probe_code/language/

parser.rs

use anyhow::{Context, Result};
use dashmap::DashMap;
use once_cell::sync::Lazy;
use std::collections::hash_map::DefaultHasher;
use std::collections::{HashMap, HashSet};
use std::hash::{Hash, Hasher};
use tree_sitter::{Node, Parser as TSParser};

use probe_code::language::factory::get_language_impl;
use probe_code::language::language_trait::LanguageImpl;
use probe_code::language::tree_cache;
use probe_code::models::CodeBlock;

// Define a static cache for line maps
static LINE_MAP_CACHE: Lazy<DashMap<String, Vec<Option<CachedNodeInfo>>>> = Lazy::new(DashMap::new);

/// Calculate a hash of the content for cache validation
fn calculate_content_hash(content: &str) -> u64 {
    let mut hasher = DefaultHasher::new();
    content.hash(&mut hasher);
    hasher.finish()
}

/// A version of NodeInfo without lifetimes for caching
#[derive(Clone, Debug)]
struct CachedNodeInfo {
    // Original node info
    start_byte: usize,
    end_byte: usize,
    start_row: usize,
    end_row: usize,
    node_kind: String,
    is_comment: bool,
    is_test: bool,                     // Test status of the original node
    original_node_is_acceptable: bool, // Was the original node an acceptable parent type?
    // Context node info (if any)
    context_node_bytes: Option<(usize, usize)>,
    context_node_rows: Option<(usize, usize)>,
    context_node_kind: Option<String>,
    context_node_is_test: Option<bool>, // Test status of the context node (if any)
    // specificity: usize, // Specificity of the original node assignment - REMOVED (unused)
    // Parent function info (if applicable)
    parent_node_type: Option<String>,
    parent_start_row: Option<usize>,
    parent_end_row: Option<usize>,
    // Representative node info - REMOVED (unused)
    // representative_start_byte: usize,
    // representative_end_byte: usize,
    // representative_start_row: usize,
    // representative_end_row: usize,
    // representative_node_kind: String,
    // is_merged_comment: bool, // Flag if this represents a merged comment+context block - REMOVED (unused)
}

impl CachedNodeInfo {
    /// Create a CachedNodeInfo from a NodeInfo and determine the representative node
    fn from_node_info(
        info: &NodeInfo<'_>,
        language_impl: &dyn LanguageImpl,
        content: &[u8],
        allow_tests: bool,
    ) -> Self {
        // Determine the representative node based on the same logic used in the live processing path
        let mut rep_node = info.node; // Default to self
                                      // let mut is_merged = false; // Removed unused variable

        if info.is_comment {
            if let Some(ctx) = info.context_node {
                if !allow_tests && !language_impl.is_test_node(&ctx, content) {
                    rep_node = ctx; // Context represents the merged block
                                    // is_merged = true; // Removed assignment to unused variable
                }
            }
        } else if !info.is_test {
            if let Some(ctx) = info.context_node {
                if !allow_tests && !language_impl.is_test_node(&ctx, content) {
                    rep_node = ctx; // Use context ancestor
                }
            }
        }
        // If info.is_test is true, rep_node remains info.node

        // Get parent function info if applicable (e.g., for struct_type nodes)
        let parent_info = if rep_node.kind() == "struct_type" {
            language_impl
                .find_parent_function(rep_node)
                .map(|parent_node| {
                    let parent_type = parent_node.kind().to_string();
                    let parent_start = parent_node.start_position().row;
                    let parent_end = parent_node.end_position().row;
                    (parent_type, parent_start, parent_end)
                })
        } else {
            None
        };

        // Check if original node is acceptable
        let original_acceptable = language_impl.is_acceptable_parent(&info.node);

        // Check if context node is a test node
        let context_test = info
            .context_node
            .map(|ctx| language_impl.is_test_node(&ctx, content));

        CachedNodeInfo {
            // Original node details
            start_byte: info.node.start_byte(),
            end_byte: info.node.end_byte(),
            start_row: info.node.start_position().row,
            end_row: info.node.end_position().row,
            node_kind: info.node.kind().to_string(),
            is_comment: info.is_comment,
            is_test: info.is_test, // Original node test status
            original_node_is_acceptable: original_acceptable,
            // Context node details
            context_node_bytes: info.context_node.map(|n| (n.start_byte(), n.end_byte())),
            context_node_rows: info
                .context_node
                .map(|n| (n.start_position().row, n.end_position().row)),
            context_node_kind: info.context_node.map(|n| n.kind().to_string()),
            context_node_is_test: context_test, // Context node test status
            // specificity: info.specificity, // Original node specificity - REMOVED (unused)
            // Parent function info
            parent_node_type: parent_info.as_ref().map(|(t, _, _)| t.clone()),
            parent_start_row: parent_info.as_ref().map(|(_, s, _)| *s),
            parent_end_row: parent_info.as_ref().map(|(_, _, e)| *e),
            // Representative node details - REMOVED (unused)
            // representative_start_byte: rep_node.start_byte(),
            // representative_end_byte: rep_node.end_byte(),
            // representative_start_row: rep_node.start_position().row,
            // representative_end_row: rep_node.end_position().row,
            // representative_node_kind: rep_node.kind().to_string(),
            // is_merged_comment: is_merged,
        }
    }
}

/// Structure to hold node information for a specific line
#[derive(Clone, Copy)]
struct NodeInfo<'a> {
    node: Node<'a>,
    is_comment: bool,
    context_node: Option<Node<'a>>, // Represents the nearest acceptable ancestor if node itself isn't one
    is_test: bool,
    specificity: usize,
}

/// Helper function to determine if we should update the line map for a given line
fn should_update_line_map<'a>(
    line_map: &[Option<NodeInfo<'a>>],
    line: usize,
    node: Node<'a>,
    is_comment: bool,
    context_node: Option<Node<'a>>,
    specificity: usize,
) -> bool {
    match &line_map[line] {
        None => true, // No existing node, always update
        Some(current) => {
            // Special case: If current node is a comment with context, and new node is the context,
            // don't replace it (preserve the comment+context relationship)
            if current.is_comment && current.context_node.is_some() {
                if let Some(ctx) = current.context_node {
                    if ctx.id() == node.id() {
                        return false;
                    }
                }
            }

