Use a libmagic wrapper to guess Content-Type headers.

README.md

@@ -18,6 +18,7 @@ This project intends to provide a well-tested and maintainable alternative to
 
 - Python 3 (>=3.6)
 - Python dependencies:
+  - file-magic
   - jinja2
   
 ## Usage

matemat/webserver/httpd.py

@@ -4,7 +4,7 @@ from typing import Any, Callable, Dict, Tuple, Type, Union
 import logging
 import os
 import socket
-import mimetypes
+import magic
 from socketserver import TCPServer
 from http.server import HTTPServer, BaseHTTPRequestHandler
 from http.cookies import SimpleCookie
@@ -308,7 +308,6 @@ class HttpHandler(BaseHTTPRequestHandler):
         if path in _PAGELET_PATHS:
             # Prepare some headers.  Those can still be overwritten by the pagelet
             headers: Dict[str, str] = {
-                'Content-Type': 'text/html',
                 'Cache-Control': 'no-cache'
             }
             # Call the pagelet function
@@ -328,6 +327,10 @@ class HttpHandler(BaseHTTPRequestHandler):
                              f'matemat_session_id={session_id}; expires={expires}')
             # Compute the body length and add the appropriate header
             headers['Content-Length'] = str(len(data))
+            # If the pagelet did not set its own Content-Type header, use libmagic to guess an appropriate one
+            if 'Content-Type' not in headers:
+                filemagic: magic.FileMagic = magic.detect_from_content(data)
+                headers['Content-Type'] = f'{filemagic.mime_type}; charset={filemagic.encoding}'
             # Send all headers set by the pagelet
             for name, value in headers.items():
                 self.send_header(name, value)
@@ -365,13 +368,12 @@ class HttpHandler(BaseHTTPRequestHandler):
                     data = f.read()
                 # File read successfully, send 'OK' header
                 self.send_response(200)
-                # TODO: Guess the MIME type.  Unfortunately this call solely relies on the file extension, not ideal?
+                # Guess the MIME type and encoding using libmagic
-                mimetype, _ = mimetypes.guess_type(filepath)
+                filemagic: magic.FileMagic = magic.detect_from_filename(filepath)
-                # Fall back to octet-stream type, if unknown
+                mimetype: str = filemagic.mime_type
-                if mimetype is None:
+                charset: str = filemagic.encoding
-                    mimetype = 'application/octet-stream'
                 # Send content type and length header
-                self.send_header('Content-Type', mimetype)
+                self.send_header('Content-Type', f'{mimetype}; charset={charset}')
                 self.send_header('Content-Length', str(len(data)))
                 self.send_header('Last-Modified', fileage.strftime('%a, %d %b %Y %H:%M:%S GMT'))
                 self.send_header('Cache-Control', 'max-age=1')

matemat/webserver/test/test_serve.py

@@ -17,7 +17,6 @@ def serve_test_pagelet_str(method: str,
                            session_vars: Dict[str, Any],
                            headers: Dict[str, str],
                            pagelet_variables: Dict[str, str]) -> Union[bytes, str, PageletResponse]:
-    headers['Content-Type'] = 'text/plain'
     return 'serve test pagelet str'
 
 
@@ -28,7 +27,7 @@ def serve_test_pagelet_bytes(method: str,
                              session_vars: Dict[str, Any],
                              headers: Dict[str, str],
                              pagelet_variables: Dict[str, str]) -> Union[bytes, str, PageletResponse]:
-    headers['Content-Type'] = 'application/octet-stream'
+    headers['Content-Type'] = 'application/x-foo-bar'
     return b'serve\x80test\xffpagelet\xfebytes'
 
 
@@ -49,7 +48,6 @@ def serve_test_pagelet_template(method: str,
                                 session_vars: Dict[str, Any],
                                 headers: Dict[str, str],
                                 pagelet_variables: Dict[str, str]) -> Union[bytes, str, PageletResponse]:
-    headers['Content-Type'] = 'text/plain'
     return TemplateResponse('test.txt', what='World')
 
 
@@ -62,7 +60,6 @@ def serve_test_pagelet_fail(method: str,
                             headers: Dict[str, str],
                             pagelet_variables: Dict[str, str]) -> Union[bytes, str, PageletResponse]:
     session_vars['test'] = 'hello, world!'
-    headers['Content-Type'] = 'text/plain'
     raise HttpException(599, 'Error expected during unit testing')
 
 
@@ -228,3 +225,17 @@ class TestServe(AbstractHttpdTest):
         self.assertIsNone(packet.pagelet)
         # Make sure a 405 Method Not Allowed header is served
         self.assertEqual(405, packet.statuscode)
+
+    def test_serve_static_libmagic(self):
+        # The correct Content-Type header must be guessed, if a pagelet does not provide one
+        self.client_sock.set_request(b'GET /just/testing/serve_pagelet_str HTTP/1.1\r\n\r\n')
+        HttpHandler(self.client_sock, ('::1', 45678), self.server)
+        packet = self.client_sock.get_response()
+        self.assertEqual('text/plain; charset=us-ascii', packet.headers['Content-Type'])
+
+    def test_serve_static_libmagic_skipped(self):
+        # The Content-Type set by a pagelet should not be overwritten
+        self.client_sock.set_request(b'GET /just/testing/serve_pagelet_bytes HTTP/1.1\r\n\r\n')
+        HttpHandler(self.client_sock, ('::1', 45678), self.server)
+        packet = self.client_sock.get_response()
+        self.assertEqual('application/x-foo-bar', packet.headers['Content-Type'])

requirements.txt

@@ -1 +1,2 @@
+file-magic
 jinja2