What is ${inlineMath('x^2 < 1', 'x')} when x is 2?
Other
Square it: ${inlineMath('2^2', '2')} first
What is when x is 2?
Square it: first
if x:\n return 1
ab
a \n b
Solve ' + + 'xx' + + ' now
Solve the equation x2=3\frac{x}{2} = 3 for x. What is the value of x?
6
3
It is wrong ', result.xml) + self.assertIn( + '', + result.xml, + ) + self.assertTrue(validate_qti_item(result.xml.encode("utf-8")).is_valid) def test_maths_in_a_table_cell(self): result = self._convert("| a |\n|---|\n| $$x^2$$ |") - self.assertIn('', result.xml) + self.assertIn( + '', + result.xml, + ) + self.assertTrue(validate_qti_item(result.xml.encode("utf-8")).is_valid) + + def test_maths_latex2mathml_rejects_converts_to_a_valid_item(self): + for markdown in ("It is $$x_$$ here.", "$${$$", "- $$a', result.xml) + self.assertIn( + '>', + result.xml, + ) + self.assertTrue(validate_qti_item(result.xml.encode("utf-8")).is_valid) class CustomInteractionTests(unittest.TestCase): diff --git a/contentcuration/contentcuration/tests/utils/qti/test_html_to_markdown.py b/contentcuration/contentcuration/tests/utils/qti/test_html_to_markdown.py index 0cf4d04783..491b563530 100644 --- a/contentcuration/contentcuration/tests/utils/qti/test_html_to_markdown.py +++ b/contentcuration/contentcuration/tests/utils/qti/test_html_to_markdown.py @@ -97,6 +97,13 @@ def test_underlined_span_becomes_a_perseus_underline(): ) +def test_formula_span_becomes_double_dollar_latex(): + assert ( + _markdown_from_html('Is even?') + == "Is $$x^2$$ even?" + ) + + def test_span_carrying_both_decorations_keeps_both(): assert ( _markdown_from_html( diff --git a/contentcuration/contentcuration/tests/utils/qti/test_mathml.py b/contentcuration/contentcuration/tests/utils/qti/test_mathml.py index 0bace05336..340ce6129e 100644 --- a/contentcuration/contentcuration/tests/utils/qti/test_mathml.py +++ b/contentcuration/contentcuration/tests/utils/qti/test_mathml.py @@ -636,7 +636,10 @@ def test_math_element_with_attributes(self): children=[Mi(children=["y"])], ) xml_str = obj.to_xml_string() - expected_xml_str = 'y' + expected_xml_str = ( + 'y' + ) self.assertEqual(xml_str, expected_xml_str) def test_mrow_nested_elements(self): @@ -899,6 +902,12 @@ def test_from_string_unregistered_tag(self): ): MathMLElement.from_string(xml_str) + def test_from_string_foreign_namespace_tag(self): + with self.assertRaisesRegex( + ValueError, "No registered class found for tag: {urn:foo}mi" + ): + MathMLElement.from_string('x') + def test_attribute_validation_error_on_creation(self): with self.assertRaises(ValueError): # Pydantic's ValidationError Mi(mathvariant="not-a-valid-variant", children=["x"]) diff --git a/contentcuration/contentcuration/tests/utils/qti/test_media.py b/contentcuration/contentcuration/tests/utils/qti/test_media.py index 9043b263cb..d015f84f8d 100644 --- a/contentcuration/contentcuration/tests/utils/qti/test_media.py +++ b/contentcuration/contentcuration/tests/utils/qti/test_media.py @@ -1,5 +1,11 @@ +from lxml import etree + +from contentcuration.tests.utils.qti.test_validation import SINGLE_HINT_CATALOG from contentcuration.tests.utils.qti.test_validation import VALID_CHOICE_ITEM +from contentcuration.utils.assessment.markdown import latex_to_mathml +from contentcuration.utils.assessment.qti.constants import MATHML_NAMESPACE from contentcuration.utils.assessment.qti.media import get_qti_media_references +from contentcuration.utils.assessment.qti.media import render_qti_latex_spans from contentcuration.utils.assessment.qti.media import rewrite_qti_media_paths from contentcuration.utils.assessment.qti.media import rewrite_qti_sized_image_paths from contentcuration.utils.assessment.qti.media import set_qti_item_language @@ -289,3 +295,58 @@ def test_strip_studio_attributes_ignores_lookalikes_in_values_and_text(): "data-other": "1", } assert paragraph.text == 'data-studio-prompt=""' + + +def _with_latex_spans(prompt, hint): + return VALID_CHOICE_ITEM.replace( + "Select the correct answer.", + f"{prompt}", + ).replace( + "', + "Hint.", + ) + ) + annotation = _root(result).find(f".