rumdl_lib/rules/
md037_spaces_around_emphasis.rs

1/// Rule MD037: No spaces around emphasis markers
2///
3/// See [docs/md037.md](../../docs/md037.md) for full documentation, configuration, and examples.
4use crate::rule::{Fix, LintError, LintResult, LintWarning, Rule, RuleCategory, Severity};
5use crate::utils::emphasis_utils::{
6    EmphasisSpan, find_emphasis_markers, find_emphasis_spans, has_doc_patterns, replace_inline_code,
7};
8use crate::utils::kramdown_utils::has_span_ial;
9use crate::utils::range_utils::LineIndex;
10use crate::utils::regex_cache::UNORDERED_LIST_MARKER_REGEX;
11use crate::utils::skip_context::{is_in_html_comment, is_in_math_context, is_in_table_cell};
12use lazy_static::lazy_static;
13use regex::Regex;
14
15lazy_static! {
16    // Reference definition pattern - matches [ref]: url "title"
17    static ref REF_DEF_REGEX: Regex = Regex::new(
18        r#"(?m)^[ ]{0,3}\[([^\]]+)\]:\s*([^\s]+)(?:\s+(?:"([^"]*)"|'([^']*)'))?$"#
19    ).unwrap();
20}
21
22/// Check if an emphasis span has spacing issues that should be flagged
23#[inline]
24fn has_spacing_issues(span: &EmphasisSpan) -> bool {
25    span.has_leading_space || span.has_trailing_space
26}
27
28/// Rule MD037: Spaces inside emphasis markers
29#[derive(Clone)]
30pub struct MD037NoSpaceInEmphasis;
31
32impl Default for MD037NoSpaceInEmphasis {
33    fn default() -> Self {
34        Self
35    }
36}
37
38impl MD037NoSpaceInEmphasis {
39    /// Check if a byte position is within a link (inline links, reference links, or reference definitions)
40    fn is_in_link(&self, ctx: &crate::lint_context::LintContext, byte_pos: usize) -> bool {
41        // Check inline and reference links
42        for link in &ctx.links {
43            if link.byte_offset <= byte_pos && byte_pos < link.byte_end {
44                return true;
45            }
46        }
47
48        // Check images (which use similar syntax)
49        for image in &ctx.images {
50            if image.byte_offset <= byte_pos && byte_pos < image.byte_end {
51                return true;
52            }
53        }
54
55        // Check reference definitions [ref]: url "title" using regex pattern
56        for m in REF_DEF_REGEX.find_iter(ctx.content) {
57            if m.start() <= byte_pos && byte_pos < m.end() {
58                return true;
59            }
60        }
61
62        false
63    }
64}
65
66impl Rule for MD037NoSpaceInEmphasis {
67    fn name(&self) -> &'static str {
68        "MD037"
69    }
70
71    fn description(&self) -> &'static str {
72        "Spaces inside emphasis markers"
73    }
74
75    fn check(&self, ctx: &crate::lint_context::LintContext) -> LintResult {
76        let content = ctx.content;
77        let _timer = crate::profiling::ScopedTimer::new("MD037_check");
78
79        // Early return: if no emphasis markers at all, skip processing
80        if !content.contains('*') && !content.contains('_') {
81            return Ok(vec![]);
82        }
83
84        // Create LineIndex for correct byte position calculations across all line ending types
85        let line_index = LineIndex::new(content.to_string());
86
87        let mut warnings = Vec::new();
88
89        // Process the content line by line
90        for (line_num, line) in content.lines().enumerate() {
91            // Skip if in code block or front matter
92            if ctx.is_in_code_block(line_num + 1) || ctx.is_in_front_matter(line_num + 1) {
93                continue;
94            }
95
96            // Skip if the line doesn't contain any emphasis markers
97            if !line.contains('*') && !line.contains('_') {
98                continue;
99            }
100
101            // Check for emphasis issues on the original line
102            self.check_line_for_emphasis_issues_fast(line, line_num + 1, &mut warnings);
103        }
104
105        // Filter out warnings for emphasis markers that are inside links, HTML comments, or math
106        let mut filtered_warnings = Vec::new();
107
108        for (line_idx, _line) in content.lines().enumerate() {
109            let line_num = line_idx + 1;
110            let line_start_pos = line_index.get_line_start_byte(line_num).unwrap_or(0);
111
112            // Find warnings for this line
113            for warning in &warnings {
114                if warning.line == line_num {
115                    // Calculate byte position of the warning
