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::regex_cache::UNORDERED_LIST_MARKER_REGEX;
10use crate::utils::skip_context::{is_in_html_comment, is_in_math_context, is_in_table_cell};
11use lazy_static::lazy_static;
12use regex::Regex;
13
14lazy_static! {
15    // Reference definition pattern - matches [ref]: url "title"
16    static ref REF_DEF_REGEX: Regex = Regex::new(
17        r#"(?m)^[ ]{0,3}\[([^\]]+)\]:\s*([^\s]+)(?:\s+(?:"([^"]*)"|'([^']*)'))?$"#
18    ).unwrap();
19}
20
21/// Check if an emphasis span has spacing issues that should be flagged
22#[inline]
23fn has_spacing_issues(span: &EmphasisSpan) -> bool {
24    span.has_leading_space || span.has_trailing_space
25}
26
27/// Rule MD037: Spaces inside emphasis markers
28#[derive(Clone)]
29pub struct MD037NoSpaceInEmphasis;
30
31impl Default for MD037NoSpaceInEmphasis {
32    fn default() -> Self {
33        Self
34    }
35}
36
37impl MD037NoSpaceInEmphasis {
38    /// Check if a byte position is within a link (inline links, reference links, or reference definitions)
39    fn is_in_link(&self, ctx: &crate::lint_context::LintContext, byte_pos: usize) -> bool {
40        // Check inline and reference links
41        for link in &ctx.links {
42            if link.byte_offset <= byte_pos && byte_pos < link.byte_end {
43                return true;
44            }
45        }
46
47        // Check images (which use similar syntax)
48        for image in &ctx.images {
49            if image.byte_offset <= byte_pos && byte_pos < image.byte_end {
50                return true;
51            }
52        }
53
54        // Check reference definitions [ref]: url "title" using regex pattern
55        for m in REF_DEF_REGEX.find_iter(ctx.content) {
56            if m.start() <= byte_pos && byte_pos < m.end() {
57                return true;
58            }
59        }
60
61        false
62    }
63}
64
65impl Rule for MD037NoSpaceInEmphasis {
66    fn name(&self) -> &'static str {
67        "MD037"
68    }
69
70    fn description(&self) -> &'static str {
71        "Spaces inside emphasis markers"
72    }
73
74    fn check(&self, ctx: &crate::lint_context::LintContext) -> LintResult {
75        let content = ctx.content;
76        let _timer = crate::profiling::ScopedTimer::new("MD037_check");
77
78        // Early return: if no emphasis markers at all, skip processing
79        if !content.contains('*') && !content.contains('_') {
80            return Ok(vec![]);
81        }
82
83        let mut warnings = Vec::new();
84
85        // Process the content line by line
86        for (line_num, line) in content.lines().enumerate() {
87            // Skip if in code block or front matter
88            if ctx.is_in_code_block(line_num + 1) || ctx.is_in_front_matter(line_num + 1) {
89                continue;
90            }
91
92            // Skip if the line doesn't contain any emphasis markers
93            if !line.contains('*') && !line.contains('_') {
94                continue;
95            }
96
97            // Check for emphasis issues on the original line
98            self.check_line_for_emphasis_issues_fast(line, line_num + 1, &mut warnings);
99        }
100
101        // Filter out warnings for emphasis markers that are inside links, HTML comments, or math
102        let mut filtered_warnings = Vec::new();
103        let mut line_start_pos = 0;
104
105        for (line_idx, line) in content.lines().enumerate() {
106            let line_num = line_idx + 1;
107
108            // Find warnings for this line
109            for warning in &warnings {
110                if warning.line == line_num {
111                    // Calculate byte position of the warning
112                    let byte_pos = line_start_pos + (warning.column - 1);
113
114                    // Skip if inside links, HTML comments, math contexts, or tables
115                    if !self.is_in_link(ctx, byte_pos)
116                        && !is_in_html_comment(content, byte_pos)
117                        && !is_in_math_context(ctx, byte_pos)
118                        && !is_in_table_cell(ctx, line_num, warning.column)
119                    {
120                        filtered_warnings.push(warning.clone());
121                    }
122                }
123            }
124
125            line_start_pos += line.len() + 1; // +1 for newline
126        }
127
128        Ok(filtered_warnings)
129    }
130
131    fn fix(&self, ctx: &crate::lint_context::LintContext) -> Result<String, LintError> {
132        let content = ctx.content;
133        let _timer = crate::profiling::ScopedTimer::new("MD037_fix");
134
135        // Fast path: if no emphasis markers, return unchanged
136        if !content.contains('*') && !content.contains('_') {
137            return Ok(content.to_string());
138        }
139
140        // First check for issues and get all warnings with fixes
141        let warnings = self.check(ctx)?;
142
143        // If no warnings, return original content
144        if warnings.is_empty() {
145            return Ok(content.to_string());
146        }
147
148        // Get all line positions to make it easier to apply fixes by warning
149        let mut line_positions = Vec::new();
150        let mut pos = 0;
151        for line in content.lines() {
152            line_positions.push(pos);
