rumdl_lib/

lint_context.rs

use crate::config::MarkdownFlavor;
use crate::rules::front_matter_utils::FrontMatterUtils;
use crate::utils::code_block_utils::{CodeBlockContext, CodeBlockUtils};
use pulldown_cmark::{BrokenLink, Event, LinkType, Options, Parser, Tag, TagEnd};
use regex::Regex;
use std::borrow::Cow;
use std::path::PathBuf;
use std::sync::LazyLock;

/// Macro for profiling sections - only active in non-WASM builds
#[cfg(not(target_arch = "wasm32"))]
macro_rules! profile_section {
    ($name:expr, $profile:expr, $code:expr) => {{
        let start = std::time::Instant::now();
        let result = $code;
        if $profile {
            eprintln!("[PROFILE] {}: {:?}", $name, start.elapsed());
        }
        result
    }};
}

#[cfg(target_arch = "wasm32")]
macro_rules! profile_section {
    ($name:expr, $profile:expr, $code:expr) => {{ $code }};
}

// Comprehensive link pattern that captures both inline and reference links
// Use (?s) flag to make . match newlines
static LINK_PATTERN: LazyLock<Regex> = LazyLock::new(|| {
    Regex::new(
        r#"(?sx)
        \[((?:[^\[\]\\]|\\.)*)\]          # Link text in group 1 (optimized - no nested brackets to prevent catastrophic backtracking)
        (?:
            \((?:<([^<>\n]*)>|([^)"']*))(?:\s+(?:"([^"]*)"|'([^']*)'))?\)  # URL in group 2 (angle) or 3 (bare), title in 4/5
            |
            \[([^\]]*)\]      # Reference ID in group 6
        )"#
    ).unwrap()
});

// Image pattern (similar to links but with ! prefix)
// Use (?s) flag to make . match newlines
static IMAGE_PATTERN: LazyLock<Regex> = LazyLock::new(|| {
    Regex::new(
        r#"(?sx)
        !\[((?:[^\[\]\\]|\\.)*)\]         # Alt text in group 1 (optimized - no nested brackets to prevent catastrophic backtracking)
        (?:
            \((?:<([^<>\n]*)>|([^)"']*))(?:\s+(?:"([^"]*)"|'([^']*)'))?\)  # URL in group 2 (angle) or 3 (bare), title in 4/5
            |
            \[([^\]]*)\]      # Reference ID in group 6
        )"#
    ).unwrap()
});

// Reference definition pattern
static REF_DEF_PATTERN: LazyLock<Regex> =
    LazyLock::new(|| Regex::new(r#"(?m)^[ ]{0,3}\[([^\]]+)\]:\s*([^\s]+)(?:\s+(?:"([^"]*)"|'([^']*)'))?$"#).unwrap());

// Pattern for bare URLs
static BARE_URL_PATTERN: LazyLock<Regex> = LazyLock::new(|| {
    Regex::new(
        r#"(https?|ftp)://[^\s<>\[\]()\\'"`]+(?:\.[^\s<>\[\]()\\'"`]+)*(?::\d+)?(?:/[^\s<>\[\]()\\'"`]*)?(?:\?[^\s<>\[\]()\\'"`]*)?(?:#[^\s<>\[\]()\\'"`]*)?"#
    ).unwrap()
});

// Pattern for email addresses
static BARE_EMAIL_PATTERN: LazyLock<Regex> =
    LazyLock::new(|| Regex::new(r"[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}").unwrap());

// Pattern for blockquote prefix in parse_list_blocks
static BLOCKQUOTE_PREFIX_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"^(\s*>+\s*)").unwrap());

/// Pre-computed information about a line
#[derive(Debug, Clone)]
pub struct LineInfo {
    /// Byte offset where this line starts in the document
    pub byte_offset: usize,
    /// Length of the line in bytes (without newline)
    pub byte_len: usize,
    /// Number of leading spaces/tabs
    pub indent: usize,
    /// Whether the line is blank (empty or only whitespace)
    pub is_blank: bool,
    /// Whether this line is inside a code block
    pub in_code_block: bool,
    /// Whether this line is inside front matter
    pub in_front_matter: bool,
    /// Whether this line is inside an HTML block
    pub in_html_block: bool,
    /// Whether this line is inside an HTML comment
    pub in_html_comment: bool,
    /// List item information if this line starts a list item
    pub list_item: Option<ListItemInfo>,
    /// Heading information if this line is a heading
    pub heading: Option<HeadingInfo>,
    /// Blockquote information if this line is a blockquote
    pub blockquote: Option<BlockquoteInfo>,
    /// Whether this line is inside a mkdocstrings autodoc block
    pub in_mkdocstrings: bool,
    /// Whether this line is part of an ESM import/export block (MDX only)
    pub in_esm_block: bool,
    /// Whether this line is a continuation of a multi-line code span from a previous line
    pub in_code_span_continuation: bool,
}

impl LineInfo {
    /// Get the line content as a string slice from the source document
    pub fn content<'a>(&self, source: &'a str) -> &'a str {
        &source[self.byte_offset..self.byte_offset + self.byte_len]
    }
}

/// Information about a list item
#[derive(Debug, Clone)]
pub struct ListItemInfo {
    /// The marker used (*, -, +, or number with . or ))
    pub marker: String,
    /// Whether it's ordered (true) or unordered (false)
    pub is_ordered: bool,
    /// The number for ordered lists
    pub number: Option<usize>,
    /// Column where the marker starts (0-based)
    pub marker_column: usize,
    /// Column where content after marker starts
    pub content_column: usize,
}

/// Heading style type
#[derive(Debug, Clone, PartialEq)]
pub enum HeadingStyle {
    /// ATX style heading (# Heading)
    ATX,
    /// Setext style heading with = underline
    Setext1,
    /// Setext style heading with - underline
    Setext2,
}

/// Parsed link information
#[derive(Debug, Clone)]
pub struct ParsedLink<'a> {
    /// Line number (1-indexed)
    pub line: usize,
    /// Start column (0-indexed) in the line
    pub start_col: usize,
    /// End column (0-indexed) in the line
    pub end_col: usize,
    /// Byte offset in document
    pub byte_offset: usize,
    /// End byte offset in document
    pub byte_end: usize,
    /// Link text
    pub text: Cow<'a, str>,
    /// Link URL or reference
    pub url: Cow<'a, str>,
    /// Whether this is a reference link [text][ref] vs inline [text](url)
    pub is_reference: bool,
    /// Reference ID for reference links
    pub reference_id: Option<Cow<'a, str>>,
    /// Link type from pulldown-cmark
    pub link_type: LinkType,
}

/// Information about a broken link reported by pulldown-cmark
#[derive(Debug, Clone)]
pub struct BrokenLinkInfo {
    /// The reference text that couldn't be resolved
    pub reference: String,
    /// Byte span in the source document
    pub span: std::ops::Range<usize>,
}

/// Parsed footnote reference (e.g., `[^1]`, `[^note]`)
#[derive(Debug, Clone)]
pub struct FootnoteRef {
    /// The footnote ID (without the ^ prefix)
    pub id: String,
    /// Line number (1-indexed)
    pub line: usize,
    /// Start byte offset in document
    pub byte_offset: usize,
    /// End byte offset in document
    pub byte_end: usize,
}

/// Parsed image information
#[derive(Debug, Clone)]
pub struct ParsedImage<'a> {
    /// Line number (1-indexed)
    pub line: usize,
    /// Start column (0-indexed) in the line
    pub start_col: usize,
    /// End column (0-indexed) in the line
    pub end_col: usize,
    /// Byte offset in document
    pub byte_offset: usize,
    /// End byte offset in document
    pub byte_end: usize,
    /// Alt text
    pub alt_text: Cow<'a, str>,
    /// Image URL or reference
    pub url: Cow<'a, str>,
    /// Whether this is a reference image ![alt][ref] vs inline ![alt](url)
    pub is_reference: bool,
    /// Reference ID for reference images
    pub reference_id: Option<Cow<'a, str>>,
    /// Link type from pulldown-cmark
    pub link_type: LinkType,
}

/// Reference definition [ref]: url "title"
#[derive(Debug, Clone)]
pub struct ReferenceDef {
    /// Line number (1-indexed)
    pub line: usize,
    /// Reference ID (normalized to lowercase)
    pub id: String,
    /// URL
    pub url: String,
    /// Optional title
    pub title: Option<String>,
    /// Byte offset where the reference definition starts
    pub byte_offset: usize,
    /// Byte offset where the reference definition ends
    pub byte_end: usize,
}

/// Parsed code span information
#[derive(Debug, Clone)]
pub struct CodeSpan {
    /// Line number where the code span starts (1-indexed)
    pub line: usize,
    /// Line number where the code span ends (1-indexed)
    pub end_line: usize,
    /// Start column (0-indexed) in the line
    pub start_col: usize,
    /// End column (0-indexed) in the line
    pub end_col: usize,
    /// Byte offset in document
    pub byte_offset: usize,
    /// End byte offset in document
    pub byte_end: usize,
    /// Number of backticks used (1, 2, 3, etc.)
    pub backtick_count: usize,
    /// Content inside the code span (without backticks)
    pub content: String,
}

/// Information about a heading
#[derive(Debug, Clone)]
pub struct HeadingInfo {
    /// Heading level (1-6 for ATX, 1-2 for Setext)
    pub level: u8,
    /// Style of heading
    pub style: HeadingStyle,
    /// The heading marker (# characters or underline)
    pub marker: String,
    /// Column where the marker starts (0-based)
    pub marker_column: usize,
    /// Column where heading text starts
    pub content_column: usize,
    /// The heading text (without markers and without custom ID syntax)
    pub text: String,
    /// Custom header ID if present (e.g., from {#custom-id} syntax)
    pub custom_id: Option<String>,
    /// Original heading text including custom ID syntax
    pub raw_text: String,
    /// Whether it has a closing sequence (for ATX)
    pub has_closing_sequence: bool,
    /// The closing sequence if present
    pub closing_sequence: String,
}

/// Information about a blockquote line
#[derive(Debug, Clone)]
pub struct BlockquoteInfo {
    /// Nesting level (1 for >, 2 for >>, etc.)
    pub nesting_level: usize,
    /// The indentation before the blockquote marker
    pub indent: String,
    /// Column where the first > starts (0-based)
    pub marker_column: usize,
    /// The blockquote prefix (e.g., "> ", ">> ", etc.)
    pub prefix: String,
    /// Content after the blockquote marker(s)
    pub content: String,
    /// Whether the line has no space after the marker
    pub has_no_space_after_marker: bool,
    /// Whether the line has multiple spaces after the marker
    pub has_multiple_spaces_after_marker: bool,
    /// Whether this is an empty blockquote line needing MD028 fix
    pub needs_md028_fix: bool,
}

/// Information about a list block
#[derive(Debug, Clone)]
pub struct ListBlock {
    /// Line number where the list starts (1-indexed)
    pub start_line: usize,
    /// Line number where the list ends (1-indexed)
    pub end_line: usize,
    /// Whether it's ordered or unordered
    pub is_ordered: bool,
    /// The consistent marker for unordered lists (if any)
    pub marker: Option<String>,
    /// Blockquote prefix for this list (empty if not in blockquote)
    pub blockquote_prefix: String,
    /// Lines that are list items within this block
    pub item_lines: Vec<usize>,
    /// Nesting level (0 for top-level lists)
    pub nesting_level: usize,
    /// Maximum marker width seen in this block (e.g., 3 for "1. ", 4 for "10. ")
    pub max_marker_width: usize,
}

use std::sync::{Arc, Mutex};

/// Character frequency data for fast content analysis
#[derive(Debug, Clone, Default)]
pub struct CharFrequency {
    /// Count of # characters (headings)
    pub hash_count: usize,
    /// Count of * characters (emphasis, lists, horizontal rules)
    pub asterisk_count: usize,
    /// Count of _ characters (emphasis, horizontal rules)
    pub underscore_count: usize,
    /// Count of - characters (lists, horizontal rules, setext headings)
    pub hyphen_count: usize,
    /// Count of + characters (lists)
    pub plus_count: usize,
    /// Count of > characters (blockquotes)
    pub gt_count: usize,
    /// Count of | characters (tables)
    pub pipe_count: usize,
    /// Count of [ characters (links, images)
    pub bracket_count: usize,
    /// Count of ` characters (code spans, code blocks)
    pub backtick_count: usize,
    /// Count of < characters (HTML tags, autolinks)
    pub lt_count: usize,
    /// Count of ! characters (images)
    pub exclamation_count: usize,
    /// Count of newline characters
    pub newline_count: usize,
}

/// Pre-parsed HTML tag information
#[derive(Debug, Clone)]
pub struct HtmlTag {
    /// Line number (1-indexed)
    pub line: usize,
    /// Start column (0-indexed) in the line
    pub start_col: usize,
    /// End column (0-indexed) in the line
    pub end_col: usize,
    /// Byte offset in document
    pub byte_offset: usize,
    /// End byte offset in document
    pub byte_end: usize,
    /// Tag name (e.g., "div", "img", "br")
    pub tag_name: String,
    /// Whether it's a closing tag (`</tag>`)
    pub is_closing: bool,
    /// Whether it's self-closing (`<tag />`)
    pub is_self_closing: bool,
    /// Raw tag content
    pub raw_content: String,
}

/// Pre-parsed emphasis span information
#[derive(Debug, Clone)]
pub struct EmphasisSpan {
    /// Line number (1-indexed)
    pub line: usize,
    /// Start column (0-indexed) in the line
    pub start_col: usize,
    /// End column (0-indexed) in the line
    pub end_col: usize,
    /// Byte offset in document
    pub byte_offset: usize,
    /// End byte offset in document
    pub byte_end: usize,
    /// Type of emphasis ('*' or '_')
    pub marker: char,
    /// Number of markers (1 for italic, 2 for bold, 3+ for bold+italic)
    pub marker_count: usize,
    /// Content inside the emphasis
    pub content: String,
}

/// Pre-parsed table row information
#[derive(Debug, Clone)]
pub struct TableRow {
    /// Line number (1-indexed)
    pub line: usize,
    /// Whether this is a separator row (contains only |, -, :, and spaces)
    pub is_separator: bool,
    /// Number of columns (pipe-separated cells)
    pub column_count: usize,
    /// Alignment info from separator row
    pub column_alignments: Vec<String>, // "left", "center", "right", "none"
}

