Python: Add test of stdlib HTTP server facilities

Just a port of the old tests, except for the fact that I learned
`cgi.FieldStorage()` _should_ be tainted when not specifying any arguments. (and
moved taint-test to own function)

Also clarified how imports of all the .*HTTPRequestHandler works in Python2

python/ql/test/experimental/library-tests/frameworks/stdlib/TestTaint.expected

@@ -2,3 +2,36 @@
 | CodeExecution.py:36 | ok   | test_additional_taint | cmd1 |
 | CodeExecution.py:37 | ok   | test_additional_taint | cmd2 |
 | CodeExecution.py:38 | ok   | test_additional_taint | cmd3 |
+| http_server.py:22 | fail | test_cgi_FieldStorage_taint | form |
+| http_server.py:24 | fail | test_cgi_FieldStorage_taint | form['key'] |
+| http_server.py:25 | fail | test_cgi_FieldStorage_taint | form['key'].value |
+| http_server.py:26 | fail | test_cgi_FieldStorage_taint | form['key'].file |
+| http_server.py:27 | fail | test_cgi_FieldStorage_taint | form['key'].filename |
+| http_server.py:28 | fail | test_cgi_FieldStorage_taint | form['key'][0] |
+| http_server.py:29 | fail | test_cgi_FieldStorage_taint | form['key'][0].value |
+| http_server.py:30 | fail | test_cgi_FieldStorage_taint | form['key'][0].file |
+| http_server.py:31 | fail | test_cgi_FieldStorage_taint | form['key'][0].filename |
+| http_server.py:32 | fail | test_cgi_FieldStorage_taint | ListComp |
+| http_server.py:34 | fail | test_cgi_FieldStorage_taint | form.getvalue(..) |
+| http_server.py:35 | fail | test_cgi_FieldStorage_taint | form.getvalue(..)[0] |
+| http_server.py:37 | fail | test_cgi_FieldStorage_taint | form.getfirst(..) |
+| http_server.py:39 | fail | test_cgi_FieldStorage_taint | form.getlist(..) |
+| http_server.py:40 | fail | test_cgi_FieldStorage_taint | form.getlist(..)[0] |
+| http_server.py:41 | fail | test_cgi_FieldStorage_taint | ListComp |
+| http_server.py:50 | fail | taint_sources | self |
+| http_server.py:52 | fail | taint_sources | self.requestline |
+| http_server.py:54 | fail | taint_sources | self.path |
+| http_server.py:56 | fail | taint_sources | self.headers |
+| http_server.py:57 | fail | taint_sources | self.headers['Foo'] |
+| http_server.py:58 | fail | taint_sources | self.headers.get(..) |
+| http_server.py:59 | fail | taint_sources | self.headers.get_all(..) |
+| http_server.py:60 | fail | taint_sources | self.headers.keys() |
+| http_server.py:61 | fail | taint_sources | self.headers.values() |
+| http_server.py:62 | fail | taint_sources | self.headers.items() |
+| http_server.py:63 | fail | taint_sources | self.headers.as_bytes() |
+| http_server.py:64 | fail | taint_sources | self.headers.as_string() |
+| http_server.py:65 | fail | taint_sources | str(..) |
+| http_server.py:66 | fail | taint_sources | bytes(..) |
+| http_server.py:68 | fail | taint_sources | self.rfile |
+| http_server.py:69 | fail | taint_sources | self.rfile.read() |
+| http_server.py:78 | fail | taint_sources | form |

python/ql/test/experimental/library-tests/frameworks/stdlib/TestTaint.ql

@@ -1,2 +1,9 @@
 import experimental.dataflow.tainttracking.TestTaintLib
 import semmle.python.dataflow.new.RemoteFlowSources
+
+class WithRemoteFlowSources extends TestTaintTrackingConfiguration {
+  override predicate isSource(DataFlow::Node source) {
+    super.isSource(source) or
+    source instanceof RemoteFlowSource
+  }
+}