//{{{MATHML_NAMESPACE}}}annotation") + assert annotation.text == r"\begin{matrix}a < b & c\end{matrix}" + validation = validate_qti_item(result) + assert validation.is_valid, validation.errors + + +def test_render_latex_spans_leaves_malformed_latex_as_its_source_text(): + result = render_qti_latex_spans( + _with_latex_spans('Solve now.', "Hint.") + ) + prompt = _root(result).find(".//{*}qti-prompt") + assert list(prompt) == [] + assert prompt.text == "Solve $$x_$$ now." + validation = validate_qti_item(result) + assert validation.is_valid, validation.errors + + +def test_render_latex_spans_leaves_an_item_without_formulas_unchanged(): + assert render_qti_latex_spans(VALID_CHOICE_ITEM) == VALID_CHOICE_ITEM diff --git a/contentcuration/contentcuration/tests/utils/qti/test_validation.py b/contentcuration/contentcuration/tests/utils/qti/test_validation.py index f13d158807..0737ced441 100644 --- a/contentcuration/contentcuration/tests/utils/qti/test_validation.py +++ b/contentcuration/contentcuration/tests/utils/qti/test_validation.py @@ -47,6 +47,13 @@ def _item_xml(identifier, title, response_declaration, item_body): "", ) +SINGLE_HINT_CATALOG = ( + '' + '' + "{hint}" + "" +) + _XML_DECLARATION = '' _ENTITY_DOCTYPE = ']>' diff --git a/contentcuration/contentcuration/tests/utils/test_exercise_creation.py b/contentcuration/contentcuration/tests/utils/test_exercise_creation.py index 071bb1545e..c05bd89255 100644 --- a/contentcuration/contentcuration/tests/utils/test_exercise_creation.py +++ b/contentcuration/contentcuration/tests/utils/test_exercise_creation.py @@ -33,15 +33,18 @@ from contentcuration.tests.testdata import fileobj_exercise_graphie from contentcuration.tests.testdata import fileobj_exercise_image from contentcuration.tests.utils.qti.test_convert import _normalize_xml +from contentcuration.tests.utils.qti.test_media import _with_latex_spans from contentcuration.tests.utils.qti.test_perseus_derive import _text_item from contentcuration.tests.utils.qti.test_perseus_derive import TOLERANCE_ITEM from contentcuration.tests.utils.qti.test_validation import _item_xml +from contentcuration.tests.utils.qti.test_validation import SINGLE_HINT_CATALOG from contentcuration.tests.utils.qti.test_validation import VALID_CHOICE_ITEM from contentcuration.utils.assessment.base import resize_image from contentcuration.utils.assessment.perseus import answerless_input_ids from contentcuration.utils.assessment.perseus import PerseusExerciseGenerator from contentcuration.utils.assessment.qti.archive import hex_to_qti_id from contentcuration.utils.assessment.qti.archive import QTIExerciseGenerator +from contentcuration.utils.assessment.qti.constants import MATHML_NAMESPACE from contentcuration.utils.assessment.qti.validation import parse_qti_xml from contentcuration.utils.assessment.qti.validation import validate_qti_item @@ -147,6 +150,12 @@ def _mpo_bytes(): ) +_FORMULA = '' +FORMULA_CHOICE_ITEM = _with_latex_spans( + f"Is {_FORMULA} even?", f"Try {_FORMULA}." +).replace(">Option A<", f">{_FORMULA}<") + + SVG_BYTES = b'' @@ -2847,6 +2856,23 @@ def test_native_qti_item_published_without_studio_attributes(self): _canonical_xml(raw_data.replace(' data-studio-prompt=""', "")), ) + def test_native_qti_item_formulas_published_as_mathml(self): + item = self._create_native_qti_item(FORMULA_CHOICE_ITEM) + + self._create_qti_zip(_exercise_data([item])) + + exercise_file = self.exercise_node.files.get(preset_id=format_presets.QTI_ZIP) + zip_file = self._validate_qti_zip_structure(exercise_file) + item_xml = zip_file.read("items/item_1.xml") + root = parse_qti_xml(item_xml).getroot() + annotations = root.findall(f".