Block local file paths at Campaign server boundaries

src/govoplan_campaign/backend/path_security.py

@@ -0,0 +1,170 @@
+from __future__ import annotations
+
+from pathlib import PureWindowsPath
+from typing import Any, Iterable
+
+
+class CampaignPathSecurityError(ValueError):
+    """Raised when server-provided campaign JSON could read local files."""
+
+
+def _text(value: object) -> str:
+    return value.strip() if isinstance(value, str) else ""
+
+
+def assert_logical_relative_path(value: object, *, field: str) -> None:
+    raw = _text(value)
+    if not raw:
+        return
+    if "\x00" in raw:
+        raise CampaignPathSecurityError(f"{field} contains a NUL byte")
+
+    normalized = raw.replace("\\", "/")
+    windows_path = PureWindowsPath(raw)
+    if (
+        normalized.startswith("/")
+        or normalized.startswith("~")
+        or windows_path.is_absolute()
+        or bool(windows_path.drive)
+    ):
+        raise CampaignPathSecurityError(f"{field} must be a relative managed-file path")
+    if any(part == ".." for part in normalized.split("/")):
+        raise CampaignPathSecurityError(f"{field} must not contain parent-directory traversal")
+
+
+def is_managed_source(value: object) -> bool:
+    raw = _text(value)
+    if not raw.startswith("managed:"):
+        return False
+    parts = raw.split(":", 2)
+    return len(parts) == 3 and parts[1] in {"user", "group"} and bool(parts[2].strip())
+
+
+def _attachment_rules(raw_json: dict[str, Any]) -> Iterable[tuple[str, dict[str, Any]]]:
+    attachments = raw_json.get("attachments")
+    if isinstance(attachments, dict):
+        global_rules = attachments.get("global")
+        if isinstance(global_rules, list):
+            for index, rule in enumerate(global_rules):
+                if isinstance(rule, dict):
+                    yield f"attachments.global[{index}]", rule
+
+    entries = raw_json.get("entries")
+    inline = entries.get("inline") if isinstance(entries, dict) else None
+    if isinstance(inline, list):
+        for entry_index, entry in enumerate(inline):
+            rules = entry.get("attachments") if isinstance(entry, dict) else None
+            if not isinstance(rules, list):
+                continue
+            for rule_index, rule in enumerate(rules):
+                if isinstance(rule, dict):
+                    yield f"entries.inline[{entry_index}].attachments[{rule_index}]", rule
+    defaults = entries.get("defaults") if isinstance(entries, dict) else None
+    default_rules = defaults.get("attachments") if isinstance(defaults, dict) else None
+    if isinstance(default_rules, list):
+        for rule_index, rule in enumerate(default_rules):
+            if isinstance(rule, dict):
+                    yield f"entries.defaults.attachments[{rule_index}]", rule
+
+
+def _assert_inline_server_sources(
+    raw_json: dict[str, Any],
+    *,
+    source_filename: str | None,
+    source_base_path: str | None,
+) -> None:
+    if _text(source_filename):
+        raise CampaignPathSecurityError("source_filename is not accepted for server-managed campaigns")
+    if _text(source_base_path):
+        raise CampaignPathSecurityError("source_base_path is not accepted for server-managed campaigns")
+
+    template = raw_json.get("template")
+    template_source = template.get("source") if isinstance(template, dict) else None
+    path_fields = ("subject_path", "text_path", "html_path")
+    if isinstance(template_source, dict) and any(_text(template_source.get(key)) for key in path_fields):
+        raise CampaignPathSecurityError(
+            "template source paths are not accepted for server-managed campaigns; store template content inline"
+        )
+
+    entries = raw_json.get("entries")
+    entries_source = entries.get("source") if isinstance(entries, dict) else None
+    if isinstance(entries_source, dict) and _text(entries_source.get("path")):
+        raise CampaignPathSecurityError(
+            "entries.source.path is not accepted for server-managed campaigns; import recipients into the campaign snapshot"
+        )
+
+
+def _managed_attachment_base_paths(attachments: dict[str, Any]) -> dict[str, dict[str, Any]]:
+    assert_logical_relative_path(attachments.get("base_path"), field="attachments.base_path")
+    raw_base_paths = attachments.get("base_paths")
+    base_paths = raw_base_paths if isinstance(raw_base_paths, list) else []
+    managed: dict[str, dict[str, Any]] = {}
+    for index, item in enumerate(base_paths):
+        if not isinstance(item, dict):
+            continue
+        assert_logical_relative_path(
+            item.get("path"),
+            field=f"attachments.base_paths[{index}].path",
+        )
+        source_is_managed = is_managed_source(item.get("source"))
+        base_path_id = _text(item.get("id"))
+        if base_path_id and source_is_managed:
+            managed[base_path_id] = item
+        if item.get("unsent_warning") is True and not source_is_managed:
+            raise CampaignPathSecurityError(
+                f"attachments.base_paths[{index}] must use a managed Files source before unsent-file scanning is enabled"
+            )
+    return managed
+
+
+def _assert_managed_attachment_rules(
+    raw_json: dict[str, Any],
+    *,
+    managed_base_paths: dict[str, dict[str, Any]],
+    managed_files_available: bool,
+) -> None:
+    rules = list(_attachment_rules(raw_json))
+    if not rules:
+        return
+    if not managed_files_available:
+        raise CampaignPathSecurityError("campaign attachments require the Files module in server mode")
+
+    for field, rule in rules:
+        base_path_id = _text(rule.get("base_path_id"))
+        if not base_path_id or base_path_id not in managed_base_paths:
+            raise CampaignPathSecurityError(
+                f"{field}.base_path_id must select an attachments.base_paths entry backed by managed Files"
+            )
+        assert_logical_relative_path(rule.get("base_dir"), field=f"{field}.base_dir")
+        assert_logical_relative_path(rule.get("file_filter"), field=f"{field}.file_filter")
+
+
+def assert_server_safe_campaign_paths(
+    raw_json: dict[str, Any],
+    *,
+    source_filename: str | None = None,
+    source_base_path: str | None = None,
+    managed_files_available: bool,
+) -> None:
+    """Reject local-file references at the HTTP/server trust boundary.
+
+    File-oriented campaign loading remains available to trusted operator code
+    through the campaign loader and message builder. Persisted/API campaigns
+    must use inline templates and recipients plus managed Files attachment
+    sources, which are materialized into an isolated snapshot before use.
+    """
+
+    _assert_inline_server_sources(
+        raw_json,
+        source_filename=source_filename,
+        source_base_path=source_base_path,
+    )
+
+    attachments = raw_json.get("attachments")
+    if not isinstance(attachments, dict):
+        return
+    _assert_managed_attachment_rules(
+        raw_json,
+        managed_base_paths=_managed_attachment_base_paths(attachments),
+        managed_files_available=managed_files_available,
+    )