Skip to main content

rumdl_lib/rules/
md038_no_space_in_code.rs

1use crate::rule::{Fix, LintError, LintResult, LintWarning, Rule, RuleCategory, Severity};
2use crate::utils::mkdocs_extensions::is_inline_hilite_content;
3
4/// Rule MD038: No space inside code span markers
5///
6/// See [docs/md038.md](../../docs/md038.md) for full documentation, configuration, and examples.
7///
8/// MD038: Spaces inside code span elements
9///
10/// This rule is triggered when there are spaces inside code span elements.
11///
12/// For example:
13///
14/// ``` markdown
15/// ` some text`
16/// `some text `
17/// ` some text `
18/// ```
19///
20/// To fix this issue, remove the leading and trailing spaces within the code span markers:
21///
22/// ``` markdown
23/// `some text`
24/// ```
25///
26/// Note: Code spans containing backticks (e.g., `` `backticks` inside ``) are not flagged
27/// to avoid breaking nested backtick structures used to display backticks in documentation.
28#[derive(Debug, Clone, Default)]
29pub struct MD038NoSpaceInCode {
30    pub enabled: bool,
31}
32
33impl MD038NoSpaceInCode {
34    pub fn new() -> Self {
35        Self { enabled: true }
36    }
37
38    /// Check if a code span is part of Hugo template syntax (e.g., {{raw `...`}})
39    ///
40    /// Hugo static site generator uses backticks as part of template delimiters,
41    /// not markdown code spans. This function detects common Hugo shortcode patterns:
42    /// - {{raw `...`}} - Raw HTML shortcode
43    /// - {{< `...` >}} - Partial shortcode
44    /// - {{% `...` %}} - Shortcode with percent delimiters
45    /// - {{ `...` }} - Generic shortcode
46    ///
47    /// The detection is conservative to avoid false positives:
48    /// - Requires opening {{ pattern before the backtick
49    /// - Requires closing }} after the code span
50    /// - Handles multi-line templates correctly
51    ///
52    /// Returns true if the code span is part of Hugo template syntax and should be skipped.
53    fn is_hugo_template_syntax(
54        &self,
55        ctx: &crate::lint_context::LintContext,
56        code_span: &crate::lint_context::CodeSpan,
57    ) -> bool {
58        let start_line_idx = code_span.line.saturating_sub(1);
59        if start_line_idx >= ctx.lines.len() {
60            return false;
61        }
62
63        let start_line_content = ctx.lines[start_line_idx].content(ctx.content);
64
65        // start_col is 0-indexed character position
66        let span_start_col = code_span.start_col;
67
68        // Check if there's Hugo template syntax before the code span on the same line
69        // Pattern: {{raw ` or {{< ` or similar Hugo template patterns
70        // The code span starts at the backtick, so we need to check what's before it
71        // span_start_col is the position of the backtick (0-indexed character position)
72        // Minimum pattern is "{{ `" which has 3 characters before the backtick
73        if span_start_col >= 3 {
74            // Look backwards for Hugo template patterns
75            // Get the content up to (but not including) the backtick
76            let before_span: String = start_line_content.chars().take(span_start_col).collect();
77
78            // Check for Hugo template patterns: {{raw `, {{< `, {{% `, etc.
79            // The backtick is at span_start_col, so we check if the content before it
80            // ends with the Hugo pattern (without the backtick), and verify the next char is a backtick
81            let char_at_span_start = start_line_content.chars().nth(span_start_col).unwrap_or(' ');
82
83            // Match Hugo shortcode patterns:
84            // - {{raw ` - Raw HTML shortcode
85            // - {{< ` - Partial shortcode (may have parameters before backtick)
86            // - {{% ` - Shortcode with percent delimiters
87            // - {{ ` - Generic shortcode
88            // Also handle cases with parameters: {{< highlight go ` or {{< code ` etc.
89            // We check if the pattern starts with {{ and contains the shortcode type before the backtick
90            let is_hugo_start =
91                // Exact match: {{raw `
92                (before_span.ends_with("{{raw ") && char_at_span_start == '`')
93                // Partial shortcode: {{< ` or {{< name ` or {{< name param ` etc.
94                || (before_span.starts_with("{{<") && before_span.ends_with(' ') && char_at_span_start == '`')
95                // Percent shortcode: {{% `
96                || (before_span.ends_with("{{% ") && char_at_span_start == '`')
97                // Generic shortcode: {{ `
98                || (before_span.ends_with("{{ ") && char_at_span_start == '`');
99
100            if is_hugo_start {
101                // Check if there's a closing }} after the code span
102                // First check the end line of the code span
103                let end_line_idx = code_span.end_line.saturating_sub(1);
104                if end_line_idx < ctx.lines.len() {
105                    let end_line_content = ctx.lines[end_line_idx].content(ctx.content);
106                    let end_line_char_count = end_line_content.chars().count();
107                    let span_end_col = code_span.end_col.min(end_line_char_count);
108
109                    // Check for closing }} on the same line as the end of the code span
110                    if span_end_col < end_line_char_count {
111                        let after_span: String = end_line_content.chars().skip(span_end_col).collect();
112                        if after_span.trim_start().starts_with("}}") {
113                            return true;
114                        }
115                    }
116
117                    // Also check the next line for closing }}
118                    let next_line_idx = code_span.end_line;
119                    if next_line_idx < ctx.lines.len() {
120                        let next_line = ctx.lines[next_line_idx].content(ctx.content);
121                        if next_line.trim_start().starts_with("}}") {
122                            return true;
123                        }
124                    }
125                }
126            }
127        }
128
129        false
130    }
131
132    /// Check if content is an Obsidian Dataview inline query
133    ///
134    /// Dataview plugin uses two inline query syntaxes:
135    /// - Inline DQL: `= expression` - Starts with "= "
136    /// - Inline DataviewJS: `$= expression` - Starts with "$= "
137    ///
138    /// Examples:
139    /// - `= this.file.name` - Get current file name
140    /// - `= date(today)` - Get today's date
141    /// - `= [[Page]].field` - Access field from another page
142    /// - `$= dv.current().file.mtime` - DataviewJS expression
143    /// - `$= dv.pages().length` - Count pages
144    ///
145    /// These patterns legitimately start with a space after = or $=,
146    /// so they should not trigger MD038.
