fix: raise clear errors for missing input files and unavailable Java runtime

Objective: Users get cryptic CalledProcessError when passing non-existent
paths or running without Java installed.

Approach: Add early validation in lazy_load_data() — FileNotFoundError
for missing paths, RuntimeError with install guidance for missing Java.
Cache only positive Java probe results; catch OSError for PermissionError.

Evidence: 33 unit tests passed (30 existing + 3 new).
Before: CalledProcessError / bare FileNotFoundError from subprocess.
After: actionable error messages with specific path and install instructions.

Author: hyunhee-jo

llama_index/readers/opendataloader_pdf/base.py

@@ -4,6 +4,7 @@
 import logging
 import re
 import shutil
+import subprocess
 import tempfile
 from collections import defaultdict
 from pathlib import Path
@@ -14,6 +15,34 @@
 
 logger = logging.getLogger(__name__)
 
+
+_java_found: Optional[bool] = None
+
+
+def _java_available() -> bool:
+    """Return True if a working Java runtime is found on the system PATH.
+
+    A positive result is cached for the lifetime of the process; a negative
+    result is re-probed on every call so that installing Java mid-session
+    is recognised without a restart.
+    """
+    global _java_found  # noqa: PLW0603
+    if _java_found:
+        return True
+    try:
+        result = subprocess.run(
+            ["java", "-version"],
+            capture_output=True,
+            text=True,
+            check=False,
+            timeout=10,
+        )
+        _java_found = result.returncode == 0
+        return _java_found
+    except (OSError, subprocess.TimeoutExpired):
+        return False
+
+
 _FORMAT_TO_EXT = {
     "json": "json",
     "text": "txt",
@@ -180,6 +209,19 @@ def lazy_load_data(
         else:
             paths = [str(p) for p in file_path]
 
+        for p in paths:
+            if not Path(p).exists():
+                raise FileNotFoundError(f"Input path does not exist: {p}")
+
+        if not _java_available():
+            raise RuntimeError(
+                "Java is not found on the system PATH. "
+                "OpenDataLoader PDF requires Java 11+. "
+                "Install Java from https://adoptium.net/ and ensure "
+                "'java' is on your system PATH. "
+                "Verify with: java -version"
+            )
+
         ext = _FORMAT_TO_EXT[fmt]
 
         try:

tests/test_readers_opendataloader_pdf.py

@@ -8,6 +8,33 @@
 from llama_index.core.schema import Document
 from llama_index.readers.opendataloader_pdf import OpenDataLoaderPDFReader
 
+# Save original before any monkeypatching.
+_original_path_exists = Path.exists
+
+
+@pytest.fixture(autouse=True)
+def _bypass_input_validation(monkeypatch):
+    """Bypass file-existence and Java checks in unit tests.
+
+    Tests that verify validation behaviour override these explicitly.
+    Only dummy PDF paths are faked; all other Path.exists calls delegate
+    to the real implementation.
+    """
+    monkeypatch.setattr(
+        "llama_index.readers.opendataloader_pdf.base._java_available",
+        lambda: True,
+    )
+    monkeypatch.setattr(
+        "llama_index.readers.opendataloader_pdf.base._java_found", None
+    )
+
+    def _fake_exists(self):
+        if self.suffix == ".pdf":
+            return True
+        return _original_path_exists(self)
+
+    monkeypatch.setattr("pathlib.Path.exists", _fake_exists)
+
 
 # ---------------------------------------------------------------------------
 # TestInit: Verify default and custom parameter values
@@ -552,3 +579,47 @@ def mock_import(name, *args, **kwargs):
         with patch("builtins.__import__", side_effect=mock_import):
             with pytest.raises(ImportError, match="opendataloader_pdf"):
                 list(reader.lazy_load_data(file_path="doc.pdf"))
+
+
+# ---------------------------------------------------------------------------
+# TestFileNotFound: Missing input file raises FileNotFoundError
+# ---------------------------------------------------------------------------
+class TestFileNotFound:
+    """Test that missing input files raise FileNotFoundError."""
+
+    @pytest.fixture(autouse=True)
+    def _real_path_exists(self, monkeypatch):
+        """Restore real Path.exists so file-not-found is testable."""
+        monkeypatch.setattr("pathlib.Path.exists", _original_path_exists)
+
+    def test_single_missing_file(self) -> None:
+        reader = OpenDataLoaderPDFReader()
+        with pytest.raises(
+            FileNotFoundError,
+            match=r"Input path does not exist: nonexistent\.pdf",
+        ):
+            list(reader.lazy_load_data(file_path="nonexistent.pdf"))
+
+    def test_missing_file_in_list(self) -> None:
+        reader = OpenDataLoaderPDFReader()
+        with pytest.raises(
+            FileNotFoundError,
+            match=r"Input path does not exist: nonexistent\.pdf",
+        ):
+            list(reader.lazy_load_data(file_path=["nonexistent.pdf"]))
+
+
+# ---------------------------------------------------------------------------
+# TestJavaCheck: Missing Java raises RuntimeError
+# ---------------------------------------------------------------------------
+class TestJavaCheck:
+    """Test that missing Java runtime raises RuntimeError."""
+
+    @patch(
+        "llama_index.readers.opendataloader_pdf.base._java_available",
+        return_value=False,
+    )
+    def test_no_java_raises_runtime_error(self, _: MagicMock) -> None:
+        reader = OpenDataLoaderPDFReader()
+        with pytest.raises(RuntimeError, match="Java is not found"):
+            list(reader.lazy_load_data(file_path="doc.pdf"))