rumdl_lib/rules/
md053_link_image_reference_definitions.rs

1use crate::rule::{LintError, LintResult, LintWarning, Rule, Severity};
2use crate::rule_config_serde::RuleConfig;
3use crate::utils::range_utils::calculate_line_range;
4use fancy_regex::Regex as FancyRegex;
5use lazy_static::lazy_static;
6use regex::Regex;
7use serde::{Deserialize, Serialize};
8use std::collections::{HashMap, HashSet};
9
10lazy_static! {
11    // Link reference format: [text][reference]
12    // REMOVED: static ref LINK_REFERENCE_REGEX: FancyRegex = FancyRegex::new(r"\[([^\]]*)\]\s*\[([^\]]*)\]").unwrap();
13
14    // Image reference format: ![text][reference]
15    // REMOVED: static ref IMAGE_REFERENCE_REGEX: FancyRegex = FancyRegex::new(r"!\[([^\]]*)\]\s*\[([^\]]*)\]").unwrap();
16
17    // Shortcut reference links: [reference] - must not be followed by another bracket
18    // Allow references followed by punctuation like colon, period, comma (e.g., "[reference]:", "[reference].")
19    // Don't exclude references followed by ": " in the middle of a line (only at start of line)
20    static ref SHORTCUT_REFERENCE_REGEX: FancyRegex =
21        FancyRegex::new(r"(?<!\!)\[([^\]]+)\](?!\[)").unwrap();
22
23    // REMOVED: Empty reference links: [text][] or ![text][]
24    // static ref EMPTY_LINK_REFERENCE_REGEX: Regex = Regex::new(r"\[([^\]]+)\]\s*\[\s*\]").unwrap();
25    // static ref EMPTY_IMAGE_REFERENCE_REGEX: Regex = Regex::new(r"!\[([^\]]+)\]\s*\[\s*\]").unwrap();
26
27    // Link/image reference definition format: [reference]: URL
28    static ref REFERENCE_DEFINITION_REGEX: Regex =
29        Regex::new(r"^\s*\[([^\]]+)\]:\s+(.+)$").unwrap();
30
31    // Multi-line reference definition continuation pattern
32    static ref CONTINUATION_REGEX: Regex = Regex::new(r"^\s+(.+)$").unwrap();
33
34    // Code block regex - support indented code blocks for MkDocs tabs
35    static ref CODE_BLOCK_START_REGEX: Regex = Regex::new(r"^(\s*)(`{3,}|~{3,})").unwrap();
36    static ref CODE_BLOCK_END_REGEX: Regex = Regex::new(r"^(\s*)(`{3,}|~{3,})\s*$").unwrap();
37}
38
39/// Configuration for MD053 rule
40#[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]
41#[serde(rename_all = "kebab-case")]
42pub struct MD053Config {
43    /// List of reference names to keep even if unused
44    #[serde(default = "default_ignored_definitions")]
45    pub ignored_definitions: Vec<String>,
46}
47
48impl Default for MD053Config {
49    fn default() -> Self {
50        Self {
51            ignored_definitions: default_ignored_definitions(),
52        }
53    }
54}
55
56fn default_ignored_definitions() -> Vec<String> {
57    Vec::new()
58}
59
60impl RuleConfig for MD053Config {
61    const RULE_NAME: &'static str = "MD053";
62}
63
64/// Rule MD053: Link and image reference definitions should be used
65///
66/// See [docs/md053.md](../../docs/md053.md) for full documentation, configuration, and examples.
67///
68/// This rule is triggered when a link or image reference definition is declared but not used
69/// anywhere in the document. Unused reference definitions can create confusion and clutter.