/// Pre-parsed bare URL information (not in links)
#[derive(Debug, Clone)]
pub struct BareUrl {
    /// Line number (1-indexed)
    pub line: usize,
    /// Start column (0-indexed) in the line
    pub start_col: usize,
    /// End column (0-indexed) in the line
    pub end_col: usize,
    /// Byte offset in document
    pub byte_offset: usize,
    /// End byte offset in document
    pub byte_end: usize,
    /// The URL string
    pub url: String,
    /// Type of URL ("http", "https", "ftp", "email")
    pub url_type: String,
}

pub struct LintContext<'a> {
    pub content: &'a str,
    pub line_offsets: Vec<usize>,
    pub code_blocks: Vec<(usize, usize)>, // Cached code block ranges (not including inline code spans)
    pub lines: Vec<LineInfo>,             // Pre-computed line information
    pub links: Vec<ParsedLink<'a>>,       // Pre-parsed links
    pub images: Vec<ParsedImage<'a>>,     // Pre-parsed images
    pub broken_links: Vec<BrokenLinkInfo>, // Broken/undefined references
    pub footnote_refs: Vec<FootnoteRef>,  // Pre-parsed footnote references
    pub reference_defs: Vec<ReferenceDef>, // Reference definitions
    code_spans_cache: Mutex<Option<Arc<Vec<CodeSpan>>>>, // Lazy-loaded inline code spans
    pub list_blocks: Vec<ListBlock>,      // Pre-parsed list blocks
    pub char_frequency: CharFrequency,    // Character frequency analysis
    html_tags_cache: Mutex<Option<Arc<Vec<HtmlTag>>>>, // Lazy-loaded HTML tags
    emphasis_spans_cache: Mutex<Option<Arc<Vec<EmphasisSpan>>>>, // Lazy-loaded emphasis spans
    table_rows_cache: Mutex<Option<Arc<Vec<TableRow>>>>, // Lazy-loaded table rows
    bare_urls_cache: Mutex<Option<Arc<Vec<BareUrl>>>>, // Lazy-loaded bare URLs
    html_comment_ranges: Vec<crate::utils::skip_context::ByteRange>, // Pre-computed HTML comment ranges
    pub table_blocks: Vec<crate::utils::table_utils::TableBlock>, // Pre-computed table blocks
    pub line_index: crate::utils::range_utils::LineIndex<'a>, // Pre-computed line index for byte position calculations
    jinja_ranges: Vec<(usize, usize)>,    // Pre-computed Jinja template ranges ({{ }}, {% %})
    pub flavor: MarkdownFlavor,           // Markdown flavor being used
    pub source_file: Option<PathBuf>,     // Source file path (for rules that need file context)
}

/// Detailed blockquote parse result with all components
struct BlockquoteComponents<'a> {
    indent: &'a str,
    markers: &'a str,
    spaces_after: &'a str,
    content: &'a str,
}

/// Parse blockquote prefix with detailed components using manual parsing
#[inline]
fn parse_blockquote_detailed(line: &str) -> Option<BlockquoteComponents<'_>> {
    let bytes = line.as_bytes();
    let mut pos = 0;

    // Parse leading whitespace (indent)
    while pos < bytes.len() && (bytes[pos] == b' ' || bytes[pos] == b'\t') {
        pos += 1;
    }
    let indent_end = pos;

    // Must have at least one '>' marker
    if pos >= bytes.len() || bytes[pos] != b'>' {
        return None;
    }

    // Parse '>' markers
    while pos < bytes.len() && bytes[pos] == b'>' {
        pos += 1;
    }
    let markers_end = pos;

    // Parse spaces after markers
    while pos < bytes.len() && (bytes[pos] == b' ' || bytes[pos] == b'\t') {
        pos += 1;
    }
    let spaces_end = pos;

    Some(BlockquoteComponents {
        indent: &line[0..indent_end],
        markers: &line[indent_end..markers_end],
        spaces_after: &line[markers_end..spaces_end],
        content: &line[spaces_end..],
    })
}

impl<'a> LintContext<'a> {
    pub fn new(content: &'a str, flavor: MarkdownFlavor, source_file: Option<PathBuf>) -> Self {
        #[cfg(not(target_arch = "wasm32"))]
        let profile = std::env::var("RUMDL_PROFILE_QUADRATIC").is_ok();
        #[cfg(target_arch = "wasm32")]
        let profile = false;

        let line_offsets = profile_section!("Line offsets", profile, {
            let mut offsets = vec![0];
            for (i, c) in content.char_indices() {
                if c == '\n' {
                    offsets.push(i + 1);
                }
            }
            offsets
        });

        // Detect code blocks once and cache them
        let code_blocks = profile_section!("Code blocks", profile, CodeBlockUtils::detect_code_blocks(content));

        // Pre-compute HTML comment ranges ONCE for all operations
        let html_comment_ranges = profile_section!(
            "HTML comment ranges",
            profile,
            crate::utils::skip_context::compute_html_comment_ranges(content)
        );

        // Pre-compute autodoc block ranges for MkDocs flavor (avoids O(n²) scaling)
        let autodoc_ranges = profile_section!("Autodoc block ranges", profile, {
            if flavor == MarkdownFlavor::MkDocs {
                crate::utils::mkdocstrings_refs::detect_autodoc_block_ranges(content)
            } else {
                Vec::new()
            }
        });

        // Pre-compute line information (without headings/blockquotes yet)
        let mut lines = profile_section!(
            "Basic line info",
            profile,
            Self::compute_basic_line_info(
                content,
                &line_offsets,
                &code_blocks,
                flavor,
                &html_comment_ranges,
                &autodoc_ranges,
            )
        );

        // Detect HTML blocks BEFORE heading detection
        profile_section!("HTML blocks", profile, Self::detect_html_blocks(content, &mut lines));

        // Detect ESM import/export blocks in MDX files BEFORE heading detection
        profile_section!(
            "ESM blocks",
            profile,
            Self::detect_esm_blocks(content, &mut lines, flavor)
        );

        // Now detect headings and blockquotes
        profile_section!(
            "Headings & blockquotes",
            profile,
            Self::detect_headings_and_blockquotes(content, &mut lines, flavor, &html_comment_ranges)
        );

        // Parse code spans early so we can exclude them from link/image parsing
        let code_spans = profile_section!("Code spans", profile, Self::parse_code_spans(content, &lines));

        // Mark lines that are continuations of multi-line code spans
        // This is needed for parse_list_blocks to correctly handle list items with multi-line code spans
        for span in &code_spans {
            if span.end_line > span.line {
                // Mark lines after the first line as continuations
                for line_num in (span.line + 1)..=span.end_line {
                    if let Some(line_info) = lines.get_mut(line_num - 1) {
                        line_info.in_code_span_continuation = true;
                    }
                }
            }
        }

        // Parse links, images, references, and list blocks
        let (links, broken_links, footnote_refs) = profile_section!(
            "Links",
            profile,
            Self::parse_links(content, &lines, &code_blocks, &code_spans, flavor, &html_comment_ranges)
        );

        let images = profile_section!(
            "Images",
            profile,
            Self::parse_images(content, &lines, &code_blocks, &code_spans, &html_comment_ranges)
        );

        let reference_defs = profile_section!("Reference defs", profile, Self::parse_reference_defs(content, &lines));

        let list_blocks = profile_section!("List blocks", profile, Self::parse_list_blocks(content, &lines));

        // Compute character frequency for fast content analysis
        let char_frequency = profile_section!("Char frequency", profile, Self::compute_char_frequency(content));

        // Pre-compute table blocks for rules that need them (MD013, MD055, MD056, MD058, MD060)
        let table_blocks = profile_section!(
            "Table blocks",
            profile,
            crate::utils::table_utils::TableUtils::find_table_blocks_with_code_info(
                content,
                &code_blocks,
                &code_spans,
                &html_comment_ranges,
            )
        );

        // Pre-compute LineIndex once for all rules (eliminates 46x content cloning)
        let line_index = profile_section!(
            "Line index",
            profile,
            crate::utils::range_utils::LineIndex::new(content)
        );

        // Pre-compute Jinja template ranges once for all rules (eliminates O(n×m) in MD011)
        let jinja_ranges = profile_section!(
            "Jinja ranges",
            profile,
            crate::utils::jinja_utils::find_jinja_ranges(content)
        );

        Self {
            content,
            line_offsets,
            code_blocks,
            lines,
            links,
            images,
            broken_links,
            footnote_refs,
            reference_defs,
            code_spans_cache: Mutex::new(Some(Arc::new(code_spans))),
            list_blocks,
            char_frequency,
            html_tags_cache: Mutex::new(None),
            emphasis_spans_cache: Mutex::new(None),
            table_rows_cache: Mutex::new(None),
            bare_urls_cache: Mutex::new(None),
            html_comment_ranges,
            table_blocks,
            line_index,
            jinja_ranges,
            flavor,
            source_file,
        }
    }

    /// Get code spans - computed lazily on first access
    pub fn code_spans(&self) -> Arc<Vec<CodeSpan>> {
        let mut cache = self.code_spans_cache.lock().expect("Code spans cache mutex poisoned");

        Arc::clone(cache.get_or_insert_with(|| Arc::new(Self::parse_code_spans(self.content, &self.lines))))
    }

    /// Get HTML comment ranges - pre-computed during LintContext construction
    pub fn html_comment_ranges(&self) -> &[crate::utils::skip_context::ByteRange] {
        &self.html_comment_ranges
    }

    /// Get HTML tags - computed lazily on first access
    pub fn html_tags(&self) -> Arc<Vec<HtmlTag>> {
        let mut cache = self.html_tags_cache.lock().expect("HTML tags cache mutex poisoned");

        Arc::clone(cache.get_or_insert_with(|| {
            Arc::new(Self::parse_html_tags(
                self.content,
                &self.lines,
                &self.code_blocks,
                self.flavor,
            ))
        }))
    }

    /// Get emphasis spans - computed lazily on first access
    pub fn emphasis_spans(&self) -> Arc<Vec<EmphasisSpan>> {
        let mut cache = self
            .emphasis_spans_cache
            .lock()
            .expect("Emphasis spans cache mutex poisoned");

        Arc::clone(
            cache.get_or_insert_with(|| {
                Arc::new(Self::parse_emphasis_spans(self.content, &self.lines, &self.code_blocks))
            }),
        )
    }

    /// Get table rows - computed lazily on first access
    pub fn table_rows(&self) -> Arc<Vec<TableRow>> {
        let mut cache = self.table_rows_cache.lock().expect("Table rows cache mutex poisoned");

        Arc::clone(cache.get_or_insert_with(|| Arc::new(Self::parse_table_rows(self.content, &self.lines))))
    }

    /// Get bare URLs - computed lazily on first access
    pub fn bare_urls(&self) -> Arc<Vec<BareUrl>> {
        let mut cache = self.bare_urls_cache.lock().expect("Bare URLs cache mutex poisoned");

        Arc::clone(
            cache.get_or_insert_with(|| Arc::new(Self::parse_bare_urls(self.content, &self.lines, &self.code_blocks))),
        )
    }

    /// Map a byte offset to (line, column)
    pub fn offset_to_line_col(&self, offset: usize) -> (usize, usize) {
        match self.line_offsets.binary_search(&offset) {
            Ok(line) => (line + 1, 1),
            Err(line) => {
                let line_start = self.line_offsets.get(line.wrapping_sub(1)).copied().unwrap_or(0);
                (line, offset - line_start + 1)
            }
        }
    }

    /// Check if a position is within a code block or code span
    pub fn is_in_code_block_or_span(&self, pos: usize) -> bool {
        // Check code blocks first
        if CodeBlockUtils::is_in_code_block_or_span(&self.code_blocks, pos) {
            return true;
        }

        // Check inline code spans (lazy load if needed)
        self.code_spans()
            .iter()
            .any(|span| pos >= span.byte_offset && pos < span.byte_end)
    }

    /// Get line information by line number (1-indexed)
    pub fn line_info(&self, line_num: usize) -> Option<&LineInfo> {
        if line_num > 0 {
            self.lines.get(line_num - 1)
        } else {
            None
        }
    }

    /// Get byte offset for a line number (1-indexed)
    pub fn line_to_byte_offset(&self, line_num: usize) -> Option<usize> {
        self.line_info(line_num).map(|info| info.byte_offset)
    }

    /// Get URL for a reference link/image by its ID
    pub fn get_reference_url(&self, ref_id: &str) -> Option<&str> {
        let normalized_id = ref_id.to_lowercase();
        self.reference_defs
            .iter()
            .find(|def| def.id == normalized_id)
            .map(|def| def.url.as_str())
    }

    /// Check if a line is part of a list block
    pub fn is_in_list_block(&self, line_num: usize) -> bool {
        self.list_blocks
            .iter()
            .any(|block| line_num >= block.start_line && line_num <= block.end_line)
    }

    /// Get the list block containing a specific line
    pub fn list_block_for_line(&self, line_num: usize) -> Option<&ListBlock> {
        self.list_blocks
            .iter()
            .find(|block| line_num >= block.start_line && line_num <= block.end_line)
    }

    // Compatibility methods for DocumentStructure migration

    /// Check if a line is within a code block
    pub fn is_in_code_block(&self, line_num: usize) -> bool {
        if line_num == 0 || line_num > self.lines.len() {
            return false;
        }
        self.lines[line_num - 1].in_code_block
    }

    /// Check if a line is within front matter
    pub fn is_in_front_matter(&self, line_num: usize) -> bool {
        if line_num == 0 || line_num > self.lines.len() {
            return false;
        }
        self.lines[line_num - 1].in_front_matter
    }

    /// Check if a line is within an HTML block
    pub fn is_in_html_block(&self, line_num: usize) -> bool {
        if line_num == 0 || line_num > self.lines.len() {
            return false;
        }
        self.lines[line_num - 1].in_html_block
    }

    /// Check if a line and column is within a code span
    pub fn is_in_code_span(&self, line_num: usize, col: usize) -> bool {
        if line_num == 0 || line_num > self.lines.len() {
            return false;
        }

        // Use the code spans cache to check
        // Note: col is 1-indexed from caller, but span.start_col and span.end_col are 0-indexed
        // Convert col to 0-indexed for comparison
        let col_0indexed = if col > 0 { col - 1 } else { 0 };
        let code_spans = self.code_spans();
        code_spans.iter().any(|span| {
            // Check if line is within the span's line range
            if line_num < span.line || line_num > span.end_line {
                return false;
            }

            if span.line == span.end_line {
                // Single-line span: check column bounds
                col_0indexed >= span.start_col && col_0indexed < span.end_col
            } else if line_num == span.line {
                // First line of multi-line span: anything after start_col is in span
                col_0indexed >= span.start_col
            } else if line_num == span.end_line {
                // Last line of multi-line span: anything before end_col is in span
                col_0indexed < span.end_col
            } else {
                // Middle line of multi-line span: entire line is in span
                true
            }
        })
    }

    /// Check if a byte offset is within a code span
    #[inline]
    pub fn is_byte_offset_in_code_span(&self, byte_offset: usize) -> bool {
        let code_spans = self.code_spans();
        code_spans
            .iter()
            .any(|span| byte_offset >= span.byte_offset && byte_offset < span.byte_end)
    }

    /// Check if a byte position is within a reference definition
    /// This is much faster than scanning the content with regex for each check (O(1) vs O(n))
    #[inline]
    pub fn is_in_reference_def(&self, byte_pos: usize) -> bool {
        self.reference_defs
            .iter()
            .any(|ref_def| byte_pos >= ref_def.byte_offset && byte_pos < ref_def.byte_end)
    }

    /// Check if a byte position is within an HTML comment
    /// This is much faster than scanning the content with regex for each check (O(k) vs O(n))
    /// where k is the number of HTML comments (typically very small)
    #[inline]
    pub fn is_in_html_comment(&self, byte_pos: usize) -> bool {
        self.html_comment_ranges
            .iter()
            .any(|range| byte_pos >= range.start && byte_pos < range.end)
    }

    /// Check if a byte position is within a Jinja template ({{ }} or {% %})
    pub fn is_in_jinja_range(&self, byte_pos: usize) -> bool {
        self.jinja_ranges
            .iter()
            .any(|(start, end)| byte_pos >= *start && byte_pos < *end)
    }