//{{{MATHML_NAMESPACE}}}annotation") + self.assertEqual([a.text for a in annotations], ["x^2"] * 3) + self.assertEqual(root.xpath("//*[@data-latex]"), []) + validation = validate_qti_item(item_xml) + self.assertTrue(validation.is_valid, validation.errors) + item.refresh_from_db() + self.assertEqual(item.raw_data, FORMULA_CHOICE_ITEM) + def test_native_qti_item_media_included_and_addressed(self): # fileobj_exercise_image() writes real bytes to storage keyed by their # actual md5 checksum + "jpg" ext -- use that real checksum/ext rather @@ -3224,12 +3250,7 @@ def test_republish_replaces_stale_native_qti_archive(self): def test_native_qti_perseus_derivation(self): """A native QTI choice item is derived into a rendered Perseus item JSON.""" - catalog_info = ( - '' - '' - "First hint." - "" - ) + catalog_info = SINGLE_HINT_CATALOG.format(hint="First hint.") raw_data = VALID_CHOICE_ITEM.replace( "What is the answer to this question? " - '' + '' "x·y=z2" 'x\cdot y=z^2' # noqa W605 "\n" @@ -35,7 +35,7 @@ def test_block_math(self): ) expected = ( "Here's an equation:\n" - '' + '' "E=mc2" 'E = mc^2' "" @@ -59,7 +59,7 @@ def test_multiline_block_math(self): "$$" ) expected = ( - '' + '' "a=b+c" '$5=d+e' '\n\\begin{aligned}\na = b + c \\\\\n$5 = d + e\n\\end{aligned}\n' @@ -77,7 +77,7 @@ def test_inline_math_with_dollar_inside(self): markdown_text = "Test this $$\\text{Cost = 1.00 $USD$}$$" expected = ( "Test this " - '' + '' "Cost = 1.00 $USD$" '\\text{Cost = 1.00 $USD$}' "\n" @@ -85,19 +85,34 @@ def test_inline_math_with_dollar_inside(self): self._assert_conversion(markdown_text, expected) + def test_text_with_markup_characters(self): + markdown_text = "$$\\text{a' + "a<b+" + "c&d+" + "e < f" + '' + "\\text{a<b}+\\text{c&d}+\\text{e < f}" + "" + "" + ) + + self._assert_conversion(markdown_text, expected) + def test_multiple_math_expressions(self): """Test multiple math expressions in one document""" markdown_text = "First: $$a + b$$, then $$c \\times d$$, finally $$e^f$$." expected = ( "First: " - 'a+b' + 'a+b' 'a + b' ", then " - 'c×d' + 'c×d' 'c \\times d' ", finally " - 'ef' + 'ef' 'e^f' ".\n" ) @@ -115,17 +130,17 @@ def test_mixed_inline_and_block(self): ) expected = ( "This is inline math: " - 'a=b' + 'a=b' 'a = b' "\n" "And this is block math:\n" - '' + '' "∑i=1" "nxi=y" '\sum_{i=1}^{n} x_i = y' # noqa W605 "" "Back to text with more inline: " - 'z2' + 'z2' 'z^2' "\n" ) @@ -137,7 +152,7 @@ def test_block_sum_uses_munderover(self): markdown_text = "$$\\sum_{i=1}^{n} x_i$$" expected = ( - '' + '' "∑i=1" "nxi" '\\sum_{i=1}^{n} x_i' @@ -152,7 +167,8 @@ def test_inline_sum_uses_msubsup(self): markdown_text = "The sum $$\\sum_{i=1}^{n} x_i$$ is finite." expected = ( "The sum " - '∑i=1' + '' + "∑i=1" "nxi" '\\sum_{i=1}^{n} x_i' " is finite.\n" @@ -165,7 +181,7 @@ def test_block_prod_uses_munderover(self): markdown_text = "$$\\prod_{k=0}^{n} a_k$$" expected = ( - '' + '' "∏k=0" "nak" '\\prod_{k=0}^{n} a_k' @@ -179,7 +195,7 @@ def test_block_int_unaffected(self): markdown_text = "$$\\int_{a}^{b} f(x) dx$$" expected = ( - '' + '' "∫a" 'bf(x)dx' '\\int_{a}^{b} f(x) dx' @@ -202,7 +218,7 @@ def test_simple_inline_math(self): markdown_text = "The variable $$x$$ is unknown." expected = ( "The variable " - 'x' + 'x' 'x' " is unknown.