Merge branch '18-libmagic-mime-type-detection' into 'master'

Resolve "Use a libmagic wrapper for MIME type detection"

Closes #18

See merge request s3lph/matemat!19

.gitlab-ci.yml

@@ -1,5 +1,5 @@
 ---
-image: s3lph/matemat-ci:20180711-02
+image: s3lph/matemat-ci:20180720-01
 
 stages:
 - test

Dockerfile

@@ -2,8 +2,9 @@
 FROM python:3.6-alpine
 
 RUN mkdir -p /var/matemat/db /var/matemat/upload
+RUN apk --update add libmagic
 ADD . /
 RUN pip3 install -r /requirements.txt
 
 EXPOSE 80/tcp
-CMD [ "/usr/local/bin/python3", "-m", "matemat", "/etc/matemat.conf", "/matemat.docker.conf" ]
+CMD [ "/run.sh" ]

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

@@ -5,6 +5,7 @@ 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 +309,6 @@ 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 +328,20 @@ 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:
+                try:
+                    filemagic: magic.FileMagic = magic.detect_from_content(data)
+                    mimetype: str = filemagic.mime_type
+                    charset: str = filemagic.encoding
+                except ValueError:
+                    mimetype = 'application/octet-stream'
+                    charset = 'binary'
+                # Only append the charset if it is not "binary"
+                if charset == 'binary':
+                    headers['Content-Type'] = mimetype
+                else:
+                    headers['Content-Type'] = f'{mimetype}; charset={charset}'
             # Send all headers set by the pagelet
             for name, value in headers.items():
                 self.send_header(name, value)
@@ -365,13 +379,21 @@ 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?
-                mimetype, _ = mimetypes.guess_type(filepath)
-                # Fall back to octet-stream type, if unknown
-                if mimetype is None:
+                # Guess the MIME type by file extension, or use libmagic as fallback
+                # Use libmagic to guess the charset
+                try:
+                    exttype: str = mimetypes.guess_type(filepath)[0]
+                    filemagic: magic.FileMagic = magic.detect_from_filename(filepath)
+                    mimetype: str = exttype if exttype is not None else filemagic.mime_type
+                    charset: str = filemagic.encoding
+                except ValueError:
                     mimetype = 'application/octet-stream'
-                # Send content type and length header
+                    charset = 'binary'
+                # Send content type and length header.  Only set the charset if it's not "binary"
+                if charset == 'binary':
                     self.send_header('Content-Type', mimetype)
+                else:
+                    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')
 
 
@@ -81,6 +78,15 @@ class TestServe(AbstractHttpdTest):
         with open(forbidden, 'w') as f:
             f.write('This should not be readable')
         os.chmod(forbidden, 0)
+        # Create a CSS resource whose MIME type should be detected by file extension
+        with open(os.path.join(self.tempdir.name, 'teststyle.css'), 'w') as f:
+            f.write('.ninja { display: none; }\n')
+        # Create a file without extension (containing UTF-16 text with BOM); libmagic should take over
+        with open(os.path.join(self.tempdir.name, 'testdata'), 'wb') as f:
+            f.write(b'\xFE\xFFH\x00e\x00l\x00l\x00o\x00,\x00 \x00w\x00o\x00r\x00l\x00d\x00!\x00\n\x00')
+        # Create a file that will yield "text/plain; charset=binary"
+        with open(os.path.join(self.tempdir.name, 'testbin.txt'), 'wb') as f:
+            f.write(b'\x00\x00\x00\x00\x00\x00\x00\x00')
 
     def test_serve_pagelet_str(self):
         # Call the test pagelet that produces a 200 OK result
@@ -228,3 +234,41 @@ 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_pagelet_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_pagelet_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'])
+
+    def test_serve_static_mime_extension(self):
+        # The correct Content-Type should be guessed by file extension primarily
+        self.client_sock.set_request(b'GET /teststyle.css HTTP/1.1\r\n\r\n')
+        HttpHandler(self.client_sock, ('::1', 45678), self.server)
+        packet = self.client_sock.get_response()
+        # libmagic would say text/plain instead
+        self.assertEqual('text/css; charset=us-ascii', packet.headers['Content-Type'])
+
+    def test_serve_static_mime_magic(self):
+        # The correct Content-Type should be guessed by file extension primarily
+        self.client_sock.set_request(b'GET /testdata HTTP/1.1\r\n\r\n')
+        HttpHandler(self.client_sock, ('::1', 45678), self.server)
+        packet = self.client_sock.get_response()
+        # Extension-based would fail, as there is no extension
+        self.assertEqual('text/plain; charset=utf-16be', packet.headers['Content-Type'])
+
+    def test_serve_static_mime_magic_binary(self):
+        # The correct Content-Type should be guessed by file extension primarily
+        self.client_sock.set_request(b'GET /testbin.txt HTTP/1.1\r\n\r\n')
+        HttpHandler(self.client_sock, ('::1', 45678), self.server)
+        packet = self.client_sock.get_response()
+        # No charset should be in the header.  Yes, this is a stupid example
+        self.assertEqual('text/plain', packet.headers['Content-Type'])

requirements.txt

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

run.sh

@@ -0,0 +1,5 @@
+#!/bin/sh
+
+export LD_PRELOAD=/usr/lib/libmagic.so.1
+
+/usr/local/bin/python3 -m matemat /etc/matemat.conf /matemat.docker.conf

testing/Dockerfile

@@ -5,7 +5,7 @@ RUN useradd -d /home/matemat -m matemat
 RUN mkdir -p /var/matemat/db && chown matemat:matemat -R /var/matemat/db
 RUN mkdir -p /var/matemat/upload && chown matemat:matemat -R /var/matemat/upload
 RUN apt-get update -qy
-RUN apt-get install -y --no-install-recommends sudo openssh-client git docker.io python3-dev python3-pip python3-coverage python3-setuptools build-essential
+RUN apt-get install -y --no-install-recommends file sudo openssh-client git docker.io python3-dev python3-pip python3-coverage python3-setuptools build-essential
 RUN pip3 install wheel pycodestyle mypy
 
 WORKDIR /home/matemat