            // Special case: If new node is a comment with context, and current node is the context,
            // replace it (comment with context is more specific)
            if is_comment && context_node.is_some() {
                if let Some(ctx) = context_node {
                    if ctx.id() == current.node.id() {
                        return true;
                    }
                }
            }

            // Otherwise use specificity to decide
            specificity < current.specificity
        }
    }
}

/// Gets the previous sibling of a node in the AST
fn find_prev_sibling(node: Node<'_>) -> Option<Node<'_>> {
    let parent = node.parent()?;

    let mut cursor = parent.walk();
    let mut prev_child = None;

    for child in parent.children(&mut cursor) {
        if child.id() == node.id() {
            return prev_child;
        }
        prev_child = Some(child);
    }

    None // No previous sibling found
}

/// Find first acceptable node in a subtree
fn find_acceptable_child<'a>(node: Node<'a>, language_impl: &dyn LanguageImpl) -> Option<Node<'a>> {
    let mut cursor = node.walk();
    for child in node.children(&mut cursor) {
        if language_impl.is_acceptable_parent(&child) {
            return Some(child);
        }

        // Recursive search
        if let Some(acceptable) = find_acceptable_child(child, language_impl) {
            return Some(acceptable);
        }
    }

    None // No acceptable child found
}

/// Finds the immediate next node that follows a given node in the AST
fn find_immediate_next_node(node: Node<'_>) -> Option<Node<'_>> {
    let debug_mode = std::env::var("DEBUG").unwrap_or_default() == "1";

    // First try direct next sibling
    if let Some(next) = node.next_sibling() {
        if debug_mode {
            println!(
                "DEBUG: Found immediate next sibling: type='{}', lines={}-{}",
                next.kind(),
                next.start_position().row + 1,
                next.end_position().row + 1
            );
        }
        return Some(next);
    }

    // If no direct sibling, check parent's next sibling
    if let Some(parent) = node.parent() {
        if let Some(next_parent) = parent.next_sibling() {
            if debug_mode {
                println!(
                    "DEBUG: Found parent's next sibling: type='{}', lines={}-{}",
                    next_parent.kind(),
                    next_parent.start_position().row + 1,
                    next_parent.end_position().row + 1
                );
            }
            return Some(next_parent);
        }
    }

    if debug_mode {
        println!("DEBUG: No immediate next node found");
    }
    None
}

/// Helper function to find the context node for a comment.
/// This is a comprehensive implementation that handles all comment context finding strategies.
fn find_comment_context_node<'a>(
    comment_node: Node<'a>,
    language_impl: &dyn LanguageImpl,
    debug_mode: bool,
) -> Option<Node<'a>> {
    let start_row = comment_node.start_position().row;

    if debug_mode {
        println!(
            "DEBUG: Finding context for comment at lines {}-{}: {}",
            comment_node.start_position().row + 1,
            comment_node.end_position().row + 1,
            comment_node.kind()
        );
    }

    // Strategy 1: Try to find next non-comment sibling first (most common case for doc comments)
    let mut current_sibling = comment_node.next_sibling();

    // Skip over any comment siblings to find the next non-comment sibling
    while let Some(sibling) = current_sibling {
        if sibling.kind() == "comment"
            || sibling.kind() == "line_comment"
            || sibling.kind() == "block_comment"
            || sibling.kind() == "doc_comment"
            || sibling.kind() == "//"
        {
            // This is another comment, move to the next sibling
            current_sibling = sibling.next_sibling();
            continue;
        }

        // Found a non-comment sibling
        if language_impl.is_acceptable_parent(&sibling) {
            if debug_mode {
                println!(
                    "DEBUG: Found next non-comment sibling for comment at line {}: type='{}', lines={}-{}",
                    start_row + 1,
                    sibling.kind(),
                    sibling.start_position().row + 1,
                    sibling.end_position().row + 1
                );
            }
            return Some(sibling);
        } else {
            // If next sibling isn't acceptable, check its children
            if let Some(child) = find_acceptable_child(sibling, language_impl) {
                if debug_mode {
                    println!(
                        "DEBUG: Found acceptable child in next non-comment sibling for comment at line {}: type='{}', lines={}-{}",
                        start_row + 1,
                        child.kind(),
                        child.start_position().row + 1,
                        child.end_position().row + 1
                    );
                }
                return Some(child);
            }
        }

        // If we get here, this non-comment sibling wasn't acceptable, try the next one
        current_sibling = sibling.next_sibling();
    }

    // Strategy 2: If no acceptable next sibling, try previous sibling (for trailing comments)
    // But only if the comment is at the end of a block or if there's no next sibling
    // This helps ensure comments are associated with the code that follows them when possible
    let has_next_sibling = comment_node.next_sibling().is_some();

    if !has_next_sibling {
        if let Some(prev_sibling) = find_prev_sibling(comment_node) {
            if language_impl.is_acceptable_parent(&prev_sibling) {
                if debug_mode {
                    println!(
                        "DEBUG: Found previous sibling for comment at line {}: type='{}', lines={}-{}",
                        start_row + 1,
                        prev_sibling.kind(),
                        prev_sibling.start_position().row + 1,
                        prev_sibling.end_position().row + 1
                    );
                }
                return Some(prev_sibling);
            } else {
                // If previous sibling isn't acceptable, check its children
                if let Some(child) = find_acceptable_child(prev_sibling, language_impl) {
                    if debug_mode {
                        println!(
                            "DEBUG: Found acceptable child in previous sibling for comment at line {}: type='{}', lines={}-{}",
                            start_row + 1,
                            child.kind(),
                            child.start_position().row + 1,
                            child.end_position().row + 1
                        );
                    }
                    return Some(child);
                }
            }
        }
    }

    // Strategy 3: Check parent chain
    let mut current = comment_node;
    while let Some(parent) = current.parent() {
        if language_impl.is_acceptable_parent(&parent) {
            if debug_mode {
                println!(
                    "DEBUG: Found parent for comment at line {}: type='{}', lines={}-{}",
                    start_row + 1,
                    parent.kind(),
                    parent.start_position().row + 1,
                    parent.end_position().row + 1
                );
            }
            return Some(parent);
        }
        current = parent;
    }