\n" ) @@ -214,7 +230,7 @@ def test_simple_block_math(self): markdown_text = "$$y = mx + b$$" expected = ( - '' + '' "y=mx+b" 'y = mx + b' "" @@ -222,12 +238,37 @@ def test_simple_block_math(self): self._assert_conversion(markdown_text, expected) + def test_math_at_line_start_followed_by_text_is_inline(self): + markdown_text = "$$5$$ exactly" + expected = ( + "" + '5' + '5' + " exactly\n" + ) + + self._assert_conversion(markdown_text, expected) + + def test_line_starting_and_ending_with_math_is_inline(self): + markdown_text = "$$a$$ and $$b$$" + expected = ( + "" + 'a' + 'a' + " and " + 'b' + 'b' + "\n" + ) + + self._assert_conversion(markdown_text, expected) + def test_mo_accented(self): """Regression test for missed experimental property on mo tags""" markdown_text = "$$a_b+\\overrightarrow{abc}+\\overleftarrow{abc}\\div\\surd\\overline{abc}$$" expected = ( - '' + '' "ab+abc→+" 'abc←÷√ab' 'c―' @@ -269,6 +310,34 @@ def _assert_conversion(self, markdown_text: str, expected: str): ) +class MalformedMathTests(unittest.TestCase): + def test_inline_math_latex2mathml_rejects_renders_as_its_source(self): + for latex in ("x_", "{", " "): + with self.subTest(latex=latex): + self.assertEqual( + render_markdown(f"It is $${latex}$$ here."), + f"It is $${latex}$$ here.\n", + ) + + def test_block_math_latex2mathml_rejects_renders_as_a_paragraph(self): + self.assertEqual( + render_markdown("Before\n\n$$x_$$\n\nAfter"), + "Before\n$$x_$$\nAfter\n", + ) + + def test_source_of_rejected_math_is_escaped(self): + self.assertEqual( + render_markdown("It is $$aIt is $$a<b&c_$$ here\n", + ) + + def test_stray_ampersand_renders_as_its_source(self): + self.assertEqual( + render_markdown("It is $$a & c$$ here"), + "It is $$a & c$$ here\n", + ) + + class SizedImageTests(unittest.TestCase): """Perseus images, whose size and alignment suffixes are not valid CommonMark.""" diff --git a/contentcuration/contentcuration/utils/assessment/markdown.py b/contentcuration/contentcuration/utils/assessment/markdown.py index d14a12cd8f..0a5e0fdc49 100644 --- a/contentcuration/contentcuration/utils/assessment/markdown.py +++ b/contentcuration/contentcuration/utils/assessment/markdown.py @@ -1,8 +1,10 @@ +import logging import re import xml.etree.ElementTree as ET from latex2mathml.converter import convert from markdown_it import MarkdownIt +from markdown_it.common.utils import escapeHtml from markdown_it.renderer import RendererProtocol from markdown_it.rules_block import StateBlock from markdown_it.rules_inline import StateInline @@ -10,19 +12,25 @@ from markdown_it.utils import EnvType from markdown_it.utils import OptionsDict +from contentcuration.utils.assessment.qti.constants import MATHML_NAMESPACE from contentcuration.utils.assessment.qti.mathml.core import Annotation from contentcuration.utils.assessment.qti.mathml.core import Semantics +logger = logging.getLogger(__name__) + # The QTI 3.0 HTML profile has no or , so a decoration is carried as a style # on a . Shared with the reverse conversion in qti/html_to_markdown.py and with # the raw-HTML rewrite in qti/convert.py. STRIKETHROUGH_DECORATION = "line-through" UNDERLINE_DECORATION = "underline" +# Set by the QTI editor's TipTapEditor/extensions/Math.js. +LATEX_ATTRIBUTE = "data-latex" + # Regex patterns for $$ delimited math INLINE_PATTERN = re.compile(r"^\$\$([\s\S]+?)\$\$") -BLOCK_PATTERN = re.compile(r"^\$\$([\s\S]+?)\$\$", re.M) +BLOCK_PATTERN = re.compile(r"^\$\$((?:(?!