feat: Add path_utils.py with method resolve_repo_path

Signed-off-by: Jorge Garcia Oncins <jgarciao@redhat.com>

Author: jgarciao

CONSTITUTION.md

@@ -79,6 +79,7 @@ All code MUST consider security implications.
 - Avoid running destructive commands without explicit user confirmation
 - Use detect-secrets and gitleaks pre-commit hooks to prevent secret leakage
 - Test code MUST NOT introduce vulnerabilities into the tested systems
+- Use `utilities.path_utils.resolve_repo_path` to resolve and validate any user-supplied or parameterized file paths, preventing path-traversal and symlink-escape outside the repository root
 - JIRA ticket links are allowed in PRs and commit messages (our Jira is public)
 - Do NOT reference internal-only resources (Jenkins, Confluence, Slack threads) in code, PRs, or commit messages
 - Do NOT link embargoed or security-restricted (RH-employee-only) tickets

tests/llama_stack/conftest.py

@@ -1,6 +1,5 @@
 import os
 from collections.abc import Callable, Generator
-from pathlib import Path
 from typing import Any
 
 import httpx
@@ -810,7 +809,6 @@ def vector_store(
             try:
                 vector_store_upload_doc_sources(
                     doc_sources=doc_sources,
-                    repo_root=Path(request.config.rootdir).resolve(),
                     llama_stack_client=unprivileged_llama_stack_client,
                     vector_store=vector_store,
                     vector_io_provider=vector_io_provider,

tests/llama_stack/utils.py

@@ -20,6 +20,7 @@
     LLS_CORE_POD_FILTER,
 )
 from utilities.exceptions import UnexpectedResourceCountError
+from utilities.path_utils import resolve_repo_path
 from utilities.resources.llama_stack_distribution import LlamaStackDistribution
 
 LOGGER = get_logger(name=__name__)
@@ -280,26 +281,24 @@ def vector_store_create_file_from_path(
 
 def vector_store_upload_doc_sources(
     doc_sources: Any,
-    repo_root: Path,
     llama_stack_client: LlamaStackClient,
     vector_store: Any,
     vector_io_provider: str,
 ) -> None:
     """Upload parametrized document sources (URLs and repo-local paths) to a vector store.
 
-    Resolves each local path under ``repo_root`` and re-resolves directory entries to avoid
-    symlink escape outside the repository.
+    Resolves each local path via ``resolve_repo_path`` and re-resolves directory entries
+    to avoid symlink escape outside the repository.
 
     Args:
         doc_sources: List of URL or path strings (repo-relative or absolute under repo root).
-        repo_root: Resolved repository root; local paths must resolve under this directory.
         llama_stack_client: Client used for file and vector store APIs.
         vector_store: Target vector store (must expose ``id``).
         vector_io_provider: Provider id for log context only.
 
     Raises:
         TypeError: If ``doc_sources`` is not a list.
-        ValueError: If a local path resolves outside ``repo_root``.
+        ValueError: If a local path resolves outside the repo root.
         FileNotFoundError: If a file or non-empty directory source is missing.
     """
     if not isinstance(doc_sources, list):
@@ -310,7 +309,6 @@ def vector_store_upload_doc_sources(
         vector_store.id,
         doc_sources,
     )
-    repo_root_resolved = repo_root.resolve()
     for source in doc_sources:
         if source.startswith(("http://", "https://")):
             vector_store_create_file_from_url(
@@ -319,25 +317,14 @@ def vector_store_upload_doc_sources(
                 vector_store=vector_store,
             )
             continue
-        raw_path = Path(source)  # noqa: FCN001
-        resolved_source = raw_path.resolve() if raw_path.is_absolute() else (repo_root_resolved / raw_path).resolve()
-        if not resolved_source.is_relative_to(repo_root_resolved):
-            raise ValueError(
-                f"doc_sources path must be under repo root ({repo_root_resolved}): {source!r}",
-            )
-        source_path = resolved_source
+        source_path = resolve_repo_path(source=source)
 
         if source_path.is_dir():
             files = sorted(source_path.iterdir())
             if not files:
                 raise FileNotFoundError(f"No files found in directory: {source_path}")
             for file_path in files:
-                file_path_resolved = file_path.resolve(strict=True)
-                if not file_path_resolved.is_relative_to(repo_root_resolved):
-                    raise ValueError(
-                        f"doc_sources directory entry must resolve under repo root "
-                        f"({repo_root_resolved}): {file_path!r} -> {file_path_resolved!r}",
-                    )
+                file_path_resolved = resolve_repo_path(source=file_path)
                 if not file_path_resolved.is_file():
                     continue
                 vector_store_create_file_from_path(

utilities/path_utils.py

@@ -0,0 +1,43 @@
+"""Path resolution and validation utilities for repo-relative file access."""
+
+from pathlib import Path
+
+_REPO_ROOT = Path(__file__).resolve().parent.parent
+
+
+def resolve_repo_path(source: str | Path, repo_root: Path | None = None) -> Path:
+    """Turn a repo-relative or absolute path into a safe, resolved absolute path.
+
+    Ensures the final path lives inside the repository root. Symlinks are fully
+    resolved before the check so that a symlink pointing outside the repo is rejected.
+
+    Accepted (returns resolved absolute path)::
+
+        resolve_repo_path("tests/data/sample.pdf")          # relative to repo root
+        resolve_repo_path("tests/data/../data/sample.pdf")   # normalised, still under root
+        resolve_repo_path("/home/user/repo/tests/data/f.txt")  # absolute, under root
+
+    Rejected (raises ``ValueError``)::
+
+        resolve_repo_path("../../etc/passwd")                # escapes repo root
+        resolve_repo_path("/tmp/evil.txt")                   # absolute, outside root
+
+    Args:
+        source: A repo-relative string/path or an absolute path.
+        repo_root: Repository root to validate against. Defaults to the detected
+            repo root (parent of the ``utilities/`` package).
+
+    Returns:
+        The resolved absolute path, guaranteed to be under ``repo_root``.
+
+    Raises:
+        ValueError: If the resolved path falls outside the repo root.
+    """
+    repo_root_resolved = (repo_root or _REPO_ROOT).resolve()
+    raw = Path(source)  # noqa: FCN001
+    resolved = raw.resolve() if raw.is_absolute() else (repo_root_resolved / raw).resolve()
+    if not resolved.is_relative_to(repo_root_resolved):
+        raise ValueError(
+            f"Path must be under repo root ({repo_root_resolved}): {source!r}",
+        )
+    return resolved