    /// Check if content has any instances of a specific character (fast)
    pub fn has_char(&self, ch: char) -> bool {
        match ch {
            '#' => self.char_frequency.hash_count > 0,
            '*' => self.char_frequency.asterisk_count > 0,
            '_' => self.char_frequency.underscore_count > 0,
            '-' => self.char_frequency.hyphen_count > 0,
            '+' => self.char_frequency.plus_count > 0,
            '>' => self.char_frequency.gt_count > 0,
            '|' => self.char_frequency.pipe_count > 0,
            '[' => self.char_frequency.bracket_count > 0,
            '`' => self.char_frequency.backtick_count > 0,
            '<' => self.char_frequency.lt_count > 0,
            '!' => self.char_frequency.exclamation_count > 0,
            '\n' => self.char_frequency.newline_count > 0,
            _ => self.content.contains(ch), // Fallback for other characters
        }
    }

    /// Get count of a specific character (fast)
    pub fn char_count(&self, ch: char) -> usize {
        match ch {
            '#' => self.char_frequency.hash_count,
            '*' => self.char_frequency.asterisk_count,
            '_' => self.char_frequency.underscore_count,
            '-' => self.char_frequency.hyphen_count,
            '+' => self.char_frequency.plus_count,
            '>' => self.char_frequency.gt_count,
            '|' => self.char_frequency.pipe_count,
            '[' => self.char_frequency.bracket_count,
            '`' => self.char_frequency.backtick_count,
            '<' => self.char_frequency.lt_count,
            '!' => self.char_frequency.exclamation_count,
            '\n' => self.char_frequency.newline_count,
            _ => self.content.matches(ch).count(), // Fallback for other characters
        }
    }

    /// Check if content likely contains headings (fast)
    pub fn likely_has_headings(&self) -> bool {
        self.char_frequency.hash_count > 0 || self.char_frequency.hyphen_count > 2 // Potential setext underlines
    }

    /// Check if content likely contains lists (fast)
    pub fn likely_has_lists(&self) -> bool {
        self.char_frequency.asterisk_count > 0
            || self.char_frequency.hyphen_count > 0
            || self.char_frequency.plus_count > 0
    }

    /// Check if content likely contains emphasis (fast)
    pub fn likely_has_emphasis(&self) -> bool {
        self.char_frequency.asterisk_count > 1 || self.char_frequency.underscore_count > 1
    }

    /// Check if content likely contains tables (fast)
    pub fn likely_has_tables(&self) -> bool {
        self.char_frequency.pipe_count > 2
    }

    /// Check if content likely contains blockquotes (fast)
    pub fn likely_has_blockquotes(&self) -> bool {
        self.char_frequency.gt_count > 0
    }

    /// Check if content likely contains code (fast)
    pub fn likely_has_code(&self) -> bool {
        self.char_frequency.backtick_count > 0
    }

    /// Check if content likely contains links or images (fast)
    pub fn likely_has_links_or_images(&self) -> bool {
        self.char_frequency.bracket_count > 0 || self.char_frequency.exclamation_count > 0
    }

    /// Check if content likely contains HTML (fast)
    pub fn likely_has_html(&self) -> bool {
        self.char_frequency.lt_count > 0
    }

    /// Get HTML tags on a specific line
    pub fn html_tags_on_line(&self, line_num: usize) -> Vec<HtmlTag> {
        self.html_tags()
            .iter()
            .filter(|tag| tag.line == line_num)
            .cloned()
            .collect()
    }

    /// Get emphasis spans on a specific line
    pub fn emphasis_spans_on_line(&self, line_num: usize) -> Vec<EmphasisSpan> {
        self.emphasis_spans()
            .iter()
            .filter(|span| span.line == line_num)
            .cloned()
            .collect()
    }

    /// Get table rows on a specific line
    pub fn table_rows_on_line(&self, line_num: usize) -> Vec<TableRow> {
        self.table_rows()
            .iter()
            .filter(|row| row.line == line_num)
            .cloned()
            .collect()
    }

    /// Get bare URLs on a specific line
    pub fn bare_urls_on_line(&self, line_num: usize) -> Vec<BareUrl> {
        self.bare_urls()
            .iter()
            .filter(|url| url.line == line_num)
            .cloned()
            .collect()
    }

    /// Find the line index for a given byte offset using binary search.
    /// Returns (line_index, line_number, column) where:
    /// - line_index is the 0-based index in the lines array
    /// - line_number is the 1-based line number
    /// - column is the byte offset within that line
    #[inline]
    fn find_line_for_offset(lines: &[LineInfo], byte_offset: usize) -> (usize, usize, usize) {
        // Binary search to find the line containing this byte offset
        let idx = match lines.binary_search_by(|line| {
            if byte_offset < line.byte_offset {
                std::cmp::Ordering::Greater
            } else if byte_offset > line.byte_offset + line.byte_len {
                std::cmp::Ordering::Less
            } else {
                std::cmp::Ordering::Equal
            }
        }) {
            Ok(idx) => idx,
            Err(idx) => idx.saturating_sub(1),
        };

        let line = &lines[idx];
        let line_num = idx + 1;
        let col = byte_offset.saturating_sub(line.byte_offset);

        (idx, line_num, col)
    }

    /// Check if a byte offset is within a code span using binary search
    #[inline]
    fn is_offset_in_code_span(code_spans: &[CodeSpan], offset: usize) -> bool {
        // Since spans are sorted by byte_offset, use partition_point for binary search
        let idx = code_spans.partition_point(|span| span.byte_offset <= offset);

        // Check the span that starts at or before our offset
        if idx > 0 {
            let span = &code_spans[idx - 1];
            if offset >= span.byte_offset && offset < span.byte_end {
                return true;
            }
        }

        false
    }

    /// Parse all links in the content
    fn parse_links(
        content: &'a str,
        lines: &[LineInfo],
        code_blocks: &[(usize, usize)],
        code_spans: &[CodeSpan],
        flavor: MarkdownFlavor,
        html_comment_ranges: &[crate::utils::skip_context::ByteRange],
    ) -> (Vec<ParsedLink<'a>>, Vec<BrokenLinkInfo>, Vec<FootnoteRef>) {
        use crate::utils::skip_context::{is_in_html_comment_ranges, is_mkdocs_snippet_line};
        use std::collections::HashSet;

        let mut links = Vec::with_capacity(content.len() / 500);
        let mut broken_links = Vec::new();
        let mut footnote_refs = Vec::new();

        // Track byte positions of links found by pulldown-cmark
        let mut found_positions = HashSet::new();

        // Use pulldown-cmark's streaming parser with BrokenLink callback
        // The callback captures undefined references: [text][undefined], [shortcut], [text][]
        // This automatically handles:
        // - Escaped links (won't generate events)
        // - Links in code blocks/spans (won't generate Link events)
        // - Images (generates Tag::Image instead)
        // - Reference resolution (dest_url is already resolved!)
        // - Broken references (callback is invoked)
        // - Wiki-links (enabled via ENABLE_WIKILINKS)
        let mut options = Options::empty();
        options.insert(Options::ENABLE_WIKILINKS);
        options.insert(Options::ENABLE_FOOTNOTES);

        let parser = Parser::new_with_broken_link_callback(
            content,
            options,
            Some(|link: BrokenLink<'_>| {
                broken_links.push(BrokenLinkInfo {
                    reference: link.reference.to_string(),
                    span: link.span.clone(),
                });
                None
            }),
        )
        .into_offset_iter();

        let mut link_stack: Vec<(
            usize,
            usize,
            pulldown_cmark::CowStr<'a>,
            LinkType,
            pulldown_cmark::CowStr<'a>,
        )> = Vec::new();
        let mut text_chunks: Vec<(String, usize, usize)> = Vec::new(); // (text, start, end)

        for (event, range) in parser {
            match event {
                Event::Start(Tag::Link {
                    link_type,
                    dest_url,
                    id,
                    ..
                }) => {
                    // Link start - record position, URL, and reference ID
                    link_stack.push((range.start, range.end, dest_url, link_type, id));
                    text_chunks.clear();
                }
                Event::Text(text) if !link_stack.is_empty() => {
                    // Track text content with its byte range
                    text_chunks.push((text.to_string(), range.start, range.end));
                }
                Event::Code(code) if !link_stack.is_empty() => {
                    // Include inline code in link text (with backticks)
                    let code_text = format!("`{code}`");
                    text_chunks.push((code_text, range.start, range.end));
                }
                Event::End(TagEnd::Link) => {
                    if let Some((start_pos, _link_start_end, url, link_type, ref_id)) = link_stack.pop() {
                        // Skip if in HTML comment
                        if is_in_html_comment_ranges(html_comment_ranges, start_pos) {
                            text_chunks.clear();
                            continue;
                        }

                        // Find line and column information
                        let (line_idx, line_num, col_start) = Self::find_line_for_offset(lines, start_pos);

                        // Skip if this link is on a MkDocs snippet line
                        if is_mkdocs_snippet_line(lines[line_idx].content(content), flavor) {
                            text_chunks.clear();
                            continue;
                        }

                        let (_, _end_line_num, col_end) = Self::find_line_for_offset(lines, range.end);

                        let is_reference = matches!(
                            link_type,
                            LinkType::Reference | LinkType::Collapsed | LinkType::Shortcut
                        );

                        // Extract link text directly from source bytes to preserve escaping
                        // Text events from pulldown-cmark unescape \] → ], which breaks MD039
                        let link_text = if start_pos < content.len() {
                            let link_bytes = &content.as_bytes()[start_pos..range.end.min(content.len())];

                            // Find MATCHING ] by tracking bracket depth for nested brackets
                            // An unescaped bracket is one NOT preceded by an odd number of backslashes
                            // Brackets inside code spans (between backticks) should be ignored
                            let mut close_pos = None;
                            let mut depth = 0;
                            let mut in_code_span = false;

                            for (i, &byte) in link_bytes.iter().enumerate().skip(1) {
                                // Count preceding backslashes
                                let mut backslash_count = 0;
                                let mut j = i;
                                while j > 0 && link_bytes[j - 1] == b'\\' {
                                    backslash_count += 1;
                                    j -= 1;
                                }
                                let is_escaped = backslash_count % 2 != 0;

                                // Track code spans - backticks toggle in/out of code
                                if byte == b'`' && !is_escaped {
                                    in_code_span = !in_code_span;
                                }

                                // Only count brackets when NOT in a code span
                                if !is_escaped && !in_code_span {
                                    if byte == b'[' {
                                        depth += 1;
                                    } else if byte == b']' {
                                        if depth == 0 {
                                            // Found the matching closing bracket
                                            close_pos = Some(i);
                                            break;
                                        } else {
                                            depth -= 1;
                                        }
                                    }
                                }
                            }

                            if let Some(pos) = close_pos {
                                Cow::Borrowed(std::str::from_utf8(&link_bytes[1..pos]).unwrap_or(""))
                            } else {
                                Cow::Borrowed("")
                            }
                        } else {
                            Cow::Borrowed("")
                        };

                        // For reference links, use the actual reference ID from pulldown-cmark
                        let reference_id = if is_reference && !ref_id.is_empty() {
                            Some(Cow::Owned(ref_id.to_lowercase()))
                        } else if is_reference {
                            // For collapsed/shortcut references without explicit ID, use the link text
                            Some(Cow::Owned(link_text.to_lowercase()))
                        } else {
                            None
                        };

                        // WORKAROUND: pulldown-cmark bug with escaped brackets
                        // Check for escaped image syntax: \![text](url)
                        // The byte_offset points to the '[', so we check 2 bytes back for '\!'
                        let has_escaped_bang = start_pos >= 2
                            && content.as_bytes().get(start_pos - 2) == Some(&b'\\')
                            && content.as_bytes().get(start_pos - 1) == Some(&b'!');

                        // Check for escaped bracket: \[text](url)
                        // The byte_offset points to the '[', so we check 1 byte back for '\'
                        let has_escaped_bracket =
                            start_pos >= 1 && content.as_bytes().get(start_pos - 1) == Some(&b'\\');

                        if has_escaped_bang || has_escaped_bracket {
                            text_chunks.clear();
                            continue; // Skip: this is escaped markdown, not a real link
                        }

                        // Track this position as found
                        found_positions.insert(start_pos);

                        links.push(ParsedLink {
                            line: line_num,
                            start_col: col_start,
                            end_col: col_end,
                            byte_offset: start_pos,
                            byte_end: range.end,
                            text: link_text,
                            url: Cow::Owned(url.to_string()),
                            is_reference,
                            reference_id,
                            link_type,
                        });

                        text_chunks.clear();
                    }
                }
                Event::FootnoteReference(footnote_id) => {
                    // Capture footnote references like [^1], [^note]
                    // Skip if in HTML comment
                    if is_in_html_comment_ranges(html_comment_ranges, range.start) {
                        continue;
                    }

                    let (_, line_num, _) = Self::find_line_for_offset(lines, range.start);
                    footnote_refs.push(FootnoteRef {
                        id: footnote_id.to_string(),
                        line: line_num,
                        byte_offset: range.start,
                        byte_end: range.end,
                    });
                }
                _ => {}
            }
        }

        // Also find undefined references using regex
        // These are patterns like [text][ref] that pulldown-cmark didn't parse as links
        // because the reference is undefined
        for cap in LINK_PATTERN.captures_iter(content) {
            let full_match = cap.get(0).unwrap();
            let match_start = full_match.start();
            let match_end = full_match.end();

            // Skip if this was already found by pulldown-cmark (it's a valid link)
            if found_positions.contains(&match_start) {
                continue;
            }

            // Skip if escaped
            if match_start > 0 && content.as_bytes().get(match_start - 1) == Some(&b'\\') {
                continue;
            }

            // Skip if it's an image
            if match_start > 0 && content.as_bytes().get(match_start - 1) == Some(&b'!') {
                continue;
            }

            // Skip if in code block
            if CodeBlockUtils::is_in_code_block(code_blocks, match_start) {
                continue;
            }

            // Skip if in code span
            if Self::is_offset_in_code_span(code_spans, match_start) {
                continue;
            }

            // Skip if in HTML comment
            if is_in_html_comment_ranges(html_comment_ranges, match_start) {
                continue;
            }

            // Find line and column information
            let (line_idx, line_num, col_start) = Self::find_line_for_offset(lines, match_start);

            // Skip if this link is on a MkDocs snippet line
            if is_mkdocs_snippet_line(lines[line_idx].content(content), flavor) {
                continue;
            }

            let (_, _end_line_num, col_end) = Self::find_line_for_offset(lines, match_end);

            let text = cap.get(1).map_or("", |m| m.as_str());

            // Only process reference links (group 6)
            if let Some(ref_id) = cap.get(6) {
                let ref_id_str = ref_id.as_str();
                let normalized_ref = if ref_id_str.is_empty() {
                    Cow::Owned(text.to_lowercase()) // Implicit reference
                } else {
                    Cow::Owned(ref_id_str.to_lowercase())
                };

                // This is an undefined reference (pulldown-cmark didn't parse it)
                links.push(ParsedLink {
                    line: line_num,
                    start_col: col_start,
                    end_col: col_end,
                    byte_offset: match_start,
                    byte_end: match_end,
                    text: Cow::Borrowed(text),
                    url: Cow::Borrowed(""), // Empty URL indicates undefined reference
                    is_reference: true,
                    reference_id: Some(normalized_ref),
                    link_type: LinkType::Reference, // Undefined references are reference-style
                });
            }
        }