147    fn is_dataview_expression(content: &str) -> bool {
148        // Inline DQL: starts with "= " (equals followed by space)
149        // Inline DataviewJS: starts with "$= " (dollar-equals followed by space)
150        content.starts_with("= ") || content.starts_with("$= ")
151    }
152
153    /// Check if a code span is likely part of a nested backtick structure
154    fn is_likely_nested_backticks(&self, ctx: &crate::lint_context::LintContext, span_index: usize) -> bool {
155        // If there are multiple code spans on the same line, and there's text
156        // between them that contains "code" or other indicators, it's likely nested
157        let code_spans = ctx.code_spans();
158        let current_span = &code_spans[span_index];
159        let current_line = current_span.line;
160
161        // Look for other code spans on the same line
162        let same_line_spans: Vec<_> = code_spans
163            .iter()
164            .enumerate()
165            .filter(|(i, s)| s.line == current_line && *i != span_index)
166            .collect();
167
168        if same_line_spans.is_empty() {
169            return false;
170        }
171
172        // Check if there's content between spans that might indicate nesting
173        // Get the line content
174        let line_idx = current_line - 1; // Convert to 0-based
175        if line_idx >= ctx.lines.len() {
176            return false;
177        }
178
179        let line_content = &ctx.lines[line_idx].content(ctx.content);
180
181        // For each pair of adjacent code spans, check what's between them
182        for (_, other_span) in &same_line_spans {
183            let start_char = current_span.end_col.min(other_span.end_col);
184            let end_char = current_span.start_col.max(other_span.start_col);
185
186            if start_char < end_char {
187                // Convert character positions to byte offsets for string slicing
188                let char_indices: Vec<(usize, char)> = line_content.char_indices().collect();
189                let start_byte = char_indices.get(start_char).map(|(i, _)| *i);
190                let end_byte = char_indices.get(end_char).map_or(line_content.len(), |(i, _)| *i);
191
192                if let Some(start_byte) = start_byte
193                    && start_byte < end_byte
194                    && end_byte <= line_content.len()
195                {
196                    let between = &line_content[start_byte..end_byte];
197                    // If there's text containing "code" or similar patterns between spans,
198                    // it's likely they're showing nested backticks
199                    if between.contains("code") || between.contains("backtick") {
200                        return true;
201                    }
202                }
203            }
204        }
205
206        false
207    }
208}
209
210impl Rule for MD038NoSpaceInCode {
211    fn name(&self) -> &'static str {
212        "MD038"
213    }
214
215    fn description(&self) -> &'static str {
216        "Spaces inside code span elements"
217    }
218
219    fn category(&self) -> RuleCategory {
220        RuleCategory::Other
221    }
222
223    fn check(&self, ctx: &crate::lint_context::LintContext) -> LintResult {
224        if !self.enabled {
225            return Ok(vec![]);
226        }
227
228        let mut warnings = Vec::new();
229
230        // Use centralized code spans from LintContext
231        let code_spans = ctx.code_spans();
232        for (i, code_span) in code_spans.iter().enumerate() {
233            // Skip code spans that are inside fenced/indented code blocks
234            if let Some(line_info) = ctx.lines.get(code_span.line - 1) {
235                if line_info.in_code_block {
236                    continue;
237                }
238                // Skip multi-line code spans inside MkDocs containers where pulldown-cmark
239                // misinterprets indented fenced code block markers as code spans.
240                // Covers admonitions, tabs, HTML markdown blocks, and PyMdown blocks.
241                if (line_info.in_mkdocs_container() || line_info.in_pymdown_block) && code_span.content.contains('\n') {
242                    continue;
243                }
244            }
245
246            let code_content = &code_span.content;
247
248            // Skip empty code spans
249            if code_content.is_empty() {
250                continue;
251            }
252
253            // Early check: if no leading/trailing whitespace, skip
254            let has_leading_space = code_content.chars().next().is_some_and(char::is_whitespace);
255            let has_trailing_space = code_content.chars().last().is_some_and(char::is_whitespace);
256
257            if !has_leading_space && !has_trailing_space {
258                continue;
259            }
260
261            let trimmed = code_content.trim();
262
263            // Check if there are leading or trailing spaces
264            if code_content != trimmed {
265                // CommonMark behavior: if there is exactly ONE space at start AND ONE at end,
266                // and the content after trimming is non-empty, those spaces are stripped.
267                // We should NOT flag this case since the spaces are intentionally stripped.
268                // See: https://spec.commonmark.org/0.31.2/#code-spans
269                //
270                // Examples:
271                // ` text ` → "text" (spaces stripped, NOT flagged)
272                // `  text ` → " text" (extra leading space remains, FLAGGED)
273                // ` text  ` → "text " (extra trailing space remains, FLAGGED)
274                // ` text` → " text" (no trailing space to balance, FLAGGED)
275                // `text ` → "text " (no leading space to balance, FLAGGED)
276                if has_leading_space && has_trailing_space && !trimmed.is_empty() {
277                    let leading_spaces = code_content.len() - code_content.trim_start().len();
278                    let trailing_spaces = code_content.len() - code_content.trim_end().len();
279
280                    // Exactly one space on each side - CommonMark strips them
281                    if leading_spaces == 1 && trailing_spaces == 1 {
282                        continue;
283                    }
284                }
285                // Check if the content itself contains backticks - if so, skip to avoid
286                // breaking nested backtick structures
287                if trimmed.contains('`') {
288                    continue;
289                }
290
291                // Skip inline R code in Quarto/RMarkdown: `r expression`
292                // This is a legitimate pattern where space is required after 'r'
293                if ctx.flavor == crate::config::MarkdownFlavor::Quarto
294                    && trimmed.starts_with('r')
295                    && trimmed.len() > 1
296                    && trimmed.chars().nth(1).is_some_and(char::is_whitespace)
297                {
298                    continue;
299                }
300
301                // Skip InlineHilite syntax in MkDocs: `#!python code`
302                // The space after the language specifier is legitimate
303                if ctx.flavor == crate::config::MarkdownFlavor::MkDocs && is_inline_hilite_content(trimmed) {
304                    continue;
305                }
306
307                // Skip Dataview inline queries in Obsidian: `= expression` or `$= expression`
308                // Dataview plugin uses these patterns for inline DQL and DataviewJS queries.
309                // The space after = or $= is part of the syntax, not a spacing error.
