rumdl_lib/rules/md013_line_length.rs
1/// Rule MD013: Line length
2///
3/// See [docs/md013.md](../../docs/md013.md) for full documentation, configuration, and examples.
4use crate::rule::{LintError, LintResult, LintWarning, Rule, RuleCategory, Severity};
5use crate::utils::mkdocs_admonitions;
6use crate::utils::mkdocs_attr_list::is_standalone_attr_list;
7use crate::utils::mkdocs_snippets::is_snippet_block_delimiter;
8use crate::utils::mkdocs_tabs;
9use crate::utils::range_utils::calculate_excess_range;
10use crate::utils::regex_cache::{IMAGE_REF_PATTERN, LINK_REF_PATTERN, URL_PATTERN};
11use crate::utils::table_utils::TableUtils;
12use crate::utils::text_reflow::{
13 BlockquoteLineData, blockquote_continuation_style, dominant_blockquote_prefix, is_self_contained_display_math_line,
14 join_soft_break_lines, reflow_blockquote_content, split_into_sentences,
15};
16use pulldown_cmark::LinkType;
17use toml;
18
19mod block_builder;
20mod helpers;
21pub mod md013_config;
22use crate::rules::md030_list_marker_space::MD030Config;
23use crate::utils::is_template_directive_only;
24use block_builder::{Block, BlockBuilder};
25use helpers::{
26 extract_list_marker_and_content, has_hard_break, is_github_alert_marker, is_horizontal_rule, is_html_only_line,
27 is_list_item, is_setext_heading_text_line, is_setext_underline_content, is_standalone_link_or_image_line,
28 is_unwrappable_line, source_list_marker, split_into_segments, standalone_link_ends_paragraph,
29 trim_preserving_hard_break,
30};
31pub use md013_config::MD013Config;
32use md013_config::{CjkSoftBreak, LengthMode, ReflowMode};
33
34#[cfg(test)]
35mod tests;
36use unicode_width::UnicodeWidthStr;
37
38#[derive(Clone, Default)]
39pub struct MD013LineLength {
40 pub(crate) config: MD013Config,
41 /// MD030 list-marker spacing, applied when reflowing list items so the rewrite
42 /// uses the configured post-marker spacing rather than a hard-coded single
43 /// space. Defaults to MD030's defaults (a single space everywhere), which
44 /// reproduces the previous behaviour exactly. See [`MD030Config::expected_spaces`].
45 pub(crate) list_spacing: MD030Config,
46}
47
48/// Blockquote paragraph line collected for reflow, with original line index for range computation.
49struct CollectedBlockquoteLine {
50 line_idx: usize,
51 data: BlockquoteLineData,
52}
53
54/// A deliberately conservative MDG step candidate.
55///
56/// Gherkin keywords are localized, so recognizing the word after the marker is
57/// not reliable without the complete dialect table. MDG represents steps as
58/// unordered Markdown list items; withholding reflow from all such items in the
59/// flavor is safer than splitting a step and changing its Gherkin text.
60fn is_potential_mdg_step(ctx: &crate::lint_context::LintContext, line_num: usize) -> bool {
61 let Some(item) = ctx.list_item_on_line(line_num) else {
62 return false;
63 };
64 !item.is_ordered() && matches!(item.marker_char(), Some('*' | '-' | '+'))
65}
66
67/// Whether line `line_num` (1-based) is touched on either boundary by a code
68/// span crossing more than one line: a span containing the newline that ends
69/// the line before it, or the newline that ends it.
70///
71/// A renderer reads the line break inside a code span as one space, so such a
72/// line is code however it is spelled, and a `$$...$$` expression on it is no
73/// display block. A span that begins and ends on the line itself does not
74/// matter. `flags` holds one entry per line of the document, indexed here by
75/// the 1-based line number, and comes from
76/// [`crate::utils::text_reflow::lines_touching_multiline_code_span`], which
77/// reads code spans on their own rather than alongside math delimiters: a
78/// `$$...$$` pair closes wherever the next `$$` sits, backtick or not, so a
79/// backtick between them that in fact opens a span reaching past the line
80/// needs a parser that isn't also deciding where the math closes.
81fn line_touches_multiline_code_span(flags: &[bool], line_num: usize) -> bool {
82 line_num
83 .checked_sub(1)
84 .and_then(|idx| flags.get(idx))
85 .copied()
86 .unwrap_or(false)
87}
88
89/// Whether any of the lines `start_idx..=end_idx` (0-indexed) is a definition
90/// list's term, or a definition's marker line holding its text.
91///
92/// The container reflow paths join these into prose: a term is absorbed into
93/// the text around it, and a marker loses the spacing that sets the
94/// definition's content column. A definition's later paragraph is prose at
95/// the definition's indentation, which the paths keep, and a list or
96/// blockquote nested inside a definition holds none of these lines, so both
97/// still reflow.
98fn holds_definition_list(ctx: &crate::lint_context::LintContext, start_idx: usize, end_idx: usize) -> bool {
99 (start_idx..=end_idx).any(|idx| {
100 let line_num = idx + 1;
101 ctx.is_definition_term(line_num)
102 || ctx
103 .definition_text_at(line_num)
104 .is_some_and(|text| text.start_line == line_num && text.marker_prefix_len.is_some())
105 })
106}
107
108impl MD013LineLength {
109 pub fn new(line_length: usize, code_blocks: bool, tables: bool, headings: bool, strict: bool) -> Self {
110 Self {
111 config: MD013Config {
112 line_length: crate::types::LineLength::new(line_length),
113 code_blocks,
114 code_spans: true,
115 tables,
116 headings,
117 math_blocks: true,
118 bracket_display_math: false,
119 paragraphs: true, // Default to true for backwards compatibility
120 blockquotes: true, // Default to true for backwards compatibility
121 strict,
122 stern: false,
123 heading_line_length: None,
124 code_block_line_length: None,
125 reflow: false,
126 reflow_mode: ReflowMode::default(),
127 length_mode: LengthMode::default(),
128 abbreviations: Vec::new(),
129 require_sentence_capital: true,
130 ignore_link_urls: true,
131 atomic_spans: true,
132 reflow_break_link_text: false,
133 reflow_length_exemptions: false,
134 cjk_soft_break: CjkSoftBreak::default(),
135 },
136 list_spacing: MD030Config::default(),
137 }
138 }
139
140 pub fn from_config_struct(config: MD013Config) -> Self {
141 Self {
142 config,
143 list_spacing: MD030Config::default(),
144 }
145 }
146
147 /// Return a clone with code block checking disabled.
148 /// Used for doc comment linting where code blocks are Rust code managed by rustfmt.
149 pub fn with_code_blocks_disabled(&self) -> Self {
150 let mut clone = self.clone();
151 clone.config.code_blocks = false;
152 clone
153 }
154
155 /// Convert MD013 LengthMode to text_reflow ReflowLengthMode
156 /// Normalized set of reference labels defined in the document.
157 ///
158 /// Passed to reflow so a bare shortcut reference (`[text]`) is treated as an
159 /// atomic link only when its label is actually defined; an undefined
160 /// bracketed run reflows as literal prose.
161 fn defined_reference_labels(ctx: &crate::lint_context::LintContext) -> std::collections::HashSet<String> {
162 ctx.reference_definitions()
163 .iter()
164 .map(|d| crate::utils::text_reflow::normalize_reference_label(&d.id))
165 .collect()
166 }
167
168 /// Build the reflow options shared by every MD013 fix path.
169 ///
170 /// `line_length` varies per call site (some subtract a list-marker or
171 /// blockquote prefix); every other field is derived uniformly from the
172 /// effective config and the document flavor. Callers that need a different
173 /// `max_list_continuation_indent` override it via struct update.
174 fn reflow_options(
175 ctx: &crate::lint_context::LintContext,
176 config: &MD013Config,
177 line_length: usize,
178 ) -> crate::utils::text_reflow::ReflowOptions {
179 crate::utils::text_reflow::ReflowOptions {
180 line_length,
181 break_on_sentences: true,
182 preserve_breaks: false,
183 sentence_per_line: config.reflow_mode == ReflowMode::SentencePerLine,
184 semantic_line_breaks: config.reflow_mode == ReflowMode::SemanticLineBreaks,
185 abbreviations: config.abbreviations_for_reflow(),
186 length_mode: config.reflow_length_mode(),
187 attr_lists: ctx.flavor.supports_attr_lists(),
188 myst_roles: ctx.flavor.supports_myst_roles(),
189 require_sentence_capital: config.require_sentence_capital,
190 max_list_continuation_indent: if ctx.flavor.requires_strict_list_indent() {
191 Some(4)
192 } else {
193 None
194 },
195 defined_references: Some(Self::defined_reference_labels(ctx)),
196 atomic_spans: config.atomic_spans,
197 break_link_text: config.reflow_break_link_text,
198 length_exemptions: config.length_exemptions_for_reflow(),
199 cjk_soft_break: config.cjk_soft_break,
200 }
201 }
202
203 fn should_ignore_line(
204 &self,
205 line: &str,
206 _lines: &[&str],
207 current_line: usize,
208 ctx: &crate::lint_context::LintContext,
209 ) -> bool {
210 if self.config.strict {
211 return false;
212 }
213
214 // Quick check for common patterns before expensive regex
215 let trimmed = line.trim();
216
217 // Only skip if the entire line is a URL (quick check first)
218 if (trimmed.starts_with("http://") || trimmed.starts_with("https://")) && URL_PATTERN.is_match(trimmed) {
219 return true;
220 }
221
222 // Only skip if the entire line is an image reference (quick check first)
223 if trimmed.starts_with("![") && trimmed.ends_with(']') && IMAGE_REF_PATTERN.is_match(trimmed) {
224 return true;
225 }
226
227 // Note: link reference definitions are handled as always-exempt (even in strict mode)
228 // in the main check loop, so they don't need to be checked here.
229
230 // Code blocks with long strings (only check if in code block)
231 if ctx.line_info(current_line + 1).is_some_and(|info| info.in_code_block)
232 && !trimmed.is_empty()
233 && !line.contains(' ')
234 && !line.contains('\t')
235 {
236 return true;
237 }
238
239 false
240 }
241
242 /// Check if rule should skip based on provided config (used for inline config support)
243 fn should_skip_with_config(&self, ctx: &crate::lint_context::LintContext, config: &MD013Config) -> bool {
244 // Skip if content is empty
245 if ctx.content.is_empty() {
246 return true;
247 }
248
249 // For sentence-per-line, semantic-line-breaks, or normalize mode, never skip based on line length
250 if config.reflow
251 && (config.reflow_mode == ReflowMode::SentencePerLine
252 || config.reflow_mode == ReflowMode::SemanticLineBreaks
253 || config.reflow_mode == ReflowMode::Normalize)
254 {
255 return false;
256 }
257
258 // Use the smallest applicable budget across line/heading/code-block
259 // contexts so a stricter context-specific limit doesn't get masked by
260 // the document-wide budget.
261 let min_limit = config.min_effective_line_length();
262 if min_limit.is_unlimited() {
263 return true;
264 }
265 let min_limit_bytes = min_limit.get();
266
267 // Quick check: if total content is shorter than the smallest line limit,
268 // definitely skip.
269 if ctx.content.len() <= min_limit_bytes {
270 return true;
271 }
272
273 // Skip if no line exceeds the smallest applicable limit.
274 !ctx.lines.iter().any(|line| line.byte_len > min_limit_bytes)
275 }
276
277 fn normalize_mode_needs_reflow<'a, I>(&self, lines: I, config: &MD013Config) -> bool
278 where
279 I: IntoIterator<Item = &'a str>,
280 {
281 let mut line_count = 0;
282 let check_length = !config.line_length.is_unlimited();
283
284 for line in lines {
285 line_count += 1;
286 if check_length && self.calculate_effective_length(line) > config.line_length.get() {
287 return true;
288 }
289 }
290
291 line_count > 1
292 }
293}
294
295impl Rule for MD013LineLength {
296 fn name(&self) -> &'static str {
297 "MD013"
298 }
299
300 fn description(&self) -> &'static str {
301 "Line length should not be excessive"
302 }
303
304 fn check(&self, ctx: &crate::lint_context::LintContext) -> LintResult {
305 // Use pre-parsed inline config from LintContext
306 let config_override = ctx.inline_config().get_rule_config("MD013");
307
308 // Apply configuration override if present
309 let effective_config = if let Some(json_config) = config_override {
310 if let Some(obj) = json_config.as_object() {
311 let mut config = self.config.clone();
312 if let Some(line_length) = obj.get("line_length").and_then(serde_json::Value::as_u64) {
313 config.line_length = crate::types::LineLength::new(line_length as usize);
314 }
315 if let Some(code_blocks) = obj.get("code_blocks").and_then(serde_json::Value::as_bool) {
316 config.code_blocks = code_blocks;
317 }
318 if let Some(code_spans) = obj.get("code_spans").and_then(serde_json::Value::as_bool) {
319 config.code_spans = code_spans;
320 }
321 if let Some(tables) = obj.get("tables").and_then(serde_json::Value::as_bool) {
322 config.tables = tables;
323 }
324 if let Some(headings) = obj.get("headings").and_then(serde_json::Value::as_bool) {
325 config.headings = headings;
326 }
327 if let Some(math_blocks) = obj
328 .get("math_blocks")
329 .or_else(|| obj.get("math-blocks"))
330 .and_then(serde_json::Value::as_bool)
331 {
332 config.math_blocks = math_blocks;
333 }
334 if let Some(enabled) = obj
335 .get("bracket_display_math")
336 .or_else(|| obj.get("bracket-display-math"))
337 .and_then(serde_json::Value::as_bool)
338 {
339 config.bracket_display_math = enabled;
340 }
341 if let Some(blockquotes) = obj.get("blockquotes").and_then(serde_json::Value::as_bool) {
342 config.blockquotes = blockquotes;
343 }
344 if let Some(strict) = obj.get("strict").and_then(serde_json::Value::as_bool) {
345 config.strict = strict;
346 }
347 if let Some(stern) = obj.get("stern").and_then(serde_json::Value::as_bool) {
348 config.stern = stern;
349 }
350 if let Some(v) = obj
351 .get("ignore_link_urls")
352 .or_else(|| obj.get("ignore-link-urls"))
353 .or_else(|| obj.get("semantic_link_understanding"))
354 .or_else(|| obj.get("semantic-link-understanding"))
355 .and_then(serde_json::Value::as_bool)
356 {
357 config.ignore_link_urls = v;
358 }
359 if let Some(reflow) = obj.get("reflow").and_then(serde_json::Value::as_bool) {
360 config.reflow = reflow;
361 }
362 if let Some(reflow_mode) = obj.get("reflow_mode").and_then(|v| v.as_str()) {
363 config.reflow_mode = match reflow_mode {
364 "default" => ReflowMode::Default,
365 "normalize" => ReflowMode::Normalize,
366 "sentence-per-line" => ReflowMode::SentencePerLine,
367 "semantic-line-breaks" => ReflowMode::SemanticLineBreaks,
368 _ => ReflowMode::default(),
369 };
370 }
371 config
372 } else {
373 self.config.clone()
374 }
375 } else {
376 self.config.clone()
377 };
378
379 // Fast early return using should_skip with EFFECTIVE config (after inline overrides)
380 // But don't skip if we're in reflow mode with Normalize or SentencePerLine
381 if self.should_skip_with_config(ctx, &effective_config)
382 && !(effective_config.reflow
383 && (effective_config.reflow_mode == ReflowMode::Normalize
384 || effective_config.reflow_mode == ReflowMode::SentencePerLine
385 || effective_config.reflow_mode == ReflowMode::SemanticLineBreaks))
386 {
387 return Ok(Vec::new());
388 }
389
390 // Direct implementation without DocumentStructure
391 let mut warnings = Vec::new();
392
393 // Special handling: line_length = 0 means "no line length limit"
394 // Skip all line length checks, but still allow reflow if enabled
395 let skip_length_checks = effective_config.line_length.is_unlimited();
396
397 // Pre-filter lines that could be problematic to avoid processing all lines.
398 // Use the smallest applicable budget across line/heading/code-block contexts
399 // so candidates aren't dropped when a stricter context-specific budget applies.
400 let prefilter_limit = effective_config.min_effective_line_length();
401 let prefilter_skip = prefilter_limit.is_unlimited();
402 let mut candidate_lines = Vec::new();
403 if !skip_length_checks && !prefilter_skip {
404 for (line_idx, line_info) in ctx.lines.iter().enumerate() {
405 // Skip front matter - it should never be linted
406 if line_info.in_front_matter {
407 continue;
408 }
409
410 // Quick length check first
411 if line_info.byte_len > prefilter_limit.get() {
412 candidate_lines.push(line_idx);
413 }
414 }
415 }
416
417 // If no candidate lines and not in normalize or sentence-per-line mode, early return
418 if candidate_lines.is_empty()
419 && !(effective_config.reflow
420 && (effective_config.reflow_mode == ReflowMode::Normalize
421 || effective_config.reflow_mode == ReflowMode::SentencePerLine
422 || effective_config.reflow_mode == ReflowMode::SemanticLineBreaks))
423 {
424 return Ok(warnings);
425 }
426
427 let lines = ctx.raw_lines();
428
429 // Whether a 1-indexed line is part of a heading. A setext heading's text
430 // is the whole paragraph its underline ends, and the parser records the
431 // heading itself only on the last of those lines, so the earlier ones
432 // are recognized through `is_setext_heading_text`. Both are O(1)
433 // per-line fields, so check them directly at each use site instead of
434 // materializing a full-document HashSet (an extra O(n) pass and
435 // allocation on a rule that runs on virtually every file).
436 let is_heading_line_num = |line_number: usize| -> bool {
437 line_number
438 .checked_sub(1)
439 .and_then(|idx| ctx.lines.get(idx))
440 .is_some_and(|line| line.heading.is_some() || line.is_setext_heading_text)
441 };
442
443 // Use pre-computed table blocks from context
444 // We need this for both the table skip check AND the paragraphs check
445 let table_blocks = &ctx.table_blocks;
446 let mut table_lines_set = std::collections::HashSet::new();
447 for table in table_blocks {
448 table_lines_set.insert(table.header_line + 1);
449 table_lines_set.insert(table.delimiter_line + 1);
450 for &line in &table.content_lines {
451 table_lines_set.insert(line + 1);
452 }
453 }
454
455 let defined_references = Self::defined_reference_labels(ctx);
456
457 // Process candidate lines for line length checks
458 'line_loop: for &line_idx in &candidate_lines {
459 let line_number = line_idx + 1;
460 let line = lines[line_idx];
461
462 // Calculate actual line length (used in warning messages)
463 let effective_length = self.calculate_effective_length(line);
464
465 // Pick the context-specific limit: heading > code-block > paragraph.
466 // Headings dominate over code-block context if a setext underline ever
467 // overlaps a fenced range (defensive — these are mutually exclusive in
468 // practice, but the explicit ordering documents intent).
469 let is_heading_line = is_heading_line_num(line_number);
470 let in_code_block = ctx.line_info(line_number).is_some_and(|info| info.in_code_block);
471 let line_limit = if is_heading_line {
472 effective_config.effective_heading_line_length().get()
473 } else if in_code_block {
474 effective_config.effective_code_block_line_length().get()
475 } else {
476 effective_config.line_length.get()
477 };
478
479 // A context-specific limit of 0 means "unlimited for this context".
480 if line_limit == 0 {
481 continue;
482 }
483
484 // Stern mode: like default, but the trailing-token forgiveness is
485 // disabled — a line with whitespace that exceeds the limit is a
486 // violation even if the excess is the final token. The "unwrappable"
487 // line exemption (single token, optionally prefixed by # or >) is
488 // still honored. Strict overrides stern entirely.
489 if effective_config.stern && !effective_config.strict && is_unwrappable_line(line) {
490 continue;
491 }
492
493 // Trailing-token forgiveness: only in default mode (not strict, not stern).
494 // If the line only exceeds the limit because of a long token at the end
495 // (URL, link chain, identifier), it passes. This matches markdownlint's
496 // behavior: line.replace(/\S*$/u, "#")
497 let check_length = if effective_config.strict || effective_config.stern {
498 effective_length
499 } else {
500 match line.rfind(char::is_whitespace) {
501 Some(pos) => {
502 let ws_char = line[pos..].chars().next().unwrap();
503 let prefix_end = pos + ws_char.len_utf8();
504 self.calculate_string_length(&line[..prefix_end]) + 1
505 }
506 None => 1, // No whitespace — entire line is a single token
507 }
508 };
509
510 // Skip lines where the check length is within the limit
511 if check_length <= line_limit {
512 continue;
513 }
514
515 // Ignore inline link/image URLs: suppress when excess comes entirely from inline URLs.
516 // Disabled by `strict` (all forgiveness off) and by `ignore_link_urls = false`
517 // (count link/image URLs toward the line length so such lines are flagged).
518 if !effective_config.strict && effective_config.ignore_link_urls {
519 let length_without_urls = self.length_without_inline_link_urls(effective_length, line_number, ctx);
520 if length_without_urls <= line_limit {
521 continue;
522 }
523 }
524
525 // Inline code spans cannot be wrapped, so reflow cannot shorten a line
526 // whose excess length is one. When code-span checking is disabled,
527 // suppress a violation that would fit once inline code spans are excluded.