70///
71/// ## Supported Reference Formats
72///
73/// This rule handles the following reference formats:
74///
75/// - **Full reference links/images**: `[text][reference]` or `![text][reference]`
76/// - **Collapsed reference links/images**: `[text][]` or `![text][]`
77/// - **Shortcut reference links**: `[reference]` (must be defined elsewhere)
78/// - **Reference definitions**: `[reference]: URL "Optional Title"`
79/// - **Multi-line reference definitions**:
80///   ```markdown
81///   [reference]: URL
82///      "Optional title continued on next line"
83///   ```
84///
85/// ## Configuration Options
86///
87/// The rule supports the following configuration options:
88///
89/// ```yaml
90/// MD053:
91///   ignored_definitions: []  # List of reference definitions to ignore (never report as unused)
92/// ```
93///
94/// ## Performance Optimizations
95///
96/// This rule implements various performance optimizations for handling large documents:
97///
98/// 1. **Caching**: The rule caches parsed definitions and references based on content hashing
99/// 2. **Efficient Reference Matching**: Uses HashMaps for O(1) lookups of definitions
100/// 3. **Smart Code Block Handling**: Efficiently skips references inside code blocks/spans
101/// 4. **Lazy Evaluation**: Only processes necessary portions of the document
102///
103/// ## Edge Cases Handled
104///
105/// - **Case insensitivity**: References are matched case-insensitively
106/// - **Escaped characters**: Properly processes escaped characters in references
107/// - **Unicode support**: Handles non-ASCII characters in references and URLs
108/// - **Code blocks**: Ignores references inside code blocks and spans
109/// - **Special characters**: Properly handles references with special characters
110///
111/// ## Fix Behavior
112///
113/// This rule does not provide automatic fixes. Unused references must be manually reviewed
114/// and removed, as they may be intentionally kept for future use or as templates.
115#[derive(Clone)]
116pub struct MD053LinkImageReferenceDefinitions {
117    config: MD053Config,
118}
119
120impl MD053LinkImageReferenceDefinitions {
121    /// Create a new instance of the MD053 rule
122    pub fn new() -> Self {
123        Self {
124            config: MD053Config::default(),
125        }
126    }
127
128    /// Create a new instance with the given configuration
129    pub fn from_config_struct(config: MD053Config) -> Self {
130        Self { config }
131    }
132
133    /// Check if a pattern is likely NOT a markdown reference
134    /// Returns true if this pattern should be skipped
135    fn is_likely_not_reference(text: &str) -> bool {
136        // Don't skip pure numeric patterns - they could be footnote references like [1]
137        // Only skip numeric ranges like [1:3], [0:10], etc.
138        if text.contains(':') && text.chars().all(|c| c.is_ascii_digit() || c == ':') {
139            return true;
140        }
141
142        // Skip glob/wildcard patterns like [*], [...], [**]
143        if text == "*" || text == "..." || text == "**" {
144            return true;
145        }
146
147        // Skip patterns that are just punctuation or operators
148        if text.chars().all(|c| !c.is_alphanumeric() && c != ' ') {
149            return true;
150        }
151
152        // Skip very short non-word patterns (likely operators or syntax)
153        // But allow single digits (could be footnotes) and single letters
154        if text.len() <= 2 && !text.chars().all(|c| c.is_alphanumeric()) {
155            return true;
156        }
157
158        // Skip descriptive patterns with colon like [default: the project root]
159        // But allow simple numeric ranges which are handled above
160        if text.contains(':') && text.contains(' ') {
161            return true;
162        }
163
164        // Skip alert/admonition patterns like [!WARN], [!NOTE], etc.
165        if text.starts_with('!') {
166            return true;
167        }
168
169        // Note: We don't filter out patterns with backticks because backticks in reference names
170        // are valid markdown syntax, e.g., [`dataclasses.InitVar`] is a valid reference name
171
172        // Also don't filter out references with dots - these are legitimate reference names
173        // like [tool.ruff] or [os.path] which are valid markdown references
174
175        // Note: We don't filter based on word count anymore because legitimate references
176        // can have many words, like "python language reference for import statements"
177        // Word count filtering was causing false positives where valid references were
178        // being incorrectly flagged as unused
179
180        false
181    }
182
183    /// Unescape a reference string by removing backslashes before special characters.
184    ///
185    /// This allows matching references like `[example\-reference]` with definitions like
186    /// `[example-reference]: http://example.com`
187    ///
188    /// Returns the unescaped reference string.
189    fn unescape_reference(reference: &str) -> String {
190        // Remove backslashes before special characters
191        reference.replace("\\", "")
192    }
193
194    /// Find all link and image reference definitions in the content.
195    ///
196    /// This method returns a HashMap where the key is the normalized reference ID and the value is a vector of (start_line, end_line) tuples.