116                    let byte_pos = line_start_pos + (warning.column - 1);
117
118                    // Skip if inside links, HTML comments, math contexts, or tables
119                    if !self.is_in_link(ctx, byte_pos)
120                        && !is_in_html_comment(content, byte_pos)
121                        && !is_in_math_context(ctx, byte_pos)
122                        && !is_in_table_cell(ctx, line_num, warning.column)
123                    {
124                        filtered_warnings.push(warning.clone());
125                    }
126                }
127            }
128        }
129
130        Ok(filtered_warnings)
131    }
132
133    fn fix(&self, ctx: &crate::lint_context::LintContext) -> Result<String, LintError> {
134        let content = ctx.content;
135        let _timer = crate::profiling::ScopedTimer::new("MD037_fix");
136
137        // Fast path: if no emphasis markers, return unchanged
138        if !content.contains('*') && !content.contains('_') {
139            return Ok(content.to_string());
140        }
141
142        // First check for issues and get all warnings with fixes
143        let warnings = self.check(ctx)?;
144
145        // If no warnings, return original content
146        if warnings.is_empty() {
147            return Ok(content.to_string());
148        }
149
150        // Create LineIndex for correct byte position calculations across all line ending types
151        let line_index = LineIndex::new(content.to_string());
152
153        // Apply fixes
154        let mut result = content.to_string();
155        let mut offset: isize = 0;
156
157        // Sort warnings by position to apply fixes in the correct order
158        let mut sorted_warnings: Vec<_> = warnings.iter().filter(|w| w.fix.is_some()).collect();
159        sorted_warnings.sort_by_key(|w| (w.line, w.column));
160
161        for warning in sorted_warnings {
162            if let Some(fix) = &warning.fix {
163                // Calculate the absolute position in the file
164                let line_start = line_index.get_line_start_byte(warning.line).unwrap_or(0);
165                let abs_start = line_start + warning.column - 1;
166                let abs_end = abs_start + (fix.range.end - fix.range.start);
167
168                // Apply fix with offset adjustment
169                let actual_start = (abs_start as isize + offset) as usize;
170                let actual_end = (abs_end as isize + offset) as usize;
171
172                // Make sure we're not out of bounds
173                if actual_start < result.len() && actual_end <= result.len() {
174                    // Replace the text
175                    result.replace_range(actual_start..actual_end, &fix.replacement);
176                    // Update offset for future replacements
177                    offset += fix.replacement.len() as isize - (fix.range.end - fix.range.start) as isize;
178                }
179            }
180        }
181
182        Ok(result)
183    }
184
185    /// Get the category of this rule for selective processing
186    fn category(&self) -> RuleCategory {
187        RuleCategory::Emphasis
188    }
189
190    /// Check if this rule should be skipped
191    fn should_skip(&self, ctx: &crate::lint_context::LintContext) -> bool {
192        ctx.content.is_empty() || !ctx.likely_has_emphasis()
193    }
194
195    fn as_any(&self) -> &dyn std::any::Any {
196        self
197    }
198
199    fn from_config(_config: &crate::config::Config) -> Box<dyn Rule>
200    where
201        Self: Sized,
202    {
203        Box::new(MD037NoSpaceInEmphasis)
204    }
205}
206
207impl MD037NoSpaceInEmphasis {
208    /// Optimized line checking for emphasis spacing issues
209    #[inline]
210    fn check_line_for_emphasis_issues_fast(&self, line: &str, line_num: usize, warnings: &mut Vec<LintWarning>) {
211        // Quick documentation pattern checks
212        if has_doc_patterns(line) {
213            return;
214        }
215
216        // Optimized list detection with fast path
217        if (line.starts_with(' ') || line.starts_with('*') || line.starts_with('+') || line.starts_with('-'))
218            && UNORDERED_LIST_MARKER_REGEX.is_match(line)
219        {
220            if let Some(caps) = UNORDERED_LIST_MARKER_REGEX.captures(line)
221                && let Some(full_match) = caps.get(0)
222            {