528 if !effective_config.code_spans {
529 let code_span_width: usize = ctx
530 .code_spans()
531 .iter()
532 .filter(|span| span.line == line_number && span.end_line == line_number)
533 .map(|span| self.calculate_string_length(&ctx.content[span.byte_offset..span.byte_end]))
534 .sum();
535 if effective_length.saturating_sub(code_span_width) <= line_limit {
536 continue;
537 }
538 }
539
540 // Skip mkdocstrings and pymdown blocks (already handled by LintContext)
541 if ctx.lines[line_idx].in_mkdocstrings || ctx.lines[line_idx].in_pymdown_block {
542 continue;
543 }
544
545 // Skip MyST comments (% comment) — structural lines, not prose
546 if ctx.lines[line_idx].is_myst_comment {
547 continue;
548 }
549
550 // Link reference definitions are always exempt, even in strict mode.
551 // There's no way to shorten them without breaking the URL.
552 // Also check after stripping list markers, since list items may
553 // contain link ref defs as their content.
554 {
555 let trimmed = line.trim();
556 if trimmed.starts_with('[') && trimmed.contains("]:") && LINK_REF_PATTERN.is_match(trimmed) {
557 continue;
558 }
559 if is_list_item(trimmed) {
560 let (_, content) = extract_list_marker_and_content(trimmed);
561 let content_trimmed = content.trim();
562 if content_trimmed.starts_with('[')
563 && content_trimmed.contains("]:")
564 && LINK_REF_PATTERN.is_match(content_trimmed)
565 {
566 continue;
567 }
568 }
569 }
570
571 // Skip various block types efficiently
572 if !effective_config.strict {
573 // Lines whose only content is a link/image are exempt.
574 // After stripping list markers, blockquote markers, and emphasis,
575 // if only a link or image remains, there is no way to shorten it.
576 if is_standalone_link_or_image_line(ctx, line_number) {
577 continue;
578 }
579
580 // Lines consisting entirely of HTML tags are exempt.
581 // Badge lines, images with attributes, and similar inline HTML
582 // are long due to URLs in attributes and can't be meaningfully shortened.
583 if is_html_only_line(line) {
584 continue;
585 }
586
587 // Skip setext heading underlines
588 if !line.trim().is_empty() && line.trim().chars().all(|c| c == '=' || c == '-') {
589 continue;
590 }
591
592 // Skip block elements according to config flags
593 // The flags mean: true = check these elements, false = skip these elements
594 // So we skip when the flag is FALSE and the line is in that element type
595 if (!effective_config.headings && is_heading_line_num(line_number))
596 || (!effective_config.code_blocks
597 && ctx.line_info(line_number).is_some_and(|info| info.in_code_block))
598 || (!effective_config.tables && table_lines_set.contains(&line_number))
599 || (!effective_config.math_blocks && self.line_is_display_math(line_number, ctx, &effective_config))
600 || ctx.line_info(line_number).is_some_and(|info| info.in_html_block)
601 || ctx.line_info(line_number).is_some_and(|info| info.in_html_comment)
602 || ctx.line_info(line_number).is_some_and(|info| info.in_esm_block)
603 || ctx.line_info(line_number).is_some_and(|info| info.in_jsx_expression)
604 || ctx.line_info(line_number).is_some_and(|info| info.in_jsx_block)
605 || ctx.line_info(line_number).is_some_and(|info| info.in_mdx_comment)
606 || ctx.line_info(line_number).is_some_and(|info| info.in_pymdown_block)
607 {
608 continue;
609 }
610
611 // Check if this is a paragraph/regular text line
612 // If paragraphs = false, skip lines that are NOT in special blocks
613 // Blockquote content is treated as paragraph text, so it's not
614 // included in the special blocks list here.
615 if !effective_config.paragraphs {
616 let is_special_block = is_heading_line_num(line_number)
617 || ctx.line_info(line_number).is_some_and(|info| info.in_code_block)
618 || table_lines_set.contains(&line_number)
619 || ctx.line_info(line_number).is_some_and(|info| info.in_html_block)
620 || ctx.line_info(line_number).is_some_and(|info| info.in_html_comment)
621 || ctx.line_info(line_number).is_some_and(|info| info.in_esm_block)
622 || ctx.line_info(line_number).is_some_and(|info| info.in_jsx_expression)
623 || ctx.line_info(line_number).is_some_and(|info| info.in_jsx_block)
624 || ctx.line_info(line_number).is_some_and(|info| info.in_mdx_comment)
625 || ctx
626 .line_info(line_number)
627 .is_some_and(super::super::lint_context::types::LineInfo::in_mkdocs_container);
628
629 // Skip regular paragraph text when paragraphs = false
630 if !is_special_block {
631 continue;
632 }
633 }
634
635 // Skip blockquote lines when blockquotes = false.
636 // Also skip lazy continuation lines that belong to a blockquote
637 // (lines without `>` prefix that follow a blockquote line).
638 if !effective_config.blockquotes {
639 if ctx.lines[line_number - 1].blockquote.is_some() {
640 continue;
641 }
642 // Check for lazy continuation: scan backwards through
643 // non-blank lines to find if this paragraph started with
644 // a blockquote marker
645 if !line.trim().is_empty() {
646 let mut scan = line_number.saturating_sub(2);
647 loop {
648 if ctx.lines[scan].blockquote.is_some() {
649 // Found a blockquote ancestor — this is a lazy continuation
650 continue 'line_loop;
651 }
652 if lines[scan].trim().is_empty() || scan == 0 {
653 break;
654 }
655 scan -= 1;
656 }
657 }
658 }
659
660 // Skip lines that are only a URL, image ref, or link ref
661 if self.should_ignore_line(line, lines, line_idx, ctx) {
662 continue;
663 }
664 }
665
666 // In sentence-per-line mode, check if this is a single long sentence
667 // If so, emit a warning without a fix (user must manually rephrase)
668 if effective_config.reflow_mode == ReflowMode::SentencePerLine {
669 let sentences = split_into_sentences(
670 line.trim(),
671 Some(&defined_references),
672 effective_config.require_sentence_capital,
673 );
674 if sentences.len() == 1 {
675 // Single sentence that's too long - warn but don't auto-fix
676 let message = format!("Line length {effective_length} exceeds {line_limit} characters");
677
678 let (start_line, start_col, end_line, end_col) =
679 calculate_excess_range(line_number, line, line_limit);
680
681 warnings.push(LintWarning {
682 rule_name: Some(self.name().to_string()),
683 message,
684 line: start_line,
685 column: start_col,
686 end_line,
687 end_column: end_col,
688 severity: Severity::Warning,
689 fix: None, // No auto-fix for long single sentences
690 });
691 continue;
692 }
693 // Multiple sentences will be handled by paragraph-based reflow
694 continue;
695 }
696
697 // In semantic-line-breaks mode, skip per-line checks —
698 // all reflow is handled at the paragraph level with cascading splits
699 if effective_config.reflow_mode == ReflowMode::SemanticLineBreaks {
700 continue;
701 }
702
703 // Don't provide fix for individual lines when reflow is enabled
704 // Paragraph-based fixes will be handled separately
705 let fix = None;
706
707 let message = format!("Line length {effective_length} exceeds {line_limit} characters");
708
709 // Calculate precise character range for the excess portion
710 let (start_line, start_col, end_line, end_col) = calculate_excess_range(line_number, line, line_limit);
711
712 warnings.push(LintWarning {
713 rule_name: Some(self.name().to_string()),
714 message,
715 line: start_line,
716 column: start_col,
717 end_line,
718 end_column: end_col,
719 severity: Severity::Warning,
720 fix,
721 });
722 }
723
724 // If reflow is enabled, generate paragraph-based fixes
725 if effective_config.reflow {
726 let paragraph_warnings = self.generate_paragraph_fixes(ctx, &effective_config, lines);
727 // Merge paragraph warnings with line warnings, removing duplicates
728 for mut pw in paragraph_warnings {
729 if ctx.flavor == crate::config::MarkdownFlavor::MDG
730 && (pw.line..=pw.end_line).any(|line_number| is_potential_mdg_step(ctx, line_number))
731 {
732 // Keep reporting the excessive line, but do not offer a fix
733 // whose replacement would split a Gherkin step across lines.
734 pw.fix = None;
735 }
736 // Remove any line warnings that overlap with this paragraph
737 warnings.retain(|w| w.line < pw.line || w.line > pw.end_line);
738 warnings.push(pw);
739 }
740 }
741
742 Ok(warnings)
743 }
744
745 fn fix(&self, ctx: &crate::lint_context::LintContext) -> Result<String, LintError> {
746 // For CLI usage, apply fixes from warnings
747 // LSP will use the warning-based fixes directly
748 let warnings = self.check(ctx)?;
749 let warnings =
750 crate::utils::fix_utils::filter_warnings_by_inline_config(warnings, ctx.inline_config(), self.name());
751
752 // If there are no fixes, return content unchanged
753 if !warnings.iter().any(|w| w.fix.is_some()) {
754 return Ok(ctx.content.to_string());
755 }
756
757 // Apply warning-based fixes
758 crate::utils::fix_utils::apply_warning_fixes(ctx.content, &warnings)
759 .map_err(|e| LintError::FixFailed(format!("Failed to apply fixes: {e}")))
760 }
761
762 fn as_any(&self) -> &dyn std::any::Any {
763 self
764 }
765
766 fn category(&self) -> RuleCategory {
767 RuleCategory::Whitespace
768 }
769
770 fn should_skip(&self, ctx: &crate::lint_context::LintContext) -> bool {
771 self.should_skip_with_config(ctx, &self.config)
772 }
773
774 crate::impl_rule_config_sections!(MD013Config);
775
776 fn config_aliases(&self) -> Option<std::collections::HashMap<String, String>> {
777 let mut aliases = std::collections::HashMap::new();
778 aliases.insert("enable_reflow".to_string(), "reflow".to_string());
779 aliases.insert("strict_sentences".to_string(), "require-sentence-capital".to_string());
780 aliases.insert("strict-sentences".to_string(), "require-sentence-capital".to_string());
781 // Kept in step with the `alias` attributes on `MD013Config::ignore_link_urls`.
782 // Serde accepts these spellings, so a config using one is honored; without
783 // them here the key validator reports a documented, working key as unknown.
784 aliases.insert(
785 "semantic-link-understanding".to_string(),
786 "ignore-link-urls".to_string(),
787 );
788 aliases.insert(
789 "semantic_link_understanding".to_string(),
790 "ignore-link-urls".to_string(),
791 );
792 Some(aliases)
793 }
794
795 fn from_config(config: &crate::config::Config) -> Box<dyn Rule>
796 where
797 Self: Sized,
798 {
799 let rule_config = MD013Config::from_document_config(config);
800 let mut rule = Self::from_config_struct(rule_config);
801 // Pull list-marker spacing from MD030 (via the shared serde config loader)
802 // so reflow rewrites list items with the configured spacing rather than a
803 // hard-coded single space.
804 rule.list_spacing = crate::rule_config_serde::load_rule_config::<MD030Config>(config);
805 Box::new(rule)
806 }
807}
808
809impl MD013LineLength {
810 /// True when `line_num` (1-indexed) sits inside a `$$` span covering more
811 /// than one line, as seen by the byte-level math parser.
812 fn line_in_multiline_math_span(&self, line_num: usize, ctx: &crate::lint_context::LintContext) -> bool {
813 ctx.math_spans()
814 .iter()
815 .any(|span| span.is_display && span.end_line > span.line && (span.line..=span.end_line).contains(&line_num))
816 }
817
818 /// True when `line_num` (1-indexed) holds nothing but display math, whether
819 /// that is one line of a multi-line block, a delimiter line, or a whole line
820 /// that is a single complete `$$...$$` span. This is what `math-blocks =
821 /// false` exempts from the length check.
822 ///
823 /// rumdl models math twice and the two models miss different containers, so
824 /// this consults both. `math_spans()` is byte-level and sees a block opened on
825 /// a list marker line (`- $$`), which the line-level map cannot because that
826 /// line does not begin with `$$`. `LineInfo::in_math_block` is line-level and
827 /// sees a four-space-indented block inside a footnote, which the byte-level
828 /// parser reads as an indented code block. Taking the union only ever adds
829 /// coverage: neither signal fires on ordinary prose, and an unmatched `$$`
830 /// opener is flagged by neither, so a stray delimiter cannot exempt the rest
831 /// of the document.
832 fn line_is_display_math(
833 &self,
834 line_num: usize,
835 ctx: &crate::lint_context::LintContext,
836 config: &MD013Config,
837 ) -> bool {
838 self.line_holds_only_multiline_math(line_num, ctx)
839 || ctx.line_info(line_num).is_some_and(|info| info.in_math_block)
840 || (config.bracket_display_math
841 && ctx
842 .bracket_display_math_lines()
843 .multiline
844 .get(line_num.saturating_sub(1))
845 .copied()
846 .unwrap_or(false))
847 || (config.bracket_display_math
848 && ctx
849 .bracket_display_math_lines()
850 .standalone
851 .get(line_num.saturating_sub(1))
852 .copied()
853 .unwrap_or(false))
854 }
855
856 /// True when a multi-line `$$` span covers `line_num` (1-indexed) and the line
857 /// holds nothing besides that math.
858 ///
859 /// A delimiter line can carry Markdown outside the delimiter: `$$ trailing
860 /// prose` closes a block and then continues in prose, and `leading prose $$`
861 /// opens one at the end of a sentence. That prose is ordinary text and counts
862 /// toward the line's length like any other, so exempting the whole line would
863 /// hide arbitrarily long prose behind a delimiter. The line-level map already
864 /// leaves such mixed lines unflagged; this is the byte-level half of the same
865 /// judgement.
866 fn line_holds_only_multiline_math(&self, line_num: usize, ctx: &crate::lint_context::LintContext) -> bool {
867 let Some(info) = ctx.line_info(line_num) else {
868 return false;
869 };
870 let line = info.content(ctx.content);
871
872 ctx.math_spans().iter().any(|span| {
873 if !span.is_display || span.end_line <= span.line || !(span.line..=span.end_line).contains(&line_num) {
874 return false;
875 }
876 if line_num == span.line {
877 let before = line.get(..span.byte_offset.saturating_sub(info.byte_offset));
878 if !before.is_none_or(|before| Self::only_structure_precedes_math(before, info)) {
879 return false;
880 }
881 }
882 if line_num == span.end_line {
883 let after = line.get(span.byte_end.saturating_sub(info.byte_offset)..);
884 if !after.is_none_or(|after| after.trim().is_empty()) {
885 return false;
886 }
887 }
888 true
889 })
890 }
891
892 /// True when the text before a block's opening delimiter is only the structure
893 /// the block sits in: indentation, a blockquote marker, or the list marker
894 /// introducing it. Such a block still owns its whole line.
895 fn only_structure_precedes_math(before: &str, info: &crate::lint_context::LineInfo) -> bool {
896 let after_marker =
897 crate::utils::blockquote::parse_blockquote_prefix(before).map_or(before, |prefix| prefix.content);
898 if after_marker.trim().is_empty() {
899 return true;
900 }
901 info.list_item.as_ref().is_some_and(|item| {
902 let mut chars = before.chars();
903 chars.by_ref().take(item.content_column).count() == item.content_column && chars.as_str().trim().is_empty()
904 })
905 }
906
907 /// True when `line_num` (1-indexed) falls inside a display-math block that
908 /// spans more than one line.
909 ///
910 /// Line breaks carry meaning inside such a block: a TeX `%` comment runs to
911 /// the end of its line, so joining the lines pulls whatever followed on later
912 /// lines into the comment and drops it from the rendered equation, which can
913 /// also leave an environment unclosed. Reflow therefore leaves these blocks
914 /// alone regardless of the `math_blocks` setting, which governs only whether
915 /// their length is reported.
916 ///
917 /// This is `line_is_display_math` minus the case of a whole line that is one
918 /// complete `$$...$$` span, which
919 /// [`crate::utils::text_reflow::is_self_contained_display_math_line`]
920 /// recognizes. Such a line is a block of its own: it has no internal line
921 /// breaks to lose, and reflow keeps it on the line it was written on
922 /// through that recognizer rather than through this one.
923 fn line_in_multiline_math_block(
924 &self,
925 line_num: usize,
926 ctx: &crate::lint_context::LintContext,
927 config: &MD013Config,
928 ) -> bool {
929 self.line_in_multiline_math_span(line_num, ctx)
930 || ctx.line_info(line_num).is_some_and(|info| {
931 info.in_math_block && !is_self_contained_display_math_line(info.content(ctx.content))
932 })
933 || (config.bracket_display_math
934 && ctx
935 .bracket_display_math_lines()
936 .multiline
937 .get(line_num.saturating_sub(1))
938 .copied()
939 .unwrap_or(false))
940 }
941
942 fn line_is_standalone_bracket_math(
943 &self,
944 line_num: usize,
945 ctx: &crate::lint_context::LintContext,
946 config: &MD013Config,
947 ) -> bool {
948 config.bracket_display_math
949 && ctx
950 .bracket_display_math_lines()
951 .standalone
952 .get(line_num.saturating_sub(1))
953 .copied()
954 .unwrap_or(false)
955 }
956
957 /// True when `line_num` (1-based) sits inside a structure whose lines must be
958 /// preserved verbatim (code block, front matter, HTML/JSX/MDX block, MkDocs
959 /// container, div marker, multi-line math block, ...). Used to keep blockquote
960 /// reflow from touching quoted-looking text embedded in such structures.
961 fn line_in_verbatim_context(
962 &self,
963 line_num: usize,
964 ctx: &crate::lint_context::LintContext,
965 config: &MD013Config,
966 ) -> bool {
967 if self.line_in_multiline_math_block(line_num, ctx, config) {
968 return true;
969 }
970 ctx.line_info(line_num).is_some_and(|info| {
971 info.in_code_block
972 || info.in_front_matter
973 || info.in_html_block
974 || info.in_html_comment
975 || info.in_esm_block
976 || info.in_jsx_expression
977 || info.in_jsx_block
978 || info.in_mdx_comment
979 || info.in_mkdocstrings
980 || info.in_pymdown_block
981 || info.in_mkdocs_container()
982 || info.is_div_marker
983 })
984 }
985
986 fn is_blockquote_content_boundary(
987 &self,
988 content: &str,
989 line_num: usize,
990 ctx: &crate::lint_context::LintContext,
991 config: &MD013Config,
992 ) -> bool {
993 let trimmed = content.trim();
994
995 trimmed.is_empty()
996 || self.line_in_verbatim_context(line_num, ctx, config)
997 || trimmed.starts_with('#')
998 || trimmed.starts_with("```")
999 || trimmed.starts_with("~~~")
1000 || trimmed.starts_with('>')
1001 || TableUtils::is_potential_table_row_with_flavor(content, ctx.flavor)
1002 || is_list_item(trimmed)
1003 || is_horizontal_rule(content)
1004 // A setext underline ends the quoted paragraph. The text decides
1005 // it: under quoted paragraph text the run is an underline, a dash
1006 // run anywhere else is a thematic break, and stopping at an equals
1007 // run that is neither only leaves text unreflowed.
1008 || is_setext_underline_content(content)
1009 || (trimmed.starts_with('[') && content.contains("]:"))
1010 || is_template_directive_only(content)
1011 || is_standalone_attr_list(content)
1012 || is_snippet_block_delimiter(content)
1013 || is_github_alert_marker(trimmed)
1014 || is_html_only_line(content)
1015 || self.line_is_standalone_bracket_math(line_num, ctx, config)
1016 || standalone_link_ends_paragraph(ctx, line_num, config)
1017 }
1018
1019 #[allow(clippy::too_many_arguments)]
1020 fn generate_blockquote_paragraph_fix(
1021 &self,
1022 ctx: &crate::lint_context::LintContext,
1023 config: &MD013Config,
1024 lines: &[&str],
1025 start_idx: usize,
1026 line_ending: &str,
1027 // Extra indent (spaces) to prepend to the emitted `>` prefix so a blockquote
1028 // nested in a list item moves with its parent's widened marker. Zero unless a
1029 // non-default MD030 widened an ancestor list item.