153            pos += line.len() + 1; // +1 for the newline
154        }
155
156        // Apply fixes
157        let mut result = content.to_string();
158        let mut offset: isize = 0;
159
160        // Sort warnings by position to apply fixes in the correct order
161        let mut sorted_warnings: Vec<_> = warnings.iter().filter(|w| w.fix.is_some()).collect();
162        sorted_warnings.sort_by_key(|w| (w.line, w.column));
163
164        for warning in sorted_warnings {
165            if let Some(fix) = &warning.fix {
166                // Calculate the absolute position in the file
167                let line_start = line_positions.get(warning.line - 1).copied().unwrap_or(0);
168                let abs_start = line_start + warning.column - 1;
169                let abs_end = abs_start + (fix.range.end - fix.range.start);
170
171                // Apply fix with offset adjustment
172                let actual_start = (abs_start as isize + offset) as usize;
173                let actual_end = (abs_end as isize + offset) as usize;
174
175                // Make sure we're not out of bounds
176                if actual_start < result.len() && actual_end <= result.len() {
177                    // Replace the text
178                    result.replace_range(actual_start..actual_end, &fix.replacement);
179                    // Update offset for future replacements
180                    offset += fix.replacement.len() as isize - (fix.range.end - fix.range.start) as isize;
181                }
182            }
183        }
184
185        Ok(result)
186    }
187
188    /// Get the category of this rule for selective processing
189    fn category(&self) -> RuleCategory {
190        RuleCategory::Emphasis
191    }
192
193    /// Check if this rule should be skipped
194    fn should_skip(&self, ctx: &crate::lint_context::LintContext) -> bool {
195        ctx.content.is_empty() || !ctx.likely_has_emphasis()
196    }
197
198    fn as_any(&self) -> &dyn std::any::Any {
199        self
200    }
201
202    fn from_config(_config: &crate::config::Config) -> Box<dyn Rule>
203    where
204        Self: Sized,
205    {
206        Box::new(MD037NoSpaceInEmphasis)
207    }
208}
209
210impl MD037NoSpaceInEmphasis {
211    /// Optimized line checking for emphasis spacing issues
212    #[inline]
213    fn check_line_for_emphasis_issues_fast(&self, line: &str, line_num: usize, warnings: &mut Vec<LintWarning>) {
214        // Quick documentation pattern checks
215        if has_doc_patterns(line) {
216            return;
217        }
218
219        // Optimized list detection with fast path
220        if (line.starts_with(' ') || line.starts_with('*') || line.starts_with('+') || line.starts_with('-'))
221            && UNORDERED_LIST_MARKER_REGEX.is_match(line)
222        {
223            if let Some(caps) = UNORDERED_LIST_MARKER_REGEX.captures(line)
224                && let Some(full_match) = caps.get(0)
225            {
226                let list_marker_end = full_match.end();
227                if list_marker_end < line.len() {
228                    let remaining_content = &line[list_marker_end..];
229
230                    if self.is_likely_list_item_fast(remaining_content) {
231                        self.check_line_content_for_emphasis_fast(
232                            remaining_content,
233                            line_num,
234                            list_marker_end,
235                            warnings,
236                        );
237                    } else {
238                        self.check_line_content_for_emphasis_fast(line, line_num, 0, warnings);
239                    }
240                }
241            }
242            return;
243        }
244
245        // Check the entire line
246        self.check_line_content_for_emphasis_fast(line, line_num, 0, warnings);
247    }
248
249    /// Fast list item detection with optimized logic
250    #[inline]
251    fn is_likely_list_item_fast(&self, content: &str) -> bool {
252        let trimmed = content.trim();
253
254        // Early returns for obvious cases
255        if trimmed.is_empty() || trimmed.len() < 3 {
256            return false;
257        }
258
259        // Quick word count using bytes
260        let word_count = trimmed.split_whitespace().count();
261
262        // Short content ending with * is likely emphasis
263        if word_count <= 2 && trimmed.ends_with('*') && !trimmed.ends_with("**") {
264            return false;
265        }
266
267        // Long content (4+ words) without emphasis is likely a list
268        if word_count >= 4 {
269            // Quick check: if no emphasis markers, it's a list
270            if !trimmed.contains('*') && !trimmed.contains('_') {
271                return true;
272            }
273        }
274
275        // For ambiguous cases, default to emphasis (more conservative)
276        false
277    }
278
279    /// Optimized line content checking for emphasis issues
280    fn check_line_content_for_emphasis_fast(
281        &self,
282        content: &str,
283        line_num: usize,
284        offset: usize,
285        warnings: &mut Vec<LintWarning>,
286    ) {
287        // Replace inline code to avoid false positives with emphasis markers inside backticks
288        let processed_content = replace_inline_code(content);
289