    // Strategy 4: Look for any immediate next node
    if let Some(next_node) = find_immediate_next_node(comment_node) {
        if language_impl.is_acceptable_parent(&next_node) {
            if debug_mode {
                println!(
                    "DEBUG: Using immediate next acceptable node: type='{}', lines={}-{}",
                    next_node.kind(),
                    next_node.start_position().row + 1,
                    next_node.end_position().row + 1
                );
            }
            return Some(next_node);
        }

        // Look for acceptable child in the next node
        if let Some(child) = find_acceptable_child(next_node, language_impl) {
            if debug_mode {
                println!(
                    "DEBUG: Found acceptable child in next node: type='{}', lines={}-{}",
                    child.kind(),
                    child.start_position().row + 1,
                    child.end_position().row + 1
                );
            }
            return Some(child);
        }
    }

    if debug_mode {
        println!("DEBUG: No related node found for the comment");
    }
    None
}

/// Process a node and its children in a single pass, building a comprehensive line-to-node map.
/// This version passes the nearest acceptable ancestor context down the tree.
#[allow(clippy::too_many_arguments)]
fn process_node<'a>(
    node: Node<'a>,
    line_map: &mut Vec<Option<NodeInfo<'a>>>,
    _extension: &str, // Keep extension if needed by language_impl methods, otherwise remove
    language_impl: &dyn LanguageImpl,
    content: &[u8],
    allow_tests: bool,
    debug_mode: bool,
    current_ancestor: Option<Node<'a>>, // The nearest acceptable ancestor found so far
) {
    let start_row = node.start_position().row;
    let end_row = node.end_position().row;

    // Skip nodes that are outside the file bounds (e.g., if file content changed during processing)
    if start_row >= line_map.len() {
        return;
    }

    // Determine node type and test status
    let is_comment = node.kind() == "comment"
        || node.kind() == "line_comment"
        || node.kind() == "block_comment"
        || node.kind() == "doc_comment"
        || node.kind() == "//"; // Example for some languages

    // Check if the node itself represents test code
    let is_test = !allow_tests && language_impl.is_test_node(&node, content);

    // Calculate node specificity (smaller is more specific)
    let line_coverage = end_row.saturating_sub(start_row) + 1;
    let byte_coverage = node.end_byte().saturating_sub(node.start_byte());
    let specificity = line_coverage * 1000 + (byte_coverage / 100); // Example specificity calculation

    // Determine the context_node for this node
    let context_node = if is_comment {
        // For comments, find the related code node (e.g., the function it documents)
        // This function might still need to look up/around, but doesn't use the ancestor cache.
        find_comment_context_node(node, language_impl, debug_mode)
    } else {
        // For non-comments, if the node itself isn't an acceptable block boundary,
        // use the ancestor context passed down. Otherwise, it defines its own context (None).
        if !language_impl.is_acceptable_parent(&node) {
            current_ancestor
        } else {
            None // This node is an acceptable parent, it starts a new context.
        }
    };

    // Update the line map for each line covered by this node
    // Ensure end_row does not exceed line_map bounds
    let effective_end_row = std::cmp::min(end_row, line_map.len().saturating_sub(1));
    for line in start_row..=effective_end_row {
        // Determine if this node is a better fit for the line than the current entry
        let should_update =
            should_update_line_map(line_map, line, node, is_comment, context_node, specificity);

        if should_update {
            // Store info about the node covering this line
            line_map[line] = Some(NodeInfo {
                node,
                is_comment,
                context_node, // Store the determined context (parent ancestor or None)
                is_test,      // Store the test status of this specific node
                specificity,
            });
        }
    }

    // Determine the ancestor context to pass down to children
    let next_ancestor = if language_impl.is_acceptable_parent(&node) {
        // If this node is an acceptable parent, it becomes the context for its children
        Some(node)
    } else {
        // Otherwise, children inherit the same context as this node
        current_ancestor
    };

    // Process children recursively (depth-first traversal)
    let mut cursor = node.walk();
    for child in node.children(&mut cursor) {
        process_node(
            child,
            line_map,
            _extension,
            language_impl,
            content,
            allow_tests,
            debug_mode,
            next_ancestor, // Pass the determined ancestor context down
        );
    }
}

/// Process a cached line map to extract code blocks
fn process_cached_line_map(
    cached_line_map: &[Option<CachedNodeInfo>],
    line_numbers: &HashSet<usize>,
    _language_impl: &dyn LanguageImpl, // Not used directly, logic relies on cached info
    _content: &str,                    // Not used directly, logic relies on cached info
    allow_tests: bool,
    debug_mode: bool,
) -> Result<Vec<CodeBlock>> {
    let mut code_blocks: Vec<CodeBlock> = Vec::new();
    // Use a HashSet to track the start/end rows of blocks already added
    let mut seen_block_spans: HashSet<(usize, usize)> = HashSet::new();

    // Process each line number using the cached map
    for &line in line_numbers {
        let line_idx = line.saturating_sub(1); // Adjust for 0-based indexing

        if debug_mode {
            println!("DEBUG: Processing line {line} from cache");
        }

        if line_idx >= cached_line_map.len() {
            if debug_mode {
                println!("DEBUG: Line {line} is out of bounds (Cache)");
            }
            continue;
        }

        if let Some(info) = &cached_line_map[line_idx] {
            if debug_mode {
                println!(
                    "DEBUG: Found cached node info for line {}: original_type='{}', original_lines={}-{}, is_comment={}, is_test={}, context_kind={:?}, context_lines={:?}",
                    line,
                    info.node_kind,
                    info.start_row + 1,
                    info.end_row + 1,
                    info.is_comment,
                    info.is_test,
                    info.context_node_kind,
                    info.context_node_rows.map(|(s, e)| (s + 1, e + 1))
                );
            }

            // Determine which block to potentially create based on cached info
            let mut potential_block: Option<CodeBlock> = None;
            let mut block_key: Option<(usize, usize)> = None; // Key for seen_block_spans