\$\$)[\s\S])+)\$\$[ \t]*$", re.M) # Perseus extends the CommonMark image with a trailing size and alignment: #  @@ -120,14 +128,34 @@ def math_block_func( return True -def _convert(latex, inline=True): - # Remove the namespace declaration for cleaner output - markup = convert(latex, display="inline" if inline else "block").replace( - ' xmlns="http://www.w3.org/1998/Math/MathML"', "" - ) - # By default latex2mathml encodes operators that don't need to be encoded - # so we parse it with ElementTree and turn it back into a string here for consistency. - math_element = ET.fromstring(markup) +# latex2mathml 3.78.1 writes the argument of \text and its kin into without +# escaping it; the only character reference it emits there is for a space. +MTEXT_CONTENT_PATTERN = re.compile(r"(]*)?(?)(.*?)()", re.S) +UNESCAPED_AMPERSAND_PATTERN = re.compile(r"&(?!#x[0-9A-Fa-f]+;)") + + +def _escape_mtext(match): + text = UNESCAPED_AMPERSAND_PATTERN.sub("&", match[2]).replace("<", "<") + return f"{match[1]}{text}{match[3]}" + + +def latex_to_mathml(latex, inline=True): + """Return ``None`` when latex2mathml cannot convert ``latex``.""" + try: + markup = convert(latex, display="inline" if inline else "block") + markup = MTEXT_CONTENT_PATTERN.sub(_escape_mtext, markup) + # ElementTree serializes a parsed default namespace as ns0: prefixes, so parse + # without it and declare it on the root afterwards. + markup = markup.replace(f' xmlns="{MATHML_NAMESPACE}"', "") + # By default latex2mathml encodes operators that don't need to be encoded + # so we parse it with ElementTree and turn it back into a string here for consistency. + math_element = ET.fromstring(markup) + # latex2mathml 3.78.1 raises a dozen exception classes with no shared base on + # malformed input, as well as StopIteration, ValueError and IndexError, and emits + # a stray & unescaped, which fails the parse. + except Exception: + logger.warning("Could not convert LaTeX to MathML: %r", latex) + return None # Create LaTeX annotation latex_annotation_element = Annotation( @@ -140,10 +168,15 @@ def _convert(latex, inline=True): semantics_element.append(child) semantics_element.append(latex_annotation_element) math_element.append(semantics_element) + math_element.attrib = {"xmlns": MATHML_NAMESPACE, **math_element.attrib} return ET.tostring(math_element, encoding="unicode") +def latex_source(latex): + return f"$${latex}$$" + + def render_math_inline( self: RendererProtocol, tokens: list[Token], @@ -152,7 +185,8 @@ def render_math_inline( env: EnvType, ) -> str: """Render inline math to MathML""" - return _convert(tokens[idx].content) + latex = tokens[idx].content + return latex_to_mathml(latex) or escapeHtml(latex_source(latex)) def render_math_block( @@ -163,7 +197,11 @@ def render_math_block( env: EnvType, ) -> str: """Render block math to MathML""" - return _convert(tokens[idx].content, inline=False) + latex = tokens[idx].content + return ( + latex_to_mathml(latex, inline=False) + or f"{escapeHtml(latex_source(latex))}\n" + ) def texmath_to_mathml_plugin(md: MarkdownIt) -> None: diff --git a/contentcuration/contentcuration/utils/assessment/qti/archive.py b/contentcuration/contentcuration/utils/assessment/qti/archive.py index dfb2ef5ae6..3910f582ff 100644 --- a/contentcuration/contentcuration/utils/assessment/qti/archive.py +++ b/contentcuration/contentcuration/utils/assessment/qti/archive.py @@ -29,6 +29,7 @@ from contentcuration.utils.assessment.qti.imsmanifest import Resource from contentcuration.utils.assessment.qti.imsmanifest import Resources from contentcuration.utils.assessment.qti.media import get_qti_media_references +from contentcuration.utils.assessment.qti.media import render_qti_latex_spans from contentcuration.utils.assessment.qti.media import rewrite_qti_media_paths from contentcuration.utils.assessment.qti.media import rewrite_qti_sized_image_paths from contentcuration.utils.assessment.qti.media import set_qti_item_language @@ -139,6 +140,7 @@ def _create_native_qti_item(self, assessment_item) -> Optional[Tuple[str, bytes] # depending only on which editor wrote them. item_xml = set_qti_item_language(item_xml, self._node_language()) item_xml = strip_studio_attributes(item_xml) + item_xml = render_qti_latex_spans(item_xml) self._add_resource( QTIResource( diff --git a/contentcuration/contentcuration/utils/assessment/qti/base.py b/contentcuration/contentcuration/utils/assessment/qti/base.py index 9196fab4c5..113b7cefe4 100644 --- a/contentcuration/contentcuration/utils/assessment/qti/base.py +++ b/contentcuration/contentcuration/utils/assessment/qti/base.py @@ -18,6 +18,7 @@ from pydantic import PrivateAttr from .constants import Dir +from .constants import MATHML_NAMESPACE from .fields import entry_pattern as srcset_entry_pattern from contentcuration.utils.assessment.qti.fields import BCP47Language @@ -212,7 +213,8 @@ def _ensure_registry_complete(cls): def get_class_for_tag(cls, tag_name: str) -> Optional[Type]: """Get the registered class for a given tag name""" cls._ensure_registry_complete() - return getattr(cls, "_registry", {}).get(tag_name) + local_name = tag_name.removeprefix(f"{{{MATHML_NAMESPACE}}}") + return getattr(cls, "_registry", {}).get(local_name) @lru_cache(maxsize=1) diff --git a/contentcuration/contentcuration/utils/assessment/qti/constants.py b/contentcuration/contentcuration/utils/assessment/qti/constants.py index 05d530734a..27bd391605 100644 --- a/contentcuration/contentcuration/utils/assessment/qti/constants.py +++ b/contentcuration/contentcuration/utils/assessment/qti/constants.py @@ -1,6 +1,8 @@ from enum import Enum +MATHML_NAMESPACE = "http://www.w3.org/1998/Math/MathML" + # QTI Constants for Outcome Declarations diff --git a/contentcuration/contentcuration/utils/assessment/qti/html_to_markdown.py b/contentcuration/contentcuration/utils/assessment/qti/html_to_markdown.py index 25af3f3c87..1975919b10 100644 --- a/contentcuration/contentcuration/utils/assessment/qti/html_to_markdown.py +++ b/contentcuration/contentcuration/utils/assessment/qti/html_to_markdown.py @@ -25,6 +25,8 @@ from le_utils.constants import exercises from lxml import etree +from contentcuration.utils.assessment.markdown import LATEX_ATTRIBUTE +from contentcuration.utils.assessment.markdown import latex_source from contentcuration.utils.assessment.markdown import STRIKETHROUGH_DECORATION from contentcuration.utils.assessment.markdown import UNDERLINE_DECORATION from contentcuration.utils.assessment.qti.media import img_pixel_size @@ -67,7 +69,7 @@ def _render_math(el): _localname(annotation) == "annotation" and annotation.get("encoding") == "application/x-tex" ): - return "$${}$${}".format(annotation.text or "", suffix) + return latex_source(annotation.text or "") + suffix logger.warning("MathML element without an application/x-tex annotation; dropping") return "" @@ -102,10 +104,14 @@ def _render_span(el): """A decorated span is how a strikethrough and an underline are carried. ``render_markdown`` writes the first; the editor writes either, since the QTI - HTML profile has no ```` and no ````. Any other span unwraps, keeping its - content: whatever a foreign QTI package styles a span with has nothing in - Perseus markdown to say it. + HTML profile has no ```` and