310                if ctx.flavor == crate::config::MarkdownFlavor::Obsidian && Self::is_dataview_expression(code_content) {
311                    continue;
312                }
313
314                // Check if this is part of Hugo template syntax (e.g., {{raw `...`}})
315                // Hugo uses backticks as part of template delimiters, not markdown code spans
316                if self.is_hugo_template_syntax(ctx, code_span) {
317                    continue;
318                }
319
320                // Check if this might be part of a nested backtick structure
321                // by looking for other code spans nearby that might indicate nesting
322                if self.is_likely_nested_backticks(ctx, i) {
323                    continue;
324                }
325
326                warnings.push(LintWarning {
327                    rule_name: Some(self.name().to_string()),
328                    line: code_span.line,
329                    column: code_span.start_col + 1, // Convert to 1-indexed
330                    end_line: code_span.line,
331                    end_column: code_span.end_col, // Don't add 1 to match test expectation
332                    message: "Spaces inside code span elements".to_string(),
333                    severity: Severity::Warning,
334                    fix: Some(Fix {
335                        range: code_span.byte_offset..code_span.byte_end,
336                        replacement: format!(
337                            "{}{}{}",
338                            "`".repeat(code_span.backtick_count),
339                            trimmed,
340                            "`".repeat(code_span.backtick_count)
341                        ),
342                    }),
343                });
344            }
345        }
346
347        Ok(warnings)
348    }
349
350    fn fix(&self, ctx: &crate::lint_context::LintContext) -> Result<String, LintError> {
351        let content = ctx.content;
352        if !self.enabled {
353            return Ok(content.to_string());
354        }
355
356        // Early return if no backticks in content
357        if !content.contains('`') {
358            return Ok(content.to_string());
359        }
360
361        // Get warnings to identify what needs to be fixed
362        let warnings = self.check(ctx)?;
363        let warnings =
364            crate::utils::fix_utils::filter_warnings_by_inline_config(warnings, ctx.inline_config(), self.name());
365        if warnings.is_empty() {
366            return Ok(content.to_string());
367        }
368
369        // Collect all fixes and sort by position (reverse order to avoid position shifts)
370        let mut fixes: Vec<(std::ops::Range<usize>, String)> = warnings
371            .into_iter()
372            .filter_map(|w| w.fix.map(|f| (f.range, f.replacement)))
373            .collect();
374
375        fixes.sort_by_key(|(range, _)| std::cmp::Reverse(range.start));
376
377        // Apply fixes - only allocate string when we have fixes to apply
378        let mut result = content.to_string();
379        for (range, replacement) in fixes {
380            result.replace_range(range, &replacement);
381        }
382
383        Ok(result)
384    }
385
386    /// Check if content is likely to have code spans
387    fn should_skip(&self, ctx: &crate::lint_context::LintContext) -> bool {
388        !ctx.likely_has_code()
389    }
390
391    fn as_any(&self) -> &dyn std::any::Any {
392        self
393    }
394
395    fn from_config(_config: &crate::config::Config) -> Box<dyn Rule>
396    where
397        Self: Sized,
398    {
399        Box::new(MD038NoSpaceInCode { enabled: true })
400    }
401}
402
403#[cfg(test)]
404mod tests {
405    use super::*;
406
407    #[test]
408    fn test_md038_readme_false_positives() {
409        // These are the exact cases from README.md that are incorrectly flagged
410        let rule = MD038NoSpaceInCode::new();
411        let valid_cases = vec![
412            "3. `pyproject.toml` (must contain `[tool.rumdl]` section)",
413            "#### Effective Configuration (`rumdl config`)",
414            "- Blue: `.rumdl.toml`",
415            "### Defaults Only (`rumdl config --defaults`)",
416        ];
417
418        for case in valid_cases {
419            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Standard, None);
420            let result = rule.check(&ctx).unwrap();
421            assert!(
422                result.is_empty(),
423                "Should not flag code spans without leading/trailing spaces: '{}'. Got {} warnings",
424                case,
425                result.len()
426            );
427        }
428    }
429
430    #[test]
431    fn test_md038_valid() {
432        let rule = MD038NoSpaceInCode::new();
433        let valid_cases = vec![
434            "This is `code` in a sentence.",
435            "This is a `longer code span` in a sentence.",
436            "This is `code with internal spaces` which is fine.",
437            "Code span at `end of line`",
438            "`Start of line` code span",
439            "Multiple `code spans` in `one line` are fine",
440            "Code span with `symbols: !@#$%^&*()`",
441            "Empty code span `` is technically valid",
442        ];
443        for case in valid_cases {
444            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Standard, None);
445            let result = rule.check(&ctx).unwrap();
446            assert!(result.is_empty(), "Valid case should not have warnings: {case}");
447        }
448    }
449
450    #[test]
451    fn test_md038_invalid() {
452        let rule = MD038NoSpaceInCode::new();
453        // Flag cases that violate CommonMark:
454        // - Space only at start (no matching end space)
455        // - Space only at end (no matching start space)
456        // - Multiple spaces at start or end (extra space will remain after CommonMark stripping)
457        let invalid_cases = vec![
458            // Unbalanced: only leading space
459            "This is ` code` with leading space.",
460            // Unbalanced: only trailing space
461            "This is `code ` with trailing space.",
462            // Multiple leading spaces (one will remain after CommonMark strips one)
463            "This is `  code ` with double leading space.",
464            // Multiple trailing spaces (one will remain after CommonMark strips one)
465            "This is ` code  ` with double trailing space.",
466            // Multiple spaces both sides
467            "This is `  code  ` with double spaces both sides.",
468        ];
469        for case in invalid_cases {
470            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Standard, None);
471            let result = rule.check(&ctx).unwrap();
472            assert!(!result.is_empty(), "Invalid case should have warnings: {case}");
473        }
474    }
475
476    #[test]
477    fn test_md038_valid_commonmark_stripping() {
478        let rule = MD038NoSpaceInCode::new();
479        // These cases have exactly ONE space at start AND ONE at end.
480        // CommonMark strips both, so these should NOT be flagged.
481        // See: https://spec.commonmark.org/0.31.2/#code-spans
482        let valid_cases = vec![
483            "Type ` y ` to confirm.",
484            "Use ` git commit -m \"message\" ` to commit.",
485            "The variable ` $HOME ` contains home path.",
486            "The pattern ` *.txt ` matches text files.",
487            "This is ` random word ` with unnecessary spaces.",
488            "Text with ` plain text ` is valid.",
489            "Code with ` just code ` here.",
490            "Multiple ` word ` spans with ` text ` in one line.",
491            "This is ` code ` with both leading and trailing single space.",
492            "Use ` - ` as separator.",
493        ];
494        for case in valid_cases {
495            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Standard, None);
496            let result = rule.check(&ctx).unwrap();
497            assert!(
498                result.is_empty(),
499                "Single space on each side should not be flagged (CommonMark strips them): {case}"
500            );
501        }
502    }
503
504    #[test]
505    fn test_md038_fix() {
506        let rule = MD038NoSpaceInCode::new();
507        // Only cases that violate CommonMark should be fixed
508        let test_cases = vec![
509            // Unbalanced: only leading space - should be fixed
510            (
511                "This is ` code` with leading space.",
512                "This is `code` with leading space.",
513            ),
514            // Unbalanced: only trailing space - should be fixed
515            (
516                "This is `code ` with trailing space.",
517                "This is `code` with trailing space.",
518            ),