290        // Find all emphasis markers using optimized parsing
291        let markers = find_emphasis_markers(&processed_content);
292        if markers.is_empty() {
293            return;
294        }
295
296        // Find valid emphasis spans
297        let spans = find_emphasis_spans(&processed_content, markers);
298
299        // Check each span for spacing issues
300        for span in spans {
301            if has_spacing_issues(&span) {
302                // Calculate the full span including markers
303                let full_start = span.opening.start_pos;
304                let full_end = span.closing.end_pos();
305                let full_text = &content[full_start..full_end];
306
307                // Skip if this emphasis has a Kramdown span IAL immediately after it
308                // (no space between emphasis and IAL)
309                if full_end < content.len() {
310                    let remaining = &content[full_end..];
311                    // Check if IAL starts immediately after the emphasis (no whitespace)
312                    if remaining.starts_with('{') && has_span_ial(remaining.split_whitespace().next().unwrap_or("")) {
313                        continue;
314                    }
315                }
316
317                // Create the marker string efficiently
318                let marker_char = span.opening.as_char();
319                let marker_str = if span.opening.count == 1 {
320                    marker_char.to_string()
321                } else {
322                    format!("{marker_char}{marker_char}")
323                };
324
325                // Create the fixed version by trimming spaces from content
326                let trimmed_content = span.content.trim();
327                let fixed_text = format!("{marker_str}{trimmed_content}{marker_str}");
328
329                let warning = LintWarning {
330                    rule_name: Some(self.name()),
331                    message: format!("Spaces inside emphasis markers: {full_text:?}"),
332                    line: line_num,
333                    column: offset + full_start + 1, // +1 because columns are 1-indexed
334                    end_line: line_num,
335                    end_column: offset + full_end + 1,
336                    severity: Severity::Warning,
337                    fix: Some(Fix {
338                        range: (offset + full_start)..(offset + full_end),
339                        replacement: fixed_text,
340                    }),
341                };
342
343                warnings.push(warning);
344            }
345        }
346    }
347}
348
349#[cfg(test)]
350mod tests {
351    use super::*;
352    use crate::lint_context::LintContext;
353
354    #[test]
355    fn test_emphasis_marker_parsing() {
356        let markers = find_emphasis_markers("This has *single* and **double** emphasis");
357        assert_eq!(markers.len(), 4); // *, *, **, **
358
359        let markers = find_emphasis_markers("*start* and *end*");
360        assert_eq!(markers.len(), 4); // *, *, *, *
361    }
362
363    #[test]
364    fn test_emphasis_span_detection() {
365        let markers = find_emphasis_markers("This has *valid* emphasis");
366        let spans = find_emphasis_spans("This has *valid* emphasis", markers);
367        assert_eq!(spans.len(), 1);
368        assert_eq!(spans[0].content, "valid");
369        assert!(!spans[0].has_leading_space);
370        assert!(!spans[0].has_trailing_space);
371
372        let markers = find_emphasis_markers("This has * invalid * emphasis");
373        let spans = find_emphasis_spans("This has * invalid * emphasis", markers);
374        assert_eq!(spans.len(), 1);
375        assert_eq!(spans[0].content, " invalid ");
376        assert!(spans[0].has_leading_space);
377        assert!(spans[0].has_trailing_space);
378    }
379
380    #[test]
381    fn test_with_document_structure() {
382        let rule = MD037NoSpaceInEmphasis;
383
384        // Test with no spaces inside emphasis - should pass
385        let content = "This is *correct* emphasis and **strong emphasis**";
386        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
387        let result = rule.check(&ctx).unwrap();
388        assert!(result.is_empty(), "No warnings expected for correct emphasis");
389
390        // Test with actual spaces inside emphasis - use content that should warn
391        let content = "This is * text with spaces * and more content";
392        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
393        let result = rule.check(&ctx).unwrap();
394        assert!(!result.is_empty(), "Expected warnings for spaces in emphasis");
395
396        // Test with code blocks - emphasis in code should be ignored
397        let content = "This is *correct* emphasis\n```\n* incorrect * in code block\n```\nOutside block with * spaces in emphasis *";
398        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
399        let result = rule.check(&ctx).unwrap();
400        assert!(
401            !result.is_empty(),
402            "Expected warnings for spaces in emphasis outside code block"
403        );
404    }
405
406    #[test]
407    fn test_emphasis_in_links_not_flagged() {
408        let rule = MD037NoSpaceInEmphasis;
409        let content = r#"Check this [* spaced asterisk *](https://example.com/*test*) link.
