }
92    }
93
94    // Remove trailing punctuation from text
95    #[inline]
96    fn remove_trailing_punctuation(&self, text: &str, re: &Regex) -> String {
97        re.replace_all(text.trim(), "").to_string()
98    }
99
100    // Optimized ATX heading fix using unified regex
101    #[inline]
102    fn fix_atx_heading(&self, line: &str, re: &Regex) -> String {
103        if let Some(captures) = ATX_HEADING_UNIFIED.captures(line) {
104            let indentation = captures.get(1).unwrap().as_str();
105            let hashes = captures.get(2).unwrap().as_str();
106            let space = captures.get(3).unwrap().as_str();
107            let content = captures.get(4).unwrap().as_str();
108
109            // Check if content ends with a custom header ID like {#my-id}
110            // If so, we need to fix punctuation before the ID
111            let fixed_content = if let Some(id_pos) = content.rfind(" {#") {
112                // Has a custom ID - fix punctuation before it
113                let before_id = &content[..id_pos];
114                let id_part = &content[id_pos..];
115                let fixed_before = self.remove_trailing_punctuation(before_id, re);
116                format!("{fixed_before}{id_part}")
117            } else {
118                // No custom ID - just remove trailing punctuation
119                self.remove_trailing_punctuation(content, re)
120            };
121
122            // Preserve any trailing hashes if present
123            if let Some(trailing) = captures.get(5) {
124                return format!(
125                    "{}{}{}{}{}",
126                    indentation,
127                    hashes,
128                    space,
129                    fixed_content,
130                    trailing.as_str()
131                );
132            }
133
134            return format!("{indentation}{hashes}{space}{fixed_content}");
135        }
136
137        // Fallback if no regex matches
138        line.to_string()
139    }
140
141    // Fix a setext heading by removing trailing punctuation from the content line
142    #[inline]
143    fn fix_setext_heading(&self, content_line: &str, re: &Regex) -> String {
144        let trimmed = content_line.trim_end();
145        let mut whitespace = "";
146
147        // Preserve trailing whitespace
148        if content_line.len() > trimmed.len() {
149            whitespace = &content_line[trimmed.len()..];
150        }
151
152        // Remove punctuation and preserve whitespace
153        format!("{}{}", self.remove_trailing_punctuation(trimmed, re), whitespace)
154    }
155}
156
157impl Rule for MD026NoTrailingPunctuation {
158    fn name(&self) -> &'static str {
159        "MD026"
160    }
161
162    fn description(&self) -> &'static str {
163        "Trailing punctuation in heading"
164    }
165
166    fn should_skip(&self, ctx: &crate::lint_context::LintContext) -> bool {
167        // Skip if no heading markers
168        if !ctx.content.contains('#') {
169            return true;
170        }
171        // Skip if none of the configured punctuation exists
172        let punctuation = &self.config.punctuation;
173        !punctuation.chars().any(|p| ctx.content.contains(p))
174    }
175
176    fn as_maybe_document_structure(&self) -> Option<&dyn crate::rule::MaybeDocumentStructure> {
177        None
178    }
179
180    fn check(&self, ctx: &crate::lint_context::LintContext) -> LintResult {
181        let content = ctx.content;
182
183        // Early returns for performance
184        if content.is_empty() {
185            return Ok(Vec::new());
186        }
187
188        // Quick check for any punctuation we care about
189        // For custom punctuation, we need to check differently
190        if self.config.punctuation == DEFAULT_PUNCTUATION {
191            if !QUICK_PUNCTUATION_CHECK.is_match(content) {
192                return Ok(Vec::new());
193            }
194        } else {
195            // For custom punctuation, check if any of those characters exist
196            let has_custom_punctuation = self.config.punctuation.chars().any(|c| content.contains(c));
197            if !has_custom_punctuation {
198                return Ok(Vec::new());
199            }
200        }
201
202        // Check if we have any headings from pre-computed line info
203        let has_headings = ctx.lines.iter().any(|line| line.heading.is_some());
204        if !has_headings {
205            return Ok(Vec::new());
206        }
207
208        let mut warnings = Vec::new();
209        let re = match self.get_punctuation_regex() {
210            Ok(regex) => regex,
211            Err(_) => return Ok(warnings),
212        };
213
214        // Use pre-computed heading information from LintContext
215        for (line_num, line_info) in ctx.lines.iter().enumerate() {
216            if let Some(heading) = &line_info.heading {
217                // Skip deeply indented headings (they're code blocks)
218                if line_info.indent >= 4 && matches!(heading.style, crate::lint_context::HeadingStyle::ATX) {
219                    continue;
220                }
221
222                // LintContext already strips Kramdown IDs from heading.text
223                // So we just check the heading text directly for trailing punctuation
