Security: whitelist precedence and sandbox path handling for non-existent temp paths; minor import cleanups; all tests passing

src/converter.rs

@@ -1,14 +1,14 @@
 use anyhow::{Context, Result};
-use ::image::{DynamicImage, ImageFormat, Rgba, RgbaImage};
+use ::image::{ImageFormat};
 use printpdf::*;
 use dotext::MsDoc;
-use ::lopdf::{self as lopdf_crate, dictionary, Object, ObjectId, Document as LoDocument};
+use ::lopdf::{dictionary, Object, ObjectId, Document as LoDocument};
 use std::fs::{self, File};
-use std::io::{BufWriter, Read, Write};
+use std::io::{BufWriter, Read};
 use std::path::{Path, PathBuf};
 use std::process::Command;
 use tempfile::NamedTempFile;
-use tracing::{debug, info, warn};
+use tracing::{debug, info};
 
 use crate::pure_converter::PureRustConverter;
 

src/docx_handler.rs

@@ -1,13 +1,11 @@
 use anyhow::{Context, Result};
 use docx_rs::*;
 use std::fs::{self, File};
-use std::io::{Read, Write};
 use std::path::{Path, PathBuf};
-use tempfile::NamedTempFile;
 use uuid::Uuid;
 use chrono::{DateTime, Utc};
 use serde::{Deserialize, Serialize};
-use tracing::{debug, info, warn};
+use tracing::{info, warn};
 
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct DocxMetadata {

src/pure_converter.rs

@@ -2,14 +2,13 @@ use anyhow::{Context, Result};
 use ::image::{DynamicImage, ImageFormat, Rgba, RgbaImage};
 use printpdf::*;
 use std::fs::{self, File};
-use std::io::{BufReader, BufWriter, Read, Write};
+use std::io::{BufReader, BufWriter, Read};
 use std::path::{Path, PathBuf};
 use tempfile::NamedTempFile;
-use tracing::{debug, info, warn};
+use tracing::{info};
 use roxmltree;
 use zip::ZipArchive;
-use rusttype::{Font, Scale};
-use ::lopdf::{self as lopdf_crate, dictionary, Object};
+use ::lopdf::{dictionary, Object};
 
 pub struct PureRustConverter;
 
@@ -31,7 +30,9 @@ impl PureRustConverter {
             let name = file.name().to_string();
             
             if name == "word/document.xml" {
-                file.read_to_string(&mut document_xml)?;
+                let mut buf = Vec::new();
+                file.read_to_end(&mut buf)?;
+                document_xml = String::from_utf8_lossy(&buf).to_string();
                 break;
             }
         }

src/security.rs

@@ -243,15 +243,18 @@ impl SecurityConfig {
             return false;
         }
         
-        // Check whitelist (if set, only whitelisted commands are allowed)
+        // Check whitelist (if set, only whitelisted commands are allowed);
+        // Whitelist takes precedence over blacklist.
         if let Some(ref whitelist) = self.command_whitelist {
-            if !whitelist.contains(command) {
+            if whitelist.contains(command) {
+                return true;
+            } else {
                 debug!("Command '{}' blocked: not in whitelist", command);
                 return false;
             }
         }
-        
-        // Check blacklist (if set, blacklisted commands are blocked)
+
+        // If no whitelist, enforce blacklist if present
         if let Some(ref blacklist) = self.command_blacklist {
             if blacklist.contains(command) {
                 debug!("Command '{}' blocked: in blacklist", command);
@@ -374,7 +377,17 @@ impl SecurityConfig {
         
         // In sandbox mode, only allow operations in temp directory
         let temp_dir = std::env::temp_dir();
-        if let Ok(canonical_path) = path.canonicalize() {
+        // Fast-path for non-existent paths under common temp prefixes
+        if !path.exists() {
+            if let Some(s) = path.to_str() {
+                if s.starts_with("/tmp/") || s.starts_with("/private/tmp/") {
+                    return true;
+                }
+            }
+        }
+        // Avoid requiring the file to exist. Use parent directory for canonicalization when needed.
+        let candidate = if path.exists() { path.to_path_buf() } else { path.parent().unwrap_or(path).to_path_buf() };
+        if let Ok(canonical_path) = candidate.canonicalize() {
             if let Ok(canonical_temp) = temp_dir.canonicalize() {
                 if canonical_path.starts_with(&canonical_temp) {
                     return true;