        (links, broken_links, footnote_refs)
    }

    /// Parse all images in the content
    fn parse_images(
        content: &'a str,
        lines: &[LineInfo],
        code_blocks: &[(usize, usize)],
        code_spans: &[CodeSpan],
        html_comment_ranges: &[crate::utils::skip_context::ByteRange],
    ) -> Vec<ParsedImage<'a>> {
        use crate::utils::skip_context::is_in_html_comment_ranges;
        use std::collections::HashSet;

        // Pre-size based on a heuristic: images are less common than links
        let mut images = Vec::with_capacity(content.len() / 1000);
        let mut found_positions = HashSet::new();

        // Use pulldown-cmark for parsing - more accurate and faster
        let parser = Parser::new(content).into_offset_iter();
        let mut image_stack: Vec<(usize, pulldown_cmark::CowStr<'a>, LinkType, pulldown_cmark::CowStr<'a>)> =
            Vec::new();
        let mut text_chunks: Vec<(String, usize, usize)> = Vec::new(); // (text, start, end)

        for (event, range) in parser {
            match event {
                Event::Start(Tag::Image {
                    link_type,
                    dest_url,
                    id,
                    ..
                }) => {
                    image_stack.push((range.start, dest_url, link_type, id));
                    text_chunks.clear();
                }
                Event::Text(text) if !image_stack.is_empty() => {
                    text_chunks.push((text.to_string(), range.start, range.end));
                }
                Event::Code(code) if !image_stack.is_empty() => {
                    let code_text = format!("`{code}`");
                    text_chunks.push((code_text, range.start, range.end));
                }
                Event::End(TagEnd::Image) => {
                    if let Some((start_pos, url, link_type, ref_id)) = image_stack.pop() {
                        // Skip if in code block
                        if CodeBlockUtils::is_in_code_block(code_blocks, start_pos) {
                            continue;
                        }

                        // Skip if in code span
                        if Self::is_offset_in_code_span(code_spans, start_pos) {
                            continue;
                        }

                        // Skip if in HTML comment
                        if is_in_html_comment_ranges(html_comment_ranges, start_pos) {
                            continue;
                        }

                        // Find line and column using binary search
                        let (_, line_num, col_start) = Self::find_line_for_offset(lines, start_pos);
                        let (_, _end_line_num, col_end) = Self::find_line_for_offset(lines, range.end);

                        let is_reference = matches!(
                            link_type,
                            LinkType::Reference | LinkType::Collapsed | LinkType::Shortcut
                        );

                        // Extract alt text directly from source bytes to preserve escaping
                        // Text events from pulldown-cmark unescape \] → ], which breaks rules that need escaping
                        let alt_text = if start_pos < content.len() {
                            let image_bytes = &content.as_bytes()[start_pos..range.end.min(content.len())];

                            // Find MATCHING ] by tracking bracket depth for nested brackets
                            // An unescaped bracket is one NOT preceded by an odd number of backslashes
                            let mut close_pos = None;
                            let mut depth = 0;

                            if image_bytes.len() > 2 {
                                for (i, &byte) in image_bytes.iter().enumerate().skip(2) {
                                    // Count preceding backslashes
                                    let mut backslash_count = 0;
                                    let mut j = i;
                                    while j > 0 && image_bytes[j - 1] == b'\\' {
                                        backslash_count += 1;
                                        j -= 1;
                                    }
                                    let is_escaped = backslash_count % 2 != 0;

                                    if !is_escaped {
                                        if byte == b'[' {
                                            depth += 1;
                                        } else if byte == b']' {
                                            if depth == 0 {
                                                // Found the matching closing bracket
                                                close_pos = Some(i);
                                                break;
                                            } else {
                                                depth -= 1;
                                            }
                                        }
                                    }
                                }
                            }

                            if let Some(pos) = close_pos {
                                Cow::Borrowed(std::str::from_utf8(&image_bytes[2..pos]).unwrap_or(""))
                            } else {
                                Cow::Borrowed("")
                            }
                        } else {
                            Cow::Borrowed("")
                        };

                        let reference_id = if is_reference && !ref_id.is_empty() {
                            Some(Cow::Owned(ref_id.to_lowercase()))
                        } else if is_reference {
                            Some(Cow::Owned(alt_text.to_lowercase())) // Collapsed/shortcut references
                        } else {
                            None
                        };

                        found_positions.insert(start_pos);
                        images.push(ParsedImage {
                            line: line_num,
                            start_col: col_start,
                            end_col: col_end,
                            byte_offset: start_pos,
                            byte_end: range.end,
                            alt_text,
                            url: Cow::Owned(url.to_string()),
                            is_reference,
                            reference_id,
                            link_type,
                        });
                    }
                }
                _ => {}
            }
        }

        // Regex fallback for undefined references that pulldown-cmark treats as plain text
        for cap in IMAGE_PATTERN.captures_iter(content) {
            let full_match = cap.get(0).unwrap();
            let match_start = full_match.start();
            let match_end = full_match.end();

            // Skip if already found by pulldown-cmark
            if found_positions.contains(&match_start) {
                continue;
            }

            // Skip if the ! is escaped
            if match_start > 0 && content.as_bytes().get(match_start - 1) == Some(&b'\\') {
                continue;
            }

            // Skip if in code block, code span, or HTML comment
            if CodeBlockUtils::is_in_code_block(code_blocks, match_start)
                || Self::is_offset_in_code_span(code_spans, match_start)
                || is_in_html_comment_ranges(html_comment_ranges, match_start)
            {
                continue;
            }

            // Only process reference images (undefined references not found by pulldown-cmark)
            if let Some(ref_id) = cap.get(6) {
                let (_, line_num, col_start) = Self::find_line_for_offset(lines, match_start);
                let (_, _end_line_num, col_end) = Self::find_line_for_offset(lines, match_end);
                let alt_text = cap.get(1).map_or("", |m| m.as_str());
                let ref_id_str = ref_id.as_str();
                let normalized_ref = if ref_id_str.is_empty() {
                    Cow::Owned(alt_text.to_lowercase())
                } else {
                    Cow::Owned(ref_id_str.to_lowercase())
                };

                images.push(ParsedImage {
                    line: line_num,
                    start_col: col_start,
                    end_col: col_end,
                    byte_offset: match_start,
                    byte_end: match_end,
                    alt_text: Cow::Borrowed(alt_text),
                    url: Cow::Borrowed(""),
                    is_reference: true,
                    reference_id: Some(normalized_ref),
                    link_type: LinkType::Reference, // Undefined references are reference-style
                });
            }
        }

        images
    }

    /// Parse reference definitions
    fn parse_reference_defs(content: &str, lines: &[LineInfo]) -> Vec<ReferenceDef> {
        // Pre-size based on lines count as reference definitions are line-based
        let mut refs = Vec::with_capacity(lines.len() / 20); // ~1 ref per 20 lines

        for (line_idx, line_info) in lines.iter().enumerate() {
            // Skip lines in code blocks
            if line_info.in_code_block {
                continue;
            }

            let line = line_info.content(content);
            let line_num = line_idx + 1;

            if let Some(cap) = REF_DEF_PATTERN.captures(line) {
                let id = cap.get(1).unwrap().as_str().to_lowercase();
                let url = cap.get(2).unwrap().as_str().to_string();
                let title = cap.get(3).or_else(|| cap.get(4)).map(|m| m.as_str().to_string());

                // Calculate byte positions
                // The match starts at the beginning of the line (0) and extends to the end
                let match_obj = cap.get(0).unwrap();
                let byte_offset = line_info.byte_offset + match_obj.start();
                let byte_end = line_info.byte_offset + match_obj.end();

                refs.push(ReferenceDef {
                    line: line_num,
                    id,
                    url,
                    title,
                    byte_offset,
                    byte_end,
                });
            }
        }

        refs
    }

    /// Fast blockquote prefix parser - replaces regex for 5-10x speedup
    /// Matches: ^(\s*>\s*)(.*)
    /// Returns: Some((prefix_with_ws, content_after_prefix)) or None
    #[inline]
    fn parse_blockquote_prefix(line: &str) -> Option<(&str, &str)> {
        let trimmed_start = line.trim_start();
        if !trimmed_start.starts_with('>') {
            return None;
        }

        let leading_ws_len = line.len() - trimmed_start.len();
        let after_gt = &trimmed_start[1..];
        let content = after_gt.trim_start();
        let ws_after_gt_len = after_gt.len() - content.len();
        let prefix_len = leading_ws_len + 1 + ws_after_gt_len;

        Some((&line[..prefix_len], content))
    }

    /// Fast unordered list parser - replaces regex for 5-10x speedup
    /// Matches: ^(\s*)([-*+])([ \t]*)(.*)
    /// Returns: Some((leading_ws, marker, spacing, content)) or None
    #[inline]
    fn parse_unordered_list(line: &str) -> Option<(&str, char, &str, &str)> {
        let bytes = line.as_bytes();
        let mut i = 0;

        // Skip leading whitespace
        while i < bytes.len() && (bytes[i] == b' ' || bytes[i] == b'\t') {
            i += 1;
        }

        // Check for marker
        if i >= bytes.len() {
            return None;
        }
        let marker = bytes[i] as char;
        if marker != '-' && marker != '*' && marker != '+' {
            return None;
        }
        let marker_pos = i;
        i += 1;

        // Collect spacing after marker (space or tab only)
        let spacing_start = i;
        while i < bytes.len() && (bytes[i] == b' ' || bytes[i] == b'\t') {
            i += 1;
        }

        Some((&line[..marker_pos], marker, &line[spacing_start..i], &line[i..]))
    }

    /// Fast ordered list parser - replaces regex for 5-10x speedup
    /// Matches: ^(\s*)(\d+)([.)])([ \t]*)(.*)
    /// Returns: Some((leading_ws, number_str, delimiter, spacing, content)) or None
    #[inline]
    fn parse_ordered_list(line: &str) -> Option<(&str, &str, char, &str, &str)> {
        let bytes = line.as_bytes();
        let mut i = 0;

        // Skip leading whitespace
        while i < bytes.len() && (bytes[i] == b' ' || bytes[i] == b'\t') {
            i += 1;
        }

        // Collect digits
        let number_start = i;
        while i < bytes.len() && bytes[i].is_ascii_digit() {
            i += 1;
        }
        if i == number_start {
            return None; // No digits found
        }

        // Check for delimiter
        if i >= bytes.len() {
            return None;
        }
        let delimiter = bytes[i] as char;
        if delimiter != '.' && delimiter != ')' {
            return None;
        }
        let delimiter_pos = i;
        i += 1;

        // Collect spacing after delimiter (space or tab only)
        let spacing_start = i;
        while i < bytes.len() && (bytes[i] == b' ' || bytes[i] == b'\t') {
            i += 1;
        }

        Some((
            &line[..number_start],
            &line[number_start..delimiter_pos],
            delimiter,
            &line[spacing_start..i],
            &line[i..],
        ))
    }

    /// Pre-compute which lines are in code blocks - O(m*n) where m=code_blocks, n=lines
    /// Returns a Vec<bool> where index i indicates if line i is in a code block
    fn compute_code_block_line_map(content: &str, line_offsets: &[usize], code_blocks: &[(usize, usize)]) -> Vec<bool> {
        let num_lines = line_offsets.len();
        let mut in_code_block = vec![false; num_lines];

        // For each code block, mark all lines within it
        for &(start, end) in code_blocks {
            // Ensure we're at valid UTF-8 boundaries
            let safe_start = if start > 0 && !content.is_char_boundary(start) {
                let mut boundary = start;
                while boundary > 0 && !content.is_char_boundary(boundary) {
                    boundary -= 1;
                }
                boundary
            } else {
                start
            };

            let safe_end = if end < content.len() && !content.is_char_boundary(end) {
                let mut boundary = end;
                while boundary < content.len() && !content.is_char_boundary(boundary) {
                    boundary += 1;
                }
                boundary
            } else {
                end.min(content.len())
            };

            // Trust the code blocks detected by CodeBlockUtils::detect_code_blocks()
            // That function now has proper list context awareness (see code_block_utils.rs)
            // and correctly distinguishes between:
            // - Fenced code blocks (``` or ~~~)
            // - Indented code blocks at document level (4 spaces + blank line before)
            // - List continuation paragraphs (NOT code blocks, even with 4 spaces)
            //
            // We no longer need to re-validate here. The original validation logic
            // was causing false positives by marking list continuation paragraphs as
            // code blocks when they have 4 spaces of indentation.

            // Use binary search to find the first and last line indices
            // line_offsets is sorted, so we can use partition_point for O(log n) lookup
            // Use safe_start/safe_end (UTF-8 boundaries) for consistent line mapping
            //
            // Find the line that CONTAINS safe_start: the line with the largest
            // start offset that is <= safe_start. partition_point gives us the
            // first line that starts AFTER safe_start, so we subtract 1.
            let first_line_after = line_offsets.partition_point(|&offset| offset <= safe_start);
            let first_line = first_line_after.saturating_sub(1);
            let last_line = line_offsets.partition_point(|&offset| offset < safe_end);

            // Mark all lines in the range at once
            for flag in in_code_block.iter_mut().take(last_line).skip(first_line) {
                *flag = true;
            }
        }

        in_code_block
    }

    /// Pre-compute basic line information (without headings/blockquotes)
    fn compute_basic_line_info(
        content: &str,
        line_offsets: &[usize],
        code_blocks: &[(usize, usize)],
        flavor: MarkdownFlavor,
        html_comment_ranges: &[crate::utils::skip_context::ByteRange],
        autodoc_ranges: &[crate::utils::skip_context::ByteRange],
    ) -> Vec<LineInfo> {
        let content_lines: Vec<&str> = content.lines().collect();
        let mut lines = Vec::with_capacity(content_lines.len());

        // Pre-compute which lines are in code blocks
        let code_block_map = Self::compute_code_block_line_map(content, line_offsets, code_blocks);

        // Detect front matter boundaries FIRST, before any other parsing
        // Use FrontMatterUtils to detect all types of front matter (YAML, TOML, JSON, malformed)
        let front_matter_end = FrontMatterUtils::get_front_matter_end_line(content);

        for (i, line) in content_lines.iter().enumerate() {
            let byte_offset = line_offsets.get(i).copied().unwrap_or(0);
            let indent = line.len() - line.trim_start().len();

            // Parse blockquote prefix once and reuse it (avoid redundant parsing)
            let blockquote_parse = Self::parse_blockquote_prefix(line);

            // For blank detection, consider blockquote context
            let is_blank = if let Some((_, content)) = blockquote_parse {
                // In blockquote context, check if content after prefix is blank
                content.trim().is_empty()
            } else {
                line.trim().is_empty()
            };

            // Use pre-computed map for O(1) lookup instead of O(m) iteration
            let in_code_block = code_block_map.get(i).copied().unwrap_or(false);