224                // This correctly flags "# Heading." even if it has {#id}
225                let text_to_check = heading.text.clone();
226
227                if self.has_trailing_punctuation(&text_to_check, &re) {
228                    // Find the trailing punctuation
229                    if let Some(punctuation_match) = re.find(&text_to_check) {
230                        let line = &line_info.content;
231
232                        // For ATX headings, find the punctuation position in the line
233                        let punctuation_pos_in_text = punctuation_match.start();
234                        let text_pos_in_line = line.find(&heading.text).unwrap_or(heading.content_column);
235                        let punctuation_start_in_line = text_pos_in_line + punctuation_pos_in_text;
236                        let punctuation_len = punctuation_match.len();
237
238                        let (start_line, start_col, end_line, end_col) = calculate_match_range(
239                            line_num + 1, // Convert to 1-indexed
240                            line,
241                            punctuation_start_in_line,
242                            punctuation_len,
243                        );
244
245                        let last_char = text_to_check.chars().last().unwrap_or(' ');
246                        warnings.push(LintWarning {
247                            rule_name: Some(self.name()),
248                            line: start_line,
249                            column: start_col,
250                            end_line,
251                            end_column: end_col,
252                            message: format!("Heading '{text_to_check}' ends with punctuation '{last_char}'"),
253                            severity: Severity::Warning,
254                            fix: Some(Fix {
255                                range: self.get_line_byte_range(content, line_num + 1),
256                                replacement: if matches!(heading.style, crate::lint_context::HeadingStyle::ATX) {
257                                    self.fix_atx_heading(line, &re)
258                                } else {
259                                    self.fix_setext_heading(line, &re)
260                                },
261                            }),
262                        });
263                    }
264                }
265            }
266        }
267
268        Ok(warnings)
269    }
270
271    fn fix(&self, ctx: &crate::lint_context::LintContext) -> Result<String, LintError> {
272        let content = ctx.content;
273
274        // Fast path optimizations
275        if content.is_empty() {
276            return Ok(content.to_string());
277        }
278
279        // Quick check for punctuation
280        // For custom punctuation, we need to check differently
281        if self.config.punctuation == DEFAULT_PUNCTUATION {
282            if !QUICK_PUNCTUATION_CHECK.is_match(content) {
283                return Ok(content.to_string());
284            }
285        } else {
286            // For custom punctuation, check if any of those characters exist
287            let has_custom_punctuation = self.config.punctuation.chars().any(|c| content.contains(c));
288            if !has_custom_punctuation {
289                return Ok(content.to_string());
290            }
291        }
292
293        // Check if we have any headings from pre-computed line info
294        let has_headings = ctx.lines.iter().any(|line| line.heading.is_some());
295        if !has_headings {
296            return Ok(content.to_string());
297        }
298
299        let re = match self.get_punctuation_regex() {
300            Ok(regex) => regex,
301            Err(_) => return Ok(content.to_string()),
302        };
303
304        let lines: Vec<&str> = content.lines().collect();
305        let mut fixed_lines: Vec<String> = lines.iter().map(|&s| s.to_string()).collect();
306
307        // Use pre-computed heading information from LintContext
308        for (line_num, line_info) in ctx.lines.iter().enumerate() {
309            if let Some(heading) = &line_info.heading {
310                // Skip deeply indented headings (they're code blocks)
311                if line_info.indent >= 4 && matches!(heading.style, crate::lint_context::HeadingStyle::ATX) {
312                    continue;
313                }
314
315                // LintContext already strips custom header IDs from heading.text
316                // So we just check the heading text directly for trailing punctuation
317                let text_to_check = heading.text.clone();
318
319                // Check and fix trailing punctuation
320                if self.has_trailing_punctuation(&text_to_check, &re) {
321                    fixed_lines[line_num] = if matches!(heading.style, crate::lint_context::HeadingStyle::ATX) {
322                        self.fix_atx_heading(&line_info.content, &re)
323                    } else {
324                        self.fix_setext_heading(&line_info.content, &re)
325                    };
326                }
327            }
328        }
329
330        // Reconstruct content preserving line endings
331        let mut result = String::with_capacity(content.len());