223                let list_marker_end = full_match.end();
224                if list_marker_end < line.len() {
225                    let remaining_content = &line[list_marker_end..];
226
227                    if self.is_likely_list_item_fast(remaining_content) {
228                        self.check_line_content_for_emphasis_fast(
229                            remaining_content,
230                            line_num,
231                            list_marker_end,
232                            warnings,
233                        );
234                    } else {
235                        self.check_line_content_for_emphasis_fast(line, line_num, 0, warnings);
236                    }
237                }
238            }
239            return;
240        }
241
242        // Check the entire line
243        self.check_line_content_for_emphasis_fast(line, line_num, 0, warnings);
244    }
245
246    /// Fast list item detection with optimized logic
247    #[inline]
248    fn is_likely_list_item_fast(&self, content: &str) -> bool {
249        let trimmed = content.trim();
250
251        // Early returns for obvious cases
252        if trimmed.is_empty() || trimmed.len() < 3 {
253            return false;
254        }
255
256        // Quick word count using bytes
257        let word_count = trimmed.split_whitespace().count();
258
259        // Short content ending with * is likely emphasis
260        if word_count <= 2 && trimmed.ends_with('*') && !trimmed.ends_with("**") {
261            return false;
262        }
263
264        // Long content (4+ words) without emphasis is likely a list
265        if word_count >= 4 {
266            // Quick check: if no emphasis markers, it's a list
267            if !trimmed.contains('*') && !trimmed.contains('_') {
268                return true;
269            }
270        }
271
272        // For ambiguous cases, default to emphasis (more conservative)
273        false
274    }
275
276    /// Optimized line content checking for emphasis issues
277    fn check_line_content_for_emphasis_fast(
278        &self,
279        content: &str,
280        line_num: usize,
281        offset: usize,
282        warnings: &mut Vec<LintWarning>,
283    ) {
284        // Replace inline code to avoid false positives with emphasis markers inside backticks
285        let processed_content = replace_inline_code(content);
286
287        // Find all emphasis markers using optimized parsing
288        let markers = find_emphasis_markers(&processed_content);
289        if markers.is_empty() {
290            return;
291        }
292
293        // Find valid emphasis spans
294        let spans = find_emphasis_spans(&processed_content, markers);
295
296        // Check each span for spacing issues
297        for span in spans {
298            if has_spacing_issues(&span) {
299                // Calculate the full span including markers
300                let full_start = span.opening.start_pos;
301                let full_end = span.closing.end_pos();
302                let full_text = &content[full_start..full_end];
303
304                // Skip if this emphasis has a Kramdown span IAL immediately after it
305                // (no space between emphasis and IAL)
306                if full_end < content.len() {
307                    let remaining = &content[full_end..];
308                    // Check if IAL starts immediately after the emphasis (no whitespace)
309                    if remaining.starts_with('{') && has_span_ial(remaining.split_whitespace().next().unwrap_or("")) {
310                        continue;
311                    }
312                }
313
314                // Create the marker string efficiently
315                let marker_char = span.opening.as_char();
316                let marker_str = if span.opening.count == 1 {
317                    marker_char.to_string()
318                } else {
319                    format!("{marker_char}{marker_char}")
320                };
321
322                // Create the fixed version by trimming spaces from content
323                let trimmed_content = span.content.trim();
324                let fixed_text = format!("{marker_str}{trimmed_content}{marker_str}");
325
326                let warning = LintWarning {
327                    rule_name: Some(self.name()),
328                    message: format!("Spaces inside emphasis markers: {full_text:?}"),
