GHSL-2021-1017, GHSL-2021-1020, GHSL-2021-1021

Author: martinRenou

nbconvert/exporters/html.py

@@ -5,6 +5,7 @@
 
 import base64
 import json
+from lxml.html.clean import clean_html
 import mimetypes
 import os
 from pathlib import Path
@@ -149,6 +150,14 @@ def _template_name_default(self):
         help="Template specific theme(e.g. the name of a JupyterLab CSS theme distributed as prebuilt extension for the lab template)",
     ).tag(config=True)
 
+    sanitize_html = Bool(
+        False,
+        help=(
+            "Whether the HTML in Markdown cells and cell outputs should be sanitized."
+            "This should be set to True by nbviewer or similar tools."
+        ),
+    ).tag(config=True)
+
     embed_images = Bool(
         False, help="Whether or not to embed images as base64 in markdown cells."
     ).tag(config=True)
@@ -287,4 +296,5 @@ def resources_include_url(name):
         resources["jupyter_widgets_base_url"] = self.jupyter_widgets_base_url
         resources["widget_renderer_url"] = self.widget_renderer_url
         resources["html_manager_semver_range"] = self.html_manager_semver_range
+        resources["should_sanitize_html"] = self.sanitize_html
         return resources

nbconvert/exporters/templateexporter.py

@@ -70,10 +70,9 @@
     "get_metadata": filters.get_metadata,
     "convert_pandoc": filters.convert_pandoc,
     "json_dumps": json.dumps,
-    # browsers will parse </script>, closing a script tag early
-    # Since JSON allows escaping forward slash, this will still be parsed by JSON
-    "escape_html_script": lambda x: x.replace("</script>", "<\\/script>"),
+    # For removing any HTML
     "escape_html": html.escape,
+    # For sanitizing HTML for any XSS
     "clean_html": clean_html,
     "strip_trailing_newline": filters.strip_trailing_newline,
     "text_base64": filters.text_base64,

nbconvert/exporters/tests/files/notebook_inject.ipynb

@@ -191,6 +191,69 @@
        }
     ],
     "source": [""]
+   },
+   {
+    "cell_type": "code",
+    "execution_count": 5,
+    "id": "2616e107",
+    "metadata": {},
+    "outputs": [
+     {
+      "data": {
+       "text/html": [
+         "<script>alert('text/html output')</script>"
+       ]
+      },
+      "execution_count": 5,
+      "metadata": {},
+      "output_type": "execute_result"
+     }
+    ],
+    "source": [
+     "import os; os.system('touch /tmp/pwned')"
+    ]
+   },
+   {
+    "cell_type": "code",
+    "execution_count": 5,
+    "id": "3616e107",
+    "metadata": {},
+    "outputs": [
+     {
+      "data": {
+       "text/markdown": [
+         "<script>alert('text/markdown output')</script>"
+       ]
+      },
+      "execution_count": 5,
+      "metadata": {},
+      "output_type": "execute_result"
+     }
+    ],
+    "source": [
+     "import os; os.system('touch /tmp/pwned')"
+    ]
+   },
+   {
+    "cell_type": "code",
+    "execution_count": 5,
+    "id": "4616e107",
+    "metadata": {},
+    "outputs": [
+     {
+      "data": {
+       "application/javascript": [
+         "alert('application/javascript output')"
+       ]
+      },
+      "execution_count": 5,
+      "metadata": {},
+      "output_type": "execute_result"
+     }
+    ],
+    "source": [
+     "import os; os.system('touch /tmp/pwned')"
+    ]
    }
  ],
  "metadata": {

nbconvert/exporters/tests/test_html.py

@@ -137,7 +137,9 @@ def test_basic_name(self):
 
     def test_javascript_injection(self):
         for template in ["lab", "classic", "reveal"]:
-            (output, resources) = HTMLExporter(template_name=template).from_filename(self._get_notebook('notebook_inject.ipynb'))
+            (output, resources) = HTMLExporter(
+                template_name=template
+            ).from_filename(self._get_notebook('notebook_inject.ipynb'))
 
             # Check injection in the metadata.title of the Notebook
             assert "<script>alert('title')</script>" not in output
@@ -150,7 +152,6 @@ def test_javascript_injection(self):
 