1030 ancestor_shift: isize,
1031 ) -> (Option<LintWarning>, usize) {
1032 let Some(start_bq) = ctx.lines.get(start_idx).and_then(|line| line.blockquote.as_deref()) else {
1033 return (None, start_idx + 1);
1034 };
1035 let target_level = start_bq.nesting_level;
1036 let defined_references = Self::defined_reference_labels(ctx);
1037
1038 let mut collected: Vec<CollectedBlockquoteLine> = Vec::new();
1039 let mut i = start_idx;
1040
1041 while i < lines.len() {
1042 if !collected.is_empty() && has_hard_break(&collected[collected.len() - 1].data.content) {
1043 break;
1044 }
1045
1046 let line_num = i + 1;
1047 if line_num > ctx.lines.len() {
1048 break;
1049 }
1050
1051 if lines[i].trim().is_empty() {
1052 break;
1053 }
1054
1055 let line_bq = ctx.lines[i].blockquote.as_deref();
1056 if let Some(bq) = line_bq {
1057 if bq.nesting_level != target_level {
1058 break;
1059 }
1060
1061 if self.is_blockquote_content_boundary(&bq.content, line_num, ctx, config) {
1062 break;
1063 }
1064
1065 collected.push(CollectedBlockquoteLine {
1066 line_idx: i,
1067 data: BlockquoteLineData::explicit(trim_preserving_hard_break(&bq.content), bq.prefix.clone()),
1068 });
1069 i += 1;
1070 continue;
1071 }
1072
1073 let lazy_content = lines[i].trim_start();
1074 if self.is_blockquote_content_boundary(lazy_content, line_num, ctx, config) {
1075 break;
1076 }
1077
1078 collected.push(CollectedBlockquoteLine {
1079 line_idx: i,
1080 data: BlockquoteLineData::lazy(trim_preserving_hard_break(lazy_content)),
1081 });
1082 i += 1;
1083 }
1084
1085 if collected.is_empty() {
1086 return (None, start_idx + 1);
1087 }
1088
1089 let next_idx = i;
1090 let paragraph_start = collected[0].line_idx;
1091 let end_line = collected[collected.len() - 1].line_idx;
1092 let line_data: Vec<BlockquoteLineData> = collected.iter().map(|l| l.data.clone()).collect();
1093 let paragraph_text = join_soft_break_lines(
1094 &line_data.iter().map(|d| d.content.as_str()).collect::<Vec<_>>(),
1095 config.cjk_soft_break,
1096 );
1097
1098 // A colon-led line with a line of the paragraph before it opens a
1099 // definition, and joining the lines would flatten the definition list
1100 // into prose. The paragraph's first line is prose whatever it starts
1101 // with, since a definition needs a term on the line before it.
1102 let contains_definition_list = line_data
1103 .iter()
1104 .skip(1)
1105 .any(|d| crate::utils::text_reflow::is_definition_list_marker(&d.content));
1106 if contains_definition_list || holds_definition_list(ctx, paragraph_start, end_line) {
1107 return (None, next_idx);
1108 }
1109
1110 let contains_snippets = line_data.iter().any(|d| is_snippet_block_delimiter(&d.content));
1111 if contains_snippets {
1112 return (None, next_idx);
1113 }
1114
1115 let needs_reflow = match config.reflow_mode {
1116 ReflowMode::Normalize => {
1117 self.normalize_mode_needs_reflow(line_data.iter().map(|d| d.content.as_str()), config)
1118 }
1119 ReflowMode::SentencePerLine => {
1120 let sentences = split_into_sentences(
1121 ¶graph_text,
1122 Some(&defined_references),
1123 config.require_sentence_capital,
1124 );
1125 sentences.len() > 1 || line_data.len() > 1
1126 }
1127 ReflowMode::SemanticLineBreaks => {
1128 let sentences = split_into_sentences(
1129 ¶graph_text,
1130 Some(&defined_references),
1131 config.require_sentence_capital,
1132 );
1133 sentences.len() > 1
1134 || line_data.len() > 1
1135 || collected
1136 .iter()
1137 .any(|l| self.calculate_effective_length(lines[l.line_idx]) > config.line_length.get())
1138 }
1139 ReflowMode::Default => collected
1140 .iter()
1141 .any(|l| self.calculate_effective_length(lines[l.line_idx]) > config.line_length.get()),
1142 };
1143
1144 if !needs_reflow {
1145 return (None, next_idx);
1146 }
1147
1148 let fallback_prefix = start_bq.prefix.clone();
1149 let explicit_prefix = dominant_blockquote_prefix(&line_data, &fallback_prefix);
1150 // Shift the whole quote right to track a widened parent list item's content
1151 // column (only widening matters for nesting; a narrowed parent leaves the quote
1152 // harmlessly over-indented, which MD027/MD030 tidy).
1153 let explicit_prefix = if ancestor_shift > 0 {
1154 format!("{}{explicit_prefix}", " ".repeat(ancestor_shift as usize))
1155 } else {
1156 explicit_prefix
1157 };
1158 let continuation_style = blockquote_continuation_style(&line_data);
1159
1160 let reflow_line_length = if config.line_length.is_unlimited() {
1161 usize::MAX
1162 } else {
1163 config
1164 .line_length
1165 .get()
1166 .saturating_sub(self.calculate_string_length(&explicit_prefix))
1167 .max(1)
1168 };
1169
1170 let reflow_options = Self::reflow_options(ctx, config, reflow_line_length);
1171
1172 let reflowed_with_style =
1173 reflow_blockquote_content(&line_data, &explicit_prefix, continuation_style, &reflow_options);
1174
1175 if reflowed_with_style.is_empty() {
1176 return (None, next_idx);
1177 }
1178
1179 let reflowed_text = reflowed_with_style.join(line_ending);
1180
1181 let start_range = ctx.whole_line_byte_range(paragraph_start + 1);
1182 let end_range = if end_line == lines.len() - 1 && !ctx.content.ends_with('\n') {
1183 ctx.line_text_byte_range(end_line + 1, 1, lines[end_line].len() + 1)
1184 } else {
1185 ctx.whole_line_byte_range(end_line + 1)
1186 };
1187 let byte_range = start_range.start..end_range.end;
1188
1189 let replacement = if end_line < lines.len() - 1 || ctx.content.ends_with('\n') {
1190 format!("{reflowed_text}{line_ending}")
1191 } else {
1192 reflowed_text
1193 };
1194
1195 let original_text = &ctx.content[byte_range.clone()];
1196 if original_text == replacement {
1197 return (None, next_idx);
1198 }
1199
1200 let (warning_line, warning_end_line) = match config.reflow_mode {
1201 ReflowMode::Normalize => (paragraph_start + 1, end_line + 1),
1202 ReflowMode::SentencePerLine | ReflowMode::SemanticLineBreaks => (paragraph_start + 1, end_line + 1),
1203 ReflowMode::Default => {
1204 let violating_line = collected
1205 .iter()
1206 .find(|line| self.calculate_effective_length(lines[line.line_idx]) > config.line_length.get())
1207 .map_or(paragraph_start + 1, |line| line.line_idx + 1);
1208 (violating_line, violating_line)
1209 }
1210 };
1211
1212 let warning = LintWarning {
1213 rule_name: Some(self.name().to_string()),
1214 message: match config.reflow_mode {
1215 ReflowMode::Normalize => format!(
1216 "Paragraph could be normalized to use line length of {} characters",
1217 config.line_length.get()
1218 ),
1219 ReflowMode::SentencePerLine => {
1220 let num_sentences = split_into_sentences(
1221 ¶graph_text,
1222 Some(&defined_references),
1223 config.require_sentence_capital,
1224 )
1225 .len();
1226 if line_data.len() == 1 {
1227 format!("Line contains {num_sentences} sentences (one sentence per line required)")
1228 } else {
1229 let num_lines = line_data.len();
1230 format!(
1231 "Paragraph should have one sentence per line (found {num_sentences} sentences across {num_lines} lines)"
1232 )
1233 }
1234 }
1235 ReflowMode::SemanticLineBreaks => {
1236 let num_sentences = split_into_sentences(
1237 ¶graph_text,
1238 Some(&defined_references),
1239 config.require_sentence_capital,
1240 )
1241 .len();
1242 format!("Paragraph should use semantic line breaks ({num_sentences} sentences)")
1243 }
1244 ReflowMode::Default => format!("Line length exceeds {} characters", config.line_length.get()),
1245 },
1246 line: warning_line,
1247 column: 1,
1248 end_line: warning_end_line,
1249 end_column: lines[warning_end_line.saturating_sub(1)].chars().count() + 1,
1250 severity: Severity::Warning,
1251 fix: Some(crate::rule::Fix::new(byte_range, replacement)),
1252 };
1253
1254 (Some(warning), next_idx)
1255 }
1256
1257 /// Reflow a single list item that lives inside a blockquote.
1258 ///
1259 /// The blockquote paragraph reflow treats a list marker as a content boundary,
1260 /// so `> - long item ...` is never wrapped even though the identical top-level
1261 /// item is. This handles that case: it collects one tight, prose-only list item
1262 /// at the starting blockquote level, reflows the item body to the configured
1263 /// width, and re-emits it with the blockquote prefix preserved and continuation
1264 /// lines aligned under the list content.
1265 ///
1266 /// Sibling and nested list items end collection and are reflowed independently
1267 /// by the caller's outer loop (a nested item carries its indent folded into the
1268 /// blockquote prefix, so it reflows correctly on its own). Items that are not
1269 /// simple tight prose - those embedding a code block, table, fence, or hard
1270 /// break - are left untouched (`None`) and the cursor still advances past the
1271 /// whole item so its inner lines are never reprocessed as loose prose.
1272 #[allow(clippy::too_many_arguments)]
1273 fn generate_blockquote_list_item_fix(
1274 &self,
1275 ctx: &crate::lint_context::LintContext,
1276 config: &MD013Config,
1277 lines: &[&str],
1278 start_idx: usize,
1279 line_ending: &str,
1280 // Extra indent (spaces) to prepend to the emitted `>` prefix when this quoted
1281 // list lives inside a list item whose marker widened. Zero unless a non-default
1282 // MD030 widened an ancestor list item.
1283 ancestor_shift: isize,
1284 // One entry per document line, true where a code span crosses one of the
1285 // line's boundaries. The caller reads them off one pass over the document
1286 // and shares them across every item, so an item costs no pass of its own.
1287 code_span_touches: &[bool],
1288 ) -> (Option<LintWarning>, usize) {
1289 use crate::utils::blockquote::effective_indent_in_blockquote;
1290
1291 let Some(start_bq) = ctx.lines.get(start_idx).and_then(|line| line.blockquote.as_deref()) else {
1292 return (None, start_idx + 1);
1293 };
1294
1295 let defined_references = Self::defined_reference_labels(ctx);
1296
1297 // A `>`-prefixed line can be marked as a blockquote even inside a fenced code
1298 // block (or other verbatim structure); such content must never be reflowed.
1299 if self.line_in_verbatim_context(start_idx + 1, ctx, config) {
1300 return (None, start_idx + 1);
1301 }
1302
1303 let target_level = start_bq.nesting_level;
1304
1305 // The marker line carries the canonical blockquote prefix: its content begins
1306 // with the list marker, so no list indent has been folded into the prefix.
1307 // Track a widened parent list item's content column so the quote stays nested
1308 // (only widening can detach it; narrowing just over-indents).
1309 let bq_prefix = if ancestor_shift > 0 {
1310 format!("{}{}", " ".repeat(ancestor_shift as usize), start_bq.prefix)
1311 } else {
1312 start_bq.prefix.clone()
1313 };
1314
1315 // A thematic break opens with what looks like a bullet marker (`- - -`).
1316 // It is not a list item, and reflowing it as prose destroys the break.
1317 // The top-level reflow path applies the same exemption.
1318 if is_horizontal_rule(&start_bq.content) {
1319 return (None, start_idx + 1);
1320 }
1321
1322 let (marker, first_body) = extract_list_marker_and_content(&start_bq.content);
1323 if marker.is_empty() {
1324 return (None, start_idx + 1);
1325 }
1326 let marker_width = marker.chars().count();
1327
1328 // Continuation lines of a checkbox item align under the bullet+checkbox, but
1329 // are recognized from the bullet width, matching the top-level list reflow.
1330 let base_marker_width = ["[ ] ", "[x] ", "[X] "]
1331 .iter()
1332 .find_map(|cb| marker.find(*cb))
1333 .unwrap_or(marker_width);
1334
1335 // Collect the item: the marker line plus its tight prose continuation lines.
1336 // `end_idx` always tracks the last consumed line so the cursor advances past
1337 // the entire item, even when it turns out to be too complex to reflow safely.
1338 let first_piece = trim_preserving_hard_break(&first_body);
1339 let mut simple = !has_hard_break(&first_piece);
1340 let mut body_pieces: Vec<String> = vec![first_piece];
1341 let mut end_idx = start_idx;
1342 let mut i = start_idx + 1;
1343
1344 while i < lines.len() {
1345 let Some(bq) = ctx.lines[i].blockquote.as_deref() else {
1346 // A blank line ends the item.
1347 if lines[i].trim().is_empty() {
1348 break;
1349 }
1350 // A lazy continuation (no `>` marker) is too ambiguous to reflow
1351 // safely. Consume the whole lazy run into this item's span and leave
1352 // the item untouched, so the caller does not reflow the continuation
1353 // in isolation and leave the marker line partially fixed.
1354 simple = false;
1355 while i < lines.len() && ctx.lines[i].blockquote.is_none() && !lines[i].trim().is_empty() {
1356 end_idx = i;
1357 i += 1;
1358 }
1359 break;
1360 };
1361 if bq.nesting_level != target_level {
1362 break;
1363 }
1364
1365 let content = bq.content.as_str();
1366 if content.trim().is_empty() {
1367 // Blank quoted line ends the tight paragraph. A following indented
1368 // paragraph (loose item) is reflowed on its own by the prose path.
1369 break;
1370 }
1371
1372 let eff_indent = effective_indent_in_blockquote(lines[i], target_level, 0);
1373 if eff_indent < base_marker_width {
1374 // Dedented: a sibling list item or text outside this item. Stop here
1375 // and let the outer loop classify it.
1376 break;
1377 }
1378 if is_list_item(content) {
1379 // A nested list item: its own item, handled independently.
1380 break;
1381 }
1382
1383 // An embedded structure (code block, table, fence, nested quote, ...)
1384 // means the item is not simple prose: keep consuming so the cursor clears
1385 // the whole structure, but do not produce a fix.
1386 if self.is_blockquote_content_boundary(content, i + 1, ctx, config) {
1387 simple = false;
1388 }
1389
1390 let piece = trim_preserving_hard_break(content);
1391 if has_hard_break(&piece) {
1392 simple = false;
1393 }
1394 body_pieces.push(piece);
1395 end_idx = i;
1396 i += 1;
1397 }
1398
1399 let next_idx = end_idx + 1;
1400
1401 if !simple || holds_definition_list(ctx, start_idx, end_idx) {
1402 return (None, next_idx);
1403 }
1404
1405 let exceeds_limit =
1406 || (start_idx..=end_idx).any(|idx| self.calculate_effective_length(lines[idx]) > config.line_length.get());
1407 let body_text = join_soft_break_lines(&body_pieces, config.cjk_soft_break);
1408 let body_text = body_text.trim();
1409
1410 // A body line that is one whole `$$...$$` expression renders as a display
1411 // block, so it holds a line of its own and the prose on either side of it
1412 // is reflowed separately. Each segment carries whether it is that line.
1413 // A line touched by a code span crossing one of its boundaries is code,
1414 // not such a block. The pieces are the consecutive lines from
1415 // `start_idx`, one piece each.
1416 let body_segments: Vec<(bool, Vec<&str>)> = {
1417 let mut segments: Vec<(bool, Vec<&str>)> = Vec::new();
1418 let mut current: Vec<&str> = Vec::new();
1419 for (offset, piece) in body_pieces.iter().enumerate() {
1420 if (is_self_contained_display_math_line(piece)
1421 || self.line_is_standalone_bracket_math(start_idx + offset + 1, ctx, config))
1422 && !line_touches_multiline_code_span(code_span_touches, start_idx + offset + 1)
1423 {
1424 if !current.is_empty() {
1425 segments.push((false, std::mem::take(&mut current)));
1426 }
1427 segments.push((true, vec![piece.trim_start()]));
1428 } else {
1429 current.push(piece.as_str());
1430 }
1431 }
1432 if !current.is_empty() {
1433 segments.push((false, current));
1434 }
1435 segments
1436 };
1437 let holds_display_math = body_segments.iter().any(|(is_math, _)| *is_math);
1438
1439 // Some bodies cannot be shortened and must stay verbatim, matching the
1440 // exemptions the top-level list reflow applies: link reference definitions
1441 // always, and (in non-strict mode) standalone links/images and HTML-only
1442 // lines. Reflowing a link reference definition would split it after the
1443 // colon/URL and break the definition.
1444 let is_link_ref_def =
1445 body_text.starts_with('[') && body_text.contains("]:") && LINK_REF_PATTERN.is_match(body_text);
1446 let raw_marker_line = lines[start_idx];
1447 let body_is_unwrappable = is_link_ref_def
1448 || standalone_link_ends_paragraph(ctx, start_idx + 1, config)
1449 || (!config.strict && is_html_only_line(raw_marker_line));
1450 if body_is_unwrappable {
1451 return (None, next_idx);
1452 }
1453
1454 let needs_reflow = match config.reflow_mode {
1455 ReflowMode::Normalize => body_pieces.len() > 1 || exceeds_limit(),
1456 ReflowMode::Default => exceeds_limit(),
1457 ReflowMode::SentencePerLine => {
1458 split_into_sentences(body_text, Some(&defined_references), config.require_sentence_capital).len() > 1
1459 || body_pieces.len() > 1
1460 }
1461 ReflowMode::SemanticLineBreaks => {
1462 split_into_sentences(body_text, Some(&defined_references), config.require_sentence_capital).len() > 1
1463 || exceeds_limit()
1464 }
1465 };
1466 if !needs_reflow {
1467 return (None, next_idx);
1468 }
1469
1470 // Apply MD030 list-marker spacing in the spacing-normalizing modes, mirroring
1471 // the top-level list reflow: derive the post-marker spacing from MD030 and let
1472 // the continuation indent follow the resulting content width. Default MD030 (a
1473 // single space) leaves the marker unchanged. MkDocs keeps its rigid indent.
1474 let (marker, marker_width) = if matches!(config.reflow_mode, ReflowMode::Default | ReflowMode::Normalize)
1475 && !ctx.flavor.requires_strict_list_indent()
1476 {
1477 let is_ordered = marker.starts_with(|c: char| c.is_ascii_digit());
1478 // Bullet/number portion only (e.g. `-`, `1.`); the checkbox is content.
1479 let bullet = marker.split(' ').next().unwrap_or("");
1480 let bullet_len = bullet.chars().count();
1481 let checkbox_tail = &marker[base_marker_width..];
1482 // Decide single- vs multi-line spacing from the rewritten shape. This path
1483 // only handles a single tight prose paragraph (structural or multi-paragraph
1484 // items set `simple = false` and bail out above), so the emitted item stays
1485 // multi-line solely when the joined body wraps past one line. A multi-line
1486 // *source* that collapses onto the marker line must use MD030's single-line
1487 // spacing, matching the top-level reflow. The wrap test measures at the
1488 // single-line content column so it does not depend on the spacing chosen here.
1489 let single_col = self.calculate_string_length(&bq_prefix)
1490 + bullet_len
1491 + self.list_spacing.expected_spaces(is_ordered, false, bullet_len)
1492 + checkbox_tail.chars().count();
1493 // A display-math line always holds a line of its own, so an item that
1494 // carries one spans several lines whatever the joined body measures.
1495 let is_multi = holds_display_math
1496 || (!body_text.is_empty()
1497 && self.calculate_effective_length(&format!("{}{body_text}", " ".repeat(single_col)))
1498 > config.line_length.effective_limit());
1499 let spaces = self.list_spacing.expected_spaces(is_ordered, is_multi, bullet_len);
1500 let new_marker = format!("{bullet}{}{checkbox_tail}", " ".repeat(spaces));
1501 let width = new_marker.chars().count();
1502 (new_marker, width)
1503 } else {
1504 (marker, marker_width)
1505 };
1506
1507 let prefix_width = self.calculate_string_length(&bq_prefix) + self.calculate_string_length(&marker);
1508 let reflow_line_length = if config.line_length.is_unlimited() {
1509 usize::MAX
1510 } else {
1511 config.line_length.get().saturating_sub(prefix_width).max(1)
1512 };
1513
1514 let reflow_options = Self::reflow_options(ctx, config, reflow_line_length);
1515
1516 // A display-math segment is emitted as written; a prose segment is joined
1517 // and reflowed on its own, so the prose above and below the expression
1518 // wraps within its own paragraph.
