use anyhow::Result; use magic::{Cookie, CookieFlags}; use rusqlite::Connection; use std::io; use crate::meta_plugin::MetaPlugin; #[derive(Debug)] pub struct MagicFileMetaPlugin { buffer: Vec, max_buffer_size: usize, is_saved: bool, item_id: Option, cookie: Option, outputs: std::collections::HashMap, } impl MagicFileMetaPlugin { pub fn new( options: Option>, outputs: Option>, ) -> MagicFileMetaPlugin { // Start with default options let mut final_options = Self::new_simple().default_options(); if let Some(opts) = options { for (key, value) in opts { final_options.insert(key, value); } } // Start with default outputs let mut final_outputs = std::collections::HashMap::new(); let default_outputs = Self::new_simple().default_outputs(); for output_name in default_outputs { final_outputs.insert(output_name.clone(), serde_yaml::Value::String(output_name)); } if let Some(outs) = outputs { for (key, value) in outs { final_outputs.insert(key, value); } } let max_buffer_size = final_options.get("max_buffer_size") .and_then(|v| v.as_u64()) .unwrap_or(4096) as usize; MagicFileMetaPlugin { buffer: Vec::new(), max_buffer_size, is_saved: false, item_id: None, cookie: None, outputs: final_outputs, } } pub fn new_simple() -> MagicFileMetaPlugin { Self::new(None, None) } fn get_magic_result(&self, flags: CookieFlags) -> io::Result { // Use the existing cookie and just change flags if let Some(cookie) = &self.cookie { cookie.set_flags(flags) .map_err(|e| io::Error::new(io::ErrorKind::Other, format!("Failed to set magic flags: {}", e)))?; let result = cookie.buffer(&self.buffer) .map_err(|e| io::Error::new(io::ErrorKind::Other, format!("Failed to analyze buffer: {}", e)))?; // Clean up the result - remove extra whitespace and take first part if needed let trimmed = result.trim(); // For some magic results, we might want just the first part before semicolon or comma let cleaned = if trimmed.contains(';') { trimmed.split(';').next().unwrap_or(trimmed).trim() } else if trimmed.contains(',') && flags.contains(CookieFlags::MIME_TYPE | CookieFlags::MIME_ENCODING) { trimmed.split(',').next().unwrap_or(trimmed).trim() } else { trimmed }; Ok(cleaned.to_string()) } else { Err(io::Error::new(io::ErrorKind::Other, "Magic cookie not initialized")) } } fn save_all_magic_metadata(&mut self, conn: &Connection) -> Result<()> { if let Some(item_id) = self.item_id { // Save all three magic outputs: mime_type, mime_encoding, and file_type if let Ok(mime_type) = self.get_magic_result(CookieFlags::MIME_TYPE) { if !mime_type.is_empty() { let _ = self.save_meta(conn, item_id, "mime_type", mime_type); } } if let Ok(mime_encoding) = self.get_magic_result(CookieFlags::MIME_ENCODING) { if !mime_encoding.is_empty() { let _ = self.save_meta(conn, item_id, "mime_encoding", mime_encoding); } } if let Ok(file_type) = self.get_magic_result(CookieFlags::default()) { if !file_type.is_empty() { let _ = self.save_meta(conn, item_id, "file_type", file_type); } } self.is_saved = true; } Ok(()) } } impl MetaPlugin for MagicFileMetaPlugin { fn is_internal(&self) -> bool { true } fn initialize(&mut self, _conn: &Connection, item_id: i64) -> Result<()> { self.item_id = Some(item_id); // Initialize the magic cookie once let cookie = Cookie::open(Default::default()) .map_err(|e| anyhow::anyhow!("Failed to open magic cookie: {}", e))?; cookie.load(&[] as &[&str]) .map_err(|e| anyhow::anyhow!("Failed to load magic database: {}", e))?; self.cookie = Some(cookie); Ok(()) } fn finalize(&mut self, conn: &Connection) -> Result<()> { // Save all magic metadata if not already saved if !self.is_saved { if let Err(e) = self.save_all_magic_metadata(conn) { eprintln!("Warning: Failed to save magic metadata: {}", e); } } Ok(()) } fn update(&mut self, data: &[u8], conn: &Connection) { // Only collect up to max_buffer_size let remaining_capacity = self.max_buffer_size.saturating_sub(self.buffer.len()); if remaining_capacity > 0 { let bytes_to_copy = std::cmp::min(data.len(), remaining_capacity); self.buffer.extend_from_slice(&data[..bytes_to_copy]); // Check if we've reached our buffer limit and save if so if self.buffer.len() >= self.max_buffer_size && !self.is_saved { if let Err(e) = self.save_all_magic_metadata(conn) { eprintln!("Warning: Failed to save magic metadata early: {}", e); } } } } fn meta_name(&self) -> String { "magic_file".to_string() } fn configure_options(&mut self, options: &std::collections::HashMap) -> Result<()> { if let Some(max_buffer_size) = options.get("max_buffer_size") { if let Some(size) = max_buffer_size.as_u64() { self.max_buffer_size = size as usize; } } Ok(()) } fn outputs(&self) -> &std::collections::HashMap { &self.outputs } fn outputs_mut(&mut self) -> &mut std::collections::HashMap { &mut self.outputs } fn default_outputs(&self) -> Vec { vec!["mime_type".to_string(), "mime_encoding".to_string(), "file_type".to_string()] } fn default_options(&self) -> std::collections::HashMap { let mut options = std::collections::HashMap::new(); options.insert("max_buffer_size".to_string(), serde_yaml::Value::Number(4096.into())); options } }