197    fn find_definitions(&self, ctx: &crate::lint_context::LintContext) -> HashMap<String, Vec<(usize, usize)>> {
198        let mut definitions: HashMap<String, Vec<(usize, usize)>> = HashMap::new();
199
200        // First, add all reference definitions from context
201        for ref_def in &ctx.reference_defs {
202            // Apply unescape to handle escaped characters in definitions
203            let normalized_id = Self::unescape_reference(&ref_def.id); // Already lowercase from context
204            definitions
205                .entry(normalized_id)
206                .or_default()
207                .push((ref_def.line - 1, ref_def.line - 1)); // Convert to 0-indexed
208        }
209
210        // Handle multi-line definitions that might not be fully captured by ctx.reference_defs
211        let lines = &ctx.lines;
212        let mut i = 0;
213        while i < lines.len() {
214            let line_info = &lines[i];
215            let line = &line_info.content;
216
217            // Skip code blocks and front matter using line info
218            if line_info.in_code_block || line_info.in_front_matter {
219                i += 1;
220                continue;
221            }
222
223            // Check for multi-line continuation of existing definitions
224            if i > 0 && CONTINUATION_REGEX.is_match(line) {
225                // Find the reference definition this continues
226                let mut def_start = i - 1;
227                while def_start > 0 && !REFERENCE_DEFINITION_REGEX.is_match(&lines[def_start].content) {
228                    def_start -= 1;
229                }
230
231                if let Some(caps) = REFERENCE_DEFINITION_REGEX.captures(&lines[def_start].content) {
232                    let ref_id = caps.get(1).unwrap().as_str().trim();
233                    let normalized_id = Self::unescape_reference(ref_id).to_lowercase();
234
235                    // Update the end line for this definition
236                    if let Some(ranges) = definitions.get_mut(&normalized_id)
237                        && let Some(last_range) = ranges.last_mut()
238                        && last_range.0 == def_start
239                    {
240                        last_range.1 = i;
241                    }
242                }
243            }
244            i += 1;
245        }
246        definitions
247    }
248
249    /// Find all link and image reference reference usages in the content.
250    ///
251    /// This method returns a HashSet of all normalized reference IDs found in usage.
252    /// It leverages cached data from LintContext for efficiency.
253    fn find_usages(&self, ctx: &crate::lint_context::LintContext) -> HashSet<String> {
254        let mut usages: HashSet<String> = HashSet::new();
255
256        // 1. Add usages from cached reference links in LintContext
257        for link in &ctx.links {
258            if link.is_reference
259                && let Some(ref_id) = &link.reference_id
260            {
261                // Ensure the link itself is not inside a code block line
262                if !ctx.line_info(link.line).is_some_and(|info| info.in_code_block) {
263                    usages.insert(Self::unescape_reference(ref_id).to_lowercase());
264                }
265            }
266        }
267
268        // 2. Add usages from cached reference images in LintContext
269        for image in &ctx.images {
270            if image.is_reference
271                && let Some(ref_id) = &image.reference_id
272            {
273                // Ensure the image itself is not inside a code block line
274                if !ctx.line_info(image.line).is_some_and(|info| info.in_code_block) {
275                    usages.insert(Self::unescape_reference(ref_id).to_lowercase());
276                }
277            }
278        }
279
280        // 3. Find shortcut references [ref] not already handled by DocumentStructure.links
281        //    and ensure they are not within code spans or code blocks.
282        // Cache code spans once before the loop
283        let code_spans = ctx.code_spans();
284
285        for line_info in ctx.lines.iter() {
286            // Skip lines in code blocks or front matter
287            if line_info.in_code_block || line_info.in_front_matter {
288                continue;
289            }
290
291            // Skip lines that are reference definitions (start with [ref]: at beginning)
292            if REFERENCE_DEFINITION_REGEX.is_match(&line_info.content) {
293                continue;
294            }
295
296            // Find potential shortcut references
297            for caps in SHORTCUT_REFERENCE_REGEX.captures_iter(&line_info.content).flatten() {
298                if let Some(full_match) = caps.get(0)
299                    && let Some(ref_id_match) = caps.get(1)
300                {
301                    // Check if the match is within a code span
302                    let match_byte_offset = line_info.byte_offset + full_match.start();
303                    let in_code_span = code_spans
304                        .iter()
305                        .any(|span| match_byte_offset >= span.byte_offset && match_byte_offset < span.byte_end);
306
307                    if !in_code_span {
308                        let ref_id = ref_id_match.as_str().trim();
309
310                        // Skip patterns that are likely not markdown references
311                        if !Self::is_likely_not_reference(ref_id) {
312                            let normalized_id = Self::unescape_reference(ref_id).to_lowercase();
313                            usages.insert(normalized_id);
314                        }
315                    }
316                }
317            }
318        }
319
320        // NOTE: The complex recursive loop trying to find references within definitions
321        // has been removed as it's not standard Markdown behavior for finding *usages*.