329                    line: line_num,
330                    column: offset + full_start + 1, // +1 because columns are 1-indexed
331                    end_line: line_num,
332                    end_column: offset + full_end + 1,
333                    severity: Severity::Warning,
334                    fix: Some(Fix {
335                        range: (offset + full_start)..(offset + full_end),
336                        replacement: fixed_text,
337                    }),
338                };
339
340                warnings.push(warning);
341            }
342        }
343    }
344}
345
346#[cfg(test)]
347mod tests {
348    use super::*;
349    use crate::lint_context::LintContext;
350
351    #[test]
352    fn test_emphasis_marker_parsing() {
353        let markers = find_emphasis_markers("This has *single* and **double** emphasis");
354        assert_eq!(markers.len(), 4); // *, *, **, **
355
356        let markers = find_emphasis_markers("*start* and *end*");
357        assert_eq!(markers.len(), 4); // *, *, *, *
358    }
359
360    #[test]
361    fn test_emphasis_span_detection() {
362        let markers = find_emphasis_markers("This has *valid* emphasis");
363        let spans = find_emphasis_spans("This has *valid* emphasis", markers);
364        assert_eq!(spans.len(), 1);
365        assert_eq!(spans[0].content, "valid");
366        assert!(!spans[0].has_leading_space);
367        assert!(!spans[0].has_trailing_space);
368
369        let markers = find_emphasis_markers("This has * invalid * emphasis");
370        let spans = find_emphasis_spans("This has * invalid * emphasis", markers);
371        assert_eq!(spans.len(), 1);
372        assert_eq!(spans[0].content, " invalid ");
373        assert!(spans[0].has_leading_space);
374        assert!(spans[0].has_trailing_space);
375    }
376
377    #[test]
378    fn test_with_document_structure() {
379        let rule = MD037NoSpaceInEmphasis;
380
381        // Test with no spaces inside emphasis - should pass
382        let content = "This is *correct* emphasis and **strong emphasis**";
383        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
384        let result = rule.check(&ctx).unwrap();
385        assert!(result.is_empty(), "No warnings expected for correct emphasis");
386
387        // Test with actual spaces inside emphasis - use content that should warn
388        let content = "This is * text with spaces * and more content";
389        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
390        let result = rule.check(&ctx).unwrap();
391        assert!(!result.is_empty(), "Expected warnings for spaces in emphasis");
392
393        // Test with code blocks - emphasis in code should be ignored
394        let content = "This is *correct* emphasis\n```\n* incorrect * in code block\n```\nOutside block with * spaces in emphasis *";
395        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
396        let result = rule.check(&ctx).unwrap();
397        assert!(
398            !result.is_empty(),
399            "Expected warnings for spaces in emphasis outside code block"
400        );
401    }
402
403    #[test]
404    fn test_emphasis_in_links_not_flagged() {
405        let rule = MD037NoSpaceInEmphasis;
406        let content = r#"Check this [* spaced asterisk *](https://example.com/*test*) link.
407
408This has * real spaced emphasis * that should be flagged."#;
409        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Standard);
410        let result = rule.check(&ctx).unwrap();
411
412        // Test passed - emphasis inside links are filtered out correctly
413
414        // Only the real emphasis outside links should be flagged
415        assert_eq!(
416            result.len(),
417            1,
418            "Expected exactly 1 warning, but got: {:?}",
419            result.len()
420        );
421        assert!(result[0].message.contains("Spaces inside emphasis markers"));
422        // Should flag "* real spaced emphasis *" but not emphasis patterns inside links
423        assert!(result[0].line == 3); // Line with "* real spaced emphasis *"
424    }
425
426    #[test]
427    fn test_emphasis_in_links_vs_outside_links() {
428        let rule = MD037NoSpaceInEmphasis;
429        let content = r#"Check [* spaced *](https://example.com/*test*) and inline * real spaced * text.
