car-ast 0.15.0

//! Project-wide symbol index — scans a directory, parses all supported files,
//! and builds a cross-file symbol table with caller/callee references.

use std::collections::HashMap;
use std::fs::File;
use std::io::Read;
use std::path::{Path, PathBuf};
use std::time::{Duration, Instant};

use ignore::WalkBuilder;

use crate::types::*;
use crate::{extract_source, parse_file};

/// A symbol with its file context.
#[derive(Debug, Clone)]
pub struct IndexedSymbol {
    /// Relative path from project root.
    pub file_path: String,
    pub symbol: Symbol,
    /// Source code of this symbol.
    pub source: String,
}

/// Cross-file reference: symbol A in file X references symbol B.
#[derive(Debug, Clone)]
pub struct CrossReference {
    /// The symbol doing the referencing.
    pub from_file: String,
    pub from_symbol: String,
    /// The symbol being referenced.
    pub to_file: String,
    pub to_symbol: String,
}

/// Project-wide symbol index.
#[derive(Debug)]
pub struct ProjectIndex {
    /// All parsed files, keyed by relative path.
    pub files: HashMap<String, ParsedFile>,
    /// All symbols flattened, keyed by `file_path::name`.
    pub symbols: HashMap<String, IndexedSymbol>,
    /// Cross-file references (A calls/uses B).
    pub references: Vec<CrossReference>,
    /// Project root directory.
    pub root: PathBuf,
}

/// File extensions to scan.
const EXTENSIONS: &[&str] = &["rs", "py", "ts", "tsx", "js", "jsx", "mjs", "cjs", "go"];

const MAX_FILES: usize = 5000;
const MAX_WALL: Duration = Duration::from_secs(5);
const MAX_SOURCE_BYTES: usize = 1024 * 1024;

/// Directories to skip.
const SKIP_DIRS: &[&str] = &[
    ".cargo",
    ".git",
    ".next",
    "__pycache__",
    "bin",
    "build",
    "dist",
    "node_modules",
    "obj",
    "target",
    "vendor",
];

impl ProjectIndex {
    /// Build an index by scanning a directory.
    pub fn build(root: &Path) -> Self {
        let start = Instant::now();
        let mut index = Self {
            files: HashMap::new(),
            symbols: HashMap::new(),
            references: Vec::new(),
            root: root.to_path_buf(),
        };
        let mut truncated = false;

        // Collect source files
        let files = collect_source_files(root, start, &mut truncated);

        // Parse each file
        for file_path in &files {
            if start.elapsed() >= MAX_WALL {
                truncated = true;
                tracing::warn!(
                    reason = "wall_clock",
                    phase = "parse",
                    root = %root.display(),
                    partial_files = index.files.len(),
                    cross_references_dropped = true,
                    "ProjectIndex::build exceeded budget during parse phase",
                );
                break;
            }

            let rel_path = file_path
                .strip_prefix(root)
                .unwrap_or(file_path)
                .to_string_lossy()
                .to_string();

            let content = match read_source_file_bounded(file_path) {
                SourceRead::Content(content) => content,
                SourceRead::Oversized { bytes } => {
                    truncated = true;
                    tracing::warn!(
                        reason = "file_size",
                        phase = "parse",
                        root = %root.display(),
                        file = %file_path.display(),
                        file_bytes = bytes,
                        max_file_bytes = MAX_SOURCE_BYTES,
                        partial_files = index.files.len(),
                        cross_references_dropped = true,
                        "ProjectIndex::build skipped oversized source file during parse phase",
                    );
                    continue;
                }
                SourceRead::Unreadable => continue,
            };

            if let Some(parsed) = parse_file(&content, &rel_path) {
                // Index all symbols (top-level + children)
                for sym in &parsed.symbols {
                    let key = format!("{}::{}", rel_path, sym.name);
                    index.symbols.insert(
                        key,
                        IndexedSymbol {
                            file_path: rel_path.clone(),
                            symbol: sym.clone(),
                            source: extract_source(sym, &content),
                        },
                    );

