tree structure for comparison changed text files

src/plugins/compare/file_coverage/view/file_coverage.html

@@ -75,35 +75,15 @@
 
  {% elif feature == 'changed_text_files' %}
  <tr>
- <td data-toggle="tooltip" data-placement="right" title="text files with identical paths but different content">diverging text files</td>
+ <td data-toggle="tooltip" data-placement="right" title="text files with identical paths but different content">changed text files</td>
  <td colspan="{{ uid_list | length }}">
  <div class="list-group">
  <div role="button" data-toggle="collapse" data-target="#changed_text_files" class="list-group-item list-group-item-primary d-flex justify-content-between align-items-center" aria-expanded="false">
  show files<span class="badge badge-pill badge-primary">{{ result['plugins'][plugin][feature] | length }}</span>
  </div>
  <div id="changed_text_files" class="collapse">
  <div class="list-group list-group-flush">
- <div class="list-group-item list-group-item-action border-top p-0">
- <table class="internal-table table-sm" style="width: 100%;">
- {% for path, uid_tuple_list in result['plugins'][plugin][feature].items() | sort %}
- {% for (uid_1, uid_2) in uid_tuple_list %}
- <tr>
- <td style="width: 50%;">
- <a href="/analysis/{{ uid_1 }}" target="_blank" rel="noopener noreferrer">{{ path }}</a>
- </td>
- <td>
- <a href="/analysis/{{ uid_2 }}" target="_blank" rel="noopener noreferrer">{{ path }}</a>
- </td>
- <td style="width: 140px; text-align: right;">
- <button class="btn btn-primary" onclick=" window.open('/comparison/text_files/{{ uid_1 }}/{{ uid_2 }}','_blank')">
- show file diff
- </button>
- </td>
- </tr>
- {% endfor %}
- {% endfor %}
- </table>
- </div>
+ {{ result['plugins'][plugin][feature] | group_changed_text_files | render_changed_text_files | safe }}
  </div>
  </div>
  </div>

src/test/unit/web_interface/test_filter.py

@@ -302,6 +302,18 @@ def test_get_unique_keys_from_list_of_dicts(list_of_dicts, expected_result):
  assert flt.get_unique_keys_from_list_of_dicts(list_of_dicts) == expected_result
 
 
+@pytest.mark.parametrize(
+ ('input_dict', 'expected_result'),
+ [
+ ({}, {}),
+ ({'a': 1}, {'a': 1}),
+ ({'/a/b/c': 1, '/a/b/d': 2, '/a/e': 3}, {'a': {'b': {'c': 1, 'd': 2}, 'e': 3}}),
+ ],
+)
+def test_grp_changed_text_file_data(input_dict, expected_result):
+ assert flt.group_path_dict_by_dirs(input_dict) == expected_result
+
+
 @pytest.mark.parametrize(
  ('function', 'input_data', 'expected_output', 'error_message'),
  [

src/web_interface/components/jinja_filter.py

@@ -195,6 +195,7 @@ def _setup_filters(self): # noqa: PLR0915
  self._app.jinja_env.filters['get_canvas_height'] = flt.get_canvas_height
  self._app.jinja_env.filters['get_searchable_crypto_block'] = flt.get_searchable_crypto_block
  self._app.jinja_env.filters['get_unique_keys_from_list_of_dicts'] = flt.get_unique_keys_from_list_of_dicts
+ self._app.jinja_env.filters['group_changed_text_files'] = flt.group_path_dict_by_dirs
  self._app.jinja_env.filters['hex'] = hex
  self._app.jinja_env.filters['hide_dts_binary_data'] = flt.hide_dts_binary_data
  self._app.jinja_env.filters['infection_color'] = flt.infection_color
@@ -221,6 +222,7 @@ def _setup_filters(self): # noqa: PLR0915
  self._app.jinja_env.filters['remaining_time'] = elapsed_time
  self._app.jinja_env.filters['render_analysis_tags'] = flt.render_analysis_tags
  self._app.jinja_env.filters['render_general_information'] = self._render_general_information_table
+ self._app.jinja_env.filters['render_changed_text_files'] = flt.render_changed_text_files
  self._app.jinja_env.filters['render_query_title'] = flt.render_query_title
  self._app.jinja_env.filters['render_fw_tags'] = flt.render_fw_tags
  self._app.jinja_env.filters['replace_comparison_uid_with_hid'] = self._filter_replace_comparison_uid_with_hid