no ````. The editor also writes a formula as + an empty span carrying its LaTeX. Any other span unwraps, keeping its content: + whatever a foreign QTI package styles a span with has nothing in Perseus + markdown to say it. """ + latex = el.get(LATEX_ATTRIBUTE) + if latex is not None: + return latex_source(latex) decorations = _text_decorations(el) rendered = _render_inline(el) if UNDERLINE_DECORATION in decorations: diff --git a/contentcuration/contentcuration/utils/assessment/qti/mathml/core.py b/contentcuration/contentcuration/utils/assessment/qti/mathml/core.py index 4bd3540fd0..6f7c53e99d 100644 --- a/contentcuration/contentcuration/utils/assessment/qti/mathml/core.py +++ b/contentcuration/contentcuration/utils/assessment/qti/mathml/core.py @@ -19,13 +19,14 @@ from .fields import MathMLDisplay from .fields import MathMLForm from contentcuration.utils.assessment.qti.base import TextType - +from contentcuration.utils.assessment.qti.constants import MATHML_NAMESPACE PresentationContentListLength2 = Annotated[PresentationContentList, Len(2, 2)] PresentationContentListLength3 = Annotated[PresentationContentList, Len(3, 3)] class Math(MathMLElement): + xmlns: str = MATHML_NAMESPACE display: Optional[MathMLDisplay] = None alttext: Optional[str] = None children: PresentationContentList = Field(default_factory=list) diff --git a/contentcuration/contentcuration/utils/assessment/qti/media.py b/contentcuration/contentcuration/utils/assessment/qti/media.py index 9222eb0e2e..3b4392d813 100644 --- a/contentcuration/contentcuration/utils/assessment/qti/media.py +++ b/contentcuration/contentcuration/utils/assessment/qti/media.py @@ -4,6 +4,9 @@ from lxml import etree +from contentcuration.utils.assessment.markdown import LATEX_ATTRIBUTE +from contentcuration.utils.assessment.markdown import latex_source +from contentcuration.utils.assessment.markdown import latex_to_mathml from contentcuration.utils.assessment.qti.fields import ( entry_pattern as srcset_entry_pattern, ) @@ -18,6 +21,8 @@ ) PIXEL_LENGTH_REGEX = re.compile(r"[1-9][0-9]*") XML_LANG_ATTRIBUTE = "{http://www.w3.org/XML/1998/namespace}lang" +LATEX_SPAN_XPATH = etree.XPath(f"//*[local-name()='span'][@{LATEX_ATTRIBUTE}]") +UNCONVERTED_LATEX_TAG = "unconverted-latex" def img_pixel_size(width, height) -> Optional[Tuple[int, int]]: @@ -136,3 +141,21 @@ def strip_studio_attributes(raw_data): if name.startswith("data-studio-"): del element.attrib[name] return _serialize(doc) + + +def render_qti_latex_spans(raw_data): + if LATEX_ATTRIBUTE not in raw_data: + return raw_data + doc = parse_qti_xml(raw_data) + for span in LATEX_SPAN_XPATH(doc): + latex = span.attrib.pop(LATEX_ATTRIBUTE) + mathml = latex_to_mathml(latex) + if mathml is None: + span.text = latex_source(latex) + span.tag = UNCONVERTED_LATEX_TAG + continue + math = etree.fromstring(mathml) + math.tail = span.tail + span.getparent().replace(span, math) + etree.strip_tags(doc, UNCONVERTED_LATEX_TAG) + return _serialize(doc)
Is even?
{hint}
First hint.
Here's an equation:
Test this " - '' + '' "Cost = 1.00 $USD$" '\\text{Cost = 1.00 $USD$}' "
First: " - 'a+b' + 'a+b' 'a + b' ", then " - 'c×d' + 'c×d' 'c \\times d' ", finally " - 'ef' + 'ef' 'e^f' ".
This is inline math: " - 'a=b' + 'a=b' 'a = b' "
And this is block math:
Back to text with more inline: " - 'z2' + 'z2' 'z^2' "
The sum " - '∑i=1' + '' + "∑i=1" "nxi" '\\sum_{i=1}^{n} x_i' " is finite.
The variable " - 'x' + 'x' 'x' " is unknown.
" + '5' + '5' + " exactly
" + 'a' + 'a' + " and " + 'b' + 'b' + "
It is $${latex}$$ here.
Before
$$x_$$
After
It is $$a & c$$ here
{escapeHtml(latex_source(latex))}