                    // Index children (methods)
                    for child in &sym.children {
                        let child_key = format!("{}::{}::{}", rel_path, sym.name, child.name);
                        index.symbols.insert(
                            child_key,
                            IndexedSymbol {
                                file_path: rel_path.clone(),
                                symbol: child.clone(),
                                source: extract_source(child, &content),
                            },
                        );
                    }
                }

                index.files.insert(rel_path, parsed);
            }

            if start.elapsed() >= MAX_WALL {
                truncated = true;
                tracing::warn!(
                    reason = "wall_clock",
                    phase = "parse",
                    root = %root.display(),
                    partial_files = index.files.len(),
                    cross_references_dropped = true,
                    "ProjectIndex::build exceeded budget during parse phase",
                );
                break;
            }
        }

        // Build cross-file references
        if !truncated {
            truncated = index.build_cross_references(&files, root, start);
        }
        if truncated {
            index.references.clear();
            tracing::warn!(
                reason = "truncated",
                phase = "cross_reference",
                root = %root.display(),
                partial_files = index.files.len(),
                cross_references_dropped = true,
                "ProjectIndex::build truncated; skipping cross-reference output",
            );
        }

        index
    }

    /// Find all symbols matching a name (exact).
    pub fn find(&self, name: &str) -> Vec<&IndexedSymbol> {
        self.symbols
            .values()
            .filter(|s| s.symbol.name == name)
            .collect()
    }

    /// Find all symbols matching a name (fuzzy, case-insensitive substring).
    pub fn find_fuzzy(&self, query: &str) -> Vec<&IndexedSymbol> {
        let query_lower = query.to_lowercase();
        self.symbols
            .values()
            .filter(|s| s.symbol.name.to_lowercase().contains(&query_lower))
            .collect()
    }

    /// Get all callers of a symbol (symbols that reference it).
    pub fn callers_of(&self, symbol_name: &str) -> Vec<&CrossReference> {
        self.references
            .iter()
            .filter(|r| r.to_symbol == symbol_name)
            .collect()
    }

    /// Get all callees of a symbol (symbols it references).
    pub fn callees_of(&self, symbol_name: &str) -> Vec<&CrossReference> {
        self.references
            .iter()
            .filter(|r| r.from_symbol == symbol_name)
            .collect()
    }

    /// Build cross-file references by scanning for identifier usage.
    fn build_cross_references(&mut self, files: &[PathBuf], root: &Path, start: Instant) -> bool {
        // Collect all symbol names and their defining files
        let symbol_names: Vec<(String, String)> = self
            .symbols
            .values()
            .filter(|s| !matches!(s.symbol.kind, SymbolKind::Import | SymbolKind::Const))
            .map(|s| (s.symbol.name.clone(), s.file_path.clone()))
            .collect();

        // For each file, scan for references to symbols defined in other files
        for file_path in files {
            if cross_reference_budget_exceeded(start, root, self.files.len(), self.references.len())
            {
                return true;
            }

            let rel_path = file_path
                .strip_prefix(root)
                .unwrap_or(file_path)
                .to_string_lossy()
                .to_string();

            let content = match read_source_file_bounded(file_path) {
                SourceRead::Content(content) => content,
                SourceRead::Oversized { bytes } => {
                    tracing::warn!(
                        reason = "file_size",
                        phase = "cross_reference",
                        root = %root.display(),
                        file = %file_path.display(),
                        file_bytes = bytes,
                        max_file_bytes = MAX_SOURCE_BYTES,
                        partial_files = self.files.len(),
                        partial_references = self.references.len(),
                        cross_references_dropped = true,
                        "ProjectIndex::build skipped oversized source file during cross-reference phase",
                    );
                    return true;
                }
                SourceRead::Unreadable => continue,
            };
            if cross_reference_budget_exceeded(start, root, self.files.len(), self.references.len())
            {
                return true;
            }