322        // Usages refer to `[text][ref]`, `![alt][ref]`, `[ref]`, etc., in the main content,
323        // not references potentially embedded within the URL or title of another definition.
324
325        usages
326    }
327
328    /// Get unused references with their line ranges.
329    ///
330    /// This method uses the cached definitions to improve performance.
331    ///
332    /// Note: References that are only used inside code blocks are still considered unused,
333    /// as code blocks are treated as examples or documentation rather than actual content.
334    fn get_unused_references(
335        &self,
336        definitions: &HashMap<String, Vec<(usize, usize)>>,
337        usages: &HashSet<String>,
338    ) -> Vec<(String, usize, usize)> {
339        let mut unused = Vec::new();
340        for (id, ranges) in definitions {
341            // If this id is not used anywhere and is not in the ignored list
342            if !usages.contains(id) && !self.is_ignored_definition(id) {
343                // Only report as unused if there's exactly one definition
344                // Multiple definitions are already reported as duplicates
345                if ranges.len() == 1 {
346                    let (start, end) = ranges[0];
347                    unused.push((id.clone(), start, end));
348                }
349                // If there are multiple definitions (duplicates), don't report them as unused
350                // They're already being reported as duplicate definitions
351            }
352        }
353        unused
354    }
355
356    /// Check if a definition should be ignored (kept even if unused)
357    fn is_ignored_definition(&self, definition_id: &str) -> bool {
358        self.config
359            .ignored_definitions
360            .iter()
361            .any(|ignored| ignored.eq_ignore_ascii_case(definition_id))
362    }
363}
364
365impl Default for MD053LinkImageReferenceDefinitions {
366    fn default() -> Self {
367        Self::new()
368    }
369}
370
371impl Rule for MD053LinkImageReferenceDefinitions {
372    fn name(&self) -> &'static str {
373        "MD053"
374    }
375
376    fn description(&self) -> &'static str {
377        "Link and image reference definitions should be needed"
378    }
379
380    /// Check the content for unused and duplicate link/image reference definitions.
381    ///
382    /// This implementation uses caching for improved performance on large documents.
383    fn check(&self, ctx: &crate::lint_context::LintContext) -> LintResult {
384        // Find definitions and usages using LintContext
385        let definitions = self.find_definitions(ctx);
386        let usages = self.find_usages(ctx);
387
388        // Get unused references by comparing definitions and usages
389        let unused_refs = self.get_unused_references(&definitions, &usages);
390
391        let mut warnings = Vec::new();
392
393        // Check for duplicate definitions (case-insensitive per CommonMark spec)
394        let mut seen_definitions: HashMap<String, (String, usize)> = HashMap::new(); // lowercase -> (original, first_line)
395
396        for (definition_id, ranges) in &definitions {
397            // Skip ignored definitions for duplicate checking
398            if self.is_ignored_definition(definition_id) {
399                continue;
400            }
401
402            if ranges.len() > 1 {
403                // Multiple definitions with exact same ID (already lowercase)
404                for (i, &(start_line, _)) in ranges.iter().enumerate() {
405                    if i > 0 {
406                        // Skip the first occurrence, report all others
407                        let line_num = start_line + 1;
408                        let line_content = ctx.lines.get(start_line).map(|l| l.content.as_str()).unwrap_or("");
409                        let (start_line_1idx, start_col, end_line, end_col) =
410                            calculate_line_range(line_num, line_content);
411
412                        warnings.push(LintWarning {
413                            rule_name: Some(self.name().to_string()),
414                            line: start_line_1idx,
415                            column: start_col,
416                            end_line,
417                            end_column: end_col,
418                            message: format!("Duplicate link or image reference definition: [{definition_id}]"),
419                            severity: Severity::Warning,
420                            fix: None,
421                        });
422                    }
423                }
424            }
425
426            // Track for case-variant duplicates
427            if let Some(&(start_line, _)) = ranges.first() {
428                // Find the original case version from the line
429                if let Some(line_info) = ctx.lines.get(start_line)
430                    && let Some(caps) = REFERENCE_DEFINITION_REGEX.captures(&line_info.content)
