fix: prevent path traversal via Gitea ref/sha/base/head parameters
test / test (push) Successful in 20s
lint / lint (push) Successful in 22s
docker / lint (pull_request) Successful in 33s
docker / test (pull_request) Successful in 25s
test / test (pull_request) Successful in 38s
lint / lint (pull_request) Successful in 40s
docker / docker-test (pull_request) Successful in 15s
docker / docker-publish (pull_request) Has been skipped

The ref-like tool arguments (ref, sha, base, head) were only length-limited
and were interpolated unencoded into Gitea API URL paths (get_tree,
get_commit_diff, compare_refs). Because httpx collapses ".." path segments
(RFC 3986), a crafted value such as "../../../../owner/repo/contents/secret"
escaped the declared owner/repo prefix. In service-PAT mode this allowed a
user authorized on one repository to read arbitrary repositories the service
token could reach, and in OAuth mode it bypassed the policy engine's
per-repository rules (which never see ref values).

Two defense layers:
- arguments.py: add _validate_git_ref / GitRef that rejects ".." path
  segments, leading "/", backslashes, null bytes, control chars, whitespace,
  and "?"/"#", while preserving legitimate slash refs (feature/foo, v1.2.3).
  This is what actually closes the traversal.
- gitea_client.py: defense-in-depth urllib.parse.quote() on owner/repo
  (safe="") and ref/sha/base/head/filepath (safe="/") in every repo URL
  builder, mirroring the existing pattern in server.py.

Tests: negative cases for traversal/unsafe chars across all four fields,
positive cases for slash-containing refs, length-bound regression, and a
URL-layer confinement check. Full suite green (176 passed), coverage 85.64%.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
2026-06-14 15:57:52 +02:00
parent 90df37366f
commit 2d95e89035
3 changed files with 179 additions and 30 deletions
+45 -9
View File
@@ -2,13 +2,49 @@
from __future__ import annotations
from typing import Literal
from typing import Annotated, Literal
from pydantic import BaseModel, ConfigDict, Field, model_validator
from pydantic import AfterValidator, BaseModel, ConfigDict, Field, model_validator
_REPO_PART_PATTERN = r"^[A-Za-z0-9._-]{1,100}$"
def _validate_git_ref(value: str) -> str:
"""Validate a git ref-like value (ref/sha/base/head) against traversal.
Refs that legitimately contain ``/`` (e.g. ``feature/foo``, ``release/1.0``)
are preserved; only traversal and unsafe URL-path characters are rejected.
Args:
value: Candidate ref, sha, base, or head value.
Returns:
The unchanged value when it is safe.
Raises:
ValueError: When the value could escape the intended repository path.
"""
# Security decision: block path traversal and absolute references.
if ".." in value.split("/"):
raise ValueError("ref must not contain '..' path segments")
if value.startswith("/"):
raise ValueError("ref must not start with '/'")
if "\\" in value:
raise ValueError("ref must not contain backslashes")
if "\x00" in value:
raise ValueError("ref must not contain null bytes")
if any(ord(char) < 0x20 or ord(char) == 0x7F for char in value):
raise ValueError("ref must not contain control characters")
if any(char.isspace() for char in value):
raise ValueError("ref must not contain whitespace")
if "?" in value or "#" in value:
raise ValueError("ref must not contain '?' or '#'")
return value
GitRef = Annotated[str, AfterValidator(_validate_git_ref)]
class StrictBaseModel(BaseModel):
"""Strict model base that rejects unexpected fields."""
@@ -29,7 +65,7 @@ class RepositoryArgs(StrictBaseModel):
class FileTreeArgs(RepositoryArgs):
"""Arguments for get_file_tree."""
ref: str = Field(default="main", min_length=1, max_length=200)
ref: GitRef = Field(default="main", min_length=1, max_length=200)
recursive: bool = Field(default=False)
@@ -37,7 +73,7 @@ class FileContentsArgs(RepositoryArgs):
"""Arguments for get_file_contents."""
filepath: str = Field(..., min_length=1, max_length=1024)
ref: str = Field(default="main", min_length=1, max_length=200)
ref: GitRef = Field(default="main", min_length=1, max_length=200)
@model_validator(mode="after")
def validate_filepath(self) -> FileContentsArgs:
@@ -55,7 +91,7 @@ class SearchCodeArgs(RepositoryArgs):
"""Arguments for search_code."""
query: str = Field(..., min_length=1, max_length=256)
ref: str = Field(default="main", min_length=1, max_length=200)
ref: GitRef = Field(default="main", min_length=1, max_length=200)
page: int = Field(default=1, ge=1, le=10_000)
limit: int = Field(default=25, ge=1, le=100)
@@ -63,7 +99,7 @@ class SearchCodeArgs(RepositoryArgs):
class ListCommitsArgs(RepositoryArgs):
"""Arguments for list_commits."""
ref: str = Field(default="main", min_length=1, max_length=200)
ref: GitRef = Field(default="main", min_length=1, max_length=200)
page: int = Field(default=1, ge=1, le=10_000)
limit: int = Field(default=25, ge=1, le=100)
@@ -71,14 +107,14 @@ class ListCommitsArgs(RepositoryArgs):
class CommitDiffArgs(RepositoryArgs):
"""Arguments for get_commit_diff."""
sha: str = Field(..., min_length=7, max_length=64)
sha: GitRef = Field(..., min_length=7, max_length=64)
class CompareRefsArgs(RepositoryArgs):
"""Arguments for compare_refs."""
base: str = Field(..., min_length=1, max_length=200)
head: str = Field(..., min_length=1, max_length=200)
base: GitRef = Field(..., min_length=1, max_length=200)
head: GitRef = Field(..., min_length=1, max_length=200)
class ListIssuesArgs(RepositoryArgs):