430
431[* link *]: https://example.com/*path*"#;
432        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Standard);
433        let result = rule.check(&ctx).unwrap();
434
435        // Only the actual emphasis outside links should be flagged
436        assert_eq!(result.len(), 1);
437        assert!(result[0].message.contains("Spaces inside emphasis markers"));
438        // Should be the "* real spaced *" text on line 1
439        assert!(result[0].line == 1);
440    }
441
442    #[test]
443    fn test_issue_49_asterisk_in_inline_code() {
444        // Test for issue #49 - Asterisk within backticks identified as for emphasis
445        let rule = MD037NoSpaceInEmphasis;
446
447        // Test case from issue #49
448        let content = "The `__mul__` method is needed for left-hand multiplication (`vector * 3`) and `__rmul__` is needed for right-hand multiplication (`3 * vector`).";
449        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
450        let result = rule.check(&ctx).unwrap();
451        assert!(
452            result.is_empty(),
453            "Should not flag asterisks inside inline code as emphasis (issue #49). Got: {result:?}"
454        );
455    }
456
457    #[test]
458    fn test_issue_28_inline_code_in_emphasis() {
459        // Test for issue #28 - MD037 should not flag inline code inside emphasis as spaces
460        let rule = MD037NoSpaceInEmphasis;
461
462        // Test case 1: inline code with single backticks inside bold emphasis
463        let content = "Though, we often call this an **inline `if`** because it looks sort of like an `if`-`else` statement all in *one line* of code.";
464        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
465        let result = rule.check(&ctx).unwrap();
466        assert!(
467            result.is_empty(),
468            "Should not flag inline code inside emphasis as spaces (issue #28). Got: {result:?}"
469        );
470
471        // Test case 2: multiple inline code snippets inside emphasis
472        let content2 = "The **`foo` and `bar`** methods are important.";
473        let ctx2 = LintContext::new(content2, crate::config::MarkdownFlavor::Standard);
474        let result2 = rule.check(&ctx2).unwrap();
475        assert!(
476            result2.is_empty(),
477            "Should not flag multiple inline code snippets inside emphasis. Got: {result2:?}"
478        );
479
480        // Test case 3: inline code with underscores for emphasis
481        let content3 = "This is __inline `code`__ with underscores.";
482        let ctx3 = LintContext::new(content3, crate::config::MarkdownFlavor::Standard);
483        let result3 = rule.check(&ctx3).unwrap();
484        assert!(
485            result3.is_empty(),
486            "Should not flag inline code with underscore emphasis. Got: {result3:?}"
487        );
488
489        // Test case 4: single asterisk emphasis with inline code
490        let content4 = "This is *inline `test`* with single asterisks.";
491        let ctx4 = LintContext::new(content4, crate::config::MarkdownFlavor::Standard);
492        let result4 = rule.check(&ctx4).unwrap();
493        assert!(
494            result4.is_empty(),
495            "Should not flag inline code with single asterisk emphasis. Got: {result4:?}"
496        );
497
498        // Test case 5: actual spaces that should be flagged
499        let content5 = "This has * real spaces * that should be flagged.";
500        let ctx5 = LintContext::new(content5, crate::config::MarkdownFlavor::Standard);
501        let result5 = rule.check(&ctx5).unwrap();
502        assert!(!result5.is_empty(), "Should still flag actual spaces in emphasis");
503        assert!(result5[0].message.contains("Spaces inside emphasis markers"));
504    }
505}
rumdl_lib/rules/md037_spaces_around_emphasis.rs

rumdl_lib/rules/
md037_spaces_around_emphasis.rs