            // --- Replicate Cache Miss Logic using CachedNodeInfo ---

            // 1. Handle Comments
            if info.is_comment {
                if debug_mode {
                    println!("DEBUG: Cache: Handling comment node at line {line}");
                }
                // Check for context node
                if let (Some(ctx_rows), Some(ctx_bytes), Some(ctx_kind), Some(ctx_is_test)) = (
                    info.context_node_rows,
                    info.context_node_bytes,
                    &info.context_node_kind,
                    info.context_node_is_test,
                ) {
                    // Check test status of the context node
                    if !allow_tests && ctx_is_test {
                        if debug_mode {
                            println!(
                                "DEBUG: Cache: Skipping test context node at lines {}-{}, type: {}",
                                ctx_rows.0 + 1,
                                ctx_rows.1 + 1,
                                ctx_kind
                            );
                        }
                        // Fall through to potentially add individual comment if context is skipped
                    } else {
                        // Create a merged block
                        let merged_start_row = std::cmp::min(info.start_row, ctx_rows.0);
                        let merged_end_row = std::cmp::max(info.end_row, ctx_rows.1);
                        let merged_start_byte = std::cmp::min(info.start_byte, ctx_bytes.0);
                        let merged_end_byte = std::cmp::max(info.end_byte, ctx_bytes.1);

                        block_key = Some((merged_start_row, merged_end_row));
                        if !seen_block_spans.contains(&block_key.unwrap()) {
                            potential_block = Some(CodeBlock {
                                start_row: merged_start_row,
                                end_row: merged_end_row,
                                start_byte: merged_start_byte,
                                end_byte: merged_end_byte,
                                node_type: ctx_kind.clone(), // Use context kind for merged block
                                parent_node_type: None, // Consistent with original miss path logic
                                parent_start_row: None,
                                parent_end_row: None,
                            });
                            if debug_mode {
                                println!(
                                    "DEBUG: Cache: Potential merged block (comment + context) at lines {}-{}, type: {}",
                                    merged_start_row + 1, merged_end_row + 1, ctx_kind
                                );
                            }
                        }
                        // If we created a merged block, we don't add the individual comment later
                        // So we continue the outer loop here if the block_key was already seen or if we created a potential block
                        if seen_block_spans.contains(&block_key.unwrap())
                            || potential_block.is_some()
                        {
                            if seen_block_spans.contains(&block_key.unwrap()) && debug_mode {
                                println!(
                                    "DEBUG: Cache: Merged block span {}-{} already seen",
                                    block_key.unwrap().0 + 1,
                                    block_key.unwrap().1 + 1
                                );
                            }
                            // Add the key even if block wasn't added, to prevent reprocessing context
                            seen_block_spans.insert(block_key.unwrap());
                            // Also mark original comment span as seen if merged
                            seen_block_spans.insert((info.start_row, info.end_row));
                            if let Some(block) = potential_block {
                                code_blocks.push(block);
                            }
                            continue; // Move to next line number
                        }
                    }
                }

                // Add individual comment if not merged or if context was skipped
                if potential_block.is_none() {
                    block_key = Some((info.start_row, info.end_row));
                    if !seen_block_spans.contains(&block_key.unwrap()) {
                        potential_block = Some(CodeBlock {
                            start_row: info.start_row,
                            end_row: info.end_row,
                            start_byte: info.start_byte,
                            end_byte: info.end_byte,
                            node_type: info.node_kind.clone(),
                            parent_node_type: None,
                            parent_start_row: None,
                            parent_end_row: None,
                        });
                        if debug_mode {
                            println!(
                                "DEBUG: Cache: Potential individual comment block at lines {}-{}",
                                info.start_row + 1,
                                info.end_row + 1
                            );
                        }
                    } else if debug_mode {
                        println!(
                            "DEBUG: Cache: Individual comment span {}-{} already seen",
                            block_key.unwrap().0 + 1,
                            block_key.unwrap().1 + 1
                        );
                    }
                }
            }
            // 2. Handle Non-Comments
            else {
                // Skip original test nodes if not allowed
                if !allow_tests && info.is_test {
                    if debug_mode {
                        println!(
                            "DEBUG: Cache: Skipping original test node at lines {}-{}",
                            info.start_row + 1,
                            info.end_row + 1
                        );
                    }
                    continue; // Move to next line number
                }

                // Check for context node (ancestor)
                if let (Some(ctx_rows), Some(ctx_bytes), Some(ctx_kind), Some(ctx_is_test)) = (
                    info.context_node_rows,
                    info.context_node_bytes,
                    &info.context_node_kind,
                    info.context_node_is_test,
                ) {
                    // Check test status of the context node
                    if !allow_tests && ctx_is_test {
                        if debug_mode {
                            println!(
                                "DEBUG: Cache: Skipping test context node (ancestor) at lines {}-{}",
                                ctx_rows.0 + 1, ctx_rows.1 + 1
                            );
                        }
                        // Fall through to check original node if context is skipped
                    } else {
                        // Use context node
                        block_key = Some((ctx_rows.0, ctx_rows.1));
                        if !seen_block_spans.contains(&block_key.unwrap()) {
                            potential_block = Some(CodeBlock {
                                start_row: ctx_rows.0,
                                end_row: ctx_rows.1,
                                start_byte: ctx_bytes.0,
                                end_byte: ctx_bytes.1,
                                node_type: ctx_kind.clone(),
                                // Parent info comes from CachedNodeInfo, which derived it based on the representative node (potentially the context)
                                parent_node_type: info.parent_node_type.clone(),
                                parent_start_row: info.parent_start_row,
                                parent_end_row: info.parent_end_row,
                            });
                            if debug_mode {
                                println!(
                                    "DEBUG: Cache: Potential context node (ancestor) block at lines {}-{}",
                                    ctx_rows.0 + 1, ctx_rows.1 + 1
                                );
                            }
                        } else if debug_mode {
                            println!(
                                "DEBUG: Cache: Context node span {}-{} already seen",
                                block_key.unwrap().0 + 1,
                                block_key.unwrap().1 + 1
                            );
                        }
                        // If we used the context node (or it was already seen), skip checking the original node
                        if seen_block_spans.contains(&block_key.unwrap())
                            || potential_block.is_some()
                        {
                            seen_block_spans.insert(block_key.unwrap()); // Mark context as seen
                            if let Some(block) = potential_block {
                                code_blocks.push(block);
                            }
                            continue; // Move to next line number
                        }
                    }
                }