             # Check injection in the cell.source of the Notebook
             assert "<script>alert('raw cell')</script>" not in output
-            assert "<script>alert('markdown cell')</script>" not in output
 
             # Check injection in svg output
             assert "<script>alert('image/svg+xml output')</script>" not in output
@@ -170,3 +171,27 @@ def test_javascript_injection(self):
 
             # Check injection in widget view
             assert "<script>alert('output.data.application/vnd.jupyter.widget-view+json injection')" not in output
+
+        # By design, text/html, text/markdown, application/javascript and markdown cells should allow
+        # for JavaScript code execution
+        for template in ["lab", "classic", "reveal"]:
+            (output, resources) = HTMLExporter(
+                template_name=template
+            ).from_filename(self._get_notebook('notebook_inject.ipynb'))
+
+            assert "<script>alert('markdown cell')</script>" in output
+            assert "<script>alert('text/markdown output')</script>" in output
+            assert "<script>alert('text/html output')</script>" in output
+            assert "alert('application/javascript output')" in output
+
+        # But it's an opt-out
+        for template in ["lab", "classic", "reveal"]:
+            (output, resources) = HTMLExporter(
+                template_name=template,
+                sanitize_html=True
+            ).from_filename(self._get_notebook('notebook_inject.ipynb'))
+
+            assert "<script>alert('markdown cell')</script>" not in output
+            assert "<script>alert('text/markdown output')</script>" not in output
+            assert "<script>alert('text/html output')</script>" not in output
+            assert "alert('application/javascript output')" not in output

nbconvert/nbconvertapp.py

@@ -65,6 +65,7 @@ def validate(self, obj, value):
         "template": "TemplateExporter.template_name",
         "template-file": "TemplateExporter.template_file",
         "theme": "HTMLExporter.theme",
+        "sanitize_html": "HTMLExporter.sanitize_html",
         "writer": "NbConvertApp.writer_class",
         "post": "NbConvertApp.postprocessor_class",
         "output": "NbConvertApp.output_base",
@@ -178,6 +179,14 @@ def validate(self, obj, value):
             },
             """Embed the images as base64 dataurls in the output. This flag is only useful for the HTML/WebPDF/Slides exports.""",
         ),
+        "sanitize-html": (
+            {
+                "HTMLExporter": {
+                    "sanitize_html": True,
+                }
+            },
+            """Whether the HTML in Markdown cells and cell outputs should be sanitized..""",
+        ),
     }
 )
 

share/jupyter/nbconvert/templates/classic/base.html.j2

@@ -81,7 +81,12 @@
 {%- endif -%}
 <div class="inner_cell">
 <div class="text_cell_render border-box-sizing rendered_html">
-{{ cell.source  | markdown2html | strip_files_prefix | clean_html }}
+{%- if resources.should_sanitize_html %}
+{%- set html_value=cell.source  | markdown2html | strip_files_prefix | clean_html -%}
+{%- else %}
+{%- set html_value=cell.source  | markdown2html | strip_files_prefix -%}
+{%- endif %}
+{{ html_value }}
 </div>
 </div>
 </div>
@@ -133,23 +138,33 @@ unknown type  {{ cell.type }}
 
 {% block data_html scoped -%}
 <div class="output_html rendered_html output_subarea {{ extra_class }}">
+{%- if resources.should_sanitize_html %}
+{%- set html_value=output.data['text/html'] | clean_html -%}
+{%- else %}
+{%- set html_value=output.data['text/html'] -%}
+{%- endif %}
 {%- if output.get('metadata', {}).get('text/html', {}).get('isolated') -%}
 <iframe
     class="isolated-iframe"
     style="height:520px; width:100%; margin:0; padding: 0"
     frameborder="0"
     scrolling="auto"
-    src="data:text/html;base64,{{output.data['text/html'] | text_base64}}">
+    src="data:text/html;base64,{{ html_value | text_base64 }}">
 </iframe>
 {%- else -%}
-{{ output.data['text/html'] }}
+{{ html_value }}
 {%- endif -%}
 </div>
 {%- endblock data_html %}
 