410
411This has * real spaced emphasis * that should be flagged."#;
412        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Standard);
413        let result = rule.check(&ctx).unwrap();
414
415        // Test passed - emphasis inside links are filtered out correctly
416
417        // Only the real emphasis outside links should be flagged
418        assert_eq!(
419            result.len(),
420            1,
421            "Expected exactly 1 warning, but got: {:?}",
422            result.len()
423        );
424        assert!(result[0].message.contains("Spaces inside emphasis markers"));
425        // Should flag "* real spaced emphasis *" but not emphasis patterns inside links
426        assert!(result[0].line == 3); // Line with "* real spaced emphasis *"
427    }
428
429    #[test]
430    fn test_emphasis_in_links_vs_outside_links() {
431        let rule = MD037NoSpaceInEmphasis;
432        let content = r#"Check [* spaced *](https://example.com/*test*) and inline * real spaced * text.
433
434[* link *]: https://example.com/*path*"#;
435        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Standard);
436        let result = rule.check(&ctx).unwrap();
437
438        // Only the actual emphasis outside links should be flagged
439        assert_eq!(result.len(), 1);
440        assert!(result[0].message.contains("Spaces inside emphasis markers"));
441        // Should be the "* real spaced *" text on line 1
442        assert!(result[0].line == 1);
443    }
444
445    #[test]
446    fn test_issue_49_asterisk_in_inline_code() {
447        // Test for issue #49 - Asterisk within backticks identified as for emphasis
448        let rule = MD037NoSpaceInEmphasis;
449
450        // Test case from issue #49
451        let content = "The `__mul__` method is needed for left-hand multiplication (`vector * 3`) and `__rmul__` is needed for right-hand multiplication (`3 * vector`).";
452        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
453        let result = rule.check(&ctx).unwrap();
454        assert!(
455            result.is_empty(),
456            "Should not flag asterisks inside inline code as emphasis (issue #49). Got: {result:?}"
457        );
458    }
459
460    #[test]
461    fn test_issue_28_inline_code_in_emphasis() {
462        // Test for issue #28 - MD037 should not flag inline code inside emphasis as spaces
463        let rule = MD037NoSpaceInEmphasis;
464
465        // Test case 1: inline code with single backticks inside bold emphasis
466        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.";
467        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
468        let result = rule.check(&ctx).unwrap();
469        assert!(
470            result.is_empty(),
471            "Should not flag inline code inside emphasis as spaces (issue #28). Got: {result:?}"
472        );
473
474        // Test case 2: multiple inline code snippets inside emphasis
475        let content2 = "The **`foo` and `bar`** methods are important.";
476        let ctx2 = LintContext::new(content2, crate::config::MarkdownFlavor::Standard);
477        let result2 = rule.check(&ctx2).unwrap();
478        assert!(
479            result2.is_empty(),
480            "Should not flag multiple inline code snippets inside emphasis. Got: {result2:?}"
481        );
482
483        // Test case 3: inline code with underscores for emphasis
484        let content3 = "This is __inline `code`__ with underscores.";
485        let ctx3 = LintContext::new(content3, crate::config::MarkdownFlavor::Standard);
486        let result3 = rule.check(&ctx3).unwrap();
487        assert!(
488            result3.is_empty(),
489            "Should not flag inline code with underscore emphasis. Got: {result3:?}"
490        );
491
492        // Test case 4: single asterisk emphasis with inline code
493        let content4 = "This is *inline `test`* with single asterisks.";
494        let ctx4 = LintContext::new(content4, crate::config::MarkdownFlavor::Standard);
495        let result4 = rule.check(&ctx4).unwrap();
496        assert!(
497            result4.is_empty(),
498            "Should not flag inline code with single asterisk emphasis. Got: {result4:?}"
499        );
500
501        // Test case 5: actual spaces that should be flagged
502        let content5 = "This has * real spaces * that should be flagged.";
503        let ctx5 = LintContext::new(content5, crate::config::MarkdownFlavor::Standard);
504        let result5 = rule.check(&ctx5).unwrap();
505        assert!(!result5.is_empty(), "Should still flag actual spaces in emphasis");
506        assert!(result5[0].message.contains("Spaces inside emphasis markers"));
507    }
508}
rumdl_lib/rules/md037_spaces_around_emphasis.rs

rumdl_lib/rules/
md037_spaces_around_emphasis.rs