519            // Single space on both sides - NOT fixed (valid per CommonMark)
520            (
521                "This is ` code ` with both spaces.",
522                "This is ` code ` with both spaces.", // unchanged
523            ),
524            // Double leading space - should be fixed
525            (
526                "This is `  code ` with double leading space.",
527                "This is `code` with double leading space.",
528            ),
529            // Mixed: one valid (single space both), one invalid (trailing only)
530            (
531                "Multiple ` code ` and `spans ` to fix.",
532                "Multiple ` code ` and `spans` to fix.", // only spans is fixed
533            ),
534        ];
535        for (input, expected) in test_cases {
536            let ctx = crate::lint_context::LintContext::new(input, crate::config::MarkdownFlavor::Standard, None);
537            let result = rule.fix(&ctx).unwrap();
538            assert_eq!(result, expected, "Fix did not produce expected output for: {input}");
539        }
540    }
541
542    #[test]
543    fn test_check_invalid_leading_space() {
544        let rule = MD038NoSpaceInCode::new();
545        let input = "This has a ` leading space` in code";
546        let ctx = crate::lint_context::LintContext::new(input, crate::config::MarkdownFlavor::Standard, None);
547        let result = rule.check(&ctx).unwrap();
548        assert_eq!(result.len(), 1);
549        assert_eq!(result[0].line, 1);
550        assert!(result[0].fix.is_some());
551    }
552
553    #[test]
554    fn test_code_span_parsing_nested_backticks() {
555        let content = "Code with ` nested `code` example ` should preserve backticks";
556        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Standard, None);
557
558        println!("Content: {content}");
559        println!("Code spans found:");
560        let code_spans = ctx.code_spans();
561        for (i, span) in code_spans.iter().enumerate() {
562            println!(
563                "  Span {}: line={}, col={}-{}, backticks={}, content='{}'",
564                i, span.line, span.start_col, span.end_col, span.backtick_count, span.content
565            );
566        }
567
568        // This test reveals the issue - we're getting multiple separate code spans instead of one
569        assert_eq!(code_spans.len(), 2, "Should parse as 2 code spans");
570    }
571
572    #[test]
573    fn test_nested_backtick_detection() {
574        let rule = MD038NoSpaceInCode::new();
575
576        // Test that code spans with backticks are skipped
577        let content = "Code with `` `backticks` inside `` should not be flagged";
578        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Standard, None);
579        let result = rule.check(&ctx).unwrap();
580        assert!(result.is_empty(), "Code spans with backticks should be skipped");
581    }
582
583    #[test]
584    fn test_quarto_inline_r_code() {
585        // Test that Quarto-specific R code exception works
586        let rule = MD038NoSpaceInCode::new();
587
588        // Test inline R code - should NOT trigger warning in Quarto flavor
589        // The key pattern is "r " followed by code
590        let content = r#"The result is `r nchar("test")` which equals 4."#;
591
592        // Quarto flavor should allow R code
593        let ctx_quarto = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Quarto, None);
594        let result_quarto = rule.check(&ctx_quarto).unwrap();
595        assert!(
596            result_quarto.is_empty(),
597            "Quarto inline R code should not trigger warnings. Got {} warnings",
598            result_quarto.len()
599        );
600
601        // Test that invalid code spans (not matching CommonMark stripping) still get flagged in Quarto
602        // Use only trailing space - this violates CommonMark (no balanced stripping)
603        let content_other = "This has `plain text ` with trailing space.";
604        let ctx_other =
605            crate::lint_context::LintContext::new(content_other, crate::config::MarkdownFlavor::Quarto, None);
606        let result_other = rule.check(&ctx_other).unwrap();
607        assert_eq!(
608            result_other.len(),
609            1,
610            "Quarto should still flag non-R code spans with improper spaces"
611        );
612    }
613
614    /// Comprehensive tests for Hugo template syntax detection
615    ///
616    /// These tests ensure MD038 correctly handles Hugo template syntax patterns
617    /// without false positives, while maintaining correct detection of actual
618    /// code span spacing issues.
619    #[test]
620    fn test_hugo_template_syntax_comprehensive() {
621        let rule = MD038NoSpaceInCode::new();
622
623        // ===== VALID HUGO TEMPLATE SYNTAX (Should NOT trigger warnings) =====
624
625        // Basic Hugo shortcode patterns
626        let valid_hugo_cases = vec![
627            // Raw HTML shortcode
628            (
629                "{{raw `\n\tgo list -f '{{.DefaultGODEBUG}}' my/main/package\n`}}",
630                "Multi-line raw shortcode",
631            ),
632            (
633                "Some text {{raw ` code `}} more text",
634                "Inline raw shortcode with spaces",
635            ),
636            ("{{raw `code`}}", "Raw shortcode without spaces"),
637            // Partial shortcode
638            ("{{< ` code ` >}}", "Partial shortcode with spaces"),
639            ("{{< `code` >}}", "Partial shortcode without spaces"),
640            // Shortcode with percent
641            ("{{% ` code ` %}}", "Percent shortcode with spaces"),
642            ("{{% `code` %}}", "Percent shortcode without spaces"),
643            // Generic shortcode
644            ("{{ ` code ` }}", "Generic shortcode with spaces"),
645            ("{{ `code` }}", "Generic shortcode without spaces"),
646            // Shortcodes with parameters (common Hugo pattern)
647            ("{{< highlight go `code` >}}", "Shortcode with highlight parameter"),
648            ("{{< code `go list` >}}", "Shortcode with code parameter"),
649            // Multi-line Hugo templates
650            ("{{raw `\n\tcommand here\n\tmore code\n`}}", "Multi-line raw template"),
651            ("{{< highlight `\ncode here\n` >}}", "Multi-line highlight template"),
652            // Hugo templates with nested Go template syntax
653            (
654                "{{raw `\n\t{{.Variable}}\n\t{{range .Items}}\n`}}",
655                "Nested Go template syntax",
656            ),
657            // Edge case: Hugo template at start of line
658            ("{{raw `code`}}", "Hugo template at line start"),
659            // Edge case: Hugo template at end of line
660            ("Text {{raw `code`}}", "Hugo template at end of line"),
661            // Edge case: Multiple Hugo templates
662            ("{{raw `code1`}} and {{raw `code2`}}", "Multiple Hugo templates"),
663        ];
664
665        for (case, description) in valid_hugo_cases {
666            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Standard, None);
667            let result = rule.check(&ctx).unwrap();
668            assert!(
669                result.is_empty(),
670                "Hugo template syntax should not trigger MD038 warnings: {description} - {case}"
671            );
672        }
673
674        // ===== FALSE POSITIVE PREVENTION (Non-Hugo asymmetric spaces should be flagged) =====
675
676        // These have asymmetric spaces (leading-only or trailing-only) and should be flagged
677        // Per CommonMark spec: symmetric single-space pairs are stripped and NOT flagged
678        let should_be_flagged = vec![
679            ("This is ` code` with leading space.", "Leading space only"),
680            ("This is `code ` with trailing space.", "Trailing space only"),
681            ("Text `  code ` here", "Extra leading space (asymmetric)"),
682            ("Text ` code  ` here", "Extra trailing space (asymmetric)"),
683            ("Text `  code` here", "Double leading, no trailing"),
684            ("Text `code  ` here", "No leading, double trailing"),
685        ];
686
687        for (case, description) in should_be_flagged {
688            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Standard, None);
689            let result = rule.check(&ctx).unwrap();
690            assert!(
691                !result.is_empty(),
692                "Should flag asymmetric space code spans: {description} - {case}"
693            );
694        }
695
696        // ===== COMMONMARK SYMMETRIC SPACE BEHAVIOR (Should NOT be flagged) =====
697
698        // Per CommonMark 0.31.2: When a code span has exactly one space at start AND end,
699        // those spaces are stripped from the output. This is intentional, not an error.
700        // These cases should NOT trigger MD038.
