Prevent escaping the source doc's folder, or running arbitrary code, …

…without explicit opt-in at the command-line.

bikeshed/InputSource.py

@@ -13,6 +13,7 @@
 import requests
 import tenacity
 
+from . import config
 from .Line import Line
 
 
@@ -38,17 +39,17 @@ class InputSource:
     manager for temporarily switching to the directory of a file InputSource.
     """
 
-    def __new__(cls, sourceName: str):
+    def __new__(cls, sourceName: str, **kwargs):
         """Dispatches to the right subclass."""
         if cls != InputSource:
             # Only take control of calls to InputSource(...) itself.
             return super().__new__(cls)
 
         if sourceName == "-":
-            return StdinInputSource(sourceName)
+            return StdinInputSource(sourceName, **kwargs)
         if sourceName.startswith("https:"):
-            return UrlInputSource(sourceName)
-        return FileInputSource(sourceName)
+            return UrlInputSource(sourceName, **kwargs)
+        return FileInputSource(sourceName, **kwargs)
 
     @abstractmethod
     def __str__(self) -> str:
@@ -157,11 +158,17 @@ def relative(self, relativePath) -> UrlInputSource:
 
 
 class FileInputSource(InputSource):
-    def __init__(self, sourceName: str):
+    def __init__(self, sourceName: str, *, chroot: bool, chrootPath: Optional[str] = None):
         self.sourceName = sourceName
+        self.chrootPath = chrootPath
         self.type = "file"
         self.content = None
 
+        if chroot and self.chrootPath is None:
+            self.chrootPath = self.directory()
+        if self.chrootPath is not None:
+            self.sourceName = config.chrootPath(self.chrootPath, self.sourceName)
+
     def __str__(self) -> str:
         return self.sourceName
 
@@ -179,7 +186,7 @@ def directory(self) -> str:
         return os.path.dirname(os.path.abspath(self.sourceName))
 
     def relative(self, relativePath) -> FileInputSource:
-        return FileInputSource(os.path.join(self.directory(), relativePath))
+        return FileInputSource(os.path.join(self.directory(), relativePath), chroot=False, chrootPath=self.chrootPath)
 
     def cheaplyExists(self, relativePath) -> bool:
         return os.access(self.relative(relativePath).sourceName, os.R_OK)

bikeshed/Spec.py

@@ -58,7 +58,7 @@ def __init__(
                 "No input file specified, and no *.bs or *.src.html files found in current directory.\nPlease specify an input file, or use - to pipe from STDIN."
             )
             return
-        self.inputSource = InputSource(inputFilename)
+        self.inputSource = InputSource(inputFilename, chroot=constants.chroot)
         self.transitiveDependencies = set()
         self.debug = debug
         self.token = token

bikeshed/cli.py

@@ -75,6 +75,18 @@ def main():
         choices=["nothing", "fatal", "link-error", "warning", "everything"],
         help="Determines what sorts of errors cause Bikeshed to die (quit immediately with an error status code). Default is 'fatal'; the -f flag is a shorthand for 'nothing'",
     )
+    argparser.add_argument(
+        "--allow-nonlocal-files",
+        dest="allowNonlocalFiles",
+        action="store_true",
+        help="Allows Bikeshed to see/include files from folders higher than the one your source document is in."
+    )
+    argparser.add_argument(
+        "--allow-execute",
+        dest="allowExecute",
+        action="store_true",
+        help="Allow some features to execute arbitrary code from outside the Bikeshed codebase."
+    )
 
     subparsers = argparser.add_subparsers(title="Subcommands", dest="subparserName")
 
@@ -444,6 +456,8 @@ def main():
             constants.printMode = "console"
     else:
         constants.printMode = options.printMode
+    constants.chroot = not options.allowNonlocalFiles
+    constants.executeCode = options.allowExecute
 
     update.fixupDataFiles()
     if options.subparserName == "update":

bikeshed/config/main.py

@@ -4,6 +4,9 @@
 
 import lxml
 
+from .. import constants
+from .. import messages
+
 
 def englishFromList(items, conjunction="or"):
     # Format a list of strings into an English list.
@@ -168,7 +171,19 @@ def flatten(arr):
 