                // Check if original node itself is acceptable (and wasn't skipped as test)
                // This check happens if there was no context node, or if the context node was skipped (e.g., test)
                if potential_block.is_none() && info.original_node_is_acceptable {
                    block_key = Some((info.start_row, info.end_row));
                    if !seen_block_spans.contains(&block_key.unwrap()) {
                        potential_block = Some(CodeBlock {
                            start_row: info.start_row,
                            end_row: info.end_row,
                            start_byte: info.start_byte,
                            end_byte: info.end_byte,
                            node_type: info.node_kind.clone(),
                            // Parent info comes from CachedNodeInfo, derived based on representative node (original node in this case)
                            parent_node_type: info.parent_node_type.clone(),
                            parent_start_row: info.parent_start_row,
                            parent_end_row: info.parent_end_row,
                        });
                        if debug_mode {
                            println!(
                                "DEBUG: Cache: Potential acceptable original node block at lines {}-{}",
                                info.start_row + 1, info.end_row + 1
                            );
                        }
                    } else if debug_mode {
                        println!(
                            "DEBUG: Cache: Original acceptable node span {}-{} already seen",
                            block_key.unwrap().0 + 1,
                            block_key.unwrap().1 + 1
                        );
                    }
                }
            }

            // Add the potential block if one was determined and not already seen
            if let (Some(block), Some(key)) = (potential_block, block_key) {
                if seen_block_spans.insert(key) {
                    // Returns true if the value was not present
                    code_blocks.push(block);
                }
            }
        } else if debug_mode {
            println!("DEBUG: Cache: No cached node info found for line {line}");
        }
    }

    // Removed extra closing brace that was here

    // Sort the blocks generated from the cache
    code_blocks.sort_by_key(|block| block.start_row);

    // --- Apply the exact same deduplication logic as the cache miss path ---
    let mut final_code_blocks: Vec<CodeBlock> = Vec::new();

    // Add comments first
    for block in code_blocks
        .iter()
        .filter(|b| b.node_type.contains("comment") || b.node_type == "/*" || b.node_type == "*/")
    {
        final_code_blocks.push(block.clone());
    }

    // Add non-comments, using the improved deduplication logic
    for block in code_blocks
        .iter() // Use iter() here as we pushed clones earlier
        .filter(|b| !b.node_type.contains("comment") && b.node_type != "/*" && b.node_type != "*/")
    {
        let mut should_add = true;
        let mut blocks_to_remove: Vec<usize> = Vec::new();

        // Define important block types that should be preserved
        let important_block_types = [
            "function_declaration",
            "method_declaration",
            "function_item",
            "impl_item",
            "type_declaration",
            "struct_item",
            "block_comment", // Keep this? Seems odd for non-comment filter but matches original
        ];
        let is_important = important_block_types.contains(&block.node_type.as_str());

        // Check if this block overlaps with any of the previous blocks in final_code_blocks
        for (idx, prev_block) in final_code_blocks.iter().enumerate() {
            if prev_block.node_type.contains("comment")
                || prev_block.node_type == "/*"
                || prev_block.node_type == "*/"
            {
                continue; // Skip comments already added
            }

            let prev_is_important = important_block_types.contains(&prev_block.node_type.as_str());

            // Check if blocks overlap
            if block.start_row <= prev_block.end_row && block.end_row >= prev_block.start_row {
                // Case 1: Current block is contained within previous block
                if block.start_row >= prev_block.start_row && block.end_row <= prev_block.end_row {
                    if debug_mode {
                        println!(
                            "DEBUG: Cache Dedupe: Current block contained: type='{}', lines={}-{} (in type='{}', lines={}-{})",
                            block.node_type, block.start_row + 1, block.end_row + 1,
                            prev_block.node_type, prev_block.start_row + 1, prev_block.end_row + 1
                        );
                    }
                    if is_important && !prev_is_important {
                        if debug_mode {
                            println!("DEBUG: Cache Dedupe: Keeping important contained block");
                        }
                        // Keep both - don't remove, don't skip add
                    } else if !is_important && prev_is_important {
                        if debug_mode {
                            println!("DEBUG: Cache Dedupe: Skipping non-important contained block");
                        }
                        should_add = false;
                        break;
                    } else {
                        // Both important or both not - prefer contained (current)
                        if debug_mode {
                            println!(
                                "DEBUG: Cache Dedupe: Replacing outer block with contained block"
                            );
                        }
                        blocks_to_remove.push(idx);
                    }
                }
                // Case 2: Previous block is contained within current block
                else if prev_block.start_row >= block.start_row
                    && prev_block.end_row <= block.end_row
                {
                    if debug_mode {
                        println!(
                            "DEBUG: Cache Dedupe: Previous block contained: type='{}', lines={}-{} (contains type='{}', lines={}-{})",
                            block.node_type, block.start_row + 1, block.end_row + 1,
                            prev_block.node_type, prev_block.start_row + 1, prev_block.end_row + 1
                        );
                    }
                    if is_important && !prev_is_important {
                        if debug_mode {
                            println!("DEBUG: Cache Dedupe: Keeping important outer block");
                        }
                        // Keep both - don't skip add, continue checking
                    } else if !is_important && prev_is_important {
                        if debug_mode {
                            println!("DEBUG: Cache Dedupe: Skipping non-important outer block");
                        }
                        should_add = false;
                        break;
                    } else {
                        // Both important or both not - prefer contained (previous)
                        if debug_mode {
                            println!("DEBUG: Cache Dedupe: Skipping outer block (already have contained)");
                        }
                        should_add = false;
                        break;
                    }
                }
                // Case 3: Blocks partially overlap
                else {
                    if debug_mode {
                        println!(
                            "DEBUG: Cache Dedupe: Partial overlap: type='{}', lines={}-{} (overlaps type='{}', lines={}-{})",
                            block.node_type, block.start_row + 1, block.end_row + 1,
                            prev_block.node_type, prev_block.start_row + 1, prev_block.end_row + 1
                        );
                    }
                    // Skip current block in case of partial overlap (consistent with miss path)
                    should_add = false;
                    break;
                }
            }
        }

        // Remove blocks marked for removal (in reverse order)
        for idx in blocks_to_remove.iter().rev() {
            final_code_blocks.remove(*idx);
        }

        // Add the current block if it wasn't skipped
        if should_add {
            final_code_blocks.push(block.clone());
        }
    }

    // Final sort to maintain correct order after deduplication
    final_code_blocks.sort_by_key(|block| block.start_row);
    Ok(final_code_blocks)
} // Added missing closing brace for process_cached_line_map
  // Removed unexpected closing brace that was here

