diff --git a/src/mistral_ocr/__init__.py b/src/mistral_ocr/__init__.py
index 1e128a9..580f487 100644
--- a/src/mistral_ocr/__init__.py
+++ b/src/mistral_ocr/__init__.py
@@ -83,7 +83,11 @@ def find_pdf_files(input_path: Path) -> list[Path]:
             return []
 
     if input_path.is_dir():
-        return list(input_path.glob("**/*.pdf"))
+        return [
+            path
+            for path in input_path.rglob("*")
+            if path.suffix.lower() == ".pdf"
+        ]
 
     return []
 
diff --git a/src/mistral_ocr/ocr_utils.py b/src/mistral_ocr/ocr_utils.py
index d877e03..60dc6da 100644
--- a/src/mistral_ocr/ocr_utils.py
+++ b/src/mistral_ocr/ocr_utils.py
@@ -2,7 +2,7 @@
 
 import os
 from pathlib import Path
-from typing import NamedTuple
+from typing import Callable, NamedTuple
 
 import pyperclip
 from dotenv import load_dotenv
@@ -136,7 +136,14 @@ def save_markdown_to_file(content: str, output_path: Path) -> None:
 
 
 def process_and_save_pdf(
-    client: Mistral, input_path: Path, output_dir: Path, force: bool = False
+    client: Mistral,
+    input_path: Path,
+    output_dir: Path,
+    force: bool = False,
+    *,
+    process_func: Callable[[Mistral, Path], OCRResponse] = process_pdf_file,
+    extract_func: Callable[[OCRResponse], str] = extract_markdown_from_response,
+    save_func: Callable[[str, Path], None] = save_markdown_to_file,
 ) -> tuple[bool, str, ProcessedDocument | None]:
     """Process a single PDF file and save its markdown output.
 
@@ -160,9 +167,9 @@ def process_and_save_pdf(
                 None,
             )
 
-        ocr_response = process_pdf_file(client, input_path)
-        markdown_content = extract_markdown_from_response(ocr_response)
-        save_markdown_to_file(markdown_content, output_path)
+        ocr_response = process_func(client, input_path)
+        markdown_content = extract_func(ocr_response)
+        save_func(markdown_content, output_path)
 
         processed_doc = ProcessedDocument(
             filename=input_path.name,
diff --git a/tests/test_file_discovery.py b/tests/test_file_discovery.py
new file mode 100644
index 0000000..6be7f5a
--- /dev/null
+++ b/tests/test_file_discovery.py
@@ -0,0 +1,132 @@
+from pathlib import Path
+from typing import Any
+
+import pytest
+
+from mistral_ocr import (
+    find_pdf_files,
+    determine_output_directory,
+    create_conversion_plan,
+    process_and_save_pdf,
+)
+
+
+class DummyClient:
+    pass
+
+
+class DummyResponse:
+    pass
+
+
+@pytest.mark.parametrize(
+    "filename,expected",
+    [
+        ("doc.pdf", 1),
+        ("DOC.PDF", 1),
+        ("not_pdf.txt", 0),
+    ],
+)
+def test_find_pdf_files_file(tmp_path: Path, filename: str, expected: int) -> None:
+    file_path = tmp_path / filename
+    file_path.write_text("content")
+    result = find_pdf_files(file_path)
+    assert len(result) == expected
+
+
+def test_find_pdf_files_directory(tmp_path: Path) -> None:
+    (tmp_path / "sub").mkdir()
+    files = [
+        tmp_path / "a.pdf",
+        tmp_path / "sub" / "b.PDF",
+        tmp_path / "c.txt",
+    ]
+    for f in files:
+        f.write_text("data")
+    result = find_pdf_files(tmp_path)
+    assert set(p.name for p in result) == {"a.pdf", "b.PDF"}
+
+
+def test_find_pdf_files_missing_path(tmp_path: Path) -> None:
+    missing = tmp_path / "missing"
+    result = find_pdf_files(missing)
+    assert result == []
+
+
+@pytest.mark.parametrize(
+    "input_path,output_dir,expected",
+    [
+        ("dir", "explicit", "explicit"),
+        ("dir", None, "dir"),
+    ],
+)
+def test_determine_output_directory_directory(tmp_path: Path, input_path: str, output_dir: str | None, expected: str) -> None:
+    base = tmp_path / "dir"
+    base.mkdir()
+    arg_output = tmp_path / output_dir if output_dir else None
+    result = determine_output_directory(base, arg_output)
+    assert result == (tmp_path / expected)
+
+
+def test_determine_output_directory_file(tmp_path: Path) -> None:
+    file_path = tmp_path / "f.pdf"
+    file_path.write_text("x")
+    result = determine_output_directory(file_path, None)
+    assert result == tmp_path
+
+
+def test_create_conversion_plan_output_paths(tmp_path: Path) -> None:
+    pdf = tmp_path / "doc.pdf"
+    pdf.write_text("x")
+    plan = create_conversion_plan(pdf, None, False, False)
+    assert plan.output_dir == tmp_path
+    assert len(plan.files) == 1
+    assert plan.files[0].output_path == tmp_path / "doc.md"
+    assert plan.files[0].action == "convert"
+
+
+def test_create_conversion_plan_skip_existing(tmp_path: Path) -> None:
+    pdf = tmp_path / "doc.pdf"
+    pdf.write_text("x")
+    md = tmp_path / "doc.md"
+    md.write_text("y")
+    plan = create_conversion_plan(tmp_path, None, False, False)
+    [action] = plan.files
+    assert action.action == "skip"
+    assert action.skip_reason == "already exists"
+
+
+def test_process_and_save_pdf_custom_funcs(tmp_path: Path) -> None:
+    pdf = tmp_path / "doc.pdf"
+    pdf.write_text("x")
+    out_dir = tmp_path / "out"
+    out_dir.mkdir()
+
+    called: dict[str, Any] = {}
+
+    def fake_process(client: DummyClient, path: Path) -> DummyResponse:
+        called["process"] = path
+        return DummyResponse()
+
+    def fake_extract(resp: DummyResponse) -> str:
+        called["extract"] = resp
+        return "markdown"
+
+    def fake_save(content: str, path: Path) -> None:
+        called["save"] = (content, path)
+        path.write_text(content)
+
+    success, msg, doc = process_and_save_pdf(
+        DummyClient(),
+        pdf,
+        out_dir,
+        process_func=fake_process,
+        extract_func=fake_extract,
+        save_func=fake_save,
+    )
+
+    assert success
+    assert doc
+    assert doc.output_path.read_text() == "markdown"
+    assert called["process"] == pdf
+    assert called["save"][1] == out_dir / "doc.md"