when browsing in browser, get non-html as html

aaxelb · aaxelb · commit a3fb2b28bb57 · 2025-09-10T14:52:00.000-04:00
- replace `trove.render._rendering` module with `trove.render.rendering`
  package (separate file for each rendering class)
- make `ProtoRendering` an actual `typing.Protocol` and narrow types to
  only unicode `str`
- add `trove.render.rendering.html_wrapped.HtmlWrappedRendering` that
  puts minimal `&lt;!DOCTYPE html&gt;&lt;pre&gt;...&lt;/pre&gt;` around an inner rendering
- use `HtmlWrappedRendering` to wrap responses that aren't html or json,
  when `Accept` header allows html and query params omit `withFileName`
- use mediatype constants more consistently (leaving off charset except
  for content-type header)
diff --git a/tests/trove/render/_base.py b/tests/trove/render/_base.py
@@ -7,7 +7,7 @@
 
 from trove.trovesearch.trovesearch_gathering import trovesearch_by_indexstrategy
 from trove.render._base import BaseRenderer
-from trove.render._rendering import ProtoRendering
+from trove.render.rendering import ProtoRendering
 from trove.vocab.namespaces import RDF
 from tests.trove._input_output_tests import BasicInputOutputTestCase
 from ._inputs import UNRENDERED_RDF, UNRENDERED_SEARCH_RDF, RdfCase
@@ -66,9 +66,9 @@ def assert_outputs_equal(self, expected_output, actual_output) -> None:
             self._get_rendered_output(actual_output),
         )
 
-    def _get_rendered_output(self, rendering: ProtoRendering):
+    def _get_rendered_output(self, rendering: ProtoRendering) -> str:
         # for now, they always iter strings (update if/when bytes are in play)
-        return ''.join(rendering.iter_content())  # type: ignore[arg-type]
+        return ''.join(map(str, rendering.iter_content()))
 
 
 class TrovesearchRendererTests(TroveRendererTests):
diff --git a/tests/trove/render/test_jsonapi_renderer.py b/tests/trove/render/test_jsonapi_renderer.py
@@ -2,7 +2,7 @@
 from unittest import mock
 
 from trove.render.jsonapi import RdfJsonapiRenderer
-from trove.render._rendering import SimpleRendering
+from trove.render.rendering import SimpleRendering
 from trove.vocab.namespaces import BLARG
 from . import _base
 
diff --git a/tests/trove/render/test_jsonld_renderer.py b/tests/trove/render/test_jsonld_renderer.py
@@ -1,7 +1,7 @@
 import json
 
 from trove.render.jsonld import RdfJsonldRenderer
-from trove.render._rendering import SimpleRendering
+from trove.render.rendering import SimpleRendering
 from ._inputs import BLARG
 from . import _base
 
diff --git a/tests/trove/render/test_simple_csv_renderer.py b/tests/trove/render/test_simple_csv_renderer.py
@@ -1,5 +1,5 @@
 from trove.render.simple_csv import TrovesearchSimpleCsvRenderer
-from trove.render._rendering import SimpleRendering
+from trove.render.rendering import SimpleRendering
 from . import _base
 
 
diff --git a/tests/trove/render/test_simple_json_renderer.py b/tests/trove/render/test_simple_json_renderer.py
@@ -1,7 +1,7 @@
 import json
 
 from trove.render.simple_json import TrovesearchSimpleJsonRenderer
-from trove.render._rendering import SimpleRendering
+from trove.render.rendering import SimpleRendering
 from trove.vocab.namespaces import BLARG
 from . import _base
 
diff --git a/tests/trove/render/test_simple_tsv_renderer.py b/tests/trove/render/test_simple_tsv_renderer.py
@@ -1,5 +1,5 @@
 from trove.render.simple_tsv import TrovesearchSimpleTsvRenderer
-from trove.render._rendering import SimpleRendering
+from trove.render.rendering import SimpleRendering
 from . import _base
 
 
diff --git a/tests/trove/render/test_turtle_renderer.py b/tests/trove/render/test_turtle_renderer.py
@@ -1,7 +1,7 @@
 from primitive_metadata import primitive_rdf as rdf
 
 from trove.render.turtle import RdfTurtleRenderer
-from trove.render._rendering import SimpleRendering
+from trove.render.rendering import SimpleRendering
 from . import _base
 
 
diff --git a/tests/trove/test_doctest.py b/tests/trove/test_doctest.py
@@ -4,6 +4,7 @@
 import trove.util.frozen
 import trove.util.iris
 import trove.util.propertypath
+import trove.vocab.mediatypes
 