701        let symmetric_single_space = vec![
702            ("Text ` code ` here", "Symmetric single space - CommonMark strips"),
703            ("{raw ` code `}", "Looks like Hugo but missing opening {{"),
704            ("raw ` code `}}", "Missing opening {{ - but symmetric spaces"),
705        ];
706
707        for (case, description) in symmetric_single_space {
708            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Standard, None);
709            let result = rule.check(&ctx).unwrap();
710            assert!(
711                result.is_empty(),
712                "CommonMark symmetric spaces should NOT be flagged: {description} - {case}"
713            );
714        }
715
716        // ===== EDGE CASES: Unicode and Special Characters =====
717
718        let unicode_cases = vec![
719            ("{{raw `\n\t你好世界\n`}}", "Unicode in Hugo template"),
720            ("{{raw `\n\t🎉 emoji\n`}}", "Emoji in Hugo template"),
721            ("{{raw `\n\tcode with \"quotes\"\n`}}", "Quotes in Hugo template"),
722            (
723                "{{raw `\n\tcode with 'single quotes'\n`}}",
724                "Single quotes in Hugo template",
725            ),
726        ];
727
728        for (case, description) in unicode_cases {
729            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Standard, None);
730            let result = rule.check(&ctx).unwrap();
731            assert!(
732                result.is_empty(),
733                "Hugo templates with special characters should not trigger warnings: {description} - {case}"
734            );
735        }
736
737        // ===== BOUNDARY CONDITIONS =====
738
739        // Minimum valid Hugo pattern
740        assert!(
741            rule.check(&crate::lint_context::LintContext::new(
742                "{{ ` ` }}",
743                crate::config::MarkdownFlavor::Standard,
744                None
745            ))
746            .unwrap()
747            .is_empty(),
748            "Minimum Hugo pattern should be valid"
749        );
750
751        // Hugo template with only whitespace
752        assert!(
753            rule.check(&crate::lint_context::LintContext::new(
754                "{{raw `\n\t\n`}}",
755                crate::config::MarkdownFlavor::Standard,
756                None
757            ))
758            .unwrap()
759            .is_empty(),
760            "Hugo template with only whitespace should be valid"
761        );
762    }
763
764    /// Test interaction with other markdown elements
765    #[test]
766    fn test_hugo_template_with_other_markdown() {
767        let rule = MD038NoSpaceInCode::new();
768
769        // Hugo template inside a list
770        let content = r#"1. First item
7712. Second item with {{raw `code`}} template
7723. Third item"#;
773        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Standard, None);
774        let result = rule.check(&ctx).unwrap();
775        assert!(result.is_empty(), "Hugo template in list should not trigger warnings");
776
777        // Hugo template in blockquote
778        let content = r#"> Quote with {{raw `code`}} template"#;
779        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Standard, None);
780        let result = rule.check(&ctx).unwrap();
781        assert!(
782            result.is_empty(),
783            "Hugo template in blockquote should not trigger warnings"
784        );
785
786        // Hugo template near regular code span (should flag the regular one)
787        let content = r#"{{raw `code`}} and ` bad code` here"#;
788        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Standard, None);
789        let result = rule.check(&ctx).unwrap();
790        assert_eq!(result.len(), 1, "Should flag regular code span but not Hugo template");
791    }
792
793    /// Performance test: Many Hugo templates
794    #[test]
795    fn test_hugo_template_performance() {
796        let rule = MD038NoSpaceInCode::new();
797
798        // Create content with many Hugo templates
799        let mut content = String::new();
800        for i in 0..100 {
801            content.push_str(&format!("{{{{raw `code{i}\n`}}}}\n"));
802        }
803
804        let ctx = crate::lint_context::LintContext::new(&content, crate::config::MarkdownFlavor::Standard, None);
805        let start = std::time::Instant::now();
806        let result = rule.check(&ctx).unwrap();
807        let duration = start.elapsed();
808
809        assert!(result.is_empty(), "Many Hugo templates should not trigger warnings");
810        assert!(
811            duration.as_millis() < 1000,
812            "Performance test: Should process 100 Hugo templates in <1s, took {duration:?}"
813        );
814    }
815
816    #[test]
817    fn test_mkdocs_inline_hilite_not_flagged() {
818        // InlineHilite syntax: `#!language code` should NOT be flagged
819        // The space after the language specifier is legitimate
820        let rule = MD038NoSpaceInCode::new();
821
822        let valid_cases = vec![
823            "`#!python print('hello')`",
824            "`#!js alert('hi')`",
825            "`#!c++ cout << x;`",
826            "Use `#!python import os` to import modules",
827            "`#!bash echo $HOME`",
828        ];
829
830        for case in valid_cases {
831            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::MkDocs, None);
832            let result = rule.check(&ctx).unwrap();
833            assert!(
834                result.is_empty(),
835                "InlineHilite syntax should not be flagged in MkDocs: {case}"
836            );
837        }
838
839        // Test that InlineHilite IS flagged in Standard flavor (not MkDocs-aware)
840        let content = "`#!python print('hello')`";
841        let ctx_standard =
842            crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Standard, None);
843        let result_standard = rule.check(&ctx_standard).unwrap();
844        // In standard flavor, the content " print('hello')" has no special meaning
845        // But since "#!python print('hello')" doesn't have leading/trailing spaces, it's valid!
846        assert!(
847            result_standard.is_empty(),
848            "InlineHilite with no extra spaces should not be flagged even in Standard flavor"
849        );
850    }
851
852    #[test]
853    fn test_multibyte_utf8_no_panic() {
854        // Regression test: ensure multi-byte UTF-8 characters don't cause panics
855        // when checking for nested backticks between code spans.
856        // These are real examples from the-art-of-command-line translations.