python/ql/test/experimental/library-tests/frameworks/stdlib/http_server.py

@@ -0,0 +1,122 @@
+import sys
+import os
+import cgi
+
+if sys.version_info[0] == 2:
+    from BaseHTTPServer import BaseHTTPRequestHandler
+    from BaseHTTPServer import HTTPServer
+    from SimpleHTTPServer import SimpleHTTPRequestHandler
+    from CGIHTTPServer import CGIHTTPRequestHandler
+
+if sys.version_info[0] == 3:
+    from http.server import HTTPServer, BaseHTTPRequestHandler, SimpleHTTPRequestHandler, CGIHTTPRequestHandler
+
+
+def test_cgi_FieldStorage_taint():
+    # When a python script is invoked through CGI, the default values used by
+    # `cgi.FieldStorage` constructor makes it handle data from incoming request.
+    # You _can_ also manually set the input-data, as is shown below in `MyHandler`.
+    form = cgi.FieldStorage()
+
+    ensure_tainted(
+        form,
+
+        form['key'], # will be a list, if multiple fields named "key" are provided
+        form['key'].value,
+        form['key'].file,
+        form['key'].filename,
+        form['key'][0],
+        form['key'][0].value,
+        form['key'][0].file,
+        form['key'][0].filename,
+        [field.value for field in form['key']],
+
+        form.getvalue('key'), # will be a list, if multiple fields named "key" are provided
+        form.getvalue('key')[0],
+
+        form.getfirst('key'),
+
+        form.getlist('key'),
+        form.getlist('key')[0],
+        [field.value for field in form.getlist('key')],
+    )
+
+
+class MyHandler(BaseHTTPRequestHandler):
+
+    def taint_sources(self):
+
+        ensure_tainted(
+            self,
+
+            self.requestline,
+
+            self.path,
+
+            self.headers,
+            self.headers['Foo'],
+            self.headers.get('Foo'),
+            self.headers.get_all('Foo'),
+            self.headers.keys(),
+            self.headers.values(),
+            self.headers.items(),
+            self.headers.as_bytes(),
+            self.headers.as_string(),
+            str(self.headers),
+            bytes(self.headers),
+
+            self.rfile,
+            self.rfile.read(),
+        )
+
+        form = cgi.FieldStorage(
+            self.rfile,
+            self.headers,
+            environ={'REQUEST_METHOD': 'POST', 'CONTENT_TYPE': self.headers.get('content-type')},
+        )
+
+        ensure_tainted(form)
+
+
+    def do_GET(self):
+        # send_response will log a line to stderr
+        self.send_response(200)
+        self.send_header("Content-type", "text/plain; charset=utf-8")
+        self.end_headers()
+        self.wfile.write(b"Hello BaseHTTPRequestHandler\n")
+        self.wfile.writelines([b"1\n", b"2\n", b"3\n"])
+        print(self.headers)
+
+
+    def do_POST(self):
+        form = cgi.FieldStorage(
+            self.rfile,
+            self.headers,
+            environ={'REQUEST_METHOD': 'POST', 'CONTENT_TYPE': self.headers.get('content-type')},
+        )
+
+        if 'myfile' not in form:
+            self.send_response(422)
+            self.end_headers()
+            return
+
+        field = form['myfile']
+
+        field.file.seek(0, os.SEEK_END)
+        filesize = field.file.tell()
+
+        print("Uploaded {!r} with {} bytes".format(field.filename, filesize))
+
+        self.send_response(200)
+        self.end_headers()
+
+
+if __name__ == "__main__":
+    server = HTTPServer(("127.0.0.1", 8080), MyHandler)
+    server.serve_forever()
+
+    # Headers works case insensitvely, so self.headers['foo'] == self.headers['FOO']
+    # curl localhost:8080 --header "Foo: 1" --header "foo: 2"
+
+    # To test file submission through forms, use
+    # curl -F myfile=@<yourfile> localhost:8080