            // Get symbols defined in this file (to find which symbol contains the reference)
            let file_symbols: Vec<&IndexedSymbol> = self
                .symbols
                .values()
                .filter(|s| s.file_path == rel_path)
                .collect();

            for (target_name, target_file) in &symbol_names {
                if cross_reference_budget_exceeded(
                    start,
                    root,
                    self.files.len(),
                    self.references.len(),
                ) {
                    return true;
                }

                // Skip self-references within the same file
                if *target_file == rel_path {
                    continue;
                }
                // Skip very short names (too many false positives)
                if target_name.len() < 3 {
                    continue;
                }

                // Check if this file contains the target symbol name as a whole word
                let has_reference = match contains_identifier_bounded(&content, target_name, start)
                {
                    IdentifierSearch::Found(found) => found,
                    IdentifierSearch::BudgetExceeded => {
                        warn_cross_reference_budget_exceeded(
                            root,
                            self.files.len(),
                            self.references.len(),
                        );
                        return true;
                    }
                };
                if !has_reference {
                    continue;
                }

                // Find which symbol in this file contains the reference
                let mut referencing_symbol = None;
                for symbol in &file_symbols {
                    match contains_identifier_bounded(&symbol.source, target_name, start) {
                        IdentifierSearch::Found(true) => {
                            referencing_symbol = Some(symbol.symbol.name.clone());
                            break;
                        }
                        IdentifierSearch::Found(false) => {}
                        IdentifierSearch::BudgetExceeded => {
                            warn_cross_reference_budget_exceeded(
                                root,
                                self.files.len(),
                                self.references.len(),
                            );
                            return true;
                        }
                    }
                }
                let referencing_symbol = referencing_symbol.unwrap_or_else(|| rel_path.clone());
                if cross_reference_budget_exceeded(
                    start,
                    root,
                    self.files.len(),
                    self.references.len(),
                ) {
                    return true;
                }

                self.references.push(CrossReference {
                    from_file: rel_path.clone(),
                    from_symbol: referencing_symbol,
                    to_file: target_file.clone(),
                    to_symbol: target_name.clone(),
                });

                if cross_reference_budget_exceeded(
                    start,
                    root,
                    self.files.len(),
                    self.references.len(),
                ) {
                    return true;
                }
            }
        }

        false
    }

    /// Summary statistics.
    pub fn stats(&self) -> IndexStats {
        IndexStats {
            files: self.files.len(),
            symbols: self.symbols.len(),
            references: self.references.len(),
        }
    }
}

fn cross_reference_budget_exceeded(
    start: Instant,
    root: &Path,
    partial_files: usize,
    partial_references: usize,
) -> bool {
    if start.elapsed() < MAX_WALL {
        return false;
    }

    warn_cross_reference_budget_exceeded(root, partial_files, partial_references);
    true
}

fn warn_cross_reference_budget_exceeded(
    root: &Path,
    partial_files: usize,
    partial_references: usize,
) {
    tracing::warn!(
        reason = "wall_clock",
        phase = "cross_reference",
        root = %root.display(),
        partial_files,
        partial_references,
        cross_references_dropped = true,
        "ProjectIndex::build exceeded budget during cross-reference phase",
    );
}

#[derive(Debug, Clone)]
pub struct IndexStats {
    pub files: usize,
    pub symbols: usize,
    pub references: usize,
}

impl std::fmt::Display for IndexStats {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        write!(
            f,
            "{} files, {} symbols, {} cross-references",
            self.files, self.symbols, self.references
        )
    }
}

