temp-deepwiki/src/discover.rs

use crate::stats::DiscoveryStats;
use crate::types::FileRecord;
use anyhow::Result;
use ignore::WalkBuilder;
use std::path::Path;
use std::time::{Instant, UNIX_EPOCH};

/// Step 1: Discovery - find all files respecting ignore patterns

const DEFAULT_IGNORES: &[&str] = &[
    ".git/**",
    "node_modules/**",
    "dist/**",
    "build/**",
    "target/**",
    "**/*.lock",
    "*-lock.json",
    "*.lock",
    ".vscode/**",
    ".idea/**",
    "__pycache__/**",
    "*.pyc",
    ".DS_Store",
];

const MAX_INDEXABLE_BYTES: u64 = 2_000_000; // 2MB

pub fn discover<P: AsRef<Path>>(
    root: P,
    verbose: bool,
) -> Result<(Vec<FileRecord>, DiscoveryStats)> {
    let start = Instant::now();
    let root = root.as_ref();

    if verbose {
        println!("[Discovery] Scanning directory: {}", root.display());
    }

    let mut files = Vec::new();
    let mut skipped = 0;
    let mut total_bytes = 0u64;

    let walker = WalkBuilder::new(root)
        .standard_filters(true) // Respects .gitignore, .ignore, etc.
        .hidden(false) // Don't skip hidden files by default
        .build();

    for entry_result in walker {
        let entry = match entry_result {
            Ok(e) => e,
            Err(e) => {
                eprintln!("Error walking directory: {}", e);
                continue;
            }
        };

        // Skip directories
        if entry.file_type().map_or(true, |ft| ft.is_dir()) {
            continue;
        }

        let path = entry.path();

        // Check against default ignores
        if should_ignore(path) {
            skipped += 1;
            continue;
        }

        let metadata = match std::fs::metadata(path) {
            Ok(m) => m,
            Err(e) => {
                eprintln!("Error reading metadata for {}: {}", path.display(), e);
                continue;
            }
        };

        let size = metadata.len();

        // Skip files that are too large
        if size > MAX_INDEXABLE_BYTES {
            if verbose {
                eprintln!(
                    "[Discovery] Skipping large file: {} ({} bytes)",
                    path.display(),
                    size
                );
            }
            skipped += 1;
            continue;
        }

        total_bytes += size;

        let modified_time = metadata
            .modified()
            .ok()
            .and_then(|t| t.duration_since(UNIX_EPOCH).ok())
            .map(|d| d.as_secs())
            .unwrap_or(0);

        // Compute fingerprint (hash of content)
        let fingerprint = match compute_fingerprint(path) {
            Ok(fp) => fp,
            Err(e) => {
                eprintln!("Error computing fingerprint for {}: {}", path.display(), e);
                continue;
            }
        };

        files.push(FileRecord {
            path: path.to_path_buf(),
            size,
            modified_time,
            fingerprint,
        });
    }

    let stats = DiscoveryStats {
        files_found: files.len(),
        files_skipped: skipped,
        total_bytes,
        duration_ms: start.elapsed().as_millis() as u64,
    };

    if verbose {
        println!(
            "[Discovery] Complete: {} files found, {} skipped, {:.2} MB total",
            files.len(),
            skipped,
            total_bytes as f64 / 1_048_576.0
        );
    }

    Ok((files, stats))
}

fn should_ignore(path: &Path) -> bool {
    let path_str = path.to_string_lossy();
    let file_name = path.file_name().and_then(|n| n.to_str()).unwrap_or("");

    for pattern in DEFAULT_IGNORES {
        if pattern.ends_with("/**") {
            let prefix = pattern.trim_end_matches("/**");
            // Check if the path contains this directory
            if path_str.contains(&format!("/{}/", prefix))
                || path_str.contains(&format!("\\{}\\", prefix))
                || path_str.contains(&format!("/{}", prefix)) // At start
                || path_str.starts_with(&format!("{}\\", prefix))
                || path_str.starts_with(&format!("{}/", prefix))
            {
                return true;
            }
        } else if pattern.starts_with("**/*.") {
            let ext = pattern.trim_start_matches("**/");
            if file_name.ends_with(ext) {
                return true;
            }
        } else if pattern.starts_with("*.") {
            if file_name.ends_with(pattern.trim_start_matches('*')) {
                return true;
            }
        } else if pattern.starts_with('*') && pattern.contains('.') {
            // Pattern like *-lock.json
            let suffix = pattern.trim_start_matches('*');
            if file_name.ends_with(suffix) {
                return true;
            }
        } else if path_str.ends_with(pattern) || file_name == *pattern {
            return true;
        }
    }

    false
}

fn compute_fingerprint(path: &Path) -> Result<String> {
    let content = std::fs::read(path)?;
    let hash = blake3::hash(&content);
    Ok(hash.to_hex()[..16].to_string()) // Use first 16 chars for brevity
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_should_ignore() {
        assert!(should_ignore(Path::new("node_modules/package/index.js")));
        assert!(should_ignore(Path::new(".git/config")));
        assert!(should_ignore(Path::new("target/debug/app.exe")));
        assert!(should_ignore(Path::new("package-lock.json")));
        assert!(!should_ignore(Path::new("src/main.rs")));
        assert!(!should_ignore(Path::new("README.md")));
    }
}