1519 let mut reflowed: Vec<String> = Vec::new();
1520 for (is_math, segment) in &body_segments {
1521 if *is_math {
1522 reflowed.push(segment[0].to_string());
1523 continue;
1524 }
1525 let segment_text = join_soft_break_lines(segment, config.cjk_soft_break);
1526 let segment_text = segment_text.trim();
1527 if segment_text.is_empty() {
1528 continue;
1529 }
1530 reflowed.extend(crate::utils::text_reflow::reflow_line(segment_text, &reflow_options));
1531 }
1532 if reflowed.is_empty() {
1533 return (None, next_idx);
1534 }
1535
1536 let continuation_indent = " ".repeat(marker_width);
1537 let reflowed_text = reflowed
1538 .iter()
1539 .enumerate()
1540 .map(|(idx, line)| {
1541 if idx == 0 {
1542 format!("{bq_prefix}{marker}{line}")
1543 } else {
1544 format!("{bq_prefix}{continuation_indent}{line}")
1545 }
1546 })
1547 .collect::<Vec<_>>()
1548 .join(line_ending);
1549
1550 let start_range = ctx.whole_line_byte_range(start_idx + 1);
1551 let end_range = if end_idx == lines.len() - 1 && !ctx.content.ends_with('\n') {
1552 ctx.line_text_byte_range(end_idx + 1, 1, lines[end_idx].len() + 1)
1553 } else {
1554 ctx.whole_line_byte_range(end_idx + 1)
1555 };
1556 let byte_range = start_range.start..end_range.end;
1557
1558 let replacement = if end_idx < lines.len() - 1 || ctx.content.ends_with('\n') {
1559 format!("{reflowed_text}{line_ending}")
1560 } else {
1561 reflowed_text
1562 };
1563
1564 let original_text = &ctx.content[byte_range.clone()];
1565 if original_text == replacement {
1566 return (None, next_idx);
1567 }
1568
1569 let message = match config.reflow_mode {
1570 ReflowMode::Normalize => format!(
1571 "Paragraph could be normalized to use line length of {} characters",
1572 config.line_length.get()
1573 ),
1574 ReflowMode::SentencePerLine => {
1575 let num_sentences =
1576 split_into_sentences(body_text, Some(&defined_references), config.require_sentence_capital).len();
1577 format!("List item should have one sentence per line (found {num_sentences} sentences)")
1578 }
1579 ReflowMode::SemanticLineBreaks => {
1580 let num_sentences =
1581 split_into_sentences(body_text, Some(&defined_references), config.require_sentence_capital).len();
1582 format!("List item should use semantic line breaks ({num_sentences} sentences)")
1583 }
1584 ReflowMode::Default => format!("Line length exceeds {} characters", config.line_length.get()),
1585 };
1586
1587 let warning = LintWarning {
1588 rule_name: Some(self.name().to_string()),
1589 message,
1590 line: start_idx + 1,
1591 column: 1,
1592 end_line: end_idx + 1,
1593 end_column: lines[end_idx].chars().count() + 1,
1594 severity: Severity::Warning,
1595 fix: Some(crate::rule::Fix::new(byte_range, replacement)),
1596 };
1597
1598 (Some(warning), next_idx)
1599 }
1600
1601 /// Generate paragraph-based fixes
1602 fn generate_paragraph_fixes(
1603 &self,
1604 ctx: &crate::lint_context::LintContext,
1605 config: &MD013Config,
1606 lines: &[&str],
1607 ) -> Vec<LintWarning> {
1608 let mut warnings = Vec::new();
1609 let defined_references = Self::defined_reference_labels(ctx);
1610 // A line touched by a code span crossing one of its boundaries is code
1611 // however it is spelled, so a `$$...$$` expression on one is no display
1612 // block.
1613 let code_span_touches = crate::utils::text_reflow::lines_touching_multiline_code_span(ctx.content);
1614
1615 // Detect the content's line ending style to preserve it in replacements.
1616 // The LSP receives content from editors which may use CRLF (Windows).
1617 // Replacements must match the original line endings to avoid false positives.
1618 let line_ending = crate::utils::line_ending::detect_line_ending(ctx.content);
1619
1620 // Ancestor list-item indent shifts, innermost last. When a reflowed parent's
1621 // marker widens under a non-default MD030 (e.g. ul-multi = 3 moves the parent's
1622 // content from column 2 to 4), its nested list/blockquote children are reflowed
1623 // independently and would otherwise keep their original indent — leaving them
1624 // under the parent's new content column, where a CommonMark parser reparses them
1625 // as siblings rather than children. Each frame is (normalized marker width,
1626 // cumulative shift applied to that item's content column); a descendant adds its
1627 // innermost open ancestor's shift to its own indent so the whole subtree moves
1628 // together. With a default MD030 and no marker padding every shift is 0, so this
1629 // is inert and the output is byte-identical.
1630 let mut list_shift_stack: Vec<(usize, isize)> = Vec::new();
1631
1632 let mut i = 0;
1633 while i < lines.len() {
1634 let line_num = i + 1;
1635
1636 // Close ancestor frames whose list item has ended at this line: a non-blank
1637 // line indented less than the frame's source content column is no longer
1638 // inside that item. Blank lines alone don't close a (loose) list item.
1639 if !list_shift_stack.is_empty()
1640 && let Some(info) = ctx.lines.get(i)
1641 && !info.is_blank
1642 {
1643 while let Some(&(content_column, _)) = list_shift_stack.last() {
1644 if info.indent < content_column {
1645 list_shift_stack.pop();
1646 } else {
1647 break;
1648 }
1649 }
1650 }
1651
1652 // Handle blockquote paragraphs with style-preserving reflow.
1653 // Skip blockquotes when blockquotes=false or paragraphs=false
1654 if line_num > 0 && line_num <= ctx.lines.len() && ctx.lines[line_num - 1].blockquote.is_some() {
1655 if !config.blockquotes || !config.paragraphs {
1656 // Skip past all blockquote lines (explicit and lazy continuations).
1657 // A lazy continuation is a non-blank line without `>` that follows
1658 // a blockquote line and isn't a structural element.
1659 let mut saw_explicit_bq = false;
1660 while i < lines.len() && i < ctx.lines.len() {
1661 if ctx.lines[i].blockquote.is_some() {
1662 saw_explicit_bq = true;
1663 i += 1;
1664 } else if saw_explicit_bq
1665 && !lines[i].trim().is_empty()
1666 && !lines[i].trim_start().starts_with('#')
1667 && !lines[i].trim_start().starts_with('>')
1668 {
1669 // Lazy continuation of preceding blockquote
1670 i += 1;
1671 } else {
1672 break;
1673 }
1674 }
1675 continue;
1676 }
1677 // A blockquote nested in a list item moves with its parent when the
1678 // parent's marker widens (see `list_shift_stack`); pass that shift so the
1679 // emitted `>` prefix lands under the parent's new content column instead
1680 // of detaching into a sibling.
1681 let ancestor_shift = list_shift_stack.last().map_or(0isize, |&(_, shift)| shift);
1682 // A list item inside the blockquote needs list-aware reflow (marker +
1683 // continuation indent); plain prose goes through the paragraph path.
1684 let is_bq_list_item = ctx.lines[i]
1685 .blockquote
1686 .as_deref()
1687 .is_some_and(|bq| is_list_item(&bq.content));
1688 let (warning, next_idx) = if is_bq_list_item {
1689 self.generate_blockquote_list_item_fix(
1690 ctx,
1691 config,
1692 lines,
1693 i,
1694 line_ending,
1695 ancestor_shift,
1696 &code_span_touches,
1697 )
1698 } else {
1699 self.generate_blockquote_paragraph_fix(ctx, config, lines, i, line_ending, ancestor_shift)
1700 };
1701 if let Some(warning) = warning {
1702 warnings.push(warning);
1703 }
1704 i = next_idx;
1705 continue;
1706 }
1707
1708 // Skip special structures (but NOT MkDocs containers - those get special handling)
1709 let should_skip_due_to_line_info = ctx.line_info(line_num).is_some_and(|info| {
1710 info.in_code_block
1711 || info.in_front_matter
1712 || info.in_html_block
1713 || info.in_html_comment
1714 || info.in_esm_block
1715 || info.in_jsx_expression
1716 || info.in_jsx_block
1717 || info.in_mdx_comment
1718 || info.in_mkdocstrings
1719 || info.in_pymdown_block
1720 });
1721
1722 // Skip link reference definitions but NOT footnote definitions.
1723 // Footnote definitions (`[^id]: prose`) contain reflowable text,
1724 // while link reference definitions (`[ref]: URL`) contain URLs
1725 // that cannot be shortened.
1726 let is_link_ref_def =
1727 lines[i].trim().starts_with('[') && !lines[i].trim().starts_with("[^") && lines[i].contains("]:");
1728
1729 // A setext heading is a heading, not a paragraph: skip every line of
1730 // its text and its underline together, the way an ATX heading is
1731 // skipped just below. Reflowing any part rewrites the document's
1732 // structure - joining the underline onto the text demotes the
1733 // heading to prose, and rewrapping the text moves words across the
1734 // heading boundary. The text spans the whole paragraph the underline
1735 // ends, so walk the flag to the end of it; the line after the last
1736 // text line is the underline.
1737 if is_setext_heading_text_line(ctx, line_num) {
1738 while i < lines.len() && is_setext_heading_text_line(ctx, i + 1) {
1739 i += 1;
1740 }
1741 i += 1;
1742 continue;
1743 }
1744
1745 if should_skip_due_to_line_info
1746 || lines[i].trim().starts_with('#')
1747 || TableUtils::is_potential_table_row_with_flavor(lines[i], ctx.flavor)
1748 || lines[i].trim().is_empty()
1749 || is_horizontal_rule(lines[i])
1750 || is_template_directive_only(lines[i])
1751 || is_link_ref_def
1752 || ctx.line_info(line_num).is_some_and(|info| info.is_div_marker)
1753 || is_html_only_line(lines[i])
1754 || standalone_link_ends_paragraph(ctx, line_num, config)
1755 {
1756 i += 1;
1757 continue;
1758 }
1759
1760 // Handle footnote definitions: `[^id]: prose text that can be reflowed`
1761 // Supports multi-paragraph footnotes with code blocks, blockquotes,
1762 // tables, and lists preserved verbatim.
1763 // Validate structure: must start with `[^`, contain `]:`, and the ID
1764 // must not contain `[` or `]` (prevents false matches on nested brackets)
1765 if lines[i].trim().starts_with("[^") && lines[i].contains("]:") && {
1766 let after_caret = &lines[i].trim()[2..];
1767 after_caret
1768 .find("]:")
1769 .is_some_and(|pos| pos > 0 && !after_caret[..pos].contains(['[', ']']))
1770 } {
1771 let footnote_start = i;
1772 let line = lines[i];
1773
1774 // Extract the prefix `[^id]:`
1775 let Some(colon_pos) = line.find("]:") else {
1776 i += 1;
1777 continue;
1778 };
1779 let prefix_end = colon_pos + 2;
1780 let prefix = &line[..prefix_end];
1781
1782 // Content starts after `]: ` (with optional space)
1783 let content_start = if line[prefix_end..].starts_with(' ') {
1784 prefix_end + 1
1785 } else {
1786 prefix_end
1787 };
1788 let first_content = &line[content_start..];
1789
1790 // CommonMark footnotes use 4-space continuation indent
1791 const FN_INDENT: usize = 4;
1792
1793 // --- Line classification for footnote content ---
1794 #[derive(Debug, Clone)]
1795 enum FnLineType {
1796 Content(String),
1797 Verbatim(String, usize), // preserved text, original indent
1798 Empty,
1799 }
1800
1801 // Helper: compute visual indent (tabs = 4 spaces)
1802 let visual_indent = |s: &str| -> usize {
1803 s.chars()
1804 .take_while(|c| c.is_whitespace())
1805 .map(|c| if c == '\t' { 4 } else { 1 })
1806 .sum::<usize>()
1807 };
1808
1809 // Helper: check if a trimmed line is a fence marker (homogeneous chars)
1810 let is_fence = |s: &str| -> bool {
1811 let t = s.trim();
1812 let fence_char = t.chars().next();
1813 matches!(fence_char, Some('`') | Some('~'))
1814 && t.chars().take_while(|&c| c == fence_char.unwrap()).count() >= 3
1815 };
1816
1817 // Helper: check if a trimmed line is a setext underline
1818 let is_setext_underline = |s: &str| -> bool {
1819 let t = s.trim();
1820 !t.is_empty()
1821 && (t.chars().all(|c| c == '=' || c == ' ') || t.chars().all(|c| c == '-' || c == ' '))
1822 && t.contains(['=', '-'])
1823 };
1824
1825 // Deferred body: `[^id]:\n content` — first line has no content,
1826 // actual content starts on the next indented line
1827 let deferred_body = first_content.trim().is_empty();
1828
1829 // Collect all lines belonging to this footnote definition
1830 let mut fn_lines: Vec<FnLineType> = Vec::new();
1831 if !deferred_body {
1832 fn_lines.push(FnLineType::Content(first_content.to_string()));
1833 }
1834 let mut last_consumed = i;
1835 i += 1;
1836
1837 // Strip only the footnote continuation indent, preserving
1838 // internal indentation (e.g., code block body indent)
1839 let strip_fn_indent = |s: &str| -> String {
1840 let mut chars = s.chars();
1841 let mut stripped = 0;
1842 while stripped < FN_INDENT {
1843 match chars.next() {
1844 Some('\t') => stripped += 4,
1845 Some(c) if c.is_whitespace() => stripped += 1,
1846 _ => break,
1847 }
1848 }
1849 chars.as_str().to_string()
1850 };
1851
1852 let mut in_fenced_code = false;
1853 let mut consecutive_blanks = 0u32;
1854
1855 while i < lines.len() {
1856 let next = lines[i];
1857 let next_trimmed = next.trim();
1858
1859 // Blank line handling
1860 if next_trimmed.is_empty() {
1861 consecutive_blanks += 1;
1862 // 2+ consecutive blanks terminate the footnote
1863 if consecutive_blanks >= 2 {
1864 break;
1865 }
1866
1867 // Inside a fenced code block, blank lines are part of the code
1868 if in_fenced_code {
1869 consecutive_blanks = 0; // Don't count blanks inside code blocks
1870 fn_lines.push(FnLineType::Verbatim(String::new(), 0));
1871 last_consumed = i;
1872 i += 1;
1873 continue;
1874 }
1875
1876 // Peek ahead: if next non-blank line is indented >= FN_INDENT,
1877 // this blank is an internal paragraph separator
1878 if i + 1 < lines.len() {
1879 let peek = lines[i + 1];
1880 let peek_indent = visual_indent(peek);
1881 if !peek.trim().is_empty() && peek_indent >= FN_INDENT {
1882 fn_lines.push(FnLineType::Empty);
1883 last_consumed = i;
1884 i += 1;
1885 continue;
1886 }
1887 }
1888 // No valid continuation after blank — end of footnote
1889 break;
1890 }
1891
1892 consecutive_blanks = 0;
1893 let indent = visual_indent(next);
1894
1895 // Not indented enough — end of footnote
1896 if indent < FN_INDENT {
1897 break;
1898 }
1899
1900 // Inside a fenced code block: everything is verbatim until closing fence
1901 if in_fenced_code {
1902 fn_lines.push(FnLineType::Verbatim(strip_fn_indent(next), indent));
1903 if is_fence(next_trimmed) {
1904 in_fenced_code = false;
1905 }
1906 last_consumed = i;
1907 i += 1;
1908 continue;
1909 }
1910
1911 // Fence opener — start verbatim code block
1912 if is_fence(next_trimmed) {
1913 in_fenced_code = true;
1914 fn_lines.push(FnLineType::Verbatim(strip_fn_indent(next), indent));
1915 last_consumed = i;
1916 i += 1;
1917 continue;
1918 }
1919
1920 // A multi-line display-math block is verbatim: its line breaks
1921 // carry meaning (see `line_in_multiline_math_block`).
1922 if self.line_in_multiline_math_block(i + 1, ctx, config) {
1923 fn_lines.push(FnLineType::Verbatim(strip_fn_indent(next), indent));
1924 last_consumed = i;
1925 i += 1;
1926 continue;
1927 }
1928
1929 // Indented code block: indent >= FN_INDENT + 4 (= 8 spaces)
1930 if indent >= FN_INDENT + 4 {
1931 fn_lines.push(FnLineType::Verbatim(strip_fn_indent(next), indent));
1932 last_consumed = i;
1933 i += 1;
1934 continue;
1935 }
1936
1937 // Structural content that must be preserved verbatim
1938 if next_trimmed.starts_with('#')
1939 || is_list_item(next_trimmed)
1940 || next_trimmed.starts_with('>')
1941 || TableUtils::is_potential_table_row_with_flavor(next_trimmed, ctx.flavor)
1942 || is_setext_underline(next_trimmed)
1943 || is_horizontal_rule(next_trimmed)
1944 || crate::utils::mkdocs_footnotes::is_footnote_definition(next_trimmed)
1945 {
1946 // Preserve verbatim: blockquotes, tables, lists, setext
1947 // underlines, and horizontal rules inside the footnote
1948 if next_trimmed.starts_with('>')
1949 || TableUtils::is_potential_table_row_with_flavor(next_trimmed, ctx.flavor)
1950 || is_list_item(next_trimmed)
1951 || is_setext_underline(next_trimmed)
1952 || is_horizontal_rule(next_trimmed)
1953 {
1954 fn_lines.push(FnLineType::Verbatim(strip_fn_indent(next), indent));
1955 last_consumed = i;
1956 i += 1;
1957 continue;
1958 }
1959 // Headings, new footnote defs, link refs — end the footnote
1960 break;
1961 }
1962
1963 // Link reference definitions inside footnotes are not reflowable
1964 if next_trimmed.starts_with('[')
1965 && !next_trimmed.starts_with("[^")
1966 && next_trimmed.contains("]:")
1967 && LINK_REF_PATTERN.is_match(next_trimmed)
1968 {
1969 fn_lines.push(FnLineType::Verbatim(strip_fn_indent(next), indent));
1970 last_consumed = i;
1971 i += 1;
1972 continue;
1973 }
1974
1975 // HTML-only lines inside footnotes are not reflowable
1976 if is_html_only_line(next_trimmed) {
1977 fn_lines.push(FnLineType::Verbatim(strip_fn_indent(next), indent));
1978 last_consumed = i;
1979 i += 1;
1980 continue;
1981 }
1982
1983 // Regular prose content
1984 fn_lines.push(FnLineType::Content(next_trimmed.to_string()));
1985 last_consumed = i;
1986 i += 1;
1987 }
1988
1989 // Nothing collected or only empty lines
1990 if fn_lines.iter().all(|l| matches!(l, FnLineType::Empty)) || fn_lines.is_empty() {
1991 continue;
1992 }
1993
1994 // The footnote is rebuilt with a fixed indent and its paragraphs
1995 // joined, which a definition list inside it does not survive.
1996 if holds_definition_list(ctx, footnote_start, last_consumed) {
1997 continue;
1998 }
1999
2000 // --- Group into blocks ---
2001 #[derive(Debug)]
2002 enum FnBlock {
2003 Paragraph(Vec<String>),
2004 Verbatim(Vec<(String, usize)>), // (content, indent) preserved as-is
2005 }
2006
2007 let mut blocks: Vec<FnBlock> = Vec::new();
2008 let mut current_para: Vec<String> = Vec::new();
2009 let mut current_verbatim: Vec<(String, usize)> = Vec::new();
2010
2011 for fl in &fn_lines {
2012 match fl {
2013 FnLineType::Content(s) => {
2014 if !current_verbatim.is_empty() {
2015 blocks.push(FnBlock::Verbatim(std::mem::take(&mut current_verbatim)));
2016 }
2017 current_para.push(s.clone());
2018 }
2019 FnLineType::Verbatim(s, indent) => {
2020 if !current_para.is_empty() {
2021 blocks.push(FnBlock::Paragraph(std::mem::take(&mut current_para)));
2022 }
2023 current_verbatim.push((s.clone(), *indent));
2024 }
2025 FnLineType::Empty => {
2026 if !current_para.is_empty() {
2027 blocks.push(FnBlock::Paragraph(std::mem::take(&mut current_para)));
2028 }
2029 if !current_verbatim.is_empty() {
2030 blocks.push(FnBlock::Verbatim(std::mem::take(&mut current_verbatim)));
2031 }
2032 }
2033 }
2034 }
2035 if !current_para.is_empty() {
2036 blocks.push(FnBlock::Paragraph(current_para));
2037 }
2038 if !current_verbatim.is_empty() {
2039 blocks.push(FnBlock::Verbatim(current_verbatim));
2040 }
2041
2042 // --- Reflow paragraphs and reconstruct ---
2043 let prefix_display_width = prefix.chars().count() + 1; // +1 for space
2044 let reflow_line_length = if config.line_length.is_unlimited() {
2045 usize::MAX
2046 } else {
2047 config
2048 .line_length
2049 .get()
2050 .saturating_sub(FN_INDENT.max(prefix_display_width))
2051 .max(20)
2052 };
2053 // Footnote continuation uses a fixed 4-space indent, so list
2054 // continuation capping does not apply here.