            // Detect list items (skip if in frontmatter, in mkdocstrings block, or in HTML comment)
            let in_mkdocstrings = flavor == MarkdownFlavor::MkDocs
                && crate::utils::mkdocstrings_refs::is_within_autodoc_block_ranges(autodoc_ranges, byte_offset);
            // Use pre-computed ranges for efficiency (O(log n) vs O(file_size))
            let in_html_comment =
                crate::utils::skip_context::is_in_html_comment_ranges(html_comment_ranges, byte_offset);
            let list_item = if !(in_code_block
                || is_blank
                || in_mkdocstrings
                || in_html_comment
                || (front_matter_end > 0 && i < front_matter_end))
            {
                // Strip blockquote prefix if present for list detection (reuse cached result)
                let (line_for_list_check, blockquote_prefix_len) = if let Some((prefix, content)) = blockquote_parse {
                    (content, prefix.len())
                } else {
                    (&**line, 0)
                };

                if let Some((leading_spaces, marker, spacing, _content)) =
                    Self::parse_unordered_list(line_for_list_check)
                {
                    let marker_column = blockquote_prefix_len + leading_spaces.len();
                    let content_column = marker_column + 1 + spacing.len();

                    // According to CommonMark spec, unordered list items MUST have at least one space
                    // after the marker (-, *, or +). Without a space, it's not a list item.
                    // This also naturally handles cases like:
                    // - *emphasis* (not a list)
                    // - **bold** (not a list)
                    // - --- (horizontal rule, not a list)
                    if spacing.is_empty() {
                        None
                    } else {
                        Some(ListItemInfo {
                            marker: marker.to_string(),
                            is_ordered: false,
                            number: None,
                            marker_column,
                            content_column,
                        })
                    }
                } else if let Some((leading_spaces, number_str, delimiter, spacing, _content)) =
                    Self::parse_ordered_list(line_for_list_check)
                {
                    let marker = format!("{number_str}{delimiter}");
                    let marker_column = blockquote_prefix_len + leading_spaces.len();
                    let content_column = marker_column + marker.len() + spacing.len();

                    // According to CommonMark spec, ordered list items MUST have at least one space
                    // after the marker (period or parenthesis). Without a space, it's not a list item.
                    if spacing.is_empty() {
                        None
                    } else {
                        Some(ListItemInfo {
                            marker,
                            is_ordered: true,
                            number: number_str.parse().ok(),
                            marker_column,
                            content_column,
                        })
                    }
                } else {
                    None
                }
            } else {
                None
            };

            lines.push(LineInfo {
                byte_offset,
                byte_len: line.len(),
                indent,
                is_blank,
                in_code_block,
                in_front_matter: front_matter_end > 0 && i < front_matter_end,
                in_html_block: false, // Will be populated after line creation
                in_html_comment,
                list_item,
                heading: None,    // Will be populated in second pass for Setext headings
                blockquote: None, // Will be populated after line creation
                in_mkdocstrings,
                in_esm_block: false, // Will be populated after line creation for MDX files
                in_code_span_continuation: false, // Will be populated after code spans are parsed
            });
        }

        lines
    }

    /// Detect headings and blockquotes (called after HTML block detection)
    fn detect_headings_and_blockquotes(
        content: &str,
        lines: &mut [LineInfo],
        flavor: MarkdownFlavor,
        html_comment_ranges: &[crate::utils::skip_context::ByteRange],
    ) {
        // Regex for heading detection
        static ATX_HEADING_REGEX: LazyLock<regex::Regex> =
            LazyLock::new(|| regex::Regex::new(r"^(\s*)(#{1,6})(\s*)(.*)$").unwrap());
        static SETEXT_UNDERLINE_REGEX: LazyLock<regex::Regex> =
            LazyLock::new(|| regex::Regex::new(r"^(\s*)(=+|-+)\s*$").unwrap());

        let content_lines: Vec<&str> = content.lines().collect();

        // Detect front matter boundaries to skip those lines
        let front_matter_end = FrontMatterUtils::get_front_matter_end_line(content);

        // Detect headings (including Setext which needs look-ahead) and blockquotes
        for i in 0..lines.len() {
            if lines[i].in_code_block {
                continue;
            }

            // Skip lines in front matter
            if front_matter_end > 0 && i < front_matter_end {
                continue;
            }

            // Skip lines in HTML blocks - HTML content should not be parsed as markdown
            if lines[i].in_html_block {
                continue;
            }

            let line = content_lines[i];

            // Check for blockquotes (even on blank lines within blockquotes)
            if let Some(bq) = parse_blockquote_detailed(line) {
                let nesting_level = bq.markers.len(); // Each '>' is one level
                let marker_column = bq.indent.len();

                // Build the prefix (indentation + markers + space)
                let prefix = format!("{}{}{}", bq.indent, bq.markers, bq.spaces_after);

                // Check for various blockquote issues
                let has_no_space = bq.spaces_after.is_empty() && !bq.content.is_empty();
                // Only flag multiple literal spaces, not tabs
                // Tabs are handled by MD010 (no-hard-tabs), matching markdownlint behavior
                let has_multiple_spaces = bq.spaces_after.chars().filter(|&c| c == ' ').count() > 1;

                // Check if needs MD028 fix (empty blockquote line without proper spacing)
                // MD028 flags empty blockquote lines that don't have a single space after the marker
                // Lines like "> " or ">> " are already correct and don't need fixing
                let needs_md028_fix = bq.content.is_empty() && bq.spaces_after.is_empty();

                lines[i].blockquote = Some(BlockquoteInfo {
                    nesting_level,
                    indent: bq.indent.to_string(),
                    marker_column,
                    prefix,
                    content: bq.content.to_string(),
                    has_no_space_after_marker: has_no_space,
                    has_multiple_spaces_after_marker: has_multiple_spaces,
                    needs_md028_fix,
                });
            }

            // Skip heading detection for blank lines
            if lines[i].is_blank {
                continue;
            }

            // Check for ATX headings (but skip MkDocs snippet lines)
            // In MkDocs flavor, lines like "# -8<- [start:name]" are snippet markers, not headings
            let is_snippet_line = if flavor == MarkdownFlavor::MkDocs {
                crate::utils::mkdocs_snippets::is_snippet_section_start(line)
                    || crate::utils::mkdocs_snippets::is_snippet_section_end(line)
            } else {
                false
            };

            if !is_snippet_line && let Some(caps) = ATX_HEADING_REGEX.captures(line) {
                // Skip headings inside HTML comments (using pre-computed ranges for efficiency)
                if crate::utils::skip_context::is_in_html_comment_ranges(html_comment_ranges, lines[i].byte_offset) {
                    continue;
                }
                let leading_spaces = caps.get(1).map_or("", |m| m.as_str());
                let hashes = caps.get(2).map_or("", |m| m.as_str());
                let spaces_after = caps.get(3).map_or("", |m| m.as_str());
                let rest = caps.get(4).map_or("", |m| m.as_str());

                let level = hashes.len() as u8;
                let marker_column = leading_spaces.len();

                // Check for closing sequence, but handle custom IDs that might come after
                let (text, has_closing, closing_seq) = {
                    // First check if there's a custom ID at the end
                    let (rest_without_id, custom_id_part) = if let Some(id_start) = rest.rfind(" {#") {
                        // Check if this looks like a valid custom ID (ends with })
                        if rest[id_start..].trim_end().ends_with('}') {
                            // Split off the custom ID
                            (&rest[..id_start], &rest[id_start..])
                        } else {
                            (rest, "")
                        }
                    } else {
                        (rest, "")
                    };

                    // Now look for closing hashes in the part before the custom ID
                    let trimmed_rest = rest_without_id.trim_end();
                    if let Some(last_hash_pos) = trimmed_rest.rfind('#') {
                        // Look for the start of the hash sequence
                        let mut start_of_hashes = last_hash_pos;
                        while start_of_hashes > 0 && trimmed_rest.chars().nth(start_of_hashes - 1) == Some('#') {
                            start_of_hashes -= 1;
                        }

                        // Check if there's at least one space before the closing hashes
                        let has_space_before = start_of_hashes == 0
                            || trimmed_rest
                                .chars()
                                .nth(start_of_hashes - 1)
                                .is_some_and(|c| c.is_whitespace());

                        // Check if this is a valid closing sequence (all hashes to end of trimmed part)
                        let potential_closing = &trimmed_rest[start_of_hashes..];
                        let is_all_hashes = potential_closing.chars().all(|c| c == '#');

                        if is_all_hashes && has_space_before {
                            // This is a closing sequence
                            let closing_hashes = potential_closing.to_string();
                            // The text is everything before the closing hashes
                            // Don't include the custom ID here - it will be extracted later
                            let text_part = if !custom_id_part.is_empty() {
                                // If we have a custom ID, append it back to get the full rest
                                // This allows the extract_header_id function to handle it properly
                                format!("{}{}", rest_without_id[..start_of_hashes].trim_end(), custom_id_part)
                            } else {
                                rest_without_id[..start_of_hashes].trim_end().to_string()
                            };
                            (text_part, true, closing_hashes)
                        } else {
                            // Not a valid closing sequence, return the full content
                            (rest.to_string(), false, String::new())
                        }
                    } else {
                        // No hashes found, return the full content
                        (rest.to_string(), false, String::new())
                    }
                };

                let content_column = marker_column + hashes.len() + spaces_after.len();

                // Extract custom header ID if present
                let raw_text = text.trim().to_string();
                let (clean_text, mut custom_id) = crate::utils::header_id_utils::extract_header_id(&raw_text);

                // If no custom ID was found on the header line, check the next line for standalone attr-list
                if custom_id.is_none() && i + 1 < content_lines.len() && i + 1 < lines.len() {
                    let next_line = content_lines[i + 1];
                    if !lines[i + 1].in_code_block
                        && crate::utils::header_id_utils::is_standalone_attr_list(next_line)
                        && let Some(next_line_id) =
                            crate::utils::header_id_utils::extract_standalone_attr_list_id(next_line)
                    {
                        custom_id = Some(next_line_id);
                    }
                }

                lines[i].heading = Some(HeadingInfo {
                    level,
                    style: HeadingStyle::ATX,
                    marker: hashes.to_string(),
                    marker_column,
                    content_column,
                    text: clean_text,
                    custom_id,
                    raw_text,
                    has_closing_sequence: has_closing,
                    closing_sequence: closing_seq,
                });
            }
            // Check for Setext headings (need to look at next line)
            else if i + 1 < content_lines.len() && i + 1 < lines.len() {
                let next_line = content_lines[i + 1];
                if !lines[i + 1].in_code_block && SETEXT_UNDERLINE_REGEX.is_match(next_line) {
                    // Skip if next line is front matter delimiter
                    if front_matter_end > 0 && i < front_matter_end {
                        continue;
                    }

                    // Skip Setext headings inside HTML comments (using pre-computed ranges for efficiency)
                    if crate::utils::skip_context::is_in_html_comment_ranges(html_comment_ranges, lines[i].byte_offset)
                    {
                        continue;
                    }

                    let underline = next_line.trim();

                    let level = if underline.starts_with('=') { 1 } else { 2 };
                    let style = if level == 1 {
                        HeadingStyle::Setext1
                    } else {
                        HeadingStyle::Setext2
                    };

                    // Extract custom header ID if present
                    let raw_text = line.trim().to_string();
                    let (clean_text, mut custom_id) = crate::utils::header_id_utils::extract_header_id(&raw_text);

                    // If no custom ID was found on the header line, check the line after underline for standalone attr-list
                    if custom_id.is_none() && i + 2 < content_lines.len() && i + 2 < lines.len() {
                        let attr_line = content_lines[i + 2];
                        if !lines[i + 2].in_code_block
                            && crate::utils::header_id_utils::is_standalone_attr_list(attr_line)
                            && let Some(attr_line_id) =
                                crate::utils::header_id_utils::extract_standalone_attr_list_id(attr_line)
                        {
                            custom_id = Some(attr_line_id);
                        }
                    }

                    lines[i].heading = Some(HeadingInfo {
                        level,
                        style,
                        marker: underline.to_string(),
                        marker_column: next_line.len() - next_line.trim_start().len(),
                        content_column: lines[i].indent,
                        text: clean_text,
                        custom_id,
                        raw_text,
                        has_closing_sequence: false,
                        closing_sequence: String::new(),
                    });
                }
            }
        }
    }

    /// Detect HTML blocks in the content
    fn detect_html_blocks(content: &str, lines: &mut [LineInfo]) {
        // HTML block elements that trigger block context
        const BLOCK_ELEMENTS: &[&str] = &[
            "address",
            "article",
            "aside",
            "blockquote",
            "details",
            "dialog",
            "dd",
            "div",
            "dl",
            "dt",
            "fieldset",
            "figcaption",
            "figure",
            "footer",
            "form",
            "h1",
            "h2",
            "h3",
            "h4",
            "h5",
            "h6",
            "header",
            "hr",
            "li",
            "main",
            "nav",
            "ol",
            "p",
            "picture",
            "pre",
            "script",
            "section",
            "style",
            "table",
            "tbody",
            "td",
            "textarea",
            "tfoot",
            "th",
            "thead",
            "tr",
            "ul",
        ];

        let mut i = 0;
        while i < lines.len() {
            // Skip if already in code block or front matter
            if lines[i].in_code_block || lines[i].in_front_matter {
                i += 1;
                continue;
            }

            let trimmed = lines[i].content(content).trim_start();

            // Check if line starts with an HTML tag
            if trimmed.starts_with('<') && trimmed.len() > 1 {
                // Extract tag name safely
                let after_bracket = &trimmed[1..];
                let is_closing = after_bracket.starts_with('/');
                let tag_start = if is_closing { &after_bracket[1..] } else { after_bracket };

                // Extract tag name (stop at space, >, /, or end of string)
                let tag_name = tag_start
                    .chars()
                    .take_while(|c| c.is_ascii_alphabetic() || *c == '-' || c.is_ascii_digit())
                    .collect::<String>()
                    .to_lowercase();

                // Check if it's a block element
                if !tag_name.is_empty() && BLOCK_ELEMENTS.contains(&tag_name.as_str()) {
                    // Mark this line as in HTML block
                    lines[i].in_html_block = true;

                    // For simplicity, just mark lines until we find a closing tag or reach a blank line
                    // This avoids complex nesting logic that might cause infinite loops
                    if !is_closing {
                        let closing_tag = format!("</{tag_name}>");
                        // style and script tags can contain blank lines (CSS/JS formatting)
                        let allow_blank_lines = tag_name == "style" || tag_name == "script";
                        let mut j = i + 1;
                        while j < lines.len() && j < i + 100 {
                            // Limit search to 100 lines
                            // Stop at blank lines (except for style/script tags)
                            if !allow_blank_lines && lines[j].is_blank {
                                break;
                            }

                            lines[j].in_html_block = true;

                            // Check if this line contains the closing tag
                            if lines[j].content(content).contains(&closing_tag) {
                                break;
                            }
                            j += 1;
                        }
                    }
                }
            }

            i += 1;
        }
    }

    /// Detect ESM import/export blocks in MDX files
    /// ESM blocks consist of contiguous import/export statements at the top of the file
    fn detect_esm_blocks(content: &str, lines: &mut [LineInfo], flavor: MarkdownFlavor) {
        // Only process MDX files
        if !flavor.supports_esm_blocks() {
            return;
        }

        for line in lines.iter_mut() {
            // Skip blank lines and comments at the start
            if line.is_blank || line.in_html_comment {
                continue;
            }

            // Check if line starts with import or export
            let trimmed = line.content(content).trim_start();
            if trimmed.starts_with("import ") || trimmed.starts_with("export ") {
                line.in_esm_block = true;
            } else {
                // Once we hit a non-ESM line, we're done with the ESM block
                break;
            }
        }
    }