/// Function to parse a file and extract code blocks for the given line numbers
pub fn parse_file_for_code_blocks(
    content: &str,
    extension: &str,
    line_numbers: &HashSet<usize>,
    allow_tests: bool,
    _term_matches: Option<&HashMap<usize, HashSet<usize>>>, // Query index to line numbers
) -> Result<Vec<CodeBlock>> {
    // Get the appropriate language implementation
    let language_impl = match get_language_impl(extension) {
        Some(lang) => lang,
        None => {
            return Err(anyhow::anyhow!(format!(
                "Unsupported file type: {}",
                extension
            )))
        }
    };

    // Check for debug mode
    let debug_mode = std::env::var("DEBUG").unwrap_or_default() == "1";

    // Calculate content hash for cache key
    let content_hash = calculate_content_hash(content);
    let cache_key = format!("{extension}_{content_hash}_{allow_tests}");

    // Check if we have a cached line map
    if let Some(cached_entry) = LINE_MAP_CACHE.get(&cache_key) {
        if debug_mode {
            println!("DEBUG: Cache hit for line_map key: {cache_key}");
        }

        // Process the cached line map
        return process_cached_line_map(
            cached_entry.value(),
            line_numbers,
            language_impl.as_ref(),
            content,
            allow_tests,
            debug_mode,
        );
    }

    if debug_mode {
        println!("DEBUG: Cache miss for line_map key: {cache_key}. Generating...");
    }

    // Get the tree-sitter language
    let language = language_impl.get_tree_sitter_language();

    // Parse the file
    let mut parser = TSParser::new();
    parser.set_language(&language)?;

    // Use the tree cache to get or parse the tree
    // We use a stable identifier for the file
    let tree_cache_key = format!("file_{extension}");
    let tree = tree_cache::get_or_parse_tree(&tree_cache_key, content, &mut parser)
        .context("Failed to parse the file")?;

    let root_node = tree.root_node();

    // Check for debug mode
    let debug_mode = std::env::var("DEBUG").unwrap_or_default() == "1";

    if debug_mode {
        println!("DEBUG: Parsing file with extension: {extension}");
        println!("DEBUG: Root node type: {}", root_node.kind());

        // Log all node types in the file
        let mut node_types = HashSet::new();
        super::common::collect_node_types(root_node, &mut node_types);
        println!("DEBUG: All node types in file: {node_types:?}");
    }

    // Create a line-to-node map for the entire file
    let line_count = content.lines().count();
    let mut line_map: Vec<Option<NodeInfo>> = vec![None; line_count];

    // Build the line-to-node map with a single traversal
    if debug_mode {
        println!("DEBUG: Building line-to-node map with a single traversal");
    }

    // For large files, we could parallelize the processing, but due to thread-safety
    // constraints with the language implementation, we'll use a sequential approach
    // that's still efficient for most cases
    if debug_mode {
        println!("DEBUG: Using sequential processing for AST nodes");
    }

    // Start the traversal from the root node, passing None as the initial ancestor context
    process_node(
        root_node,
        &mut line_map,
        extension, // Pass if needed by process_node/language_impl
        language_impl.as_ref(),
        content.as_bytes(),
        allow_tests,
        debug_mode,
        None, // Initial ancestor context is None
              // REMOVED: &mut ancestor_cache,
    );

    if debug_mode {
        println!("DEBUG: Line-to-node map built successfully");
    }

    // ====================================================================
    // START: Inserted Original Block Processing Logic (Cache Miss Path)
    // ====================================================================
    // This code runs ONLY on a cache miss, after process_node generates the live line_map.
    // It generates the CodeBlocks for *this specific request* from the live NodeInfo data.

    let mut code_blocks: Vec<CodeBlock> = Vec::new();
    let mut seen_nodes: HashSet<(usize, usize)> = HashSet::new(); // Use row-based key for this original logic

    // Process each line number using the *live* precomputed map (line_map)
    for &line in line_numbers {
        // Adjust for 0-based indexing
        let line_idx = line.saturating_sub(1);

        if debug_mode {
            println!("DEBUG: Processing line {line} (Live NodeInfo)");
        }

        // Skip if line is out of bounds
        if line_idx >= line_map.len() {
            if debug_mode {
                println!("DEBUG: Line {line} is out of bounds (Live NodeInfo)");
            }
            continue;
        }

        // Get the node info for this line from the live map
        if let Some(info) = &line_map[line_idx] {
            if debug_mode {
                println!(
                    "DEBUG: Found node for line {}: type='{}', lines={}-{}",
                    line,
                    info.node.kind(),
                    info.node.start_position().row + 1,
                    info.node.end_position().row + 1
                );
            }
            let target_node = info.node;
            let start_pos = target_node.start_position();
            let end_pos = target_node.end_position();
            // Use row key consistent with original logic for seen_nodes in this block
            let node_key = (start_pos.row, end_pos.row);

            // Skip if we've already processed this node
            if seen_nodes.contains(&node_key) {
                if debug_mode {
                    println!(
                        "DEBUG: Already processed node at lines {}-{}, type: {}",
                        start_pos.row + 1,
                        end_pos.row + 1,
                        target_node.kind()
                    );
                }
                continue;
            }

            // Mark this node as seen
            seen_nodes.insert(node_key);

            // Special handling for comments (using live NodeInfo and context_node)
            if info.is_comment {
                if debug_mode {
                    println!(
                        "DEBUG: Found comment node at line {}: {}",
                        line,
                        target_node.kind()
                    );
                }

                // If we have a context node for this comment
                if let Some(context_node) = info.context_node {
                    let rel_start_pos = context_node.start_position();
                    let rel_end_pos = context_node.end_position();
                    let rel_key = (rel_start_pos.row, rel_end_pos.row);

                    // Check test status using live node and language_impl
                    // Ensure content is available here if needed by is_test_node
                    if !allow_tests && language_impl.is_test_node(&context_node, content.as_bytes())
                    {
                        if debug_mode {
                            println!(
                                "DEBUG: Skipping test context node at lines {}-{}, type: {}",
                                rel_start_pos.row + 1,
                                rel_end_pos.row + 1,
                                context_node.kind()
                            );
                        }
                    } else {
                        // Create a merged block
                        let merged_start_row = std::cmp::min(start_pos.row, rel_start_pos.row);
                        let merged_end_row = std::cmp::max(end_pos.row, rel_end_pos.row);
                        let merged_start_byte =
                            std::cmp::min(target_node.start_byte(), context_node.start_byte());
                        let merged_end_byte =
                            std::cmp::max(target_node.end_byte(), context_node.end_byte());
                        let merged_node_type = context_node.kind().to_string();

