Add/fix support for various script types in HTML documents

- Add support for application/javascript and text/javascript
- Fix support for application/json
- Add support for all unknown script types

src/warc2zim/content_rewriting/html.py

@@ -88,9 +88,14 @@ def handle_starttag(self, tag: str, attrs: AttrsList, *, auto_close: bool = Fals
         self.html_rewrite_context = tag  # default value if not overriden later on
         if tag == "script":
             script_type = self.extract_attr(attrs, "type")
-            self.html_rewrite_context = {"json": "json", "module": "js-module"}.get(
-                script_type or "", "js-classic"
-            )
+            self.html_rewrite_context = {
+                "application/json": "json",
+                "json": "json",
+                "module": "js-module",
+                "application/javascript": "js-classic",
+                "text/javascript": "js-classic",
+                "": "js-classic",
+            }.get(script_type or "", "unknown")
         elif tag == "link":
             link_rel = self.extract_attr(attrs, "rel")
             if link_rel == "modulepreload":

tests/test_html_rewriting.py

@@ -40,6 +40,14 @@
         ContentForTests(
             '<script type="json">{"window": "https://kiwix.org/path"}</script>'
         ),
+        ContentForTests(
+            '<script type="application/json">{"window": "https://kiwix.org/path"}'
+            "</script>"
+        ),
+        ContentForTests(
+            '<script type="application/i_dont_know_you">'
+            '{"window": "https://kiwix.org/path"}</script>'
+        ),
     ]
 )
 def no_rewrite_content(request):
@@ -126,6 +134,57 @@ def test_escaped_content(escaped_content, no_js_notify):
                 "}</script>"
             ),
         ),
+        ContentForTests(
+            '<script type="application/javascript">document.title="HELLO";</script>',
+            (
+                """<script type="application/javascript">"""
+                "var _____WB$wombat$assign$function_____ = function(name) "
+                ""
+                "{return (self._wb_wombat && self._wb_wombat.local_init && "
+                "self._wb_wombat.local_init(name)) || self[name]; };\n"
+                "if (!self.__WB_pmw) { self.__WB_pmw = function(obj) "
+                "{ this.__WB_source = obj; return this; } }\n"
+                "{\n"
+                """let window = _____WB$wombat$assign$function_____("window");\n"""
+                "let globalThis = _____WB$wombat$assign$function_____"
+                """("globalThis");\n"""
+                """let self = _____WB$wombat$assign$function_____("self");\n"""
+                """let document = _____WB$wombat$assign$function_____("document");\n"""
+                """let location = _____WB$wombat$assign$function_____("location");\n"""
+                """let top = _____WB$wombat$assign$function_____("top");\n"""
+                """let parent = _____WB$wombat$assign$function_____("parent");\n"""
+                """let frames = _____WB$wombat$assign$function_____("frames");\n"""
+                """let opener = _____WB$wombat$assign$function_____("opener");\n"""
+                "let arguments;\n\n"
+                """document.title="HELLO";\n"""
+                "}</script>"
+            ),
+        ),
+        ContentForTests(
+            '<script type="text/javascript">document.title="HELLO";</script>',
+            (
+                """<script type="text/javascript">"""
+                "var _____WB$wombat$assign$function_____ = function(name) "
+                "{return (self._wb_wombat && self._wb_wombat.local_init && "
+                "self._wb_wombat.local_init(name)) || self[name]; };\n"
+                "if (!self.__WB_pmw) { self.__WB_pmw = function(obj) "
+                "{ this.__WB_source = obj; return this; } }\n"
+                "{\n"
+                """let window = _____WB$wombat$assign$function_____("window");\n"""
+                "let globalThis = _____WB$wombat$assign$function_____"
+                """("globalThis");\n"""
+                """let self = _____WB$wombat$assign$function_____("self");\n"""
+                """let document = _____WB$wombat$assign$function_____("document");\n"""
+                """let location = _____WB$wombat$assign$function_____("location");\n"""
+                """let top = _____WB$wombat$assign$function_____("top");\n"""
+                """let parent = _____WB$wombat$assign$function_____("parent");\n"""
+                """let frames = _____WB$wombat$assign$function_____("frames");\n"""
+                """let opener = _____WB$wombat$assign$function_____("opener");\n"""
+                "let arguments;\n\n"
+                """document.title="HELLO";\n"""
+                "}</script>"
+            ),
+        ),
     ]
 )
 def js_rewrites(request):