857        let rule = MD038NoSpaceInCode::new();
858
859        // Greek text with code spans
860        let greek = "- Χρήσιμα εργαλεία της γραμμής εντολών είναι τα `ping`,` ipconfig`, `traceroute` και `netstat`.";
861        let ctx = crate::lint_context::LintContext::new(greek, crate::config::MarkdownFlavor::Standard, None);
862        let result = rule.check(&ctx);
863        assert!(result.is_ok(), "Greek text should not panic");
864
865        // Chinese text with code spans
866        let chinese = "- 當你需要對文字檔案做集合交、並、差運算時,`sort`/`uniq` 很有幫助。";
867        let ctx = crate::lint_context::LintContext::new(chinese, crate::config::MarkdownFlavor::Standard, None);
868        let result = rule.check(&ctx);
869        assert!(result.is_ok(), "Chinese text should not panic");
870
871        // Cyrillic/Ukrainian text with code spans
872        let cyrillic = "- Основи роботи з файлами: `ls` і `ls -l`, `less`, `head`,` tail` і `tail -f`.";
873        let ctx = crate::lint_context::LintContext::new(cyrillic, crate::config::MarkdownFlavor::Standard, None);
874        let result = rule.check(&ctx);
875        assert!(result.is_ok(), "Cyrillic text should not panic");
876
877        // Mixed multi-byte with multiple code spans on same line
878        let mixed = "使用 `git` 命令和 `npm` 工具来管理项目,可以用 `docker` 容器化。";
879        let ctx = crate::lint_context::LintContext::new(mixed, crate::config::MarkdownFlavor::Standard, None);
880        let result = rule.check(&ctx);
881        assert!(
882            result.is_ok(),
883            "Mixed Chinese text with multiple code spans should not panic"
884        );
885    }
886
887    // ==================== Obsidian Dataview Plugin Tests ====================
888
889    /// Test that Dataview inline DQL expressions are not flagged in Obsidian flavor
890    #[test]
891    fn test_obsidian_dataview_inline_dql_not_flagged() {
892        let rule = MD038NoSpaceInCode::new();
893
894        // Basic inline DQL expressions - should NOT be flagged in Obsidian
895        let valid_dql_cases = vec![
896            "`= this.file.name`",
897            "`= date(today)`",
898            "`= [[Page]].field`",
899            "`= choice(condition, \"yes\", \"no\")`",
900            "`= this.file.mtime`",
901            "`= this.file.ctime`",
902            "`= this.file.path`",
903            "`= this.file.folder`",
904            "`= this.file.size`",
905            "`= this.file.ext`",
906            "`= this.file.link`",
907            "`= this.file.outlinks`",
908            "`= this.file.inlinks`",
909            "`= this.file.tags`",
910        ];
911
912        for case in valid_dql_cases {
913            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Obsidian, None);
914            let result = rule.check(&ctx).unwrap();
915            assert!(
916                result.is_empty(),
917                "Dataview DQL expression should not be flagged in Obsidian: {case}"
918            );
919        }
920    }
921
922    /// Test that Dataview inline DataviewJS expressions are not flagged in Obsidian flavor
923    #[test]
924    fn test_obsidian_dataview_inline_dvjs_not_flagged() {
925        let rule = MD038NoSpaceInCode::new();
926
927        // Inline DataviewJS expressions - should NOT be flagged in Obsidian
928        let valid_dvjs_cases = vec![
929            "`$= dv.current().file.mtime`",
930            "`$= dv.pages().length`",
931            "`$= dv.current()`",
932            "`$= dv.pages('#tag').length`",
933            "`$= dv.pages('\"folder\"').length`",
934            "`$= dv.current().file.name`",
935            "`$= dv.current().file.path`",
936            "`$= dv.current().file.folder`",
937            "`$= dv.current().file.link`",
938        ];
939
940        for case in valid_dvjs_cases {
941            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Obsidian, None);
942            let result = rule.check(&ctx).unwrap();
943            assert!(
944                result.is_empty(),
945                "Dataview JS expression should not be flagged in Obsidian: {case}"
946            );
947        }
948    }
949
950    /// Test complex Dataview expressions with nested parentheses
951    #[test]
952    fn test_obsidian_dataview_complex_expressions() {
953        let rule = MD038NoSpaceInCode::new();
954
955        let complex_cases = vec![
956            // Nested function calls
957            "`= sum(filter(pages, (p) => p.done))`",
958            "`= length(filter(file.tags, (t) => startswith(t, \"project\")))`",
959            // choice() function
960            "`= choice(x > 5, \"big\", \"small\")`",
961            "`= choice(this.status = \"done\", \"✅\", \"⏳\")`",
962            // date functions
963            "`= date(today) - dur(7 days)`",
964            "`= dateformat(this.file.mtime, \"yyyy-MM-dd\")`",
965            // Math expressions
966            "`= sum(rows.amount)`",
967            "`= round(average(rows.score), 2)`",
968            "`= min(rows.priority)`",
969            "`= max(rows.priority)`",
970            // String operations
971            "`= join(this.file.tags, \", \")`",
972            "`= replace(this.title, \"-\", \" \")`",
973            "`= lower(this.file.name)`",
974            "`= upper(this.file.name)`",
975            // List operations
976            "`= length(this.file.outlinks)`",
977            "`= contains(this.file.tags, \"important\")`",
978            // Link references
979            "`= [[Page Name]].field`",
980            "`= [[Folder/Subfolder/Page]].nested.field`",
981            // Conditional expressions
982            "`= default(this.status, \"unknown\")`",
983            "`= coalesce(this.priority, this.importance, 0)`",
984        ];
985
986        for case in complex_cases {
987            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Obsidian, None);
988            let result = rule.check(&ctx).unwrap();
989            assert!(
990                result.is_empty(),
991                "Complex Dataview expression should not be flagged in Obsidian: {case}"
992            );
993        }
994    }
995
996    /// Test that complex DataviewJS expressions with method chains are not flagged
997    #[test]
998    fn test_obsidian_dataviewjs_method_chains() {
999        let rule = MD038NoSpaceInCode::new();
1000
1001        let method_chain_cases = vec![
1002            "`$= dv.pages().where(p => p.status).length`",
1003            "`$= dv.pages('#project').where(p => !p.done).length`",
1004            "`$= dv.pages().filter(p => p.file.day).sort(p => p.file.mtime, 'desc').limit(5)`",
1005            "`$= dv.pages('\"folder\"').map(p => p.file.link).join(', ')`",
1006            "`$= dv.current().file.tasks.where(t => !t.completed).length`",
1007            "`$= dv.pages().flatMap(p => p.file.tags).distinct().sort()`",
1008            "`$= dv.page('Index').children.map(p => p.title)`",
1009            "`$= dv.pages().groupBy(p => p.status).map(g => [g.key, g.rows.length])`",
1010        ];
1011
1012        for case in method_chain_cases {
1013            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Obsidian, None);
1014            let result = rule.check(&ctx).unwrap();
1015            assert!(
1016                result.is_empty(),
1017                "DataviewJS method chain should not be flagged in Obsidian: {case}"
1018            );
1019        }
1020    }
1021
1022    /// Test Dataview-like patterns in Standard flavor
1023    ///
1024    /// Note: The actual content `= this.file.name` starts with `=`, not whitespace,
1025    /// so it doesn't have a leading space issue. Dataview expressions only become
1026    /// relevant when their content would otherwise be flagged.
1027    ///
1028    /// To properly test the difference, we need patterns that have leading whitespace
1029    /// issues that would be skipped in Obsidian but flagged in Standard.