431                {
432                    let original_id = caps.get(1).unwrap().as_str().trim();
433                    let lower_id = original_id.to_lowercase();
434
435                    if let Some((first_original, first_line)) = seen_definitions.get(&lower_id) {
436                        // Found a case-variant duplicate
437                        if first_original != original_id {
438                            let line_num = start_line + 1;
439                            let line_content = &line_info.content;
440                            let (start_line_1idx, start_col, end_line, end_col) =
441                                calculate_line_range(line_num, line_content);
442
443                            warnings.push(LintWarning {
444                                    rule_name: Some(self.name().to_string()),
445                                    line: start_line_1idx,
446                                    column: start_col,
447                                    end_line,
448                                    end_column: end_col,
449                                    message: format!("Duplicate link or image reference definition: [{}] (conflicts with [{}] on line {})",
450                                                   original_id, first_original, first_line + 1),
451                                    severity: Severity::Warning,
452                                    fix: None,
453                                });
454                        }
455                    } else {
456                        seen_definitions.insert(lower_id, (original_id.to_string(), start_line));
457                    }
458                }
459            }
460        }
461
462        // Create warnings for unused references
463        for (definition, start, _end) in unused_refs {
464            let line_num = start + 1; // 1-indexed line numbers
465            let line_content = ctx.lines.get(start).map(|l| l.content.as_str()).unwrap_or("");
466
467            // Calculate precise character range for the entire reference definition line
468            let (start_line, start_col, end_line, end_col) = calculate_line_range(line_num, line_content);
469
470            warnings.push(LintWarning {
471                rule_name: Some(self.name().to_string()),
472                line: start_line,
473                column: start_col,
474                end_line,
475                end_column: end_col,
476                message: format!("Unused link/image reference: [{definition}]"),
477                severity: Severity::Warning,
478                fix: None, // MD053 is warning-only, no automatic fixes
479            });
480        }
481
482        Ok(warnings)
483    }
484
485    /// MD053 does not provide automatic fixes
486    fn fix(&self, ctx: &crate::lint_context::LintContext) -> Result<String, LintError> {
487        // This rule is warning-only, no automatic fixes provided
488        Ok(ctx.content.to_string())
489    }
490
491    /// Check if this rule should be skipped for performance
492    fn should_skip(&self, ctx: &crate::lint_context::LintContext) -> bool {
493        // Skip if content is empty or has no links/images
494        ctx.content.is_empty() || !ctx.likely_has_links_or_images()
495    }
496
497    fn as_any(&self) -> &dyn std::any::Any {
498        self
499    }
500
501    fn default_config_section(&self) -> Option<(String, toml::Value)> {
502        let default_config = MD053Config::default();
503        let json_value = serde_json::to_value(&default_config).ok()?;
504        let toml_value = crate::rule_config_serde::json_to_toml_value(&json_value)?;
505        if let toml::Value::Table(table) = toml_value {
506            if !table.is_empty() {
507                Some((MD053Config::RULE_NAME.to_string(), toml::Value::Table(table)))
508            } else {
509                None
510            }
511        } else {
512            None
513        }
514    }
515
516    fn from_config(config: &crate::config::Config) -> Box<dyn Rule>
517    where
518        Self: Sized,
519    {
520        let rule_config = crate::rule_config_serde::load_rule_config::<MD053Config>(config);
521        Box::new(MD053LinkImageReferenceDefinitions::from_config_struct(rule_config))
522    }
523}
524
525#[cfg(test)]
526mod tests {
527    use super::*;
528    use crate::lint_context::LintContext;
529
530    #[test]
531    fn test_used_reference_link() {
532        let rule = MD053LinkImageReferenceDefinitions::new();
533        let content = "[text][ref]\n\n[ref]: https://example.com";
534        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
535        let result = rule.check(&ctx).unwrap();
536
537        assert_eq!(result.len(), 0);
538    }
539
540    #[test]
541    fn test_unused_reference_definition() {
542        let rule = MD053LinkImageReferenceDefinitions::new();
543        let content = "[unused]: https://example.com";