 _DOCTEST_OPTIONFLAGS = (
     doctest.ELLIPSIS
@@ -15,6 +16,7 @@
     trove.util.frozen,
     trove.util.iris,
     trove.util.propertypath,
+    trove.vocab.mediatypes,
 )
 
 
diff --git a/trove/render/__init__.py b/trove/render/__init__.py
@@ -1,8 +1,7 @@
-from typing import Type
-
 from django import http
 
 from trove import exceptions as trove_exceptions
+from trove.vocab.mediatypes import strip_mediatype_parameters
 from ._base import BaseRenderer
 from .jsonapi import RdfJsonapiRenderer
 from .html_browse import RdfHtmlBrowseRenderer
@@ -25,10 +24,6 @@
     TrovesearchSimpleTsvRenderer,
 )
 
-RendersType = Type[
-    BaseRenderer | RdfHtmlBrowseRenderer | RdfJsonapiRenderer | RdfTurtleRenderer | RdfJsonldRenderer | TrovesearchSimpleCsvRenderer | TrovesearchSimpleJsonRenderer | TrovesearchSimpleTsvRenderer
-]
-
 RENDERER_BY_MEDIATYPE = {
     _renderer_type.MEDIATYPE: _renderer_type
     for _renderer_type in RENDERERS
@@ -42,7 +37,9 @@ def get_renderer_type(request: http.HttpRequest) -> type[BaseRenderer]:
     _requested_mediatype = request.GET.get('acceptMediatype')
     if _requested_mediatype:
         try:
-            _chosen_renderer_type = RENDERER_BY_MEDIATYPE[_requested_mediatype]
+            _chosen_renderer_type = RENDERER_BY_MEDIATYPE[
+                strip_mediatype_parameters(_requested_mediatype)
+            ]
         except KeyError:
             raise trove_exceptions.CannotRenderMediatype(_requested_mediatype)
     else:
diff --git a/trove/render/_base.py b/trove/render/_base.py
@@ -13,7 +13,7 @@
 from trove.vocab import mediatypes
 from trove.vocab.trove import TROVE_API_THESAURUS
 from trove.vocab.namespaces import namespaces_shorthand
-from ._rendering import ProtoRendering, SimpleRendering
+from .rendering import ProtoRendering, SimpleRendering
 
 
 @dataclasses.dataclass
@@ -61,15 +61,15 @@ def render_document(self) -> ProtoRendering:
         except NotImplementedError:
             raise NotImplementedError(f'class "{type(self)}" must implement either `render_document` or `simple_render_document`')
         else:
-            return SimpleRendering(  # type: ignore[return-value]  # until ProtoRendering(typing.Protocol) with py3.12
+            return SimpleRendering(
                 mediatype=self.MEDIATYPE,
                 rendered_content=_content,
             )
 
     @classmethod
     def render_error_document(cls, error: trove_exceptions.TroveError) -> ProtoRendering:
         # may override, but default to jsonapi
-        return SimpleRendering(  # type: ignore[return-value]  # until ProtoRendering(typing.Protocol) with py3.12
+        return SimpleRendering(
             mediatype=mediatypes.JSONAPI,
             rendered_content=json.dumps(
                 {'errors': [{  # https://jsonapi.org/format/#error-objects
diff --git a/trove/render/_html.py b/trove/render/_html.py
@@ -5,6 +5,7 @@
 from xml.etree.ElementTree import (
     Element,
     SubElement,
+    tostring as etree_tostring,
 )
 from typing import Any
 