2055 let reflow_options = crate::utils::text_reflow::ReflowOptions {
2056 max_list_continuation_indent: None,
2057 ..Self::reflow_options(ctx, config, reflow_line_length)
2058 };
2059
2060 let indent_str = " ".repeat(FN_INDENT);
2061 let mut result_lines: Vec<String> = Vec::new();
2062 let mut is_first_block = true;
2063
2064 for block in &blocks {
2065 match block {
2066 FnBlock::Paragraph(para_lines) => {
2067 let paragraph_text = join_soft_break_lines(para_lines, config.cjk_soft_break);
2068 let paragraph_text = paragraph_text.trim();
2069 if paragraph_text.is_empty() {
2070 continue;
2071 }
2072
2073 let reflowed = crate::utils::text_reflow::reflow_line(paragraph_text, &reflow_options);
2074 if reflowed.is_empty() {
2075 continue;
2076 }
2077
2078 // Blank line separator between blocks
2079 if !result_lines.is_empty() {
2080 result_lines.push(String::new());
2081 }
2082
2083 for (idx, rline) in reflowed.iter().enumerate() {
2084 if is_first_block && idx == 0 {
2085 result_lines.push(format!("{prefix} {rline}"));
2086 } else {
2087 result_lines.push(format!("{indent_str}{rline}"));
2088 }
2089 }
2090 is_first_block = false;
2091 }
2092 FnBlock::Verbatim(verb_lines) => {
2093 // Blank line separator between blocks
2094 if !result_lines.is_empty() {
2095 result_lines.push(String::new());
2096 }
2097
2098 if is_first_block {
2099 // Verbatim as first block in a deferred-body footnote
2100 if deferred_body {
2101 result_lines.push(prefix.to_string());
2102 }
2103 is_first_block = false;
2104 }
2105 for (content, _orig_indent) in verb_lines {
2106 result_lines.push(format!("{indent_str}{content}"));
2107 }
2108 }
2109 }
2110 }
2111
2112 // If nothing was produced, skip
2113 if result_lines.is_empty() {
2114 continue;
2115 }
2116
2117 let reflowed_text = result_lines.join(line_ending);
2118
2119 // Calculate byte range using last_consumed
2120 let start_range = ctx.whole_line_byte_range(footnote_start + 1);
2121 let end_range = if last_consumed == lines.len() - 1 && !ctx.content.ends_with('\n') {
2122 ctx.line_text_byte_range(last_consumed + 1, 1, lines[last_consumed].len() + 1)
2123 } else {
2124 ctx.whole_line_byte_range(last_consumed + 1)
2125 };
2126 let byte_range = start_range.start..end_range.end;
2127
2128 let replacement = if last_consumed < lines.len() - 1 || ctx.content.ends_with('\n') {
2129 format!("{reflowed_text}{line_ending}")
2130 } else {
2131 reflowed_text
2132 };
2133
2134 let original_text = &ctx.content[byte_range.clone()];
2135 let max_length = (footnote_start..=last_consumed)
2136 .map(|idx| self.calculate_effective_length(lines[idx]))
2137 .max()
2138 .unwrap_or(0);
2139 let line_limit = if config.line_length.is_unlimited() {
2140 usize::MAX
2141 } else {
2142 config.line_length.get()
2143 };
2144 if original_text != replacement && max_length > line_limit {
2145 warnings.push(LintWarning {
2146 rule_name: Some(self.name().to_string()),
2147 message: format!(
2148 "Line length {} exceeds {} characters",
2149 max_length,
2150 config.line_length.get()
2151 ),
2152 line: footnote_start + 1,
2153 column: 1,
2154 end_line: last_consumed + 1,
2155 end_column: lines[last_consumed].chars().count() + 1,
2156 severity: Severity::Warning,
2157 fix: Some(crate::rule::Fix::new(byte_range, replacement)),
2158 });
2159 }
2160 continue;
2161 }
2162
2163 // Handle MkDocs container content (admonitions and tabs) with indent-preserving reflow
2164 if ctx
2165 .line_info(line_num)
2166 .is_some_and(super::super::lint_context::types::LineInfo::in_mkdocs_container)
2167 {
2168 // Skip admonition/tab marker lines — only reflow their indented content
2169 let current_line = lines[i];
2170 if mkdocs_admonitions::is_admonition_start(current_line) || mkdocs_tabs::is_tab_marker(current_line) {
2171 i += 1;
2172 continue;
2173 }
2174
2175 let container_start = i;
2176
2177 // Detect the actual indent level from the first content line
2178 // (supports nested admonitions with 8+ spaces)
2179 let first_line = lines[i];
2180 let base_indent_len = first_line.len() - first_line.trim_start().len();
2181 let base_indent: String = " ".repeat(base_indent_len);
2182
2183 // Collect consecutive MkDocs container paragraph lines
2184 let mut container_lines: Vec<&str> = Vec::new();
2185 while i < lines.len() {
2186 let current_line_num = i + 1;
2187 let line_info = ctx.line_info(current_line_num);
2188
2189 // Stop if we leave the MkDocs container
2190 if !line_info.is_some_and(super::super::lint_context::types::LineInfo::in_mkdocs_container) {
2191 break;
2192 }
2193
2194 let line = lines[i];
2195
2196 // Stop at paragraph boundaries within the container
2197 if line.trim().is_empty() {
2198 break;
2199 }
2200
2201 // Skip list items, code blocks, headings, HTML-only lines within containers
2202 if is_list_item(line.trim())
2203 || line.trim().starts_with("```")
2204 || line.trim().starts_with("~~~")
2205 || line.trim().starts_with('#')
2206 || is_html_only_line(line)
2207 {
2208 break;
2209 }
2210
2211 container_lines.push(line);
2212 i += 1;
2213 }
2214
2215 if container_lines.is_empty() {
2216 // Must advance i to avoid infinite loop when we encounter
2217 // non-paragraph content (code block, list, heading, empty line)
2218 // at the start of an MkDocs container
2219 i += 1;
2220 continue;
2221 }
2222
2223 // Strip the base indent from each line and join for reflow
2224 let stripped_lines: Vec<&str> = container_lines
2225 .iter()
2226 .map(|line| {
2227 if line.starts_with(&base_indent) {
2228 &line[base_indent_len..]
2229 } else {
2230 line.trim_start()
2231 }
2232 })
2233 .collect();
2234 let paragraph_text = join_soft_break_lines(&stripped_lines, config.cjk_soft_break);
2235
2236 // Check if reflow is needed
2237 let needs_reflow = match config.reflow_mode {
2238 ReflowMode::Normalize => self.normalize_mode_needs_reflow(container_lines.iter().copied(), config),
2239 ReflowMode::SentencePerLine => {
2240 let sentences = split_into_sentences(
2241 ¶graph_text,
2242 Some(&defined_references),
2243 config.require_sentence_capital,
2244 );
2245 sentences.len() > 1 || container_lines.len() > 1
2246 }
2247 ReflowMode::SemanticLineBreaks => {
2248 let sentences = split_into_sentences(
2249 ¶graph_text,
2250 Some(&defined_references),
2251 config.require_sentence_capital,
2252 );
2253 sentences.len() > 1
2254 || container_lines.len() > 1
2255 || container_lines
2256 .iter()
2257 .any(|line| self.calculate_effective_length(line) > config.line_length.get())
2258 }
2259 ReflowMode::Default => container_lines
2260 .iter()
2261 .any(|line| self.calculate_effective_length(line) > config.line_length.get()),
2262 };
2263
2264 if !needs_reflow {
2265 continue;
2266 }
2267
2268 // Calculate byte range for this container paragraph
2269 let start_range = ctx.whole_line_byte_range(container_start + 1);
2270 let end_line = container_start + container_lines.len() - 1;
2271 let end_range = if end_line == lines.len() - 1 && !ctx.content.ends_with('\n') {
2272 ctx.line_text_byte_range(end_line + 1, 1, lines[end_line].len() + 1)
2273 } else {
2274 ctx.whole_line_byte_range(end_line + 1)
2275 };
2276 let byte_range = start_range.start..end_range.end;
2277
2278 // Reflow with adjusted line length (accounting for the 4-space indent)
2279 let reflow_line_length = if config.line_length.is_unlimited() {
2280 usize::MAX
2281 } else {
2282 config.line_length.get().saturating_sub(base_indent_len).max(1)
2283 };
2284 let reflow_options = Self::reflow_options(ctx, config, reflow_line_length);
2285 let reflowed = crate::utils::text_reflow::reflow_line(¶graph_text, &reflow_options);
2286
2287 // Re-add the 4-space indent to each reflowed line
2288 let reflowed_with_indent: Vec<String> =
2289 reflowed.iter().map(|line| format!("{base_indent}{line}")).collect();
2290 let reflowed_text = reflowed_with_indent.join(line_ending);
2291
2292 // Preserve trailing newline
2293 let replacement = if end_line < lines.len() - 1 || ctx.content.ends_with('\n') {
2294 format!("{reflowed_text}{line_ending}")
2295 } else {
2296 reflowed_text
2297 };
2298
2299 // Only generate a warning if the replacement is different
2300 let original_text = &ctx.content[byte_range.clone()];
2301 if original_text != replacement {
2302 warnings.push(LintWarning {
2303 rule_name: Some(self.name().to_string()),
2304 message: format!(
2305 "Line length {} exceeds {} characters (in MkDocs container)",
2306 container_lines.iter().map(|l| l.len()).max().unwrap_or(0),
2307 config.line_length.get()
2308 ),
2309 line: container_start + 1,
2310 column: 1,
2311 end_line: end_line + 1,
2312 end_column: lines[end_line].chars().count() + 1,
2313 severity: Severity::Warning,
2314 fix: Some(crate::rule::Fix::new(byte_range, replacement)),
2315 });
2316 }
2317 continue;
2318 }
2319
2320 // Helper function to detect semantic line markers
2321 let is_semantic_line = |content: &str| -> bool {
2322 let trimmed = content.trim_start();
2323 let semantic_markers = [
2324 "NOTE:",
2325 "WARNING:",
2326 "IMPORTANT:",
2327 "CAUTION:",
2328 "TIP:",
2329 "DANGER:",
2330 "HINT:",
2331 "INFO:",
2332 ];
2333 semantic_markers.iter().any(|marker| trimmed.starts_with(marker))
2334 };
2335
2336 // Helper function to detect fence markers (opening or closing)
2337 let is_fence_marker = |content: &str| -> bool {
2338 let trimmed = content.trim_start();
2339 trimmed.starts_with("```") || trimmed.starts_with("~~~")
2340 };
2341
2342 // Check if this is a list item - handle it specially
2343 let trimmed = lines[i].trim();
2344 if is_list_item(trimmed) {
2345 // Collect the entire list item including continuation lines
2346 let list_start = i;
2347 let (marker, first_content) = extract_list_marker_and_content(lines[i]);
2348 let marker_len = marker.len();
2349 // The normalized marker above is what gets re-emitted; the source marker
2350 // is where the item's content actually starts. Nested blocks move by the
2351 // difference between the two content columns, so the shift must be
2352 // measured against the source, not against the normalized width.
2353 let source_marker = source_list_marker(lines[i]);
2354 let source_content_col = source_marker.as_ref().map_or(marker_len, |m| m.content_col);
2355
2356 // Checkbox ([ ]/[x]/[X]) is inline content, not part of the list marker.
2357 // Use the base bullet/number marker width for continuation recognition
2358 // so that continuation lines at 2+ spaces are collected for "- [ ] " items.
2359 let base_marker_len = if marker.contains("[ ] ") || marker.contains("[x] ") || marker.contains("[X] ") {
2360 marker.find('[').unwrap_or(marker_len)
2361 } else {
2362 marker_len
2363 };
2364
2365 // MkDocs flavor requires at least 4 spaces for list continuation
2366 // after a blank line (multi-paragraph list items). For non-blank
2367 // continuation (lines directly following the marker line), use
2368 // the natural marker width so that 2-space indent is recognized.
2369 let item_indent = ctx.lines[i].indent;
2370 let min_continuation_indent = if ctx.flavor.requires_strict_list_indent() {
2371 // Use 4-space relative indent from the list item's nesting level
2372 item_indent + (base_marker_len - item_indent).max(4)
2373 } else {
2374 marker_len
2375 };
2376 let content_continuation_indent = base_marker_len;
2377
2378 // Track lines and their types (content, code block, fence, nested list)
2379 #[derive(Clone)]
2380 enum LineType {
2381 Content(String, usize), // content and 1-indexed line number
2382 CodeBlock(String, usize), // content and original indent
2383 SemanticLine(String), // Lines starting with NOTE:, WARNING:, etc that should stay separate
2384 SnippetLine(String), // MkDocs Snippets delimiters (-8<-) that must stay on their own line
2385 DivMarker(String), // Quarto/Pandoc div markers (::: opening or closing)
2386 AdmonitionHeader(String, usize), // header text (e.g. "!!! note") and original indent
2387 AdmonitionContent(String, usize), // body content text and original indent
2388 Table(String, usize), // GFM table row, preserved verbatim with original indent
2389 Empty,
2390 }
2391
2392 let start_idx = i;
2393 // A marker line whose content is one whole `$$...$$` expression
2394 // renders as a display block, so it keeps the line it was written
2395 // on and the prose under it starts a paragraph of its own. The
2396 // code-block carrier re-emits it unchanged after the marker.
2397 //
2398 // The marker keeps one padding space and leaves the rest at the
2399 // head of the content, and the carrier writes the marker as the
2400 // author spelled it, padding included. The carrier therefore
2401 // holds the content with its leading whitespace off, so the
2402 // padding is written once.
2403 //
2404 // A marker that cannot interrupt a paragraph, such as an
2405 // ordered one not starting at one, can sit inside a code span
2406 // the paragraph above opened, and its content is code then.
2407 // The marker line can just as well be the one that opens the
2408 // span, closing on a line still to come.
2409 let mut list_item_lines: Vec<LineType> = if (is_self_contained_display_math_line(&first_content)
2410 || self.line_is_standalone_bracket_math(i + 1, ctx, config))
2411 && !line_touches_multiline_code_span(&code_span_touches, i + 1)
2412 {
2413 vec![LineType::CodeBlock(first_content.trim_start().to_string(), marker_len)]
2414 } else {
2415 vec![LineType::Content(first_content, i + 1)]
2416 };
2417 // Set when collection stops at a nested list item or a nested
2418 // blockquote that belongs to this item. Such structure is reflowed
2419 // independently and is therefore absent from `list_item_lines`/`blocks`,
2420 // but it still keeps the emitted item spanning multiple physical lines,
2421 // which the MD030 multi-line spacing decision below must account for.
2422 let mut has_trailing_nested_structure = false;
2423 i += 1;
2424
2425 // Collect continuation lines using ctx.lines for metadata
2426 while i < lines.len() {
2427 let line_info = &ctx.lines[i];
2428
2429 // Use pre-computed is_blank from ctx
2430 if line_info.is_blank {
2431 // Empty line - check if next line is indented (part of list item)
2432 if i + 1 < lines.len() {
2433 let next_info = &ctx.lines[i + 1];
2434
2435 // Check if next line is indented enough to be continuation
2436 if !next_info.is_blank && next_info.indent >= min_continuation_indent {
2437 // This blank line is between paragraphs/blocks in the list item
2438 list_item_lines.push(LineType::Empty);
2439 i += 1;
2440 continue;
2441 }
2442 }
2443 // No indented line after blank, end of list item
2444 break;
2445 }
2446
2447 // Use pre-computed indent from ctx
2448 let indent = line_info.indent;
2449
2450 // Valid continuation must be indented at least content_continuation_indent.
2451 // For non-blank continuation, use marker_len (e.g. 2 for "- ").
2452 // MkDocs strict 4-space requirement applies only after blank lines.
2453 if indent >= content_continuation_indent {
2454 let trimmed = line_info.content(ctx.content).trim();
2455
2456 // Check for MkDocs admonition lines inside list items BEFORE
2457 // checking in_code_block. Lines inside code blocks within
2458 // admonitions have both in_admonition and in_code_block set;
2459 // admonition membership takes priority so the entire admonition
2460 // structure (including embedded code blocks) is preserved.
2461 if line_info.in_admonition {
2462 let raw_content = line_info.content(ctx.content);
2463 if mkdocs_admonitions::is_admonition_start(raw_content) {
2464 let header_text = raw_content[indent..].trim_end().to_string();
2465 list_item_lines.push(LineType::AdmonitionHeader(header_text, indent));
2466 } else {
2467 let body_text = raw_content[indent..].trim_end().to_string();
2468 list_item_lines.push(LineType::AdmonitionContent(body_text, indent));
2469 }
2470 i += 1;
2471 continue;
2472 }
2473
2474 // Use pre-computed in_code_block from ctx
2475 if line_info.in_code_block {
2476 list_item_lines.push(LineType::CodeBlock(
2477 line_info.content(ctx.content)[indent..].to_string(),
2478 indent,
2479 ));
2480 i += 1;
2481 continue;
2482 }
2483
2484 // A multi-line display-math block inside the item is verbatim:
2485 // its line breaks carry meaning (see
2486 // `line_in_multiline_math_block`), so reuse the code-block
2487 // carrier to re-emit it unchanged.
2488 if self.line_in_multiline_math_block(i + 1, ctx, config) {
2489 list_item_lines.push(LineType::CodeBlock(
2490 line_info.content(ctx.content)[indent..].to_string(),
2491 indent,
2492 ));
2493 i += 1;
2494 continue;
2495 }
2496
2497 // A blockquote nested inside the list item is reflowed by the
2498 // blockquote-aware path (it preserves the `>` prefix, including the
2499 // list indent), not as list-item prose. Collecting it as Content
2500 // would strip the markers and reflow `>` as words, collapsing the
2501 // blank `>` line and dropping `>` from wrapped continuations. End
2502 // the item here so the outer loop routes the blockquote line to
2503 // generate_blockquote_paragraph_fix. Uncollect a pending blank so
2504 // the separator between the list prose and the blockquote survives.
2505 if line_info.blockquote.is_some() {
2506 has_trailing_nested_structure = true;
2507 if matches!(list_item_lines.last(), Some(LineType::Empty)) {
2508 list_item_lines.pop();
2509 i -= 1;
2510 }
2511 break;
2512 }
2513
2514 // Check if this is a SIBLING list item (breaks parent)
2515 // Nested lists are indented >= marker_len and are PART of the parent item
2516 // Siblings are at indent < marker_len (at or before parent marker)
2517 if is_list_item(trimmed) && indent < marker_len {
2518 // This is a sibling item at same or higher level - end parent item
2519 break;
2520 }
2521
2522 // Nested list items are always processed independently
2523 // by the outer loop, so break when we encounter one.
2524 // If a blank line was collected before this, uncollect it
2525 // so the outer loop preserves the blank between parent and nested.
2526 if is_list_item(trimmed) && indent >= marker_len {
2527 has_trailing_nested_structure = true;
2528 if matches!(list_item_lines.last(), Some(LineType::Empty)) {
2529 list_item_lines.pop();
2530 i -= 1;
2531 }
2532 break;
2533 }
2534
2535 // Normal continuation vs indented code block.
2536 // Use min_continuation_indent for the threshold since
2537 // code blocks start 4 spaces beyond the expected content
2538 // level (which is min_continuation_indent for MkDocs).
2539 if indent <= min_continuation_indent + 3 {
2540 // Extract content (remove indentation and trailing whitespace)
2541 // Preserve hard breaks (2 trailing spaces) while removing excessive whitespace
2542 // See: https://github.com/rvben/rumdl/issues/76
2543 let content = trim_preserving_hard_break(&line_info.content(ctx.content)[indent..]);
2544
2545 // Check if this is a div marker (::: opening or closing)
2546 // These must be preserved on their own line, not merged into paragraphs
2547 if line_info.is_div_marker {
2548 list_item_lines.push(LineType::DivMarker(content));
2549 }
2550 // A fence marker opens or closes a code block, and a line
2551 // that is one whole `$$...$$` expression renders as a
2552 // display block. Both keep the line they were written on,
2553 // so the code-block carrier re-emits them unchanged
2554 // between the prose above and below. A line touched by a
2555 // code span crossing one of its boundaries is code, not
2556 // such a block.
2557 else if is_fence_marker(&content)
2558 || ((is_self_contained_display_math_line(&content)
2559 || self.line_is_standalone_bracket_math(i + 1, ctx, config))
2560 && !line_touches_multiline_code_span(&code_span_touches, i + 1))
2561 {
2562 list_item_lines.push(LineType::CodeBlock(content, indent));
2563 }
2564 // Check if this is a semantic line (NOTE:, WARNING:, etc.)
2565 else if is_semantic_line(&content) {
2566 list_item_lines.push(LineType::SemanticLine(content));
2567 }
2568 // Check if this is a snippet block delimiter (-8<- or --8<--)
2569 // These must be preserved on their own lines for MkDocs Snippets extension
2570 else if is_snippet_block_delimiter(&content) {
2571 list_item_lines.push(LineType::SnippetLine(content));
2572 }
2573 // Check if this is a GFM table row. Tables nested inside list
2574 // items must be preserved verbatim — joining them with prose
2575 // breaks the column structure.
2576 //
2577 // `is_potential_table_row` is intentionally permissive at the
2578 // row level: any line with `|` and 2+ cells qualifies. To avoid
2579 // misclassifying prose continuation lines that contain a literal
2580 // pipe (e.g. "use grep | sort to ..."), require one of:
2581 // - the row is pipe-bordered (`| ... |`), the canonical form
2582 // for tables nested in lists; or
2583 // - the next line is a delimiter row (this is a header); or
2584 // - the previous classified line was already a Table (this is
2585 // a continuation row).
2586 else if TableUtils::is_potential_table_row_with_flavor(&content, ctx.flavor) && {
2587 let pipe_bordered = content.trim().starts_with('|') && content.trim().ends_with('|');
2588 let next_is_delim = ctx
2589 .lines
2590 .get(i + 1)
2591 .is_some_and(|next| TableUtils::is_delimiter_row(next.content(ctx.content)));
2592 let prev_was_table = matches!(list_item_lines.last(), Some(LineType::Table(..)));
2593 pipe_bordered || next_is_delim || prev_was_table
2594 } {
2595 list_item_lines.push(LineType::Table(content, indent));
2596 } else {
2597 list_item_lines.push(LineType::Content(content, i + 1));
2598 }
2599 i += 1;
2600 } else {
2601 // indent >= min_continuation_indent + 4: indented code block
2602 list_item_lines.push(LineType::CodeBlock(
2603 line_info.content(ctx.content)[indent..].to_string(),
2604 indent,
2605 ));
2606 i += 1;
2607 }
2608 } else {
2609 // Not indented enough, end of list item
2610 break;
2611 }
2612 }
2613
2614 // Determine the output continuation indent.