/// Collect source files from a directory with ignore-aware filtering.
fn collect_source_files(root: &Path, start: Instant, truncated: &mut bool) -> Vec<PathBuf> {
    let mut files = Vec::new();

    let root_path = root.to_path_buf();
    let walker = WalkBuilder::new(root)
        .standard_filters(true)
        .require_git(false)
        .filter_entry(move |entry| {
            let is_dir = entry
                .file_type()
                .map(|file_type| file_type.is_dir())
                .unwrap_or(false);
            if !is_dir || entry.path() == root_path {
                return true;
            }

            entry
                .file_name()
                .to_str()
                .map(|name| !SKIP_DIRS.contains(&name))
                .unwrap_or(true)
        })
        .build();

    for entry in walker {
        if start.elapsed() >= MAX_WALL {
            *truncated = true;
            tracing::warn!(
                reason = "wall_clock",
                phase = "walk",
                root = %root.display(),
                partial_files = files.len(),
                cross_references_dropped = true,
                "ProjectIndex::build exceeded budget during walk phase",
            );
            break;
        }

        let entry = match entry {
            Ok(entry) => entry,
            Err(_) => continue,
        };

        let is_file = entry
            .file_type()
            .map(|file_type| file_type.is_file())
            .unwrap_or(false);
        if !is_file {
            continue;
        }

        let ext = entry
            .path()
            .extension()
            .and_then(|e| e.to_str())
            .unwrap_or("");
        if EXTENSIONS.contains(&ext) {
            if files.len() >= MAX_FILES {
                *truncated = true;
                tracing::warn!(
                    reason = "file_count",
                    phase = "walk",
                    root = %root.display(),
                    partial_files = files.len(),
                    cross_references_dropped = true,
                    "ProjectIndex::build exceeded file budget during walk phase",
                );
                break;
            }
            files.push(entry.into_path());
        }
    }

    files
}

enum SourceRead {
    Content(String),
    Oversized { bytes: u64 },
    Unreadable,
}

fn read_source_file_bounded(path: &Path) -> SourceRead {
    let file = match File::open(path) {
        Ok(file) => file,
        Err(_) => return SourceRead::Unreadable,
    };

    if let Ok(metadata) = file.metadata() {
        if metadata.len() > MAX_SOURCE_BYTES as u64 {
            return SourceRead::Oversized {
                bytes: metadata.len(),
            };
        }
    }

    let mut bytes = Vec::new();
    let mut limited = file.take(MAX_SOURCE_BYTES as u64 + 1);
    if limited.read_to_end(&mut bytes).is_err() {
        return SourceRead::Unreadable;
    }

    if bytes.len() > MAX_SOURCE_BYTES {
        return SourceRead::Oversized {
            bytes: bytes.len() as u64,
        };
    }

    match String::from_utf8(bytes) {
        Ok(content) => SourceRead::Content(content),
        Err(_) => SourceRead::Unreadable,
    }
}

enum IdentifierSearch {
    Found(bool),
    BudgetExceeded,
}

/// Check if `source` contains `name` as a whole identifier (not a substring).
#[cfg(test)]
fn contains_identifier(source: &str, name: &str) -> bool {
    match contains_identifier_bounded(source, name, Instant::now()) {
        IdentifierSearch::Found(found) => found,
        IdentifierSearch::BudgetExceeded => false,
    }
}

fn contains_identifier_bounded(source: &str, name: &str, start: Instant) -> IdentifierSearch {
    let bytes = source.as_bytes();
    let name_bytes = name.as_bytes();
    let mut pos = 0;

    while pos + name_bytes.len() <= bytes.len() {
        if start.elapsed() >= MAX_WALL {
            return IdentifierSearch::BudgetExceeded;
        }

        if let Some(found) = source[pos..].find(name) {
            let abs = pos + found;
            let before = if abs > 0 { bytes[abs - 1] } else { b' ' };
            let after_pos = abs + name_bytes.len();
            let after = if after_pos < bytes.len() {
                bytes[after_pos]
            } else {
                b' '
            };

            if !is_ident_char(before) && !is_ident_char(after) {
                return IdentifierSearch::Found(true);
            }
            pos = abs + 1;
        } else {
            break;
        }
    }
    IdentifierSearch::Found(false)
}