    /// Parse all inline code spans in the content using pulldown-cmark streaming parser
    fn parse_code_spans(content: &str, lines: &[LineInfo]) -> Vec<CodeSpan> {
        let mut code_spans = Vec::new();

        // Quick check - if no backticks, no code spans
        if !content.contains('`') {
            return code_spans;
        }

        // Use pulldown-cmark's streaming parser with byte offsets
        let parser = Parser::new(content).into_offset_iter();

        for (event, range) in parser {
            if let Event::Code(_) = event {
                let start_pos = range.start;
                let end_pos = range.end;

                // The range includes the backticks, extract the actual content
                let full_span = &content[start_pos..end_pos];
                let backtick_count = full_span.chars().take_while(|&c| c == '`').count();

                // Extract content between backticks, preserving spaces
                let content_start = start_pos + backtick_count;
                let content_end = end_pos - backtick_count;
                let span_content = if content_start < content_end {
                    content[content_start..content_end].to_string()
                } else {
                    String::new()
                };

                // Use binary search to find line number - O(log n) instead of O(n)
                // Find the rightmost line whose byte_offset <= start_pos
                let line_idx = lines
                    .partition_point(|line| line.byte_offset <= start_pos)
                    .saturating_sub(1);
                let line_num = line_idx + 1;
                let byte_col_start = start_pos - lines[line_idx].byte_offset;

                // Find end column using binary search
                let end_line_idx = lines
                    .partition_point(|line| line.byte_offset <= end_pos)
                    .saturating_sub(1);
                let byte_col_end = end_pos - lines[end_line_idx].byte_offset;

                // Convert byte offsets to character positions for correct Unicode handling
                // This ensures consistency with warning.column which uses character positions
                let line_content = lines[line_idx].content(content);
                let col_start = if byte_col_start <= line_content.len() {
                    line_content[..byte_col_start].chars().count()
                } else {
                    line_content.chars().count()
                };

                let end_line_content = lines[end_line_idx].content(content);
                let col_end = if byte_col_end <= end_line_content.len() {
                    end_line_content[..byte_col_end].chars().count()
                } else {
                    end_line_content.chars().count()
                };

                code_spans.push(CodeSpan {
                    line: line_num,
                    end_line: end_line_idx + 1,
                    start_col: col_start,
                    end_col: col_end,
                    byte_offset: start_pos,
                    byte_end: end_pos,
                    backtick_count,
                    content: span_content,
                });
            }
        }

        // Sort by position to ensure consistent ordering
        code_spans.sort_by_key(|span| span.byte_offset);

        code_spans
    }

    /// Parse all list blocks in the content (legacy line-by-line approach)
    ///
    /// Uses a forward-scanning O(n) algorithm that tracks two variables during iteration:
    /// - `has_list_breaking_content_since_last_item`: Set when encountering content that
    ///   terminates a list (headings, horizontal rules, tables, insufficiently indented content)
    /// - `min_continuation_for_tracking`: Minimum indentation required for content to be
    ///   treated as list continuation (based on the list marker width)
    ///
    /// When a new list item is encountered, we check if list-breaking content was seen
    /// since the last item. If so, we start a new list block.
    fn parse_list_blocks(content: &str, lines: &[LineInfo]) -> Vec<ListBlock> {
        // Minimum indentation for unordered list continuation per CommonMark spec
        const UNORDERED_LIST_MIN_CONTINUATION_INDENT: usize = 2;

        /// Initialize or reset the forward-scanning tracking state.
        /// This helper eliminates code duplication across three initialization sites.
        #[inline]
        fn reset_tracking_state(
            list_item: &ListItemInfo,
            has_list_breaking_content: &mut bool,
            min_continuation: &mut usize,
        ) {
            *has_list_breaking_content = false;
            let marker_width = if list_item.is_ordered {
                list_item.marker.len() + 1 // Ordered markers need space after period/paren
            } else {
                list_item.marker.len()
            };
            *min_continuation = if list_item.is_ordered {
                marker_width
            } else {
                UNORDERED_LIST_MIN_CONTINUATION_INDENT
            };
        }

        // Pre-size based on lines that could be list items
        let mut list_blocks = Vec::with_capacity(lines.len() / 10); // Estimate ~10% of lines might start list blocks
        let mut current_block: Option<ListBlock> = None;
        let mut last_list_item_line = 0;
        let mut current_indent_level = 0;
        let mut last_marker_width = 0;

        // Track list-breaking content since last item (fixes O(n²) bottleneck from issue #148)
        let mut has_list_breaking_content_since_last_item = false;
        let mut min_continuation_for_tracking = 0;

        for (line_idx, line_info) in lines.iter().enumerate() {
            let line_num = line_idx + 1;

            // Enhanced code block handling using Design #3's context analysis
            if line_info.in_code_block {
                if let Some(ref mut block) = current_block {
                    // Calculate minimum indentation for list continuation
                    let min_continuation_indent =
                        CodeBlockUtils::calculate_min_continuation_indent(content, lines, line_idx);

                    // Analyze code block context using the three-tier classification
                    let context = CodeBlockUtils::analyze_code_block_context(lines, line_idx, min_continuation_indent);

                    match context {
                        CodeBlockContext::Indented => {
                            // Code block is properly indented - continues the list
                            block.end_line = line_num;
                            continue;
                        }
                        CodeBlockContext::Standalone => {
                            // Code block separates lists - end current block
                            let completed_block = current_block.take().unwrap();
                            list_blocks.push(completed_block);
                            continue;
                        }
                        CodeBlockContext::Adjacent => {
                            // Edge case - use conservative behavior (continue list)
                            block.end_line = line_num;
                            continue;
                        }
                    }
                } else {
                    // No current list block - skip code block lines
                    continue;
                }
            }

            // Extract blockquote prefix if any
            let blockquote_prefix = if let Some(caps) = BLOCKQUOTE_PREFIX_REGEX.captures(line_info.content(content)) {
                caps.get(0).unwrap().as_str().to_string()
            } else {
                String::new()
            };

            // Track list-breaking content for non-list, non-blank lines (O(n) replacement for nested loop)
            // Skip lines that are continuations of multi-line code spans - they're part of the previous list item
            if current_block.is_some()
                && line_info.list_item.is_none()
                && !line_info.is_blank
                && !line_info.in_code_span_continuation
            {
                let line_content = line_info.content(content).trim();

                // Count pipes outside of inline code spans (to avoid confusing `||` for table)
                let pipes_outside_code = {
                    let mut count = 0;
                    let mut in_code = false;
                    for ch in line_content.chars() {
                        if ch == '`' {
                            in_code = !in_code;
                        } else if ch == '|' && !in_code {
                            count += 1;
                        }
                    }
                    count
                };

                // Check for structural separators that break lists
                let breaks_list = line_info.heading.is_some()
                    || line_content.starts_with("---")
                    || line_content.starts_with("***")
                    || line_content.starts_with("___")
                    || (pipes_outside_code > 0
                        && !line_content.contains("](")
                        && !line_content.contains("http")
                        && (pipes_outside_code > 1 || line_content.starts_with('|') || line_content.ends_with('|')))
                    || line_content.starts_with(">")
                    || (line_info.indent < min_continuation_for_tracking);

                if breaks_list {
                    has_list_breaking_content_since_last_item = true;
                }
            }

            // If this line is a code span continuation within an active list block,
            // extend the block's end_line to include this line (maintains list continuity)
            if line_info.in_code_span_continuation
                && line_info.list_item.is_none()
                && let Some(ref mut block) = current_block
            {
                block.end_line = line_num;
            }

            // Extend block.end_line for regular continuation lines (non-list-item, non-blank,
            // properly indented lines within the list). This ensures the workaround at line 2448
            // works correctly when there are multiple continuation lines before a nested list item.
            if !line_info.in_code_span_continuation
                && line_info.list_item.is_none()
                && !line_info.is_blank
                && !line_info.in_code_block
                && line_info.indent >= min_continuation_for_tracking
                && let Some(ref mut block) = current_block
            {
                block.end_line = line_num;
            }

            // Check if this line is a list item
            if let Some(list_item) = &line_info.list_item {
                // Calculate nesting level based on indentation
                let item_indent = list_item.marker_column;
                let nesting = item_indent / 2; // Assume 2-space indentation for nesting

                if let Some(ref mut block) = current_block {
                    // Check if this continues the current block
                    // For nested lists, we need to check if this is a nested item (higher nesting level)
                    // or a continuation at the same or lower level
                    let is_nested = nesting > block.nesting_level;
                    let same_type =
                        (block.is_ordered && list_item.is_ordered) || (!block.is_ordered && !list_item.is_ordered);
                    let same_context = block.blockquote_prefix == blockquote_prefix;
                    // Allow one blank line after last item, or lines immediately after block content
                    let reasonable_distance = line_num <= last_list_item_line + 2 || line_num == block.end_line + 1;

                    // For unordered lists, also check marker consistency
                    let marker_compatible =
                        block.is_ordered || block.marker.is_none() || block.marker.as_ref() == Some(&list_item.marker);

                    // O(1) check: Use the tracked variable instead of O(n) nested loop
                    // This eliminates the quadratic bottleneck from issue #148
                    let has_non_list_content = has_list_breaking_content_since_last_item;

                    // A list continues if:
                    // 1. It's a nested item (indented more than the parent), OR
                    // 2. It's the same type at the same level with reasonable distance
                    let mut continues_list = if is_nested {
                        // Nested items always continue the list if they're in the same context
                        same_context && reasonable_distance && !has_non_list_content
                    } else {
                        // Same-level items need to match type and markers
                        same_type && same_context && reasonable_distance && marker_compatible && !has_non_list_content
                    };

                    // WORKAROUND: If items are truly consecutive (no blank lines), they MUST be in the same list
                    // This handles edge cases where content patterns might otherwise split lists incorrectly
                    if !continues_list && reasonable_distance && line_num > 0 && block.end_line == line_num - 1 {
                        // Check if the previous line was a list item
                        if block.item_lines.contains(&(line_num - 1)) {
                            // They're consecutive list items - force them to be in the same list
                            continues_list = true;
                        }
                    }

                    if continues_list {
                        // Extend current block
                        block.end_line = line_num;
                        block.item_lines.push(line_num);

                        // Update max marker width
                        block.max_marker_width = block.max_marker_width.max(if list_item.is_ordered {
                            list_item.marker.len() + 1
                        } else {
                            list_item.marker.len()
                        });

                        // Update marker consistency for unordered lists
                        if !block.is_ordered
                            && block.marker.is_some()
                            && block.marker.as_ref() != Some(&list_item.marker)
                        {
                            // Mixed markers, clear the marker field
                            block.marker = None;
                        }

                        // Reset tracked state for issue #148 optimization
                        reset_tracking_state(
                            list_item,
                            &mut has_list_breaking_content_since_last_item,
                            &mut min_continuation_for_tracking,
                        );
                    } else {
                        // End current block and start a new one

                        list_blocks.push(block.clone());

                        *block = ListBlock {
                            start_line: line_num,
                            end_line: line_num,
                            is_ordered: list_item.is_ordered,
                            marker: if list_item.is_ordered {
                                None
                            } else {
                                Some(list_item.marker.clone())
                            },
                            blockquote_prefix: blockquote_prefix.clone(),
                            item_lines: vec![line_num],
                            nesting_level: nesting,
                            max_marker_width: if list_item.is_ordered {
                                list_item.marker.len() + 1
                            } else {
                                list_item.marker.len()
                            },
                        };

                        // Initialize tracked state for new block (issue #148 optimization)
                        reset_tracking_state(
                            list_item,
                            &mut has_list_breaking_content_since_last_item,
                            &mut min_continuation_for_tracking,
                        );
                    }
                } else {
                    // Start a new block
                    current_block = Some(ListBlock {
                        start_line: line_num,
                        end_line: line_num,
                        is_ordered: list_item.is_ordered,
                        marker: if list_item.is_ordered {
                            None
                        } else {
                            Some(list_item.marker.clone())
                        },
                        blockquote_prefix,
                        item_lines: vec![line_num],
                        nesting_level: nesting,
                        max_marker_width: list_item.marker.len(),
                    });

                    // Initialize tracked state for new block (issue #148 optimization)
                    reset_tracking_state(
                        list_item,
                        &mut has_list_breaking_content_since_last_item,
                        &mut min_continuation_for_tracking,
                    );
                }

                last_list_item_line = line_num;
                current_indent_level = item_indent;
                last_marker_width = if list_item.is_ordered {
                    list_item.marker.len() + 1 // Add 1 for the space after ordered list markers
                } else {
                    list_item.marker.len()
                };
            } else if let Some(ref mut block) = current_block {
                // Not a list item - check if it continues the current block

                // For MD032 compatibility, we use a simple approach:
                // - Indented lines continue the list
                // - Blank lines followed by indented content continue the list
                // - Everything else ends the list

                // Check if the last line in the list block ended with a backslash (hard line break)
                // This handles cases where list items use backslash for hard line breaks
                let prev_line_ends_with_backslash = if block.end_line > 0 && block.end_line - 1 < lines.len() {
                    lines[block.end_line - 1].content(content).trim_end().ends_with('\\')
                } else {
                    false
                };

                // Calculate minimum indentation for list continuation
                // For ordered lists, use the last marker width (e.g., 3 for "1. ", 4 for "10. ")
                // For unordered lists like "- ", content starts at column 2, so continuations need at least 2 spaces
                let min_continuation_indent = if block.is_ordered {
                    current_indent_level + last_marker_width
                } else {
                    current_indent_level + 2 // Unordered lists need at least 2 spaces (e.g., "- " = 2 chars)
                };

                if prev_line_ends_with_backslash || line_info.indent >= min_continuation_indent {
                    // Indented line or backslash continuation continues the list
                    block.end_line = line_num;
                } else if line_info.is_blank {
                    // Blank line - check if it's internal to the list or ending it
                    // We only include blank lines that are followed by more list content
                    let mut check_idx = line_idx + 1;
                    let mut found_continuation = false;

                    // Skip additional blank lines
                    while check_idx < lines.len() && lines[check_idx].is_blank {
                        check_idx += 1;
                    }

                    if check_idx < lines.len() {
                        let next_line = &lines[check_idx];
                        // Check if followed by indented content (list continuation)
                        if !next_line.in_code_block && next_line.indent >= min_continuation_indent {
                            found_continuation = true;
                        }
                        // Check if followed by another list item at the same level
                        else if !next_line.in_code_block
                            && next_line.list_item.is_some()
                            && let Some(item) = &next_line.list_item
                        {
                            let next_blockquote_prefix = BLOCKQUOTE_PREFIX_REGEX
                                .find(next_line.content(content))
                                .map_or(String::new(), |m| m.as_str().to_string());
                            if item.marker_column == current_indent_level
                                && item.is_ordered == block.is_ordered
                                && block.blockquote_prefix.trim() == next_blockquote_prefix.trim()
                            {
                                // Check if there was meaningful content between the list items (unused now)
                                // This variable is kept for potential future use but is currently replaced by has_structural_separators
                                let _has_meaningful_content = (line_idx + 1..check_idx).any(|idx| {
                                    if let Some(between_line) = lines.get(idx) {
                                        let between_content = between_line.content(content);
                                        let trimmed = between_content.trim();
                                        // Skip empty lines
                                        if trimmed.is_empty() {
                                            return false;
                                        }
                                        // Check for meaningful content
                                        let line_indent = between_content.len() - between_content.trim_start().len();