1030    #[test]
1031    fn test_standard_flavor_vs_obsidian_dataview() {
1032        let rule = MD038NoSpaceInCode::new();
1033
1034        // These Dataview expressions don't have leading whitespace (they start with "=")
1035        // so they wouldn't be flagged in ANY flavor
1036        let no_issue_cases = vec!["`= this.file.name`", "`$= dv.current()`"];
1037
1038        for case in no_issue_cases {
1039            // Standard flavor - no issue because content doesn't start with whitespace
1040            let ctx_std = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Standard, None);
1041            let result_std = rule.check(&ctx_std).unwrap();
1042            assert!(
1043                result_std.is_empty(),
1044                "Dataview expression without leading space shouldn't be flagged in Standard: {case}"
1045            );
1046
1047            // Obsidian flavor - also no issue
1048            let ctx_obs = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Obsidian, None);
1049            let result_obs = rule.check(&ctx_obs).unwrap();
1050            assert!(
1051                result_obs.is_empty(),
1052                "Dataview expression shouldn't be flagged in Obsidian: {case}"
1053            );
1054        }
1055
1056        // Test that regular code with leading/trailing spaces is still flagged in both flavors
1057        // (when not matching Dataview pattern)
1058        let space_issues = vec![
1059            "` code`", // Leading space, no trailing
1060            "`code `", // Trailing space, no leading
1061        ];
1062
1063        for case in space_issues {
1064            // Standard flavor - should be flagged
1065            let ctx_std = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Standard, None);
1066            let result_std = rule.check(&ctx_std).unwrap();
1067            assert!(
1068                !result_std.is_empty(),
1069                "Code with spacing issue should be flagged in Standard: {case}"
1070            );
1071
1072            // Obsidian flavor - should also be flagged (not a Dataview pattern)
1073            let ctx_obs = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Obsidian, None);
1074            let result_obs = rule.check(&ctx_obs).unwrap();
1075            assert!(
1076                !result_obs.is_empty(),
1077                "Code with spacing issue should be flagged in Obsidian (not Dataview): {case}"
1078            );
1079        }
1080    }
1081
1082    /// Test that regular code spans with leading space are still flagged in Obsidian
1083    #[test]
1084    fn test_obsidian_still_flags_regular_code_spans_with_space() {
1085        let rule = MD038NoSpaceInCode::new();
1086
1087        // These are NOT Dataview expressions, just regular code spans with leading space
1088        // They should still be flagged even in Obsidian flavor
1089        let invalid_cases = [
1090            "` regular code`", // Space at start, not Dataview
1091            "`code `",         // Space at end
1092            "` code `",        // This is valid per CommonMark (symmetric single space)
1093            "`  code`",        // Double space at start (not Dataview pattern)
1094        ];
1095
1096        // Only the asymmetric cases should be flagged
1097        let expected_flags = [
1098            true,  // ` regular code` - leading space, no trailing
1099            true,  // `code ` - trailing space, no leading
1100            false, // ` code ` - symmetric single space (CommonMark valid)
1101            true,  // `  code` - double leading space
1102        ];
1103
1104        for (case, should_flag) in invalid_cases.iter().zip(expected_flags.iter()) {
1105            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Obsidian, None);
1106            let result = rule.check(&ctx).unwrap();
1107            if *should_flag {
1108                assert!(
1109                    !result.is_empty(),
1110                    "Non-Dataview code span with spacing issue should be flagged in Obsidian: {case}"
1111                );
1112            } else {
1113                assert!(
1114                    result.is_empty(),
1115                    "CommonMark-valid symmetric spacing should not be flagged: {case}"
1116                );
1117            }
1118        }
1119    }
1120
1121    /// Test edge cases for Dataview pattern detection
1122    #[test]
1123    fn test_obsidian_dataview_edge_cases() {
1124        let rule = MD038NoSpaceInCode::new();
1125
1126        // Valid Dataview patterns
1127        let valid_cases = vec![
1128            ("`= x`", true),                         // Minimal DQL
1129            ("`$= x`", true),                        // Minimal DVJS
1130            ("`= `", true),                          // Just equals-space (empty expression)
1131            ("`$= `", true),                         // Just dollar-equals-space (empty expression)
1132            ("`=x`", false),                         // No space after = (not Dataview, and no leading whitespace issue)
1133            ("`$=x`", false),       // No space after $= (not Dataview, and no leading whitespace issue)
1134            ("`= [[Link]]`", true), // Link in expression
1135            ("`= this`", true),     // Simple this reference
1136            ("`$= dv`", true),      // Just dv object reference
1137            ("`= 1 + 2`", true),    // Math expression
1138            ("`$= 1 + 2`", true),   // Math in DVJS
1139            ("`= \"string\"`", true), // String literal
1140            ("`$= 'string'`", true), // Single-quoted string
1141            ("`= this.field ?? \"default\"`", true), // Null coalescing
1142            ("`$= dv?.pages()`", true), // Optional chaining
1143        ];
1144
1145        for (case, should_be_valid) in valid_cases {
1146            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Obsidian, None);
1147            let result = rule.check(&ctx).unwrap();
1148            if should_be_valid {
1149                assert!(
1150                    result.is_empty(),
1151                    "Valid Dataview expression should not be flagged: {case}"
1152                );
1153            } else {
1154                // These might or might not be flagged depending on other MD038 rules
1155                // We just verify they don't crash
1156                let _ = result;
1157            }
1158        }
1159    }
1160
1161    /// Test Dataview expressions in context (mixed with regular markdown)
1162    #[test]
1163    fn test_obsidian_dataview_in_context() {
1164        let rule = MD038NoSpaceInCode::new();
1165
1166        // Document with mixed Dataview and regular code spans
1167        let content = r#"# My Note
1168
1169The file name is `= this.file.name` and it was created on `= this.file.ctime`.
1170
1171Regular code: `println!("hello")` and `let x = 5;`
1172
1173DataviewJS count: `$= dv.pages('#project').length` projects found.
1174
1175More regular code with issue: ` bad code` should be flagged.
1176"#;
1177
1178        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Obsidian, None);
1179        let result = rule.check(&ctx).unwrap();
1180
1181        // Should only flag ` bad code` (line 9)
1182        assert_eq!(
1183            result.len(),
1184            1,
1185            "Should only flag the regular code span with leading space, not Dataview expressions"
1186        );
1187        assert_eq!(result[0].line, 9, "Warning should be on line 9");
1188    }
1189
1190    /// Test that Dataview expressions in code blocks are properly handled
1191    #[test]
1192    fn test_obsidian_dataview_in_code_blocks() {
1193        let rule = MD038NoSpaceInCode::new();
1194
1195        // Dataview expressions inside fenced code blocks should be ignored
1196        // (because they're inside code blocks, not because of Dataview logic)
1197        let content = r#"# Example
1198
1199```
1200`= this.file.name`
1201`$= dv.current()`
1202```
1203
1204Regular paragraph with `= this.file.name` Dataview.