fn is_ident_char(b: u8) -> bool {
    b.is_ascii_alphanumeric() || b == b'_'
}

#[cfg(test)]
mod tests {
    use super::*;
    use std::fs;
    #[cfg(unix)]
    use std::os::unix::fs::PermissionsExt;
    use tracing_test::traced_test;

    fn write_file(path: &Path, content: impl AsRef<str>) {
        if let Some(parent) = path.parent() {
            fs::create_dir_all(parent).unwrap();
        }
        fs::write(path, content.as_ref()).unwrap();
    }

    fn rust_fn(name: &str) -> String {
        format!("pub fn {}() {{}}\n", name)
    }

    fn rel(parts: &[&str]) -> String {
        let sep = std::path::MAIN_SEPARATOR.to_string();
        parts.join(&sep)
    }

    #[test]
    fn test_project_index() {
        let dir = tempfile::tempdir().unwrap();

        // Create a small Rust project
        fs::write(
            dir.path().join("lib.rs"),
            r#"
pub struct Parser {
    lang: String,
}

impl Parser {
    pub fn new(lang: String) -> Self {
        Self { lang }
    }

    pub fn parse(&self, source: &str) -> Vec<String> {
        vec![]
    }
}

pub fn create_parser() -> Parser {
    Parser::new("rust".into())
}
"#,
        )
        .unwrap();

        fs::write(
            dir.path().join("main.rs"),
            r#"
mod lib;

fn main() {
    let p = create_parser();
    let result = p.parse("fn foo() {}");
    println!("{:?}", result);
}
"#,
        )
        .unwrap();

        let index = ProjectIndex::build(dir.path());
        let stats = index.stats();

        assert_eq!(stats.files, 2);
        assert!(stats.symbols >= 4); // Parser, new, parse, create_parser, main

        // Find Parser
        let parsers = index.find("Parser");
        assert!(!parsers.is_empty());

        // Find cross-references: main.rs should reference create_parser from lib.rs
        let refs = index.callers_of("create_parser");
        assert!(!refs.is_empty(), "main.rs should reference create_parser");
        assert!(refs.iter().any(|r| r.from_file == "main.rs"));

        // Fuzzy search
        let fuzzy = index.find_fuzzy("pars");
        assert!(fuzzy.len() >= 2); // Parser, parse, create_parser
    }

    #[test]
    fn test_contains_identifier() {
        assert!(contains_identifier("let x = foo();", "foo"));
        assert!(!contains_identifier("let x = foobar();", "foo"));
        assert!(contains_identifier("use crate::Parser;", "Parser"));
        assert!(!contains_identifier("use crate::ParserBuilder;", "Parser"));
    }

    #[test]
    fn test_walker_honors_gitignore_outside_git_repo() {
        let dir = tempfile::tempdir().unwrap();
        write_file(&dir.path().join(".gitignore"), "ignored-by-gitignore/\n");
        write_file(&dir.path().join("src").join("lib.rs"), rust_fn("visible"));
        write_file(
            &dir.path().join("generated-local").join("generated.ts"),
            "export function includedGenerated() {}\n",
        );
        write_file(
            &dir.path().join("ignored-by-gitignore").join("generated.ts"),
            "export function generated() {}\n",
        );

        let index = ProjectIndex::build(dir.path());

        assert!(index.files.contains_key(&rel(&["src", "lib.rs"])));
        assert!(index
            .files
            .contains_key(&rel(&["generated-local", "generated.ts"])));
        assert!(!index
            .files
            .contains_key(&rel(&["ignored-by-gitignore", "generated.ts"])));
    }