2615 // Normalize/Default modes canonicalize to min_continuation_indent
2616 // (fixing over-indented continuation). Semantic/SentencePerLine
2617 // modes preserve the user's actual indent since they only fix
2618 // line breaking, not indentation.
2619 let indent_size = match config.reflow_mode {
2620 ReflowMode::SemanticLineBreaks | ReflowMode::SentencePerLine => {
2621 // Find indent of the first plain text continuation line,
2622 // skipping the marker line (index 0), nested list items,
2623 // code blocks, and blank lines.
2624 list_item_lines
2625 .iter()
2626 .enumerate()
2627 .skip(1)
2628 .find_map(|(k, lt)| {
2629 if matches!(lt, LineType::Content(..)) {
2630 Some(ctx.lines[list_start + k].indent)
2631 } else {
2632 None
2633 }
2634 })
2635 .unwrap_or(min_continuation_indent)
2636 }
2637 _ => min_continuation_indent,
2638 };
2639 // For checkbox items in mkdocs flavor, enforce minimum indent so
2640 // continuation lines use the structural list indent (4), not the
2641 // content-aligned indent (6) which Python-Markdown doesn't support
2642 let has_checkbox = base_marker_len < marker_len;
2643 let indent_size = if has_checkbox && ctx.flavor.requires_strict_list_indent() {
2644 indent_size.max(min_continuation_indent)
2645 } else {
2646 indent_size
2647 };
2648
2649 // Split list_item_lines into blocks (paragraphs, code blocks, nested lists, semantic lines, and HTML blocks)
2650 let mut builder = BlockBuilder::new_with_start_line(start_idx + 1);
2651 for line in &list_item_lines {
2652 match line {
2653 LineType::Empty => builder.feed_blank_line(),
2654 LineType::Content(content, _) => builder.feed_content(content),
2655 LineType::CodeBlock(content, indent) => builder.feed_code_line(content, *indent),
2656 LineType::SemanticLine(content) => builder.feed_semantic_line(content),
2657 LineType::SnippetLine(content) => builder.feed_snippet_line(content),
2658 LineType::DivMarker(content) => builder.feed_div_marker(content),
2659 LineType::AdmonitionHeader(header_text, indent) => {
2660 builder.feed_admonition_header(header_text, *indent)
2661 }
2662 LineType::AdmonitionContent(content, indent) => {
2663 builder.feed_admonition_content(content, *indent)
2664 }
2665 LineType::Table(content, indent) => builder.feed_table_line(content, *indent),
2666 }
2667 }
2668 let blocks = builder.finalize();
2669
2670 // Helper: check if a line (raw source or stripped content) is exempt
2671 // from line-length checks. Link reference definitions are always exempt;
2672 // standalone link/image lines are exempt when strict mode is off.
2673 // Also checks content after stripping list markers, since list item
2674 // continuation lines may contain link ref defs.
2675 let is_exempt_line = |raw_line: &str, line_num: usize| -> bool {
2676 let trimmed = raw_line.trim();
2677 // Link reference definitions: always exempt
2678 if trimmed.starts_with('[') && trimmed.contains("]:") && LINK_REF_PATTERN.is_match(trimmed) {
2679 return true;
2680 }
2681 // Also check after stripping list markers (for list item content)
2682 if is_list_item(trimmed) {
2683 let (_, content) = extract_list_marker_and_content(trimmed);
2684 let content_trimmed = content.trim();
2685 if content_trimmed.starts_with('[')
2686 && content_trimmed.contains("]:")
2687 && LINK_REF_PATTERN.is_match(content_trimmed)
2688 {
2689 return true;
2690 }
2691 }
2692 // Standalone link/image lines: exempt when not strict
2693 if standalone_link_ends_paragraph(ctx, line_num, config) {
2694 return true;
2695 }
2696 // HTML-only lines: exempt when not strict
2697 if !config.strict && is_html_only_line(raw_line) {
2698 return true;
2699 }
2700 false
2701 };
2702
2703 // Check if reflowing is needed (only for content paragraphs, not code blocks or nested lists)
2704 // Exclude link reference definitions and standalone link lines from content
2705 // so they don't pollute combined_content or trigger false reflow.
2706 let content_lines: Vec<String> = list_item_lines
2707 .iter()
2708 .filter_map(|line| {
2709 if let LineType::Content(s, line_num) = line {
2710 if is_exempt_line(s, *line_num) {
2711 return None;
2712 }
2713 Some(s.clone())
2714 } else {
2715 None
2716 }
2717 })
2718 .collect();
2719
2720 // Check if we need to reflow this list item
2721 // We check the combined content to see if it exceeds length limits
2722 let combined_content = join_soft_break_lines(&content_lines, config.cjk_soft_break)
2723 .trim()
2724 .to_string();
2725
2726 // Helper to check if we should reflow in normalize mode
2727 let should_normalize = || {
2728 // Don't normalize if the list item only contains nested lists, code blocks, or semantic lines
2729 // DO normalize if it has plain text content that spans multiple lines
2730 let has_code_blocks = blocks.iter().any(|b| matches!(b, Block::Code { .. }));
2731 let has_semantic_lines = blocks.iter().any(|b| matches!(b, Block::SemanticLine(_)));
2732 let has_snippet_lines = blocks.iter().any(|b| matches!(b, Block::SnippetLine(_)));
2733 let has_div_markers = blocks.iter().any(|b| matches!(b, Block::DivMarker(_)));
2734 let has_admonitions = blocks.iter().any(|b| matches!(b, Block::Admonition { .. }));
2735 let has_tables = blocks.iter().any(|b| matches!(b, Block::Table { .. }));
2736 let has_paragraphs = blocks.iter().any(|b| matches!(b, Block::Paragraph(_)));
2737
2738 // If we have structural blocks but no paragraphs, don't normalize
2739 if (has_code_blocks
2740 || has_semantic_lines
2741 || has_snippet_lines
2742 || has_div_markers
2743 || has_admonitions
2744 || has_tables)
2745 && !has_paragraphs
2746 {
2747 return false;
2748 }
2749
2750 // If we have paragraphs, check if they span multiple lines or there are multiple blocks
2751 if has_paragraphs {
2752 // Count only paragraphs that contain at least one non-exempt line.
2753 // Paragraphs consisting entirely of link ref defs or standalone links
2754 // should not trigger normalization.
2755 let paragraph_count = blocks
2756 .iter()
2757 .filter(|b| {
2758 if let Block::Paragraph(para_lines) = b {
2759 !para_lines
2760 .iter()
2761 .all(|(line, line_num)| is_exempt_line(line, *line_num))
2762 } else {
2763 false
2764 }
2765 })
2766 .count();
2767 if paragraph_count > 1 {
2768 // Multiple non-exempt paragraph blocks should be normalized
2769 return true;
2770 }
2771
2772 // Single paragraph block: normalize if it has multiple content lines
2773 if content_lines.len() > 1 {
2774 return true;
2775 }
2776 }
2777
2778 false
2779 };
2780
2781 // Integrate MD030 list-marker spacing (and MD007's text-aligned
2782 // continuation). In Default/Normalize modes — the modes that already
2783 // canonicalize spacing/indent — derive the number of spaces after the
2784 // marker from the configured MD030 values instead of forcing a single
2785 // space, then align continuation lines to the resulting content column.
2786 // Sentence/Semantic modes only adjust line breaks, so they keep the
2787 // marker spacing and indentation already present in the source.
2788 //
2789 // With default MD030 (a single space everywhere) the rebuilt marker is
2790 // byte-identical to the source marker, so this is a no-op and existing
2791 // behaviour is preserved; only a non-default MD030 changes the output.
2792 // The MkDocs flavor enforces a rigid structural indent (4 spaces,
2793 // capped via max_list_continuation_indent) that Python-Markdown
2794 // requires; leave its specialized handling untouched.
2795 let (marker, indent_size, code_indent_shift) =
2796 if matches!(config.reflow_mode, ReflowMode::Default | ReflowMode::Normalize)
2797 && !ctx.flavor.requires_strict_list_indent()
2798 && let Some(li) = ctx.lines[list_start].list_item.as_deref()
2799 {
2800 let bullet_len = li.marker.len();
2801 // The checkbox (e.g. `[ ] `) is content, not part of the list
2802 // marker MD030 governs; carry it over verbatim after the spacing.
2803 let checkbox_tail = marker[base_marker_len..].to_string();
2804 // Shift this item right by its ancestors' cumulative marker
2805 // widening so a nested item stays under its parent's (widened)
2806 // content column. Zero for top-level items and for the whole
2807 // tree under default MD030, where the source indent is preserved
2808 // verbatim (byte-identical output).
2809 let ancestor_shift = list_shift_stack.last().map_or(0isize, |&(_, shift)| shift);
2810 let shifted_indent = (item_indent as isize + ancestor_shift).max(0) as usize;
2811 let indent_prefix = if ancestor_shift == 0 {
2812 marker[..item_indent].to_string()
2813 } else {
2814 " ".repeat(shifted_indent)
2815 };
2816
2817 // Decide single- vs multi-line spacing from the *rewritten* shape,
2818 // not the source. A multi-line source is not enough: plain prose
2819 // continuation collapses onto the marker line during reflow, so a
2820 // two-line bullet that fits becomes a single physical line and must
2821 // use MD030's single-line spacing (otherwise MD013 emits a result
2822 // that MD030 immediately rewrites). The emitted item stays
2823 // multi-line only when reflow cannot collapse it:
2824 // - the prose wraps past the line length, or
2825 // - a structural block remains (code, table, admonition, semantic
2826 // line, snippet, div marker, HTML) that is not joinable prose, or
2827 // - more than one paragraph remains (blank-separated), or
2828 // - a nested list/blockquote follows (reflowed independently, so it
2829 // is absent from `blocks` but still keeps the item multi-line).
2830 // The wrap test uses the single-line content column so it is
2831 // independent of the spacing we are about to choose (avoiding a
2832 // circular result). `ol-align-column` ignores this flag entirely in
2833 // expected_spaces(), so ordered lists are unaffected.
2834 //
2835 // This is the rewritten-shape counterpart of MD030's
2836 // `is_multi_line_list_item` (which keys off the *source*). The two
2837 // are related but technically distinct and intentionally separate;
2838 // if the notion of "multi-line" changes in one, revisit the other.
2839 let single_col = shifted_indent
2840 + bullet_len
2841 + self.list_spacing.expected_spaces(li.is_ordered, false, bullet_len)
2842 + checkbox_tail.len();
2843 let prose_wraps = !combined_content.is_empty()
2844 && self
2845 .calculate_effective_length(&format!("{}{combined_content}", " ".repeat(single_col)))
2846 > config.line_length.effective_limit();
2847 let has_structural_block = blocks.iter().any(|b| !matches!(b, Block::Paragraph(_)));
2848 let multiple_paragraphs =
2849 blocks.iter().filter(|b| matches!(b, Block::Paragraph(_))).count() > 1;
2850 let is_multi =
2851 prose_wraps || has_structural_block || multiple_paragraphs || has_trailing_nested_structure;
2852
2853 let spaces = self.list_spacing.expected_spaces(li.is_ordered, is_multi, bullet_len);
2854 let new_marker = format!("{indent_prefix}{}{}{checkbox_tail}", li.marker, " ".repeat(spaces));
2855 let new_col = new_marker.chars().count();
2856 let shift = new_col as isize - source_content_col as isize;
2857 (new_marker, new_col, shift)
2858 } else {
2859 // MkDocs enforces a rigid structural indent, so the item is emitted
2860 // exactly as written and its nested blocks never move. Re-emitting
2861 // the normalized marker here would narrow the content column while
2862 // leaving those blocks behind.
2863 let marker = source_marker.map_or(marker, |m| m.text);
2864 (marker, indent_size, 0isize)
2865 };
2866 let expected_indent = " ".repeat(indent_size);
2867
2868 // A colon-led line with a line of its paragraph before it opens a
2869 // definition and makes the item a definition list. Joining the
2870 // lines flattens that into prose, so the item is left as the
2871 // author wrote it. The first line of each paragraph is prose
2872 // whatever it starts with, since a definition needs a term on
2873 // the line before it. Content lines arrive with the item's own
2874 // indentation already off, which is what the marker's
2875 // indentation is counted from.
2876 let contains_definition_list = blocks.iter().any(|block| match block {
2877 Block::Paragraph(para_lines) => para_lines
2878 .iter()
2879 .skip(1)
2880 .any(|(line, _)| crate::utils::text_reflow::is_definition_list_marker(line)),
2881 _ => false,
2882 });
2883
2884 let needs_reflow = !contains_definition_list
2885 && !holds_definition_list(ctx, list_start, i - 1)
2886 && match config.reflow_mode {
2887 ReflowMode::Normalize => {
2888 // Only reflow if:
2889 // 1. Any non-exempt paragraph, when joined, exceeds the limit, OR
2890 // 2. Any admonition content line exceeds the limit, OR
2891 // 3. The list item should be normalized (has multi-line plain text)
2892 let any_paragraph_exceeds = blocks.iter().any(|block| match block {
2893 Block::Paragraph(para_lines) => {
2894 if para_lines
2895 .iter()
2896 .all(|(line, line_num)| is_exempt_line(line, *line_num))
2897 {
2898 return false;
2899 }
2900 let joined = join_soft_break_lines(
2901 ¶_lines.iter().map(|(l, _)| l.as_str()).collect::<Vec<_>>(),
2902 config.cjk_soft_break,
2903 );
2904 let with_marker = format!("{}{}", " ".repeat(indent_size), joined.trim());
2905 self.calculate_effective_length(&with_marker) > config.line_length.get()
2906 }
2907 Block::Admonition {
2908 content_lines,
2909 header_indent,
2910 ..
2911 } => content_lines.iter().any(|(content, indent)| {
2912 if content.is_empty() {
2913 return false;
2914 }
2915 let with_indent = format!("{}{}", " ".repeat(*indent.max(header_indent)), content);
2916 self.calculate_effective_length(&with_indent) > config.line_length.get()
2917 }),
2918 _ => false,
2919 });
2920 if any_paragraph_exceeds {
2921 true
2922 } else {
2923 should_normalize()
2924 }
2925 }
2926 ReflowMode::SentencePerLine => {
2927 // Check if list item has multiple sentences
2928 let sentences = split_into_sentences(
2929 &combined_content,
2930 Some(&defined_references),
2931 config.require_sentence_capital,
2932 );
2933 sentences.len() > 1
2934 }
2935 ReflowMode::SemanticLineBreaks => {
2936 let sentences = split_into_sentences(
2937 &combined_content,
2938 Some(&defined_references),
2939 config.require_sentence_capital,
2940 );
2941 sentences.len() > 1
2942 || (list_start..i).any(|line_idx| {
2943 let line = lines[line_idx];
2944 let trimmed = line.trim();
2945 if trimmed.is_empty() || is_exempt_line(line, line_idx + 1) {
2946 return false;
2947 }
2948 self.calculate_effective_length(line) > config.line_length.get()
2949 })
2950 }
2951 ReflowMode::Default => {
2952 // In default mode, only reflow if any individual non-exempt line exceeds limit
2953 (list_start..i).any(|line_idx| {
2954 let line = lines[line_idx];
2955 let trimmed = line.trim();
2956 // Skip blank lines and exempt lines
2957 if trimmed.is_empty() || is_exempt_line(line, line_idx + 1) {
2958 return false;
2959 }
2960 self.calculate_effective_length(line) > config.line_length.get()
2961 })
2962 }
2963 };
2964
2965 // Record this item's frame so its nested children inherit the shift.
2966 // Only a reflowed item's marker actually moves; an unreflowed one keeps
2967 // its source position and so contributes no shift to its children. The
2968 // threshold that decides which following lines are inside this item is
2969 // the normalized marker width, NOT `source_content_col`: the collection
2970 // loop above gathers continuations by that same width, so the frame
2971 // boundary must match it or the two would disagree about ownership of
2972 // lines indented between the normalized and the source content column.
2973 list_shift_stack.push((marker_len, if needs_reflow { code_indent_shift } else { 0 }));
2974
2975 if needs_reflow {
2976 let start_range = ctx.whole_line_byte_range(list_start + 1);
2977 let end_line = i - 1;
2978 let end_range = if end_line == lines.len() - 1 && !ctx.content.ends_with('\n') {
2979 ctx.line_text_byte_range(end_line + 1, 1, lines[end_line].len() + 1)
2980 } else {
2981 ctx.whole_line_byte_range(end_line + 1)
2982 };
2983 let byte_range = start_range.start..end_range.end;
2984
2985 // Reflow each block (paragraphs only, preserve code blocks)
2986 // When line_length = 0 (no limit), use a very large value for reflow
2987 let reflow_line_length = if config.line_length.is_unlimited() {
2988 usize::MAX
2989 } else {
2990 config.line_length.get().saturating_sub(indent_size).max(1)
2991 };
2992 let reflow_options = Self::reflow_options(ctx, config, reflow_line_length);
2993
2994 let mut result: Vec<String> = Vec::new();
2995 let mut is_first_block = true;
2996
2997 for (block_idx, block) in blocks.iter().enumerate() {
2998 match block {
2999 Block::Paragraph(para_lines) => {
3000 // If every line in this paragraph is exempt (link ref defs,
3001 // standalone links), preserve the paragraph verbatim instead
3002 // of reflowing it. Reflowing would corrupt link ref defs.
3003 let all_exempt = para_lines
3004 .iter()
3005 .all(|(line, line_num)| is_exempt_line(line, *line_num));
3006
3007 if all_exempt {
3008 for (idx, (line, _)) in para_lines.iter().enumerate() {
3009 if is_first_block && idx == 0 {
3010 result.push(format!("{marker}{line}"));
3011 is_first_block = false;
3012 } else {
3013 result.push(format!("{expected_indent}{line}"));
3014 }
3015 }
3016 } else {
3017 // Split the paragraph into segments at hard break boundaries
3018 // Each segment can be reflowed independently
3019 let segments = split_into_segments(para_lines);
3020
3021 for (segment_idx, segment) in segments.iter().enumerate() {
3022 // Check if this segment ends with a hard break and what type
3023 let hard_break_type = segment.last().and_then(|(line, _)| {
3024 let line = line.strip_suffix('\r').unwrap_or(line);
3025 if line.ends_with('\\') {
3026 Some("\\")
3027 } else if line.ends_with(" ") {
3028 Some(" ")
3029 } else {
3030 None
3031 }
3032 });
3033
3034 // Join and reflow the segment (removing the hard break marker for processing)
3035 let segment_for_reflow: Vec<String> = segment
3036 .iter()
3037 .map(|(line, _)| {
3038 // Strip hard break marker (2 spaces or backslash) for reflow processing
3039 if line.ends_with('\\') {
3040 line[..line.len() - 1].trim_end().to_string()
3041 } else if line.ends_with(" ") {
3042 line[..line.len() - 2].trim_end().to_string()
3043 } else {
3044 line.clone()
3045 }
3046 })
3047 .collect();
3048
3049 let segment_text =
3050 join_soft_break_lines(&segment_for_reflow, config.cjk_soft_break)
3051 .trim()
3052 .to_string();
3053 if !segment_text.is_empty() {
3054 let reflowed =
3055 crate::utils::text_reflow::reflow_line(&segment_text, &reflow_options);
3056
3057 if is_first_block && segment_idx == 0 {
3058 // First segment of first block starts with marker
3059 result.push(format!("{marker}{}", reflowed[0]));
3060 for line in reflowed.iter().skip(1) {
3061 result.push(format!("{expected_indent}{line}"));
3062 }
3063 is_first_block = false;
3064 } else {
3065 // Subsequent segments
3066 for line in reflowed {
3067 result.push(format!("{expected_indent}{line}"));
3068 }
3069 }
3070
3071 // If this segment had a hard break, add it back to the last line
3072 // Preserve the original hard break format (backslash or two spaces)
3073 if let Some(break_marker) = hard_break_type
3074 && let Some(last_line) = result.last_mut()
3075 {
3076 last_line.push_str(break_marker);
3077 }
3078 }
3079 }
3080 }
3081
3082 // Add blank line after paragraph block if there's a next block.
3083 // Check if next block is a code block that doesn't want a preceding blank.
3084 // Also don't add blank lines before snippet lines (they should stay tight).
3085 // Only add if not already ending with one (avoids double blanks).
3086 if block_idx < blocks.len() - 1 {
3087 let next_block = &blocks[block_idx + 1];
3088 let should_add_blank = match next_block {
3089 Block::Code {
3090 has_preceding_blank, ..
3091 } => *has_preceding_blank,
3092 Block::Table {
3093 has_preceding_blank, ..