src/web_interface/filter.py

@@ -3,6 +3,7 @@
 import binascii
 import json
 import logging
+import os
 import random
 import re
 import stat
@@ -366,6 +367,20 @@ def get_unique_keys_from_list_of_dicts(list_of_dicts: list[dict]):
  return unique_keys
 
 
+def group_path_dict_by_dirs(data: dict[str, list[tuple]]) -> dict:
+ # groups paths into folders, resulting in a tree structure
+ result = {}
+ for path, value in data.items():
+ current = result
+ # path is expected to be of structure /dir1/dir2/.../file
+ *folders, file = path.split('/')
+ for dir_name in folders:
+ if dir_name:
+ current = current.setdefault(dir_name, {})
+ current[file] = value
+ return result
+
+
 def random_collapse_id():
  return ''.join(random.choice(ascii_letters) for _ in range(10))
 
@@ -387,6 +402,65 @@ def format_duration(duration: float) -> str:
  return str(timedelta(seconds=duration))
 
 
+def render_changed_text_files(changed_text_files: dict) -> str:
+ elements = []
+ for key, value in sorted(changed_text_files.items()):
+ if isinstance(value, list):
+ # file tuple list (represents a leaf/file in the file tree)
+ lines = ['<table class="internal-table table-sm" style="width: 100%;">']
+ for uid_1, uid_2 in value:
+ lines.extend(
+ [
+ '<tr>',
+ ' <td style="width: 50%;">',
+ f' <a href="/analysis/{uid_1}" target="_blank" rel="noopener noreferrer">{key}</a>',
+ ' </td>',
+ ' <td>',
+ f' <a href="/analysis/{uid_2}" target="_blank" rel="noopener noreferrer">{key}</a>',
+ ' </td>',
+ ' <td style="width: 140px; text-align: right;">',
+ f' <button class="btn btn-primary" onclick=" '
+ f" window.open('/comparison/text_files/{uid_1}/{uid_2}','_blank')\">",
+ ' show file diff',
+ ' </button>',
+ ' </td>',
+ '</tr>',
+ ]
+ )
+ lines.append('</table>')
+ element = '\n'.join(lines)
+ else:
+ # directory dict (represents an inner node/folder in the file tree)
+ inner = render_changed_text_files(value)
+ id_ = f'ctf-{os.urandom(8).hex()}'
+ count = _count_changed_files(value)
+ element = (
+ f'<div class="p-1" data-toggle="collapse" data-target="#{id_}">'
+ f' ╰ {key} <i class="fas fa-caret-down"></i>'
+ f' <span class="badge badge-pill badge-secondary">{count}</span>'
+ f'</div>\n'
+ f'<div id="{id_}" class="collapse list-group list-group-flush">\n'
+ f' {inner}\n'
+ '</div>\n'
+ )
+ elements.append(
+ f'<div class="list-group-item list-group-item-action border-top" style="padding: 0 0 0 25px">\n'
+ f'{element}\n'
+ '</div>'
+ )
+ return '\n'.join(elements)
+
+
+def _count_changed_files(ctf_dict: dict) -> int:
+ count = 0
+ for value in ctf_dict.values():
+ if isinstance(value, dict):
+ count += _count_changed_files(value)
+ elif isinstance(value, list):
+ count += len(value)
+ return count
+
+
 def render_query_title(query_title: None | str | dict):
  if query_title is None:
  return None