[feat] add input validation, enhanced error messages, and new integration tests

tests/integration_validation.rs

@@ -0,0 +1,125 @@
+// SPDX-License-Identifier: MIT
+// Validation and error-handling integration tests
+
+use std::fs;
+use std::io::Read;
+use std::path::PathBuf;
+use std::process::Command;
+
+fn bin() -> &'static str {
+    env!("CARGO_BIN_EXE_polyscribe")
+}
+
+fn manifest_path(relative: &str) -> PathBuf {
+    let mut p = PathBuf::from(env!("CARGO_MANIFEST_DIR"));
+    p.push(relative);
+    p
+}
+
+#[test]
+fn error_on_missing_input_file() {
+    let exe = bin();
+    let missing = manifest_path("input/definitely_missing_123.json");
+    let out = Command::new(exe)
+        .arg(missing.as_os_str())
+        .output()
+        .expect("failed to run polyscribe with missing input");
+    assert!(!out.status.success(), "command should fail on missing input");
+    let stderr = String::from_utf8_lossy(&out.stderr);
+    assert!(
+        stderr.contains("Input not found") || stderr.contains("No input files provided"),
+        "stderr should mention missing input; got: {}",
+        stderr
+    );
+}
+
+#[test]
+fn error_on_directory_as_input() {
+    let exe = bin();
+    // Use the repo's input directory which exists and is a directory
+    let input_dir = manifest_path("input");
+    let out = Command::new(exe)
+        .arg(input_dir.as_os_str())
+        .output()
+        .expect("failed to run polyscribe with directory input");
+    assert!(!out.status.success(), "command should fail on dir input");
+    let stderr = String::from_utf8_lossy(&out.stderr);
+    assert!(
+        stderr.contains("directory") || stderr.contains("Unsupported input type"),
+        "stderr should mention directory/unsupported; got: {}",
+        stderr
+    );
+}
+
+#[test]
+fn error_on_no_ffmpeg_present() {
+    let exe = bin();
+    // Create a tiny temp .wav file (may be empty; ffmpeg will be attempted but PATH will be empty)
+    let tmp_dir = manifest_path("target/tmp/itest_no_ffmpeg");
+    let _ = fs::remove_dir_all(&tmp_dir);
+    fs::create_dir_all(&tmp_dir).unwrap();
+    let wav = tmp_dir.join("dummy.wav");
+    fs::write(&wav, b"\0\0\0\0").unwrap();
+
+    let out = Command::new(exe)
+        .arg(wav.as_os_str())
+        .env("PATH", "") // simulate ffmpeg missing
+        .env_remove("WHISPER_MODEL")
+        .env("POLYSCRIBE_MODELS_BASE_COPY_DIR", manifest_path("models").as_os_str())
+        .arg("--language").arg("en")
+        .output()
+        .expect("failed to run polyscribe with empty PATH");
+    assert!(
+        !out.status.success(),
+        "command should fail when ffmpeg is not found"
+    );
+    let stderr = String::from_utf8_lossy(&out.stderr);
+    assert!(
+        stderr.contains("ffmpeg not found") || stderr.contains("Failed to execute ffmpeg"),
+        "stderr should mention ffmpeg not found; got: {}",
+        stderr
+    );
+}
+
+#[cfg(unix)]
+#[test]
+fn error_on_readonly_output_dir() {
+    use std::os::unix::fs::PermissionsExt;
+
+    let exe = bin();
+    let input1 = manifest_path("input/1-s0wlz.json");
+
+    // Prepare a read-only directory
+    let tmp_dir = manifest_path("target/tmp/itest_readonly_out");
+    let _ = fs::remove_dir_all(&tmp_dir);
+    fs::create_dir_all(&tmp_dir).unwrap();
+    let mut perms = fs::metadata(&tmp_dir).unwrap().permissions();
+    perms.set_mode(0o555); // read & execute, no write
+    fs::set_permissions(&tmp_dir, perms).unwrap();
+
+    let out = Command::new(exe)
+        .arg(input1.as_os_str())
+        .arg("-o")
+        .arg(tmp_dir.as_os_str())
+        .output()
+        .expect("failed to run polyscribe with read-only output dir");
+
+    // Restore perms for cleanup
+    let mut perms2 = fs::metadata(&tmp_dir).unwrap().permissions();
+    perms2.set_mode(0o755);
+    let _ = fs::set_permissions(&tmp_dir, perms2);
+
+    assert!(
+        !out.status.success(),
+        "command should fail when outputs cannot be created"
+    );
+    let stderr = String::from_utf8_lossy(&out.stderr);
+    assert!(
+        stderr.contains("Failed to create output") || stderr.contains("permission"),
+        "stderr should mention failure to create outputs; got: {}",
+        stderr
+    );
+
+    // Cleanup
+    let _ = fs::remove_dir_all(&tmp_dir);
+}