544        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
545        let result = rule.check(&ctx).unwrap();
546
547        assert_eq!(result.len(), 1);
548        assert!(result[0].message.contains("Unused link/image reference: [unused]"));
549    }
550
551    #[test]
552    fn test_used_reference_image() {
553        let rule = MD053LinkImageReferenceDefinitions::new();
554        let content = "![alt][img]\n\n[img]: image.jpg";
555        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
556        let result = rule.check(&ctx).unwrap();
557
558        assert_eq!(result.len(), 0);
559    }
560
561    #[test]
562    fn test_case_insensitive_matching() {
563        let rule = MD053LinkImageReferenceDefinitions::new();
564        let content = "[Text][REF]\n\n[ref]: https://example.com";
565        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
566        let result = rule.check(&ctx).unwrap();
567
568        assert_eq!(result.len(), 0);
569    }
570
571    #[test]
572    fn test_shortcut_reference() {
573        let rule = MD053LinkImageReferenceDefinitions::new();
574        let content = "[ref]\n\n[ref]: https://example.com";
575        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
576        let result = rule.check(&ctx).unwrap();
577
578        assert_eq!(result.len(), 0);
579    }
580
581    #[test]
582    fn test_collapsed_reference() {
583        let rule = MD053LinkImageReferenceDefinitions::new();
584        let content = "[ref][]\n\n[ref]: https://example.com";
585        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
586        let result = rule.check(&ctx).unwrap();
587
588        assert_eq!(result.len(), 0);
589    }
590
591    #[test]
592    fn test_multiple_unused_definitions() {
593        let rule = MD053LinkImageReferenceDefinitions::new();
594        let content = "[unused1]: url1\n[unused2]: url2\n[unused3]: url3";
595        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
596        let result = rule.check(&ctx).unwrap();
597
598        assert_eq!(result.len(), 3);
599
600        // The warnings might not be in the same order, so collect all messages
601        let messages: Vec<String> = result.iter().map(|w| w.message.clone()).collect();
602        assert!(messages.iter().any(|m| m.contains("unused1")));
603        assert!(messages.iter().any(|m| m.contains("unused2")));
604        assert!(messages.iter().any(|m| m.contains("unused3")));
605    }
606
607    #[test]
608    fn test_mixed_used_and_unused() {
609        let rule = MD053LinkImageReferenceDefinitions::new();
610        let content = "[used]\n\n[used]: url1\n[unused]: url2";
611        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
612        let result = rule.check(&ctx).unwrap();
613
614        assert_eq!(result.len(), 1);
615        assert!(result[0].message.contains("unused"));
616    }
617
618    #[test]
619    fn test_multiline_definition() {
620        let rule = MD053LinkImageReferenceDefinitions::new();
621        let content = "[ref]: https://example.com\n  \"Title on next line\"";
622        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
623        let result = rule.check(&ctx).unwrap();
624
625        assert_eq!(result.len(), 1); // Still unused
626    }
627
628    #[test]
629    fn test_reference_in_code_block() {
630        let rule = MD053LinkImageReferenceDefinitions::new();
631        let content = "```\n[ref]\n```\n\n[ref]: https://example.com";
632        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
633        let result = rule.check(&ctx).unwrap();
634
635        // Reference used only in code block is still considered unused
636        assert_eq!(result.len(), 1);
637    }
638
639    #[test]
640    fn test_reference_in_inline_code() {
641        let rule = MD053LinkImageReferenceDefinitions::new();
642        let content = "`[ref]`\n\n[ref]: https://example.com";
643        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
644        let result = rule.check(&ctx).unwrap();
645
646        // Reference in inline code is not a usage
647        assert_eq!(result.len(), 1);
648    }
649
650    #[test]
651    fn test_escaped_reference() {
652        let rule = MD053LinkImageReferenceDefinitions::new();
653        let content = "[example\\-ref]\n\n[example-ref]: https://example.com";
654        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
655        let result = rule.check(&ctx).unwrap();
656
657        // Should match despite escaping
658        assert_eq!(result.len(), 0);
659    }
660
661    #[test]
662    fn test_duplicate_definitions() {
663        let rule = MD053LinkImageReferenceDefinitions::new();
664        let content = "[ref]: url1\n[ref]: url2\n\n[ref]";
665        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
666        let result = rule.check(&ctx).unwrap();
667
668        // Should flag the duplicate definition even though it's used (matches markdownlint)
669        assert_eq!(result.len(), 1);
670    }
671