332        for (i, line) in fixed_lines.iter().enumerate() {
333            result.push_str(line);
334            if i < fixed_lines.len() - 1 || content.ends_with('\n') {
335                result.push('\n');
336            }
337        }
338
339        Ok(result)
340    }
341
342    fn as_any(&self) -> &dyn std::any::Any {
343        self
344    }
345
346    fn default_config_section(&self) -> Option<(String, toml::Value)> {
347        let json_value = serde_json::to_value(&self.config).ok()?;
348        Some((
349            self.name().to_string(),
350            crate::rule_config_serde::json_to_toml_value(&json_value)?,
351        ))
352    }
353
354    fn from_config(config: &crate::config::Config) -> Box<dyn Rule>
355    where
356        Self: Sized,
357    {
358        let rule_config = crate::rule_config_serde::load_rule_config::<MD026Config>(config);
359        Box::new(Self::from_config_struct(rule_config))
360    }
361}
362
363#[cfg(test)]
364mod tests {
365    use super::*;
366    use crate::lint_context::LintContext;
367
368    #[test]
369    fn test_no_trailing_punctuation() {
370        let rule = MD026NoTrailingPunctuation::new(None);
371        let content = "# This is a heading\n\n## Another heading";
372        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
373        let result = rule.check(&ctx).unwrap();
374        assert!(result.is_empty(), "Headings without punctuation should not be flagged");
375    }
376
377    #[test]
378    fn test_trailing_period() {
379        let rule = MD026NoTrailingPunctuation::new(None);
380        let content = "# This is a heading.\n\n## Another one.";
381        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
382        let result = rule.check(&ctx).unwrap();
383        assert_eq!(result.len(), 2);
384        assert_eq!(result[0].line, 1);
385        assert_eq!(result[0].column, 20);
386        assert!(result[0].message.contains("ends with punctuation '.'"));
387        assert_eq!(result[1].line, 3);
388        assert_eq!(result[1].column, 15);
389    }
390
391    #[test]
392    fn test_trailing_comma() {
393        let rule = MD026NoTrailingPunctuation::new(None);
394        let content = "# Heading,\n## Sub-heading,";
395        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
396        let result = rule.check(&ctx).unwrap();
397        assert_eq!(result.len(), 2);
398        assert!(result[0].message.contains("ends with punctuation ','"));
399    }
400
401    #[test]
402    fn test_trailing_semicolon() {
403        let rule = MD026NoTrailingPunctuation::new(None);
404        let content = "# Title;\n## Subtitle;";
405        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
406        let result = rule.check(&ctx).unwrap();
407        assert_eq!(result.len(), 2);
408        assert!(result[0].message.contains("ends with punctuation ';'"));
409    }
410
411    #[test]
412    fn test_custom_punctuation() {
413        let rule = MD026NoTrailingPunctuation::new(Some("!".to_string()));
414        let content = "# Important!\n## Regular heading.";
415        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
416        let result = rule.check(&ctx).unwrap();
417        assert_eq!(result.len(), 1, "Only exclamation should be flagged with custom config");
418        assert_eq!(result[0].line, 1);
419        assert!(result[0].message.contains("ends with punctuation '!'"));
420    }
421
422    #[test]
423    fn test_legitimate_question_mark() {
424        let rule = MD026NoTrailingPunctuation::new(Some(".,;?".to_string()));
425        let content = "# What is this?\n# This is bad.";
426        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
427        let result = rule.check(&ctx).unwrap();
428        // With custom punctuation, legitimate punctuation exceptions don't apply
429        assert_eq!(result.len(), 2, "Both should be flagged with custom punctuation");
430    }
431
432    #[test]
433    fn test_question_marks_not_in_default() {
434        let rule = MD026NoTrailingPunctuation::new(None);
435        let content = "# What is Rust?\n# How does it work?\n# Is it fast?";
436        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
437        let result = rule.check(&ctx).unwrap();
438        assert!(result.is_empty(), "Question marks are not in default punctuation list");
439    }
440
441    #[test]
442    fn test_colons_in_default() {
443        let rule = MD026NoTrailingPunctuation::new(None);
444        let content = "# FAQ:\n# API Reference:\n# Step 1:\n# Version 2.0:";
445        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
446        let result = rule.check(&ctx).unwrap();
447        assert_eq!(
448            result.len(),
449            4,
450            "Colons are in default punctuation list and should be flagged"
451        );
452    }
453
454    #[test]
455    fn test_fix_atx_headings() {
456        let rule = MD026NoTrailingPunctuation::new(None);
457        let content = "# Title.\n## Subtitle,\n### Sub-subtitle;";
458        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
459        let fixed = rule.fix(&ctx).unwrap();
460        assert_eq!(fixed, "# Title\n## Subtitle\n### Sub-subtitle");
461    }
462
463    #[test]
464    fn test_fix_setext_headings() {
465        let rule = MD026NoTrailingPunctuation::new(None);
