fix: preview of certain files were not shown as text (#1315)

e.g. javascript and json

src/test/unit/web_interface/test_filter.py

@@ -6,6 +6,7 @@
 import pytest
 
 import web_interface.filter as flt
+from test.common_helper import create_test_file_object
 
 UNSORTABLE_LIST = [[], ()]
 
@@ -496,3 +497,18 @@ def test_str_to_hex(input_, expected_result):
 )
 def test_octal_to_readable(input_, include_type, expected_result):
     assert flt.octal_to_readable(input_, include_type=include_type) == expected_result
+
+
+@pytest.mark.parametrize(
+    ('type_analysis', 'expected_result'),
+    [
+        ({'file_type': {}}, False),
+        ({'file_type': {'result': {'mime': 'image/png'}}}, True),
+        ({'file_type': {'result': {'mime': 'text/plain'}}}, True),
+        ({'file_type': {'result': {'mime': 'application/octet-stream'}}}, False),
+        ({'file_type': {'result': {'mime': 'application/javascript'}}}, True),
+    ],
+)
+def test_is_text_file_or_image(type_analysis, expected_result):
+    fo = create_test_file_object(analyses=type_analysis)
+    assert flt.is_text_file_or_image(fo) == expected_result

src/web_interface/components/ajax_routes.py

@@ -12,7 +12,12 @@
 from web_interface.file_tree.file_tree import remove_virtual_path_from_root
 from web_interface.file_tree.file_tree_node import FileTreeNode
 from web_interface.file_tree.jstree_conversion import convert_to_jstree_node
-from web_interface.filter import bytes_to_str_filter, encode_base64_filter
+from web_interface.filter import (
+    bytes_to_str_filter,
+    encode_base64_filter,
+    is_image,
+    is_text_file,
+)
 from web_interface.security.decorator import roles_accepted
 from web_interface.security.privileges import PRIVILEGES
 
@@ -89,12 +94,12 @@ def _get_nice_uid_list_html(self, input_data, root_uid):
     def ajax_get_binary(self, mime_type, uid):
         mime_type = mime_type.replace('_', '/')
         binary = self.intercom.get_binary_and_filename(uid)[0]
-        if 'text/' in mime_type:
+        if is_text_file(mime_type):
             return (
                 '<pre class="line_numbering" style="white-space: pre-wrap">'
                 f'{html.escape(bytes_to_str_filter(binary))}</pre>'
             )
-        if 'image/' in mime_type:
+        if is_image(mime_type):
             return (
                 '<div style="display: block; border: 1px solid; border-color: #dddddd; padding: 5px; '
                 f'text-align: center"><img src="data:image/{mime_type[6:]} ;base64,{encode_base64_filter(binary)}" '

src/web_interface/components/jinja_filter.py

@@ -183,6 +183,7 @@ def _setup_filters(self):
                 'hide_dts_binary_data': flt.hide_dts_binary_data,
                 'infection_color': flt.infection_color,
                 'is_list': lambda item: isinstance(item, list),
+                'is_text_file_or_image': flt.is_text_file_or_image,
                 'json_dumps': json.dumps,
                 'link_cve': flt.replace_cve_with_link,
                 'link_cwe': flt.replace_cwe_with_link,

src/web_interface/filter.py

@@ -14,7 +14,7 @@
 from re import Match
 from string import ascii_letters
 from time import localtime, strftime, struct_time, time
-from typing import Iterable, Union
+from typing import TYPE_CHECKING, Iterable, Union
 
 import packaging.version
 import semver
@@ -28,6 +28,9 @@
 from web_interface.security.authentication import user_has_privilege
 from web_interface.security.privileges import PRIVILEGES
 
+if TYPE_CHECKING:
+    from objects.file import FileObject
+
 
 def generic_nice_representation(i):  # noqa: PLR0911
     if isinstance(i, struct_time):
@@ -557,3 +560,25 @@ def _file_mode_to_type(mode: int) -> str:
 
 def str_to_hex(string: str) -> str:
     return string.encode(errors='replace').hex()
+
+
+KNOWN_TEXT_MIME_TYPES = {  # that don't start with text/
+    'application/javascript',
+    'application/json',
+    'application/pgp-keys',
+    'firmware/intel-hex',
+    'message/news',
+}
+
+
+def is_text_file(mime: str) -> bool:
+    return mime.startswith('text/') or mime in KNOWN_TEXT_MIME_TYPES
+
+
+def is_image(mime: str) -> bool:
+    return mime.startswith('image/')
+
+
+def is_text_file_or_image(fo: FileObject) -> bool:
+    mime = fo.processed_analysis.get('file_type', {}).get('result', {}).get('mime', '')
+    return is_image(mime) or is_text_file(mime)

src/web_interface/templates/show_analysis.html

@@ -360,10 +360,8 @@ <h5 class="modal-title">Add analysis to file</h5>
                         <div id="preview-content" class="m-0 mt-2 p-0"></div>
                     </div>
 
-                    {%- set is_text_preview = firmware.processed_analysis["file_type"]["result"]['mime'][0:5] == "text/" or firmware.processed_analysis["file_type"]["result"]['mime'][0:6] == "image/" %}
-
                     <script>
-                        const isTextOrImage = {{ 'true' if is_text_preview else 'false' }};
+                        const isTextOrImage = {{ 'true' if firmware | is_text_file_or_image else 'false' }};
                         let mimeType = '{{ firmware.processed_analysis["file_type"]["result"]["mime"].replace("/", "_") }}';
                     </script>
                     <script type="text/javascript" src="{{ url_for('static', filename='js/show_analysis_preview.js') }}"></script>