3094 } => *has_preceding_blank,
3095 Block::SnippetLine(_) | Block::DivMarker(_) => false,
3096 _ => true, // For all other blocks, add blank line
3097 };
3098 if should_add_blank && result.last().is_none_or(|s: &String| !s.is_empty()) {
3099 result.push(String::new());
3100 }
3101 }
3102 }
3103 Block::Code {
3104 lines: code_lines,
3105 has_preceding_blank: _,
3106 } => {
3107 // Preserve code blocks as-is with original indentation
3108 // NOTE: Blank line before code block is handled by the previous block
3109 // (see paragraph block's logic above)
3110
3111 for (idx, (content, orig_indent)) in code_lines.iter().enumerate() {
3112 if is_first_block && idx == 0 {
3113 // First line of first block gets marker
3114 result.push(format!(
3115 "{marker}{}",
3116 " ".repeat(orig_indent - marker_len) + content.as_str()
3117 ));
3118 is_first_block = false;
3119 } else if content.is_empty() {
3120 result.push(String::new());
3121 } else {
3122 // Shift nested code with the marker so it stays
3123 // aligned under content when MD030 widens spacing.
3124 result.push(format!(
3125 "{}{}",
3126 " ".repeat((*orig_indent as isize + code_indent_shift).max(0) as usize),
3127 content
3128 ));
3129 }
3130 }
3131 }
3132 Block::SemanticLine(content) => {
3133 // Preserve semantic lines (NOTE:, WARNING:, etc.) as-is on their own line.
3134 // Only add blank before if not already ending with one.
3135 if !is_first_block && result.last().is_none_or(|s: &String| !s.is_empty()) {
3136 result.push(String::new());
3137 }
3138
3139 if is_first_block {
3140 // First block starts with marker
3141 result.push(format!("{marker}{content}"));
3142 is_first_block = false;
3143 } else {
3144 // Subsequent blocks use expected indent
3145 result.push(format!("{expected_indent}{content}"));
3146 }
3147
3148 // Add blank line after semantic line if there's a next block.
3149 // Only add if not already ending with one.
3150 if block_idx < blocks.len() - 1 {
3151 let next_block = &blocks[block_idx + 1];
3152 let should_add_blank = match next_block {
3153 Block::Code {
3154 has_preceding_blank, ..
3155 } => *has_preceding_blank,
3156 Block::Table {
3157 has_preceding_blank, ..
3158 } => *has_preceding_blank,
3159 Block::SnippetLine(_) | Block::DivMarker(_) => false,
3160 _ => true, // For all other blocks, add blank line
3161 };
3162 if should_add_blank && result.last().is_none_or(|s: &String| !s.is_empty()) {
3163 result.push(String::new());
3164 }
3165 }
3166 }
3167 Block::SnippetLine(content) => {
3168 // Preserve snippet delimiters (-8<-) as-is on their own line
3169 // Unlike semantic lines, snippet lines don't add extra blank lines
3170 if is_first_block {
3171 // First block starts with marker
3172 result.push(format!("{marker}{content}"));
3173 is_first_block = false;
3174 } else {
3175 // Subsequent blocks use expected indent
3176 result.push(format!("{expected_indent}{content}"));
3177 }
3178 // No blank lines added before or after snippet delimiters
3179 }
3180 Block::DivMarker(content) => {
3181 // Preserve div markers (::: opening or closing) as-is on their own line
3182 if is_first_block {
3183 result.push(format!("{marker}{content}"));
3184 is_first_block = false;
3185 } else {
3186 result.push(format!("{expected_indent}{content}"));
3187 }
3188 }
3189 Block::Html {
3190 lines: html_lines,
3191 has_preceding_blank: _,
3192 } => {
3193 // Preserve HTML blocks exactly as-is with original indentation
3194 // NOTE: Blank line before HTML block is handled by the previous block
3195
3196 for (idx, line) in html_lines.iter().enumerate() {
3197 if is_first_block && idx == 0 {
3198 // First line of first block gets marker
3199 result.push(format!("{marker}{line}"));
3200 is_first_block = false;
3201 } else if line.is_empty() {
3202 // Preserve blank lines inside HTML blocks
3203 result.push(String::new());
3204 } else {
3205 // Preserve lines with their original content (already includes indentation)
3206 result.push(format!("{expected_indent}{line}"));
3207 }
3208 }
3209
3210 // Add blank line after HTML block if there's a next block.
3211 // Only add if not already ending with one (avoids double blanks
3212 // when the HTML block itself contained a trailing blank line).
3213 if block_idx < blocks.len() - 1 {
3214 let next_block = &blocks[block_idx + 1];
3215 let should_add_blank = match next_block {
3216 Block::Code {
3217 has_preceding_blank, ..
3218 } => *has_preceding_blank,
3219 Block::Html {
3220 has_preceding_blank, ..
3221 } => *has_preceding_blank,
3222 Block::Table {
3223 has_preceding_blank, ..
3224 } => *has_preceding_blank,
3225 Block::SnippetLine(_) | Block::DivMarker(_) => false,
3226 _ => true, // For all other blocks, add blank line
3227 };
3228 if should_add_blank && result.last().is_none_or(|s: &String| !s.is_empty()) {
3229 result.push(String::new());
3230 }
3231 }
3232 }
3233 Block::Table {
3234 lines: table_lines,
3235 has_preceding_blank: _,
3236 } => {
3237 // Preserve table rows verbatim with their original indentation.
3238 // Reflowing rows would corrupt column alignment and inject `|`
3239 // characters mid-paragraph (issue #590).
3240 // The leading blank line is emitted by the previous block.
3241 for (idx, (content, orig_indent)) in table_lines.iter().enumerate() {
3242 if is_first_block && idx == 0 {
3243 // First line of first block gets the list marker
3244 result.push(format!(
3245 "{marker}{}",
3246 " ".repeat(orig_indent.saturating_sub(marker_len)) + content.as_str()
3247 ));
3248 is_first_block = false;
3249 } else {
3250 // Shift nested table rows with the marker so they
3251 // stay aligned when MD030 widens marker spacing.
3252 result.push(format!(
3253 "{}{}",
3254 " ".repeat((*orig_indent as isize + code_indent_shift).max(0) as usize),
3255 content
3256 ));
3257 }
3258 }
3259
3260 // Add blank line after table block if there's a next block.
3261 if block_idx < blocks.len() - 1 {
3262 let next_block = &blocks[block_idx + 1];
3263 let should_add_blank = match next_block {
3264 Block::Code {
3265 has_preceding_blank, ..
3266 } => *has_preceding_blank,
3267 Block::Html {
3268 has_preceding_blank, ..
3269 } => *has_preceding_blank,
3270 Block::Table {
3271 has_preceding_blank, ..
3272 } => *has_preceding_blank,
3273 Block::SnippetLine(_) | Block::DivMarker(_) => false,
3274 _ => true,
3275 };
3276 if should_add_blank && result.last().is_none_or(|s: &String| !s.is_empty()) {
3277 result.push(String::new());
3278 }
3279 }
3280 }
3281 Block::Admonition {
3282 header,
3283 header_indent,
3284 content_lines: admon_lines,
3285 } => {
3286 // Reconstruct admonition block with header at original indent
3287 // and body content reflowed to fit within the line length limit
3288
3289 // Add blank line before admonition if not first block
3290 if !is_first_block && result.last().is_none_or(|s: &String| !s.is_empty()) {
3291 result.push(String::new());
3292 }
3293
3294 // Output the header at its original indent
3295 let header_indent_str = " ".repeat(*header_indent);
3296 if is_first_block {
3297 result.push(format!(
3298 "{marker}{}",
3299 " ".repeat(header_indent.saturating_sub(marker_len)) + header.as_str()
3300 ));
3301 is_first_block = false;
3302 } else {
3303 result.push(format!("{header_indent_str}{header}"));
3304 }
3305
3306 // Derive body indent from the first non-empty content line's
3307 // stored indent, falling back to header_indent + 4 for
3308 // empty-body admonitions
3309 let body_indent = admon_lines
3310 .iter()
3311 .find(|(content, _)| !content.is_empty())
3312 .map_or(header_indent + 4, |(_, indent)| *indent);
3313 let body_indent_str = " ".repeat(body_indent);
3314
3315 // Segment body content into code blocks (verbatim) and
3316 // text paragraphs (reflowable), separated by blank lines.
3317 // Code lines store (content, orig_indent) to reconstruct
3318 // internal indentation relative to body_indent.
3319 enum AdmonSegment {
3320 Text(Vec<String>),
3321 Code(Vec<(String, usize)>),
3322 }
3323
3324 let mut segments: Vec<AdmonSegment> = Vec::new();
3325 let mut current_text: Vec<String> = Vec::new();
3326 let mut current_code: Vec<(String, usize)> = Vec::new();
3327 let mut in_admon_code = false;
3328 // Track the opening fence character so closing fences
3329 // must match (backticks close backticks, tildes close tildes)
3330 let mut fence_char: char = '`';
3331
3332 // Opening fences: ``` or ~~~ followed by optional info string
3333 let get_opening_fence = |s: &str| -> Option<(char, usize)> {
3334 let t = s.trim_start();
3335 if t.starts_with("```") {
3336 Some(('`', t.bytes().take_while(|&b| b == b'`').count()))
3337 } else if t.starts_with("~~~") {
3338 Some(('~', t.bytes().take_while(|&b| b == b'~').count()))
3339 } else {
3340 None
3341 }
3342 };
3343 // Closing fences: ONLY fence chars + optional trailing spaces
3344 let get_closing_fence = |s: &str| -> Option<(char, usize)> {
3345 let t = s.trim();
3346 if t.starts_with("```") && t.bytes().all(|b| b == b'`') {
3347 Some(('`', t.len()))
3348 } else if t.starts_with("~~~") && t.bytes().all(|b| b == b'~') {
3349 Some(('~', t.len()))
3350 } else {
3351 None
3352 }
3353 };
3354 let mut fence_len: usize = 3;
3355
3356 for (content, orig_indent) in admon_lines {
3357 if in_admon_code {
3358 // Closing fence must use the same character, be
3359 // at least as long, and have no info string
3360 if let Some((ch, len)) = get_closing_fence(content)
3361 && ch == fence_char
3362 && len >= fence_len
3363 {
3364 current_code.push((content.clone(), *orig_indent));
3365 in_admon_code = false;
3366 segments.push(AdmonSegment::Code(std::mem::take(&mut current_code)));
3367 continue;
3368 }
3369 current_code.push((content.clone(), *orig_indent));
3370 } else if let Some((ch, len)) = get_opening_fence(content) {
3371 if !current_text.is_empty() {
3372 segments.push(AdmonSegment::Text(std::mem::take(&mut current_text)));
3373 }
3374 in_admon_code = true;
3375 fence_char = ch;
3376 fence_len = len;
3377 current_code.push((content.clone(), *orig_indent));
3378 } else if content.is_empty() {
3379 if !current_text.is_empty() {
3380 segments.push(AdmonSegment::Text(std::mem::take(&mut current_text)));
3381 }
3382 } else {
3383 current_text.push(content.clone());
3384 }
3385 }
3386 if in_admon_code && !current_code.is_empty() {
3387 segments.push(AdmonSegment::Code(std::mem::take(&mut current_code)));
3388 }
3389 if !current_text.is_empty() {
3390 segments.push(AdmonSegment::Text(std::mem::take(&mut current_text)));
3391 }
3392
3393 // Build reflow options once for all text segments
3394 let admon_reflow_length = if config.line_length.is_unlimited() {
3395 usize::MAX
3396 } else {
3397 config.line_length.get().saturating_sub(body_indent).max(1)
3398 };
3399
3400 let admon_reflow_options = Self::reflow_options(ctx, config, admon_reflow_length);
3401
3402 // Output each segment
3403 for segment in &segments {
3404 // Blank line before each segment (after the header or previous segment)
3405 result.push(String::new());
3406
3407 match segment {
3408 AdmonSegment::Code(lines) => {
3409 for (line, orig_indent) in lines {
3410 if line.is_empty() {
3411 // Preserve blank lines inside code blocks
3412 result.push(String::new());
3413 } else {
3414 // Reconstruct with body_indent + any extra
3415 // indentation the line had beyond body_indent
3416 let extra = orig_indent.saturating_sub(body_indent);
3417 let indent_str = " ".repeat(body_indent + extra);
3418 result.push(format!("{indent_str}{line}"));
3419 }
3420 }
3421 }
3422 AdmonSegment::Text(lines) => {
3423 let paragraph_text =
3424 join_soft_break_lines(lines, config.cjk_soft_break).trim().to_string();
3425 if paragraph_text.is_empty() {
3426 continue;
3427 }
3428 let reflowed = crate::utils::text_reflow::reflow_line(
3429 ¶graph_text,
3430 &admon_reflow_options,
3431 );
3432 for line in &reflowed {
3433 result.push(format!("{body_indent_str}{line}"));
3434 }
3435 }
3436 }
3437 }
3438
3439 // Add blank line after admonition if there's a next block
3440 if block_idx < blocks.len() - 1 {
3441 let next_block = &blocks[block_idx + 1];
3442 let should_add_blank = match next_block {
3443 Block::Code {
3444 has_preceding_blank, ..
3445 } => *has_preceding_blank,
3446 Block::Table {
3447 has_preceding_blank, ..
3448 } => *has_preceding_blank,
3449 Block::SnippetLine(_) | Block::DivMarker(_) => false,
3450 _ => true,
3451 };
3452 if should_add_blank && result.last().is_none_or(|s: &String| !s.is_empty()) {
3453 result.push(String::new());
3454 }
3455 }
3456 }
3457 }
3458 }
3459
3460 let reflowed_text = result.join(line_ending);
3461
3462 // Preserve trailing newline
3463 let replacement = if end_line < lines.len() - 1 || ctx.content.ends_with('\n') {
3464 format!("{reflowed_text}{line_ending}")
3465 } else {
3466 reflowed_text
3467 };
3468
3469 // Get the original text to compare
3470 let original_text = &ctx.content[byte_range.clone()];
3471
3472 // Physical-line-length scan, shared by the Normalize-mode gate and its
3473 // message. The list-item reflow preserves code blocks, HTML blocks,
3474 // admonition headers, fence markers, semantic markers, and snippet/div
3475 // markers verbatim; only paragraph content and admonition bodies are
3476 // restructured. Only those lines drive the length warning, so that
3477 // preserved-but-overlong content does not keep the paragraph-level
3478 // warning alive when the reflow would not fix that line.
3479 let should_count_for_length = |line_idx: usize| -> bool {
3480 let line = lines[line_idx];
3481 let trimmed = line.trim();
3482 if trimmed.is_empty() || is_exempt_line(line, line_idx + 1) {
3483 return false;
3484 }
3485 let info = &ctx.lines[line_idx];
3486 if info.in_code_block || info.in_html_block {
3487 return false;
3488 }
3489 if info.in_admonition && mkdocs_admonitions::is_admonition_start(line) {
3490 return false;
3491 }
3492 if is_fence_marker(line) || is_semantic_line(line) {
3493 return false;
3494 }
3495 if is_snippet_block_delimiter(line) {
3496 return false;
3497 }
3498 if line.trim_start().starts_with(":::") {
3499 return false;
3500 }
3501 true
3502 };
3503 let max_physical_length = (list_start..i)
3504 .filter(|&idx| should_count_for_length(idx))
3505 .map(|idx| self.calculate_effective_length(lines[idx]))
3506 .max()
3507 .unwrap_or(0);
3508 // `line-length = 0` means "no limit", so no physical line can be
3509 // "over"; the message below then describes a structural join rather
3510 // than a length violation.
3511 let any_paragraph_line_over =
3512 !config.line_length.is_unlimited() && max_physical_length > config.line_length.get();
3513
3514 // Normalize mode reflows list-item prose just like paragraphs:
3515 // joining continuation lines and re-wrapping to `line-length`.
3516 // `prose_changed` is true only when the reflow alters the words or
3517 // line breaks, not when it would merely re-indent continuation
3518 // lines or trim trailing whitespace. Comparing the texts with each
3519 // line's leading and trailing whitespace removed isolates "did the
3520 // words/line breaks change" from "did the surrounding whitespace
3521 // change". Continuation indentation is MD077's responsibility and
3522 // trailing whitespace is MD009's; an MD013 warning for either would
3523 // both duplicate those rules and resurface a persistent advisory on
3524 // already-fitting items that users disable MD013 fixing to avoid.
3525 let prose_changed = {
3526 let stripped = |text: &str| text.lines().map(str::trim).collect::<Vec<_>>().join("\n");
3527 stripped(original_text) != stripped(&replacement)
3528 };
3529 // Warn when the reflow rewraps prose (the normalize feature for
3530 // list items), or when a physical line genuinely exceeds the limit
3531 // and the reflow can change something (a true length violation,
3532 // even if all that changes is the continuation indent). A line that
3533 // is already optimal in both respects produces no warning.
3534 let gate_ok = prose_changed || (any_paragraph_line_over && original_text != replacement);
3535 if gate_ok {
3536 // Generate an appropriate message based on why reflow is needed
3537 let message = match config.reflow_mode {
3538 ReflowMode::SentencePerLine => {
3539 let num_sentences = split_into_sentences(
3540 &combined_content,
3541 Some(&defined_references),
3542 config.require_sentence_capital,
3543 )
3544 .len();
3545 let num_lines = content_lines.len();
3546 if num_lines == 1 {
3547 // Single line with multiple sentences
3548 format!("Line contains {num_sentences} sentences (one sentence per line required)")
3549 } else {
3550 // Multiple lines - could be split sentences or mixed
3551 format!(
3552 "Paragraph should have one sentence per line (found {num_sentences} sentences across {num_lines} lines)"
3553 )
3554 }
3555 }
3556 ReflowMode::SemanticLineBreaks => {
3557 let num_sentences = split_into_sentences(
3558 &combined_content,
3559 Some(&defined_references),
3560 config.require_sentence_capital,
3561 )
3562 .len();
3563 format!("Paragraph should use semantic line breaks ({num_sentences} sentences)")
3564 }
3565 ReflowMode::Normalize => {
3566 // When a physical line genuinely exceeds the limit, report
3567 // it as a length violation. Otherwise the reflow is a
3568 // structural normalization (joining/re-wrapping multi-line
3569 // content that already fits), mirroring the paragraph path.
3570 if any_paragraph_line_over {
3571 format!(
3572 "Line length {} exceeds {} characters",
3573 max_physical_length,
3574 config.line_length.get()
3575 )
3576 } else {
3577 format!(
3578 "List item could be normalized to use line length of {} characters",
3579 config.line_length.get()
3580 )
3581 }
3582 }
3583 ReflowMode::Default => {
3584 // Report the actual longest non-exempt line, not the combined content
3585 let max_length = (list_start..i)
3586 .filter(|&line_idx| {
3587 let line = lines[line_idx];
3588 let trimmed = line.trim();
3589 !trimmed.is_empty() && !is_exempt_line(line, line_idx + 1)
3590 })
3591 .map(|line_idx| self.calculate_effective_length(lines[line_idx]))
3592 .max()
3593 .unwrap_or(0);
3594 format!(
3595 "Line length {} exceeds {} characters",
3596 max_length,
3597 config.line_length.get()
3598 )
3599 }
3600 };
3601
3602 warnings.push(LintWarning {
3603 rule_name: Some(self.name().to_string()),
3604 message,
3605 line: list_start + 1,
3606 column: 1,
3607 end_line: end_line + 1,
3608 end_column: lines[end_line].chars().count() + 1,
3609 severity: Severity::Warning,
3610 fix: Some(crate::rule::Fix::new(byte_range, replacement)),
3611 });
3612 }
3613 }
3614 continue;
3615 }
3616
3617 // A definition's text reflows below, inside the definition. The rest
3618 // of a definition list is left as written: a term is one line of its
3619 // own, and a line holding a definition's marker but none of its text
3620 // opens a block the definition nests (a list, a quote, a code block)
3621 // or leaves the text for the lines below it.
3622 let definition_text = ctx.definition_text_at(line_num).cloned();
3623 if definition_text.is_none() && ctx.is_in_definition_list(line_num) {
3624 i += 1;
3625 continue;
3626 }
3627
3628 // Found start of a paragraph - collect all lines in it
3629 let paragraph_start = i;
3630 let mut paragraph_lines = vec![lines[i]];
3631 i += 1;
3632
3633 while i < lines.len() {
3634 let next_line = lines[i];
3635 let next_line_num = i + 1;
3636 let next_trimmed = next_line.trim();
3637
3638 // Stop at paragraph boundaries
3639 if next_trimmed.is_empty()
3640 || ctx.line_info(next_line_num).is_some_and(|info| info.in_code_block)
3641 || ctx.line_info(next_line_num).is_some_and(|info| info.in_front_matter)
3642 || ctx.line_info(next_line_num).is_some_and(|info| info.in_html_block)
3643 || ctx.line_info(next_line_num).is_some_and(|info| info.in_html_comment)
3644 || ctx.line_info(next_line_num).is_some_and(|info| info.in_esm_block)
3645 || ctx.line_info(next_line_num).is_some_and(|info| info.in_jsx_expression)
3646 || ctx.line_info(next_line_num).is_some_and(|info| info.in_jsx_block)
3647 || ctx.line_info(next_line_num).is_some_and(|info| info.in_mdx_comment)
3648 || ctx
3649 .line_info(next_line_num)
3650 .is_some_and(super::super::lint_context::types::LineInfo::in_mkdocs_container)
3651 || (next_line_num > 0
3652 && next_line_num <= ctx.lines.len()
3653 && ctx.lines[next_line_num - 1].blockquote.is_some())
3654 || next_trimmed.starts_with('#')
3655 // A setext heading ends the paragraph before it. The flag is
3656 // set on every line of the heading's text, so the paragraph
3657 // stops at the first of them and absorbs no heading text.