1205"#;
1206
1207        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Obsidian, None);
1208        let result = rule.check(&ctx).unwrap();
1209
1210        // Should not flag anything - code blocks are skipped, and inline Dataview is valid
1211        assert!(
1212            result.is_empty(),
1213            "Dataview in code blocks should be ignored, inline Dataview should be valid"
1214        );
1215    }
1216
1217    /// Test Dataview with Unicode content
1218    #[test]
1219    fn test_obsidian_dataview_unicode() {
1220        let rule = MD038NoSpaceInCode::new();
1221
1222        let unicode_cases = vec![
1223            "`= this.日本語`",                  // Japanese field name
1224            "`= this.中文字段`",                // Chinese field name
1225            "`= \"Привет мир\"`",               // Russian string
1226            "`$= dv.pages('#日本語タグ')`",     // Japanese tag
1227            "`= choice(true, \"✅\", \"❌\")`", // Emoji in strings
1228            "`= this.file.name + \" 📝\"`",     // Emoji concatenation
1229        ];
1230
1231        for case in unicode_cases {
1232            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Obsidian, None);
1233            let result = rule.check(&ctx).unwrap();
1234            assert!(
1235                result.is_empty(),
1236                "Unicode Dataview expression should not be flagged: {case}"
1237            );
1238        }
1239    }
1240
1241    /// Test that Dataview detection doesn't break regular equals patterns
1242    #[test]
1243    fn test_obsidian_regular_equals_still_works() {
1244        let rule = MD038NoSpaceInCode::new();
1245
1246        // Regular code with equals signs should still work normally
1247        let valid_regular_cases = vec![
1248            "`x = 5`",       // Assignment (no leading space)
1249            "`a == b`",      // Equality check
1250            "`x >= 10`",     // Comparison
1251            "`let x = 10`",  // Variable declaration
1252            "`const y = 5`", // Const declaration
1253        ];
1254
1255        for case in valid_regular_cases {
1256            let ctx = crate::lint_context::LintContext::new(case, crate::config::MarkdownFlavor::Obsidian, None);
1257            let result = rule.check(&ctx).unwrap();
1258            assert!(
1259                result.is_empty(),
1260                "Regular code with equals should not be flagged: {case}"
1261            );
1262        }
1263    }
1264
1265    /// Test fix behavior doesn't break Dataview expressions
1266    #[test]
1267    fn test_obsidian_dataview_fix_preserves_expressions() {
1268        let rule = MD038NoSpaceInCode::new();
1269
1270        // Content with Dataview expressions and one fixable issue
1271        let content = "Dataview: `= this.file.name` and bad: ` fixme`";
1272        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Obsidian, None);
1273        let fixed = rule.fix(&ctx).unwrap();
1274
1275        // Should fix ` fixme` but preserve `= this.file.name`
1276        assert!(
1277            fixed.contains("`= this.file.name`"),
1278            "Dataview expression should be preserved after fix"
1279        );
1280        assert!(
1281            fixed.contains("`fixme`"),
1282            "Regular code span should be fixed (space removed)"
1283        );
1284        assert!(!fixed.contains("` fixme`"), "Bad code span should have been fixed");
1285    }
1286
1287    /// Test multiple Dataview expressions on same line
1288    #[test]
1289    fn test_obsidian_multiple_dataview_same_line() {
1290        let rule = MD038NoSpaceInCode::new();
1291
1292        let content = "Created: `= this.file.ctime` | Modified: `= this.file.mtime` | Count: `$= dv.pages().length`";
1293        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Obsidian, None);
1294        let result = rule.check(&ctx).unwrap();
1295
1296        assert!(
1297            result.is_empty(),
1298            "Multiple Dataview expressions on same line should all be valid"
1299        );
1300    }
1301
1302    /// Performance test: Many Dataview expressions
1303    #[test]
1304    fn test_obsidian_dataview_performance() {
1305        let rule = MD038NoSpaceInCode::new();
1306
1307        // Create content with many Dataview expressions
1308        let mut content = String::new();
1309        for i in 0..100 {
1310            content.push_str(&format!("Field {i}: `= this.field{i}` | JS: `$= dv.current().f{i}`\n"));
1311        }
1312
1313        let ctx = crate::lint_context::LintContext::new(&content, crate::config::MarkdownFlavor::Obsidian, None);
1314        let start = std::time::Instant::now();
1315        let result = rule.check(&ctx).unwrap();
1316        let duration = start.elapsed();
1317
1318        assert!(result.is_empty(), "All Dataview expressions should be valid");
1319        assert!(
1320            duration.as_millis() < 1000,
1321            "Performance test: Should process 200 Dataview expressions in <1s, took {duration:?}"
1322        );
1323    }
1324
1325    /// Test is_dataview_expression helper function directly
1326    #[test]
1327    fn test_is_dataview_expression_helper() {
1328        // Valid Dataview patterns
1329        assert!(MD038NoSpaceInCode::is_dataview_expression("= this.file.name"));
1330        assert!(MD038NoSpaceInCode::is_dataview_expression("= "));
1331        assert!(MD038NoSpaceInCode::is_dataview_expression("$= dv.current()"));
1332        assert!(MD038NoSpaceInCode::is_dataview_expression("$= "));
1333        assert!(MD038NoSpaceInCode::is_dataview_expression("= x"));
1334        assert!(MD038NoSpaceInCode::is_dataview_expression("$= x"));
1335
1336        // Invalid Dataview patterns
1337        assert!(!MD038NoSpaceInCode::is_dataview_expression("=")); // No space after =
1338        assert!(!MD038NoSpaceInCode::is_dataview_expression("$=")); // No space after $=
1339        assert!(!MD038NoSpaceInCode::is_dataview_expression("=x")); // No space
1340        assert!(!MD038NoSpaceInCode::is_dataview_expression("$=x")); // No space
1341        assert!(!MD038NoSpaceInCode::is_dataview_expression(" = x")); // Leading space before =
1342        assert!(!MD038NoSpaceInCode::is_dataview_expression("x = 5")); // Assignment, not Dataview
1343        assert!(!MD038NoSpaceInCode::is_dataview_expression("== x")); // Double equals
1344        assert!(!MD038NoSpaceInCode::is_dataview_expression("")); // Empty
1345        assert!(!MD038NoSpaceInCode::is_dataview_expression("regular")); // Regular text
1346    }
1347
1348    /// Test Dataview expressions work alongside other Obsidian features (tags)
1349    #[test]
1350    fn test_obsidian_dataview_with_tags() {
1351        let rule = MD038NoSpaceInCode::new();
1352
1353        // Document using both Dataview and Obsidian tags
1354        let content = r#"# Project Status
1355
1356Tags: #project #active
1357
1358Status: `= this.status`
1359Count: `$= dv.pages('#project').length`
1360
1361Regular code: `function test() {}`
1362"#;
1363
1364        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Obsidian, None);
1365        let result = rule.check(&ctx).unwrap();
1366
1367        // Nothing should be flagged
1368        assert!(
1369            result.is_empty(),
1370            "Dataview expressions and regular code should work together"
1371        );
1372    }
1373
1374    #[test]
1375    fn test_unicode_between_code_spans_no_panic() {
1376        // Verify that multi-byte characters between code spans do not cause panics
1377        // or incorrect slicing in the nested-backtick detection logic.
1378        let rule = MD038NoSpaceInCode::new();
1379
1380        // Multi-byte character (U-umlaut = 2 bytes) between two code spans
1381        let content = "Use `one` \u{00DC}nited `two` for backtick examples.";
1382        let ctx = crate::lint_context::LintContext::new(content, crate::config::MarkdownFlavor::Standard, None);
1383        let result = rule.check(&ctx);
1384        // Should not panic; any warnings or lack thereof are acceptable
1385        assert!(result.is_ok(), "Should not panic with Unicode between code spans");
1386
1387        // CJK characters (3 bytes each) between code spans
1388        let content_cjk = "Use `one` \u{4E16}\u{754C} `two` for examples.";
1389        let ctx_cjk = crate::lint_context::LintContext::new(content_cjk, crate::config::MarkdownFlavor::Standard, None);
1390        let result_cjk = rule.check(&ctx_cjk);
1391        assert!(result_cjk.is_ok(), "Should not panic with CJK between code spans");
1392    }
1393}