 def scriptPath(*pathSegs):
     startPath = os.path.dirname(os.path.dirname(os.path.realpath(__file__)))
-    return os.path.join(startPath, *pathSegs)
+    path = os.path.join(startPath, *pathSegs)
+    return path
+
+
+def chrootPath(chrootPath, path):
+    chrootPath = os.path.abspath(chrootPath)
+    path = os.path.abspath(path)
+    if not path.startswith(chrootPath):
+        messages.die(f"Attempted to access a file ({path}) outside the source document's directory ({chrootPath}). See --allow-nonlocal-files.")
+        raise Exception()
+    else:
+        return path
+
 
 
 def doEvery(s, action, lastTime=None):

bikeshed/config/retrieve.py

@@ -63,7 +63,7 @@ def _fail(self, location, str, okayToFail):
 )
 
 
-def retrieveBoilerplateFile(doc, name, group=None, status=None, error=True):
+def retrieveBoilerplateFile(doc, name, group=None, status=None, error=True, allowLocal=True):
     # Looks in three or four locations, in order:
     # the folder the spec source is in, the group's boilerplate folder, the megagroup's boilerplate folder, and the generic boilerplate folder.
     # In each location, it first looks for the file specialized on status, and then for the generic file.
@@ -77,7 +77,7 @@ def retrieveBoilerplateFile(doc, name, group=None, status=None, error=True):
             status = doc.md.rawStatus
     megaGroup, status = splitStatus(status)
 
-    searchLocally = doc.md.localBoilerplate[name]
+    searchLocally = allowLocal and doc.md.localBoilerplate[name]
 
     def boilerplatePath(*segs):
         return scriptPath("boilerplate", *segs)
@@ -101,13 +101,13 @@ def boilerplatePath(*segs):
                 # We should remove this after giving specs time to react to the warning:
                 sources.append(doc.inputSource.relative(f))
     if group:
-        sources.append(InputSource(boilerplatePath(group, statusFile)))
-        sources.append(InputSource(boilerplatePath(group, genericFile)))
+        sources.append(InputSource(boilerplatePath(group, statusFile), chroot=False))
+        sources.append(InputSource(boilerplatePath(group, genericFile), chroot=False))
     if megaGroup:
-        sources.append(InputSource(boilerplatePath(megaGroup, statusFile)))
-        sources.append(InputSource(boilerplatePath(megaGroup, genericFile)))
-    sources.append(InputSource(boilerplatePath(statusFile)))
-    sources.append(InputSource(boilerplatePath(genericFile)))
+        sources.append(InputSource(boilerplatePath(megaGroup, statusFile), chroot=False))
+        sources.append(InputSource(boilerplatePath(megaGroup, genericFile), chroot=False))
+    sources.append(InputSource(boilerplatePath(statusFile), chroot=False))
+    sources.append(InputSource(boilerplatePath(genericFile), chroot=False))
 
     # Watch all the possible sources, not just the one that got used, because if
     # an earlier one appears, we want to rebuild.

bikeshed/constants.py

@@ -9,6 +9,8 @@
 biblioDisplay = StringEnum("index", "inline")
 specClass = None
 testAnnotationURL = "https://test.csswg.org/harness/annotate.js"
+chroot = True
+executeCode = False
 
 
 def errorLevelAt(target):

bikeshed/extensions.py

@@ -1,8 +1,9 @@
 from . import config
+from . import constants
 from .h import *  # noqa: F401
 from .messages import *  # noqa: F401
 
 
 def load(doc):
-    code = config.retrieveBoilerplateFile(doc, "bs-extensions")
+    code = config.retrieveBoilerplateFile(doc, "bs-extensions", allowLocal=constants.executeCode)
     exec(code, globals())

bikeshed/inlineTags/__init__.py

@@ -1,11 +1,15 @@
 from subprocess import PIPE, Popen
 
+from .. import constants
 from ..h import *
 from ..messages import *
 
 
 def processTags(doc):
     for el in findAll("[data-span-tag]", doc):
+        if not constants.executeCode:
+            die("Found an inline code tag, but arbitrary code execution isn't allowed. See the --allow-execute flag.")
+            return
         tag = el.get("data-span-tag")
         if tag not in doc.md.inlineTagCommands:
             die("Unknown inline tag '{0}' found:\n  {1}", tag, outerHTML(el), el=el)