@@ -13,10 +14,12 @@
 
 __all__ = ('HtmlBuilder',)
 
+HTML_DOCTYPE = '<!DOCTYPE html>'
+
 
 @dataclasses.dataclass
 class HtmlBuilder:
-    given_root: Element
+    given_root: Element = dataclasses.field(default_factory=lambda: Element('html'))
     _: dataclasses.KW_ONLY
     _nested_elements: list[Element] = dataclasses.field(default_factory=list)
     _heading_depth: int = 0
@@ -67,3 +70,9 @@ def leaf(self, tag_name: str, *, text: str | None = None, attrs: dict | None = N
             _leaf_element.text = text.unicode_value
         elif text is not None:
             _leaf_element.text = text
+
+    def as_html_doc(self) -> str:
+        return '\n'.join((
+            HTML_DOCTYPE,
+            etree_tostring(self.root_element, encoding='unicode', method='html'),
+        ))
diff --git a/trove/render/_rendering.py b/trove/render/_rendering.py
diff --git a/trove/render/_simple_trovesearch.py b/trove/render/_simple_trovesearch.py
@@ -9,7 +9,7 @@
 from trove.vocab.jsonapi import JSONAPI_LINK_OBJECT
 from trove.vocab.namespaces import TROVE, RDF
 from ._base import BaseRenderer
-from ._rendering import ProtoRendering, SimpleRendering
+from .rendering import ProtoRendering, SimpleRendering
 if TYPE_CHECKING:
     from trove.util.json import JsonObject
 
@@ -30,7 +30,7 @@ def simple_multicard_rendering(self, cards: Iterator[tuple[str, JsonObject]]) ->
         raise NotImplementedError
 
     def unicard_rendering(self, card_iri: str, osfmap_json: JsonObject) -> ProtoRendering:
-        return SimpleRendering(  # type: ignore[return-value]
+        return SimpleRendering(
             mediatype=self.MEDIATYPE,
             rendered_content=self.simple_unicard_rendering(card_iri, osfmap_json),
         )
@@ -41,7 +41,7 @@ def multicard_rendering(self, card_pages: Iterator[dict[str, JsonObject]]) -> Pr
             for _page in card_pages
             for _card_iri, _card_contents in _page.items()
         )
-        return SimpleRendering(  # type: ignore[return-value]
+        return SimpleRendering(
             mediatype=self.MEDIATYPE,
             rendered_content=self.simple_multicard_rendering(_cards),
         )
diff --git a/trove/render/html_browse.py b/trove/render/html_browse.py
@@ -12,7 +12,6 @@
 from urllib.parse import quote, urlsplit, urlunsplit
 from xml.etree.ElementTree import (
     Element,
-    tostring as etree_tostring,
     fromstring as etree_fromstring,
 )
 
@@ -60,12 +59,10 @@
 
 _PHI = (math.sqrt(5) + 1) / 2
 
-_HTML_DOCTYPE = '<!DOCTYPE html>'
-
 
 @dataclasses.dataclass
 class RdfHtmlBrowseRenderer(BaseRenderer):
-    MEDIATYPE: ClassVar[str] = 'text/html; charset=utf-8'
+    MEDIATYPE: ClassVar[str] = mediatypes.HTML
     __current_data: rdf.RdfTripleDictionary = dataclasses.field(init=False)
     __visiting_iris: set[str] = dataclasses.field(init=False)
     __hb: HtmlBuilder = dataclasses.field(init=False)
@@ -82,7 +79,7 @@ def is_data_blended(self) -> bool | None:
 
     # override BaseRenderer
     def simple_render_document(self) -> str:
-        self.__hb = HtmlBuilder(Element('html'))
+        self.__hb = HtmlBuilder()
         self.render_html_head()
         _body_attrs = {
             'class': 'BrowseWrapper',
@@ -92,10 +89,7 @@ def simple_render_document(self) -> str:
             self.render_nav()
             self.render_main()
             self.render_footer()
-        return '\n'.join((
-            _HTML_DOCTYPE,
-            etree_tostring(self.__hb.root_element, encoding='unicode', method='html'),
-        ))
+        return self.__hb.as_html_doc()
 