672    #[test]
673    fn test_fix_returns_original() {
674        // MD053 is warning-only, fix should return original content
675        let rule = MD053LinkImageReferenceDefinitions::new();
676        let content = "[used]\n\n[used]: url1\n[unused]: url2\n\nMore content";
677        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
678        let fixed = rule.fix(&ctx).unwrap();
679
680        assert_eq!(fixed, content);
681    }
682
683    #[test]
684    fn test_fix_preserves_content() {
685        // MD053 is warning-only, fix should preserve all content
686        let rule = MD053LinkImageReferenceDefinitions::new();
687        let content = "Content\n\n[unused]: url\n\nMore content";
688        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
689        let fixed = rule.fix(&ctx).unwrap();
690
691        assert_eq!(fixed, content);
692    }
693
694    #[test]
695    fn test_fix_does_not_remove() {
696        // MD053 is warning-only, fix should not remove anything
697        let rule = MD053LinkImageReferenceDefinitions::new();
698        let content = "[unused1]: url1\n[unused2]: url2\n[unused3]: url3";
699        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
700        let fixed = rule.fix(&ctx).unwrap();
701
702        assert_eq!(fixed, content);
703    }
704
705    #[test]
706    fn test_special_characters_in_reference() {
707        let rule = MD053LinkImageReferenceDefinitions::new();
708        let content = "[ref-with_special.chars]\n\n[ref-with_special.chars]: url";
709        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
710        let result = rule.check(&ctx).unwrap();
711
712        assert_eq!(result.len(), 0);
713    }
714
715    #[test]
716    fn test_find_definitions() {
717        let rule = MD053LinkImageReferenceDefinitions::new();
718        let content = "[ref1]: url1\n[ref2]: url2\nSome text\n[ref3]: url3";
719        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
720        let defs = rule.find_definitions(&ctx);
721
722        assert_eq!(defs.len(), 3);
723        assert!(defs.contains_key("ref1"));
724        assert!(defs.contains_key("ref2"));
725        assert!(defs.contains_key("ref3"));
726    }
727
728    #[test]
729    fn test_find_usages() {
730        let rule = MD053LinkImageReferenceDefinitions::new();
731        let content = "[text][ref1] and [ref2] and ![img][ref3]";
732        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
733        let usages = rule.find_usages(&ctx);
734
735        assert!(usages.contains("ref1"));
736        assert!(usages.contains("ref2"));
737        assert!(usages.contains("ref3"));
738    }
739
740    #[test]
741    fn test_ignored_definitions_config() {
742        // Test with ignored definitions
743        let config = MD053Config {
744            ignored_definitions: vec!["todo".to_string(), "draft".to_string()],
745        };
746        let rule = MD053LinkImageReferenceDefinitions::from_config_struct(config);
747
748        let content = "[todo]: https://example.com/todo\n[draft]: https://example.com/draft\n[unused]: https://example.com/unused";
749        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
750        let result = rule.check(&ctx).unwrap();
751
752        // Should only flag "unused", not "todo" or "draft"
753        assert_eq!(result.len(), 1);
754        assert!(result[0].message.contains("unused"));
755        assert!(!result[0].message.contains("todo"));
756        assert!(!result[0].message.contains("draft"));
757    }
758
759    #[test]
760    fn test_ignored_definitions_case_insensitive() {
761        // Test case-insensitive matching of ignored definitions
762        let config = MD053Config {
763            ignored_definitions: vec!["TODO".to_string()],
764        };
765        let rule = MD053LinkImageReferenceDefinitions::from_config_struct(config);
766
767        let content = "[todo]: https://example.com/todo\n[unused]: https://example.com/unused";
768        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
769        let result = rule.check(&ctx).unwrap();
770
771        // Should only flag "unused", not "todo" (matches "TODO" case-insensitively)
772        assert_eq!(result.len(), 1);
773        assert!(result[0].message.contains("unused"));
774        assert!(!result[0].message.contains("todo"));
775    }
776
777    #[test]
778    fn test_default_config_section() {
779        let rule = MD053LinkImageReferenceDefinitions::default();
780        let config_section = rule.default_config_section();
781
782        assert!(config_section.is_some());
783        let (name, value) = config_section.unwrap();
784        assert_eq!(name, "MD053");
785
786        // Should contain the ignored_definitions option with default empty array
787        if let toml::Value::Table(table) = value {
788            assert!(table.contains_key("ignored-definitions"));
789            assert_eq!(table["ignored-definitions"], toml::Value::Array(vec![]));
