fix

Signed-off-by: Jan Kowalleck <[email protected]>

Author: jkowalleck

cyclonedx_py/_internal/utils/mimetypes.py

@@ -19,22 +19,40 @@
 from os.path import splitext
 from typing import Optional
 
+_MIME_TEXT_PLAIN = 'text/plain'
+
 _MAP_EXT_MIME = {
     # https://www.iana.org/assignments/media-types/media-types.xhtml
+    '.csv': 'text/csv',
+    '.htm': 'text/html',
+    '.html': 'text/html',
     '.md': 'text/markdown',
     '.txt': 'text/plain',
     '.rst': 'text/prs.fallenstein.rst',
+    '.xml': 'text/xml',  # not `application/xml` -- our scope is text!
+    # license-specific files
+    '.license': _MIME_TEXT_PLAIN,
+    '.licence': _MIME_TEXT_PLAIN,
     # add more mime types. pull-requests welcome!
 }
 
+_LICENSE_FNAME_BASE = ('licence', 'license')
+_LICENSE_FNAME_EXT = (
+    '.apache',
+    '.bsd',
+    '.gpl',
+    '.mit',
+)
+
 
 def guess_type(file_name: str) -> Optional[str]:
     """
     The stdlib `mimetypes.guess_type()` is inconsistent, as it depends heavily on type registry in the env/os.
     Therefore, this polyfill exists.
     """
-    ext = splitext(file_name)[1].lower()
-    return _MAP_EXT_MIME.get(
-        ext,
-        _stdlib_guess_type(file_name)[0]
-    )
+    file_name_l = file_name.lower()
+    base, ext = splitext(file_name_l)
+    if base in _LICENSE_FNAME_BASE and ext in _LICENSE_FNAME_EXT:
+        return _MIME_TEXT_PLAIN
+    return _MAP_EXT_MIME.get(ext) \
+        or _stdlib_guess_type(file_name_l)[0]