     def render_html_head(self) -> None:
         with self.__hb.nest('head'):
diff --git a/trove/render/rendering/__init__.py b/trove/render/rendering/__init__.py
@@ -0,0 +1,4 @@
+from .proto import ProtoRendering
+from .simple import SimpleRendering
+
+__all__ = ('ProtoRendering', 'SimpleRendering')
diff --git a/trove/render/rendering/html_wrapped.py b/trove/render/rendering/html_wrapped.py
@@ -0,0 +1,20 @@
+import dataclasses
+import html
+from typing import Iterator
+
+from trove.vocab import mediatypes
+from trove.render._html import HTML_DOCTYPE
+from .proto import ProtoRendering
+
+
+@dataclasses.dataclass
+class HtmlWrappedRendering(ProtoRendering):
+    inner_rendering: ProtoRendering
+    mediatype: str = mediatypes.HTML
+
+    def iter_content(self) -> Iterator[str]:
+        yield HTML_DOCTYPE
+        yield '<pre>'
+        for _content in self.inner_rendering.iter_content():
+            yield html.escape(_content)
+        yield '</pre>'
diff --git a/trove/render/rendering/proto.py b/trove/render/rendering/proto.py
@@ -0,0 +1,16 @@
+from typing import (
+    Iterator,
+    Protocol,
+)
+
+__all__ = ('ProtoRendering',)
+
+
+class ProtoRendering(Protocol):
+    '''protocol for all renderings
+    '''
+    mediatype: str  # required attribute
+
+    def iter_content(self) -> Iterator[str]:
+        '''`iter_content`: (only) required method
+        '''
diff --git a/trove/render/rendering/simple.py b/trove/render/rendering/simple.py
@@ -0,0 +1,17 @@
+from collections.abc import Generator
+import dataclasses
+
+from .proto import ProtoRendering
+
+__all__ = ('SimpleRendering',)
+
+
+@dataclasses.dataclass
+class SimpleRendering(ProtoRendering):
+    '''for simple pre-rendered string content
+    '''
+    mediatype: str
+    rendered_content: str = ''
+
+    def iter_content(self) -> Generator[str]:
+        yield self.rendered_content
diff --git a/trove/render/rendering/streamable.py b/trove/render/rendering/streamable.py
@@ -0,0 +1,18 @@
+from collections.abc import Iterator
+import dataclasses
+
+from trove import exceptions as trove_exceptions
+from .proto import ProtoRendering
+
+
+@dataclasses.dataclass
+class StreamableRendering(ProtoRendering):
+    mediatype: str
+    content_stream: Iterator[str] = iter(())
+    _started_already: bool = False
+
+    def iter_content(self) -> Iterator[str]:
+        if self._started_already:
+            raise trove_exceptions.CannotRenderStreamTwice
+        self._started_already = True
+        yield from self.content_stream
diff --git a/trove/render/simple_csv.py b/trove/render/simple_csv.py
@@ -20,7 +20,8 @@
 from trove.vocab import osfmap
 from trove.vocab.namespaces import TROVE
 from ._simple_trovesearch import SimpleTrovesearchRenderer
-from ._rendering import StreamableRendering, ProtoRendering
+from .rendering import ProtoRendering
+from .rendering.streamable import StreamableRendering
 if TYPE_CHECKING:
     from trove.util.trove_params import BasicTroveParams
     from trove.util.json import JsonValue, JsonObject
@@ -47,7 +48,7 @@ def multicard_rendering(self, card_pages: Iterator[dict[str, JsonObject]]) -> Pr
             card_pages,
             trove_params=getattr(self.response_focus, 'search_params', None),
         )
-        return StreamableRendering(  # type: ignore[return-value]
+        return StreamableRendering(
             mediatype=self.MEDIATYPE,
             content_stream=csv_stream(self.CSV_DIALECT, _doc.header(), _doc.rows()),
         )
diff --git a/trove/render/simple_json.py b/trove/render/simple_json.py
diff --git a/trove/views/_base.py b/trove/views/_base.py
diff --git a/trove/views/_responder.py b/trove/views/_responder.py
diff --git a/trove/vocab/mediatypes.py b/trove/vocab/mediatypes.py