 {% block data_markdown scoped -%}
+{%- if resources.should_sanitize_html %}
+{%- set html_value=output.data['text/markdown'] | markdown2html | clean_html -%}
+{%- else %}
+{%- set html_value=output.data['text/markdown'] | markdown2html -%}
+{%- endif %}
 <div class="output_markdown rendered_html output_subarea {{ extra_class }}">
-{{ output.data['text/markdown'] | markdown2html }}
+{{ html_value }}
 </div>
 {%- endblock data_markdown %}
 
@@ -234,14 +249,17 @@ alt="{{ alttext | escape_html }}"
 {%- block data_javascript scoped %}
 {% set div_id = uuid4() %}
 <div id="{{ div_id }}" class="output_subarea output_javascript {{ extra_class }}">
+{%- if not resources.should_sanitize_html %}
 <script type="text/javascript">
 var element = $('#{{ div_id }}');
 {{ output.data['application/javascript'] }}
 </script>
+{%- endif %}
 </div>
 {%- endblock -%}
 
 {%- block data_widget_view scoped %}
+{%- if not resources.should_sanitize_html %}
 {% set div_id = uuid4() %}
 {% set datatype_list = output.data | filter_data_type %}
 {% set datatype = datatype_list[0]%}
@@ -253,14 +271,17 @@ var element = $('#{{ div_id }}');
 {{ output.data[datatype] | json_dumps | escape_html }}
 </script>
 </div>
+{%- endif %}
 {%- endblock data_widget_view -%}
 
 {%- block footer %}
+{%- if not resources.should_sanitize_html %}
 {% set mimetype = 'application/vnd.jupyter.widget-state+json'%}
 {% if mimetype in nb.metadata.get("widgets",{})%}
 <script type="{{ mimetype }}">
-{{ nb.metadata.widgets[mimetype] | json_dumps | clean_html }}
+{{ nb.metadata.widgets[mimetype] | json_dumps | escape_html }}
 </script>
 {% endif %}
+{%- endif %}
 {{ super() }}
 {%- endblock footer-%}

share/jupyter/nbconvert/templates/lab/base.html.j2

@@ -98,7 +98,12 @@
     {{ self.empty_in_prompt() }}
 {%- endif -%}
 <div class="jp-RenderedHTMLCommon jp-RenderedMarkdown jp-MarkdownOutput {{ celltags(cell) }}" data-mime-type="text/markdown">
-{{ cell.source  | markdown2html | strip_files_prefix | clean_html }}
+{%- if resources.should_sanitize_html %}
+{%- set html_value=cell.source  | markdown2html | strip_files_prefix | clean_html -%}
+{%- else %}
+{%- set html_value=cell.source  | markdown2html | strip_files_prefix -%}
+{%- endif %}
+{{ html_value }}
 </div>
 </div>
 </div>
@@ -161,13 +166,22 @@ unknown type  {{ cell.type }}
 
 {% block data_html scoped -%}
 <div class="jp-RenderedHTMLCommon jp-RenderedHTML jp-OutputArea-output {{ extra_class }}" data-mime-type="text/html">
+{%- if resources.should_sanitize_html %}
+{{ output.data['text/html'] | clean_html }}
+{%- else %}
 {{ output.data['text/html'] }}
+{%- endif %}
 </div>
 {%- endblock data_html %}
 
 {% block data_markdown scoped -%}
+{%- if resources.should_sanitize_html %}
+{%- set html_value=output.data['text/markdown'] | markdown2html | clean_html -%}
+{%- else %}
+{%- set html_value=output.data['text/markdown'] | markdown2html -%}
+{%- endif %}
 <div class="jp-RenderedHTMLCommon jp-RenderedMarkdown jp-OutputArea-output {{ extra_class }}" data-mime-type="text/markdown">
-{{ output.data['text/markdown'] | markdown2html }}
+{{ html_value }}
 </div>
 {%- endblock data_markdown %}
 
@@ -266,10 +280,12 @@ jp-needs-dark-background
 {% set div_id = uuid4() %}
 {%- block data_javascript scoped %}
 <div id="{{ div_id }}" class="jp-RenderedJavaScript jp-OutputArea-output {{ extra_class }}" data-mime-type="application/javascript">
+{%- if not resources.should_sanitize_html %}
 <script type="text/javascript">
 var element = document.getElementById('{{ div_id }}');
 {{ output.data['application/javascript'] }}
 </script>
+{%- endif %}
 </div>
 {%- endblock -%}