                                        // Structural separators (code fences, headings, etc.) are meaningful and should BREAK lists
                                        if trimmed.starts_with("```")
                                            || trimmed.starts_with("~~~")
                                            || trimmed.starts_with("---")
                                            || trimmed.starts_with("***")
                                            || trimmed.starts_with("___")
                                            || trimmed.starts_with(">")
                                            || trimmed.contains('|') // Tables
                                            || between_line.heading.is_some()
                                        {
                                            return true; // These are structural separators - meaningful content that breaks lists
                                        }

                                        // Only properly indented content continues the list
                                        line_indent >= min_continuation_indent
                                    } else {
                                        false
                                    }
                                });

                                if block.is_ordered {
                                    // For ordered lists: don't continue if there are structural separators
                                    // Check if there are structural separators between the list items
                                    let has_structural_separators = (line_idx + 1..check_idx).any(|idx| {
                                        if let Some(between_line) = lines.get(idx) {
                                            let trimmed = between_line.content(content).trim();
                                            if trimmed.is_empty() {
                                                return false;
                                            }
                                            // Check for structural separators that break lists
                                            trimmed.starts_with("```")
                                                || trimmed.starts_with("~~~")
                                                || trimmed.starts_with("---")
                                                || trimmed.starts_with("***")
                                                || trimmed.starts_with("___")
                                                || trimmed.starts_with(">")
                                                || trimmed.contains('|') // Tables
                                                || between_line.heading.is_some()
                                        } else {
                                            false
                                        }
                                    });
                                    found_continuation = !has_structural_separators;
                                } else {
                                    // For unordered lists: also check for structural separators
                                    let has_structural_separators = (line_idx + 1..check_idx).any(|idx| {
                                        if let Some(between_line) = lines.get(idx) {
                                            let trimmed = between_line.content(content).trim();
                                            if trimmed.is_empty() {
                                                return false;
                                            }
                                            // Check for structural separators that break lists
                                            trimmed.starts_with("```")
                                                || trimmed.starts_with("~~~")
                                                || trimmed.starts_with("---")
                                                || trimmed.starts_with("***")
                                                || trimmed.starts_with("___")
                                                || trimmed.starts_with(">")
                                                || trimmed.contains('|') // Tables
                                                || between_line.heading.is_some()
                                        } else {
                                            false
                                        }
                                    });
                                    found_continuation = !has_structural_separators;
                                }
                            }
                        }
                    }

                    if found_continuation {
                        // Include the blank line in the block
                        block.end_line = line_num;
                    } else {
                        // Blank line ends the list - don't include it
                        list_blocks.push(block.clone());
                        current_block = None;
                    }
                } else {
                    // Check for lazy continuation - non-indented line immediately after a list item
                    // But only if the line has sufficient indentation for the list type
                    let min_required_indent = if block.is_ordered {
                        current_indent_level + last_marker_width
                    } else {
                        current_indent_level + 2
                    };

                    // For lazy continuation to apply, the line must either:
                    // 1. Have no indentation (true lazy continuation)
                    // 2. Have sufficient indentation for the list type
                    // BUT structural separators (headings, code blocks, etc.) should never be lazy continuations
                    let line_content = line_info.content(content).trim();
                    let is_structural_separator = line_info.heading.is_some()
                        || line_content.starts_with("```")
                        || line_content.starts_with("~~~")
                        || line_content.starts_with("---")
                        || line_content.starts_with("***")
                        || line_content.starts_with("___")
                        || line_content.starts_with(">")
                        || (line_content.contains('|')
                            && !line_content.contains("](")
                            && !line_content.contains("http")
                            && (line_content.matches('|').count() > 1
                                || line_content.starts_with('|')
                                || line_content.ends_with('|'))); // Tables

                    // Allow lazy continuation if we're still within the same list block
                    // (not just immediately after a list item)
                    let is_lazy_continuation = !is_structural_separator
                        && !line_info.is_blank
                        && (line_info.indent == 0 || line_info.indent >= min_required_indent);

                    if is_lazy_continuation {
                        // Additional check: if the line starts with uppercase and looks like a new sentence,
                        // it's probably not a continuation
                        let content_to_check = if !blockquote_prefix.is_empty() {
                            // Strip blockquote prefix to check the actual content
                            line_info
                                .content(content)
                                .strip_prefix(&blockquote_prefix)
                                .unwrap_or(line_info.content(content))
                                .trim()
                        } else {
                            line_info.content(content).trim()
                        };

                        let starts_with_uppercase = content_to_check.chars().next().is_some_and(|c| c.is_uppercase());

                        // If it starts with uppercase and the previous line ended with punctuation,
                        // it's likely a new paragraph, not a continuation
                        if starts_with_uppercase && last_list_item_line > 0 {
                            // This looks like a new paragraph
                            list_blocks.push(block.clone());
                            current_block = None;
                        } else {
                            // This is a lazy continuation line
                            block.end_line = line_num;
                        }
                    } else {
                        // Non-indented, non-blank line that's not a lazy continuation - end the block
                        list_blocks.push(block.clone());
                        current_block = None;
                    }
                }
            }
        }

        // Don't forget the last block
        if let Some(block) = current_block {
            list_blocks.push(block);
        }

        // Merge adjacent blocks that should be one
        merge_adjacent_list_blocks(content, &mut list_blocks, lines);

        list_blocks
    }

    /// Compute character frequency for fast content analysis
    fn compute_char_frequency(content: &str) -> CharFrequency {
        let mut frequency = CharFrequency::default();

        for ch in content.chars() {
            match ch {
                '#' => frequency.hash_count += 1,
                '*' => frequency.asterisk_count += 1,
                '_' => frequency.underscore_count += 1,
                '-' => frequency.hyphen_count += 1,
                '+' => frequency.plus_count += 1,
                '>' => frequency.gt_count += 1,
                '|' => frequency.pipe_count += 1,
                '[' => frequency.bracket_count += 1,
                '`' => frequency.backtick_count += 1,
                '<' => frequency.lt_count += 1,
                '!' => frequency.exclamation_count += 1,
                '\n' => frequency.newline_count += 1,
                _ => {}
            }
        }

        frequency
    }

    /// Parse HTML tags in the content
    fn parse_html_tags(
        content: &str,
        lines: &[LineInfo],
        code_blocks: &[(usize, usize)],
        flavor: MarkdownFlavor,
    ) -> Vec<HtmlTag> {
        static HTML_TAG_REGEX: LazyLock<regex::Regex> =
            LazyLock::new(|| regex::Regex::new(r"(?i)<(/?)([a-zA-Z][a-zA-Z0-9]*)(?:\s+[^>]*?)?\s*(/?)>").unwrap());

        let mut html_tags = Vec::with_capacity(content.matches('<').count());

        for cap in HTML_TAG_REGEX.captures_iter(content) {
            let full_match = cap.get(0).unwrap();
            let match_start = full_match.start();
            let match_end = full_match.end();

            // Skip if in code block
            if CodeBlockUtils::is_in_code_block_or_span(code_blocks, match_start) {
                continue;
            }

            let is_closing = !cap.get(1).unwrap().as_str().is_empty();
            let tag_name_original = cap.get(2).unwrap().as_str();
            let tag_name = tag_name_original.to_lowercase();
            let is_self_closing = !cap.get(3).unwrap().as_str().is_empty();

            // Skip JSX components in MDX files (tags starting with uppercase letter)
            // JSX components like <Chart />, <MyComponent> should not be treated as HTML
            if flavor.supports_jsx() && tag_name_original.chars().next().is_some_and(|c| c.is_uppercase()) {
                continue;
            }

            // Find which line this tag is on
            let mut line_num = 1;
            let mut col_start = match_start;
            let mut col_end = match_end;
            for (idx, line_info) in lines.iter().enumerate() {
                if match_start >= line_info.byte_offset {
                    line_num = idx + 1;
                    col_start = match_start - line_info.byte_offset;
                    col_end = match_end - line_info.byte_offset;
                } else {
                    break;
                }
            }

            html_tags.push(HtmlTag {
                line: line_num,
                start_col: col_start,
                end_col: col_end,
                byte_offset: match_start,
                byte_end: match_end,
                tag_name,
                is_closing,
                is_self_closing,
                raw_content: full_match.as_str().to_string(),
            });
        }

        html_tags
    }

    /// Parse emphasis spans in the content
    fn parse_emphasis_spans(content: &str, lines: &[LineInfo], code_blocks: &[(usize, usize)]) -> Vec<EmphasisSpan> {
        static EMPHASIS_REGEX: LazyLock<regex::Regex> =
            LazyLock::new(|| regex::Regex::new(r"(\*{1,3}|_{1,3})([^*_\s][^*_]*?)(\*{1,3}|_{1,3})").unwrap());

        let mut emphasis_spans = Vec::with_capacity(content.matches('*').count() + content.matches('_').count() / 4);

        for cap in EMPHASIS_REGEX.captures_iter(content) {
            let full_match = cap.get(0).unwrap();
            let match_start = full_match.start();
            let match_end = full_match.end();

            // Skip if in code block
            if CodeBlockUtils::is_in_code_block_or_span(code_blocks, match_start) {
                continue;
            }

            let opening_markers = cap.get(1).unwrap().as_str();
            let content_part = cap.get(2).unwrap().as_str();
            let closing_markers = cap.get(3).unwrap().as_str();

            // Validate matching markers
            if opening_markers.chars().next() != closing_markers.chars().next()
                || opening_markers.len() != closing_markers.len()
            {
                continue;
            }

            let marker = opening_markers.chars().next().unwrap();
            let marker_count = opening_markers.len();

            // Find which line this emphasis is on
            let mut line_num = 1;
            let mut col_start = match_start;
            let mut col_end = match_end;
            for (idx, line_info) in lines.iter().enumerate() {
                if match_start >= line_info.byte_offset {
                    line_num = idx + 1;
                    col_start = match_start - line_info.byte_offset;
                    col_end = match_end - line_info.byte_offset;
                } else {
                    break;
                }
            }

            emphasis_spans.push(EmphasisSpan {
                line: line_num,
                start_col: col_start,
                end_col: col_end,
                byte_offset: match_start,
                byte_end: match_end,
                marker,
                marker_count,
                content: content_part.to_string(),
            });
        }

        emphasis_spans
    }

    /// Parse table rows in the content
    fn parse_table_rows(content: &str, lines: &[LineInfo]) -> Vec<TableRow> {
        let mut table_rows = Vec::with_capacity(lines.len() / 20);

        for (line_idx, line_info) in lines.iter().enumerate() {
            // Skip lines in code blocks or blank lines
            if line_info.in_code_block || line_info.is_blank {
                continue;
            }

            let line = line_info.content(content);
            let line_num = line_idx + 1;

            // Check if this line contains pipes (potential table row)
            if !line.contains('|') {
                continue;
            }

            // Count columns by splitting on pipes
            let parts: Vec<&str> = line.split('|').collect();
            let column_count = if parts.len() > 2 { parts.len() - 2 } else { parts.len() };

            // Check if this is a separator row
            let is_separator = line.chars().all(|c| "|:-+ \t".contains(c));
            let mut column_alignments = Vec::new();

            if is_separator {
                for part in &parts[1..parts.len() - 1] {
                    // Skip first and last empty parts
                    let trimmed = part.trim();
                    let alignment = if trimmed.starts_with(':') && trimmed.ends_with(':') {
                        "center".to_string()
                    } else if trimmed.ends_with(':') {
                        "right".to_string()
                    } else if trimmed.starts_with(':') {
                        "left".to_string()
                    } else {
                        "none".to_string()
                    };
                    column_alignments.push(alignment);
                }
            }

            table_rows.push(TableRow {
                line: line_num,
                is_separator,
                column_count,
                column_alignments,
            });
        }

        table_rows
    }

    /// Parse bare URLs and emails in the content
    fn parse_bare_urls(content: &str, lines: &[LineInfo], code_blocks: &[(usize, usize)]) -> Vec<BareUrl> {
        let mut bare_urls = Vec::with_capacity(content.matches("http").count() + content.matches('@').count());

        // Check for bare URLs (not in angle brackets or markdown links)
        for cap in BARE_URL_PATTERN.captures_iter(content) {
            let full_match = cap.get(0).unwrap();
            let match_start = full_match.start();
            let match_end = full_match.end();

            // Skip if in code block
            if CodeBlockUtils::is_in_code_block_or_span(code_blocks, match_start) {
                continue;
            }

            // Skip if already in angle brackets or markdown links
            let preceding_char = if match_start > 0 {
                content.chars().nth(match_start - 1)
            } else {
                None
            };
            let following_char = content.chars().nth(match_end);

            if preceding_char == Some('<') || preceding_char == Some('(') || preceding_char == Some('[') {
                continue;
            }
            if following_char == Some('>') || following_char == Some(')') || following_char == Some(']') {
                continue;
            }

            let url = full_match.as_str();
            let url_type = if url.starts_with("https://") {
                "https"
            } else if url.starts_with("http://") {
                "http"
            } else if url.starts_with("ftp://") {
                "ftp"
            } else {
                "other"
            };

            // Find which line this URL is on
            let mut line_num = 1;
            let mut col_start = match_start;
            let mut col_end = match_end;
            for (idx, line_info) in lines.iter().enumerate() {
                if match_start >= line_info.byte_offset {
                    line_num = idx + 1;
                    col_start = match_start - line_info.byte_offset;
                    col_end = match_end - line_info.byte_offset;
                } else {
                    break;
                }
            }

            bare_urls.push(BareUrl {
                line: line_num,
                start_col: col_start,
                end_col: col_end,
                byte_offset: match_start,
                byte_end: match_end,
                url: url.to_string(),
                url_type: url_type.to_string(),
            });
        }

        // Check for bare email addresses
        for cap in BARE_EMAIL_PATTERN.captures_iter(content) {
            let full_match = cap.get(0).unwrap();
            let match_start = full_match.start();
            let match_end = full_match.end();

            // Skip if in code block
            if CodeBlockUtils::is_in_code_block_or_span(code_blocks, match_start) {
                continue;
            }

            // Skip if already in angle brackets or markdown links
            let preceding_char = if match_start > 0 {
                content.chars().nth(match_start - 1)
            } else {
                None
            };
            let following_char = content.chars().nth(match_end);

            if preceding_char == Some('<') || preceding_char == Some('(') || preceding_char == Some('[') {
                continue;
            }
            if following_char == Some('>') || following_char == Some(')') || following_char == Some(']') {
                continue;
            }

            let email = full_match.as_str();

            // Find which line this email is on
            let mut line_num = 1;
            let mut col_start = match_start;
            let mut col_end = match_end;
            for (idx, line_info) in lines.iter().enumerate() {
                if match_start >= line_info.byte_offset {
                    line_num = idx + 1;
                    col_start = match_start - line_info.byte_offset;
                    col_end = match_end - line_info.byte_offset;
                } else {
                    break;
                }
            }

            bare_urls.push(BareUrl {
                line: line_num,
                start_col: col_start,
                end_col: col_end,
                byte_offset: match_start,
                byte_end: match_end,
                url: email.to_string(),
                url_type: "email".to_string(),
            });
        }

        bare_urls
    }
}

/// Merge adjacent list blocks that should be treated as one
fn merge_adjacent_list_blocks(content: &str, list_blocks: &mut Vec<ListBlock>, lines: &[LineInfo]) {
    if list_blocks.len() < 2 {
        return;
    }

    let mut merger = ListBlockMerger::new(content, lines);
    *list_blocks = merger.merge(list_blocks);
}