    #[test]
    fn test_walker_honors_ignore_file() {
        let dir = tempfile::tempdir().unwrap();
        write_file(&dir.path().join(".ignore"), "ignored.rs\n");
        write_file(&dir.path().join("visible.rs"), rust_fn("visible"));
        write_file(&dir.path().join("ignored.rs"), rust_fn("ignored"));

        let index = ProjectIndex::build(dir.path());

        assert!(index.files.contains_key("visible.rs"));
        assert!(!index.files.contains_key("ignored.rs"));
    }

    #[test]
    fn test_walker_honors_git_info_exclude_without_global_config() {
        let dir = tempfile::tempdir().unwrap();
        write_file(
            &dir.path().join(".git").join("info").join("exclude"),
            "excluded.rs\n",
        );
        write_file(&dir.path().join("visible.rs"), rust_fn("visible"));
        write_file(&dir.path().join("excluded.rs"), rust_fn("excluded"));

        let index = ProjectIndex::build(dir.path());

        assert!(index.files.contains_key("visible.rs"));
        assert!(!index.files.contains_key("excluded.rs"));
    }

    #[test]
    fn test_walker_skips_hidden_paths() {
        let dir = tempfile::tempdir().unwrap();
        write_file(&dir.path().join("visible.rs"), rust_fn("visible"));
        write_file(&dir.path().join(".hidden.rs"), rust_fn("hidden_file"));
        write_file(
            &dir.path().join(".hidden_dir").join("secret.rs"),
            rust_fn("hidden_dir"),
        );

        let index = ProjectIndex::build(dir.path());

        assert!(index.files.contains_key("visible.rs"));
        assert!(!index.files.contains_key(".hidden.rs"));
        assert!(!index
            .files
            .contains_key(&rel(&[".hidden_dir", "secret.rs"])));
    }

    #[test]
    fn test_walker_prunes_skip_dirs() {
        let dir = tempfile::tempdir().unwrap();
        write_file(&dir.path().join("src").join("lib.rs"), rust_fn("visible"));
        write_file(
            &dir.path().join("node_modules").join("pkg").join("index.js"),
            "export function ignored() {}\n",
        );
        write_file(
            &dir.path().join("target").join("debug").join("generated.rs"),
            rust_fn("target_generated"),
        );
        write_file(&dir.path().join("bin").join("cli.rs"), rust_fn("bin_cli"));
        write_file(
            &dir.path().join("obj").join("generated.rs"),
            rust_fn("obj_generated"),
        );

        let index = ProjectIndex::build(dir.path());

        assert!(index.files.contains_key(&rel(&["src", "lib.rs"])));
        assert!(!index
            .files
            .contains_key(&rel(&["node_modules", "pkg", "index.js"])));
        assert!(!index
            .files
            .contains_key(&rel(&["target", "debug", "generated.rs"])));
        assert!(!index.files.contains_key(&rel(&["bin", "cli.rs"])));
        assert!(!index.files.contains_key(&rel(&["obj", "generated.rs"])));
    }

    #[test]
    fn test_walker_cascades_subdir_gitignore() {
        let dir = tempfile::tempdir().unwrap();
        for project in ["app_one", "app_two"] {
            let project_dir = dir.path().join(project);
            write_file(&project_dir.join(".gitignore"), "ignored-by-gitignore/\n");
            write_file(&project_dir.join("src").join("lib.rs"), rust_fn("visible"));
            write_file(
                &project_dir.join("generated-local").join("generated.rs"),
                rust_fn("included_generated"),
            );
            write_file(
                &project_dir
                    .join("ignored-by-gitignore")
                    .join("generated.rs"),
                rust_fn("generated"),
            );
        }

        let index = ProjectIndex::build(dir.path());

        assert!(index
            .files
            .contains_key(&rel(&["app_one", "src", "lib.rs"])));
        assert!(index
            .files
            .contains_key(&rel(&["app_two", "src", "lib.rs"])));
        assert!(index
            .files
            .contains_key(&rel(&["app_one", "generated-local", "generated.rs"])));
        assert!(index
            .files
            .contains_key(&rel(&["app_two", "generated-local", "generated.rs"])));
        assert!(!index.files.contains_key(&rel(&[
            "app_one",
            "ignored-by-gitignore",
            "generated.rs"
        ])));
        assert!(!index.files.contains_key(&rel(&[
            "app_two",
            "ignored-by-gitignore",
            "generated.rs"
        ])));
    }

