Improving MIME type detection for some common file types when they are missing from the system MIME database

Also allow the user to specify their own database via mime.types in the
kitty config directory. See #3056

docs/changelog.rst

@@ -24,6 +24,9 @@ To update |kitty|, :doc:`follow the instructions <binary>`.
 - Fix selections created by dragging upwards not being auto-cleared when
   screen contents change (:pull:`3028`)
 
+- Allow adding MIME definitions to kitty by placing a ``mime.types`` file in
+  the kitty config directory (:iss:`3056`)
+
 
 0.19.1 [2020-10-06]
 -------------------

docs/open_actions.rst

@@ -81,7 +81,11 @@ lines. The various available criteria are:
 
 ``mime``
     A comma separated list of MIME types, for example: ``text/*, image/*,
-    application/pdf``
+    application/pdf``. You can add MIME types to kitty by creating the
+    :file:`mime.types` in the kitty configuration directory. Useful if your
+    system MIME database does not have definitions you need. This file is
+    in the standard format of one definition per line, like: ``text/plain rst
+    md``.
 
 ``ext``
     A comma separated list of file extensions, for example: ``jpeg, tar.gz``

kittens/diff/collect.py

@@ -7,7 +7,7 @@ import re
 from contextlib import suppress
 from functools import lru_cache
 from hashlib import md5
-from mimetypes import guess_type
+from kitty.guess_mime_type import guess_type
 from typing import TYPE_CHECKING, Dict, List, Set, Optional, Iterator, Tuple, Union
 
 if TYPE_CHECKING:
@@ -136,7 +136,7 @@ def sanitize(text: str) -> str:
 
 @lru_cache(maxsize=1024)
 def mime_type_for_path(path: str) -> str:
-    return guess_type(path)[0] or 'application/octet-stream'
+    return guess_type(path) or 'application/octet-stream'
 
 
 @lru_cache(maxsize=1024)

kittens/icat/main.py

@@ -3,7 +3,6 @@
 # License: GPL v3 Copyright: 2017, Kovid Goyal <kovid at kovidgoyal.net>
 
 import contextlib
-import mimetypes
 import os
 import re
 import socket
@@ -18,6 +17,7 @@ from typing import (
     Dict, Generator, List, NamedTuple, Optional, Pattern, Tuple, Union
 )
 
+from kitty.guess_mime_type import guess_type
 from kitty.cli import parse_args
 from kitty.cli_stub import IcatCLIOptions
 from kitty.constants import appname
@@ -273,7 +273,7 @@ def process(path: str, args: IcatCLIOptions, parsed_opts: ParsedOpts, is_tempfil
 def scan(d: str) -> Generator[Tuple[str, str], None, None]:
     for dirpath, dirnames, filenames in os.walk(d):
         for f in filenames:
-            mt = mimetypes.guess_type(f)[0]
+            mt = guess_type(f)
             if mt and mt.startswith('image/'):
                 yield os.path.join(dirpath, f), mt
 

kitty/complete.py

@@ -394,10 +394,10 @@ def complete_files_and_dirs(
 
 
 def complete_icat_args(ans: Completions, opt: Optional[OptionDict], prefix: str, unknown_args: Delegate) -> None:
-    from mimetypes import guess_type
+    from .guess_mime_type import guess_type
 
     def icat_file_predicate(filename: str) -> bool:
-        mt = guess_type(filename)[0]
+        mt = guess_type(filename)
         if mt and mt.startswith('image/'):
             return True
         return False

kitty/guess_mime_type.py

@@ -0,0 +1,45 @@
+#!/usr/bin/env python
+# vim:fileencoding=utf-8
+# License: GPLv3 Copyright: 2020, Kovid Goyal <kovid at kovidgoyal.net>
+
+import os
+from contextlib import suppress
+from typing import Optional
+
+known_extensions = {
+    'asciidoc': 'text/asciidoctor',
+    'conf': 'text/config',
+    'md': 'text/markdown',
+    'pyj': 'text/rapydscript-ng',
+    'recipe': 'text/python',
+    'rst': 'text/restructured-text',
+    'toml': 'text/toml',
+    'vim': 'text/vim',
+    'yaml': 'text/yaml',
+}
+
+
+def is_rc_file(path: str) -> bool:
+    name = os.path.basename(path)
+    return '.' not in name and name.endswith('rc')
+
+
+def guess_type(path: str) -> Optional[str]:
+    if not hasattr(guess_type, 'inited'):
+        setattr(guess_type, 'inited', True)
+        from mimetypes import init
+        init(None)
+        from kitty.constants import config_dir
+        local_defs = os.path.join(config_dir, 'mime.types')
+        if os.path.exists(local_defs):
+            init((local_defs,))
+    from mimetypes import guess_type as stdlib_guess_type
+    mt = None
+    with suppress(Exception):
+        mt = stdlib_guess_type(path)[0]
+    if not mt:
+        ext = path.rpartition('.')[-1].lower()
+        mt = known_extensions.get(ext)
+    if not mt and is_rc_file(path):
+        mt = 'text/plain'
+    return mt

kitty/open_actions.py

@@ -17,6 +17,7 @@ from .config import KeyAction, parse_key_action
 from .constants import config_dir
 from .typing import MatchType
 from .utils import expandvars, log_error
+from .guess_mime_type import guess_type
 
 
 class MatchCriteria(NamedTuple):
@@ -75,12 +76,8 @@ def url_matches_criterion(purl: 'ParseResult', url: str, unquoted_path: str, mc:
 
     if mc.type == 'mime':
         import fnmatch
-        from mimetypes import guess_type
-        try:
-            mt = guess_type(unquoted_path)[0]
-        except Exception:
-            return False
-        if mt is None:
+        mt = guess_type(unquoted_path)
+        if not mt:
             return False
         mt = mt.lower()
         for mpat in mc.value.split(','):