/// Helper struct to manage the complex logic of merging list blocks
struct ListBlockMerger<'a> {
    content: &'a str,
    lines: &'a [LineInfo],
}

impl<'a> ListBlockMerger<'a> {
    fn new(content: &'a str, lines: &'a [LineInfo]) -> Self {
        Self { content, lines }
    }

    fn merge(&mut self, list_blocks: &[ListBlock]) -> Vec<ListBlock> {
        let mut merged = Vec::with_capacity(list_blocks.len());
        let mut current = list_blocks[0].clone();

        for next in list_blocks.iter().skip(1) {
            if self.should_merge_blocks(&current, next) {
                current = self.merge_two_blocks(current, next);
            } else {
                merged.push(current);
                current = next.clone();
            }
        }

        merged.push(current);
        merged
    }

    /// Determine if two adjacent list blocks should be merged
    fn should_merge_blocks(&self, current: &ListBlock, next: &ListBlock) -> bool {
        // Basic compatibility checks
        if !self.blocks_are_compatible(current, next) {
            return false;
        }

        // Check spacing and content between blocks
        let spacing = self.analyze_spacing_between(current, next);
        match spacing {
            BlockSpacing::Consecutive => true,
            BlockSpacing::SingleBlank => self.can_merge_with_blank_between(current, next),
            BlockSpacing::MultipleBlanks | BlockSpacing::ContentBetween => {
                self.can_merge_with_content_between(current, next)
            }
        }
    }

    /// Check if blocks have compatible structure for merging
    fn blocks_are_compatible(&self, current: &ListBlock, next: &ListBlock) -> bool {
        current.is_ordered == next.is_ordered
            && current.blockquote_prefix == next.blockquote_prefix
            && current.nesting_level == next.nesting_level
    }

    /// Analyze the spacing between two list blocks
    fn analyze_spacing_between(&self, current: &ListBlock, next: &ListBlock) -> BlockSpacing {
        let gap = next.start_line - current.end_line;

        match gap {
            1 => BlockSpacing::Consecutive,
            2 => BlockSpacing::SingleBlank,
            _ if gap > 2 => {
                if self.has_only_blank_lines_between(current, next) {
                    BlockSpacing::MultipleBlanks
                } else {
                    BlockSpacing::ContentBetween
                }
            }
            _ => BlockSpacing::Consecutive, // gap == 0, overlapping (shouldn't happen)
        }
    }

    /// Check if unordered lists can be merged with a single blank line between
    fn can_merge_with_blank_between(&self, current: &ListBlock, next: &ListBlock) -> bool {
        // Check if there are structural separators between the blocks
        // If has_meaningful_content_between returns true, it means there are structural separators
        if has_meaningful_content_between(self.content, current, next, self.lines) {
            return false; // Structural separators prevent merging
        }

        // Only merge unordered lists with same marker across single blank
        !current.is_ordered && current.marker == next.marker
    }

    /// Check if ordered lists can be merged when there's content between them
    fn can_merge_with_content_between(&self, current: &ListBlock, next: &ListBlock) -> bool {
        // Do not merge lists if there are structural separators between them
        if has_meaningful_content_between(self.content, current, next, self.lines) {
            return false; // Structural separators prevent merging
        }

        // Only consider merging ordered lists if there's no structural content between
        current.is_ordered && next.is_ordered
    }

    /// Check if there are only blank lines between blocks
    fn has_only_blank_lines_between(&self, current: &ListBlock, next: &ListBlock) -> bool {
        for line_num in (current.end_line + 1)..next.start_line {
            if let Some(line_info) = self.lines.get(line_num - 1)
                && !line_info.content(self.content).trim().is_empty()
            {
                return false;
            }
        }
        true
    }

    /// Merge two compatible list blocks into one
    fn merge_two_blocks(&self, mut current: ListBlock, next: &ListBlock) -> ListBlock {
        current.end_line = next.end_line;
        current.item_lines.extend_from_slice(&next.item_lines);

        // Update max marker width
        current.max_marker_width = current.max_marker_width.max(next.max_marker_width);

        // Handle marker consistency for unordered lists
        if !current.is_ordered && self.markers_differ(&current, next) {
            current.marker = None; // Mixed markers
        }

        current
    }

    /// Check if two blocks have different markers
    fn markers_differ(&self, current: &ListBlock, next: &ListBlock) -> bool {
        current.marker.is_some() && next.marker.is_some() && current.marker != next.marker
    }
}

/// Types of spacing between list blocks
#[derive(Debug, PartialEq)]
enum BlockSpacing {
    Consecutive,    // No gap between blocks
    SingleBlank,    // One blank line between blocks
    MultipleBlanks, // Multiple blank lines but no content
    ContentBetween, // Content exists between blocks
}

/// Check if there's meaningful content (not just blank lines) between two list blocks
fn has_meaningful_content_between(content: &str, current: &ListBlock, next: &ListBlock, lines: &[LineInfo]) -> bool {
    // Check lines between current.end_line and next.start_line
    for line_num in (current.end_line + 1)..next.start_line {
        if let Some(line_info) = lines.get(line_num - 1) {
            // Convert to 0-indexed
            let trimmed = line_info.content(content).trim();

            // Skip empty lines
            if trimmed.is_empty() {
                continue;
            }

            // Check for structural separators that should separate lists (CommonMark compliant)

            // Headings separate lists
            if line_info.heading.is_some() {
                return true; // Has meaningful content - headings separate lists
            }

            // Horizontal rules separate lists (---, ***, ___)
            if is_horizontal_rule(trimmed) {
                return true; // Has meaningful content - horizontal rules separate lists
            }

            // Tables separate lists (lines containing | but not in URLs or code)
            // Simple heuristic: tables typically have | at start/end or multiple |
            if trimmed.contains('|') && trimmed.len() > 1 {
                // Don't treat URLs with | as tables
                if !trimmed.contains("](") && !trimmed.contains("http") {
                    // More robust check: tables usually have multiple | or | at edges
                    let pipe_count = trimmed.matches('|').count();
                    if pipe_count > 1 || trimmed.starts_with('|') || trimmed.ends_with('|') {
                        return true; // Has meaningful content - tables separate lists
                    }
                }
            }

            // Blockquotes separate lists
            if trimmed.starts_with('>') {
                return true; // Has meaningful content - blockquotes separate lists
            }

            // Code block fences separate lists (unless properly indented as list content)
            if trimmed.starts_with("```") || trimmed.starts_with("~~~") {
                let line_indent = line_info.byte_len - line_info.content(content).trim_start().len();

                // Check if this code block is properly indented as list continuation
                let min_continuation_indent = if current.is_ordered {
                    current.nesting_level + current.max_marker_width + 1 // +1 for space after marker
                } else {
                    current.nesting_level + 2
                };

                if line_indent < min_continuation_indent {
                    // This is a standalone code block that separates lists
                    return true; // Has meaningful content - standalone code blocks separate lists
                }
            }

            // Check if this line has proper indentation for list continuation
            let line_indent = line_info.byte_len - line_info.content(content).trim_start().len();

            // Calculate minimum indentation needed to be list continuation
            let min_indent = if current.is_ordered {
                current.nesting_level + current.max_marker_width
            } else {
                current.nesting_level + 2
            };

            // If the line is not indented enough to be list continuation, it's meaningful content
            if line_indent < min_indent {
                return true; // Has meaningful content - content not indented as list continuation
            }

            // If we reach here, the line is properly indented as list continuation
            // Continue checking other lines
        }
    }

    // Only blank lines or properly indented list continuation content between blocks
    false
}

/// Check if a line is a horizontal rule (---, ***, ___)
fn is_horizontal_rule(trimmed: &str) -> bool {
    if trimmed.len() < 3 {
        return false;
    }

    // Check for three or more consecutive -, *, or _ characters (with optional spaces)
    let chars: Vec<char> = trimmed.chars().collect();
    if let Some(&first_char) = chars.first()
        && (first_char == '-' || first_char == '*' || first_char == '_')
    {
        let mut count = 0;
        for &ch in &chars {
            if ch == first_char {
                count += 1;
            } else if ch != ' ' && ch != '\t' {
                return false; // Non-matching, non-whitespace character
            }
        }
        return count >= 3;
    }
    false
}

/// Check if content contains patterns that cause the markdown crate to panic
#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_empty_content() {
        let ctx = LintContext::new("", MarkdownFlavor::Standard, None);
        assert_eq!(ctx.content, "");
        assert_eq!(ctx.line_offsets, vec![0]);
        assert_eq!(ctx.offset_to_line_col(0), (1, 1));
        assert_eq!(ctx.lines.len(), 0);
    }

    #[test]
    fn test_single_line() {
        let ctx = LintContext::new("# Hello", MarkdownFlavor::Standard, None);
        assert_eq!(ctx.content, "# Hello");
        assert_eq!(ctx.line_offsets, vec![0]);
        assert_eq!(ctx.offset_to_line_col(0), (1, 1));
        assert_eq!(ctx.offset_to_line_col(3), (1, 4));
    }

    #[test]
    fn test_multi_line() {
        let content = "# Title\n\nSecond line\nThird line";
        let ctx = LintContext::new(content, MarkdownFlavor::Standard, None);
        assert_eq!(ctx.line_offsets, vec![0, 8, 9, 21]);
        // Test offset to line/col
        assert_eq!(ctx.offset_to_line_col(0), (1, 1)); // start
        assert_eq!(ctx.offset_to_line_col(8), (2, 1)); // start of blank line
        assert_eq!(ctx.offset_to_line_col(9), (3, 1)); // start of 'Second line'
        assert_eq!(ctx.offset_to_line_col(15), (3, 7)); // middle of 'Second line'
        assert_eq!(ctx.offset_to_line_col(21), (4, 1)); // start of 'Third line'
    }

    #[test]
    fn test_line_info() {
        let content = "# Title\n    indented\n\ncode:\n```rust\nfn main() {}\n```";
        let ctx = LintContext::new(content, MarkdownFlavor::Standard, None);

        // Test line info
        assert_eq!(ctx.lines.len(), 7);

        // Line 1: "# Title"
        let line1 = &ctx.lines[0];
        assert_eq!(line1.content(ctx.content), "# Title");
        assert_eq!(line1.byte_offset, 0);
        assert_eq!(line1.indent, 0);
        assert!(!line1.is_blank);
        assert!(!line1.in_code_block);
        assert!(line1.list_item.is_none());

        // Line 2: "    indented"
        let line2 = &ctx.lines[1];
        assert_eq!(line2.content(ctx.content), "    indented");
        assert_eq!(line2.byte_offset, 8);
        assert_eq!(line2.indent, 4);
        assert!(!line2.is_blank);

        // Line 3: "" (blank)
        let line3 = &ctx.lines[2];
        assert_eq!(line3.content(ctx.content), "");
        assert!(line3.is_blank);

        // Test helper methods
        assert_eq!(ctx.line_to_byte_offset(1), Some(0));
        assert_eq!(ctx.line_to_byte_offset(2), Some(8));
        assert_eq!(ctx.line_info(1).map(|l| l.indent), Some(0));
        assert_eq!(ctx.line_info(2).map(|l| l.indent), Some(4));
    }

    #[test]
    fn test_list_item_detection() {
        let content = "- Unordered item\n  * Nested item\n1. Ordered item\n   2) Nested ordered\n\nNot a list";
        let ctx = LintContext::new(content, MarkdownFlavor::Standard, None);

        // Line 1: "- Unordered item"
        let line1 = &ctx.lines[0];
        assert!(line1.list_item.is_some());
        let list1 = line1.list_item.as_ref().unwrap();
        assert_eq!(list1.marker, "-");
        assert!(!list1.is_ordered);
        assert_eq!(list1.marker_column, 0);
        assert_eq!(list1.content_column, 2);

        // Line 2: "  * Nested item"
        let line2 = &ctx.lines[1];
        assert!(line2.list_item.is_some());
        let list2 = line2.list_item.as_ref().unwrap();
        assert_eq!(list2.marker, "*");
        assert_eq!(list2.marker_column, 2);

        // Line 3: "1. Ordered item"
        let line3 = &ctx.lines[2];
        assert!(line3.list_item.is_some());
        let list3 = line3.list_item.as_ref().unwrap();
        assert_eq!(list3.marker, "1.");
        assert!(list3.is_ordered);
        assert_eq!(list3.number, Some(1));

        // Line 6: "Not a list"
        let line6 = &ctx.lines[5];
        assert!(line6.list_item.is_none());
    }

    #[test]
    fn test_offset_to_line_col_edge_cases() {
        let content = "a\nb\nc";
        let ctx = LintContext::new(content, MarkdownFlavor::Standard, None);
        // line_offsets: [0, 2, 4]
        assert_eq!(ctx.offset_to_line_col(0), (1, 1)); // 'a'
        assert_eq!(ctx.offset_to_line_col(1), (1, 2)); // after 'a'
        assert_eq!(ctx.offset_to_line_col(2), (2, 1)); // 'b'
        assert_eq!(ctx.offset_to_line_col(3), (2, 2)); // after 'b'
        assert_eq!(ctx.offset_to_line_col(4), (3, 1)); // 'c'
        assert_eq!(ctx.offset_to_line_col(5), (3, 2)); // after 'c'
    }

    #[test]
    fn test_mdx_esm_blocks() {
        let content = r##"import {Chart} from './snowfall.js'
export const year = 2023

# Last year's snowfall

In {year}, the snowfall was above average.
It was followed by a warm spring which caused
flood conditions in many of the nearby rivers.

<Chart color="#fcb32c" year={year} />
"##;

        let ctx = LintContext::new(content, MarkdownFlavor::MDX, None);

        // Check that lines 1 and 2 are marked as ESM blocks
        assert_eq!(ctx.lines.len(), 10);
        assert!(ctx.lines[0].in_esm_block, "Line 1 (import) should be in_esm_block");
        assert!(ctx.lines[1].in_esm_block, "Line 2 (export) should be in_esm_block");
        assert!(!ctx.lines[2].in_esm_block, "Line 3 (blank) should NOT be in_esm_block");
        assert!(
            !ctx.lines[3].in_esm_block,
            "Line 4 (heading) should NOT be in_esm_block"
        );
        assert!(!ctx.lines[4].in_esm_block, "Line 5 (blank) should NOT be in_esm_block");
        assert!(!ctx.lines[5].in_esm_block, "Line 6 (text) should NOT be in_esm_block");
    }

    #[test]
    fn test_mdx_esm_blocks_not_detected_in_standard_flavor() {
        let content = r#"import {Chart} from './snowfall.js'
export const year = 2023

# Last year's snowfall
"#;

        let ctx = LintContext::new(content, MarkdownFlavor::Standard, None);

        // ESM blocks should NOT be detected in Standard flavor
        assert!(
            !ctx.lines[0].in_esm_block,
            "Line 1 should NOT be in_esm_block in Standard flavor"
        );
        assert!(
            !ctx.lines[1].in_esm_block,
            "Line 2 should NOT be in_esm_block in Standard flavor"
        );
    }
}