466        let content = "Title.\n======\n\nSubtitle,\n---------";
467        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
468        let fixed = rule.fix(&ctx).unwrap();
469        assert_eq!(fixed, "Title\n======\n\nSubtitle\n---------");
470    }
471
472    #[test]
473    fn test_fix_preserves_trailing_hashes() {
474        let rule = MD026NoTrailingPunctuation::new(None);
475        let content = "# Title. #\n## Subtitle, ##";
476        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
477        let fixed = rule.fix(&ctx).unwrap();
478        assert_eq!(fixed, "# Title #\n## Subtitle ##");
479    }
480
481    #[test]
482    fn test_indented_headings() {
483        let rule = MD026NoTrailingPunctuation::new(None);
484        let content = "   # Title.\n  ## Subtitle.";
485        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
486        let result = rule.check(&ctx).unwrap();
487        assert_eq!(result.len(), 2, "Indented headings (< 4 spaces) should be checked");
488    }
489
490    #[test]
491    fn test_deeply_indented_ignored() {
492        let rule = MD026NoTrailingPunctuation::new(None);
493        let content = "    # This is code.";
494        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
495        let result = rule.check(&ctx).unwrap();
496        assert!(result.is_empty(), "Deeply indented lines (4+ spaces) should be ignored");
497    }
498
499    #[test]
500    fn test_multiple_punctuation() {
501        let rule = MD026NoTrailingPunctuation::new(None);
502        let content = "# Title...";
503        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
504        let result = rule.check(&ctx).unwrap();
505        assert_eq!(result.len(), 1);
506        assert_eq!(result[0].column, 8); // Points to first period
507    }
508
509    #[test]
510    fn test_empty_content() {
511        let rule = MD026NoTrailingPunctuation::new(None);
512        let content = "";
513        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
514        let result = rule.check(&ctx).unwrap();
515        assert!(result.is_empty());
516    }
517
518    #[test]
519    fn test_no_headings() {
520        let rule = MD026NoTrailingPunctuation::new(None);
521        let content = "This is just text.\nMore text with punctuation.";
522        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
523        let result = rule.check(&ctx).unwrap();
524        assert!(result.is_empty(), "Non-heading lines should not be checked");
525    }
526
527    #[test]
528    fn test_get_punctuation_regex() {
529        let rule = MD026NoTrailingPunctuation::new(Some("!?".to_string()));
530        let regex = rule.get_punctuation_regex().unwrap();
531        assert!(regex.is_match("text!"));
532        assert!(regex.is_match("text?"));
533        assert!(!regex.is_match("text."));
534    }
535
536    #[test]
537    fn test_regex_caching() {
538        let rule1 = MD026NoTrailingPunctuation::new(Some("!".to_string()));
539        let rule2 = MD026NoTrailingPunctuation::new(Some("!".to_string()));
540
541        // Both should get the same cached regex
542        let _regex1 = rule1.get_punctuation_regex().unwrap();
543        let _regex2 = rule2.get_punctuation_regex().unwrap();
544
545        // Check cache has the entry
546        let cache = PUNCTUATION_REGEX_CACHE.read().unwrap();
547        assert!(cache.contains_key("!"));
548    }
549
550    #[test]
551    fn test_config_from_toml() {
552        let mut config = crate::config::Config::default();
553        let mut rule_config = crate::config::RuleConfig::default();
554        rule_config
555            .values
556            .insert("punctuation".to_string(), toml::Value::String("!?".to_string()));
557        config.rules.insert("MD026".to_string(), rule_config);
558
559        let rule = MD026NoTrailingPunctuation::from_config(&config);
560        let content = "# Title!\n# Another?";
561        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
562        let result = rule.check(&ctx).unwrap();
563        assert_eq!(result.len(), 2, "Custom punctuation from config should be used");
564    }
565
566    #[test]
567    fn test_fix_removes_punctuation() {
568        let rule = MD026NoTrailingPunctuation::new(None);
569        let content = "# Title.   \n## Subtitle,  ";
570        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
571        let fixed = rule.fix(&ctx).unwrap();
572        // The current implementation doesn't preserve trailing whitespace after punctuation removal
573        assert_eq!(fixed, "# Title\n## Subtitle");
574    }
575
576    #[test]
577    fn test_final_newline_preservation() {
578        let rule = MD026NoTrailingPunctuation::new(None);
579        let content = "# Title.\n";
580        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
581        let fixed = rule.fix(&ctx).unwrap();
582        assert_eq!(fixed, "# Title\n");
583
584        let content_no_newline = "# Title.";
585        let ctx2 = LintContext::new(content_no_newline, crate::config::MarkdownFlavor::Standard);
586        let fixed2 = rule.fix(&ctx2).unwrap();
587        assert_eq!(fixed2, "# Title");
588    }
589}
rumdl_lib/rules/md026_no_trailing_punctuation.rs

rumdl_lib/rules/
md026_no_trailing_punctuation.rs