790        } else {
791            panic!("Expected TOML table");
792        }
793    }
794
795    #[test]
796    fn test_fix_with_ignored_definitions() {
797        // MD053 is warning-only, fix should not remove anything even with ignored definitions
798        let config = MD053Config {
799            ignored_definitions: vec!["template".to_string()],
800        };
801        let rule = MD053LinkImageReferenceDefinitions::from_config_struct(config);
802
803        let content = "[template]: https://example.com/template\n[unused]: https://example.com/unused\n\nSome content.";
804        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
805        let fixed = rule.fix(&ctx).unwrap();
806
807        // Should keep everything since MD053 doesn't fix
808        assert_eq!(fixed, content);
809    }
810
811    #[test]
812    fn test_duplicate_definitions_exact_case() {
813        let rule = MD053LinkImageReferenceDefinitions::new();
814        let content = "[ref]: url1\n[ref]: url2\n[ref]: url3";
815        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
816        let result = rule.check(&ctx).unwrap();
817
818        // Should have 2 duplicate warnings (for the 2nd and 3rd definitions)
819        // Plus 1 unused warning
820        let duplicate_warnings: Vec<_> = result.iter().filter(|w| w.message.contains("Duplicate")).collect();
821        assert_eq!(duplicate_warnings.len(), 2);
822        assert_eq!(duplicate_warnings[0].line, 2);
823        assert_eq!(duplicate_warnings[1].line, 3);
824    }
825
826    #[test]
827    fn test_duplicate_definitions_case_variants() {
828        let rule = MD053LinkImageReferenceDefinitions::new();
829        let content =
830            "[method resolution order]: url1\n[Method Resolution Order]: url2\n[METHOD RESOLUTION ORDER]: url3";
831        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
832        let result = rule.check(&ctx).unwrap();
833
834        // Should have 2 duplicate warnings (for the 2nd and 3rd definitions)
835        // Note: These are treated as exact duplicates since they normalize to the same ID
836        let duplicate_warnings: Vec<_> = result.iter().filter(|w| w.message.contains("Duplicate")).collect();
837        assert_eq!(duplicate_warnings.len(), 2);
838
839        // The exact duplicate messages don't include "conflicts with"
840        // Only case-variant duplicates with different normalized forms would
841        assert_eq!(duplicate_warnings[0].line, 2);
842        assert_eq!(duplicate_warnings[1].line, 3);
843    }
844
845    #[test]
846    fn test_duplicate_and_unused() {
847        let rule = MD053LinkImageReferenceDefinitions::new();
848        let content = "[used]\n[used]: url1\n[used]: url2\n[unused]: url3";
849        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
850        let result = rule.check(&ctx).unwrap();
851
852        // Should have 1 duplicate warning and 1 unused warning
853        let duplicate_warnings: Vec<_> = result.iter().filter(|w| w.message.contains("Duplicate")).collect();
854        let unused_warnings: Vec<_> = result.iter().filter(|w| w.message.contains("Unused")).collect();
855
856        assert_eq!(duplicate_warnings.len(), 1);
857        assert_eq!(unused_warnings.len(), 1);
858        assert_eq!(duplicate_warnings[0].line, 3); // Second [used] definition
859        assert_eq!(unused_warnings[0].line, 4); // [unused] definition
860    }
861
862    #[test]
863    fn test_duplicate_with_usage() {
864        let rule = MD053LinkImageReferenceDefinitions::new();
865        // Even if used, duplicates should still be reported
866        let content = "[ref]\n\n[ref]: url1\n[ref]: url2";
867        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
868        let result = rule.check(&ctx).unwrap();
869
870        // Should have 1 duplicate warning (no unused since it's referenced)
871        let duplicate_warnings: Vec<_> = result.iter().filter(|w| w.message.contains("Duplicate")).collect();
872        let unused_warnings: Vec<_> = result.iter().filter(|w| w.message.contains("Unused")).collect();
873
874        assert_eq!(duplicate_warnings.len(), 1);
875        assert_eq!(unused_warnings.len(), 0);
876        assert_eq!(duplicate_warnings[0].line, 4);
877    }
878
879    #[test]
880    fn test_no_duplicate_different_ids() {
881        let rule = MD053LinkImageReferenceDefinitions::new();
882        let content = "[ref1]: url1\n[ref2]: url2\n[ref3]: url3";
883        let ctx = LintContext::new(content, crate::config::MarkdownFlavor::Standard);
884        let result = rule.check(&ctx).unwrap();
885
886        // Should have no duplicate warnings, only unused warnings
887        let duplicate_warnings: Vec<_> = result.iter().filter(|w| w.message.contains("Duplicate")).collect();
888        assert_eq!(duplicate_warnings.len(), 0);
889    }
890}
rumdl_lib/rules/md053_link_image_reference_definitions.rs

rumdl_lib/rules/
md053_link_image_reference_definitions.rs
