Python: Model file threat-model

python/ql/lib/semmle/python/frameworks/Stdlib.model.yml

@@ -17,6 +17,8 @@ extensions:
 
       # if no argument is given, the default is to use sys.argv[1:]
       - ['argparse.ArgumentParser', 'Member[parse_args,parse_known_args].WithArity[0].ReturnValue', 'commandargs']
+
+      - ['os', 'Member[read].ReturnValue', 'file']
   - addsTo:
       pack: codeql/python-all
       extensible: summaryModel

python/ql/lib/semmle/python/frameworks/Stdlib.qll

@@ -1499,13 +1499,17 @@ module StdlibPrivate {
    * See https://docs.python.org/3/library/functions.html#open
    */
   private class OpenCall extends FileSystemAccess::Range, Stdlib::FileLikeObject::InstanceSource,
-    DataFlow::CallCfgNode
+    ThreatModelSource::Range, DataFlow::CallCfgNode
   {
     OpenCall() { this = getOpenFunctionRef().getACall() }
 
     override DataFlow::Node getAPathArgument() {
       result in [this.getArg(0), this.getArgByName("file")]
     }
+
+    override string getThreatModel() { result = "file" }
+
+    override string getSourceType() { result = "open()" }
   }
 
   /**

python/ql/test/library-tests/frameworks/stdlib/FileSystemAccess.py

@@ -5,25 +5,25 @@ import stat
 import tempfile
 import shutil
 
-open("file")  # $ getAPathArgument="file"
-open(file="file")  # $ getAPathArgument="file"
+open("file")  # $ getAPathArgument="file" threatModelSource[file]=open(..)
+open(file="file")  # $ getAPathArgument="file" threatModelSource[file]=open(..)
 
 o = open
 
-o("file")  # $ getAPathArgument="file"
-o(file="file")  # $ getAPathArgument="file"
+o("file")  # $ getAPathArgument="file" threatModelSource[file]=o(..)
+o(file="file")  # $ getAPathArgument="file" threatModelSource[file]=o(..)
 
 
-builtins.open("file")  # $ getAPathArgument="file"
-builtins.open(file="file")  # $ getAPathArgument="file"
+builtins.open("file")  # $ getAPathArgument="file" threatModelSource[file]=builtins.open(..)
+builtins.open(file="file")  # $ getAPathArgument="file" threatModelSource[file]=builtins.open(..)
 
 
-io.open("file")  # $ getAPathArgument="file"
-io.open(file="file")  # $ getAPathArgument="file"
+io.open("file")  # $ getAPathArgument="file" threatModelSource[file]=io.open(..)
+io.open(file="file")  # $ getAPathArgument="file" threatModelSource[file]=io.open(..)
 io.open_code("file")  # $ getAPathArgument="file"
 io.FileIO("file")  # $ getAPathArgument="file"
 
-f = open("path") # $ getAPathArgument="path"
+f = open("path") # $ getAPathArgument="path" threatModelSource[file]=open(..)
 f.write("foo") # $ getAPathArgument="path" fileWriteData="foo"
 lines = ["foo"]
 f.writelines(lines) # $ getAPathArgument="path" fileWriteData=lines
@@ -87,8 +87,8 @@ def test_fspath():
         os.fspath(path=TAINTED_STRING), # $ tainted
     )
 
-os.open("path", os.O_RDONLY) # $ getAPathArgument="path"
-os.open(path="path", flags=os.O_RDONLY) # $ getAPathArgument="path"
+os.open("path", os.O_RDONLY) # $ getAPathArgument="path" SPURIOUS: threatModelSource[file]=os.open(..)
+os.open(path="path", flags=os.O_RDONLY) # $ getAPathArgument="path" SPURIOUS: threatModelSource[file]=os.open(..)
 
 os.access("path", os.R_OK) # $ getAPathArgument="path"
 os.access(path="path", mode=os.R_OK) # $ getAPathArgument="path"

python/ql/test/library-tests/frameworks/stdlib/threat_models.py

@@ -48,6 +48,19 @@ ensure_tainted(
     input(), # $ tainted threatModelSource[stdin]=input()
 )
 
+########################################
+# reading data from files
+########################################
+
+ensure_tainted(
+    open("foo"), # $ tainted threatModelSource[file]=open(..) getAPathArgument="foo"
+    open("foo").read(), # $ tainted threatModelSource[file]=open(..) getAPathArgument="foo"
+    open("foo").readline(), # $ tainted threatModelSource[file]=open(..) getAPathArgument="foo"
+    open("foo").readlines(), # $ tainted threatModelSource[file]=open(..) getAPathArgument="foo"
+
+    os.read(os.open("foo"), 1024), # $ tainted threatModelSource[file]=os.read(..) SPURIOUS: threatModelSource[file]=os.open(..) getAPathArgument="foo"
+)
+
 ########################################
 # socket
 ########################################