                        seen_nodes.insert(rel_key); // Mark context as seen too

                        code_blocks.push(CodeBlock {
                            start_row: merged_start_row,
                            end_row: merged_end_row,
                            start_byte: merged_start_byte,
                            end_byte: merged_end_byte,
                            node_type: merged_node_type.clone(),
                            parent_node_type: None, // Keep consistent with original logic here
                            parent_start_row: None,
                            parent_end_row: None,
                        });

                        if debug_mode {
                            println!(
                                "DEBUG: Added merged block (comment + context) at lines {}-{}, type: {}",
                                merged_start_row + 1,
                                merged_end_row + 1,
                                merged_node_type
                            );
                        }
                        continue; // Skip adding individual comment
                    }
                }

                // Add individual comment if not merged
                code_blocks.push(CodeBlock {
                    start_row: start_pos.row,
                    end_row: end_pos.row,
                    start_byte: target_node.start_byte(),
                    end_byte: target_node.end_byte(),
                    node_type: target_node.kind().to_string(),
                    parent_node_type: None,
                    parent_start_row: None,
                    parent_end_row: None,
                });
                if debug_mode {
                    println!(
                        "DEBUG: Added individual comment block at lines {}-{}",
                        start_pos.row + 1,
                        end_pos.row + 1
                    );
                }
                continue; // Skip rest for comments
            }

            // Skip test nodes (using live check)
            if info.is_test {
                // is_test flag was set during process_node
                if debug_mode {
                    println!(
                        "DEBUG: Skipping test node at lines {}-{}",
                        start_pos.row + 1,
                        end_pos.row + 1
                    );
                }
                continue;
            }

            // Check if line is within an existing block (this check might be redundant with seen_nodes)
            // Keep consistent with original logic if it was there
            let mut existing_block = false;
            for block in &code_blocks {
                if line > block.start_row + 1 && line <= block.end_row + 1 {
                    if debug_mode {
                        println!(
                            "DEBUG: Line {} is within existing block: type='{}', lines={}-{}",
                            line,
                            block.node_type,
                            block.start_row + 1,
                            block.end_row + 1
                        );
                    }
                    existing_block = true;
                    break;
                }
            }
            if existing_block {
                continue;
            }

            // Check context node (acceptable ancestor)
            if let Some(context_node) = info.context_node {
                // context_node was set during process_node
                let rel_start_pos = context_node.start_position();
                let rel_end_pos = context_node.end_position();
                let rel_key = (rel_start_pos.row, rel_end_pos.row);

                // Ensure content is available if needed by is_test_node
                if !allow_tests && language_impl.is_test_node(&context_node, content.as_bytes()) {
                    if debug_mode {
                        println!(
                            "DEBUG: Skipping test context node (ancestor) at lines {}-{}",
                            rel_start_pos.row + 1,
                            rel_end_pos.row + 1
                        );
                    }
                } else {
                    if debug_mode {
                        println!(
                            "DEBUG: Using context node (ancestor) at lines {}-{}",
                            rel_start_pos.row + 1,
                            rel_end_pos.row + 1
                        );
                    }
                    seen_nodes.insert(rel_key); // Mark context as seen

                    // Get parent function info if applicable (e.g., for struct_type nodes)
                    let parent_info = if context_node.kind() == "struct_type" {
                        language_impl
                            .find_parent_function(context_node)
                            .map(|parent_node| {
                                let parent_type = parent_node.kind().to_string();
                                let parent_start = parent_node.start_position().row;
                                let parent_end = parent_node.end_position().row;
                                (parent_type, parent_start, parent_end)
                            })
                    } else {
                        None
                    };

                    code_blocks.push(CodeBlock {
                        start_row: rel_start_pos.row,
                        end_row: rel_end_pos.row,
                        start_byte: context_node.start_byte(),
                        end_byte: context_node.end_byte(),
                        node_type: context_node.kind().to_string(),
                        parent_node_type: parent_info.as_ref().map(|(t, _, _)| t.clone()),
                        parent_start_row: parent_info.as_ref().map(|(_, s, _)| *s),
                        parent_end_row: parent_info.as_ref().map(|(_, _, e)| *e),
                    });
                    continue; // Skip adding target_node if context was added
                }
            }

            // Check if target_node itself is acceptable (using live check)
            if language_impl.is_acceptable_parent(&target_node) {
                if debug_mode {
                    println!(
                        "DEBUG: Adding acceptable parent node at lines {}-{}",
                        start_pos.row + 1,
                        end_pos.row + 1
                    );
                }

                // Get parent function info if applicable (e.g., for struct_type nodes)
                let parent_info = if target_node.kind() == "struct_type" {
                    language_impl
                        .find_parent_function(target_node)
                        .map(|parent_node| {
                            let parent_type = parent_node.kind().to_string();
                            let parent_start = parent_node.start_position().row;
                            let parent_end = parent_node.end_position().row;
                            (parent_type, parent_start, parent_end)
                        })
                } else {
                    None
                };

                code_blocks.push(CodeBlock {
                    start_row: start_pos.row,
                    end_row: end_pos.row,
                    start_byte: target_node.start_byte(),
                    end_byte: target_node.end_byte(),
                    node_type: target_node.kind().to_string(),
                    parent_node_type: parent_info.as_ref().map(|(t, _, _)| t.clone()),
                    parent_start_row: parent_info.as_ref().map(|(_, s, _)| *s),
                    parent_end_row: parent_info.as_ref().map(|(_, _, e)| *e),
                });
                continue; // Skip fallback if acceptable parent added
            }

            // Fallback: Add the node found for the line if no context/acceptable parent logic applied
            if debug_mode {
                println!(
                    "DEBUG: Adding node via fallback at lines {}-{}",
                    start_pos.row + 1,
                    end_pos.row + 1
                );
            }