    #[test]
    fn test_walker_includes_root_even_if_named_skip() {
        let parent = tempfile::tempdir().unwrap();
        let root = parent.path().join("node_modules");
        write_file(&root.join("index.js"), "export function visible() {}\n");
        write_file(
            &root.join("node_modules").join("nested.js"),
            "export function hidden() {}\n",
        );

        let index = ProjectIndex::build(&root);

        assert!(index.files.contains_key("index.js"));
        assert!(!index
            .files
            .contains_key(&rel(&["node_modules", "nested.js"])));
    }

    #[cfg(unix)]
    #[test]
    fn test_build_does_not_panic_on_unreadable_dir() {
        let dir = tempfile::tempdir().unwrap();
        let unreadable = dir.path().join("unreadable");
        write_file(&dir.path().join("visible.rs"), rust_fn("visible"));
        write_file(&unreadable.join("hidden.rs"), rust_fn("hidden"));
        fs::set_permissions(&unreadable, fs::Permissions::from_mode(0o000)).unwrap();

        let result = std::panic::catch_unwind(|| ProjectIndex::build(dir.path()));

        fs::set_permissions(&unreadable, fs::Permissions::from_mode(0o755)).unwrap();
        let index = result.unwrap();
        assert!(index.files.contains_key("visible.rs"));
    }

    #[test]
    #[traced_test]
    fn test_build_caps_at_max_files_and_warns() {
        let dir = tempfile::tempdir().unwrap();
        for i in 0..(MAX_FILES + 50) {
            write_file(&dir.path().join(format!("file_{i:05}.rs")), "");
        }

        let mut truncated = false;
        let files = collect_source_files(dir.path(), Instant::now(), &mut truncated);

        assert_eq!(files.len(), MAX_FILES);
        assert!(truncated);
        assert!(logs_contain("reason=\"file_count\""));
        assert!(logs_contain("phase=\"walk\""));
    }

    #[test]
    #[traced_test]
    fn test_oversized_source_truncates_warns_and_drops_xrefs() {
        let dir = tempfile::tempdir().unwrap();
        write_file(&dir.path().join("a.rs"), "pub fn target_fn() {}\n");
        write_file(
            &dir.path().join("b.rs"),
            "use crate::a::target_fn;\npub fn caller() { target_fn(); }\n",
        );
        write_file(
            &dir.path().join("oversized.rs"),
            format!(
                "pub fn oversized() {{}}\n// {}\n",
                "x".repeat(MAX_SOURCE_BYTES + 1)
            ),
        );

        let index = ProjectIndex::build(dir.path());

        assert!(index.files.contains_key("a.rs"));
        assert!(index.files.contains_key("b.rs"));
        assert!(index.references.is_empty());
        assert!(logs_contain("reason=\"file_size\""));
        assert!(logs_contain("cross_references_dropped=true"));
    }

    #[test]
    #[traced_test]
    fn test_truncated_build_emits_xref_dropped_warn() {
        let dir = tempfile::tempdir().unwrap();
        for i in 0..(MAX_FILES + 1) {
            write_file(
                &dir.path().join(format!("file_{i:05}.rs")),
                if i == 0 {
                    rust_fn("visible")
                } else {
                    String::new()
                },
            );
        }

        let index = ProjectIndex::build(dir.path());

        assert_eq!(index.files.len(), MAX_FILES);
        assert!(index.references.is_empty());
        assert!(logs_contain("reason=\"file_count\""));
        assert!(logs_contain("phase=\"walk\""));
        assert!(logs_contain("cross_references_dropped=true"));
    }
}