3658 // That also protects the underline, which can only follow the
3659 // last text line, so absorbing the construct and joining it
3660 // into prose is unreachable from here. `is_horizontal_rule`
3661 // below catches a `---` underline only by coincidence (3+
3662 // dashes are also a thematic break); `=` and short `-` runs
3663 // have no such overlap.
3664 || is_setext_heading_text_line(ctx, next_line_num)
3665 || TableUtils::is_potential_table_row_with_flavor(next_line, ctx.flavor)
3666 || is_list_item(next_trimmed)
3667 || is_horizontal_rule(next_line)
3668 || (next_trimmed.starts_with('[') && next_line.contains("]:"))
3669 || is_template_directive_only(next_line)
3670 || is_standalone_attr_list(next_line)
3671 || is_snippet_block_delimiter(next_line)
3672 || ctx.line_info(next_line_num).is_some_and(|info| info.is_div_marker)
3673 || is_html_only_line(next_line)
3674 || self.line_in_multiline_math_block(next_line_num, ctx, config)
3675 // A line that is one whole `$$...$$` expression renders as a
3676 // display block, so it ends the paragraph above it and is
3677 // reflowed on its own. A line touched by a code span crossing
3678 // one of its boundaries is code, not such a block.
3679 || ((is_self_contained_display_math_line(next_line)
3680 || self.line_is_standalone_bracket_math(next_line_num, ctx, config))
3681 && !line_touches_multiline_code_span(&code_span_touches, next_line_num))
3682 || standalone_link_ends_paragraph(ctx, next_line_num, config)
3683 // A definition's text ends where the parser ends it.
3684 || definition_text
3685 .as_ref()
3686 .is_some_and(|text| next_line_num > text.end_line)
3687 {
3688 break;
3689 }
3690
3691 // Check if the previous line ends with a hard break (2+ spaces or backslash)
3692 if i > 0 && has_hard_break(lines[i - 1]) {
3693 // Don't include lines after hard breaks in the same paragraph
3694 break;
3695 }
3696
3697 paragraph_lines.push(next_line);
3698 i += 1;
3699 }
3700
3701 // Compute the common leading indent of all non-empty paragraph lines,
3702 // but only when those lines are structurally inside a list block.
3703 // Indented continuation lines that follow a nested list arrive here
3704 // with their structural indentation intact (e.g. 2 spaces for a
3705 // top-level list item). Stripping the indent before reflow and
3706 // re-applying it afterward prevents the fixer from moving those
3707 // lines to column 0.
3708 //
3709 // The list-block guard is essential: top-level paragraphs that happen
3710 // to start with spaces (insignificant in Markdown) must NOT have those
3711 // spaces preserved or injected by the fixer.
3712 let common_indent: String = if ctx.is_in_list_block(paragraph_start + 1) {
3713 let min_len = paragraph_lines
3714 .iter()
3715 .filter(|l| !l.trim().is_empty())
3716 .map(|l| l.len() - l.trim_start().len())
3717 .min()
3718 .unwrap_or(0);
3719 paragraph_lines
3720 .iter()
3721 .find(|l| !l.trim().is_empty())
3722 .map(|l| l[..min_len].to_string())
3723 .unwrap_or_default()
3724 } else {
3725 String::new()
3726 };
3727
3728 // What the reflowed lines are written after. A definition's text
3729 // starting on its marker line keeps the marker and the spacing the
3730 // author gave it, and its other lines are indented to where the text
3731 // starts, which is the definition's content column, or to four
3732 // columns when that is less. Text starting on a later line, a later
3733 // paragraph of the definition or the part after a hard break, keeps
3734 // the indentation its first line has. Everything else takes the
3735 // common indent: the first line and the rest alike.
3736 let (first_prefix, rest_indent) = match &definition_text {
3737 Some(text) => {
3738 let indent_width = |line: &str| {
3739 crate::utils::calculate_indentation_width_default(&line[..line.len() - line.trim_start().len()])
3740 };
3741 let text_lines = &lines[text.start_line - 1..text.end_line];
3742 let marker_prefix = text.marker_prefix_len.map(|len| &text_lines[0][..len]);
3743 // The colon takes one column, and a tab after it reaches the
3744 // next tab stop. The other lines take at least four columns
3745 // even after a narrower marker (`: text`): they continue the
3746 // paragraph at any indentation in CommonMark, but
3747 // Python-Markdown keeps a line indented less than four in the
3748 // definition only when no other definition follows it
3749 // directly, and otherwise reads the text as terms of a new
3750 // list.
3751 let continuation = match marker_prefix {
3752 Some(prefix) => {
3753 crate::utils::calculate_indentation_width_default(&prefix.replacen(':', " ", 1)).max(4)
3754 }
3755 None => indent_width(text_lines[0]),
3756 };
3757 // A hard break splits the text into parts reflowed one at a
3758 // time. With a line after the break indented less than the
3759 // continuation, reflowing one part would leave the parts
3760 // indented differently, and Python-Markdown ends the
3761 // definition at the first line indented less than the lines
3762 // before it, so the text is left as written.
3763 let split_by_hard_break = text_lines[..text_lines.len() - 1]
3764 .iter()
3765 .any(|line| has_hard_break(line));
3766 if split_by_hard_break && text_lines[1..].iter().any(|line| indent_width(line) < continuation) {
3767 i = paragraph_start + paragraph_lines.len();
3768 continue;
3769 }
3770 match marker_prefix.filter(|_| text.start_line == paragraph_start + 1) {
3771 Some(prefix) => (prefix.to_string(), " ".repeat(continuation)),
3772 None => {
3773 let first_line = lines[paragraph_start];
3774 let indent = &first_line[..first_line.len() - first_line.trim_start().len()];
3775 (indent.to_string(), indent.to_string())
3776 }
3777 }
3778 }
3779 None => (common_indent.clone(), common_indent.clone()),
3780 };
3781
3782 // Combine paragraph lines into a single string for processing.
3783 // This must be done BEFORE the needs_reflow check for sentence-per-line mode.
3784 let paragraph_text = if definition_text.is_some() {
3785 let stripped: Vec<&str> = paragraph_lines
3786 .iter()
3787 .enumerate()
3788 .map(|(idx, l)| {
3789 if idx == 0 {
3790 &l[first_prefix.len()..]
3791 } else {
3792 l.trim_start()
3793 }
3794 })
3795 .collect();
3796 join_soft_break_lines(&stripped, config.cjk_soft_break)
3797 } else if common_indent.is_empty() {
3798 join_soft_break_lines(¶graph_lines, config.cjk_soft_break)
3799 } else {
3800 let stripped: Vec<&str> = paragraph_lines
3801 .iter()
3802 .map(|l| {
3803 if l.starts_with(common_indent.as_str()) {
3804 &l[common_indent.len()..]
3805 } else {
3806 l.trim_start()
3807 }
3808 })
3809 .collect();
3810 join_soft_break_lines(&stripped, config.cjk_soft_break)
3811 };
3812
3813 // A colon-led line with a line of the paragraph before it opens a
3814 // definition, and joining the lines would flatten the definition
3815 // list into prose, so the paragraph is skipped. Its first line is
3816 // prose whatever it starts with, since a definition needs a term on
3817 // the line before it. The indentation a marker is allowed is
3818 // counted from the block's own content, so a list item's
3819 // indentation comes off first.
3820 let contains_definition_list = paragraph_lines.iter().skip(1).any(|line| {
3821 let content = line.strip_prefix(rest_indent.as_str()).unwrap_or(line.trim_start());
3822 crate::utils::text_reflow::is_definition_list_marker(content)
3823 });
3824
3825 if contains_definition_list {
3826 // Don't reflow definition lists - skip this paragraph
3827 i = paragraph_start + paragraph_lines.len();
3828 continue;
3829 }
3830
3831 // Skip reflowing if this paragraph contains MkDocs Snippets markers
3832 // Snippets blocks (-8<- ... -8<-) should be preserved exactly
3833 let contains_snippets = paragraph_lines.iter().any(|line| is_snippet_block_delimiter(line));
3834
3835 if contains_snippets {
3836 // Don't reflow Snippets blocks - skip this paragraph
3837 i = paragraph_start + paragraph_lines.len();
3838 continue;
3839 }
3840
3841 // Leave a line of a multi-line display-math block as it is, where
3842 // joining lines would corrupt the equation (see
3843 // `line_in_multiline_math_block`). Only the first line has to be
3844 // asked about: such a line ends the paragraph above it, so a
3845 // paragraph reaching here holds one only when it starts on one.
3846 //
3847 // Only that line is passed over, not the rest of what was collected
3848 // with it: prose written directly under the closing delimiter is an
3849 // ordinary paragraph and still reflows.
3850 if self.line_in_multiline_math_block(paragraph_start + 1, ctx, config) {
3851 i = paragraph_start + 1;
3852 continue;
3853 }
3854
3855 // A line that is one whole `$$...$$` expression is a display block:
3856 // it keeps the line it was written on, and the prose under it is an
3857 // ordinary paragraph that still reflows. The line above ended at
3858 // this one, so a paragraph reaching here holds one only when it
3859 // starts on one. A paragraph can start on a line touched by a code
3860 // span crossing one of its boundaries, under a hard break the span
3861 // holds, and its first line is code then.
3862 if (is_self_contained_display_math_line(lines[paragraph_start])
3863 || self.line_is_standalone_bracket_math(paragraph_start + 1, ctx, config))
3864 && !line_touches_multiline_code_span(&code_span_touches, paragraph_start + 1)
3865 {
3866 i = paragraph_start + 1;
3867 continue;
3868 }
3869
3870 // Check if this paragraph needs reflowing
3871 let needs_reflow = match config.reflow_mode {
3872 ReflowMode::Normalize => self.normalize_mode_needs_reflow(paragraph_lines.iter().copied(), config),
3873 ReflowMode::SentencePerLine => {
3874 // In sentence-per-line mode, check if the JOINED paragraph has multiple sentences
3875 // Note: we check the joined text because sentences can span multiple lines
3876 let sentences = split_into_sentences(
3877 ¶graph_text,
3878 Some(&defined_references),
3879 config.require_sentence_capital,
3880 );
3881
3882 // Always reflow if multiple sentences on one line
3883 if sentences.len() > 1 {
3884 true
3885 } else if paragraph_lines.len() > 1 {
3886 // For single-sentence paragraphs spanning multiple lines:
3887 // Reflow if they COULD fit on one line (respecting line-length constraint)
3888 if config.line_length.is_unlimited() {
3889 // No line-length constraint - always join single sentences
3890 true
3891 } else {
3892 // Only join if it fits within line-length.
3893 // paragraph_text has the prefix stripped, so add back
3894 // the prefix of the first line, the one line the
3895 // joined sentence is written on.
3896 let effective_length =
3897 self.calculate_effective_length(¶graph_text) + first_prefix.len();
3898 effective_length <= config.line_length.get()
3899 }
3900 } else {
3901 false
3902 }
3903 }
3904 ReflowMode::SemanticLineBreaks => {
3905 let sentences = split_into_sentences(
3906 ¶graph_text,
3907 Some(&defined_references),
3908 config.require_sentence_capital,
3909 );
3910 // Reflow if multiple sentences, multiple lines, or any line exceeds limit
3911 sentences.len() > 1
3912 || paragraph_lines.len() > 1
3913 || paragraph_lines
3914 .iter()
3915 .any(|line| self.calculate_effective_length(line) > config.line_length.get())
3916 }
3917 ReflowMode::Default => {
3918 // In default mode, only reflow if lines exceed limit
3919 paragraph_lines
3920 .iter()
3921 .any(|line| self.calculate_effective_length(line) > config.line_length.get())
3922 }
3923 };
3924
3925 if needs_reflow {
3926 // Calculate byte range for this paragraph
3927 // Use whole_line_range for each line and combine
3928 let start_range = ctx.whole_line_byte_range(paragraph_start + 1);
3929 let end_line = paragraph_start + paragraph_lines.len() - 1;
3930
3931 // For the last line, we want to preserve any trailing newline
3932 let end_range = if end_line == lines.len() - 1 && !ctx.content.ends_with('\n') {
3933 // Last line without trailing newline - use line_text_range
3934 ctx.line_text_byte_range(end_line + 1, 1, lines[end_line].len() + 1)
3935 } else {
3936 // Not the last line or has trailing newline - use whole_line_range
3937 ctx.whole_line_byte_range(end_line + 1)
3938 };
3939
3940 let byte_range = start_range.start..end_range.end;
3941
3942 // Check if the paragraph ends with a hard break and what type
3943 let hard_break_type = paragraph_lines.last().and_then(|line| {
3944 let line = line.strip_suffix('\r').unwrap_or(line);
3945 if line.ends_with('\\') {
3946 Some("\\")
3947 } else if line.ends_with(" ") {
3948 Some(" ")
3949 } else {
3950 None
3951 }
3952 });
3953
3954 // Reflow the paragraph
3955 // When line_length = 0 (no limit), use a very large value for reflow
3956 let reflow_line_length = if config.line_length.is_unlimited() {
3957 usize::MAX
3958 } else {
3959 config.line_length.get().saturating_sub(rest_indent.len()).max(1)
3960 };
3961 let reflow_options = Self::reflow_options(ctx, config, reflow_line_length);
3962 let mut reflowed = crate::utils::text_reflow::reflow_line(¶graph_text, &reflow_options);
3963
3964 // Re-apply the prefix to each non-empty reflowed line so that the
3965 // replacement preserves the original structural indentation.
3966 for (idx, line) in reflowed.iter_mut().enumerate() {
3967 let prefix = if idx == 0 { &first_prefix } else { &rest_indent };
3968 if !line.is_empty() && !prefix.is_empty() {
3969 *line = format!("{prefix}{line}");
3970 }
3971 }
3972
3973 // If the original paragraph ended with a hard break, preserve it
3974 // Preserve the original hard break format (backslash or two spaces)
3975 if let Some(break_marker) = hard_break_type
3976 && !reflowed.is_empty()
3977 {
3978 let last_idx = reflowed.len() - 1;
3979 if !has_hard_break(&reflowed[last_idx]) {
3980 reflowed[last_idx].push_str(break_marker);
3981 }
3982 }
3983
3984 let reflowed_text = reflowed.join(line_ending);
3985
3986 // Preserve trailing newline if the original paragraph had one
3987 let replacement = if end_line < lines.len() - 1 || ctx.content.ends_with('\n') {
3988 format!("{reflowed_text}{line_ending}")
3989 } else {
3990 reflowed_text
3991 };
3992
3993 // Get the original text to compare
3994 let original_text = &ctx.content[byte_range.clone()];
3995
3996 // Only generate a warning if the replacement is different from the original
3997 if original_text != replacement {
3998 // Determine which line ranges and messages to report based on the reflow mode.
3999 let warnings_to_report: Vec<(usize, usize, String)> = match config.reflow_mode {
4000 ReflowMode::Default => {
4001 // In default mode, report a warning for *every* line in the paragraph
4002 // that exceeds the limit. Each warning will carry the same paragraph-level
4003 // fix, making all of them auto-fixable.
4004 paragraph_lines
4005 .iter()
4006 .enumerate()
4007 .filter(|(_, line)| self.calculate_effective_length(line) > config.line_length.get())
4008 .map(|(idx, _)| {
4009 let violating_line = paragraph_start + idx + 1;
4010 (
4011 violating_line,
4012 violating_line,
4013 format!("Line length exceeds {} characters", config.line_length.get()),
4014 )
4015 })
4016 .collect()
4017 }
4018 ReflowMode::Normalize => {
4019 // In normalize mode, report the whole paragraph as needing normalization.
4020 vec![(
4021 paragraph_start + 1,
4022 end_line + 1,
4023 format!(
4024 "Paragraph could be normalized to use line length of {} characters",
4025 config.line_length.get()
4026 ),
4027 )]
4028 }
4029 ReflowMode::SentencePerLine => {
4030 // In sentence-per-line mode, highlight the entire paragraph that needs reformatting.
4031 let num_sentences = split_into_sentences(
4032 ¶graph_text,
4033 Some(&defined_references),
4034 config.require_sentence_capital,
4035 )
4036 .len();
4037 let message = if paragraph_lines.len() == 1 {
4038 // Single line with multiple sentences
4039 format!("Line contains {num_sentences} sentences (one sentence per line required)")
4040 } else {
4041 // Multiple lines - could be split sentences or mixed
4042 let num_lines = paragraph_lines.len();
4043 format!(
4044 "Paragraph should have one sentence per line (found {num_sentences} sentences across {num_lines} lines)"
4045 )
4046 };
4047 vec![(paragraph_start + 1, paragraph_start + paragraph_lines.len(), message)]
4048 }
4049 ReflowMode::SemanticLineBreaks => {
4050 // In semantic-line-breaks mode, highlight the entire paragraph.
4051 let num_sentences = split_into_sentences(
4052 ¶graph_text,
4053 Some(&defined_references),
4054 config.require_sentence_capital,
4055 )
4056 .len();
4057 vec![(
4058 paragraph_start + 1,
4059 paragraph_start + paragraph_lines.len(),
4060 format!("Paragraph should use semantic line breaks ({num_sentences} sentences)"),
4061 )]
4062 }
4063 };
4064
4065 // Generate the actual lint warnings. All warnings for this paragraph
4066 // share the same paragraph-level fix.
4067 for (w_start, w_end, msg) in warnings_to_report {
4068 warnings.push(LintWarning {
4069 rule_name: Some(self.name().to_string()),
4070 message: msg,
4071 line: w_start,
4072 column: 1,
4073 end_line: w_end,
4074 end_column: lines[w_end.saturating_sub(1)].chars().count() + 1,
4075 severity: Severity::Warning,
4076 fix: Some(crate::rule::Fix::new(byte_range.clone(), replacement.clone())),
4077 });
4078 }
4079 }
4080 }
4081 }
4082
4083 warnings
4084 }
4085
4086 /// Calculate string length based on the configured length mode
4087 fn calculate_string_length(&self, s: &str) -> usize {
4088 match self.config.length_mode {
4089 LengthMode::Chars => s.chars().count(),
4090 LengthMode::Visual => s.width(),
4091 LengthMode::Bytes => s.len(),
4092 }
4093 }
4094
4095 /// Calculate effective line length
4096 ///
4097 /// Returns the actual display length of the line using the configured length mode.
4098 fn calculate_effective_length(&self, line: &str) -> usize {
4099 self.calculate_string_length(line)
4100 }
4101
4102 /// Calculate line length with inline link/image URLs removed.
4103 ///
4104 /// For each inline link `[text](url)` or image `` on the line,
4105 /// computes the "savings" from removing the URL portion (keeping only `[text]`
4106 /// or `![alt]`). Returns `effective_length - total_savings`.
4107 ///
4108 /// Handles nested constructs (e.g., `[](url)`) by only counting the
4109 /// outermost construct to avoid double-counting.
4110 fn length_without_inline_link_urls(
4111 &self,
4112 effective_length: usize,
4113 line_number: usize,
4114 ctx: &crate::lint_context::LintContext,
4115 ) -> usize {
4116 let line_range = ctx.line_content_byte_range(line_number);
4117 let line_byte_end = line_range.end;
4118
4119 // Collect inline links/images on this line: (byte_offset, byte_end, text_only_display_len)
4120 let mut constructs: Vec<(usize, usize, usize)> = Vec::new();
4121
4122 for link in ctx.links_on_line(line_number) {
4123 if link.is_reference {
4124 continue;
4125 }
4126 if !matches!(link.link_type, LinkType::Inline) {
4127 continue;
4128 }
4129 if link.byte_end > line_byte_end {
4130 continue;
4131 }
4132 let text_only_len = 2 + self.calculate_string_length(&link.text);
4133 constructs.push((link.byte_offset, link.byte_end, text_only_len));
4134 }
4135
4136 for image in ctx.images_on_line(line_number) {
4137 if image.is_reference {
4138 continue;
4139 }
4140 if !matches!(image.link_type, LinkType::Inline) {
4141 continue;
4142 }
4143 if image.byte_end > line_byte_end {
4144 continue;
4145 }
4146 let text_only_len = 3 + self.calculate_string_length(&image.alt_text);
4147 constructs.push((image.byte_offset, image.byte_end, text_only_len));
4148 }
4149
4150 if constructs.is_empty() {
4151 return effective_length;
4152 }
4153
4154 // Sort by byte offset to handle overlapping/nested constructs
4155 constructs.sort_by_key(|&(start, _, _)| start);
4156
4157 let mut total_savings: usize = 0;
4158 let mut last_end: usize = 0;
4159
4160 for (start, end, text_only_len) in &constructs {
4161 // Skip constructs nested inside a previously counted one
4162 if *start < last_end {
4163 continue;
4164 }
4165 // Full construct length in configured length mode
4166 let full_source = &ctx.content[*start..*end];
4167 let full_len = self.calculate_string_length(full_source);
4168 total_savings += full_len.saturating_sub(*text_only_len);
4169 last_end = *end;
4170 }
4171
4172 effective_length.saturating_sub(total_savings)
4173 }
4174}