            // Get parent function info if applicable (e.g., for struct_type nodes)
            let parent_info = if target_node.kind() == "struct_type" {
                language_impl
                    .find_parent_function(target_node)
                    .map(|parent_node| {
                        let parent_type = parent_node.kind().to_string();
                        let parent_start = parent_node.start_position().row;
                        let parent_end = parent_node.end_position().row;
                        (parent_type, parent_start, parent_end)
                    })
            } else {
                None
            };

            code_blocks.push(CodeBlock {
                start_row: start_pos.row,
                end_row: end_pos.row,
                start_byte: target_node.start_byte(),
                end_byte: target_node.end_byte(),
                node_type: target_node.kind().to_string(),
                parent_node_type: parent_info.as_ref().map(|(t, _, _)| t.clone()),
                parent_start_row: parent_info.as_ref().map(|(_, s, _)| *s),
                parent_end_row: parent_info.as_ref().map(|(_, _, e)| *e),
            });
        } else if debug_mode {
            println!("DEBUG: No node info found for line {line} (Live NodeInfo)");
        }
    } // End loop over line_numbers

    // Sort and deduplicate the blocks generated from live data
    code_blocks.sort_by_key(|block| block.start_row);

    // Apply the improved deduplication logic
    let mut final_code_blocks: Vec<CodeBlock> = Vec::new();

    // Add comments first
    for block in code_blocks
        .iter()
        .filter(|b| b.node_type.contains("comment") || b.node_type == "/*" || b.node_type == "*/")
    {
        final_code_blocks.push(block.clone());
    }

    // Add non-comments, using the improved deduplication logic
    for block in code_blocks
        .iter()
        .filter(|b| !b.node_type.contains("comment") && b.node_type != "/*" && b.node_type != "*/")
    {
        let mut should_add = true;
        let mut blocks_to_remove: Vec<usize> = Vec::new();

        // Define important block types that should be preserved
        let important_block_types = [
            "function_declaration",
            "method_declaration",
            "function_item",
            "impl_item",
            "type_declaration",
            "struct_item",
            "block_comment",
        ];
        let is_important = important_block_types.contains(&block.node_type.as_str());

        // Check if this block overlaps with any of the previous blocks
        for (idx, prev_block) in final_code_blocks.iter().enumerate() {
            if prev_block.node_type.contains("comment")
                || prev_block.node_type == "/*"
                || prev_block.node_type == "*/"
            {
                continue; // Skip comments
            }

            let prev_is_important = important_block_types.contains(&prev_block.node_type.as_str());

            // Check if blocks overlap
            if block.start_row <= prev_block.end_row && block.end_row >= prev_block.start_row {
                // Case 1: Current block is contained within previous block
                if block.start_row >= prev_block.start_row && block.end_row <= prev_block.end_row {
                    if debug_mode {
                        println!(
                            "DEBUG: Current block is contained within previous block: type='{}', lines={}-{} (contained in type='{}', lines={}-{})",
                            block.node_type, block.start_row + 1, block.end_row + 1,
                            prev_block.node_type, prev_block.start_row + 1, prev_block.end_row + 1
                        );
                    }

                    // If current block is important and previous block is not, keep both
                    if is_important && !prev_is_important {
                        if debug_mode {
                            println!(
                                "DEBUG: Keeping important block type: {node_type}",
                                node_type = block.node_type
                            );
                        }
                        // Don't remove any blocks, don't set should_add to false
                    }
                    // If previous block is important and current block is not, skip current block
                    else if !is_important && prev_is_important {
                        if debug_mode {
                            println!("DEBUG: Skipping non-important block in favor of important block: {node_type}", node_type = prev_block.node_type);
                        }
                        should_add = false;
                        break;
                    }
                    // Otherwise, prefer the more specific (contained) block
                    else {
                        blocks_to_remove.push(idx);
                    }
                }
                // Case 2: Previous block is contained within current block
                else if prev_block.start_row >= block.start_row
                    && prev_block.end_row <= block.end_row
                {
                    if debug_mode {
                        println!(
                            "DEBUG: Previous block is contained within current block: type='{}', lines={}-{} (contains type='{}', lines={}-{})",
                            block.node_type, block.start_row + 1, block.end_row + 1,
                            prev_block.node_type, prev_block.start_row + 1, prev_block.end_row + 1
                        );
                    }

                    // If current block is important and previous block is not, keep both
                    if is_important && !prev_is_important {
                        if debug_mode {
                            println!(
                                "DEBUG: Keeping important block type: {node_type}",
                                node_type = block.node_type
                            );
                        }
                        // Don't set should_add to false, continue checking other blocks
                    }
                    // If previous block is important and current block is not, skip current block
                    else if !is_important && prev_is_important {
                        if debug_mode {
                            println!("DEBUG: Skipping non-important block in favor of important block: {node_type}", node_type = prev_block.node_type);
                        }
                        should_add = false;
                        break;
                    }
                    // Otherwise, skip current block as it's less specific
                    else {
                        should_add = false;
                        break;
                    }
                }
                // Case 3: Blocks partially overlap
                else {
                    if debug_mode {
                        println!(
                            "DEBUG: Blocks partially overlap: type='{}', lines={}-{} (overlaps with type='{}', lines={}-{})",
                            block.node_type, block.start_row + 1, block.end_row + 1,
                            prev_block.node_type, prev_block.start_row + 1, prev_block.end_row + 1
                        );
                    }
                    // Skip current block in case of partial overlap
                    should_add = false;
                    break;
                }
            }
        }

        // Remove any blocks that should be replaced
        for idx in blocks_to_remove.iter().rev() {
            final_code_blocks.remove(*idx);
        }

        if should_add {
            final_code_blocks.push(block.clone());
        }
    }

    // Final sort to maintain correct order
    final_code_blocks.sort_by_key(|block| block.start_row);

    // ====================================================================
    // END: Inserted Original Block Processing Logic (Cache Miss Path)
    // ====================================================================

    // Convert the original line_map to a cacheable format with representative node info
    let cacheable_line_map: Vec<Option<CachedNodeInfo>> = line_map
        .iter()
        .map(|opt_node_info| {
            opt_node_info.map(|node_info| {
                CachedNodeInfo::from_node_info(
                    &node_info,
                    language_impl.as_ref(),
                    content.as_bytes(),
                    allow_tests,
                )
            })
        })
        .collect();

    // Store the cacheable version in the cache (as you already have)
    LINE_MAP_CACHE.insert(cache_key.clone(), cacheable_line_map);
    if debug_mode {
        println!("DEBUG: Stored generated line_map in cache key: {cache_key}");
    }

    // Return the blocks generated from the LIVE data in this cache miss path
    Ok(final_code_blocks)
}